Att G - IHC_FocusGroupHandouts

Attachment G
Form Approved
OMB No. 0920-1067
Exp. Date: xx/xx/20xx

Principles of Analytic Validation of
Immunohistochemical Assays
Summary of Recommendations
Guideline Statement

Strength of Recommendation

1. Laboratories must validate all IHC tests before placing into clinical
service.

Recommendation

Note: Such means include (but are not necessarily limited to):
Correlating the new test’s results with the morphology and expected results;
Comparing the new test’s results with the results of prior testing of the
same tissues with a validated assay in the same laboratory;
Comparing the new test’s results with the results of testing the same tissue
validation set in another laboratory using a validated assay;
Comparing the new test’s results with previously validated
non-immunohistochemical tests; or
Testing previously graded tissue challenges from a formal proficiency
testing program (if available) and comparing the results with the graded
responses.

2. For initial validation of every assay used clinically, with the exception
of HER2/neu, ER, and PgR (for which established validation
guidelines already exist), laboratories should achieve at least 90%
overall concordance between the new test and the comparator test
or expected results. If concordance is less than 90%, laboratories
need to investigate the cause of low concordance.

Recommendation

3. For initial analytic validation of nonpredictive factor assays,
laboratories should test a minimum of 10 positive and 10 negative
tissues. When the laboratory medical director determines that fewer
than 20 validation cases are sufficient for a specific marker (eg, rare
antigen), the rationale for that decision needs to be documented.

Expert Consensus Opinion

Note: The validation set should include high and low expressors for positive
cases when appropriate, and should span the expected range of clinical
results (expression levels) for markers that are reported quantitatively.

4. For initial analytic validation of all laboratory-developed predictive
marker assays (with the exception of HER2/neu, ER and PgR),
laboratories should test a minimum of 20 positive and 20 negative
tissues. When the laboratory medical director determines that fewer
than 40 validation tissues are sufficient for a specific marker, the
rationale for that decision needs to be documented.

Expert Consensus Opinion

Note: Positive cases in the validation set should span the expected range of
clinical results (expression levels). This recommendation does not apply to
any marker for which a separate validation guideline already exists.

5. For a marker with both predictive and nonpredictive applications,
laboratories should validate it as a predictive marker if it is used as
such.

Recommendation

CDC estimates the average public reporting burden for this collection of information as 60 minutes per response, including the time for reviewing instructions,
searching existing data/information sources, gathering and maintaining the data/information needed, and completing and reviewing the collection of
information. An agency may not conduct or sponsor, and a person is not required to respond to a collection of information unless it displays a currently valid
OMB control number. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing
this burden to CDC/ATSDR Information Collection Review Office, 1600 Clifton Road NE, MS D-74, Atlanta, Georgia 30333; ATTN: PRA (0920-1067)
© 2015 College of American Pathologists. All rights reserved.
23685.060215

cap.org

Guideline Statement

Strength of Recommendation

6. When possible, laboratories should use validation tissues that have
been processed using the same fixative and processing methods as
cases that will be tested clinically.

Recommendation

7. If IHC is regularly done on cytologic specimens that are not processed Expert Consensus Opinion
in the same manner as the tissues used for assay validation
(eg, alcohol-fixed cell blocks, air-dried smears, formalin postfixed
specimens), laboratories should test a sufficient number of such
cases to ensure that assays consistently achieve expected results.
The laboratory medical director is responsible for determining the
number of positive and negative cases and the number of predictive
and nonpredictive markers to test.
8. If IHC is regularly done on decalcified tissues, laboratories should
test a sufficient number of such tissues to ensure that assays
consistently achieve expected results. The laboratory medical
director is responsible for determining the number of positive and
negative issues and the number of predictive and nonpredictive
markers to test.

Expert Consensus Opinion

9. Laboratories may use whole sections, TMAs and/or MTBs in their
validation sets as appropriate. Whole sections should be used if
TMAs/MTBs are not appropriate for the targeted antigen or if the
laboratory medical director cannot confirm that the fixation and
processing of TMAs/ MTBs is similar to clinical specimens.

Recommendation

10. When a new reagent lot is placed into clinical service for an existing
Expert Consensus Opinion
validated assay, laboratories should confirm the assay’s performance
with at least 1 known positive case and 1 known negative case.
11. Laboratories should confirm assay performance with at least 2
known positive and 2 known negative cases when an existing
validated assay has changed in any one of the following ways:
Antibody dilution;
Antibody vendor (same clone);
Incubation or retrieval times (same method).

Expert Consensus Opinion

12. Laboratories should confirm assay performance by testing a
sufficient number of cases to ensure that assays consistently
achieve expected results when any of the following have changed:
Fixative type;
Antigen retrieval method (eg, change in pH, different buffer,
different heat platform);
Antigen detection system;
Tissue processing or testing equipment;
Environmental conditions of testing (eg, laboratory relocation);
Laboratory water supply.

Expert Consensus Opinion

The laboratory medical director is responsible for determining how
many predictive and nonpredictive markers and how many positive
and negative tissues to test.
13. Laboratories should run a full revalidation (equivalent to initial
analytic validation) when the antibody clone is changed for an
existing validated assay.

Expert Consensus Opinion

14. The laboratory must document all validations and verifications in
compliance with regulatory and accreditation requirements.

Expert Consensus Opinion

Abbreviations: IHC, immunohistochemistry; ER, estrogen receptor; PgR, progesterone receptor; TMA, tissue microarray; MTB, multitissue
block
Source: Fitzgibbons PL, Bradley LA, Fatheree LA, et al. Principles of analytic validation of immunohistochemical assays: Guideline from the
College of American Pathologists Pathology and Laboratory Quality Center. Arch Pathol Lab Med. 2014;138(11):1432–1443.
© 2015 College of American Pathologists. All rights reserved.
23685.060215

cap.org

CAP Laboratory Improvement Programs

Principles of Analytic Validation
of Immunohistochemical Assays
Guideline From the College of American Pathologists Pathology
and Laboratory Quality Center
Patrick L. Fitzgibbons, MD; Linda A. Bradley, PhD; Lisa A. Fatheree, BS, SCT(ASCP); Randa Alsabeh, MD;
Regan S. Fulton, MD, PhD; Jeffrey D. Goldsmith, MD; Thomas S. Haas, DO; Rouzan G. Karabakhtsian, MD, PhD;
Patti A. Loykasek, HT(ASCP); Monna J. Marolt, MD; Steven S. Shen, MD, PhD; Anthony T. Smith, MLS; Paul E. Swanson, MD

 Context.—Laboratories must validate all assays before
they can be used to test patient specimens, but currently
there are no evidence-based guidelines regarding validation of immunohistochemical assays.
Objective.—To develop recommendations for initial
analytic validation and revalidation of immunohistochemical assays.
Design.—The College of American Pathologists Pathology and Laboratory Quality Center convened a panel of
pathologists and histotechnologists with expertise in
immunohistochemistry to develop validation recommendations. A systematic evidence review was conducted to
address key questions. Electronic searches identified 1463
publications, of which 126 met inclusion criteria and were
extracted. Individual publications were graded for quality,
Accepted for publication February 3, 2014.
Published as an Early Online Release March 19, 2014.
Supplemental digital content is available for this article at www.
archivesofpathology.org in the November 2014 table of contents.
From the Department of Pathology, St. Jude Medical Center,
Fullerton, California (Dr Fitzgibbons); the Department of Pathology
and Laboratory Medicine, Women & Infants Hospital/Brown
University, Providence, Rhode Island (Dr Bradley); the College of
American Pathologists, Northfield, Illinois (Ms Fatheree and Mr
Smith); the Department of Pathology, Kaiser Permanente - Los
Angeles Medical Center, Los Angeles, California (Dr Alsabeh);
PhenoPath Laboratories, Seattle, Washington (Dr Fulton); the
Department of Pathology, Beth Israel Deaconess Medical Center,
Boston, Massachusetts (Dr Goldsmith); the Department of Pathology,
Mercy Hospital, Janesville, Wisconsin (Dr Haas); the Department of
Pathology, Montefiore Medical Center, New York, New York (Dr
Karabakhtsian); Regional Medical Laboratory, St John’s Medical
Center, Tulsa, Oklahoma (Ms Loykasek); the Department of
Pathology, University of Minnesota Medical Center, Fairview,
Minneapolis (Dr Marolt); the Department of Pathology, The
Methodist Hospital, Houston, Texas (Dr Shen); and the Department
of Pathology, University of Washington Medical Center, Seattle (Dr
Swanson).
Authors’ disclosures of potential conflicts of interest and author
contributions are found in the appendix at the end of this article.
Reprints: Patrick L. Fitzgibbons, MD, Department of Pathology, St.
Jude Medical Center, 101 E. Valencia Mesa Dr, Fullerton, CA 92835
(e-mail: [email protected]).
For additional questions and comments, contact the Pathology and
Laboratory Quality Center at [email protected].
1432 Arch Pathol Lab Med—Vol 138, November 2014

and the key question findings for strength of evidence.
Recommendations were derived from strength of evidence,
open comment feedback, and expert panel consensus.
Results.—Fourteen guideline statements were established
to help pathology laboratories comply with validation and
revalidation requirements for immunohistochemical assays.
Conclusions.—Laboratories must document successful
analytic validation of all immunohistochemical tests before
applying to patient specimens. The parameters for cases
included in validation sets, including number, expression
levels, fixative and processing methods, should take into
account intended use and should be sufficient to ensure
that the test accurately measures the analyte of interest in
specimens tested in that laboratory. Recommendations are
also provided for confirming assay performance when
there are changes in test methods, reagents, or equipment.
(Arch Pathol Lab Med. 2014;138:1432–1443; doi:
10.5858/arpa.2013-0610-CP)

I

mmunohistochemical (IHC) testing is an essential component of the pathologic evaluation of many specimens
and increasingly provides key information that helps
determine how patients are treated. As with any test,
laboratories must ensure that IHC test results are accurate
and reproducible and that the test performs as intended.
Laboratories subject to US regulations are required by the
Clinical Laboratory Improvement Amendments of 1988
(CLIA) to verify the performance characteristics of any assay
used in patient testing before it is placed into clinical
service.1,2
Before reporting patient results for unmodified US Food
and Drug Administration (FDA)–cleared or FDA-approved
tests, laboratories must demonstrate performance characteristics for accuracy, precision, and reportable range of test
results that are comparable to those established by the
manufacturer. The laboratory medical director must determine the extent to which these performance specifications
are verified, based on the method, testing conditions, and
personnel performing the test. Manufacturers of FDAapproved or FDA-cleared test kits may provide the user
with recommendations and directions for verifying that the
Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al

kit is performing according to the manufacturer’s specification. Typically, this is performed by testing known positive
and negative samples that either are supplied by the
manufacturer or have been tested by a validated reference-laboratory method.
Laboratories that introduce non–FDA-approved or non–
FDA-cleared tests (laboratory-developed tests) or modify
FDA-cleared or FDA-approved test systems (laboratorymodified tests) must, before reporting patient test results,
establish performance specifications for accuracy, precision,
analytic sensitivity, analytic specificity, reportable range, and
reference intervals.1 For tests that are reported qualitatively
or semi-quantitatively (most IHC tests), reportable range
and reference intervals are generally not applicable.
Good laboratory practice requires establishing optimal
antibody concentration and antigen retrieval and detection
methods. Analytic validation follows assay optimization and
is done by testing an appropriate tissue set to determine
analytic sensitivity and specificity. For tests without a gold
standard referent test, this usually involves determining
overall concordance with an appropriate comparator.
Validation procedures are intended to reasonably assure
that the test performs as expected. Once validation has been
completed, assays must be regularly monitored to detect
changes in analytic performance, usually by daily quality
control, periodic proficiency testing, and comparing positivity rates for selected markers (eg, hormone receptors,
HER2/neu) with expected positivity rates. Ongoing monitoring of assay performance is as important as initial assay
validation.
Although IHC test methods have steadily improved with
the introduction of automated staining platforms and
improved antigen retrieval and detection systems, results
are still affected by various preanalytic and analytic factors,
and the need for assay validation and ongoing monitoring
has not diminished. Assay validation is particularly important when a polymer-based detection system is used and a
negative reagent control is omitted. The College of
American Pathologists (CAP) Laboratory Accreditation
Program (LAP) accepts omission of this control, but only
if the assay has been properly validated (LAP checklist
ANP.22570).3
Unfortunately, recent studies4,5 have found significant
interlaboratory variation in validation practices and revealed
that many laboratories do not follow consistent procedures
when validating IHC assays. Comments received during the
open comment period for this guideline also revealed a
surprising lack of understanding among some respondents
of requirements for analytic validation. To address this
important shortfall in laboratory practice, the CAP convened
representatives to systematically review the published data
and develop evidence-based recommendations for analytic
validation of IHC assays.

Conflict of Interest Policy
Before acceptance on the expert or advisory panel, potential
members completed the CAP conflict of interest disclosure process,
whose policy and form (in effect April 2010) require disclosure of
material financial interest in or potential for benefit of significant
value from the guideline’s development or its recommendations 12
months prior through the time of publication. Potential members
completed the conflict of interest disclosure form, listing any
relationship that could be interpreted as constituting an actual,
potential, or apparent conflict. Everyone was required to disclose
conflicts before beginning and continuously throughout the
project’s timeline. One expert panel member (R.S.F.) was recused
from discussion and voting on the recommendation pertaining to
tissue microarrays, and one (T.S.H.) was recused from voting on
recommendations pertaining to potential increased antibody usage.
Expert panel members’ disclosed conflicts are listed in the
Appendix. The CAP provided funding for the administration of
the project; no industry funds were used in the development of the
guideline. All panel members volunteered their time and were not
compensated for their involvement. Please see the supplemental
digital content for full details on the conflict of interest policy.

Objective
The panel addressed the overarching question, ‘‘What is needed
for initial analytic assay validation before placing any IHC test into
clinical service and what are the revalidation requirements?’’ The
scope questions are as follows:
1. When and how should validation assess analytic sensitivity,
analytic specificity, accuracy (assay concordance), and precision
(interrun and interoperator variability)?
2. What is the minimum number of positive and negative cases
that need to be tested to analytically validate an IHC assay for
its intended use(s)?
3. What parameters should be specified for the tissues used in the
validation set?
4. How do certain preanalytic variables influence analytic validation?
5. What conditions require assay revalidation?

Literature Search and Selection
Electronic searches of the English language–published literature
in Ovid MEDLINE, US National Library of Medicine PubMed, and
Elsevier Scopus databases were initially conducted for the time
period spanning January 2004 to May 2012; an update was
conducted through May 2013. In addition to peer-reviewed journal
articles, the search identified books, book chapters, and published
abstracts from English-language sources. Bibliographies of included articles were hand searched, and additional information was
sought through targeted grey literature electronic searches (eg,
Google) and review of laboratory compliance and guidance Web
sites (eg, Clinical and Laboratory Standards Institute, FDA,
National Guideline Clearinghouse, Wiley Cochrane Library).

Inclusion Criteria

METHODS
A detailed description of the methods and systematic review
(including the 7 key questions, quality assessment, and complete
analysis of the evidence) used to create this guideline can be
found in the supplemental digital content available at www.
archivesofpathology.org in the November 2014 table of contents.

Panel Composition
The CAP Pathology and Laboratory Quality Center (the Center)
convened expert and advisory panels consisting of members with
expertise in immunohistochemistry. Panel members included
Arch Pathol Lab Med—Vol 138, November 2014

pathologists, histotechnologists, methodologists, and CAP staff.
CAP approved the appointment of the project chair (P.L.F.) and
panel members.

Published studies were selected for full-text review if they met
each of the following criteria:
1. English-language articles/documents that addressed IHC and
provided data or information relevant to 1 or more key
questions;
2. Study designs that included validation, method comparison,
cohort or case-control studies, clinical trials, and systematic
reviews, as well as qualitative information from consensus
guidelines, regulatory documents, and US or international
proficiency testing reports; and

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al 1433

3. Articles/documents focused on the clinical use of IHC for
identification of predictive and nonpredictive markers and
analytic variables.

Exclusion Criteria
Editorials, letters, commentaries, and invited opinions were not
included in the study. Articles were also excluded if the full article
was not available in English, did not address any key question, and/
or focused primarily on assay optimization, quality control or
quality assurance, basic or nonhuman research, nontissue immunoassays, preanalytic and postanalytic variables, or clinical validation only.

Table 1.
Grade

Description

Convincing

Two or more level 1a or level 2b studies
(study design and execution) that had an
appropriate number and distribution of
challengesc and reported consistentd and
generalizablee results.
One level 1 or level 2 study that had an
appropriate number and distribution of
challenges and reported generalizable
results.
Two or more level 1 or level 2 studies that
lacked the appropriate number and
distribution of challenges OR were
consistent but not generalizable.
Combinations of level 1 or level 2 studies
that show unexplained inconsistencies OR
1 or more level 3f or level 4g studies OR
expert opinion.

Adequate

Quality Assessment
Grading the quality of individual studies was performed from
study design–specific criteria by the methodology consultant
(L.A.B.), with input as needed from the expert panel. The aim of
analytic validation is to determine a test’s ability to accurately and
reliably detect the antigen or marker of interest in specimens
consistent with those to be tested in clinical practice.6 Analytic
validity studies have a different design, compared to studies of
diagnostic accuracy or therapeutic interventions. For this reason,
the criteria needed to assess the quality of analytic validity studies
are different. Quality in this context is considered to be essentially
equivalent to internal validity and is assessed on the basis of study
design and execution, analyses, and reporting.6 The strength of
evidence for individual key questions or outcomes was assessed by
using published criteria.6 The criteria included the quality and
execution of studies, the quantity of data (number and size of
studies), and the consistency and generalizability of the evidence
across studies.6 Strength of evidence was graded convincing,
adequate, or inadequate (Table 1).

Assessing the Strength of Recommendations
Development of recommendations requires that the panel review
the identified evidence and make a series of key judgments. Grades
for strength of recommendations were developed by the CAP
Pathology and Laboratory Quality Center and are described in
Table 2.

Guideline Revision
This guideline will be reviewed every 4 years, or earlier in the
event of publication of substantive and high-quality evidence that
could potentially alter the original guideline recommendations. If
necessary, the entire panel will reconvene to discuss potential
changes. When appropriate, the panel will recommend revision of
the guideline to CAP for review and approval.

Disclaimer
The CAP developed the Pathology and Laboratory Quality
Center as a forum to create and maintain evidence-based practice
guidelines and consensus statements. Practice guidelines and
consensus statements reflect the best available evidence and expert
consensus supported in practice. They are intended to assist
physicians and patients in clinical decision making and to identify
questions and settings for further research. With the rapid flow of
scientific information, new evidence may emerge between the time
a practice guideline or consensus statement is developed and when
it is published or read. Guidelines and statements are not
continually updated and may not reflect the most recent evidence.
Guidelines and statements address only the topics specifically
identified therein and are not applicable to other interventions,
diseases, or stages of diseases. Furthermore, guidelines and
statements cannot account for individual variation among patients
and cannot be considered inclusive of all proper methods of care or
exclusive of other treatments. It is the responsibility of the treating
physician or other health care provider, relying on independent
experience and knowledge, to determine the best course of
treatment for the patient. Accordingly, adherence to any practice
guideline or consensus statement is voluntary, with the ultimate
1434 Arch Pathol Lab Med—Vol 138, November 2014

Grades for Strength of Evidence

Inadequate

From Teutsch et al.6 Reprinted with permission from Macmillan
Publishers Ltd.
a
Level 1 study: Collaborative study using a large panel of wellcharacterized samples; summary data from external proficiencytesting schemes or interlaboratory comparisons.
b
Level 2 study: High-quality peer-reviewed studies (eg, method
comparisons, validation studies).
c
Based on number of possible response categories and required
confidence in results.
d
Consistency assessed by using central estimates/ranges or testing for
result homogeneity.
e
Generalizability is the extension of findings and conclusions from 1
study to other settings.
f
Level 3 study: Lower-quality peer-reviewed studies OR expert panel–
reviewed US Food and Drug Administration summaries.
g
Level 4 study: Unpublished or non–peer-reviewed data.

determination regarding its application to be made by the physician
in light of each patient’s individual circumstances and preferences.
CAP makes no warranty, express or implied, regarding guidelines
and statements and specifically excludes any warranties of
merchantability and fitness for a particular use or purpose. CAP
assumes no responsibility for any injury or damage to persons or
property arising out of or related to any use of this statement or for
any errors or omissions.

RESULTS
Of the 1463 studies identified by electronic searches, 126
met inclusion criteria and underwent data extraction. These
included 122 published peer-reviewed articles, 2 book
chapters, and 2 grey literature documents. Among the
extracted documents, 43 did not meet minimum quality
standards, presented incomplete data or data that were not
in useable formats, or included only information based on
expert opinion. These articles were not included in analyses
or narrative summaries. The expert panel met 28 times by
teleconference Webinar from June 2010 through September
2013 and met in person on May 11 and May 12, 2013, to
review evidence to date and draft recommendations.
Additional work was completed via electronic mail. An
open comment period was held from July 8 through July 29,
2013. Eighteen draft recommendations and 5 methodology
questions were posted online on the CAP Web site.
A total of 1071 comments were received from 263
respondents (‘‘agree’’ and ‘‘disagree’’ responses were also
captured). Twelve of 18 draft recommendations achieved
more than 80% agreement; only 2 had less than 70%
agreement. Each expert panel member was assigned 1 to 2
draft recommendations for which to review all comments
Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al

Table 2.
Designation

Grades for Strength of Recommendations
Recommendation

Rationale

Strong recommendation

Recommend for or against a particular analytic
validation practice (can include must or
should).

Recommendation

Recommend for or against a particular analytic
validation practice (can include should or
may).

Expert consensus opinion

Recommend for or against a particular analytic
validation practice (can include should or
may).

Strength of evidence is convincing, based on
consistent, generalizable, good-quality
evidence; further studies are unlikely to
change the conclusions.
Strength of evidence is adequate, based on
limitations in the quality of evidence;
further studies may change the
conclusions.
Important validation element to address but
strength of evidence is inadequate; gaps in
knowledge may require further studies.

received and provide an overall summary to the rest of the
panel. Three draft recommendations were maintained with
the original language; 5 were modified with minor changes
for clarification and/or further explanation within the
manuscript, and 6 were considered extremely discordant
with major revisions made accordingly for a total of 14 final
recommendations. Resolution of all changes was obtained
by majority consensus of the panel. The final recommendations were approved by the expert panel with a formal
vote (with specific abstentions from R.S.F. and T.S.H.). The
panel considered laboratory redundancy, efficiency, and
feasibility throughout the whole process. Formal cost
analysis or cost effectiveness was not performed.
An independent review panel, masked to the expert panel
and vetted through the conflict of interest process, provided
final review of the guideline and recommended it for
approval by the CAP. The final recommendations are
summarized in Table 3.
Guideline Statements
1: Recommendation.—Laboratories must validate all
immunohistochemical tests before placing into clinical
service.
Note: Such means include (but are not necessarily limited to):
1. Correlating the new test’s results with the morphology
and expected results;
2. Comparing the new test’s results with the results of prior
testing of the same tissues with a validated assay in the
same laboratory;
3. Comparing the new test’s results with the results of
testing the same tissue validation set in another
laboratory using a validated assay;
4. Comparing the new test’s results with previously
validated non-immunohistochemical tests; or
5. Testing previously graded tissue challenges from a
formal proficiency testing program (if available) and
comparing the results with the graded responses.
The strength of evidence was adequate to support when
analytic validation should be done and that it should include
determination of analytic sensitivity and specificity (or
concordance in the absence of a gold standard referent test)
and precision (eg, interrun and interoperator) as part of
validation. The evidence was inadequate (ie, evidence was
not available or did not permit a conclusion to be reached)
to assess the precision of IHC assays in practice or how
validation should be done with regard to the listed
approaches, but did show that these approaches have been
used. The panel found that analytic validation provides a net
benefit for the overall performance and safety of IHC tests
Arch Pathol Lab Med—Vol 138, November 2014

by contributing to the avoidance of potential harms related
to analytic false-positive and false-negative test results.
Laboratories are required by CLIA (section 493.1253) to
validate the performance characteristics of all assays used in
patient testing, in order to ensure that the results are
accurate and reproducible.7 This includes establishment of
the analytic validity of all non–FDA-cleared/approved (or
‘‘laboratory-developed’’) tests.7 For qualitative assays such
as IHC, validation usually requires comparing a new assay’s
results with a reference standard and calculating estimates
of analytic sensitivity and specificity; however, because there
are no gold standard referent tests for most IHC assays,
laboratories must use another means of demonstrating that
the assay performs as expected.8–10 Publications addressing
IHC validation include independent comparisons of a new
test’s results to clinical outcomes, other validated IHC tests
(intralaboratory or interlaboratory), or previously characterized tissue validation sets.9,11–19 Non-immunohistochemical
tests may include in situ hybridization, flow cytometry, and
molecular, cytogenetic, or microbiologic studies. Laboratories may use a combination of comparison methods when
appropriate.
When correlating the new test’s results with expected
results, positive and negative tissues pertinent to each
intended clinical use must be included in the validation set.
Normal tissues (with 100% positive staining expected)
cannot comprise the entire validation set for markers
primarily used in diagnosing neoplasms, but may be used
in conjunction with neoplastic and lesional tissue as
appropriate. In some cases a section of tissue may contain
both antigen-positive cells and negative internal control
cells, and therefore serve as both a positive and negative
validation challenge. The laboratory medical director must
determine the most appropriate selection of tissues in the
validation set, but the validation set must not consist solely
of the same tissues used for antibody optimization.
Although not currently available for many markers, excess
tissue previously used in a proficiency testing or interlaboratory comparison program could also be used for assay
validation. Tissue from previously graded proficiency-testing
challenges could be tested and the results compared with
the graded responses from the program.
This recommendation applies to all assays in clinical use
(including those for pathogen-specific antigens such as
cytomegalovirus and Helicobacter pylori) irrespective of the
regulatory status of the primary antibody (eg, in vitro
diagnostic, analyte-specific reagent).
2: Recommendation.—For initial validation of every
assay used clinically, with the exception of HER2/neu,
estrogen receptor (ER), and progesterone receptor (PgR)

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al 1435

Table 3.

Guideline Statements and Strength of Recommendations
Strength of
Recommendation

Guideline Statement
1. Laboratories must validate all IHC tests before placing into clinical service.
Note: Such means include (but are not necessarily limited to):
Correlating the new test’s results with the morphology and expected results;
Comparing the new test’s results with the results of prior testing of the same tissues with a
validated assay in the same laboratory;
Comparing the new test’s results with the results of testing the same tissue validation set in
another laboratory using a validated assay;
Comparing the new test’s results with previously validated non-immunohistochemical tests; or
Testing previously graded tissue challenges from a formal proficiency testing program (if
available) and comparing the results with the graded responses.
2. For initial validation of every assay used clinically, with the exception of HER2/neu, ER, and PgR
(for which established validation guidelines already exist), laboratories should achieve at least
90% overall concordance between the new test and the comparator test or expected results. If
concordance is less than 90%, laboratories need to investigate the cause of low concordance.
3. For initial analytic validation of nonpredictive factor assays, laboratories should test a minimum of
10 positive and 10 negative tissues. When the laboratory medical director determines that fewer
than 20 validation cases are sufficient for a specific marker (eg, rare antigen), the rationale for that
decision needs to be documented.
Note: The validation set should include high and low expressors for positive cases when
appropriate and should span the expected range of clinical results (expression levels) for
markers that are reported quantitatively.
4. For initial analytic validation of all laboratory-developed predictive marker assays (with the
exception of HER2/neu, ER, and PgR), laboratories should test a minimum of 20 positive and 20
negative tissues. When the laboratory medical director determines that fewer than 40 validation
tissues are sufficient for a specific marker, the rationale for that decision needs to be documented.
Note: Positive cases in the validation set should span the expected range of clinical results
(expression levels). This recommendation does not apply to any marker for which a separate
validation guideline already exists.
5. For a marker with both predictive and nonpredictive applications, laboratories should validate it as
a predictive marker if it is used as such.
6. When possible, laboratories should use validation tissues that have been processed by using the
same fixative and processing methods as cases that will be tested clinically.
7. If IHC is regularly done on cytologic specimens that are not processed in the same manner as the
tissues used for assay validation (eg, alcohol-fixed cell blocks, air-dried smears, formalin-postfixed
specimens), laboratories should test a sufficient number of such cases to ensure that assays
consistently achieve expected results. The laboratory medical director is responsible for
determining the number of positive and negative cases and the number of predictive and
nonpredictive markers to test.
8. If IHC is regularly done on decalcified tissues, laboratories should test a sufficient number of such
tissues to ensure that assays consistently achieve expected results. The laboratory medical director
is responsible for determining the number of positive and negative tissues and the number of
predictive and nonpredictive markers to test.
9. Laboratories may use whole sections, TMAs, and/or MTBs in their validation sets as appropriate.
Whole sections should be used if TMAs/MTBs are not appropriate for the targeted antigen or if the
laboratory medical director cannot confirm that the fixation and processing of TMAs/ MTBs is
similar to clinical specimens.
10. When a new reagent lot is placed into clinical service for an existing validated assay, laboratories
should confirm the assay’s performance with at least 1 known positive case and 1 known negative
case.
11. Laboratories should confirm assay performance with at least 2 known positive and 2 known
negative cases when an existing validated assay has changed in any one of the following ways:
Antibody dilution;
Antibody vendor (same clone);
Incubation or retrieval times (same method).
12. Laboratories should confirm assay performance by testing a sufficient number of cases to ensure
that assays consistently achieve expected results when any of the following have changed:
Fixative type;
Antigen retrieval method (eg, change in pH, different buffer, different heat platform);
Antigen detection system;
Tissue processing or testing equipment;
Environmental conditions of testing (eg, laboratory relocation);
Laboratory water supply.
The laboratory medical director is responsible for determining how many predictive and
nonpredictive markers and how many positive and negative tissues to test.
13. Laboratories should run a full revalidation (equivalent to initial analytic validation) when the
antibody clone is changed for an existing validated assay.
14. The laboratory must document all validations and verifications in compliance with regulatory and
accreditation requirements.

Recommendation

Recommendation

Expert consensus opinion

Expert consensus opinion

Recommendation
Recommendation
Expert consensus opinion

Expert consensus opinion

Recommendation

Expert consensus opinion
Expert consensus opinion

Expert consensus opinion

Expert consensus opinion
Expert consensus opinion

Abbreviations: ER, estrogen receptor; IHC, immunohistochemistry; MTBs, multitissue blocks; PgR, progesterone receptor; TMAs, tissue microarrays.

1436 Arch Pathol Lab Med—Vol 138, November 2014

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al

(for which established validation guidelines already exist),
laboratories should achieve at least 90% overall concordance
between the new test and the comparator test or expected
results. If concordance is less than 90%, laboratories need to
investigate the cause of low concordance.
Strength of evidence was adequate to support a 90%
(versus 95%) overall concordance benchmark for analytic
validation of IHC tests (excepting HER2/neu, ER, PgR).
Supporting evidence for this recommendation is obtained
from published IHC validation studies, method comparisons, and proficiency testing or interlaboratory comparisons.
Examples include the following:
1. Median overall concordance in a 2-year interlaboratory
comparison of CD117 IHC and target results was
87.6%.20
2. Median overall concordance in 5 comparisons of
different HER2/neu IHC tests was 89.0% (range, 74%–
92%), with 2 of 5 studies greater than 90% concordant.13–16,19
3. Median overall concordance in 5 comparisons of HER2/
neu IHC tests to HER2/neu in situ hybridization tests was
88.2% (range, 66%–94%), with 2 of 5 comparisons
greater than 90% concordant.17,20–22
4. Median overall concordance in 6 comparisons of IHC
tests (PTEN [phosphatase and tensin homologue deleted
on chromosome 10], ER, PR, HER2/neu, MPT64, p16) to
alternative referent tests (eg, RNA expression, clinical
diagnosis) was 91.4% (range, 74%–99%), with 3 of 6
studies greater than 90% concordant.12,17,21–23
Summary concordance estimates (using a random effects
model) provided similar concordance estimates, but heterogeneity was high (I2 . 75% in all cases; P , .001) and could
not be explained by analysis of selected covariates (eg, tissue
type, antibody, study quality grade). The number of studies
was too small to allow analysis of the many possible
covariates.
These data illustrate the challenge of achieving an overall
concordance of 95%, even in large studies of IHC tests with
guidance recommending stringent protocol standards (ie,
HER2/neu, ER, PgR).10,24–26 Overall concordance of 90% was
achieved in nearly half of the above analyzed comparisons,
all of which were subject to many sources of variation (eg,
sample type; ischemic time; fixation, antigen retrieval, and
staining protocols; scoring). Therefore, laboratory validation
studies designed to minimize differences in such variables
would have a higher probability of meeting a 90%
concordance benchmark.
If the overall concordance estimate in an assay validation
study is less than 90%, laboratories should calculate positive
and negative concordance rates as well as the discordance
(using the McNemar test when sample size is appropriate)
to help investigate the cause of low concordance. The
McNemar test assesses the significance of the difference
between the discordant results (false positives and negatives) in a 2 3 2 contingency table. Refer to the supplemental
digital content for more information and link to available
resources.
3: Expert Consensus Opinion.—For initial analytic
validation of nonpredictive factor assays, laboratories should
test a minimum of 10 positive and 10 negative tissues. When
the laboratory medical director determines that fewer than
20 validation cases are sufficient for a specific marker (eg,
Arch Pathol Lab Med—Vol 138, November 2014

rare antigen), the rationale for that decision needs to be
documented.
Note: The validation set should include high and low
expressors for positive cases when appropriate and should
span the expected range of clinical results (expression levels)
for markers that are reported quantitatively.
Strength of evidence was inadequate to support the
recommended number of validation samples, but was
adequate to support distinguishing nonpredictive from
predictive IHC tests and using different numbers of
validation samples for each.
A key criterion for determining the number of samples
needed to validate an IHC assay is the test’s intended use:
whether it is used alone or as part of a test panel and
interpreted only in the context of other morphologic and
clinical data (most nonpredictive markers) or as a standalone test reported to physicians as independent diagnostic
information that may directly determine treatment (most
predictive markers and selected pathogen-specific assays,
such as viral antigens in transplant patients), for which the
risk of an incorrect result must be minimized.5,8,27 Some
tests can fall into both categories. Other criteria for
determining the number of validation samples include the
complexity of interpretation (ie, multiple test outcomes and
result categories require more samples) and the number and
range of control materials available.8 For example, an IHC
test with 3 or more result categories would require a larger
number of samples to ensure validation than one interpreted only as positive or negative.8
Validity in laboratory practice must be based on objective
observations. The most practical objective guidance for
determining the size of a validation set is statistical analysis.
Not surprisingly, the more samples that are run in a
validation set, the higher the likelihood that the concordance estimate reflects the test’s ‘‘true’’ concordance;
increasing the number of samples in a validation set
increases the confidence that the assay performs as
expected. Table 4 illustrates overall concordance estimates
with 95% confidence interval (CI) for 10 and 20 sample
validation sets with 0 to 2 observed discordant results.
Using a 10-sample validation set, the overall concordance
estimate (ie, the level of agreement between 2 tests) reaches
the 90% concordance benchmark with only 1 discordant
result. This concordance estimate has a 95% CI (the range of
values that has a 95% chance of including the ‘‘true’’
concordance) of 57% to 100%. Using a 20-sample validation
set, overall concordance meets the 90% benchmark with 2
or fewer discordant results and a 95% CI of 69% to 98%.
Both the ‘‘true’’ concordance and the number of validation
samples have an impact on the probability that a test will
reach or exceed the overall concordance benchmark of 90%.
For example, if the 95% concordance estimate (1 discordant
result) in the 20-sample validation set is a ‘‘true’’ representation of the relationship between the 2 tests, the probability
of achieving the 90% benchmark would be very high (92%).
The probability of achieving the benchmark if the 90%
concordance estimate in the 20-sample set is a ‘‘true’’
representation would be 68% (Stat Trek Binomial Calculator, http://stattrek.com/online-calculator/binomial.aspx; accessed November 7, 2013).28
With this in mind, the panel determined that use of 10
samples (5 negative and 5 positive) in a validation set for a
nonpredictive marker assay provides unacceptably broad
CIs with either 100% (CI, 68%–100%) or 90% (CI, 57%–
100%) concordance estimates. For predictive markers,

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al 1437

Table 4.

Validation Using 10- and 20-Tissue Validation Sets Against a 90% Concordance Benchmark
Concordance Estimate, % (95% CI)

No.

0 Discordant

1 Discordant

2 Discordant

10
20

100 (68–100)
100 (81–100)

90 (57–100)
95 (75–100)

80 (48–95)
90 (69–98)

Abbreviations: CI, confidence interval; No., number of validation tissues.

however, the critical relationship between the antibody/
testing method and the actual presence of the target analyte
for purposes of guiding specific therapeutic intervention or
predicting treatment response requires an even higher level
of confidence (see recommendation No. 4).
Although analytic assay validation principles are independent of the frequency of testing or the availability of
appropriate validation samples, the panel recognized that it
may be difficult for some laboratories to obtain the
recommended minimum number of positive validation
specimens for rare antigens. Working with other laboratories to pool positive cases or using validation sets prepared
by other laboratories may allow laboratories to meet this
recommendation.
The laboratory medical director is ultimately responsible
for demonstrating the validity of each assay and in selected
instances may determine that a validation set smaller than
20 samples is sufficient. In such cases, the medical director
must also provide and document an objective rationale for
this determination.
For validation results that do not meet the 90% standard,
the medical director will be responsible for determining
both the basis for this result and the appropriate mitigation
(testing of additional tissues, change in test conditions, or
use of a different antibody). In general, assays that cannot be
validated against this standard should not be used in clinical
practice.
Some nonpredictive markers are reported quantitatively.
Examples include, but are not limited to, immunoglobulin
G4 (IgG4) in sclerosing inflammatory disorders, activated
caspase 3 or Microtubule-associated protein 1 light chain 3
in ischemia or sepsis, and Phosphohistone H3 as a surrogate
of mitotic figure count. For such markers, we recommend
that the validation set include high and low expressors to
ensure test accuracy over the analytic range.
4: Expert Consensus Opinion.—For initial analytic
validation of all laboratory-developed predictive marker
assays (with the exception of HER2/neu, ER, and PgR),
laboratories should test a minimum of 20 positive and 20
negative cases. When the laboratory medical director
determines that fewer than 40 validation cases are sufficient
for a specific marker, the rationale for that decision needs to
be documented.
Note: Positive cases in the validation set should span the
expected range of clinical results (expression levels). This
recommendation does not apply to any marker for which a
separate validation guideline already exists.
Strength of evidence was inadequate to support the
recommended number of validation samples, but was
adequate to support distinguishing nonpredictive from
predictive IHC tests and using different numbers of
validation samples for each.
The statistical argument is updated here for predictive
factor assays. Table 5 provides overall concordance estimates with 95% CIs for a 40-tissue validation set and for a
1438 Arch Pathol Lab Med—Vol 138, November 2014

20-tissue set for those who will compute positive and
negative concordance estimates.
Using a 40-sample validation set, the overall concordance
estimates meet the 90% benchmark with 4 or fewer
discordances. The ‘‘true’’ concordance between the 2 assays
has only a 5% chance of falling outside the 95% CIs of the
concordance estimates, and can be lower or higher than the
estimate. If the 95% to 100% concordance estimates for the
40-sample validation set are a ‘‘true’’ representation of the
relationship between the 2 tests, the validation results
would meet the benchmark more than 95% of the time with
0 to 2 observed discordant results. The probabilities of
meeting the benchmark if the 92.5% or 90% concordance
estimates are a ‘‘true’’ representation would be 82%
(approximation) and 63%, respectively (Binomial Calculator,
Stat Trek; http://stattrek.com/).
In a 40-sample validation that does not meet the
benchmark, analyses such as the McNemar test may help
determine whether an observed difference in the offdiagonal represents a significant bias between the new
and referent tests. Table 6 provides an example. In this case,
the j statistic showed ‘‘substantial’’ agreement, but the
overall concordance estimate (87.5%) missed the benchmark by a small margin. The positive concordance of 75%
suggests false negatives could be occurring in the new test,
but the McNemar test is not significant, indicating that the 5
discordant results all in a single cell could have happened by
chance.
Some laboratories may choose to validate predictive tests
with tissue sets larger than the recommended minimum. For
validation sets of 80 samples or more, the McNemar test is
more useful in documenting whether observed differences/
biases between the tests are significant. For example, for an
80-tissue validation set in which the numbers in each of the
4 cells in Table 6 are doubled, the McNemar result for 10 to
0 asymmetry on the off-diagonal would be significant (P ¼
.004).
For validation results that do not meet the 90% standard,
the laboratory medical director will be responsible for
determining both the basis for this result and the
appropriate mitigation (testing of additional tissues, change
in test conditions).
5: Recommendation.—For a marker with both predictive
and nonpredictive applications, laboratories should validate
it as a predictive marker if it is used as such.
Strength of evidence was adequate to support the use of
the higher validation standard (eg, number of samples) in
the case of a marker with both nonpredictive and predictive
intended uses.
Immunohistochemical assays have a variety of clinical
applications including cell, tissue, or microbiologic identification, tumor diagnosis and prognosis, genetic and cancer
risk assessment, and prediction of response to targeted
therapies (predictive markers).
Although most IHC assays are interpretable only within
the context of the clinical and histologic evaluation of the
Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al

Table 5.

Validation Using a 40-Tissue Validation Set (20 Positive and 20 Negative)
Against a 90% Concordance Benchmark
Concordance Estimate, % (95% CI)

No.

0 Discordant

1 Discordant

2 Discordant

3 Discordant

4 Discordant

20
40

100 (81–100)
100 (90–100)

95 (75–100)
97.5 (86–100)

90 (69–98)
95 (83–99)

85 (63–96)
92.5 (79–98)

80 (58–92)
90 (76–97)

Abbreviations: CI, confidence interval; No., number of validation tissues.

specific case, the results of predictive factor testing often
directly influence how patients are managed. Some IHC
assays are used for more than 1 purpose—the same antigen
may be assessed to determine a patient’s eligibility for a
targeted therapy as well as part of a panel in determining
tumor type.
Assay validation procedures must take into account the
test’s intended uses. When a marker will be used in both
predictive and nonpredictive applications, assay validation
should follow the recommendation for predictive markers
because of its greater stringency.
When assessing the analytic validity of a predictive
marker, cases should be selected to ensure that the new
assay is concordant with its comparator over the expected
range of clinical results. When validating the same marker
for nonpredictive uses, cases should be selected to ensure
that the test has acceptable concordance. Assays, such as ER
or CD117 (c-KIT), that have been optimized to detect low
levels of antigen for predictive uses could have high falsepositive results (low negative concordance) when used as a
lineage marker. Laboratories may choose to perform
separate validations for the marker’s predictive and nonpredictive applications.
6: Recommendation.—When possible, laboratories
should use validation tissues that have been processed with
the same fixative and processing methods as cases that will
be tested clinically.
Strength of evidence was inadequate to address the
influence of fixation, the type of decalcification solution,
the time in decalcification solution, or validation tissues
processed in another laboratory on analytic validation;
however, the strength of evidence was adequate to support
that laboratories should, whenever possible, use the same
fixative and processing methods as cases tested clinically, in
order to validate using representative specimens.
Fixative type, fixation time, tissue processing, and other
preanalytic variables significantly affect the performance
characteristics of IHC assays. To reduce the risk of falsenegative and false-positive comparisons, validation materials should be handled in a manner similar to clinical
specimens. Reference laboratories that test tissues from
outside facilities usually cannot control differences in
specimen handling and processing but should consider
such differences when interpreting results.
Key criteria in grading the quality and strength of
evidence for analytic validation include the internal validity
of the studies and the consistency and generalizability of the
results.6,29 To generalize the laboratory’s analytic validation
results, the tissues included in a validation set must be
representative of the specimens received in routine practice
and must provide a representative range of expression
intensities and patterns.
Although it is ideal if validation materials are identical to
patient test specimens (eg, formalin-fixed tissue sections;
Arch Pathol Lab Med—Vol 138, November 2014

cell blocks from cytologic specimens initially fixed in
alcohol; decalcified tissues), it is generally not practical to
maintain complete validation sets specific for all possible
specimen types, fixatives, and times in decalcification
solution. It is reasonable for laboratories to test a selected
panel of common markers to show that specimens of
different type or processed differently exhibit equivalent
immunoreactivity (LAP checklist ANP.22550).3
Note that there have been reports of false-positive and
false-negative reactions for some markers after alcohol
fixation. Although there are currently few data on this
subject and more evidence is needed, the laboratory medical
director should consider this possibility when selecting
markers for the panel.
7: Expert Consensus Opinion.—If IHC is regularly done
on cytologic specimens that are not processed in the same
manner as the tissues used for assay validation (eg, alcoholfixed cell blocks, air-dried smears, formalin-postfixed
specimens), laboratories should test a sufficient number of
such cases to ensure that assays consistently achieve
expected results. The laboratory medical director is responsible for determining the number of positive and negative
cases and the number of predictive and nonpredictive
markers to test.
The strength of evidence was inadequate to address the
criteria and number of samples needed for validation with
cytology specimens.
Laboratories typically optimize and validate their IHC
assays by using formalin-fixed, paraffin-embedded tissues
but may use cytologic specimens in some circumstances;
however, cytologic specimens usually have different fixation
and processing methods and these factors may have
unknown effects on IHC test results. Although separate
validation of all markers on all potential cytologic specimens
is generally not feasible, laboratories should determine

Table 6. 2 3 2 Contingency Table of a 40-Tissue
Validation Set That Did Not Meet the Benchmark
With Associated Statistical Testsa–c
Comparator Test
New Test
Positive
Negative
Total

Positive
15
5
20

Negative
0
20
20

Total
15
25
40

a

Overall concordance: 35 of 40 ¼ 87.5% (does not meet 90%
benchmark); positive concordance: 15 of 20 ¼ 75%; negative
concordance: 20 of 20 ¼ 100%.
b
j: 0.75; McNemar test: P ¼ .13.
c
The j statistic shows ‘‘substantial’’ agreement, but the overall
concordance estimate misses the 90% benchmark. Positive concordance of 75% could suggest that false negatives are occurring in the
new test, but the McNemar test is not significant, indicating that the 5
discordant results all in a single cell could have happened by chance.

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al 1439

whether cytologic specimens have equivalent immunoreactivity to routinely processed, formalin-fixed tissue.
To assess the extent to which differences in cytologic
specimen types and processing steps influence IHC test
results, laboratories should test a selected set of commonly
ordered markers (eg, keratin, CD45, S100, ER) in a set of
cytologic specimen types used for IHC staining. The results
should be correlated with expected results in routinely
processed (control) tissues and with other applicable test
results (eg, surgical specimen of primary neoplasm). The
laboratory medical director must determine the number of
cases and markers to test, bearing in mind the possibility of
spurious results in alcohol-fixed materials. This assessment
should be repeated when there is a change in cytologic
fixative, collection media, sample preparation, or processing.
If an assay has not been fully validated on cytologic
specimens, laboratories may include a disclaimer in their
report that results should be interpreted with caution.
No primary studies, systematic evidence reviews, or
qualitative documents were identified that addressed the
specific question regarding the number and type of cytology
specimens that are needed in a validation set for a new IHC
assay. Studies30–36 were identified that compared cytology
specimens to formalin-fixed tissue sections for ER, PgR,
and/or HER2/neu IHC testing. Most concordance estimates
were high (90%), but the studies were small and used
different fixatives, fixation times, and cytology specimen
types (eg, smears, thin-layer, cell blocks). No two studies
could be directly compared.
8: Expert Consensus Opinion.—If IHC is regularly
performed on decalcified tissues, laboratories should test a
sufficient number of such tissues to ensure that assays
consistently achieve expected results. The laboratory medical director is responsible for determining the number of
positive and negative tissues and the number of predictive
and nonpredictive markers to test.
The strength of evidence was inadequate to address the
criteria and number of samples needed for validation with
decalcified specimens.
Decalcifying solutions vary in their effects on retention
and integrity of nucleic acids and proteins. Results of IHC
testing on decalcified specimens are unpredictable because
of wide variations in specimen types and sizes, the length of
time specimens are held in decalcification solution, and the
particular solution(s) used. Although separate validation of
all markers on all potential decalcified specimen types is not
feasible, laboratories should determine the extent to which
their decalcification procedures affect test results, particularly among specimen types that commonly have IHC
testing, such as bone marrow biopsy samples.
No primary studies, systematic evidence reviews, or
qualitative documents (eg guidelines, consensus meeting
reports) were identified that address the specific question
regarding the number of decalcified bone marrow specimens
from positive and negative cases needed in a validation set
for a new IHC assay. Nine articles and documents25,26,37–43
addressed the potential influence of decalcification as a
modifier in the analytic validation process. Some authors26,38–40 report variability in decalcification protocols
and in preservation of antigenicity in IHC tests. Two IHC
guidelines recommend interpreting IHC results on decalcified samples with caution because of the possibility of
antigen (and tissue) loss, but others report good morphology
and successful staining with protocols using different
fixatives, acid or EDTA decalcification, and paraffin or resin
1440 Arch Pathol Lab Med—Vol 138, November 2014

embedding.37,40,42,43Although the evidence was inadequate,
these observations emphasize the need for a defined
protocol and a validation plan that will ensure robust and
reproducible IHC results in decalcified specimens.
Compared with other specimens, bone marrow biopsy
samples are more consistent in size and in the time needed
for decalcification, and are usually subject to standardized
processing and decalcification protocols. To assess the
influence of their decalcification procedure on IHC test
results in bone marrows, laboratories should test a selected
set of commonly ordered markers (eg, CD3, CD20, CD138)
in a series of cases. The results may be correlated with
expected results in routinely processed (control) tissues and
with other applicable test results (eg, flow cytometry, IHC
testing of lymph node in same patient). The laboratory
medical director must determine the number of cases and
markers to test. This assessment should be repeated when
there is a change in decalcifying solution or fixative type.
For specimen types other than bone marrow samples,
laboratories may include a disclaimer in their reports that
the assay has not been fully validated on decalcified tissues
and that results should be interpreted with caution given the
possibility of false negativity on decalcified specimens (LAP
checklist ANP.22985).3
9: Recommendation.—Laboratories may use whole
sections, tissue microarrays (TMAs), and/or multitissue
blocks (MTBs) in their validation sets as appropriate. Whole
sections should be used if TMAs/MTBs are not appropriate
for the targeted antigen or if the laboratory medical director
cannot confirm that the fixation and processing of TMAs/
MTBs is similar to clinical specimens.
Strength of evidence was adequate to support TMA usage;
however, there are many variables to be considered and
thorough validation is needed for each marker. Strength of
evidence was inadequate to recommend the routine use of
TMA samples.
Whole sections usually provide more antigen-positive
cells and negative internal control cells within each section
than TMAs/MTBs, but the latter can be designed to contain
multiple previously tested positive and negative tissues. This
allows for comparison of results in multiple tissues tested
with an identical assay protocol and, when properly
selected, a cost-effective validation strategy. Because of the
small size of each tissue sample, however, TMAs and MTBs
may be inappropriate for antigens with limited tissue
expression, heterogeneous distribution, or restricted compartmentalization within tissues. The laboratory director
must use information from the literature and clinical
judgment to determine if TMAs or MTBs are useful for
validating a given assay.
Comparisons of overall concordance between IHC assays
performed on whole sections and TMAs have been done
with at least 9 markers, but primarily with ER, PgR, and
HER2/neu.44–55 Summary estimates of concordance (random
effects model) were computed, but heterogeneity was high
across the studies (I2 . 75; P , .001), and specific sources of
heterogeneity could not be identified. Consequently, concordance is reported as ranges and median values for
specific markers, all in breast cancer tissues.
Median overall concordance estimates for ER, PgR, and
HER2/neu were 95% (range, 84%–99%), 91% (range, 81%–
93%), and 93% (range, 73%–100%), respectively, but
concordance estimates in our review only met or exceeded
the 90% standard in about two-thirds of cases. Comparisons
Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al

of overall concordance for ER and PgR from an earlier
systematic review were 97% and 93%, respectively.52
10: Expert Consensus Opinion.—When a new reagent
lot is placed into clinical service for an existing validated
assay, laboratories should confirm the assay’s performance
with at least 1 known positive case and 1 known negative
case.
The strength of evidence was inadequate to address
conditions requiring assay revalidation and whether revalidation should be the same as initial validation.
Confirmation that assay performance has not changed is
necessary when a new lot of primary antibody or antigen
retrieval or detection reagent is used. For predictive markers,
testing both high and low expressors may be useful.
Including a weakly positive sample is recommended when
there is a specified cut point for positivity (eg, ER) (LAP
checklist COM.30450).3 Including 2 positive cases (1 weak
and 1 strong) should be considered for new reagent lots of
predictive marker antibodies.
11: Expert Consensus Opinion.—Laboratories should
confirm assay performance with at least 2 known positive
and 2 known negative cases when an existing validated
assay has changed in any one of the following ways:
1. Antibody dilution;
2. Antibody vendor (same clone);
3. Incubation or retrieval times (same method).
The strength of evidence was inadequate to address
conditions requiring assay revalidation and whether revalidation should be the same as initial validation.
Confirmation that assay performance has not changed is
necessary when there are minor changes to the assay
method. Public comments received on this recommendation
were more contentious than for most other recommendations. Some argued that these changes fundamentally
change the nature of the assay and therefore should require
full assay revalidation, while others noted that the number
of cases needed to ensure the assay is performing as
expected will vary by antibody. The importance of not
replacing the pathologist’s judgment with arbitrary minimum numbers was also stressed. From the comments
received, the panel concluded that re-assessing assays with
at least 2 positive and 2 negative cases was a reasonable
compromise in ensuring assay performance and provides
the laboratory medical director flexibility to increase the
number as needed.
For predictive markers, laboratories testing both high and
low expressors may be useful. Including weakly positive
samples is recommended when there is a specified cut point
for positivity (eg, ER). Major changes in antibody dilution or
incubation times (as defined by the laboratory) may warrant
testing more than 2 negative and 2 positive cases.
12: Expert Consensus Opinion.—Laboratories should
confirm assay performance by testing a sufficient number of
cases to ensure that assays consistently achieve expected
results when any of the following have changed:
1. Fixative type;
2. Antigen retrieval method (eg, change in pH, different
buffer, different heat platform);
3. Antigen detection system;
4. Tissue processing or testing equipment;
5. Environmental conditions of testing (eg, laboratory
relocation);
Arch Pathol Lab Med—Vol 138, November 2014

6. Laboratory water supply.
The laboratory medical director is responsible for determining the number of positive and negative cases and the
number of predictive and nonpredictive markers to test.
The strength of evidence was inadequate to address
conditions requiring assay revalidation and whether revalidation should be the same as initial validation.
Recommendations 10 and 11 apply to changes in 1
antibody or assay, but this recommendation applies to
changes that affect most or all of a laboratory’s assays. Full
revalidation of every assay in this situation is not practical,
but an assessment is needed to ensure that results of testing
under new conditions are comparable to the results of prior
testing. The laboratory medical director must determine the
extent of this testing based on the nature of the change. A
representative panel of predictive and nonpredictive markers could be selected to assess the impact of the change.
Based on those results, more thorough testing may be
needed, particularly for predictive markers, but if results on
this panel are acceptable, remaining assays could be verified
less rigorously. Markers selected for testing should include
those with different immunolocalizations (ie, nuclear,
membranous, cytoplasmic) as appropriate for the laboratory.
When feasible, comparing the results of staining after the
change with the slides from initial assay validation may help
to determine if the intensity of staining has changed.
Laboratories are required to verify method performance
specifications after an instrument is moved to ensure that
the test system was not affected by the relocation process or
environmental changes (LAP checklist COM.40000).3
13: Expert Consensus Opinion.—Laboratories should
run a full revalidation (equivalent to initial analytic
validation) when the antibody clone is changed for an
existing validated assay.
The strength of evidence was inadequate to address
conditions requiring assay revalidation and whether revalidation should be the same as initial validation.
Although a limited re-assessment of assay performance is
sufficient when there are minor changes in assay conditions
(eg, antibody dilution or incubation time), introduction of a
different antibody clone represents a fundamental change to
the assay and requires complete revalidation. This is because
different antibody clones are raised against different
epitopes on the target protein and their performance
characteristics may significantly vary. This phenomenon is
exemplified by the expression of TTF-1 (thyroid transcription factor 1) in carcinomas other than those of thyroid or
pulmonary origin. Multiple studies56–58 have shown low
levels of expression in metastatic and primary colorectal
carcinomas, carcinomas of gynecologic origin, and glial
neoplasms, using the SPT24 clone. By contrast, the 8G7G3/1
clone is uniformly negative in these tumor types. Similar
data exist for CDX2.59
14: Expert Consensus Opinion.—The laboratory must
document all validations and verifications in compliance
with regulatory and accreditation requirements.
For laboratories subject to US regulations, CLIA specifies
that ‘‘records of the laboratory’s establishment and verification of method performance specifications must be
retained for the period of time the test system is in use by
the laboratory, but not less than 2 years.’’ 1 Laboratories
accredited by CAP must retain records of method performance specifications while the method is in use and for at

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al 1441

least 2 years after discontinuation of the method (LAP
checklist COM.40000).3
In addition to written procedures that describe their
validation and revalidation processes, laboratories should
have documentation, signed by the laboratory medical
director, of the validation, verification, or revalidation
studies and approval of each test for its intended clinical
use(s).
Note on Evidence Analysis for Revalidation Recommendations
(No.10–No.13).—No objective evidence was identified that
addressed requirements for revalidating IHC assays when
there are changes to an existing validated assay (eg, new
reagent lot, change in antibody dilution, changes in
equipment). Refer to the full analysis of key question 6
and key question 7 regarding revalidation in the supplemental digital content for further discussion of the evidence.
CONCLUSION
Physicians and patients rely on accurate diagnostic and
prognostic testing in the clinical laboratory. Established
guidelines for validating and revalidating immunohistochemistry tests used on clinical specimens are important in
ensuring accuracy, reproducibility, and consistency of test
results. The potential harms of false-positive and falsenegative results due to inadequate validation need to be
recognized and addressed. This guideline is intended to help
laboratories improve the accuracy of testing and reassure
clinicians and patients that accepted procedures from
evidence-based and expert consensus–based recommendations are being followed. Direction for re-assessing assays
when changes have occurred or when results are not as
expected is also provided.
We thank the Center advisors Raouf Nakhleh, MD, Sandi
Larsen, MBA, MT(ASCP), and John Olsen, MD, as well as advisory
panel members Richard W. Brown, MD, Richard N. Eisen, MD, and
Hadi Yaziji, MD.
References
1. US Department of Health and Human Services. Clinical laboratory
improvement amendments of 1988: final rule. Fed Regist. 1992;57(40):7001–
7186. Codified at 42 CFR §1405–494.
2. Immunology Branch, Division of Clinical Laboratory Devices, Office of
Device Evaluation. 3.9: Manufacturers’ recommendations for verification of IHC
performance by the user. In: Guidance for Submission of Immunohistochemistry
Applications to the FDA. Center for Devices and Radiological Health, US Food
and Drug Administration; 1998. http://www.fda.gov/downloads/MedicalDevices/
DeviceRegulationandGuidance/GuidanceDocuments/ucm094015.pdf. Accessed
September 1, 2013.
3. College of American Pathologists. CAP Laboratory accreditation checklists.
http://www.cap.org/apps/cap.portal. Accessed September 1, 2013.
4. Hardy LB, Fitzgibbons PL, Goldsmith JD, et al. Immunohistochemistry
validation procedures and practices: a College of American Pathologists survey of
727 laboratories. Arch Pathol Lab Med. 2013;137(1):19–25.
5. Nakhleh RE, Grimm EE, Idowu MO, Souers RJ, Fitzgibbons PL. Laboratory
compliance with the American Society of Clinical Oncology/College of
American Pathologists guidelines for human epidermal growth factor receptor 2
testing: a College of American Pathologists survey of 757 laboratories. Arch
Pathol Lab Med. 2010;134(5):728–734.
6. Teutsch SM, Bradley LA, Palomaki GE, et al. The Evaluation of Genomic
Applications in Practice and Prevention (EGAPP) Initiative: methods of the
EGAPP Working Group. Genet Med. 2009;11(1):3–14.
7. US Department of Health and Human Services. Medicare, Medicaid and
CLIA programs: regulations implementing the Clinical Laboratory Improvement
Amendments of 1988: final rule. Fed Regist. 1992;57:7002–7186.
8. Clinical Laboratory Standards Institute. Quality assurance for design
control and implementation of immunohistochemistry assays: approved guideline, second edition. In: CLSI Document I/LA28-A2. Wayne, PA: Clinical and
Laboratory Standards Institute; 2011.
9. Dowsett M, Hanna WM, Kockx M, et al. Standardization of HER2 testing:
results of an international proficiency-testing ring study. Mod Pathol. 2007;20(5):
584–591.

1442 Arch Pathol Lab Med—Vol 138, November 2014

10. Wolff AC, Hammond ME, Schwartz JN, et al. American Society of Clinical
Oncology/College of American Pathologists guideline recommendations for
human epidermal growth factor receptor 2 testing in breast cancer. Arch Pathol
Lab Med. 2007;131(1):18–43.
11. Allred DC, Carlson RW, Berry DA, et al. NCCN Task Force report: estrogen
receptor and progesterone receptor testing in breast cancer by immunohistochemistry. J Natl Compr Canc Netw [quiz in J Natl Compr Canc Netw. 2009;
7(suppl 6):S22–S23]. 2009;7(suppl 6):S1–S21.
12. Baba K, Dyrhol-Riise AM, Sviland L, et al. Rapid and specific diagnosis of
tuberculous pleuritis with immunohistochemistry by detecting Mycobacterium
tuberculosis complex specific antigen MPT64 in patients from a HIV endemic
area. Appl Immunohistochem Mol Morphol. 2008;16(6):554–561.
13. Boers JE, Meeuwissen H, Methorst N. HER2 status in gastro-oesophageal
adenocarcinomas assessed by two rabbit monoclonal antibodies (SP3 and 4B5)
and two in situ hybridization methods (FISH and SISH). Histopathology. 2011;
58(3):383–394.
14. Mayr D, Heim S, Werhan C, Zeindl-Eberhart E, Kirchner T. Comprehensive
immunohistochemical analysis of Her-2/neu oncoprotein overexpression in
breast cancer: HercepTest (Dako) for manual testing and Her-2/neuTest 4B5
(Ventana) for Ventana BenchMark automatic staining system with correlation to
results of fluorescence in situ hybridization (FISH). Virchows Arch. 2009;454(3):
241–248.
15. Moelans CB, Kibbelaar RE, van den Heuvel MC, Castigliego D, de Weger
RA, van Diest PJ. Validation of a fully automated HER2 staining kit in breast
cancer. Cell Oncol. 2010;32(1–2):149–155.
16. O’Grady A, Allen D, Happerfield L, et al. An immunohistochemical and
fluorescence in situ hybridization-based comparison between the Oracle HER2
Bond Immunohistochemical System, Dako HercepTest, and Vysis PathVysion
HER2 FISH using both commercially validated and modified ASCO/CAP and
United Kingdom HER2 IHC scoring guidelines. Appl Immunohistochem Mol
Morphol. 2010;18(6):489–493.
17. Phillips T, Murray G, Wakamiya K, et al. Development of standard estrogen
and progesterone receptor immunohistochemical assays for selection of patients
for antihormonal therapy. Appl Immunohistochem Mol Morphol. 2007;15(3):
325–331.
18. Rhodes A, Jasani B, Anderson E, Dodson AR, Balaton AJ. Evaluation of
HER-2/neu immunohistochemical assay sensitivity and scoring on formalin-fixed
and paraffin-processed cell lines and breast tumors: a comparative study
involving results from laboratories in 21 countries. Am J Clin Pathol. 2002;
118(3):408–417.
19. van der Vegt B, de Bock GH, Bart J, Zwartjes NG, Wesseling J. Validation
of the 4B5 rabbit monoclonal antibody in determining Her2/neu status in breast
cancer. Mod Pathol. 2009;22(7):879–886.
20. Dorfman DM, Bui MM, Tubbs RR, et al. The CD117 immunohistochemistry tissue microarray survey for quality assurance and interlaboratory
comparison: a College of American Pathologists Cell Markers Committee study.
Arch Pathol Lab Med. 2006;130(6):779–782.
21. Jordan RC, Lingen MW, Perez-Ordonez B, et al. Validation of methods for
oropharyngeal cancer HPV status determination in US cooperative group trials.
Am J Surg Pathol. 2012;36(7):945–954.
22. Lotan TL, Gurel B, Sutcliffe S, et al. PTEN protein loss by immunostaining:
analytic validation and prognostic indicator for a high risk surgical cohort of
prostate cancer patients. Clin Cancer Res. 2011;17(20):6563–6573.
23. Lehmann-Che J, Amira-Bouhidel F, Turpin E, et al. Immunohistochemical
and molecular analyses of HER2 status in breast cancers are highly concordant
and complementary approaches. Br J Cancer. 2011;104(11):1739–1746.
24. Fitzgibbons PL, Murphy DA, Hammond ME, Allred DC, Valenstein PN.
Recommendations for validating estrogen and progesterone receptor immunohistochemistry assays. Arch Pathol Lab Med. 2010;134(6):930–935.
25. Hammond ME, Hayes DF, Dowsett M, et al. American Society of Clinical
Oncology/College of American Pathologists guideline recommendations for
immunohistochemical testing of estrogen and progesterone receptors in breast
cancer. Arch Pathol Lab Med. 2010;134(6):907–922.
26. Torlakovic EE, Naresh K, Kremer M, van der Walt J, Hyjek E, Porwit A. Call
for a European programme in external quality assurance for bone marrow
immunohistochemistry; report of a European Bone Marrow Working Group pilot
study. J Clin Pathol. 2009;62(6):547–551.
27. US Department of Health and Human Services. Medical devices:
classification/reclassification of immunochemistry reagents and kits. Fed Regist.
1998;63(106):30132–30142. Codified at 21 CFR §864. Doc. No. 94P–0341.
28. Wolff AC, Hammond EH, Hicks DG, et al. Recommendations for human
epidermal growth factor receptor 2 testing in breast cancer: American Society of
Clinical Oncology/College of American Pathologists clinical practice guideline
update. Arch Pathol Lab Med. 2014;138:241–256. doi:10.5858/arpa.
2013-0953-SA.
29. Sun F, Bruening W, Erinoff E, Schoelles KM. Addressing Challenges in
Genetic Test Evaluation: Evaluation Frameworks and Assessment of Analytic
Validity. Methods research report (prepared by the ECRI Institute Evience-Based
Practice Center under contract No. HHSA 290-20007-10063-I). Rockville, MD:
Agency for Healthcare Research and Quality; June 2011. AHRQ Publication No.
11-EHC048-EF.
30. Ferguson J, Chamberlain P, Cramer HM, Wu HH. ER, PR, and Her2
immunocytochemistry on cell-transferred cytologic smears of primary and

Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al

metastatic breast carcinomas: a comparison study with formalin-fixed cell blocks
and surgical biopsies. Diagn Cytopathol. 2013;41(7):575–581.
31. Gong Y, Symmans WF, Krishnamurthy S, Patel S, Sneige N. Optimal
fixation conditions for immunocytochemical analysis of estrogen receptor in
cytologic specimens of breast carcinoma. Cancer. 2004;102(1):34–40.
32. Hanley KZ, Birdsong GG, Cohen C, Siddiqui MT. Immunohistochemical
detection of estrogen receptor, progesterone receptor, and human epidermal
growth factor receptor 2 expression in breast carcinomas: comparison on cell
block, needle-core, and tissue block preparations. Cancer. 2009;117(4):279–288.
33. Kumar SK, Gupta N, Rajwanshi A, Joshi K, Singh G. Immunochemistry for
oestrogen receptor, progesterone receptor and HER2 on cell blocks in primary
breast carcinoma. Cytopathology. 2012;23(3):181–186.
34. Nishimura R, Aogi K, Yamamoto T, et al. Usefulness of liquid-based
cytology in hormone receptor analysis of breast cancer specimens. Virchows
Arch. 2011;458(2):153–158.
35. Pegolo E, Machin P, Riosa F, Bassini A, Deroma L, Di Loreto C. Hormone
receptor and human epidermal growth factor receptor 2 status evaluation on
ThinPrep specimens from breast carcinoma: correlation with histologic sections
determination. Cancer Cytopathol. 2012;120(3):196–205.
36. Shabaik A, Lin G, Peterson M, et al. Reliability of Her2/neu, estrogen
receptor, and progesterone receptor testing by immunohistochemistry on cell
block of FNA and serous effusions from patients with primary and metastatic
breast carcinoma. Diagn Cytopathol. 2011;39(5):328–332.
37. Adegboyega PA, Gokhale S. Effect of decalcification on the immunohistochemical expression of ABH blood group isoantigens. Appl Immunohistochem
Mol Morphol. 2003;11(2):194–197.
38. Arber JM, Arber DA, Jenkins KA, Battifora H. Effect of decalcification and
fixation in paraffin-section immunohistochemistry. Appl Immunohistochem.
1996;4(4):241–248.
39. Bussolati G, Leonardo E. Technical pitfalls potentially affecting diagnoses
in immunohistochemistry. J Clin Pathol. 2008;61(11):1184–1192.
40. Fend F, Tzankov A, Bink K, et al. Modern techniques for the diagnostic
evaluation of the trephine bone marrow biopsy: methodological aspects and
applications. Prog Histochem Cytochem. 2008;42(4):203–252.
41. Hsi ED. A practical approach for evaluating new antibodies in the clinical
immunohistochemistry laboratory. Arch Pathol Lab Med. 2001;125(2):289–294.
42. Wittenburg G, Volkel C, Mai R, Lauer G. Immunohistochemical
comparison of differentiation markers on paraffin and plastic embedded human
bone samples. J Physiol Pharmacol. 2009;60(suppl 8):43–49.
43. Zustin J, Boddin K, Tsourlakis MC, et al. HER-2/neu analysis in breast
cancer bone metastases. J Clin Pathol. 2009;62(6):542–546.
44. Batistatou A, Televantou D, Bobos M, et al. Evaluation of current
prognostic and predictive markers in breast cancer: a validation study of tissue
microarrays. Anticancer Res. 2013;33(5):2139–2145.
45. Drev P, Grazio SF, Bracko M. Tissue microarrays for routine diagnostic
assessment of HER2 status in breast carcinoma. Appl Immunohistochem Mol
Morphol. 2008;16(2):179–184.

46. Fons G, Hasibuan SM, van der Velden J, ten Kate FJ. Validation of tissue
microarray technology in endometrioid cancer of the endometrium. J Clin Pathol.
2007;60(5):500–503.
47. Graham AD, Faratian D, Rae F, Thomas JSJ. Tissue microarray technology
in the routine assessment of HER-2 status in invasive breast cancer: a prospective
study of the use of immunohistochemistry and fluorescence in situ hybridization.
Histopathology. 2008;52:847–855.
48. Gulbahce HE, Gamez R, Dvorak L, Forster C, Varghese L. Concordance
between tissue microarray and whole-section estrogen receptor expression and
intratumoral heterogeneity. Appl Immunohistochem Mol Morphol. 2012;20:340–
343.
49. Henriksen KL, Rasmussen BB, Lykkesfeldt AE, Moller S, Ejlertsen B,
Mouridsen HT. Semi-quantitative scoring of potentially predictive markers for
endocrine treatment of breast cancer: a comparison between whole sections and
tissue microarrays. J Clin Pathol. 2007;60(4):397–404.
50. Jones S, Prasad ML. Comparative evaluation of high-throughput small-core
(0.6-mm) and large-core (2-mm) thyroid tissue microarray: is larger better? Arch
Pathol Lab Med. 2012;136(2):199–203.
51. Kwon MJ, Nam ES, Cho SJ, et al. Comparison of tissue microarray and full
section in immunohistochemistry of gastrointestinal stromal tumors. Pathol Int.
2009;59(12):851–856.
52. Nofech-Mozes S, Vella ET, Dhesy-Thind S, et al. Systematic review on
hormone receptor testing in breast cancer. Appl Immunohistochem Mol Morphol.
2012;20(3):214–263.
53. Soiland H, Skaland I, van Diermen B, et al. Androgen receptor
determination in breast cancer: a comparison of the dextran-coated charcoal
method and quantitative immunohistochemical analysis. Appl Immunohistochem Mol Morphol. 2008;16(4):362–370.
54. Thomson TA, Zhou C, Chu C, Knight B. Tissue microarray for routine
analysis of breast biomarkers in the clinical laboratory. Am J Clin Pathol. 2009;
132(6):899–905.
55. Warnberg F, Amini RM, Goldman M, Jirstrom K. Quality aspects of the
tissue microarray technique in a population-based cohort with ductal carcinoma
in situ of the breast. Histopathology. 2008;53(6):642–649.
56. Comperat E, Zhang F, Perrotin C, et al. Variable sensitivity and specificity of
TTF-1 antibodies in lung metastatic adenocarcinoma of colorectal origin. Mod
Pathol. 2005;18(10):1371–1376.
57. Kristensen MH, Nielsen S, Vyberg M. Thyroid transcription factor-1 in
primary CNS tumors. Appl Immunohistochem Mol Morphol. 2011;19(5):437–
443.
58. Zhang PJ, Gao HG, Pasha TL, Litzky L, Livolsi VA. TTF-1 expression in
ovarian and uterine epithelial neoplasia and its potential significance, an
immunohistochemical assessment with multiple monoclonal antibodies and
different secondary detection systems. Int J Gynecol Pathol. 2009;28(1):10–18.
59. Borrisholt M, Nielsen S, Vyberg M. Demonstration of CDX2 is highly
antibody dependant. Appl Immunohistochem Mol Morphol. 2013;21(1):64–72.

APPENDIX
Disclosed Interests and Activities June 2010 to September 2013
Name

Interest/Activity Type

Linda A. Bradley, PhD

Consultancy

Regan S. Fulton, MD, PhD

Board or advisory board
Consultancy
Grants
Patent received or pending

Jeffrey D. Goldsmith, MD
Thomas S. Haas, DO

Ownership or beneficial
ownership of stock
Lecture fee paid by entity
Expert witness
Consultancy
Board or advisory board
Lecture fee paid by entity

Patti A. Loykasek,
HTL(ASCP), QIHC
Paul E. Swanson, MD

Board or advisory board
Consultancy
Consultancy
Lecture fee paid by entity

Arch Pathol Lab Med—Vol 138, November 2014

Entity
Blue Cross Blue Shield Association
American College of Medical Genetics Foundation
Center for Medical Technology Policy
Gerson Lehrman Group
National Institutes of Health-Small Business Innovation Research
Grant (application pending)
United States Patent and Trademark Office Application
(application pending)
Array Science, LLC
United States and Canadian Academy of Pathology
Various
Biocare Medical, Concord, California
Newcomer Histology Supply, Middleton, Wisconsin
Biocare Medical, Concord, California
Leica Microsystems, Buffalo Grove, Illinois
Leica Microsystems, Buffalo Grove, Illinois
Biocare Medical, Concord, California
National Society for Histotechnology
Clover Park College
PhenoPath Laboratory
American Society of Clinical Pathology
College of American Pathologists
Analytic Validation of Immunohistochemical Assays—Fitzgibbons et al 1443

Supplemental Digital Content* | Methodology |
February 2015

Principles of Analytic Validation
for Immunohistochemical Assays
Guideline from the Pathology and Laboratory
Quality Center
Corresponding Author:
Patrick L. Fitzgibbons, MD
Authors:
Linda A. Bradley, PhD
Lisa A. Fatheree, SCT(ASCP)
Anthony T. Smith, ML
Archives Early Online Release: Principles of Analytic Validation of Immunohistochemical
Assays
* The Supplemental Digital Content was not copyedited by Archives of Pathology and
Laboratory Medicine.

College of American Pathologists | 325 Waukegan Rd. | Northfield, IL 60093 | 800-323-4040 | cap.org

325 Waukegan Rd. |Northfield, IL 60093
t: 800-323-4040 | cap.org
Version no.

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

METHODS USED TO PRODUCE THE GUIDELINE
Panel Composition
The College of American Pathologists (CAP) Pathology and Laboratory Quality Center
(the Center) convened an expert and advisory panel consisting of pathologists and
histotechnologists with expertise in implementing and performing immunohistochemical
(IHC) assays. CAP approved the appointment of the project chair (PLF) and panel
members. These panel members served as the Technical Expert Panel (TEP) for the
systematic evidence review.
Conflict of Interest (COI) Policy
Prior to acceptance on the expert or advisory panel, potential members completed the CAP conflict of
interest (COI) disclosure process, whose policy and form (in effect April 2010) requires disclosure of
material financial interest in, or potential for benefit of significant value from, the guideline’s development
or its recommendations 12 months prior through the time of publication. The potential members
completed the COI disclosure form, listing any relationship that could be interpreted as constituting an
actual, potential, or apparent conflict. The CAP Center uses the following criteria:
Nominees who have the following conflicts may be excused from the panel:
a. Stock or equity interest in a commercial entity that would likely be affected by the guideline or white
paper
b. Royalties or licensing fees from products that would likely be affected by the guideline or
white paper
c. Employee of a commercial entity that would likely be affected by the guideline or white paper
Nominees who have the following potentially manageable direct conflicts may be appointed to the panel:
a. Patents for products covered by the guideline or white paper
b. Member of an advisory board of a commercial entity that would be affected by the guideline or white
paper
c. Payments to cover costs of clinical trials, including travel expenses associated directly with the trial
d. Reimbursement from commercial entity for travel to scientific or educational meetings
Everyone was required to disclose conflicts prior to beginning and continuously throughout the
project’s timeline. One expert panel member (RSF) was recused from discussion and voting on the
recommendation pertaining to tissue microarrays (TMAs). One expert panel member (TSH) was
recused from voting on the recommendations pertaining to potential increased antibody usage.
Expert panel members’ disclosed conflicts are listed in the appendix of the manuscript. The CAP
provided funding for the administration of the project; no industry funds were used in the development
of the guideline. All panel members volunteered their time and were not compensated for their
involvement.
CAP Expert Panel Literature Review and Analysis
The expert panel met 28 times through teleconference webinars from June 2010 through September
2013. Additional work was completed via electronic mail and the panel met in person May 11-12, 2013 to
review evidence to date and draft recommendations.
All expert panelists participated in the systematic evidence review (SER) level of title-abstract
and full-text review. Chair PLF and panelists PES and RSF performed the audit of data extraction.
Panelist RSF was recused from performing any audit on articles pertaining to TMAs. All articles were
available as discussion or background references. All members of the expert panel participated in
developing draft recommendations, reviewing open comment feedback, finalizing and approving

© 2015 College of American Pathologists. All rights reserved.

PAGE 2

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

recommendations and writing/editing of the manuscript except as noted for RSF and TSH.
Peer Review
An open comment period was held from July 8 through July 29, 2013. Eighteen draft recommendations
and five methodology questions were posted online on the CAP Web site www.cap.org. An
announcement was sent to the following societies deemed to have interest:
American Society for Clinical Pathology (ASCP) Association for Molecular Pathology (AMP) Society for
Immunohistochemistry
National Society for Histotechnology (NSH) American Society of Cytopathology (ASC)
Association of Directors of Anatomic and Surgical Pathology (ADASP) Association of Pathology Chairs
(APC)
Clinical Laboratory Management Association (CLMA)
US Food and Drug Administration (FDA)
Centers for Medicare and Medicaid Services (CMS) Canadian Association of Pathologists (CAP-APC)
United States & Canadian Academy of Pathology (USCAP)
United Kingdom National External Quality Assessment Service (UK NEQAS) Nordic IHC Quality Control
(NordiQC)
Canadian IHC Quality Control (CIQC)
The website received 1,071 comments in total (Agree and Disagree responses were also captured).
Twelve of 18 recommendations achieved more than 80% agreement; only 2 had less than 70%
agreement. Each expert panel member was assigned 1-2 draft recommendations for which to review all
comments received and provide an overall summary to the rest of the panel. Following panel discussion,
a secondary internal review by the CAP Surgical Pathology and Immunohistochemistry Resource
Committees and the final quality of evidence assessment, the panel members determined whether to
maintain the original draft recommendation as is, revise it with minor language change, or consider it as a
major recommendation change. Three draft recommendations were maintained with the original
language; five were modified with minor changes for clarification and/or further explanation within the
manuscript and six were considered extremely discordant with major revisions made accordingly for a
total of 14 final recommendations. Resolution of all changes was obtained by majority consensus of the
panel using nominal group technique (rounds of email discussion and multiple edited recommendations)
amongst the panel members. The final recommendations were approved by the expert panel with a
formal vote (minus RSF on the recommendation regarding TMAs and TSH on potential increased
antibody usage). The panel considered laboratory redundancy, efficiency and feasibility throughout the
whole process. Formal cost analysis or cost effectiveness was not performed.

An independent review panel (IRP) was assembled to review the guideline and recommend approval to
the CAP. The IRP was masked to the expert panel and vetted through the COI process.

Assessing the Strength of Recommendations
The central question that the panel addressed in developing the guideline was “What is needed for initial
analytic assay validation before placing any immunohistochemical test into clinical service, and what are
the revalidation requirements?”
Development of recommendations requires that the panel review the identified evidence and make a
series of key judgments:
1) What are the significant findings related to each KQ or outcome? Determine which components of
analytic validation (e.g., overall and positive/negative concordance from comparisons, precision,
robustness) have a regulatory requirement and/or evidence that support a specific action and/or
method for the validation process.

© 2015 College of American Pathologists. All rights reserved.

PAGE 3

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

2) What is the overall strength of evidence supporting each KQ or outcome? Strength of evidence is
graded as Convincing, Adequate or Inadequate, based on four published criteria (SER, Figure 2).
Strength of evidence is a key element in determining the strength of a recommendation.
3) What is the strength of each recommendation? There are many methods for determining the strength
of a recommendation based on the strength of evidence and the magnitude of net benefit or harm.
However, such methods have rarely (if ever) been applied to analytic validity, and certainly not to
recommendations on component parts of the analytic validation process. Therefore, the method for
determining strength of recommendation has been modified for this application (Table 1), and is
based on the strength of evidence and the likelihood that further studies will change the conclusions.
Recommendations not supported by evidence (i.e., evidence was missing or Insufficient to permit a
conclusion to be reached) may be made based on consensus expert opinion. Another potential
consideration is the likelihood that additional studies need to fill gaps in knowledge will be conducted.
4) What is the net balance of benefits and harms? The consideration of net balance of benefits and
harms will focus on the core recommendation to perform analytic validation before offering a test in
practice.

© 2015 College of American Pathologists. All rights reserved.

PAGE 4

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Table 1: Grades for Strength of Recommendations*

Designation

Recommendation

Rationale

Strong
Recommendation

Recommend For or Against a particular
analytic validation practice (Can include
must or should)

Strength of evidence is
Convincing based on
consistent, generalizable, good
quality evidence; further
studies are unlikely to change
the conclusions

Recommendation

Recommend For or Against a particular
analytic validation practice (Can include
should or may)

Strength of evidence is
Adequate based on limitations
in the quality of evidence;
further studies may change the
conclusions

Expert Consensus
Opinion

Recommend For or Against a particular
analytic validation practice
(Can include should or may)

Important validation element
to address but strength of
evidence is Inadequate; gaps in
knowledge may require further
studies

*Modified by the CAP Pathology and Laboratory Quality Center

Dissemination Plans
CAP will host an IHC Validation Resource web page which will include a link to manuscript and
supplemental digital content; summary of recommendations, teaching PowerPoint, frequently asked
question (FAQ) document and a free archived webinar. The guideline will be promoted and presented at
various professional society meetings including the College of American Pathologists, the United States
and Canadian Academy of Pathology (USCAP), the National Society for Histotechnologists (NSH), the
American Society of Clinical Pathology (ASCP) and the American Society of Cytopathology (ASC).
SYSTEMATIC EVIDENCE REVIEW (SER)
The objectives of the SER were to investigate the optimal performance characteristics of IHC tests and
determine how they can be achieved and measured. If of sufficient quality, findings from this review
could provide an evidence base to support development of the clinical guideline. The scope of the SER
and the key questions (KQs) were established by the TEP in consultation with a methodologist.
Search and Selection
®
Electronic searches of the English language published literature in Ovid MEDLINE , U.S.
National Library of Medicine PubMed, and Elsevier Scopus databases were initially conducted
for the time period January 2004 to May 2012; an update was conducted through May 2013. The search
utilized the following MeSH terms and keywords:

© 2015 College of American Pathologists. All rights reserved.

PAGE 5

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

MeSH Terms
Immunohistochemistry, Immunoenzyme
Techniques,
Validation Studies as Topic, Reproducibility of
Results,
Sensitivity and Specificity, Validation
Studies,
Evaluation Studies as Topic, Observer
Variation,
Clinical Laboratory Techniques,
Laboratories,
Hospital, Pathology, “Tumor Markers,
Biological”,
Ki-67 Antigen, Cyclin-Dependent Kinase
Inhibitor p16, “Receptor, erbB-2”,
“Receptors, Progesterone”,
“Receptors, Estrogen”, Vimentin

Keywords
Immunohistochemistry, IHC,
Immunocytochemistry,
Immunoperoxidase, Antigen
retrieval, Antigen detection,
Validation, Standardization, Inter-run
variance, Inter-operator variance,
Controls, Analytic variance,
Signature molecules, Molecular tests and
assays, Cytokeratin, CK 5/6, CK7, CK20,
CD5, CD10, CD20, CD45, CD99, CD117,
p63, Cyclin D1, bcl1, bcl2, actin, desim,
chromogranin, cadherin, estrogen recepto
progesterone receptor, HER2, erbB2, S10
TTF-1, vimentin, MIB-1, PTEN, Ki-67.

Bibliographies of included articles were hand searched, and additional information was sought
through targeted grey literature electronic searches (e.g., Google) and review of laboratory
compliance and guidance websites (e.g., Clinical and Laboratory Standards Institute, US Food and
Drug Administration (FDA), National Guidelines Clearinghouse, Wiley Cochrane Library).
Two reviewers were used at all levels of review (e.g., title/abstract, full article) and for data/information
extraction. Conflicts were resolved by discussion or referred to the panel Chair for a decision. When
article abstracts or document summaries were not available or a conflict was not resolved, full articles
were reviewed.
Selection at all levels was based on predetermined inclusion/exclusion
criteria. Included were:

•

English-language articles/documents that addressed IHC and provided data or information relevant
to one or more KQs;
• Study designs included validation, method comparison, cohort, or case-controlled studies, clinical
trials, and systematic reviews, as well as qualitative information from consensus guidelines,
regulatory documents or US and international proficiency testing reports; and
• Articles/documents focused on the clinical use of IHC for identification of non-FDA approved
predictive and non-predictive markers and analytic variables.
Not included were:

•

Non-English-language article/document or an English-language abstract or summary without a full
article/document available in English;

•
•
•

Article/document involves IHC but does not address any KQ;
Publications with high risk of bias, such as editorials, letters, commentary, invited opinion; and
Article/documents focused on non-human research, non-tissue IHC (immunoassays, serologic
studies), assay optimization or quality control/quality assurance, pre- or post- analytic variables, or
clinical validation.

© 2015 College of American Pathologists. All rights reserved.

PAGE 6

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Outcomes of Interest
Outcomes of interest for assessing analytic validity include analytic sensitivity (detection rate), analytic
specificity (1-false positive rate), reliability (e.g., repeatability of test results) and assay robustness (e.g.,
resistance to small changes in pre-analytic or analytic variables). Computing estimates of analytic
sensitivity and specificity requires a “gold standard” or well-characterized referent assay (or set of
referent specimens with antigen status characterized by previous testing) against which to compare the
1-3
index, or new, IHC test.
Among IHC assays, such “gold standard” referent assays are likely to be the exception rather than the
1
rule. Even HER2 IHC and FISH assays have no “gold standard” at present, as no assay currently
3
available is perfectly accurate in identifying overexpression of this protein.
Consequently, the metric for IHC validation results is most often overall concordance between the results
of the new and referent assay(s) for a specific set of validation tissues, or between the results of the new
test with previous results for a characterized set of validation tissues. Estimates of positive and negative
concordance may also be computed.
We sought quantitative data from primary studies (e.g., validation studies, method comparisons), and
systematic reviews of such studies, on concordance, repeatability, reproducibility, and robustness factors
(e.g., sample types, fixation). In addition, we sought qualitative information relevant to IHC validation or
validation standards from regulatory materials, existing evidence- informed and/or consensus guidelines,
and referenced review articles from credible sources.
Data Extraction and Management
The data elements from an included article/document were extracted by one reviewer into standard data
formats and tables developed using systematic review database software (DistillerSR, Evidence
Partners Inc., Ottawa, Canada); a second reviewer confirmed accuracy and completeness. In all cases,
the methodologist acted as either the primary or secondary reviewer. Any discrepancies in data
extraction were resolved by discussion with the Methodologist. A bibliographic database was

established in EndNote (Thomson Reuters, Carlsbad, CA) to track all literature identified
and reviewed during the study.
Environmental Scan
In 2009, CAP recommended strengthening the oversight of laboratory developed tests (LDTs). CAP’s
proposed changes would incorporate oversight of claims of clinical validity, and specify scientific and
regulatory standards to be applied to all LDTs. Risk would be determined based on claims made,
potential risk to patients, and the extent to which a test’s results could be used in the determination of
diagnosis or treatment. The FDA convened a public meeting in July 2010 to discuss issues and
stakeholder concerns surrounding LDT oversight. As of submission date of
4,5
the manuscript (October 2013), no further information is available.
Quality Assessment
Grading the quality of individual studies was performed based on study design-specific criteria by the
methodology consultant, with input as needed from the TEP. Quality assessments were summarized for
each study and recorded in the database. The aim of analytic validation is to determine a test’s ability to
accurately and reliably detect the antigen or marker of interest in
2,6
specimens consistent with those to be tested in clinical practice.”
Analytic validity studies have a
different design compared to studies of diagnostic accuracy or therapeutic interventions. For this reason,
the criteria needed to assess the quality of analytic validity studies are different.
Quality in this context is considered to be essentially equivalent to internal validity, and is assessed
2
based on study design, execution, analyses and reporting. Discordant decisions were resolved through
discussion or third-party adjudication.

© 2015 College of American Pathologists. All rights reserved.

PAGE 7

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

The hierarchy of data sources and criteria for grading quantitative studies were based on published
2,7
methods (Appendix, Table 1).
Studies were rated: Good (no features that suggest flaws or bias); Fair
(susceptible to some bias, but flaws not sufficient to invalidate results); or
Poor (significant flaws suggesting bias of various types that might invalidate results)(Appendix, Table 2).
8-11
Qualitative articles/documents were also assessed using published methods.
The quality criteria
included credibility (e.g., sources, level of review, potential for bias),
transferability (i.e., potential for broader application) dependability (e.g., findings stable over time or
and/or different methods) and confirmability (i.e., findings consistent and/or verified). Documents were
rated: Good (e.g., published/peer-reviewed, from an informed consensus process or
professional/advisory committee report); Fair (e.g., from credible source with unknown level of peer
review, report/guideline from known expert(s) with no observed bias, otherwise Good documents with a
flaw or bias); or Poor (e.g., document lacking information on source, peer review, potential bias,
referencing, or updating; or having multiple flaws or possible biases).
2
The strength of evidence for individual KQs or outcomes was assessed using published criteria. The
criteria included the quality and execution of studies, the quantity of data (number and size of studies)
2
and the consistency and generalizability of the evidence across studies. Strength of evidence was
graded Convincing, Adequate or Inadequate (Table 2).
Table 2. Grades for Strength of Evidence
Convincing
a

Two or more Level 1 or 2 studies (study design and execution) that had an appropriate number and
b

c

d

distribution of challenges and reported consistent and generalizable results.
One Level 1 or 2 study that had an appropriate number and distribution of challenges and reported
generalizable results.
Adequate
Two or more Level 1 or 2 studies that lacked the appropriate number and distribution of challenges
OR were consistent but not generalizable.
Inadequate
Combinations of Level 1 or 2 studies that show unexplained inconsistencies OR one or more lower
quality studies (Level 3 or 4) OR expert opinion.
a
Table 1 in the Appendix provides the hierarchy of data sources for analytic validation that define Level 1 through Level 4.
b Based on number of possible response categories and required confidence in results.
c Consistency can be assessed formally by testing for homogeneity, or, when data are limited, less formally using central estimates and
range of values.
d Generalizability is the extension of findings and conclusions from one study to other settings. Reprinted by permission from Macmillan
Publishers Ltd: Genetics in Medicine2, copyright 2009

Data Analysis
Both quantitative and qualitative methods could be used. Qualitative analysis focuses on identification of
themes and patterns within and among non-study related articles and documents, descriptive narrative,
10,12,13

Quantitative analyses were involved collection of data from
content and/or logical analysis.
validation or method comparison studies into simple data tables or
contingency tables (2x2 or 3x3).

© 2015 College of American Pathologists. All rights reserved.

PAGE 8

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Estimates of overall and positive and negative concordance with 95% confidence intervals
(CI) can be computed from the contingency tables (Figure 1, Table 3). Overall concordance,
also known as percent agreement, is a measure used for comparison of the results of the
new test to
14

those obtained using a non-gold standard referent assay (or an “imperfect standard”). This
measure is based on the major diagonal (Figure 1, upper left cell to lower right cell). The Kappa
statistic can be used to test if the major diagonal counts are significantly larger than those
expected by chance alone (BMDP Statistical Software, Los Angeles, CA). Negative
concordance
14

measures the proportion of “negative” samples in which the index test is negative. Positive
concordance measures the proportion of “positive” samples in which the index test is
14

positive. These last two measures are analogous to analytic sensitivity and specificity, but
are used in situations in which the “true” status (marker negative or positive) is not known.
Discordance is a measure based on the “off” diagonal (Figure 1, upper right to lower left) of
the contingency table that focuses on discrepancies between results from different assays. In
data sets of sufficient size, McNemar’s test may be used to determine whether a discordant
result between the two tests in one direction (e.g., referent negative and new test positive) is
equal to a discordant result in the other direction. A significant value (p < 0.05) indicates a lack
of symmetry and a potential bias between the two assays. McNemar’s test can be performed
on data from a 2x2 table (GraphPad Quick Calc,
http://www.graphpad.com/quickcalcs/McNemar1.cfm) or extended to three dimensions for a
3x3 table (BMDP Statistical Software).
Assay robustness may be tested by comparison of results between a “standard” IHC
component (e.g., fixative 10% neutral buffer formalin) and an alternative (e.g., other fixative)
and is generally measured by concordance with a 95% CI. For all comparisons, summary
estimates of concordance (random effects model) may be possible, with assessment of
heterogeneity and potential for publication bias (Comprehensive Meta-Analysis, Biostat Inc).
Precision, or
repeatability, is a measure of result agreement between specimens tested on different
14,15

days.
Reproducibility is a measure of agreement between a set of test results interpreted by different
14,15

Both are generally
pathologists (i.e., inter-rater) or performed in different laboratories.
reported as percent concordance with a 95% CI and/or Kappa statistic.

© 2015 College of American Pathologists. All rights reserved.

PAGE 9

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Figure 1. Comparison of a new or index IHC to a validated IHC or alternative method in a
2x2 contingency table

Referent
IHC Positive

Referent
IHC Negative

TP

FP

FN

TN

Total index
negative

Total positive

Total negative

Total N

Index

Total index

IHC positive
Index
IHC negative

positive

Abbreviation: IHC=Immunohistochemical; TP=True Positive; TN= True Negative; FP=False Positive; FN=
False Negative; N= Number

Results
Among the 1,463 citations identified by electronic and hand searches, 126 were selected for
inclusion. These included 122 published peer-reviewed articles, 2 book chapters and 2 grey
literature documents (Appendix – Figure 1). Among the extracted documents, 43
articles/documents did not meet minimum quality standards, presented incomplete data or data
that were not in useable formats, and included only information based on expert opinion. These
articles were not included in analyses or narrative summaries. Three general categories of
articles/documents were identified.
The first category was published validation and/or method comparison studies on clinical
IHC assays. The second category included published, web-based and proprietary
guidelines addressing IHC standardization or best practices in general, or guidance on
validation and standardization of specific IHC assays (e.g., HER2, ER, PgR). These
guidelines were largely qualitative reports based on varying combinations and levels of
evidence review and expert opinion. The third category consisted of reported studies on
inter-laboratory comparisons, external proficiency testing for common IHC assays or
laboratory surveys reporting current laboratory validation practices.

Table 3. Measures of Analytic Validity

Computation from
2x2 Table

Computation from 3x3 Table

Overall concordance
or percent agreement

TP + TN / TP + FP + FN +
TN

Sum of concordant cells (major
diagonal) / Total N1

Overall discordance

FP + FN / TP + FP + FN +
TN

Sum of 5 discordant cells / Total N

Positive and negative
concordance or
percent agreement

Positive = TP / (TP + FN)
Negative = TN / (TN + FP)

Not applicable unless 3x3 table can
be collapsed2 to 2x2 or all 2+
samples are excluded

Measure

© 2015 College of American Pathologists. All rights reserved.

PAGE 10

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

1

Some studies using tests that report equivocal results (e.g., 3+ positive, 2+ equivocal and 0-1+ negative) include all results as
relevant to understanding the relationship between the two tests. However, a major guideline notes that equivocal cases are not
expected to be 95% concordant, and cells with discordant results may be omitted. 2 Collapsed by authors’ classification of
equivocals as positive or negative.
Abbreviation: TP=True Positive; TN= True Negative; FP=False Positive; FN= False Negative; N= Number

KQ 1: When and how should IHC validation assess analytic sensitivity, analytic specificity and
precision (e.g., inter-run, inter-operator)?
Note: Such means include (but are not necessarily limited to):
• Correlating the new test’s results with the morphology and expected results;
• Comparing the new test’s results with the results of prior testing of the same tissues with
a validated assay in the same laboratory;
• Comparing the new test’s results with the results of testing the same tissue validation set
in another laboratory using a validated assay;
• Comparing the new test’s results with previously validated non-immunohistochemical tests;
or
• Testing previously graded tissue challenges from a formal proficiency testing
program and comparing the results with the graded responses.
Laboratories are required by the Clinical Laboratory Improvement Amendments of 1988 (Sec. 493.1253)
to validate the performance characteristics of all assays used in patient testing, in order to ensure that
16

the results are accurate and reproducible. “Validation means confirmation by examination and
provision of objective evidence that the particular requirements for a specific intended use can be
consistently fulfilled.

17

This includes establishment of the analytic validity of all non FDA16

cleared/approved (or “laboratory developed”) tests.

Analytic validity has been defined as the ability to accurately and reliably identify or measure the marker
2,6

of interest in specimens that are representative of the clinical population to be tested. The concept of
validation specimens that are “representative of the patients to be tested” is a key accepted premise or
18

“first principle” of assay validation. The key criteria in grading the quality and strength of evidence for
analytic validation include the internal validity of the studies and the consistency and generalizability of
2,19

To achieve generalizability of the laboratory’s analytic validation results, the tissues
the results.
included in a validation set must be typical of the
specimens received in routine practice and must provide a representative range of expression
intensities and patterns.
The strength of evidence was Adequate to support Recommendation 6: that laboratories
should, whenever possible, use the same fixative and processing methods as cases tested
clinically, in order to validate using representative specimens.
Components of analytic validity applicable to IHC assays are accuracy, analytic sensitivity
(detection rate) and specificity (1-false positive rate), concordance (overall, positive, negative)
2,6,15,16

Analytic sensitivity and specificity are
and precision (repeatability, reproducibility).
estimated by comparing a new assay’s results with a “gold” standard referent test or validated
tissue set. However, “gold” standard” referent tests for IHC assays are rare. For example, no
confirmatory or “gold standard” test currently exists for HER2, ER and PgR IHC and these
1,3,15,20

results do not represent “truth”.
A HER2 in-situ hybridization assay (e.g., FISH, CISH,
SISH) can only indirectly validate a HER2 IHC test, because a nucleic acid based assay does
not measure the same analyte.

© 2015 College of American Pathologists. All rights reserved.

PAGE 11

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Therefore, laboratories must use other approaches to demonstrate assay performance. Primary
validation and method comparison studies and key published professional guidelines described IHC
3,15,18,21-39

They included comparisons of a new test’s results to:
validation approaches.
clinical outcomes; to other validated IHC tests, to or other referent tests (intra- or inter- laboratory); or to
20,22,30-32,34,40-51

Based on these studies,
tissue validation sets previously characterized by consensus.
the standard metrics for IHC validation results are overall concordance between the results of the new
and referent assay(s), the Kappa statistic, and positive and negative concordance for assays with binary
results (positive, negative) that can be entered into a 2x2 table (Table 3). Quality grades for studies
referenced here were 2 Good, 22 Fair, and 6 Poor; grades for 8 other articles/documents were 2 Good
and 6 Fair.
The strength of evidence was Adequate to support the KQ 1 outcome of when analytic validation should
be done, and that it should include analytic sensitivity and specificity (or concordance in absence of a
“gold” standard referent test).
The evidence was Inadequate (i.e., evidence was not available or did not permit a conclusion to be
reached) for the KQ 1 outcome of how validation should be done with regard to the listed approaches,
but did show that these approaches have been used.
The precision of an IHC assay, or result repeatability, is the extent of agreement among results (i.e.,
positive/negative results, staining patterns/localization, level of expression) obtained by replicate testing
14,15

Reproducibility assesses the
of tissue specimens under specified conditions.
extent of agreement among results obtained by replicate testing of specimen sets between laboratories,
14,15

testing platforms or readers.
Evaluation of precision is an element required by CLIA, and CLSI IHCspecific guidance states that IHC assay validation requires acceptable
precision in the analytical (e.g., result repeatability over days) and postanalytical/interpretive (e.g., inter15,16

operator reproducibility) phases.

However, no studies were identified that provided data on assay repeatability over two or more
days. One guidance document recommended running validation samples over multiple days,
37

with no more than 20 samples tested in one day. Based on a recent CAP survey, the
proportion of laboratories that agree with“…validation cases tested on multiple days to assess
between-run precision” was 53% and 57% for non-predictive and predictive assays,
52

respectively. Since over half of laboratories support this, a possible reason for lack of
identified studies may be that this step is considered too routine for inclusion in publications.
Another possibility is that studies containing this information were published in the early years
of IHC testing and were not captured in the post-2004 search.
A small number of studies and guidance documents addressed reproducibility. Two guidance
documents have called for ongoing monitoring of the competency of histotechnologists and
3,37

One recommended that the laboratory
pathologists by measuring inter-rater reproducibility.
director determine the timing and standards for competency testing, while another called for
3,38

95% concordance as the standard for inter-operator or inter-laboratory reproducibility.

Five

49,53-56

studies were identified that reported inter-rater and/or inter-laboratory reproducibility.
However, the differences between the study protocols were so numerous that no conclusions
were possible. For example, the studies tested different markers (HER2, PTEN, multiple),
compared different numbers of raters (2 to 6) and laboratories (2-3), and variably expressed
results as coefficients of variation, percent concordance, Kappa statistic, weighted Kappa
statistic and “composite ratings.” No raw data were available to allow reanalysis.

© 2015 College of American Pathologists. All rights reserved.

PAGE 12

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Quality grades for studies referenced here were 3 Fair and 2 Poor; 1 document was graded
Good and 3 Fair.
The strength of evidence for the KQ 1 outcome of precision was Adequate to support inclusion
of precision (e.g., inter-run and inter-operator) as part of validation. The evidence was
Inadequate to assess the precision of IHC assays in practice.
The strength of evidence was Adequate to support Recommendation 1: “Laboratories must
validate all immunohistochemical tests before placing into clinical service.”
The panel found that analytic validation provides a net benefit for the overall performance
and safety of IHC tests by contributing to the avoidance of potential harms related to analytic
false positive and false negative test results.
KQ 2 and KQ 3: What is the minimum number of positive cases (KQ 2) and negative cases (KQ
3) that need to be tested to analytically validate an immunohistochemical assay? Does the minimum
number differ depending on whether the IHC assay:
Is primarily used to identify cell lineage (i.e., non-predictive markers)?
Is used to direct patient treatment (i.e., predictive markers)?
Is used to identify an infectious organism?
Is used to identify rare antigens?
Is done on cytology specimens?
Is done on decalcified specimens?
“The perennial question is, ‘How many samples do I need to run to validate a given test?’
Unfortunately, the answer is always the same—it depends. It depends on “…how the test is
to be used, which performance criteria are most critical for the intended use, and the
confidence
level that is required for good medical practice, implying that medical judgment is required.”

57

A first step in addressing this question is to consider what criteria are most likely to impact the
number of samples needed to validate IHC assays overall, and for the specific intended uses
and specimen types listed above.
Intended Use
Class I tests have been defined as interpreted by pathologists in the context of histomorphologic,
15,58-

cytomorphologic and clinical data and reported as one part of a panel of tests or clinical evaluation.
60

Class I tests may also be referred to as non-predictive or qualitative, though they may have a

quantitatively defined threshold (e.g., >10% reactive cells).

59

58

stand-alone tests with no routine morphologic correlates.

In contrast, Class II tests are generally

Class II test results are reported to physicians
15,59,60

as independent diagnostic information, and may influence treatment decisions.
tests fall into Class II.

Predictive IHC

Based on intended use, tests could be classified as predictive or non-predictive for purposes of validation
standards. Of course, some tests can fall into both categories, depending on intended use. For example,
CD117 can be considered Class I as an acute leukemia marker of myeloid differentiation, and Class II in
assessing a stromal gastroesophageal tumor to determine the
61

patient’s eligibility for imatinib treatment. Other criteria for determining number of validation
samples include the complexity of interpretation (i.e., multiple outcomes require more samples) and
feasibility (i.e., the number and range of control materials may be limited, especially for some non-

© 2015 College of American Pathologists. All rights reserved.

PAGE 13

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

15

predictive tests). In addition, the observed concordance and possible bias between
tests in the initial validation may necessitate further testing and, possibly, additional validation
59

specimens.

No studies were identified that addressed the four specific intended uses listed in KQ 2 and KQ 3, but
classifying tests’ intended use as predictive or non-predictive provides a rationale for determining the
number of samples needed for validation. Due to the potential for direct impact on clinical management, it
is not surprising that predictive tests appear to require higher certainty
18,37

in the quality of validation results.

Strength of evidence was Adequate to support an outcome of KQ 2 and KQ 3, the decision to distinguish
between non-predictive (Class I) and predictive (Class II) IHC tests in determining the recommended
number of validation samples.
Strength of evidence was Adequate to support the separation of Recommendation 3 and
Recommendation 4 in order to distinguish between non-predictive and predictive IHC tests for
determining the recommended number of validation samples.
Strength of evidence was Adequate to support Recommendation 5, regarding use of the higher
validation standard (e.g., number of samples) in the case of a marker with both non-predictive and
predictive intended uses.
Information on Numbers of Samples for Validation
Available information on the recommended number of samples needed for validation was limited.
Suggested numbers were found in four professional society clinical guidelines (quality grade Fair), two
consensus meeting reports (grade Fair), and one CLSI approved guideline
3,15,18,37,38,59,62

(grade Fair).

Note that four of these documents focused on specific predictive tests

(HER2, ER, PgR), and three on IHC assays in general.
samples:

3,15,18,37,38,59,62

Guidance on numbers of

38

Minimum 25 samples, 10 high, 10 intermediate, 5 negative
25-100 samples (no breakdown)

3,62
59

50-100 samples, 25-50 positive with an unspecified mix of weak positives, 25-50 negative
15,18,37

≥ 80 samples, ≥ 40 positive (10 weak positive), ≥ 40 negative

In the absence of clear guidance on the number of validation samples to run, the Methodologist
requested help from Women & Infants Hospital statistician (Glenn E Palomaki, PhD) to develop tables to
assist the panel in discussing this important question. Practical guidance on the size of a validation set
can be provided by statistical analysis. Simply put, the more samples that are run in a validation set, the
higher the likelihood that the concordance estimate reflects the test’s “true” concordance. But to apply
and test this approach, it was necessary to determine what concordance benchmark would be used. The
concordance benchmarks commonly mentioned in guidance documents are 90% and 95%. We reviewed
available validation and method comparison studies to identify data that might support the selection of a
benchmark.
Determining a Concordance Benchmark
Supporting evidence was identified in studies and documents reporting “real world” concordance data
from IHC validation studies, method comparisons and proficiency testing or interlaboratory comparisons.
The following is a summary of analyses. More detailed data can be found in the Appendix, Tables 3-5.
61

Data were analyzed from a two-year inter-laboratory comparison of CD117 IHC testing. Ten blinded
tissues were run in 2004 by 63 laboratories, and again in 2005 by 90 laboratories. The set included

© 2015 College of American Pathologists. All rights reserved.

PAGE 14

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

four gastrointestinal stromal tumors (GIST) positive for CD117 and six tumors that were negative by
histopathologic diagnosis. For the combined 1,530 challenges, the concordance estimate between
the laboratory responses and the target diagnosis was 88% (95% CI 86-89; k=0.75). Results for 2004
and 2005 were not statistically different. Positive concordance was 98% and negative concordance
was 81%. The McNemar’s statistic was p<0.001, confirming that the observed asymmetry in
discordant results (12 false negatives and 177 false positives) was significant. Possible explanations
included the presence of necrotic foci or CD-117 positive mast cells in normally CD117 negative
tumors (e.g., leiomyosarcoma) or the variability in primary antibodies and antigen retrieval methods for
tests between laboratories.
Data from comparisons of HER2 IHC assays were analyzed. Median overall concordance in 5
comparisons of different HER2 IHC tests was 89% (range 74–93%), with 2 of 5 studies greater than
22,30-32,34

Note that concordance estimates and
90% concordant (Appendix, Table 3).
associated Kappa and McNemars statistics were computed from 3x3 contingency tables (BMDP
Statistical Software, Los Angeles, CA).
The summary concordance estimate (random effects model) was similar at 88.1% (95% CI 81.32

92.7), but heterogeneity was high (I =89, p < 0.001), and could not be explained by analysis of
selected covariates (e.g., tissue type, study size, study quality grade). The number of studies was too
small to allow analysis for the many possible covariates. One study was rated Good and 4 Fair. The
McNemar’s p values < 0.05 indicate a significant difference/bias between the false positive and false
negative discordant results in a number of these comparisons. Such information can be helpful for
next steps in validation.
Data were analyzed from comparisons between HER2 IHC assays and in situ hybridization tests (e.g.,
FISH). Median overall concordance in 7 comparisons from the four identified studies in breast cancer
tissue was 89% (range 66–94%), with 2 of 7 studies > 90% concordant
31,34,42,49

(Appendix, Table 4).
Three studies used The HER2 4B5 primary antibody and three
used CB11. Within the limitations of the small number of studies, the results for each antibody were
consistent with the overall estimate. The summary concordance estimate (random effects model) was
2

similar at 88% (95% CI 81-93), but heterogeneity was high (I =89, p < 0.001),
and could not be explained by analysis of selected covariates (e.g., tissue type, study size, study
quality grade). The number of studies was too small to allow analysis for the many possible
covariates. There was a suggestion of publication bias (Egger’s p=0.002) that became insignificant
when the largest study was removed (a LDT with the lowest concordance of 66%,
42

k=0.37 and McNemar’s p<0.001).

The quality grade for all studies was Fair.

The median concordance estimate for 4 comparisons in 3 studies of HER2 IHC and in situ
hybridization in gastric cancers was 95% (range 88-98%), with 3 of 4 studies >90%
22,43,44

concordant.

The grade for the studies was 2 Good, 1 Fair and 1 Poor.

Analyses of data from comparisons between HER2 IHC tests and alternative referent tests. Median
overall concordance from 4 studies of IHC tests (ER, PR, HER2, p16) compared to alternative referent
tests (e.g., RNA expression, clinical diagnosis, consensus results) was 87%
20,40,45,46

(range 72–95%), with 1 of 4 studies >90% concordant (Appendix, Table 5).

These data illustrate the challenge of achieving an overall concordance of 95%, even in relatively large
studies almost entirely made up of IHC tests with guidance recommending stringent protocol standards
3,37,39,59

An overall concordance standard that is too stringent could have the
(i.e., HER2, ER, PgR).
effect of delaying or preventing successful validation, particularly for
non-predictive tests. Overall concordance of 90% was achieved in nearly half of the above analyzed

© 2015 College of American Pathologists. All rights reserved.

PAGE 15

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

comparisons, all of which were subject to many sources of variation (e.g., sample type; ischemic time;
fixation, antigen retrieval and staining protocols; scoring). Therefore, laboratory validation studies
designed to minimize differences in such variables would have a higher probability of meeting a 90%
concordance benchmark.
Strength of evidence was considered Adequate to support the adoption of a 90% (versus 95%) overall
concordance benchmark as an outcome for KQ 2 and KQ 3.
Strength of evidence was Adequate to support Recommendation 2 for a 90% overall concordance
benchmark for analytic validation of IHC tests (excepting HER2, ER, PgR).
Considering the number of tissues needed for a validation set
The basic statistical premise is that the more samples that are run in a validation set, the higher the
likelihood that the concordance estimate reflects the “true” performance of the test. As an example, 3
discordant results would be expected in a 10 sample validation set for a test with a “true” concordance of
70%. However, only 1 discordant result could be observed by chance, resulting in a concordance overestimate of 90%. In a 20 sample validation set, 6 discordant results would be expected for the test with a
“true” concordance of 70%. Observation of only 2 discordant samples could occur by chance, but the
likelihood would be low.
Of course, the premise of “..the more samples the better..” has to be balanced by laboratory feasibility
issues such as costs and resources. It is also important to keep the goal in mind – to keep false
validation failures low while identifying assays that are truly not performing well.
Table 6 in the Appendix is an example of those considered by the panel. With a 10 sample validation set,
the benchmark is reached with only 1 discordant result. The concordance estimate is 90% with a lower
95% confidence limit (L95%) of 57%. The “true” concordance could be lower or higher than 90%, but
there is only a small chance (about 5%) that it will be lower than 57%. The validation fails with 2
discordant results. Even with a “true” concordance of 80%, a 10 sample validation set has a greater than
1 in 3 chance of meeting the 90% benchmark, compared to a 1 in 5 chance in a 20 sample validation set.
A 20 sample validation set allows 2 discordant results for a 90% concordance estimate with a L95% of
74%, a more confident result.
Consideration of a 20 sample (10 positive, 10 negative) validation set for non-predictive tests
Overall concordance estimates meet the benchmark with 0, 1 or 2 observed discordant results
among the total set of 20 tissues (Table 4). The “true” concordance between the two assays has only a
5% chance of falling outside the 95% CI of each concordance estimate, and can be lower or higher than
the estimate. If the 100% or 95% concordance estimates (0, 1 observed discordant results) are a “true”
representation of the relationship between the two tests, the validation result would meet the benchmark
more than 92% of the time (Table 5). If the 90% concordance estimate is “true”, the probability of meeting
the benchmark would be 68%.
For validation results that do not meet the benchmark, it may not be useful to perform the McNemar’s
test in a small validation set (e.g., 20 tissues). The McNemar’s test is based solely on discordant results,
which are likely to be few in a small validation set. Therefore, a non- significant McNemar’s test could be
due to true symmetry between the number of discordant results, or to asymmetry on the off-diagonal but
with insufficient numbers to show statistical
significance (i.e., underpowered to find even important differences between the tests). In many
cases, a visual inspection of the results in a 2x2 or 3x3 table will identify a potential explanation for the
validation failure.
The laboratory medical director will determine any corrective action and how many additional tissues
should be tested.

© 2015 College of American Pathologists. All rights reserved.

PAGE 16

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Table 4. Validation Using a 20 Tissue Validation Set (10 Positive and 10 Negative) against
a
a 90% Concordance Benchmark

Number of
validation
tissues
20 Total

0 discordant
Concordance
estimate
(95% CI)
100%
(81-100)

1 discordant
Concordance
estimate
(95% CI)
95%
(75-100)

2 discordant
Concordance
estimate
(95% CI)
90%
(69-98)

a

Concordance estimates with 95% CI stratified by number of observed discordant samples
Abbreviation: CI= confidence interval

Consideration of a 40 sample (20 positive, 20 negative) validation set for predictive tests The
statistical argument is updated here for predictive factor assays. Table 6 provides overall concordance
estimates with 95% CIs for the 40 tissue validation set, as well as the 20 tissue sets for those who will
compute positive and negative concordance estimates. Overall concordance estimates (Table 6, shaded
row) meet the benchmark with 0 to 4 observed discordant results among the total set of 40 tissues. The
“true” concordance between the two assays can be lower or higher than the estimate, but has only a 5%
chance of falling outside the 95% CI of the concordance estimate (L95% is 76% for a 90% concordance
estimate).
If the 95-100% concordance estimates (0, 1, 2 observed discordant results) are a “true” representation of
the relationship between the two tests, the validation results would meet the benchmark more than 95%
of the time (Table 5). The probabilities of meeting the benchmark if the 92.5% and 90% concordance
estimates are “true” would be 82% (approximation) and 68%, respectively. The positive (or negative)
concordance estimates among 20 tissues (bottom row) meet or exceed the same benchmark with 0, 1, or
2 discordant results.
Table 5. The percent probability of meeting or exceeding a specified benchmark concordance
rate based on the number of specimens in the validation set and the “true” concordance rate of
a

the assay

Tissues in the Validation Set

a

20

40

21
40
68
92
99
7
18
39
74
94

8
26
63
95
>99
1
5
22
68
95

“True”
concordance
rate
80
85
90
95
98
80
85
90
95
98

Benchmark
Concordance
Rate
90%

95%

StatTrek.com Binomial Calculator and consistent with Wolff et al., 201318

© 2015 College of American Pathologists. All rights reserved.

PAGE 17

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Table 6. Validation Using a 40 Tissue Validation Set (20 Positive and 20 Negative) against a 90%
a

Concordance Benchmark

Number of
validation
tissues
40
Total
20
Positive or
Negative

0 discordant
Concordance
estimate (95%
CI)
100%
(90-100)

1 discordant
Concordance
estimate (95%
CI)
97.5%
(86-100)

2 discordant
Concordance
estimate (95%
CI)
95%
(83-99)

3 discordant
Concordance
estimate (95%
CI)
92.5%
(79-98)

4 discordant
Concordance
estimate (95%
CI)
90%
(76-97)

100%
(81-100)

95%
(75-100)

90%
(69-98)

85%
(63-96)

80%
(58-92)

a

Concordance estimates with 95% CI stratified by number of observed discordant samples Abbreviation: CI= confidence
interval

In a 40 sample validation that does not meet the benchmark, analyses such as the McNemar’s test and
kappa statistic may help determine whether an observed difference in the off-diagonal represents a
significant bias between the new and referent tests (Figure 2). In this case, the kappa statistic showed
“substantial” agreement, but the overall concordance estimate missed the benchmark by a small margin.
The positive concordance of 75% suggests false negatives could be occurring in the new test. The
McNemar’s p was 0.13 (not significant), indicating that the 5 discordant results all in a single cell could
have happened by chance. Alternatively, the test could be underpowered.

Figure 2. A 2x2 contingency table of a 40 tissue validation set that did not meet the benchmark
(results entered into a 2x2 contingency table) with associated statistical tests

New IHC Result
Positive
Negative

Referent Result
Positive
15
5
20

Referent Result
Negative
0
20
20

16
24
40

Overall concordance = 35/ 40 = 87.5% - Does not meet the 90% benchmark k = 0.75
McNemar’s p = 0.13, not significant
Positive concordance = 15/20 = 75%
Negative concordance = 20/20 = 100%
Abbreviation: IHC= immunohistochemical

Some laboratories may choose to validate predictive tests with tissue sets larger than the
recommended minimum. For validation sets of 80 samples or more, the McNemar’s test is
more useful in documenting whether observed differences/biases between the tests are
significant. For example, for an 80 tissue validation set in which the numbers in each of the 4
cells in Figure 2 are doubled, the McNemar’s result for 10 to 0 asymmetry on the off-diagonal
would be significant (P=0.004).

© 2015 College of American Pathologists. All rights reserved.

PAGE 18

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

The laboratory medical director will determine any corrective action and how many additional tissues
should be tested.
Strength of evidence was Inadequate to support Recommendation 3 and Recommendation 4 in
determining the recommended number of validation samples.
Number of specimens in a validation set for IHC tests performed on cytologic specimens.
No primary studies, systematic evidence reviews or qualitative documents were identified that addressed
the specific question regarding the number and type of cytology specimens that are needed in a
validation set for a new IHC assay. One guideline did recommend that each laboratory should validate
15

IHC assays for cytological specimens separately from those for surgical specimens.

However, studies were identified that compared cytology specimens to FFPE histologic sections for ER,
63-68

Concordance estimates and Kappa statistics
PgR and/or HER2 IHC testing (Appendix, Tables 7-9).
were consistently high at≥ 90% and >0.75, respectively. The lack of a significant finding by the
McNemar’s test may be partly related to small sample size (4 of 5 data sets had 50 or less samples), but
positive and negative concordance rates were also reasonably consistent. However, the studies were
few, generally small, and used different fixatives, fixation times, and cytology specimens (e.g., smears,
ThinPrep, cell blocks). In 3 studies only about 90% of samples were assessable. No two studies could be
directly compared.
The strength of evidence was Inadequate ( i.e., evidence was not available or did not permit a conclusion
to be reached) to address the KQ 2 and KQ 3 outcome of number of samples needed for validation with
cytology specimens.
Number of specimens in a validation set for IHC tests performed on decalcified specimens
No primary studies, systematic evidence reviews or qualitative documents (e.g., guidelines, consensus
meeting reports) were identified that addressed the specific question regarding the number of decalcified
bone marrow specimens from positive and negative cases needed in a validation set for a new IHC
assay.
Nine articles and documents addressed the potential influence of decalcification as a modifier in the
15,39,48,69-74

analytic validation process.

Some reported significant variability in decalcification protocols
70-73

(e.g., decalcification solutions, time in solution) and in preservation of antigenicity in IHC tests.
inter-laboratory survey in Europe reported that 68% of laboratories used the same protocols for
73

decalcified bone biopsies as for non-decalcified tissues.

One

Two IHC guidelines recommend interpreting
15,39

IHC results on decalcified samples with caution regarding the possibility of antigen (and tissue) loss.
However, others reported good morphology and successful staining with protocols using different
fixatives, acid or EDTA decalcification, and paraffin or resin embedding.

© 2015 College of American Pathologists. All rights reserved.

48,69,72,74

PAGE 19

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

These variable observations emphasize the need for a defined protocol and a validation plan that will
ensure robust and reproducible IHC results in decalcified specimens.
The strength of evidence was Inadequate to address the KQ 2 and KQ 3 outcome of number of samples
needed for validation with decalcified specimens.

KQ 4. What parameters should be specified for the tissues used in the validation set?
Set ratio of immunoreactive versus non-immunoreactive? Set ratio of high expressors versus low
expressors?
Set ratio of neoplastic versus non-neoplastic (when appropriate)?
Should a minimum tissue size or minimum quantity of cells be specified?
No primary studies, systematic evidence reviews or qualitative documents (e.g., guidelines, consensus
meeting reports) were identified that addressed the specific question regarding the parameters that
should be specified in validation sets with regard to neoplastic versus non- neoplastic tissues.
Several guidelines have suggested a 50:50 ratio of immunoreactive versus non-immunoreactive
3,15,18,37

Information on number of low or weak expressors versus high expressors is similarly
tissues.
unspecified. In a recent CAP survey, participating laboratories reported that the median proportion of
positive validation cases that were “weakly or focally” positive was 20% for non- predictive (N=195
respondents) and predictive (N=141) assays.
th

52

non-predictive assay validation was 7 (10 -90

The reported median number of positive samples run for

th

centiles=2-20), of which 1-2 would be weakly positive.
th

th

For predictive assay validation, the median number of positives samples was 10 (10 -90 centiles=230), of which 2 would be weakly positive. It appears this approach would lead to low certainty regarding
validation results.
There was no specific guidance on sample size, but of 34 reviewed studies that reported whole section
size, the results were 18%, 47% and 21%, respectively, for 3 um, 4 um and 5 um; the remaining 5
23,24,26-28,30,31,42,44,46,49,56,66,67,69,75-87

studies reported ranges of 2-4 um (N=3)or 4-6 um (N=22).

15,34,41,79,88-91
Reports from 8 studies on core size for TMAs ranged from 0.6 to 3 mm.
No other articles
addressed minimum tissue size or quantity of cells. A related question was raised about the comparison
of TMAs with different sizes and number of cores to whole sections.
The strength of evidence was Inadequate to address other KQ 4 outcomes regarding four specific
parameters for tissues in a validation set.
Comparisons of concordance between IHC assays performed on whole sections and TMAs
Comparisons of overall concordance between IHC assays performed on whole sections and TMAs have
21,23- 29,33,35,36,92

been done with at least 9 markers, but primarily with ER, PgR and HER2.
Summary
estimates of concordance (random effects model) were computed, but heterogeneity was high across the
2
studies (I >75; p < 0.001), and specific sources of heterogeneity could not be identified. Consequently,
concordance is reported as ranges with median values.
The median overall concordance estimate was 93% (range 73-100%)(Appendix, Table 10). Data were
stratified by study quality, marker (Appendix, Table 11) and core size (Appendix, Table
12) as possible sources of heterogeneity. All results were consistent between quality scores,
markers and core sizes. Concordance estimates met or exceeded the 90% benchmark in about two
thirds of cases. Table 13 provides limited data on other markers. The quality of studies was 8 Fair and 4
Poor.
Strength of evidence was Inadequate to recommend the routine use of TMA samples. Strength of

© 2015 College of American Pathologists. All rights reserved.

PAGE 20

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

evidence was Adequate to support the conclusion that TMA samples have been
successfully utilized in IHC tests, but there are many variables to be considered and thorough
validation is needed for each marker.
The strength of evidence was Adequate to support Recommendation 9 regarding the need for careful
validation to determine if TMAs are appropriate for the targeted antigen and the fixation and processing
is similar to clinical specimens.
KQ 5. How do the following modifiers influence analytic validation?
Type of fixative
Type of decalcification solution Time in decalcification solution
Validation tissues processed in another laboratory
No primary studies, systematic evidence reviews or qualitative documents (e.g., guidelines, consensus
meeting reports) were identified that addressed the specific question regarding the potential influence on
validation of tissues processed in another laboratory.
Nine articles and documents addressed the potential influence of the type and timing of decalcification
15,39,48,69-74

as a modifier in the analytic validation process.
decalcification protocols (e.g., decalcification solutions, time in

Some reported significant variability in

70-73

solution) and in preservation of antigenicity in IHC tests.
Two IHC guidelines recommend interpreting
IHC results on decalcified samples with caution regarding the possibility of antigen
15,39

However, others reported good morphology and successful staining with
(and tissue) loss.
protocols using different fixatives, acid or EDTA decalcification, and paraffin or resin
48,69,72,74

embedding.
These observations emphasize the need for a defined protocol and a validation
plan that will ensure robust and reproducible IHC results in decalcified specimens.
Strength of evidence was Inadequate to address the KQ5 outcomes regarding the influence of the type
of decalcification solution, the time in decalcification solution, or validation tissues processed in another
laboratory on analytic validation.
The influence of the type of fixative on analytic validation
The authors of a 2011 article reviewed 39 primary studies that investigated preanalytical variables
93

identified by a literature survey. Among 15 preanalytical variables with the potential to impact IHC
assays were time to fixation (cold ischemic time), fixative type (e.g., concentration, pH), and time in
fixative. Studies have shown that fixation delay of more than 12 hours affects the extent and intensity of
93

immunostaining, possibly leading to false negative results.

Another report found that delays of even 118,93

2 hours may decrease signal intensity in ER, PgR and HER2.
less than 1 hour delay when possible, but

One IHC guideline recommends a

39

certainly as short a delay as possible.

The most commonly recommended fixative is 10% neutral buffered formalin (NBF), but most studies
have focused on a narrow range of IHC assays (e.g., ER, PgR, HER2) in one tissue. The fixative used
can affect the extent and intensity of staining as well as nonspecific background staining, and antigen
93

specific effects have been reported. Time in fixative can also affect the extent, distribution and intensity
of staining, and may be antigen dependent. Fixation for limited periods beyond 72 hours has not resulted
in a reduction in assay sensitivity in several studies assays, and effective antigen retrieval may maintain
immunoreactivity even after fixation for several days.

76,92,94,95

The available data are, with some exceptions, focused on IHC hormone markers that help inform

© 2015 College of American Pathologists. All rights reserved.

PAGE 21

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

treatment options for women with breast cancer. However, this review is intended to provide information
to inform recommendations on analytic validation for a wide range of non-predictive and predictive
markers. The available data may, in fact, be applicable to a wide range of antigens. In the meantime,
however, careful validation will help determine when antigen specific protocol changes may be needed
for these preanalytic variables.
Strength of evidence was Inadequate ( i.e., evidence was not available or did not permit a conclusion to
be reached) to address the KQ 5 outcome regarding the influence of fixation on analytic validation.
Strength of evidence was Adequate to support that laboratories should, whenever possible, use the
same fixative and processing methods as cases tested clinically, in order to validate using representative
tissues.
KQ 6: Which of the following conditions require assay revalidation?
New lot of antibody
Change in antibody clone
Change in antibody dilution
Change in type of fixative
Change in antigen retrieval method
Change in antigen detection system Change in instrumentation
Change in water supply
Laboratory relocation
Assay no longer performing as expected
KQ 7: Does assay revalidation have the same requirements as initial assay validation?
Available information on the conditions or changes that require assay revalidation was limited.
In general, revalidation was recommended for “any significant changes to an assay/test system”
or “any deviation from a standardized method” This recommendation was found in four professional
society clinical guidelines (quality grade Fair), two consensus meeting reports (grade Fair), and one CLSI
3,15,18,37-39,62

Note that four of these
approved guideline (grade Fair).
documents focused on specific predictive tests (HER2, ER, PgR), and three on IHC assays in
3,15,18,37-39,62

general.
(Table 7).

Some of these documents also recommended revalidation for specific changes

38,59

Two guidelines recommended a limited revalidation for a new primary antibody lot.
Among CAP
Survey responders, 64% believed revalidation should be done for a new lot of primary antibody in
52

predictive tests, but whether a full or limited validation was not questioned.

Two guidelines

3,39

recommended scheduled revalidation, one semi-annually and one annually.
No guidelines addressed
change in antibody dilution, change in water supply, laboratory relocation, and assay no longer
performing as expected.
No primary studies with data supporting the consensus expert opinions were identified. Three of the
expert consensus guidelines were informed by an evidence review, but no references supported the
3,18,37

This guidance is based on qualitative information derived from expert
guidance about revalidation.
opinion and principles of good laboratory practice. It is possible that studies documenting clinically
significant result variation based on the effects of the listed changes predate 2004, or would need
different search terms to be identified.
No specific information was identified that addressed whether the requirements of revalidation are the

© 2015 College of American Pathologists. All rights reserved.

PAGE 22

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

same as initial assay validation. The term “revalidation” is not included in the CLSI Harmonized
Terminology Database.

14

Table 7. Referenced guidance on specific changes requiring revalidation and
responses from laboratories who agreed revalidation of predictive tests should be
done for those changes

Specific changes requiring IHC revalidation

2010 CAP Survey52
Non-HER2 predictive assays
% responding revalidation should
be done (N)

Modification of a commercial kit15
Primary antibody clone

NA

15,37,39,59

NA

Primary antibody provider59

NA

Change between in-house primary antibody dilution and
pre-dilution59

NA

Fixative/fixation method15
Antigen retrieval method

15,37,39,59

Detection system15,37,39,59
Instrumentation
Autostainer 51

74 (295)
80 (294)
81 (293)
78 (296)
Tissue processor, 55 (292)

Addition/change in imaging system51

NA

Relaxation of quality management procedures 37

NA

Abbreviation: N = number of respondents for that question; 2NA =this change was not part of the survey

The strength of evidence was Inadequate to address KQ 6 on conditions requiring
assay revalidation and KQ 7 on whether revalidation should be the same as initial
validation.
The strength of evidence was Inadequate to support Recommendation 10, Recommendation
11, Recommendation 12 or Recommendation 13.

© 2015 College of American Pathologists. All rights reserved.

PAGE 23

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

REFERENCES
1.
2.
3.

4.
5.

6.

7.
8.

9.
10.
11.
12.
13.
14.
15.

16.

17.

18.

19.

20.
21.

Marchio C, Dowsett M, Reis-Filho JS. Revisiting the technical validation of tumour biomarker assays:
how to open a Pandora's box. BMC Med. 2011;9:41.
Teutsch SM, Bradley LA, Palomaki GE, et al. The Evaluation of Genomic Applications in Practice and
Prevention (EGAPP) Initiative: methods of the EGAPP Working Group. Genet Med 2009;11(1):3- 14.
Wolff AC, Hammond ME, Schwartz JN, et al. American Society of Clinical Oncology/College of American
Pathologists guideline recommendations for human epidermal growth factor receptor 2 testing in breast
cancer. Arch Pathol Lab Med. 2007;131(1):18-43.
Paxton A. Prepping for a firmer FDA hand in regulating LDTs. CAP Today. 2010;24(9):5-12.
US Food and Drug Administration, Center for Devices and Radiological Health. Oversight of Laboratory
Developed Tests (LDTs). 2010. Available at:
http://www.fda.gov/medicaldevices/newsevents/workshopsconferences/ucm212830.htm. Accessed
October 4, 2013.
Haddow JE, Palomaki GE. ACCE: a model process for evaluating data on emerging genetic tests. In:
Human Genome Epidemiology: A scientific foundation for using genetic information to improve health
and prevent disease. Oxford:Oxford University Press; 2003:217-233.
Whiting PF, Weswood ME, Rutjes AW, et al. Evaluation of QUADAS, a tool for the quality assessment of
diagnostic accuracy studies. BMC Med Res Methodol. 2006;6:9.
Agency for Healthcare Research and Quality, Robert Wood Johnson Foundation. Conference on
Qualitative Methods in Health Services Research. Rockville, MD, December 4, 1998. Available at:
http://archive.ahrq.gov/about/cods/codsqual.htm. Accessed October 16, 2012.
Bowen GA. Document Analysis as a Qualitative Research Method. Qual Res. 2009;9(2):27-40.
Leys M. Health care policy: qualitative evidence and health technology assessment. Health Policy
2003;65(3):217-226.
Murphy E, Dingwall R, Greatbatch D, Parker S, Watson P. Qualitative research methods in health
technology assessment: a review of the literature. Health Technol Assess. 1998;2(16):iii-ix, 1-274.
Grant MJ, Booth A. A typology of reviews: an analysis of 14 review types and associated methodologies.
Health Info Libr J. 2009;26(2):91-108.
Centre for Reviews and Dissemination. Systematic Reviews: CRD’s guidance for undertaking reviews in
health care. York, England:CRD, University of York, 2009.
Clinical Laboratory Standards Institute (CLSI) Harmonized Terminology Database. Available at:
http://login.clsi.org/HTDatabase.cfm. Accessed May 29, 2013.
Clinical Laboratory Standards Institute. Quality assurance for design control and implementation of
immunohistochemistry assays: approved guideline, second edition. In: CLSI Document I/LA28-A2.
Wayne, PA: Clinical and Laboratory Standards Institute; 2011.
Department of Health and Human Services. Medical Devices: Classification/reclassification of
immunochemistry reagents and kits. Fed Regist. 1998;63(106):30132-30142. Codified at 21 CFR 864.
Doc. No. 94P–0341.
US Food and Drug Administration. Title 21 CFR § 820.3(z). Available at:
http://www.gpo.gov/fdsys/pkg/CFR-2012-title21-vol8/pdf/CFR-2012-title21-vol8-chapI-subchapH.pdf.
Accessed September 2, 2013.
Wolff AC, Hammond EH, Hicks DG, et al. Recommendations for human epidermal growth factor
receptor 2 testing in breast cancer: American Society of Clinical Oncology - College of American
Pathologists clinical practice guideline update. In press. 2013.
Sun F, Bruening W, Erinoff E, Schoelles KM. Addressing challenges in genetic test evaluation.
Evaluation frameworks and assessment of analytic validity. Methods research report (Prepared by the
ECRI Institute Evience-based Practice Center under contract no. HHSA 290-20007-10063-I.) AHRQ
Publication No. 11-EHC048-EF. Rockville, MD: Agency for Healthcare Research and Quality. June
2011.
Dowsett M, Hanna WM, Kockx M, et al. Standardization of HER2 testing: results of an international
proficiency-testing ring study. Mod Pathol. 2007;20(5):584-591.
Batistatou A, Televantou D, Bobos M, et al. Evaluation of current prognostic and predictive markers in
breast cancer: a validation study of tissue microarrays. Anticancer Res. 2013;33(5):2139-2145.

© 2015 College of American Pathologists. All rights reserved.

PAGE 24

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

22. Boers JE, Meeuwissen H, Methorst N. HER2 status in gastro-oesophageal adenocarcinomas assessed
by two rabbit monoclonal antibodies (SP3 and 4B5) and two in situ hybridization methods (FISH and
SISH). Histopathology. 2011;58(3):383-394.
23. Drev P, Grazio SF, Bracko M. Tissue microarrays for routine diagnostic assessment of HER2 status in
breast carcinoma. Appl Immunohistochem Mol Morphol. 2008;16(2):179-184.
24. Fons G, Hasibuan SM, van der Velden J, ten Kate FJ. Validation of tissue microarray technology in
endometrioid cancer of the endometrium. J Clin Pathol. 2007;60(5):500-503.
25. Graham AD, Faratian D, Rae F, Thomas JSJ. Tissue microarray technology in the routine assessment of
HER-2 status in invasive breast cancer: a prospective study of the use of immunohistochemistry and
fluorescence in situ hybridization. Histopathology. 2008;52:847-855.
26. Gulbahce HE, Gamez R, Dvorak L, Forster C, Varghese L. Concordance between tissue microarray and
whole-section estrogen receptor expression and intratumoral heterogeneity. Appl Immunohistochem Mol
Morphol. 2012;20:340-343.
27. Henriksen KL, Rasmussen BB, Lykkesfeldt AE, Moller S, Ejlertsen B, Mouridsen HT. Semi- quantitative
scoring of potentially predictive markers for endocrine treatment of breast cancer: a comparison between
whole sections and tissue microarrays. J Clin Pathol. 2007;60(4):397-404.
28. Jones S, Prasad ML. Comparative evaluation of high-throughput small-core (0.6-mm) and large-core (2mm) thyroid tissue microarray: is larger better? Arch Pathol Lab Med. 2012;136(2):199-203.
29. Kwon MJ, Nam ES, Cho SJ, et al. Comparison of tissue microarray and full section in
immunohistochemistry of gastrointestinal stromal tumors. Pathol Int. 2009;59(12):851-856.
30. Mayr D, Heim S, Werhan C, Zeindl-Eberhart E, Kirchner T. Comprehensive immunohistochemical
analysis of Her-2/neu oncoprotein overexpression in breast cancer: HercepTest (Dako) for manual
testing and Her-2/neuTest 4B5 (Ventana) for Ventana BenchMark automatic staining system with
correlation to results of fluorescence in situ hybridization (FISH). Virchows Arch. 2009;454(3):241-248.
31. Moelans CB, Kibbelaar RE, van den Heuvel MC, Castigliego D, de Weger RA, van Diest PJ. Validation
of a fully automated HER2 staining kit in breast cancer. Cell Oncol. 2010;32(1-2):149-155.
32. O'Grady A, Allen D, Happerfield L, et al. An immunohistochemical and fluorescence in situ hybridizationbased comparison between the Oracle HER2 Bond Immunohistochemical System, Dako HercepTest,
and Vysis PathVysion HER2 FISH using both commercially validated and modified ASCO/CAP and
United Kingdom HER2 IHC scoring guidelines. Appl Immunohistochem Mol Morphol. 2010;18(6):489493.
33. Thomson TA, Zhou C, Chu C, Knight B. Tissue microarray for routine analysis of breast biomarkers in
the clinical laboratory. Am J Clin Pathol. 2009;132(6):899-905.
34. van der Vegt B, de Bock GH, Bart J, Zwartjes NG, Wesseling J. Validation of the 4B5 rabbit monoclonal
antibody in determining Her2/neu status in breast cancer. Mod Pathol. 2009;22(7):879-886.
35. Warnberg F, Amini RM, Goldman M, Jirstrom K. Quality aspects of the tissue microarray technique in a
population-based cohort with ductal carcinoma in situ of the breast. Histopathology. 2008;53(6):642- 649.
36. Soiland H, Skaland I, van Diermen B, et al. Androgen receptor determination in breast cancer: a
comparison of the dextran-coated charcoal method and quantitative immunohistochemical analysis. Appl
Immunohistochem Molecul Morphol. 2008;16(4):362-370.
37. Fitzgibbons PL, Murphy DA, Hammond ME, Allred DC, Valenstein PN. Recommendations for validating
estrogen and progesterone receptor immunohistochemistry assays. Arch Pathol Lab Med.
2010;134(6):930-935.
38. Goldstein NS, Hewitt SM, Taylor CR, Yaziji H, Hicks DG, Members of Ad-Hoc Committee On
Immunohistochemistry Standardization. Recommendations for improved standardization of
immunohistochemistry. Appl Immunohistochem Mol Morphol. 2007;15(2):124-133.
39. Hammond ME, Hayes DF, Dowsett M, et al. American Society of Clinical Oncology/College of American
Pathologists guideline recommendations for immunohistochemical testing of estrogen and progesterone
receptors in breast cancer. Arch Pathol Lab Med. 2010;134(6):907-922.
40. Baba K, Dyrhol-Riise AM, Sviland L, et al. Rapid and specific diagnosis of tuberculous pleuritis with
immunohistochemistry by detecting Mycobacterium tuberculosis complex specific antigen MPT64 in
patients from a HIV endemic area. Appl Immunohistochem Mol Morphol. 2008;16(6):554-561.
41. Phillips T, Murray G, Wakamiya K, et al. Development of standard estrogen and progesterone receptor
immunohistochemical assays for selection of patients for antihormonal therapy. Appl Immunohistochem
Mol Morphol. 2007;15(3):325-331.

© 2015 College of American Pathologists. All rights reserved.

PAGE 25

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

42. Grimm EE, Schmidt RA, Swanson PE, Dintzis SM, Allison KH. Achieving 95% cross- methodological
concordance in HER2 testing: causes and implications of discordant cases. Am J Clin Pathol.
2010;134(2):284-292.
43. Hofmann M, Stoss O, Shi D, et al. Assessment of a HER2 scoring system for gastric cancer: results
from a validation study. Histopathology. 2008;52(7):797-805.
44. Sornmayura P, Rerkamnuaychoke B, Jinawath A, Euanorasetr C. Human epidermal growth-factor
receptor 2 overexpression in gastric carcinoma in Thai patients. J Med Assoc Thai. 2012;95(1):88-95.
45. Jordan RC, Lingen MW, Perez-Ordonez B, et al. Validation of methods for oropharyngeal cancer HPV
status determination in US cooperative group trials. Am J Surg Pathol. 2012;36(7):945-954.
46. Lehmann-Che J, Amira-Bouhidel F, Turpin E, et al. Immunohistochemical and molecular analyses of
HER2 status in breast cancers are highly concordant and complementary approaches. Br J Cancer.
2011;104(11):1739-1746.
47. Lotan TL, Gurel B, Sutcliffe S, et al. PTEN protein loss by immunostaining: analytic validation and
prognostic indicator for a high risk surgical cohort of prostate cancer patients. Clin Cancer Res.
2011;17(20):6563-6573.
48. Zustin J, Boddin K, Tsourlakis MC, et al. HER-2/neu analysis in breast cancer bone metastases. J Clin
Pathol. 2009;62(6):542-546.
49. Powell WC, Hicks DG, Prescott N, et al. A new rabbit monoclonal antibody (4B5) for the immunohistochemical (IHC) determination of the HER2 status in breast cancer: comparison with CB11,
fluorescence in situ hybridization (FISH), and interlaboratory reproducibility. Appl Immunohistochem Mol
Morphol. 2007;15(1):94-102.
50. Rhodes A, Jasani B, Anderson E, Dodson AR, Balaton AJ. Evaluation of HER-2/neu
immunohistochemical assay sensitivity and scoring on formalin-fixed and paraffin-processed cell lines
and breast tumors: a comparative study involving results from laboratories in 21 countries. Am J Clin
Pathol. 2002;118(3):408-417.
51. Allred DC, Carlson RW, Berry DA, et al. NCCN Task Force Report: Estrogen Receptor and
Progesterone Receptor Testing in Breast Cancer by Immunohistochemistry. J Natl Compr Canc Netw.
2009;7(Suppl 6):S1-S21; quiz S22-23.
52. Hardy LB, Fitzgibbons PL, Goldsmith JD, et al. Immunohistochemistry validation procedures and
practices: a College of American Pathologists survey of 727 laboratories. Arch Pathol Lab Med.
2013;137(1):19-25.
53. Emerson LL, Tripp SR, Baird BC, Layfield LJ, Rohr LR. A comparison of immunohistochemical stain
quality in conventional and rapid microwave processed tissues. Am J Clin Pathol. 2006;125(2):176- 183.
54. Gustavson MD, Bourke-Martin B, Reilly D, et al. Standardization of HER2 immunohistochemistry in
breast cancer by automated quantitative analysis. Arch Pathol Lab Med. 2009;133(9):1413-1419.
55. Ruschoff J, Dietel M, Baretton G, et al. HER2 diagnostics in gastric cancer-guideline validation and
development of standardized immunohistochemical testing. Virchows Arch. 2010;457(3):299-307.
56. Sangale Z, Prass C, Carlson A, et al. A robust immunohistochemical assay for detecting PTEN
expression in human tumors. Appl Immunohistochem Mol Morphol. 2011;19(2):173-183.
57. Jennings L, Van Deerlin VM, Gulley ML. Recommended principles and practices for validating clinical
molecular pathology tests. Arch Pathol Lab Med. 2009;133(5):743-755.
58. Department of Health and Human Services. Medical Devices: Classification/reclassification of
immunochemistry reagents and kits. Fed Regist. 1998;63(106):30132-30142. Codified at 21 CFR 864.
Doc. No. 94P–0341.
59. Torlakovic EE, Riddell R, Banerjee D, et al. Canadian Association of Pathologists-Association
canadienne des pathologistes National Standards Committee/Immunohistochemistry: best practice
recommendations for standardization of immunohistochemistry tests. Am J Clin Pathol.
2010;133(3):354- 365.
60. Hsi ED. A practical approach for evaluating new antibodies in the clinical immunohistochemistry
laboratory. Arch Pathol Lab Med. 2001;125(2):289-294.
61. Dorfman DM, Bui MM, Tubbs RR, et al. The CD117 immunohistochemistry tissue microarray survey for
quality assurance and interlaboratory comparison: a College of American Pathologists Cell Markers
Committee study. Arch Pathol Lab Med. 2006;130(6):779-782.
62. Hanna W, O'Malley F P, Barnes P, et al. Updated recommendations from the Canadian National
Consensus Meeting on HER2/neu testing in breast cancer. Curr Oncol. 2007;14(4):149-153.

© 2015 College of American Pathologists. All rights reserved.

PAGE 26

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

63. Ferguson J, Chamberlain P, Cramer HM, Wu HH. ER, PR, and Her2 immunocytochemistry on celltransferred cytologic smears of primary and metastatic breast carcinomas: a comparison study with
formalin-fixed cell blocks and surgical biopsies. Diagn Cytopathol. 2013;41(7):575-581.
64. Gong Y, Symmans WF, Krishnamurthy S, Patel S, Sneige N. Optimal fixation conditions for
immunocytochemical analysis of estrogen receptor in cytologic specimens of breast carcinoma. Cancer.
2004;102(1):34-40.
65. Kumar SK, Gupta N, Rajwanshi A, Joshi K, Singh G. Immunochemistry for oestrogen receptor,
progesterone receptor and HER2 on cell blocks in primary breast carcinoma. Cytopathol.
2012;23(3):181- 186.
66. Nishimura R, Aogi K, Yamamoto T, et al. Usefulness of liquid-based cytology in hormone receptor
analysis of breast cancer specimens. Virchows Arch. 2011;458(2):153-158.
67. Pegolo E, Machin P, Riosa F, Bassini A, Deroma L, Di Loreto C. Hormone receptor and human
epidermal growth factor receptor 2 status evaluation on ThinPrep specimens from breast carcinoma:
correlation with histologic sections determination. Cancer Cytopathol. 2012;120(3):196-205.
68. Shabaik A, Lin G, Peterson M, et al. Reliability of Her2/neu, estrogen receptor, and progesterone
receptor testing by immunohistochemistry on cell block of FNA and serous effusions from patients with
primary and metastatic breast carcinoma. Diagn Cytopathol. 2011;39(5):328-332.
69. Adegboyega PA, Gokhale S. Effect of decalcification on the immunohistochemical expression of ABH
blood group isoantigens. Appl Immunohistochem Mol Morphol. 2003;11(2):194-197.
70. Arber JM, Arber DA, Jenkins KA, Battifora H. Effect of decalcification and fixation in paraffin- section
immunohistochemistry. Appl Immunohistochem. 1996;4(4):241-248.
71. Bussolati G, Leonardo E. Technical pitfalls potentially affecting diagnoses in immunohistochemistry. J
Clin Pathol. 2008;61(11):1184-1192.
72. Fend F, Tzankov A, Bink K, et al. Modern techniques for the diagnostic evaluation of the trephine bone
marrow biopsy: methodological aspects and applications. Prog Histochem Cytochem.. 2008;42(4):203252.
73. Torlakovic EE, Naresh K, Kremer M, van der Walt J, Hyjek E, Porwit A. Call for a European programme
in external quality assurance for bone marrow immunohistochemistry; report of a European Bone
Marrow Working Group pilot study. J Clin Pathol. 2009;62(6):547-551.
74. Wittenburg G, Volkel C, Mai R, Lauer G. Immunohistochemical comparison of differentiation markers on
paraffin and plastic embedded human bone samples. J Physiol Pharmacol. 2009;60(Suppl 8):43-49.
75. Dowsett M, Nielsen TO, A'Hern R, et al. Assessment of Ki67 in breast cancer: recommendations from
the International Ki67 in Breast Cancer working group. J Natl Cancer Inst. 2011;103(22):1656-1664.
76. Arber DA. Effect of prolonged formalin fixation on the immunohistochemical reactivity of breast markers.
Appl Immunohistochem Mol Morphol. 2002;10(2):183-186.
77. Djordjevic B, Hennessy BT, Li J, et al. Clinical assessment of PTEN loss in endometrial carcinoma:
immunohistochemistry outperforms gene sequencing. Mod Pathol. 2012;25(5):699-708.
78. Manion E, Hornick JL, Lester SC, Brock JE. A comparison of equivocal immunohistochemical results
with anti-HER2/neu antibodies A0485 and SP3 with corresponding FISH results in routine clinical
practice. Am J Clin Pathol. 2011;135(6):845-851.
79. Vaughan MM, Toth K, Chintala S, Rustum YM. Double immunohistochemical staining method for HIF1alpha and its regulators PHD2 and PHD3 in formalin-fixed paraffin-embedded tissues. Appl
Immunohistochem Mol Morphol. 2010;18(4):375-381.
80. Gazziero A, Guzzardo V, Aldighieri E, Fassina A. Morphological quality and nucleic acid preservation in
cytopathology. J Clin Pathol. 2009;62(5):429-434.
81. Hansen TP, Nielsen O, Fenger C. Optimization of antibodies for detection of the mismatch repair
proteins MLH1, MSH2, MSH6, and PMS2 using a biotin-free visualization system. Appl
Immunohistochem Mol Morphol. 2006;14(1):115-121.
82. Ikeda K, Tate G, Suzuki T, Mitsuya T. Comparison of immunocytochemical sensitivity between formalinfixed and alcohol-fixed specimens reveals the diagnostic value of alcohol-fixed cytocentrifuged
preparations in malignant effusion cytology. Am J Clin Pathol. 2011;136(6):934-942.
83. Kovacs A, Stenman G. HER2-testing in 538 consecutive breast cancer cases using FISH and
immunohistochemistry. Pathol Res Pract. 2010;206(1):39-42.
84. Takai H, Kato A, Ishiguro T, et al. Optimization of tissue processing for immunohistochemistry for the
detection of human glypican-3. Acta Histochem. 2010;112(3):240-250.

© 2015 College of American Pathologists. All rights reserved.

PAGE 27

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

85. Roepman P, Horlings HM, Krijgsman O, et al. Microarray-based determination of estrogen receptor,
progesterone receptor, and HER2 receptor status in breast cancer. Clin Cancer Res. 2009;15(22):70037011.
86. Wong SC, Chan JK, Lo ES, et al. The contribution of bifunctional SkipDewax pretreatment solution,
rabbit monoclonal antibodies, and polymer detection systems in immunohistochemistry. Arch Pathol Lab
Med. 2007;131(7):1047-1055.
87. Linderoth J, Ehinger M, Akerman M, et al. Tissue microarray is inappropriate for analysis of BCL6
expression in diffuse large B-cell lymphoma. Eur J Haematol. 2007;79(2):146-149.
88. Fitzgibbons PL, Murphy DA, Dorfman DM, et al. Interlaboratory comparison of immunohistochemical
testing for HER2: results of the 2004 and 2005 College of American Pathologists HER2
Immunohistochemistry Tissue Microarray Survey. Arch Pathol Lab Med. 2006;130(10):1440-1445.
89. Lin Y, Hatem J, Wang J, et al. Tissue microarray-based immunohistochemical study can significantly
underestimate the expression of HER2 and progesterone receptor in ductal carcinoma in situ of the
breast. Biotech Histochem. 2011;86(5):345-350.
90. Lourenco HM, Pereira TP, Fonseca RR, Cardoso PM. HER2/neu detection by immunohistochemistry:
optimization of in-house protocols. Appl Immunohistochem Mol Morphol. 2009;17(2):151-157.
91. Ricardo SAV, Milanezi F, Carvalho ST, Leitao DRA, Schmitt FCL. HER2 evaluation using the novel
rabbit monoclonal antibody SP3 and CISH in tissue microarrays of invasive breast carcinomas. J Clin
Pathol. 2007;60(9):1001-1005.
92. Nofech-Mozes S, Vella ET, Dhesy-Thind S, et al. Systematic review on hormone receptor testing in
breast cancer. Appl Immunohistochem Mol Morphol. 2012;20(3):214-263.
93. Engel KB, Moore HM. Effects of preanalytical variables on the detection of proteins by
immunohistochemistry in formalin-fixed, paraffin-embedded tissue. Arch Pathol Lab Med.
2011;135(5):537-543.
94. Tong LC, Nelson N, Tsourigiannis J, Mulligan AM. The effect of prolonged fixation on the
immunohistochemical evaluation of estrogen receptor, progesterone receptor, and HER2 expression in
invasive breast cancer: a prospective study. Am J Surg Pathol. 2011;35(4):545-552.
95. Ibarra JA, Rogers LW. Fixation time does not affect expression of HER2/neu: a pilot study. Am J Clin
Pathol. 2010;134(4):594-596.
96. Moher D, Liberati A, Tetzlaff J, Altman D. Preferred reporting items for systematic reviews and metaanalyses: the PRISMA statement. PLoS Med. 2009;6(7):e1000097.

© 2015 College of American Pathologists. All rights reserved.

PAGE 28

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

APPENDIX
Appendix- Figure 1: Literature Review Results
Adapted with permission from Moher et al.96

Records identified through database
searching
(n = 1,393)

Additional records identified through
other sources
(n =98)

fter duplicates removed (n =1,463)

cords screened (n
=1,463)
Full-text articles assessed
for eligibility
(n =268)
Studies included for
extraction and grading (n
= 126)

Records excluded
(n =1,195)
Full-text articles excluded
(n =142)

Data extraction articles
excluded*
(n = 43)

*Excluded based on expert opinion, did not meet minimum quality standards, presented incomplete data or data that were not in useable formats
© 2015 College of American Pathologists. All rights reserved.

PAGE 31

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix - Table 1. Hierarchies of Data Sources for Analytic Validation

2

Level 1
• Collaborative study using a large panel of well characterized samples
• Summary data from external proficiency testing schemes or inter-laboratory
comparisons
Level 2
• High quality peer-reviewed studies (see Table 2)
• Method comparisons
• Validation studies
Level 3
• Lower quality peer-reviewed studies (see Table 2)
• Expert panel reviewed FDA summaries
Level 4
• Unpublished and/or non-peer reviewed research, clinical laboratory, or manufacturer data
• Studies on performance of the same basic methodology, but used to test for
a different target
Reprinted by permission from Macmillan Publishers Ltd: Genetics in Medicine2, copyright 2009

© 2015 College of American Pathologists. All rights reserved.

PAGE 32

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays
Appendix - Table 2. Criteria for Assessing Quality of Individual Analytic Validation Studies (internal validity)

1. Adequate descriptions of the index test (test under evaluation)
• Source and inclusion of positive and negative control materials
• Reproducibility of test results
• Quality control/assurance measures
2. Adequate descriptions of the referent test
• Specific methods/platforms evaluated
• Number of positive samples and negative controls tested
3. Adequate descriptions of the basis for the “right answer”
• Comparison to a “gold standard” reference test
• Consensus (e.g., external proficiency testing)
• Characterized control materials (e.g., National Institute of Standards
and Technology, sequenced)
4. Avoidance of biases
• Blinded testing and interpretation
• Specimens represent routinely analyzed clinical specimens in all aspects
(e.g., collection, transport, processing)
• Reporting of test failures and uninterpretable or indeterminate results
5. Analysis of data
• Point estimates of analytic sensitivity and specificity with 95% confidence intervals
• Sample size and power calculations addressed
Reprinted by permission from Macmillan Publishers Ltd: Genetics in Medicine2, copyright 2009

© 2015 College of American Pathologists. All rights reserved.

PAGE 33

2

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix –Table 3. Summary data on comparisons of concordance between IHC tests for HER2

c

IHC 2

% Conc
(95% CI)

Kappa,
McNemars

Pathway Her2/neu, 4B5

Pathway Her2/neu, CB11

436/467

93.4
(91-95)

FFPE
WS

Ventana, 4B5

Ventana, SP3

134/146

31

FFPE
WS

Oracle Auto,
CB11

Hercep
Test

32

FFPE
b
WS

Oracle Auto

FFPE
WS

Ventana 4B5

IHC 1

Van der Vegt,
34
2009

FFPE
TMA

22

Studies

Boers, 2011

Moelans, 2010

O’Grady, 2010

Mayr, 2009

30

2x2 Table

3x3 Table
Concordance

Sample
a
Type

e

Grade

0.75,
<0.001

100

Fair

92.0
(86-95)

0.66,
0.002

100

Fair

195/219

89.0
(84-93)

0.78
<0.001

100

Good

Hercep
Test

386/445

86.7
(83-90)

0.77,
<0.001

100

Fair

Dako,
Hercep
Test

96/130

73.8
(66-81)

0.60,
0.004

97.1

Fair

a

All breast cancer except O’Grady, 2010. b Gastroesophageal tumor. c Scoring system is 3+ positive, 2+ equivocal, 0-1+ negative; calculation
of overall concordance by addition of 3 cells on the major diagonal / total N. d Recalculation of concordance after excluding all 2+ cells. e
Quality grade for individual studies.
Abbreviation: Conc=concordance; FFPE=Formalin-fixed paraffin embedded; TMA= tissue microarray; WS=whole section

© 2015 College of American Pathologists. All rights reserved.

d

(minus 2+)
% Conc

PAGE 34

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix –Table 4. Summary data on concordance estimates from comparisons between HER IHC and in situ hybridization tests
Tissue
Sample
type

Data
b
Analysis

Conc
cellsc /
Total N

2x2d
[a]

[b]

[c]

[d]

Conc
(%)

Grade

IHC

ISHa

Van der
Vegt, 200934

Fair

Her2/neu,
CB11

FISH

BrCa
TMA

2x2

444/473

62

17

12

382

94.0

Van der
Vegt, 200934

Fair

Her-2/neu,
4B5

FISH

BrCa
TMA

2x2

436/466

54

4

19

389

Powell 2007,
(Site 1+2)49

Fair

Pathway
Her-2 CB11 FISH

2x2

279/322

149

13

30

Powell 2007,
(Site 1+2)49

Fair

Her-2
Benchmark
auto, 4B5

FISH

BrCa WS

2x2

288/322

155

27

7

Moelans,
201031

Fair

HercepTest
Manual

CISH

BrCa WS

3x3

Moelans,
201031

Fair

Oracle Auto, CISH
CB11

BrCa WS

Grimm,
201042

Fair

HER2 LDT,
4B5

FISH

Boers,
201122

Good

Ventana,
4B5
4B5

Boers,
201122

Good

Hofmann,
200843
Sornmayura
201244

Study

Conc
95%
Kf

McNemars p

91-96

0.77

0.46

93.6

91-95

0.80

0.003

77

86.6

82-90

0.73

0.015

133

89.4

85-92

0.79

0.001

183/219

85.8

80-90

0.72

0.001

3x3

183/219

83.6

78-88

0.66

0.004

BrCa WS

3x3

457/697

--

65.6

62-69

0.37

<0.001

SISH

GI Ca
WS

143/146

21

2

1

122

98.0

94-99

0.92

1.00

2x2

Ventana,
SP3
SP3

SISH

141/146

17

0

5

124

96.6

92-99

0.85

0.07

Poor

HercepTest
Manual
Manual

FISH

11

139

93.5

88-96

0.73

0.003

Fair

HER2 LDT,
4B5
4B5

FISH

GI Ca
WS
Gast Ca
WS
Gast Ca
WS

19

156

87.7

82-92

0.49

0.008

BrCaWS

2x2
2x3

157/168

2x2

171/195

18
15

--

0
5

a

CI

ISH = In situ hybridization. b Data entered into 2x2 or 3x3 contingency tables. c Conc=concordant cells. d Cells in a 2x2 table. e Conc=concordance.
k=Kappa statistic.
Abbreviation: IHC=immunohistochemistry; BrCa=breast cancer; GICa= gastrointestinal cancer; Gast Ca= gastric cancer; TMA=tissue microarray; WS= whole
section

f

© 2015 College of American Pathologists. All rights reserved.

PAGE 35

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix – Table 5. Summary data on comparisons of concordance between IHC and alternative referent tests

Study

Grade

IHC

Referent

Tissue

Data
Analysi
sa

Conc
cellsb /
Total
N

2x2
c

[b
]

[c]

Conc
d
(%)

[a]

LehmannChe,
201146

Fair

Benchmar
k HER2

QRT-PCR,
panel consensus

BrCa

3x3

444/446

Jordan,
201245

Fair

p16

QRT-PCR p16,
HPV quant
PCR, HPV ISH

OSCC

2x2

204/233

141

24

5

Fair

Anti-BCG

TB diagnosis

Pleural bx

2x2

31/36

20

0

5

Fair

HercepTest
HER2

Consensus

BrCa WS

3x3

65/90

Baba,
200840
Dowsett,
200720

[d]

Con
c
95%
CI

McNemar
sp

95.3

93-97

0.87

0.87

62

87.5

83-91

0.72

0.72

11

86.1

71-94

0.71

0.71

72.2

62-80

0.56

0.56

a

Data entered into 2x2 or 3x3 contingency tables. b Conc=concordant cells. c Cells in a 2x2 table. d Conc=concordance.
k=Kappa statistic
Abbreviation: IHC=immunohistochemistry; BrCa=breast cancer; OSCC=Oropharyngeal squamous cell carcinoma; bx=biopsy; WS= whole section

e

© 2015 College of American Pathologists. All rights reserved.

Ke

PAGE 36

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix – Table 6. Considering the characteristics of validation sets with different numbers of
samples1

© 2015 College of American Pathologists. All rights reserved.

PAGE 37

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix – Table 7. Summary data on concordance between ER IHC performed on cytology samples and histologic sections

Study
Gong,
64

2004
Kumar,

N
Pos

N
Neg

Total
N

Tissue

32

15

47

BrCa

65

20

30

50

BrCa

66

66

16

82

BrCa

63

22

16

38

67

85

16

101

21

18

39

2011
Nishimura,
2011
Ferguson,
2012
Pegolo,
2012
Shabaik,
68
2012

d

BrCa

f

BrCa

h

BrCa

Comparator

Referent

Cytologic

Histologic
sections
Histologic
sections

Concordance
(95% CI)
91%
(79-97)
90%
(78-96)

Histologic
sections

a

smears
FNA cell
b

block
PreserveCy
t
FNA
e

Smears
Cytolyt
ThinPrep
FNA cell
g
block

kappa

McNemar’s
p

0.79

0.13

0.79

0.37

98%
(91-99)

0.93

0.48

97%
100%

Histologic
sections

97%
(85-99)

0.95

1.0

95%
100%

Tissue sections

98%
(93-99)

0.92

0.48

100%
87%

Tissue sections

92%
(79-98)

0.85

0.25

86%
100%

a

Abbott method (10% formalin-methanol-acetone -20C); no antigen retrieval. Addition of AR improved intensity without increasing false positives.
10% buffered formalin overnight.
c
FNA immediately into PreserveCyt Solution, ThinPrep slides
d
38/47 (81%) had ≥ 50 cells
e
FNA on alcohol fixed direct smears using cell transfer technique
f
101/111 (91%) assessable
g
FNA/serous effusions FFPE cell blocks
h
39/42 (93%) assessable
Abbreviation: IHC=immunohistochemistry; BrCa=breast cancer
b

© 2015 College of American Pathologists. All rights reserved.

PAGE 38

Pos/Neg
conc
89%
100%
80%
97%

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix – Table 8. Summary data on concordance between PgR IHC performed on cytology samples and histologic sections

N
Pos

N
Neg

Total
N

Tissue

65

17

33

50

BrCa

66

58

24

82

BrCa

63

19

23

42

c

BrCa

67

75

24

99

15

24

39

Study
Kumar,
2011
Nishimura,
2011
Ferguson,
2012
Pegolo,
2012
Shabaik,
68
2012

e

BrCa

f

BrCa

Comparator

Referent

FNA cell

Histologic
sections
Histologic
sections

Concordance
(95% CI)
94%
(83-99)
95%
(88-98)

Histologic
sections

a

block
PreservCyt/
b

ThinPrep
FNA

d

Smears
Cytolyt
ThinPrep
FNA cell
g
block

kappa

McNemar’s
p

0.86

1.0

0.88

0.62

95%
(83-99)

0.90

0.48

89%
100%

Tissue sections

91%
(83-95)

0.76

0.50

92%
87%

Tissue sections

92%
(79-98)

0.83

0.25

80%
100%

a

10% buffered formalin overnight
Immediately into PreserveCyt Solution, ThinPrep slides
c
42/47 (89%) had ≥ 50 cells
d
FNA on alcohol fixed direct smears using cell transfer technique
e
99/111 (89%) assessable
f
39/42 (93%) assessable
g
FNA/serous effusions FFPE cell blocks
Abbreviation: IHC=immunohistochemistry; BrCa=breast cancer
b

© 2015 College of American Pathologists. All rights reserved.

PAGE 39

Pos/Neg
conc
88%
97%
95%
96%

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix – Table 9. Summary data on concordance between HER2 IHC performed on cytology samples and histologic
sections

Study
Kumar,
201165
Pegolo,
201267

N
3+

N 2+

N
Neg

12

NR

38

9

NR

91

Total N

Tissue

Comparator

Referent

50

BrCa

FNA cell blocka

100b

BrCa

Cytolyt ThinPrep

Histologic
sections
Tissue
sections

Concordance
(95% CI)
90%
(78-96)
100%
(96-100)

kappa

McNemar’s p

0.75

0.37

1.0

NS

†

3x3 contingency table
10% buffered formalin overnight
b
100/111 (90%) assessable
c
Conc=concordance
Abbreviation: IHC=immunohistochemistry; BrCa=breast cancer
a

40© 2015 College of American Pathologists. All rights reserved.

PAGE 40

Pos/Neg
concc
92%
89%
100%
100%

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

a

Appendix-Table 10. Summary data on concordance between IHC performed on whole sections and TMA

Study

Marker

Graham, 2008
Jones, 2012

25

28

Warnberg, 2008
Fons, 2006

35

24

Soiland, 2008
23

Drev, 2008

26

Gulbahce, 2012
Kwon, 2009

29

Henriksen, 2007

27
23

Drev, 2008 (pilot)
Thomson, 2009

33

Batistatou, 2013

21

Concordance (%)
between WS & TMA

McNemars p

Study Grade

kappa

HER2

BrCa

73.1

0.56

<0.001

Fair

CK19

Thyroid ca

83.1

0.17

0.03

Poor

ER

BrCa

84.2

0.65

0.70

Fair

ER

Endometrioid

89.5

0.78

0.13

Fair

BrCa

89.9

0.74

<0.001

Fair

HER2

BrCa

91.7

0.71

<0.001

Fair

ER

BrCa

94.5

0.85

0.30

Poor

CD34

GIST

95.5

0.93

NR

Fair

ER

BrCa

96.4

NR

NR

Poor

HER2

BrCa

96.9

0.90

0.56

Fair

ER

BrCa

98.7

NR

NR

Poor

HER2

BrCa

100.0

1.0

Not sig

Fair

Androgen
receptor

36

Tissue

Median = 93.1%
a

To avoid bias in the overall concordance range and median value related to a sample set being tested for multiple markers or for multiple TMA core sizes, the, comparisons were reduced
from 12 in this table. Only one comparison was included from each sample set. When multiple core sizes were reported, 0.6 mm cores were selected. When multiple markers were
reported, the selection order was ER/PR, HER2 and then the most common marker.
Abbreviation: IHC=immunohistochemistry; BrCa=breast cancer; GIST=gastrointestinal stromal tumor

© 2015 College of American Pathologists. All rights reserved.

PAGE 41

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix- Table 11. Summary data on whole section versus TMA, stratified by IHC marker

Marker

Number of
studies

Tissue

Concordance Range

Median Concordance
between WS & TMA

Concordance >90%

ER

5 of 6

5 BrCa, 1
endometrioid

84.2 – 98.7

5 BrCa = 95.4%
th
6 , k=0.97

67%

PR

4 of 5

4 BrCa, 1
endometrioid

81.5 – 92.6

4 BrCa = 90.8%
th
5 , k=0.90

60%

HER2 IHC

6

BrCa

73.1 - 100

92.6%

67%

HER2 FISH

2

BrCa

NA

98.6%

100%

Comparisons of overall concordance between whole sections and TMA for ER and PgR from an earlier systematic review were 97% and 93%.92
Abbreviation: BrCa= breast cancer; IHC=immunohistochemistry; TMA=tissue microarray; WS = whole section; NA= not applicable

© 2015 College of American Pathologists. All rights reserved.

PAGE 42

Supplemental Digital Content: Principles of Analytic Validation for IHC Assays

Appendix- Table 12. Summary data on whole section versus TMA, stratified by TMA core size

Core sizes

Number of
studies

Concordance
Range

Concordance
>90%

Median Concordance
between WS & TMA

0.6

17

73.1 – 98.7

92.1%

1.0 – 2.0

8

80.4 - 100

92.2%

3.0

10

74.6 – 96.4

92.5%

59%
50%
60%

*These proportions are not statistically different (p >0.5; Fisher’s exact test)
Abbreviation: TMA=tissue microarray; WS = whole section

Appendix- Table 13. Available data on other markers tested on whole sections versus TMA samples

Number of
studies

Tissue

0.6 mm
Cores

2.0 mm
Cores

3.0 mm
Cores

1

BrCa

--

--

--

1

BrCa

95.5%

92.5%

89.5%

1

GIST

74.6%

86.6%

94.0%

CK19

1

Thyroid ca

80.4%

83.1%

--

HBME1

1

Thyroid ca

92.9%

95.0%

--

1

BrCa

--

--

--

1

GIST

74.6%

86.6%

94.0%

1

Endometrioid

--

--

--

1

GIST

74.6%

77.6%

92.5%

Marker
Androgen
receptor
CD 34

Ki-67

P53

Abbreviation: TMA = tissue microarray; BrCa=breast cancer; GIST=gastrointestinal stromal tumor

© 2015 College of American Pathologists. All rights reserved.

PAGE 43

Objectives

Principles of Analytic
Validation of
Immunohistochemistry
Assays

• Apply evidence-based guidelines to ensure each
Immunohistochemistry (IHC) assay is validated prior to
reporting on patient samples
• Recognize the requirements for revalidation
• Understand possible differences in validation requirements
based on variations in fixative or specimen type
• Understand how the quality of evidence impacts the
recommendations related to the validation statements

Published : Archives of Pathology and
Laboratory Medicine

March 19, 2014

Pathology and
Laboratory Quality
Center

2

Introduction

Background

• Laboratories are required to validate all assays before testing
patient specimens.
• There is significant variation in validation practices for IHC
assays.
• Current guidelines exist only for HER2 and ER/PgR.

3

Validation Practices –
Non Predictive Factor Assays
Procedures
Lab has written validation

Validation Practices Non Predictive Factor Assays
Yes

No

68%

28%

Procedures

procedure?
Procedure specifies # validation

54%

44%

46%

46%

Yes

No

Change in antigen retrieval method?

71%

25%

Change in detection method?

74%

23%

Change in instrumentation?

74%

24%

Change in fixative?

65%

30%

cases?
Procedure specifies when
revalidation needed?
Cytology specimens addressed?
37%
63%
Hardy et al. Arch Pathol
Lab Med 2013;137:19-25

Hardy et al. Arch Pathol Lab Med 2013;137:19-25

Principles of Analytic Validation for IHC Assays:
Expert and Advisory Panel

Introduction

•

Chair:
Patrick Fitzgibbons, MD

CAP convened expert and advisory panels to
systematically review published data and develop

Expert Panel Members

evidence-based recommendations
•

Randa Alsabeh, MD

Closely followed IOM Clinical Practice Guidelines
o

Transparency

o

Manage conflicts of interest

o

Multidisciplinary panel

o

Patient advocate (N/A for this panel)

o

Systematic Review

o

Regan Fulton, MD, PhD
Jeffrey Goldsmith, MD

Advisory Panel Members:
Raouf Nakhleh, MD, Center
Richard Brown, MD
Richard Eisen, MD
Hadi Yaziji, MD

Thomas Haas, DO
Rouzan Karabakhtsian, MD, PhD

Staff:

Patti Loykasek, HT(ASCP)QIHC

Lisa Fatheree, SCT(ASCP)

Monna Marolt, MD

Tony Smith, MLS

Steven Shen, MD, PhD

Considered judgment

Paul Swanson, MD

Consultant Methodologist:
Linda Bradley, PhD

7

Systematic Evidence Review

Introduction

• Identify Key Questions

•

Overarching questions:

• Literature search
1.

• Data extraction

What is needed for initial analytic assay validation before
placing any immunohistochemical test into clinical service?

• Develop proposed recommendations
• Open comment period

2.

• Considered judgment process

What are the revalidation requirements?

9

10

Scope Questions

Scope Questions (cont.)

1.

2.

When and how should validation assess

What is the minimum number of positive and negative
cases needed to analytically validate an IHC assay for its

• analytic sensitivity

intended use(s)?

• analytic specificity

• Non-predictive markers

• accuracy (assay concordance)

• Predictive markers

• precision (inter-run and inter-operator variability)?

• Identifying infectious organisms
• Rare antigens

Should expression levels be specified for positive cases?

11

12

Scope Questions (cont.)

Scope Questions (cont.)

3.

4.

What parameters should be specified for the tissues used
in the validation set?

How do the following preanalytic variables influence
analytic validation?
• Type of fixative

• Cytology specimens

• Type of decalcification solution

• Minimum tissue size or minimum quantity of cells

• Time in decalcification solution
• Validation tissues processed in another laboratory

• Neoplastic vs. non-neoplastic tissues

5.

What conditions require assay revalidation?

13

14

Systematic Evidence Review

Systematic Evidence Review

• Literature search

• Evidence Evaluation
o Quality (rate strength of evidence)

o January 2004 – May 2013
o Quantity
o 1,463 studies met inclusion criteria
o Consistency

→ Reviewed by panel

o 126 studies identified for full data extraction

15

16

Quality Assessment
•

Grades for Strength of Evidence

Individual studies graded on specific criteria by the

Grade
Convincing

methodology consultant (LAB)
•

Criteria included:
o Quality and execution of studies

Adequate

o Quantity of data (number and size of studies)
o Consistency and generalizability of the evidence across studies

Inadequate

• Adequate descriptions of the test
• Adequate descriptions of the basis for the “right answer”

Description
Level 1 or 2 studies with an appropriate number and
distribution of challenges and reported consistent and
generalizable results.
Level 1 or 2 studies that lacked the appropriate number and
distribution of challenges OR were consistent but not
generalizable.
Combinations of Level 1 or 2 studies that show unexplained
inconsistencies, OR one or more lower quality studies (Level
3 or 4), OR expert opinion.

• Reproducibility of test results
Level 1: Collaborative study using a large panel of well-characterized samples; summary data from external
proficiency testing schemes or inter-laboratory comparisons
Level 2: High quality peer-reviewed studies
Level 3: Lower quality peer-reviewed studies OR expert panel reviewed FDA summaries
Level 4: Unpublished or non-peer reviewed data

• Avoidance of biases
• Analysis of data

17

Grades for Strength of Recommendation
Designation
Strong
Recommendation

Recommendation

Expert Consensus
Opinion

18

Systematic Evidence Review

Rationale
Strength of evidence is Convincing based on
consistent, generalizable, good quality evidence;
further studies are unlikely to change the
conclusions

• Open comment period (July 2013):
o 18 draft recommendations and 5 methodology questions

Strength of evidence is Adequate based on
limitations in the quality of evidence; further studies
may change the conclusions

o 263 respondents; 1,037 comments

Important validation element to address but strength
of evidence is Inadequate; gaps in knowledge may
require further studies

19

20

Open Comment Period

Systematic Evidence Review

100

• Considered judgment process

90

o Panel reviews and considers

80

%
Agreement

70

• Feedback

60

• Quality/quantity/consistency of evidence

50

• Benefits/harms

40

• Value versus cost/burdens

30

• Regulatory requirements
• Expert opinion

20
10
0
1

2

3

4

5

6

7

8

9

10 11 12 13 14 15 16 17 18

o 14 final recommendations

Original Draft Proposed Recommendation
Final Recommendations Combined/Condensed into 14 Total
22

ASCO/CAP HER2 Guideline
Recommendations Summary of Changes

ASCO/CAP HER2 Guideline Recommendations
Summary of Changes

Initial Test Validation

2007
25–100 samples

Concordance

2013
20(+), 20(-) for FDA-approved assays
40(+), 40(-) for LDTs
Not applicable if assay was previously validated
and lab has successful PT performance

2007
If <95% for any result
category, cases with that
test result must be
automatically reflexed to
alternative method

2013
Specific concordance requirements are
not required
Laboratories must comply with
accreditation and PT requirements

The Guidelines

Guideline 1
Recommendation: Laboratories must validate all
immunohistochemical tests before placing into clinical service.
Note: Such means include (but are not necessarily limited to):
• Correlating the new test’s results with the morphology and expected
results;
• Comparing the new test’s results with the results of prior testing of
the same tissues with a validated assay in the same laboratory;
• Comparing the new test’s results with the results of testing the same
tissue validation set in another laboratory using a validated assay;
• Comparing the new test’s results with previously validated nonimmunohistochemical tests; or
• Testing previously graded tissue challenges from a formal proficiency

© 2015 College of American Pathologists. All rights reserved.

testing program (if available) and comparing the results with the
25

26

graded responses.

Guideline 1

Rationale 1

• Strength of Evidence:

o

• Validation set should include:

Adequate to support when analytic validation should be done and
that it should include determination of concordance and precision

o Inadequate to assess how validation should be done with regard to

o Positive, negative, and low positive tissues
o Should not be all normal tissues
o Should reflect the intended use of the assay

the listed approaches, but did show that these approaches have
been used.

• Positive and negative cell types on the same section could be

• Rationale: Analytic validation provides a net benefit for the
overall performance and safety of IHC tests by contributing to the

used as separate challenges

avoidance of potential harms related to analytic false positive and
false negative test results.

27

28

Guideline 2

Guideline 2

Recommendation: For initial validation of every assay used

•

clinically (with the exception of HER2, ER and PgR, for which

Strength of evidence
o Adequate to support a 90% (versus 95%) overall concordance

established validation guidelines already exist), laboratories

benchmark for analytic validation of IHC tests (except HER2, ER,

should achieve at least 90% overall concordance between the

PgR)

new test and the comparator test or expected results. If

•

concordance is less than 90%, laboratories need to investigate

Median overall concordance in a two-year inter-laboratory
comparison of CD117 IHC and target results was 87.6% (Hsi,

the cause of low concordance.

2001)

•

Median overall concordance in 5 comparisons of different
HER2 IHC tests was 89.0% (range 74–92%), with 2 of 5
studies >90% concordant. (Boers, 2011; Mayr, 2009; Moelans, 2010; O’Grady, 2010; van der
Vegt, 2009)

29

30

Guideline 2 continued
•

Guideline 3
Expert Consensus Opinion: For initial analytic validation of

Median overall concordance in 5 comparisons of HER2 IHC

non-predictive factor assays, laboratories should test a

tests to HER2 ISH tests was 88.2% (range 66– 94%), with 2

minimum of 10 positive and 10 negative tissues. When the

of 5 comparisons >90% concordant (Dorfman, 2006; Jordan, 2012; Lotan, 2011;
•

Phillips 2007)

laboratory medical director determines that fewer than 20

Median overall concordance in 6 comparisons of IHC tests

validation cases are sufficient for a specific marker (e.g., rare

(PTEN, ER, PR, HER2, MPT64, p16) to alternative referent

antigen), the rationale for that decision needs to be

tests (e.g., RNA expression, clinical diagnosis) was 91.4%

documented.

(range 74–99%), with 3 of 6 studies >90% concordant (Phillips,
2007; Baba, 2008, Lehmann-Che, 2011)

o Note: The validation set should include high and low expressors for
positive cases when appropriate, and should span the expected
range of clinical results (expression levels) for markers that are
reported quantitatively.
© 2014 College of American Pathologists. All rights reserved.

31

32

Guideline 3

Validation Using 10 and 20 Tissue Validation
Sets against a 90% Concordance Benchmark

• Strength of Evidence
Concordance estimate (95% CI)

o Inadequate to support the recommended number of validation
samples.

o Adequate to support the distinction between non-predictive and
predictive IHC tests and the use of different numbers.

# of
validation
tissues

0 discordant

1 discordant

2 discordant

10

100% (68-100)

90% (57-100)

80% (48-95)

20

100% (81-100)

95% (75-100)

90% (69-98)

Concordance estimates with 95% confidence intervals stratified by number of observed
discordant samples

33

Guideline 4

34

Guideline 4

Expert Consensus Opinion: For initial analytic validation of all
o Strength of Evidence

laboratory-developed predictive marker assays, laboratories
should test a minimum of 20 positive and 20 negative tissues.

• Inadequate to support the recommended number of

When the laboratory medical director determines that fewer

validation samples.

than 40 validation tissues are sufficient for a specific marker,
the rationale for that decision needs to be documented.

• Adequate to support the distinction between non-predictive
and predictive IHC tests and the use of different numbers.

o Note: Positive cases in the validation set should span the expected
range of clinical results (expression levels). This recommendation
does not apply to any marker for which a separate validation
guideline already exists.

35

36

2x2 contingency table of a 40 tissue validation set that did not
meet the benchmark (results entered into a 2x2 contingency
table) with associated statistical tests

(95% CI)
Validation Using aConcordance
40 Tissue estimate
Validation
Set (20
# of
1
2
3
4
Positive
and0 20 Negative)
against
a 90%
validation discordan discordan discordan discordan discordan
Concordance
Benchmark
tissues
t
t
t
t
t
20

100% (81100)

40

100% (90100)

95% (75100)

New IHC
Result
Positive
Negative

90% (69-98) 85% (63-96) 80% (58-92)

97.5% (86- 95% (83-99) 92.5% (79- 90% (76-97)
100)
98)

Referent
Result
Positive

Referent
Result
Negative

15
5
20

0
20
20

15
25
40

Overall concordance: 35/40=87.5% (does not meet 90% benchmark)
Kappa: 0.75
McNemar’s p: 0.13, not significant
Positive concordance: 15/20 = 75%
Negative concordance: 20/20 = 100%

Concordance estimates with 95% confidence intervals stratified by number of observed discordant samples

37

Guideline 5

38

Guideline 6
Recommendation: When possible, laboratories should use

Recommendation: For a marker with both predictive and

validation tissues that have been processed using the same

non-predictive applications, laboratories should validate it

fixative and processing methods as cases that will be tested

as a predictive marker if it is used as such

clinically.

•

Strength of evidence:

•

o Adequate to support the use of the higher validation standard (e.g.,

Strength of evidence
o Adequate to support that laboratories should, whenever

number of samples) in the case of a marker with both non-

possible, use the same fixative and processing methods as

predictive and predictive intended uses.

cases tested clinically, in order to validate using representative
specimens.

39

40

Guideline 6
•
•

Guideline 7

Can be difficult in reference laboratories that receive

Expert Consensus Opinion: If IHC is regularly done on

tissues with disparate fixation protocols

cytologic specimens that are not processed in the same
manner as the tissues used for assay validation (e.g., alcohol-

Focused validation with a small number of markers may be

fixed cell blocks, air-dried smears, formalin post-fixed

appropriate

specimens), laboratories should test a sufficient number of
such cases to ensure that assays consistently achieve
expected results. The laboratory medical director is responsible
for determining the number of positive and negative cases and
the number of predictive and non-predictive markers to test.

42

41

Guideline 7

Guideline 8

• Strength of evidence

Expert Consensus Opinion: If IHC is regularly done on

o Inadequate to address the criteria and number of samples needed
for validation with cytology specimens.

decalcified tissues, laboratories should test a sufficient
number of such tissues to ensure that assays consistently

• Focused validation on representative antibodies used on

achieve expected results. The laboratory medical director is

cytologic specimens would be appropriate

responsible for determining the number of positive and
negative tissues and the number of predictive and non-

• A disclaimer in the report (especially in the case of negative

predictive markers to test.

results) may be appropriate if assays cannot be feasibly
validated:
o “Immunohistochemistry on cytologic specimens has not been
sufficiently validated; these results should be interpreted with
caution.”
43

44

Guideline 8

Guideline 9

• Strength of evidence:

Recommendation: Laboratories may use whole sections,

o Inadequate to address the criteria and number of samples

tissue microarrays (TMAs) and/or multitissue blocks

needed for validation with decalcified specimens.

(MTBs) in their validation sets as appropriate. Whole

• Focused validation on representative antibodies used on

sections should be used if TMAs/MTBs are not appropriate

decalcified specimens would be appropriate

for the targeted antigen or if the laboratory medical director

• A disclaimer in the report (especially in the case of negative

cannot confirm that the fixation and processing of TMAs/

results) may be appropriate if assays cannot be feasibly

MTBs is similar to clinical specimens.

validated (ANP.22985)

45

Guideline 9

46

Revalidation Secondary to Assay Modification
Antibody Specific:
1. Least:

• Strength of evidence

•

o Adequate to support TMA usage; however there are many

New antibody Lot

All Assays (one tier):
•
•

o
o
o

2. Moderate:

variables to be considered and thorough validation is needed
for each marker.

•

o Inadequate to recommend the routine use of TMA samples.

•

Antibody dilution
Antibody vendor
(same clone)

• TMAs / MTBs can be very useful in many circumstances.
Beware of:

•

•
•

Antibody incubation or

o Proteins with high levels of heterogeneity (gastric Her2)

antigen retrieval times

o Limited tissue expression (e.g. bcl-6)

(same A.R. method)

Fixative type
Antigen retrieval method

•

pH change
buffer type
heat type

Antigen detection system
Tissue processing
equipment
Environmental conditions
o
o

location
water supply

3. Most:
•

New antibody clone

47

48

Evidence for Revalidation Guidelines 10-13

Guideline 10
Expert Consensus Opinion: When a new reagent lot is
placed into clinical service for an existing validated assay,

• Strength of evidence

laboratories should confirm the assay’s performance with

o Inadequate to address conditions requiring assay revalidation
and whether revalidation should be the same as initial

at least 1 known positive case and 1 known negative

validation.

case.
•

Laboratories may want to include low-expressors,
especially with predictive markers

49

Guideline 11

50

Guideline 12

Expert Consensus Opinion: Laboratories should confirm assay

Expert Consensus Opinion: Laboratories should confirm assay

performance with at least 2 known positive and 2 known

performance by testing a sufficient number of cases to ensure

negative cases when an existing validated assay has changed

that assays consistently achieve expected results when any of

in any one of the following ways:

the following have changed:

• Antibody dilution

• Fixative type

• Antibody vendor (same clone)

• Antigen retrieval method (e.g., change in pH, different buffer,
different heat platform)

• Incubation or retrieval times (same method)

• Antigen detection system
•

Laboratories may want to include low-expressors, especially with
predictive markers

• Tissue processing or testing equipment
• Environmental conditions of testing (e.g. laboratory relocation)
• Laboratory water supply

51

52

Guideline 12

Guideline 13

The laboratory medical director is responsible for determining
how many predictive and non-predictive markers and how many
positive and negative tissues to test.

Expert Consensus Opinion: Laboratories should run a full
revalidation (equivalent to initial analytic validation) when
the antibody clone is changed for an existing validated
assay.

• Reasonable approach:
o Selection of antibodies from menu with:
– Variable clinical uses (predictive and non-predictive)
– Variable antigen localizations
– Variable antibody types (monoclonal / polyclonal, etc.)

53

54

Guideline 14

Summary

Expert Consensus Opinion: The laboratory must document

• Physicians and patients rely on accurate diagnostic and
prognostic testing in the clinical laboratory.

all validations and verifications in compliance with
regulatory and accreditation requirements.

• Analytic validation is essential to ensuring that an assay
performs as expected, accurately identifies and/or quantifies
the targeted analyte, and minimizes the chances of false
positive or false negative results.
• Established guidelines are important to improve the
reproducibility and consistency of the test results.

55

56

References

Disclaimer
IHC Validation Teaching PowerPoint Copyright

Early Online Release March 19, 2014
Effective March 19, 2014
Copyright of the line-by-line text and the teaching PowerPoint of
the Principles of Analytic Validation of Immunohistochemistry Assays
belongs to CAP.

Archives of Pathology and Laboratory Medicine
http://www.archivesofpathology.org/doi/pdf/10.5858/arpa.2013-0610-CP

Permission to reprint manuscript guidelines text for any purpose (e.g.,
educational or commercial) requires written permission by Archives
The guideline recommendations must be reproduced without modification,
edits or changes to text.

57

58

FAQs

Topic:
Date:

Principles of Analytic Validation of Immunohistochemical Assays
April 22, 2014

Why is this guideline needed? Is there any evidence that patients have been harmed by
incorrect immunohistochemistry tests?
There is ample evidence that improper immunohistochemistry (IHC) tests have led to patient
harm. In perhaps the best documented example, nearly 400 of 1,000 breast cancers tested in one
laboratory in Newfoundland from 1997-2005 initially classified as ER negative were subsequently
found to be ER positive. Because of the incorrect test results, these patients did not receive
appropriate therapy and more than 100 died. A governmental inquiry determined that the high
error rate was due to improper testing practices. The American Society of Clinical Oncology
(ASCO) and the College of American Pathologists (CAP) guideline for hormone receptor and
HER2 testing in breast cancer were a direct result of well documented testing inaccuracies.
How will the guideline be enforced? What happens if a laboratory doesn’t follow the
guideline?
As with any clinical evidence-based guideline they are not mandatory. These recommendations
may be incorporated into future versions of the CAP Laboratory Accreditation Program (LAP)
Checklist; however, they are not currently required by LAP or any regulatory or accrediting
agency. It is encouraged that laboratories adopt these evidence-based recommendations.
When validating an estrogen receptor (ER) assay, must we use only breast cancers for
validation tissues?
No. Since ER is most frequently used to assess eligibility for hormonal therapy in patients with
breast cancer, positive and negative breast cancers should comprise at least part of the validation
set, but other ER positive and negative tissue types could be included.
How do these recommendations apply to assays for pathogen-specific antigens (e.g.,
Helicobacter pylori)?
Assays for infectious organisms are similar to predictive marker assays in that the results can
directly influence patient treatment, but selection of validation sets can be quite challenging when
the organism is rarely encountered. The option of using normal tissues for positive cases is also
not applicable. For selected organisms, including H. pylori, Cryptococcus spp, cytomegalovirus
and herpes simplex I/II, histologic features may be sufficiently characteristic to provide “expected”
positive cases for validation purposes, but for true analytic validation, concurrent culture evidence
of specific infection or either retrospective or prospective molecular confirmation of the formalin
fixed paraffin embedded sample may be required.
For rare antigens, do laboratory directors have the flexibility to use fewer validation
samples as they deem appropriate?
Yes. Following public comment and independent peer review of the draft recommendations, it
was determined that the guideline should not be too prescriptive and that the medical director
must have the discretion to modify the recommended steps in cases where it is not possible to
gather a full validation set. Several of the final recommendations include the caveat that the
laboratory medical director may decide that fewer cases are sufficient for a specific marker (e.g.,
rare antigen); however the rationale for that decision needs to be documented. If the laboratory is
unable to find sufficient cases to provide reasonable confidence that test results are valid, the
director is responsible for the decision to offer that test.

FAQs

Are normal tissues prohibited in validation sets?
No. Normal tissues may be used in conjunction with neoplastic and lesional tissue as appropriate,
but the guideline specify that normal tissues cannot comprise the entire validation set for markers
that are primarily used in diagnosing neoplasms. If the marker will be used to determine cell
lineage in neoplasms, at least some of the tissues in the validation set should be neoplasms with
positive and negative expression for that marker.
What is the difference between a tissue microarray (TMA) and a multitissue block (MTB)?
The terms are not always used consistently and TMAs and MTBs are not necessarily different.
TMA often refers to a tissue block constructed using a commercially available instrument that
results in uniform cores while MTBs may be assembled manually; these are sometimes referred
to as “sausage blocks” or “spring rolls.”
If we temporarily move our laboratory while the existing one is being remodeled, do we
have to revalidate all assays after both moves?
A complete revalidation of all assays is not required when equipment is moved, but a limited
assessment of a selection of assays is recommended following laboratory relocation. In this
situation, re-assessment of assay performance would apply to both moves. After each move, the
laboratory medical director should select a group of assays that encompass different clinical uses
(i.e., predictive and non-predictive markers, pathogen-specific markers, etc) and
immunolocalizations (i.e., nuclear, cytoplasmic and membranous) and compare results of testing
after the move with the results of testing done previously. The number of validation tissues tested
should be determined by the director.
Does the guideline address validation of research use only (RUO) antibodies?
Not specifically, but the principles of analytic validation described in the guideline apply to all
antibodies that may be used in patient testing.
Could you give some advice on the interpretation of the following terminology for IHC
tests?:
1. Accuracy/Precision (Repeat measurement of samples at various concentrations or activities)
2. Sensitivity (Lower limit of detection)
3. Specificity
4. Reportable Range (Analytic Measurement Range)
5. CLIA requirements to determine test performance specifications apply to all lab tests
including all IHC assays, but the nature of these assays is such that some of them aren’t
relevant. For instance, reportable range and reference intervals are generally not applicable
to tests that are reported qualitatively or semi-quantitatively, which represents most IHC
tests.
•

With respect to determining accuracy, precision, analytical sensitivity and analytical
specificity, CLIA distinguishes between FDA approved and laboratory-developed tests
(LDTs). For FDA-approved test kits, laboratories must demonstrate performance
characteristics that are comparable to those established by the manufacturer (often
called “verification”). Manufacturers may provide users with directions and/or materials for
this verification. By contrast, laboratories must “establish” their own performance
specifications for LDTs. For IHC assays, accuracy, analytic sensitivity and specificity are
determined by analytic assay validation, which is theoretically done by testing a validation
tissue set against a gold standard. Since the majority of IHC tests do not have a "gold
standard" referent test, analytic sensitivity and specificity are determined by measuring
positive and negative concordances with an appropriate comparator. This may be
another validated IHC assay (i.e., different clone), testing done in another lab with a

FAQs

validated assay, a different test (e.g., ISH), or even clinical outcome if you have the
resources. For most laboratories and tests, it’s some combination of the first two.
•

In our literature review we could not find strong evidence to say how IHC assay precision
(inter-run and inter-operator) should be measured. Until stronger evidence is available,
the laboratory director must determine the extent to which these performance
specifications are established based on the method, testing conditions and personnel
performing the test.

Aren’t commercially available antibodies already validated for clinical use by
manufacturers?
The guideline applies to analytic validation of assays, not antibodies. An antibody marketed as an
FDA Class I in vitro diagnostic device may be produced following current good manufacturing
practices and with documentation of specificity, but if the laboratory’s assay is improperly
designed or is not performed correctly (e.g. incorrect antibody dilution, inadequate antigen
retrieval, wrong buffer, incorrect scoring system used), the test results will be incorrect. For
antibodies marketed as “analyte specific reagents,” the laboratory performing the test must
establish the performance characteristics of the clinical assay.
Does the guideline apply to validation of controls?
No. The guideline applies to assays, not antibodies or controls.
Can negative internal cells be used as a negative tissue test or do the negative validation
samples need to be separate tissue samples?
In some cases a section of tissue may contain both antigen-positive cells and negative internal
control cells, and therefore serve as both a positive and negative validation challenge. When
validating a new antibody lot with one positive and one negative case, for example, a single
control slide that contains both antigen-positive and antigen-negative cells might be sufficient.
Does the guideline apply to assays that have been in use in the laboratory for many years
or do they only apply to newly introduced assays?
The guideline applies to all assays used on patient specimens. CLIA requires laboratories to
verify the performance characteristics of all assays before issuing results on patient specimens.
Thus, even if an assay has been in use, if there is no documentation that validation was ever
done, the laboratory may not be compliant with federal law and could be subject to citation by an
accrediting agency.
Do we have to revalidate every existing assay to provide the number of cases
recommended?
Revalidation of existing assays would not be expected if a previous validation was performed, but
the Medical Director should determine if the previous validation was sufficient.
Must all tissues from a validation set be acquired by and processed in the laboratory
validating the IHC panel?
No. This would be ideal but is not possible for many laboratories, especially reference
laboratories, and may be impossible for some rare antigens.
How long must laboratories do validations on all the antibodies they currently use?
For each assay, initial validation is done once and not repeated unless the assay is changed.
Validation records should be retained indefinitely to demonstrate to future inspectors that it was
done.
Some laboratories use microwave fixation to decrease processing time. How does this
reduced fixation time influence IHC results?
This specific issue was not addressed in the guideline, but because any change to a procedure

FAQs

can introduce variation in test results, assays done on microwave fixed tissues should be
compared to routinely fixed and processed specimens to determine if IHC results are affected.
Is a single daily control slide sufficient for validation?
No. Daily quality control is essential to ensure the assay has not changed and continues to
perform as expected, but this is not a substitute for initial assay validation.
REFERENCES:
1. Fitzgibbons PL, Bradley LA, Fatheree LA, et al. Principles of analytic validation of
immunohistochemical assays: guideline from the College of American Pathologists Pathology
and Laboratory Quality Center. Arch Pathol Lab Med. 2014;138(11):1432-1443.

REVIEW ARTICLE

Principles of Analytic Validation of Clinical
Immunohistochemistry Assays
Jeffrey D. Goldsmith, MD,* Patrick L. Fitzgibbons, MD,w
and Paul E. Swanson, MDz

Abstract: All assays performed in anatomic and clinical pathology
laboratories must be validated before they are placed into clinical
service. This review summarizes strategies for validation of clinical
immunohistochemistry assays, and is chiefly based on the recently
released guideline released by The College of American
Pathologists.
Key Words: clinical immunohistochemistry, analytic validation,
practice guideline

(Adv Anat Pathol 2015;22:384–387)

I

n the current practice of anatomic pathology, immunohistochemistry (IHC) is a critical ancillary test that aids in
the accurate diagnosis of a host of neoplastic and nonneoplastic conditions. In addition, IHC is being increasingly used to predict response to therapy and screen for
inherited diseases. In the last decades of the 20th century,
IHC assays were being developed that could be reproducibly performed on paraffin-embedded, formalin-fixed tissues; these methods were developed as “home-brew” assays,
more appropriately termed “laboratory developed tests.”
As such, assay conditions often varied significantly between
laboratories. As IHC became more widespread and its use
expanded to industry, detection methods became more
standardized. However, as many preanalytic factors may
affect the results of IHC tests, assay conditions still may
vary significantly between laboratories.
Many IHC laboratories continue to use laboratory
developed tests; as preanalytic factors may significantly
affect assay results, robust and standardized analytic validation before use on patient samples is required, particularly for those assays with quantitative results or for IHC

From the *Department of Pathology, Beth Israel Deaconess Medical
Center, Children’s Hospital Boston, and Harvard Medical School,
Boston, MA; wDepartment of Pathology, St. Jude Medical Center,
Fullerton, CA; and zDepartment of Pathology and Laboratory
Medicine, University of Calgary and Foothills Medical Centre,
Calgary, AB, Canada.
In various combinations, authors have received honoraria and travel
expense reimbursement for speaking at the 2011 (PS), 2013 (PF,
JG), 2014 (PF, JG), and 2015 (PF, JG) College of American
Pathologists Annual Meetings regarding The Guidelines cited in
this article. J.D.G. and P.E.S. spoke at the 2014 annual meeting of
the American Society of Clinical Pathology regarding these guidelines. J.D.G. also spoke at the 2014 annual meetings of the American Society of Cytopathology and the United States and Canadian
Academy of Pathology. In addition, all authors were on the committee that produced the aforementioned guidelines and received
reimbursement from the College of American Pathologists for
expenses incurred as a result of committee participation.
Reprints: Jeffrey D. Goldsmith, MD, Department of Pathology, Beth
Israel Deaconess Medical Center, 330 Brookline Ave., Boston, MA
02215 (e-mail: [email protected]).
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

tests that predict responsiveness to specific therapies.
Indeed, analytic validation of all clinical laboratory tests,
including IHC, is required by the Clinical Laboratory
Improvement Amendments of 1988.1 Despite both this
regulatory mandate and the common sense notion that
quality testing is predicated on carefully validated
methodology, up to 28% of surveyed IHC laboratories did
not have a written procedure for initial assay validation at
the time a recent interlaboratory practice survey.2 The same
survey noted that laboratories in compliance with CLIA’88
validation requirements nonetheless followed strikingly
variable IHC assay analytic validation practices. To
address these challenges to the uniformity and quality of
diagnostic IHC, the College of American Pathologists
(CAP) convened a panel of experts in 2012 with the charge
of creating an evidence-based guideline that would serve as
a standard for analytic validation of IHC assays. The
resulting recommendations were published in 2014.3
With these introductory comments in mind, we herein
review the relevant concepts behind analytic validation with
particular focus on analytic validation of IHC assays. The
authors of this review served on the expert panel that created above-mentioned guidelines; however, this article has
been created without input from the CAP.

GENERAL CONSIDERATIONS
The United States Food and Drug Administration
defines “validation” as “confirmation by examination and
provision of objective evidence that the particular requirements for a specific intended use can be consistently
fulfilled.”4 In other words, analytic validation is a process
that confirms that a test has the expected level of sensitivity,
specificity, and reproducibility for its intended use. In the
context of the clinical pathology laboratory, validation is
achieved by comparing the test’s result with a known gold
standard. However, a vast majority of IHC assays do not
have a gold standard referent test that can be feasibly
obtained by most laboratories. As such, most laboratories
must compare their results to comparators that are not
considered gold standards in the strict sense. Such comparators tend to fall in the following 4 categories.
(1) Morphology and expected results according to the
medical literature: This comparator is frequently used
when new assays are being initiated. Typically, the
medical director of the laboratory performs a review of
the literature pertinent to the new assay. From those
data, a set of validation cases is chosen, typically from
the laboratory archives from cases fixed and processed
in the same manner as those that will be run on patient
samples.
(2) Previous results from a previously validated assay from
the same laboratory: This method is often used if the
assay conditions change to such an extent that merits

384 | www.anatomicpathology.com
Adv Anat Pathol  Volume 22, Number 6, November 2015
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Adv Anat Pathol



Volume 22, Number 6, November 2015

some sort of revalidation (see below). For example, if a
manufacturer discontinues a primary antibody and it is
replaced with a different primary antibody clone, this
change is considered a fundamental modification to the
assay that requires complete revalidation. In this
circumstance, the use of results obtained from previously validated assays from the same laboratory as a
comparator would be a reasonable approach.
(3) Another laboratory’s results from the same validation
set using a previously validated assay: This method is
particularly useful for assays that are difficult to
validate. In this situation, interlaboratory comparison
allows the laboratory to directly compare results from a
previously validated assay on the same tissues.
(4) Previously validated results from a sufficiently validated
nonimmunohistochemical assay: As noted above, this
comparator applies to very few assays, but is often the
most robust validation method. Examples include
chromogenic or fluorescent in situ hybridization
(CISH/FISH) for Her2-neu as applied to Her2 IHC,
flow cytometric analyses for markers such as CD3,
CD20, and other common hematopoietic analytes, and
mutation testing for the BRAF V600E mutation as
compared with mutation specific b-raf IHC.

CONCORDANCE AND SIZE OF VALIDATION SET
The desired level of concordance between the new
assay and the comparator is tightly related to the size of the
validation set. This is due to the fact that both of these
parameters have a hand in determining the confidence
interval for a particular level of concordance. The confidence interval, generally set at 95%, is the statistical value
that determines the level of confidence that the observed
concordance level reflects the true performance of the test.
Thus, as the size of the validation set increases, the level of
confidence that the observed concordance is the true value
increases. For an example, see Table 1; this table shows that
the 95% confidence intervals are smaller and overall confidence levels higher with a validation set composed of 40
cases compared with 20 case validation set. Thus, as a
general rule, a larger validation set is desirable, whenever
possible. Of course, larger validation sets can be difficult to
obtain, especially in smaller laboratories.
The size of the validation set should also be dictated by
the intended clinical use of the assay. The clinical use of
IHC assays fall into 2 general groups. The first are markers
that are interpreted in the context of the morphologic
findings and are typically used as ancillary stains for diagnosis (eg, cytokeratin 7, cytokeratin 20, TTF-1, GATA-3,
etc.). The second group of stains includes those that are
interpreted without regard to the histologic context; many
of these markers give predictive information about the
sensitivity of a tumor to various treatments (eg, Her2 IHC
on breast carcinoma, b-raf mutation-specific IHC on melanoma). Markers that are used for histologic diagnosis and
are interpreted in a histologic context have less direct

Analytic Validation of Clinical IHC Assays

clinical impact than predictive markers that result in an
actionable result that is independent of the morphologic
context. Thus, the size of the validation set for a predictive
marker should be larger than that prepared for a diagnostic
marker. The expansion of the size of the validation set for
predictive markers increases the confidence that the
observed concordance level truly reflects the desired level of
concordance. As such, the CAP Guideline mandates that
the size of the validation set for predictive markers should
contain at least 40 challenges, whereas nonpredictive/diagnostic markers should have at least 20 challenges.
Depending on the resources available, expansion of the size
beyond the prescribed amount of the validation set is
optimal and would add additional assurance that the assay
will behave as expected.
In theory, the level of aggregated positive and negative
concordance between the new test and the comparator
should be 100%. However, this is not practically obtainable
due to a number of factors including, but not limited to,
preanalytic factors, intratumoral heterogeneity of analyte
expression and the quality of the originally validated
method or comparator set. For these reasons, the Guideline
set the desired level of concordance at 90%; this was chiefly
based on evidence from concordance data between Her2
IHC and Her2-neu FISH, in which concordance levels
higher than 90% were not feasible for a majority of
laboratories.5–7

COMPOSITION OF THE VALIDATION SET
As a general rule, the composition of the validation set
should reflect the intended clinical use of the assay. Not
only should relevant positive cases be included, but also
judicious inclusion of cases that show lack of expression of
the analyte of interest should be part of the validation set.
For example, TTF-1 is a transcription factor that is often
used as an ancillary test in the workup of metastatic carcinoma of unknown origin. It is expressed in a majority of
small cell carcinomas of the lung, most primary pulmonary
adenocarcinomas, and many types of primary epithelial
tumors of the thyroid gland. Inclusion of tumor types that
are known to be positive for TTF-1 should be part of the
validation set. In addition, tumor types that are known to
be TTF-1 negative and are in the histologic differential
diagnosis of either metastatic pulmonary adenocarcinoma
and metastatic thyroid carcinoma should be included. Such
examples of clinically relevant TTF-1 negative carcinomas
might include ductal carcinoma of the breast, colorectal
carcinoma, and pancreatic ductal adenocarcinoma. Inclusion of clinically relevant cases in the validation set adds
additional assurance that the validation accurately reflects
the performance of the assay when performed on patient
samples.
Occasionally, assays are used in more than 1 clinical
context. In this circumstance, it would be wise to tailor the
validation set to reflect all potential clinical uses. For
example, CD30 is a marker that is often used to diagnose

TABLE 1. Comparison of Concordance Rates and 95% Confidence Intervals for Validation Sets Composed of 20 and 40 Tissues

No. Validation
Tissues

Concordance for 0
Discordant Cases

Concordance for 1
Discordant Case

Concordance for 2
Discordant Cases

20
40

100% (81%-100%)
100% (90%-100%)

95% (75%-100%)
97.5% (86%-100%)

90% (69%-98%)
95% (83%-99%)

Copyright

r

2015 Wolters Kluwer Health, Inc. All rights reserved.

www.anatomicpathology.com |

Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

385

Goldsmith et al

Hodgkin lymphoma and various germ cell tumors, such as
embryonal carcinoma. In this circumstance, the validation
set should include cases of Hodgkin lymphoma, in which
Reed-Sternberg cells are the expected positive cells, and
cases of embryonal carcinoma. Relevant negative cases
include nodular lymphocyte-predominant Hodgkin lymphoma and CD30-negative primary mediastinal diffuse
large B-cell lymphoma, which are known mimics of
Hodgkin lymphoma and are CD30 negative. In addition,
expected CD30-negative cases in the differential diagnosis
of embryonal carcinoma should be included in the validation set that might include seminoma and yolk sac
tumor.

FORMAT OF VALIDATION CHALLENGES
Classically, validation is achieved by applying single
tissue sections on slides analogous to the practice on patient
samples. More recently, tissue microarrays have been used
as a more efficient and cost-effective method of displaying
multiple challenges on a single microscopic slide.8–12 Tissue
microarrays are usually an acceptable method of validation.
However, caution should be exercised with assays that are
known to show significant heterogeneity of staining.
Examples of this include bcl-6 staining in normal tonsillar
tissue. Bcl-6 expression is limited to germinal center cells; as
such, a tissue microarray would not be an effective method.
Similarly, CD15 and CD30 validation using classic Hodgkin lymphomas should not be performed using tissue
microarrays, as the CD15-positive and CD30-positive
Reed-Sternberg cells are very often heterogenously distributed within lesional tissue in this tumor type.

PREANALYTIC CONSIDERATIONS
Once tissues become devitalized at the time of biopsy
or resection, they are fixed, processed, and prepared for
microscopic diagnosis. This process can differ between
laboratories and, in fact, may vary within a particular
laboratory depending on the specimen type. These variations in tissue processing and handling may have dramatic
effects on IHC results. For example, for some antibodies,
acidic decalcification solutions can change the avidity of the
primary antibody for its epitope(s).13 Although it is
impossible to control for all possible preanalytic factors
during validation, attention to major causes of preanalytic
variation should be taken into account. Some of the major
preanalytic factors that may impact results include fixative
type and preparation method (ie, cytologic preparations vs.
formalin-fixed, paraffin-embedded tissues).14,15
If tissues fixed in alternative fixatives or tissues
exposed to decalcifying solutions are to be used for IHC,
efforts should be made to ensure that the results are clinically valid. A reasonable approach would be to validate a
subset of assays that are often run on decalcified samples.
Examples of such assays might include cytokeratins, CD45,
S-100, and estrogen receptor.
Similarly, if IHC is run on cytologic preparations,
including smears, cytospins, cell blocks, and ThinPrep
preparations (or core samples submitted with aspirate fluid
or other preparation to the cytology laboratory in CytoLyt
or other nonformalin solutions), reasonable efforts should
be made to assure that these assays perform adequately
before they are used on patient samples. The selection of
markers tested and number of cases included in these

Adv Anat Pathol



Volume 22, Number 6, November 2015

separate validation studies must be determined by the laboratory medical director.

REVALIDATION AFTER CHANGES TO ASSAY
CONDITIONS
Once initial assay validation is successfully completed
and a test is placed in clinical service, it is common for assay
conditions to change. When that occurs, some sort of
revalidation is needed to assure that the assays perform as
expected. In general, changes to assay conditions fall into 3
categories. The first, and perhaps most straightforward, is a
change to the antibody clone. As different antibody clones
target different epitope(s), changes in antibody clone are
considered a fundamental change to the assay. In this circumstance, full analytic revalidation is required.
The second category includes modifications to assay
conditions that are common to all assays in the laboratory.
Examples include changes to detection chemistry, water
supply, antigen retrieval solution(s), and tissue processing
equipment. When such changes occur, it is not necessary to
fully revalidate all assays affected by the change. Instead, it
is reasonable to choose a representative sample of assays
run in the laboratory and compare cases prepared with the
modified assay conditions with examples representative of
original conditions. If the subset of modified assays performs as expected, it would be reasonable to assume that
the remaining assays will perform adequately. If, however,
significant changes to the assay conditions are necessary to
achieve expected results, more extensive revalidation may
be necessary.
The final set of condition changes that merit revalidation are changes that apply to single assays. Examples of
this might include changes to antibody lot, primary antibody dilution, primary antibody incubation time, and
change of primary antibody vendor using the same clone.
Of these changes, a new antibody lot (same clone) often
results in minimal perturbation of the assay. As such, verification of continued expected assay results is achieved by
running 1 known positive and 1 known negative case. It
may be judicious to include a third case that shows a lowpositive reaction as an additional indication of appropriate
assay performance. Changes to primary antibody dilution,
incubation time, and vendor are more substantive changes
to the assay. In these circumstances, it is reasonable to run 2
known positive and 2 known negative cases to assure
continued assay performance; again, it may be wise to run a
fifth, low positive, case to assure appropriate assay
sensitivity.

CONCLUSIONS
IHC is a critical ancillary test in the modern anatomic
pathology laboratory that often has significant impact on
patient care. To be assured of accurate results, robust
analytic validation must be performed on all assays before
their use on clinical samples. This review summarizes best
practices for analytic validation for IHC assays and outlines an approach for revalidation necessitated by changes
to assay conditions after successful completion of initial
validation procedures.
REFERENCES
1. US Department of Health and Human Services. Clinical
laboratory improvement amendments of 1988: Final Rule. Fed
Regist. 1992;57:7001–7186.

386 | www.anatomicpathology.com
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.
Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

Adv Anat Pathol



Volume 22, Number 6, November 2015

2. Hardy LB, Fitzgibbons PL, Goldsmith JD, et al. Immunohistochemistry validation procedures and practices: a College of
American Pathologists survey of 727 laboratories. Arch Pathol
Lab Med. 2013;137:19–25.
3. Fitzgibbons PL, Bradley LA, Fatheree LA, et al. Principles of
analytic validation of immunohistochemical assays: Guideline
from the College of American Pathologists Pathology and
Laboratory Quality Center. Arch Pathol Lab Med. 2014;138:
1432–1443.
4. Institute CLS. Quality Assurance for Design Control and
Implementation of Immunohistochemistry Assays: Approved
Guideline (CLSI Document I/LA28-A2), 2nd ed. Wayne, PA:
Clinical and Laboratory Standards Institute; 2011.
5. Mayr D, Heim S, Werhan C, et al. Comprehensive immunohistochemical analysis of Her-2/neu oncoprotein overexpression in breast cancer: HercepTest (Dako) for manual testing
and Her-2/neuTest 4B5 (Ventana) for Ventana BenchMark
automatic staining system with correlation to results of
fluorescence in situ hybridization (FISH). Virchows Arch. 2009;
454:241–248.
6. Rhodes A, Jasani B, Anderson E, et al. Evaluation of HER-2/
neu immunohistochemical assay sensitivity and scoring on
formalin-fixed and paraffin-processed cell lines and breast
tumors: a comparative study involving results from laboratories in 21 countries. Am J Clin Pathol. 2002;118:408–417.
7. van der Vegt B, de Bock GH, Bart J, et al. Validation of the
4B5 rabbit monoclonal antibody in determining Her2/neu
status in breast cancer. Mod Pathol. 2009;22:879–886.

Copyright

r

Analytic Validation of Clinical IHC Assays

8. Fons G, Hasibuan SM, van der Velden J, et al. Validation of
tissue microarray technology in endometrioid cancer of the
endometrium. J Clin Pathol. 2007;60:500–503.
9. Gulbahce HE, Gamez R, Dvorak L, et al. Concordance
between tissue microarray and whole-section estrogen receptor
expression and intratumoral heterogeneity. Appl Immunohistochem Mol Morphol. 2012;20:340–343.
10. Kwon MJ, Nam ES, Cho SJ, et al. Comparison of tissue
microarray and full section in immunohistochemistry of
gastrointestinal stromal tumors. Pathol Int. 2009;59:851–856.
11. Thomson TA, Zhou C, Chu C, et al. Tissue microarray for
routine analysis of breast biomarkers in the clinical laboratory.
Am J Clin Pathol. 2009;132:899–905.
12. Dessauvagie BF, Thomas C, Robinson C, et al. Validation of
mitosis counting by automated phosphohistone H3 (PHH3)
digital image analysis in a breast carcinoma tissue microarray.
Pathology. 2015;47:329–334.
13. Athanasou NA, Quinn J, Heryet A, et al. Effect of decalcification agents on immunoreactivity of cellular antigens.
J Clin Pathol. 1987;40:874–878.
14. Hanley KZ, Birdsong GG, Cohen C, et al. Immunohistochemical detection of estrogen receptor, progesterone receptor, and
human epidermal growth factor receptor 2 expression in breast
carcinomas: comparison on cell block, needle-core, and tissue
block preparations. Cancer. 2009;117:279–288.
15. Nishimura R, Aogi K, Yamamoto T, et al. Usefulness of
liquid-based cytology in hormone receptor analysis of breast
cancer specimens. Virchows Arch. 2011;458:153–158.

2015 Wolters Kluwer Health, Inc. All rights reserved.

www.anatomicpathology.com |

Copyright r 2015 Wolters Kluwer Health, Inc. All rights reserved.

387