COMMENT EVERYLIFE FOUND Attachment 1 8-26-13

Recommendations for the Development of Rare Disease Drugs
using the Accelerated Approval Pathway and for Qualifying
Biomarkers as Primary Endpoints in Pivotal Clinical Studies

Final
August 18, 2013

1

TABLE OF CONTENTS

I.

INTRODUCTION 3

II.

BACKGROUND

5

III. THE QUALIFICATION PROCESS SHOULD TAKE INTO ACCOUNT THE SPECIFIC CONTEXT OF THE
DISEASE, ITS RARITY AND OTHER FACTORS WHEN EVALUATING A BIOMARKER FOR THE ACCELERATED
APPROVAL PATHWAY.
7
A.
B.

THE KEY CONSIDERATIONS FOR THE DISEASE ASSESSMENT: ................................................................................................. 8
SUMMARY ............................................................................................................................................................................. 9

IV. ESTABLISH A BIOMARKER QUALIFICATION REQUEST PROCESS FOR INDIVIDUAL DRUGS AND
DISEASES
10
V. SUMMARY LIST OF QUALIFICATION CONSIDERATIONS OF A DISEASE/DRUG/BIOMARKER SET FOR
ACCELERATED APPROVAL IN RARE DISEASES
11
A. DISEASE CONSIDERATIONS ..................................................................................................................................................... 12
B. DRUG CONSIDERATIONS ......................................................................................................................................................... 12
C. BIOMARKER CONSIDERATIONS ............................................................................................................................................... 12
D. PRECLINICAL CONSIDERATIONS ............................................................................................................................................. 12
E. CLINICAL DATA CONSIDERATIONS FOR BIOMARKER QUALIFICATION ...................................................................................... 12
VI. DETAILED DESCRIPTION OF THE SCIENTIFIC CONSIDERATIONS FOR THE QUALIFICATION OF
BIOMARKER ENDPOINTS USING DISEASE, DRUG, BIOMARKER, PRECLINICAL AND CLINICAL
INFORMATION
13
A.
B.

DISEASE CONSIDERATIONS REGARDING THE UNDERLYING DISEASE PATHOPHYSIOLOGY THAT SUPPORT PREDICTIVE VALUE13
DRUG CONSIDERATIONS FOR PHARMACOLOGY OF THE TREATMENT AND ITS RELATIONSHIP TO THE DISEASE AND BIOMARKER
13
C. BIOMARKER CONSIDERATIONS REGARDING THE SPECIFIC SCIENTIFIC BASES OF BIOMARKERS OR INTERMEDIARY CLINICAL
ENDPOINTS THAT SUPPORT THE LIKELIHOOD OF PREDICTIVE VALUE ........................................................................................... 17
D. PRECLINICAL EVIDENCE TO SUPPORT USE OF A DISEASE/DRUG/BIOMARKER SET FOR ACCELERATED APPROVAL ................ 19
E. AVAILABLE CLINICAL EVIDENCE TO SUPPORT USE OF A DISEASE/DRUG/BIOMARKER SET FOR ACCELERATED APPROVAL ... 21
VII. CLINICAL TRIAL DESIGN CONSIDERATIONS
A.
B.

22

PIVOTAL STUDY TRIAL DESIGN: CONCEPTS FOR STUDIES USING A PRIMARY SURROGATE MARKER ENDPOINT ................... 23
POST-MARKETING CONFIRMATORY STUDIES IN RARE DISEASE INDICATIONS ...................................................................... 25

VIII. EXAMPLES OF SUCCESSFUL USE OF BIOMARKER ENDPOINTS DURING DEVELOPMENT OF RARE
DISEASE DRUGS
26
IX. CONCLUSIONS AND RECOMMENDATIONS
X.

REFERENCES

28

29

2

I.

INTRODUCTION

According to the National Institutes of Health, there are thousands of rare diseases which together affect
as many as 30 million Americans.1 These are often serious or life-threatening diseases and yet only there
are only about 400 drugs currently approved for rare diseases.2 Developing drugs for rare disease can be
challenging due to specific disease characteristics such as small heterogeneous patient populations, long
time-frames for disease progression, a poor understanding of disease natural history, and a lack of prior
clinical studies. Recent advances in medical science have enhanced our understanding of these
disorders at the biochemical and pathophysiologic level and have created more opportunities to address
unmet needs3 by developing specific therapeutic options for rare disease patients. Improving the
development process for these diseases is now an important part of assuring that many of the rarest and
most difficult-to-treat rare diseases have specific drugs developed.
In the United States, the regulatory approval of any new drug is based on its benefit-risk ratio, which is
assessed in part through the conduct of adequate and well-controlled clinical trials in the relevant patient
population. “Benefit” refers to a demonstrated improvement in clinical function as measured using a
clinical endpoint, which is defined broadly as the way a patient feels, functions, or survives; “risk” refers
to the safety profile of the drug. The requirement for clinical endpoints defined by the “feels, functions
or survives” paradigm can be challenging as some promising investigational treatments for serious or
life-threatening diseases may not be practically assessed for a demonstrated improvement on clinical
endpoints for many reasons. The challenge with clinical endpoints can occur in studies of the extremely
small or heterogeneous patient populations or in diseases characterized by long periods with subclinical
or slow progression and or substantial irreversible damage at the time of diagnosis.
To assist in the difficult challenge of transforming scientific discoveries into new drug treatment for
patients with serious or life-threatening disorders, the U.S. Food and Drug Administration (FDA),
Congress, and the public have all endorsed the need for flexibility in the regulatory review process for
rare disease drugs. This regulatory flexibility has been designed to speed access to new drugs, while
preserving standards for safety and efficacy. Perhaps the best expression of this flexibility is the creation
of the Accelerated Approval (AA) pathway in regulations promulgated by the FDA in 1992.4 Initially,
these regulations created by the FDA were a response to the AIDS crisis, and the growing public
pressure to accelerate access to medications. Congress codified the AA pathway in 1997, with the
passage of the FDA Modernization Act (FDAMA).
The AA regulations specify when evaluating drugs for serious and life-threatening diseases with
substantial unmet medical need, the FDA may approve a treatment based on an efficacy evaluation using
a surrogate endpoint or a clinical endpoint earlier than survival or irreversible morbidity that is
“reasonably likely to predict clinical benefit.” Drugs granted AA are approved with the stipulation that
confirmatory studies to verify the clinical benefit may be required as a condition of continuing
marketing authorization.5 The regulatory flexibility offered under the AA pathway incorporates our
belief that when a patient has a lethal or devastating disease, the benefit-risk assessment must account
for the severity of the disorder and the degree of unmet medical need, i.e., the availability (or not) of
effective alternative treatment options. It is important to emphasize, however, that AA is valuable
specifically for those situations in which timely standard approval via a clinical endpoint is unlikely or
3

impossible due to practical, scientific, or ethical reasons.
The use of biomarkers in the AA pathway is not intended to substitute for clinical endpoint-based
studies in diseases with sufficient patients and readily measured clinical endpoints that may change in
reasonable timeframes. Studies based on clinical endpoints are preferable in the development process
when feasible. In rare diseases, often the population size and heterogeneity, the nature of the disease
and the limited historical clinical data can make traditional studies with clinical endpoints, difficult or
impossible to conduct.
Although the AA regulations have been used to great advantage for diseases such as AIDS and cancer,
there has been more limited success in applying this pathway to treatments for non-oncology rare
diseases.6,7 Despite substantial scientific insights into the pathogenesis and pathophysiology of many
rare disorders, the translation of scientific discoveries into effective medicines for these disorders has
been notably more challenging and at times impossible under the current regulatory framework, even
when the understanding of the science underlying the disease is far superior to that for many multi-genic
common diseases.
Rare disease treatments are sometimes approved via the standard pathway using unvalidated biomarker
endpoints due in part to flexibility in the drug review process. 8 In most cases, this flexibility is based on
either regulatory precedent for the endpoint, or the existence of substantial clinical data with a prior
treatment. For example, the regulatory precedent for the use of ammonia levels in urea cycle drug
development is based on prior approvals for phenylbutyrate and other urea cycle drugs, and allowed
ammonia levels to be used in the recent approval of glycerol phenylbutyrate. One recent example in
which substantial prior clinical outcome data was important was the approval of sapropterin based on
the use of blood phenylalanine levels in phenylketonuria (PKU). Blood phenylalanine levels were
considered predictive based on published studies of intellectual outcomes in PKU during dietary therapy
and were used in the management of PKU. However, if there are no regulatory precedents for use of the
relevant specific biomarker, and if there is no substantial prior clinical data to support the predictive
value of the biomarker as a surrogate, then qualification of a new biomarker for use as a primary
surrogate endpoint in a pivotal study can be difficult or impossible. 9
Although the AA pathway is rarely used to approve non-oncology rare disease treatments, many rare
diseases pose similar levels of severity, lethality, and unmet medical need as diseases treated by drugs
that currently do access the AA pathway, such as AIDS and cancer. This is largely because the current
benefit-risk assessment framework as applied to rare diseases has not adequately addressed uncertainty
about the level of evidence necessary to rely on a novel biomarker endpoint.4, 10 The expectation for
demonstrated clinical outcome data has made the use of the pathway challenging for rare diseases,
despite the often excellent science that exists regarding the underlying disease, drug and biomarker. A
re-evaluation of the type and quality of data required for biomarker qualification is needed to create a
more relevant AA pathway for rare diseases when the other science is strongly supportive of the
biomarker but when clinical outcome data is limited or non-existent.
The 2012 passage of the Food and Drug Administration Safety and Innovation Act (FDASIA)11
amended the AA provisions to reflect recent advances in science and to enhance the application of the
AA pathway to drugs for rare disorders, with the intention of expediting the development and approval
4

of new treatments. FDASIA extends FDA’s authority to take into account other available endpoints to
qualify for AA and requires the development of more relevant guidance on the types of evidence that
may be acceptable in support of using a novel surrogate endpoint. In addition, the law also contains
provisions to incorporate the patients’ benefit-risk preferences into a structured evaluation process.
Together, these provisions create a significant and valuable opportunity to advance the translation of
promising scientific discoveries into new treatments for rare disorders.
This paper provides recommendations for an effective and detailed scientific framework to improve the
relevance of the AA pathway for rare diseases with unmet medical need where sufficiently strong
science exists. It outlines the types and levels of information that increase the predictive value of
biomarker endpoints, as well as the scientific bases sufficient to merit utilization of AA for rare diseases
using novel biomarker endpoints. The scientific framework cannot substitute for good judgment or
accommodate all the complexities of the science behind so many rare diseases, but the framework does
provide a more predictable structure for development and review of potential biomarker endpoints.
Specifically, the goals of this paper are to:
o Establish the key considerations for rare disease factors that should impact qualification for
Fast Track and the use of the AA pathway
o Provide for the use of a disease survey early in development to help characterize the factors
considered important for qualification of the disease for AA and in drug development
o Provide a reasonable scientific framework for considerations regarding the disease, drug,
biomarker and other data that can support adequate qualification of a biomarker as a
surrogate primary endpoint that is “reasonably likely to predict clinical benefit”
o Define a regulatory process for assessing biomarkers in rare diseases earlier in the
development process to encourage investment in new treatments: the Biomarker
Qualification Request
Ultimately, the adoption and incorporation of these recommendations for a scientific qualification
framework into the regulatory process will create an opportunity to increase the number of available
treatments for rare diseases when there is adequate science to support development while maintaining
safety and effectiveness standards.
II.

BACKGROUND

Section 901 of FDASIA, entitled “Enhancement of Accelerated Patient Access to New Medical
Treatments,” amended the AA provisions found in Section 506 of the U.S. Public Health Service Act. 12
These changes are intended to take advantage of the significant advances in science over the last several
decades to increase the application of AA to drugs for serious, life-threatening, and rare disorders. New
assay techniques and methodologies, including advances in genomics, molecular biology, and
bioinformatics, have allowed us to better understand the pathophysiology of disease and to develop new
therapies. Per FDASIA, Title IX, Section 901 (1) C:
“As a result of these remarkable scientific and medical advances, the FDA should be encouraged
to implement more broadly effective processes for the expedited development and review of
5

innovative new medicines intended to address unmet medical needs for serious or lifethreatening diseases or conditions including those for rare diseases or conditions, using a broad
range of surrogate or clinical endpoints and modern scientific tools earlier in the drug
development cycle when appropriate. This may result in fewer, smaller or shorter clinical trials
for the intended patient population or targeted subpopulation without compromising or altering
the high standards of the FDA for approval of drugs.”
FDASIA also underscores the importance of taking the context of the specific disease state targeted by
the drug into account when conducting benefit-risk determinations. In particular, the section addressing
AA was amended to note the FDA should consider the “ severity, rarity, or prevalence of the
condition and the availability or lack of alternative treatments” when reviewing a product which
demonstrates an effect on a surrogate endpoint.
The legislation also expands the list of potential information to consider when assessing the predictive
value of a surrogate endpoint. Specifically, it states the evidence “may include epidemiological,
pathophysiological, therapeutic, pharmacologic, or other evidence developed using biomarkers, for
example, or other scientific methods or tools.” These changes are significant, as they increase breadth
of data that may be used to provide reasonable inferences into the predictability of benefit.
FDASIA requires the development of guidance to clarify the considerations unique to the application of
the AA pathway to review drugs for rare disorders. Such guidance will address issues that may arise
under the AA and Fast Track processes outlined in Section 506 of the Federal Food Drug and Cosmetic
Act for drugs designated for a rare disease or condition under section 526 of the Act (21 U.S.C. 360bb).
This supports the earlier inclusion of rarity as a factor for AA and provides support for the principle that
rarity alone leads to significant development issues that this guidance should address and help to
resolve: “In developing such guidance, the Secretary … shall also consider any unique issues
associated with very rare diseases [emphasis added].”
FDASIA further directs the FDA to consider how the rarity of a disease alters the type of data that might
be available to ascertain the suitability of a surrogate endpoint, indicating that, in some cases, limited
data regarding the pathophysiology of disease or the pharmacology of the drug might be sufficient:
“The Secretary shall consider how to incorporate novel approaches to the review of surrogate
endpoints based on pathophysiologic and pharmacologic evidence in such guidance, especially in
instances where the low prevalence of a disease renders the existence or collection of other types
of data unlikely or impractical.”13
FDASIA emphasizes that the AA pathway should be accessible to severe and very rare diseases.
Additionally, it calls for a broader-based approach to the application of scientific data when assessing
the viability of a biomarker as a primary surrogate endpoint for AA. For example, in situations where
long-term clinical outcome data do not exist due to the severity and/or rarity of the disease (e.g., lack of
qualified trial participants, time constraints, etc.), other scientific criteria may be used if they are
believed to sufficiently meet the “reasonably likely” standard.
To ensure the changes outlined in FDASIA are implemented effectively, a working group was formed
(the authors) to develop recommendations to better address the qualification of surrogate endpoints as
6

“reasonably likely to predict clinical benefit” for rare disease using a scientific framework. The
scientific framework is intended to include a broader array of considerations regarding the disease, drug,
biomarker and experimental data that help improve the biologic understanding of the biomarker in the
context of the disease and drug. The fundamental assumption of this proposal is that the more known
about the scientific basis of a disease, the drug, and the biomarker endpoint, the better the ability to
assess predictive power of the biomarker.
The AA pathway is critically important for the development of treatments for rare disorders that are
more challenging to study using routine clinical study than many fast-track-eligible common disorders.14
However, the Food and Drug Modernization Act (FDAMA) did not specifically delineate considerations
unique to the review of treatments for rare diseases under AA. Subsequently, FDASIA does now call for
the consideration of additional factors specific to rare and very rare disorders when considering
regulatory flexibility under the AA pathway. Not all of these rare disease considerations need to be met
in order to qualify for review under the AA pathway; rather, they should be seen as a list of specific
factors that warrant consideration when assessing the need for regulatory flexibility in accessing the AA
pathway and for the qualification of a biomarker as a primary surrogate endpoint in a pivotal clinical
study.
III. THE QUALIFICATION PROCESS SHOULD TAKE INTO ACCOUNT THE SPECIFIC
CONTEXT OF THE DISEASE, ITS RARITY AND OTHER FACTORS WHEN
EVALUATING A BIOMARKER FOR THE ACCELERATED APPROVAL PATHWAY.
The regulations outlining AA did not specifically delineate considerations unique to the review of
treatments for rare diseases under AA. In contrast, FDASIA does call for the consideration of additional
factors specific to rare and very rare disorders when considering regulatory flexibility under the AA
pathway. A number of aspects that merit particular deliberation are outlined in the section that follows.
The proposed list of considerations during the qualification process:
1. Extremely high unmet medical need
2. Extreme rarity of the disorder
3. Lack of any prior clinical studies or clinical data
4. Slowly progressive diseases or low event rate
5. Diseases with delays between irreversible pathologic damage and clinical diagnosis
6. Lack of readily measurable, recognized clinical endpoints due to unusual clinical disease biology
The assessment should be specific to the disease targeted for treatment, including an assessment of
whether and to what degree the targeted disorder can be characterized as serious, life-threatening, and
rare or very rare. This assessment should include a disease survey designed in consultation with the
FDA and key stakeholders such as patient groups and physicians with relevant expertise. The disease
survey results should be discussed and examined with the FDA as part of the review and evaluation
process, validated by patients and patient groups as reflective of their collective view, and verified by
experienced physicians in the specialty as consistent with their experience. This survey should consider
whether and to what extent characteristics of the disorder targeted by the treatment meet the criteria of
7

serious, life-threatening or rare or very rare, thus meriting flexibility for qualification of the biomarker
under the AA pathway.
A.

The key considerations for the disease assessment:


Extremely high unmet medical need
Diseases with devastating and severe outcomes and no approved treatments deserve particular
consideration with regard to utility of AA. The unmet medical need in these diseases greatly
impacts the benefit-risk calculations made by physicians, patients, and caregivers, and these
preferences should be weighed as part of the regulatory review process.



Extreme rarity
Rare disorders affecting very small populations or genetic subpopulations present especially
difficult challenges that have a negative synergistic effect on drug development, such as:
1) The lack of available patients to be enrolled in clinical trials which incorporate clinical
endpoints, negatively impacting a study’s ability to reach a reasonable level of power to detect a
statistically significant change.
2) The need to include a significant fraction of the total available population of patients in
clinical studies, leading to the need to accept heterogeneous populations in terms of age, severity
and presence of specific clinical disease symptoms, as well as stage of disease progression
3) The limited market potential for the drug, resulting in small or non-existent financial
incentives to invest in the development of treatments for extremely rare disorders, particularly
without some degree of confidence that AA is available early in the program’s life before
significant work has begun



Lack of any prior clinical studies or formally collected clinical data
Very rare diseases with no existing treatments have often never been studied in clinical trials. As
a result, surrounding medical literature may be limited to case reports and small sets of patients.
Frequently, rare disease patients are evaluated only in terms of disease management, and not for
clinical endpoint assessments. The lack of regulatory precedents for endpoints relevant to a rare
disease often makes the determination of how to evaluate a disease or treatment effect difficult or
intractable.



Slow disease progression with significant irreversible symptoms or rare severe events
Many rare diseases have long and/or unpredictable timeframes for progression, making it
difficult or impossible to conduct clinical studies within a reasonable timeframe (i.e., less than 1
year), which creates a compelling need for the use of alternative biomarker endpoints. In some
cases, this may be because the event rate is low, even if these events are very severe.
Additionally, if the clinical manifestations of the disease are irreversible and the goal of the
therapy is stabilization, achieving sufficient power to detect the difference between placebo and
treated patients is far more difficult. In this situation, biomarkers that are directly in the line of
the pathophysiologic process could provide a valuable assessment of treatment effect that can
reasonably predict clinical benefit.
8



Significant delay between the onset of irreversible pathologic damage and clinical diagnosis
Untreated rare diseases often have challenging biology in which the disease process initiates and
progresses without clear clinical expression or diagnosis. By the time disease progression has
allowed the patient to be clinically diagnosed, severe late-stage and irreversible damage has
already occurred. This problem is particularly common in neurological disorders, for example, in
which the plasticity and compensatory powers of the brain continuously adapt to the declining
brain condition to maintain function. As a result, the appearance of normality is maintained
despite substantial disease progression until the adaptation can no longer compensate, at which
point the patient rapidly declines. Studying the treatment of a disease during this type of early
prodromal period is difficult or impossible with clinical endpoints. The slow and inconsistent
clinical change, if any, will be undetectable, and waiting until the patient declines may make the
treatment less effective. In some cases, treatment should begin years before disease
manifestations are evident, but earlier asymptomatic diagnosis may not be advocated if no
treatment is available.



Lack of readily measurable, recognized clinical endpoints due to unusual clinical disease
manifestations
A distinct challenge for some rare diseases is the atypical nature of their clinical outcomes, even
when the underlying cause and primary pathophysiology is understood. The disease may not
readily fit into existing clinical models and previously identified clinical endpoints may not be
applicable. For example, in autosomal recessive dystrophic epidermolysis bullosa, a genetic
deficiency causes the epidermal and dermal layers to split and blister. The disease process
cannot be readily measured using intermediate clinical endpoints short of major clinical events
like hospitalizations for infections which are infrequent and avoided via symptomatic care. In
many cases, the non-specific palliative treatments utilized can confound the process of clinical
evaluation, as it would of course be unethical to deny such supportive care. In rare diseases, there
may be long-term downstream clinical outcomes such as hospitalizations that could be described
and appreciated, but conducting the controlled clinical study over the timeframe required will
likely be impractical or impossible as their occurrence may be too variable or their frequency
insufficient.

B.

Summary

An effective application of AA should reflect several important criteria. First, the therapy must be
intended to treat a serious or life-threatening disorder where there is unmet medical need. Second, in
cases of rare diseases, if the disease itself displays any of the specific criteria outlined above, enhanced
flexibility in the qualification of the biomarker endpoint to allow utilization of the AA pathway is
merited.
Third, patients, patient groups and physicians with appropriate expertise should evaluate the disease and
relevant benefit-risk by formally assessing the unmet need and risk-acceptance of affected patients
through direct disease survey or other acceptable means. These considerations are intended as a guide
for determining the need for additional regulatory flexibility in the biomarker qualification process to
enhance development of and access to innovative treatments. This type of disease survey and analyses
9

may also provide opportunities to identify and support a possible clinical endpoint for use in clinical
studies as well as support the qualification of a biomarker.
IV.

ESTABLISH A BIOMARKER QUALIFICATION REQUEST PROCESS FOR
INDIVIDUAL DRUGS AND DISEASES

In the evaluation process for choosing products to develop, sponsors consistently will review the
potential clinical development pathway and the possible regulatory strategies for approval. The lack of
accepted biomarker endpoints for a rare disease with difficult to measure clinical disease manifestations,
will mean that the pathway will be considered too difficult to warrant development and investment. The
tendency to develop additional drugs for rare diseases for which drugs have already been approved, is in
part driven by the certainty of the development pathway and the endpoints. The acceptability of a
biomarker endpoint is occurring too late in the process, typically at the End-of-Phase 2 meeting, and is a
considerable barrier to the development of many novel drugs for untreated rare diseases. Unfortunately
without development pathway predictability, these drugs may never even enter the development process
but will rather languish in academia.
If the biomarker qualification determination can be made earlier in the development process, before
considerable investment in IND-enabling work, the application of the AA pathway would greatly
increase investment in research and development and accelerate the availability of new treatments for
the most difficult untreated rare diseases. There is currently no regulatory process for qualifying a
surrogate endpoint for a specific disease and drug, until typically at the End-of-Phase 2 meeting, late in
the process. Establishment of a BioMarker Qualification Request for individual drugs and diseases
could allow discussions before an IND is developed and help guide appropriate research before
substantial investment has occurred.
The proposed request would be made in parallel or before a pre-IND meeting via a process similar to
that for the pre-IND meeting. A briefing book would be prepared along with questions for the FDA, and
the data considered. Based on this meeting and discussion, the FDA could agree that for this disease and
drug in the specific proposed context, that the biomarker could be used as a primary endpoint with a set
of reasonable assumptions, or that it might be qualified if certain specified data were obtained or
bolstered in the package, or the FDA might decline to qualify the biomarker under any circumstances
due to a specific set of scientific concerns for that biomarker in that specific context of use. The timing
for this process need not be restricted to the pre-IND stage, and could occur later in the process if this is
convenient. The only key goal is to help provide the option of earlier certainty in development, and the
potential then to raise the sufficient funding to develop a rare disease drug when the pathway is clear.
The proposed new request process is different from the currently available biomarker qualification
process that is focused on broadly used biomarkers for multiple diseases, as in defining biomarkers for
renal injury in drug development for many drugs. The current process operates via the Office of
Translational Sciences within the FDA and involves multiple stakeholders, many iterations of evaluation
and multiple years. The current process does not accommodate individual drugs and individual
biomarkers. Given no specific avenue for this process for a specific drug, it is difficult to engage with
review divisions at the pre-IND stage on this topic. The proposed BioMarker Qualification Request
would be managed by the appropriate review division, with consultation of the Office of Translational
10

Sciences, and could occur as early as the pre-IND stage. Managed well, this process could open the
door to drug development in some of the most difficult, serious diseases that are not being studied
frequently enough today.
V.

SUMMARY LIST OF QUALIFICATION CONSIDERATIONS OF A
DISEASE/DRUG/BIOMARKER SET FOR ACCELERATED APPROVAL IN RARE
DISEASES

At the Biomarker Qualification Request meeting, a briefing book would be prepared that would collect
the relevant data to support the qualification based on information set forth in FDASIA. The main
principle behind these considerations is that the more that is known about the pathophysiology of the
disease, the pharmacology of the drug, the science behind the biomarker, and the data in both animal
models and humans with the biomarker, the better the predictive value for reaching the “reasonably
likely to predict clinical benefit” standard required by the AA pathway. Currently, the information
required to support qualification of a biomarker as a surrogate primary endpoint for use in a pivotal
clinical study has not been well described and is developed on a case-by-case basis16,17. Unfortunately, a
“case-by-case” approach to review without any specified guidance does not provide adequate regulatory
predictability and diminishes the potential investment in early development work when the probability
of using AA is uncertain. In addition, the emphasis on the availability of prior clinical outcome data to
support the use of a biomarker as a primary surrogate endpoint renders AA essentially inaccessible for
many rare diseases.
Novel biomarker endpoints should be acceptable under the AA pathway when the novel biomarker can
be shown to be “reasonably likely” to predict clinical outcomes. However, achieving this standard has
been difficult because of the limited or lack of clinical outcome data. The rest of the data supporting the
relevance of a biomarker as a measure of a drug’s effect of a disease may have little impact on the
qualification process, despite its scientific relevance to reaching the “reasonably likely to predict”
standard. To solve this problem, a scientific framework should be developed that establishes a broader
set of scientific considerations for qualifying a biomarker as a surrogate endpoint, without requiring
prior clinical outcome data. This is particular important when such outcome data are impossible or
impractical to collect as noted in FDASIA.18
For this reason, FDASIA calls for the consideration of novel approaches to qualifying biomarkers on
pathophysiologic and pharmacologic criteria when other types of information are not available. The
development of clear qualification considerations will encourage better early development work by
assuring a more comprehensive evaluation of a biomarker at a pre-IND stage meeting. The data will
support the basic underlying science from disease to drug to biomarker in assessing “reasonably likely to
predict clinical benefit.”
Listed below are the proposed pathophysiologic and pharmacologic considerations to help drive
confidence in a biomarker’s predictive value. It should be noted that while these considerations are not
absolute requirements, they should be viewed as data points to support the use of a biomarker for AA in
that specific context of use. In the subsequent section, these considerations are further described in
detail.
11

Specific overall biomarker qualification considerations:
A. Disease Considerations
 Cause of disease clearly understood: distinct pathophysiologic cause based on a measurable
entity or single gene disorder
 Pathophysiology mechanisms relating to clinical outcome reasonably understood
 There is no known major alternative pathway of disease that is not assessable
B. Drug Considerations
 Drug mechanism of action is direct and known
 Drug pharmacokinetics, pharmacodynamics and metabolism are relevant to the disease
process being treated and can be accurately and readily measured
 Drug can be made reproducibly with appropriate quality to provide a consistent treatment
effect
C. Biomarker Considerations
 Biomarker has reasonable biologic stability and a direct relationship to an important
pathophysiologic pathway
 Sampling compartment for biomarker predicts the important disease compartment/tissue
 Biomarker assay is a valid and reproducible: Sensitive, accurate, precise and specific with a
sufficient dynamic range to calibrate biomarker change with pathology
 Accepted clinical physiologic measures may be considered predictive if the measure is
associated with major clinical problems in other diseases even if not considered a clinical
outcome themselves
D. Preclinical Considerations
 The model should be relevant to the pathophysiologic basis for the disease
 Magnitude and type of treatment effect is relevant and substantial relative to the human
disease state
 Preclinical treatment studies show dynamic dose-response relationship of the biomarker on
pathophysiology and/or clinical effect
 Preclinical studies show a meaningful clinical or physiologic effect on the disease if the
models reflect human disease reasonably accurately
 Measurement of the biomarker compartment should be confirmed to reflect tissue
compartments of interest
E. Clinical data considerations for biomarker qualification
 The biomarker predicts clinical severity or progression in a cross-sectional clinical survey,
natural history study or in preliminary investigational studies
 The dynamic range of the biomarker is sufficiently broad to assess the full spectrum of
severity or the appropriate difference between normal and disease states
 The biomarker shows predictive value for other similar rare diseases with comparable
pathophysiology
12

VI.

DETAILED DESCRIPTION OF THE SCIENTIFIC CONSIDERATIONS FOR THE
QUALIFICATION OF BIOMARKER ENDPOINTS USING DISEASE, DRUG,
BIOMARKER, PRECLINICAL AND CLINICAL INFORMATION

A. Disease considerations regarding the underlying disease pathophysiology that support
predictive value
To understand the scientific basis behind how a drug’s effects on a biomarker relate to disease
outcomes, a clear understanding of the pathophysiologic pathways involved in disease pathogenesis,
particularly related to the root cause of disease and their relationship to clinical outcomes are important.
The greater the clarity of the underlying scientific basis and the pathophysiologic processes for a
disease, the greater the confidence regarding the interpretation of a biomarker for this disease. The
utility of animal models such as gene knockout models for diseases of monogenic origin should be
considered particularly relevant in this context. Data from the clinical literature, in vitro studies, and
relevant comparable diseases should be provided as supportive evidence of the current understanding of
pathophysiology.

B.



The cause of the disease is clear. The specific and distinct root cause of disease, such as a
specific genetic defect, the presence of a particular autoantibody, or similar specific biological
change is known or understood based on basic science, preclinical or human data.



The pathophysiology of the disease is generally understood. When there is an understanding
of how biochemical or pathological processes result in a disease manifestation or group of
manifestations, the predictive value of the biomarker is increased. There may be aspects or
secondary pathways that have not been fully understood but at least one of the major pathways of
interest should be known.



There is no known major alternative pathway of disease pathogenesis that is not assessable.
When there is no evidence of an alternative disease pathway, the predictive value of a biomarker
is enhanced. The existence of alternative pathways that are poorly understood can cause
uncertainty regarding how a biomarker might impact overall outcome.19 However, exclusion of
all possible pathophysiologic processes is impractical and unnecessary, particularly if the root
cause of disease pathogenesis leading to important clinical manifestations has been identified.
Drug considerations for pharmacology of the treatment and its relationship to the disease
and biomarker

An understanding of the basic structure, delivery and actions of a drug can also enhance a biomarker’s
predictive value. Drugs with direct and well-understood mechanisms of action provide greater
confidence in the plausibility of a relationship between a biomarker effect and a clinical outcome. In
addition, information about a compound’s distribution at appropriate effective concentrations to sites of
action and the basis for this action can further support the likelihood of a cause-effect relationship.
Conversely, when the basis for the drug’s action is unclear or its distribution to the relevant site of action
cannot be established, then the basis for the drug’s action and the understanding of the changes in the
13

biomarker or disease are less certain. The data on the pharmacology of a particular drug, then, can help
to provide greater certainty that its action on a specified biomarker comes through a pathophysiologic
process associated with a major clinical outcome. What follows is a list of the specific drug
considerations that will enhance a biomarker’s predictive value.


The drug’s structure and identity are clear and its production at developmental scale in the
correct active form is reproducible, particularly in the critical aspects relevant to its absorption,
distribution, metabolism, and action, so that each study using the agent is relevant.



The mechanism of action of the drug is direct and understood. An understanding of the
drug’s mechanisms contributes to greater certainty about the interpretability of the relationship of
the drug’s action to the biomarker and relevant clinical outcomes become more predictable. Such
mechanisms could include replacement for a deficiency, enhancement of a deficient activity,
induction of a specific protein, or synthetic process by a mechanism demonstrable in vitro in cell
lines or in highly relevant animal models.



The specific pharmacologic action and activity of the drug can be demonstrated in in vitro
or in vivo systems to provide confidence on the effective concentration, distribution to the site of
action, uptake, and action.



Studies of the drug in models demonstrate relevant absorption, distribution, metabolism
and excretion (ADME). When the drug’s ADME are consistent with delivering the drug to the
site of action at a relevant concentration consistent with the plausibility of action on the target
tissue or tissues, this increases the likelihood that the effect on a biomarker and disease state are
connected to clinical outcome.

.

14

Table 1. Examples of pathophysiologic maps from cause, to pathophysiology to clinical physiology to clinical outcome
Disease

Mucopolysaccharidosis
type 1 (MPS 1)

Phenylketonuria

Myasthenia gravis

Duchenne muscular
dystrophy

Alpha Dystroglycan
related muscular
dystrophy

Fabry Disease

Cause

1o Pathophys.

2o Pathophys.

Clinical Physiology

Early Clinical

Late Clinical

(gene or protein level)

(cell level)

(tissue level)

(system/organ)

(integrated systems)

(final major outcome/events)

 GAG infiltration of
upper airway tissue
 GAG infiltration of
Accumulation of
lungs, liver, rib and
IDUA gene mutations
heparan sulfate and
spine development
Reduce iduronidase
dermatan sulfate GAG
 Synovial storage
enzymatic activity
in cells and tissues
 Thick heart valve
 Abnormal bone
formation








Sleep apnea, ↓O2
Impaired PFT
Joint ROM defect
Nerve compression
MR on Echo
Dysostosis multiplex







 Sleep deprivation
 Pulmonary
insufficiency
 Difficult hand mobility
 Enlarged heart

 Joint pain, stiffness,

contractures

 Reduced growth rate


Defect in PAH gene
that expresses PAH
that metabolizes Phe

↓Phe destruction leads  ↑Phenylalanine causes  White matter
 Mild cognitive
cytotoxic effects
abnormalities
to ↑Phenylalanine in
impairment
 Myelin abnormalities  Altered neuro function
blood

Antibody to the AchR

Inhibition of Achbased signaling

 Muscle weakness

Genetic defect in
dystrophin gene

Deficiency of
dystrophin protein

 Rupture of myofibrils
 Myopathy
 Centrilobular nuclei

Defective binding to
Hypo-glycosylation of extracellular matrix,
alpha dystroglycan
sarcolemmal
membrane instability

Mutation αgalactosidase gene

Accumulation GL3
in lysosome

 Stem cell regenerative
defect
 Muscle cell death
 Myopathy, fibrosis

 Multiple cells storage
(cardiomyocytes,
podocytes etc)
 Small vessels storage

 Drooping eyelids
 Weak legs
 Muscle weakness
 Heart abnormality
 Decreased FVC












Right heart failure
Hospitalization/oxygen
Increased respiratory infections
Unable to do ADL
Carpal tunnel syndrome requiring
surgery
Congestive heart failure
Wheelchair bound
Orthopedic interventions
Short stature

 Advanced cognitive impairment

 Difficulty keeping eyes
 Wheelchair bound
open for vision
 Difficulty walking
 Gower’s sign
 Difficult walking
 Decreased play

Impaired PFT

Decreased balance,
walking, climbing
stairs, rising from chair 
Decreased ejection

fraction (echo)
Brain abnormalities in 
a subset of patients
PNS
CNS

Eye

GI tract

Heart

Kidney

Skin

 Loss of ambulation
 Heart failure
 Death

Respiratory
insufficiency
Muscle weakness,
impaired mobility
Enlarged heart
Cognitive impairment






Ventilatory support
Wheelchair bound
Congestive heart failure
Decreased IQ and behavioral
abnormalities

Acroparesthesia
Heat intolerance
Angiokeratoma
Proteinuria
Arrhythmia






Renal failure
Cardiac insufficiency
Stroke
Deafness

15

Table2. Biomarker types organized by biological level and compared via stages with examples
Biomarker Type

Pathophysiologic
Process or Stage

General Examples

Specific Examples

Pros

Cons

Genetic marker

1o Cause

Presence of a gene mutation

CF mutations

Measure presence of
gene

Not a function

RNA/gene
expression

1o pathophysiologic

Expression of aberrant RNA
RNA splicing error
Presence of new gene expression

Friedrich’s ataxia
Fragile X

Direct impact on
gene expression

Unclear about
downstream effect

Enzyme or
protein level

1o pathophysiologic

Enzyme activity in tissue
Protein in circulation

Alpha-1-antitrypsin

Direct measure of
active compound

Difficult to verify
tissue effect

1o pathophysiologic

Blood level of an accumulating metabolite due
to a 1o block
Decrease in level of critical needed biochemical

Phenylalanine in PKU
BH4 in BH4 deficiency

Directly toxic
compound or active
compound

Not a measure of
tissue effect

2o pathophysiologic

Increase in secondary metabolite that is toxic or
part of pathophysiology but not from original
defect

Succinyl-lactone in
tyrosinemia I
Homogentisic acid in
alkaptonuria

Directly measure of
toxic effector

Cannot always
measure downstream
toxicity

Biochemical
Secondary
Biochemical

Biopsy

Ex vivo explant
X-ray/Imaging

Variability of
biopsies,
representative
sampling, variable
assay methods
Failed : questionable
validity of an ex vivo
assessment

2 pathophysiologic

Presence of abnormal cells or marker
Pathological change in structure

GL3 granules in Fabry
Dystrophin in Duchenne

Direct measure of
disease or absence of
protein

2o pathophysiologic

Evaluate a cell removed from the patient for a
phenotype or function

CGD/y-interferon

None

X-ray ricket score

Bone structure is
nature of disease

X-ray does not show
function exactly

FVC in CF
Muscle strength in DMD
or HIBM

Measure of a
physical function that
is directly relevant

Not strictly a clinical
outcome and hard to
gauge size of effect
with clinical outcome

Measure of a
patient’s function

Need to interpret
magnitude of change
for relevance to
patient

o

2o pathophysiologic

Clinical
Physiology tests

1 clinical effect

Clinical function

2o clinical effect or
intermediate
clinical measure

o

Bone structure
Presence of abnormal lesions
Change in size
Visual appearance like fundoscopy
Tests used in clinical evaluations of clinical
conditions dependent primarily on a single
tissue/organ
EMG, EKG, NCV, BAER, hand held
dynamometry

Tests that study integrated multiple body 6min walk test
systems/organs, Pulmonary function
Walking speed
tests, sleep apnea, muscle function

16

C.

Biomarker considerations regarding the specific scientific bases of biomarkers or
intermediary clinical endpoints that support the likelihood of predictive value

Biomarkers as surrogate primary endpoints have had both successes and failures in their ability to
accurately predict clinical benefit.19 The large variety of biomarkers and disease contexts can
make the systematic scientific evaluation process difficult, but there are specific points of
supporting information that can enhance the likelihood of real predictive value. While statistical
correlations established through large interventional outcomes studies have frequently been used to
develop predictive relationships, correlations alone do not provide predictive value for a biomarker
that can be evaluated based on its biology. The biological bases of biomarkers and their
relationship to the pathophysiology of disease represent a valuable and critical insight into
predictive value.
Biomarkers can represent any point along the pathophysiologic process, from primary disease
cause to just before clinical outcome, and different considerations exist for different types of
surrogates (see Table 1 for examples). A map of the pathophysiologic pathway from primary
cause, primary and secondary pathophysiologic processes, primary and secondary clinical effects,
early clinical outcomes and final clinical outcomes should be prepared to help establish the basis
for the relationship of the biomarker, and to provide a structure for verifying the degrees of
evidentiary support that exist for these steps. Understanding the precise process level for the
biomarker and the type of biomarker is important in guiding the type of information required about
a biomarker and its position within the pathophysiology of the disease and reflecting the drug’s
mechanism of action (Table 2).


The biomarker is directly in line within the pathophysiological map for at least one of
the major pathophysiologic pathways. This is a critically important factor as reviewed by
Fleming. 20
o An effect on a biomarker close to the primary pathophysiologic cause of the disease is
likely to be predictive of a meaningful impact on the disease and is less prone to
unknown links or secondary parameters.
o If the biomarker is part of a secondary pathophysiologic process, the process must be
demonstrated to be important and critical to the clinically important pathophysiology,
and the links to the primary pathophysiology should be demonstrated in model studies.
o

A combined effect on multiple secondary pathologic biomarkers, particularly in the
setting of an effect on a primary pathophysiologic mechanism, should provide greater
confidence in predicting a clinically meaningful effect.

o A biomarker or intermediate clinical biomarker close to a major pathophysiologic
clinical outcome should also be considered relevant to and predictive of a specific
clinical outcome.
o A biomarker should be matched with the most appropriate stage of disease and used in
that context. A biomarker of early disease pathophysiology may no longer be relevant

17

once significant, irreversible damage to an organ has occurred (e.g. malformation or
failure), and conversely, a biomarker of late disease pathophysiology may not be
informative early in the disease course.
o The biomarker should not have other unpredictable parallel pathophysiologic pathways
that could confound the interpretation of the biomarker. This could include a pathway
for metabolizing the biomarker or creating the marker that is a normal biologically
variant factor. Controls for this issue should be considered in study designs.


The sampling compartment for the biomarker predicts the disease compartment. The
site of sampling, whether blood, urine, cerebrospinal fluid, an X-ray/image, or a biopsy,
must reflect the relevant disease compartment, even if sampling the disease compartment is
not possible. For example, if a blood biomarker were being studied for a CNS indication,
studies in model systems should show that the blood compartment sufficiently reflects the
brain disease state. Confirmation within a clinical cross-sectional survey and within
preclinical models is valuable. The disease process could occur in multiple compartments
beyond the sampled compartment, but in this case, it should be shown that the sampling
compartment is relevant and correlated with other compartments across a variety of
therapeutic situations or that it is at least predictive of the important compartment in
preclinical models.



Changes in the biomarker are sensitive and specific to changes in patient condition
with a sufficient dynamic range between normal and abnormal patients. The assay
should be able to distinguish abnormal from normal with sufficient precision and accuracy
to be a reliable tool in the clinical setting. To ensure that various gradations of abnormality
in specimens are accurately detected, the difference between abnormal and normal should
be large relative to the standard deviation or coefficient of variation of the assay in a
clinical study setting. When possible, these data should come from untreated patient
specimens and be comparable to normal persons of similar age and type. If collected in
preclinical models, the ability to detect an abnormal result with sufficient dynamic range
should be convincing. The dynamic range must be adequate to assure that biological
variation between patients or the assay methodology could not overshadow the relevant
changes in the biomarker. Changes in preclinical models with treatment should substantiate
the characteristics of dynamic range and responsiveness to change over time.



The biomarker should demonstrate reasonable biologic stability. If the disease state is
relatively stable and no change in physiology is occurring, the biomarker’s relative level
should not dramatically change. Obtaining these data may require testing a group of
individuals or preclinical models over a significant timeframe.



The assay methodology for measuring the biomarker should be validated using
reasonable and relevant criteria. In order for approval to be based on a biomarker, the assay
must reliably measure the biomarker’s value in humans. This is of particular importance for
tissue biopsy analysis, as well as for other techniques in which complex samples are
18

analyzed using tools that may be prone to variation from the sampling process, the reagents
or signal detection.


Accepted clinical physiologic measures may be considered a predictive biomarker if
they are directly relevant to major clinical problems in the rare disease. Many clinical
physiological tests are used to assess and treat a specific clinical condition in common and
rare diseases. These tests are routinely accepted for use in clinical practice for the
diagnosis and management of clinical conditions in other common diseases with similar
pathophysiology. For example, measures for joint range of motion, sleep apnea, heart
enlargement by echocardiography, and similar tests have been associated with clinical
outcomes and are actively used to initiate treatment in common diseases. If the
comparability between the pathophysiologic processes, disease characterization, or
outcomes can be demonstrated between the rare disease and common diseases, these tests
should be considered to have predictive value. The magnitude of the disease present and
changes expected should be shown to compare well with clinically significant changes
observed in other common diseases where the clinical physiologic measure is used.
Although the tests may not have been previously used as primary endpoints, if national
standards for the diagnosis and management of the disease condition exist for other
diseases, then the comparability of the disease process need only be supported through the
use of scientific literature or testing in order to support the use of the test as a biomarker in
a rare disorder. Examples of such tests include pulmonary function tests, sleep apnea
testing (e.g. apnea-hypopnea index), echocardiography (assessing cardiomegaly or poor
ejection fraction), nerve conduction velocity, or similar clinical physiology tests that
capture important clinical physiology used for the diagnosis and management of
conditions. Included in this category could be physiologic measures normally accepted as
clinical endpoints but for which the magnitude of the change might be too small to
represent a clinical benefit, thought the direction of the change is positive for the patient.

D. Preclinical Evidence to support use of a disease/drug/biomarker set for accelerated
approval
For rare diseases, model disease treatment data are often essential to demonstrating an important
effect of a treatment on a disease. The proper conduct of preclinical studies can be important to
establish a platform of data and a framework for understanding how the disease and drug interact,
as well as how the biomarker’s behavior is predictive. Certain critical sets of data should be
obtained to support the biomarker in in vitro and preclinical model experiments. The more
appropriate and comparable the model is to human disease, the better it may predict human
disease. In the absence of strongly predictive clinical models, a model that demonstrates the
treatment and biomarker effects at the level of pathophysiology and pharmacology is sufficient, as
preclinical models often do not express every aspect of clinical disease or progression in the same
manner as in humans. Clearly, data derived only in animal models of less certain relationship to a
disease, must be supported by other types of data in order to allow qualification. Exclusive reliance
animal models is not optimal as this can lead to a failure in the predictive value of the biomarker
when the model does not fully reflect the human disease state.
19

The best possible data setting, whether in vitro or preclinical, should be sought to support the
considerations provided below. However, for some diseases there is no opportunity to make a
preclinical model and only in vitro models may be available or valid. When measuring the clinical
effects in the models is impractical or irrelevant, the data on the preclinical models can be based on
the pharmacologic or pathophysiological changes in the model.
Key preclinical data to support the predictive value:


The model should be relevant to the pathophysiologic basis for the disease. This should
be accomplished through a comparison of pathophysiology/genetics and using microscopic,
biochemical and (if present) clinical disease assessments. Clinical disease varies in models
and may not be the same in every respect due to the differences in species and effects of
changes, but applicability of the model can still be demonstrated if relevant
pathophysiologic changes can be assessed.



There should be a broad and dynamic dose-response relationship over the wide range
of disease and treatment effects demonstrating how changes in the biomarker reflect
changes in the model disease level. When possible, it is critically important to establish the
level of biomarker improvement associated with a potentially clinically meaningful change
in disease level (Figure 1). The dose-response relationship should also be established for
suboptimal therapeutic dose levels to demonstrate the biomarker’s sensitivity in evaluating
drug effects that are low and not likely to predict benefit. The impact of any adverse
responses (e.g., an immune response to the therapeutic agent) should be evaluated for its
impact on the biomarker and treatment effect to show how the biomarker predicts the
treatment effect at the biochemical or pathologic level when altered by this condition. For
example, if antibodies to a drug interfere with efficacy, the biomarker should reflect this
decrease. If there is no clear relationship between the amount of model disease reduction
and clinical outcome, then a relative comparison toward the degree of normalization of the
pathophysiology should be used as the best estimate of a meaningful treatment effect.



Measurement of the biomarker compartment should be confirmed to reflect tissue
compartments relevant to the disease state. A comparison of the dose-response
relationship should establish the relationship between the sampled compartment levels and
the pathology in tissues associated with adverse clinical outcomes. For example, a blood
test should correlate with muscle pathology for a muscle disease treatment. A spinal fluid
measurement should be shown to reflect the brain pathology, in a brain disease treatment.
In particular, suboptimal levels of treatment should be used to determine whether the
biomarker reasonably reflects the pathologic outcome for tissues relevant to disease
outcomes.



Biomarker changes predict clinical changes in existing relevant models. Although
many preclinical models do not show comparable clinical disease to humans,
demonstration of the predictive value of the biomarker on treatment outcome in clinical
measures can still help provide support for greater predictive value.
20

Figure 1. Possible dose response relationships between a biomarker and clinical status Understanding the
biomarker-disease relationship is important and can be established to some degree in preclinical studies, with support
from cross-sectional or natural history studies. The graph shows how different shapes of the curve can provide very
different interpretations of the change in clinical status (C1, C2 and C3) for a similar change in the biomarker (b1, b2).
Establishing this relationship is an important part of interpreting the change in a biomarker in a clinical study setting
and having this data is therefore important in the qualification process.

E.

Available clinical evidence to support use of a disease/drug/biomarker set for
accelerated approval

The collection of clinical data has been an especially difficult barrier to access to the AA pathway
for rare diseases, due to lack of historical data, insufficient patient numbers, and time to establish
firm relationships between a biomarker and clinical outcomes. Optimally, clinical data with an
effective treatment are required to develop a predictive relationship for clinical outcomes. When
clinical outcome data does exist for the predictive value of a biomarker in a rare disease, these data
are important to the assessment in the qualification of a biomarker endpoint for use as a surrogate
primary endpoint. In most cases, however, longitudinal treatment studies with other agents have
never been conducted, and there is limited useful clinical outcome information available from
natural history studies. In these cases, other types of data must be sought when practical to support
the qualification of a surrogate.
In the absence of clinical outcome data, significant information can be obtained from crosssectional survey studies of patients using a biomarker and known clinical condition and assessment
measures. These studies can be conducted prior to the investment in manufacturing a drug, or
before clinical development has begun, to assist in the determination of the predictive value of the
biomarker when reasonable and practical. Ideally, the studies should include patients of different
ages, severity and stages of disease. This cross-sectional survey data can often be larger in patient
number and broader in scope than the type of data provided by a natural history study, especially if
21

the long-term retrospective data can be also collected during the cross-sectional evaluation.
Natural history data can be enormously helpful in assessing a disease and planning a development
program both in supporting a biomarker and in understanding the disease. However, such studies
are costly, take a long time to complete, and can be prone to selection bias, making it exceedingly
difficult to collect the kind and quality of data required for assessment of biomarkers in a time
frame that allows for a real impact on rare disease programs. Nonetheless, both natural history and
cross-sectional data can be very useful in supporting the clinical relevance of a biomarker
endpoint.
The clinical data to support a biomarker should focus on the following important considerations:






VII.

The biomarker predicts clinical severity or progression as assessed using other clinical
measures of disease. A cross-sectional clinical survey study or retrospective medical chart
survey or natural history study can yield data to support a relationship between the
magnitude and change over time of the biomarker measures and
severity/progression/disease level by other clinical measures. The survey can provide
multiple types of data to establish a reasonable relationship and dynamic range for the
biomarker and a clinical parameter. This can be done early in a program before a drug
exists or before an IND is submitted.
The sensitivity and dynamic range of the biomarker is sufficiently broad to elucidate
the important part of the spectrum of severity of the disease using the biomarker. Patients
with mild disease or severe disease can be distinguished from each other and from normal
patients.
The biomarker shows predictive value for other similar diseases. Clinical data with
approved drugs from similar diseases, for which studies have been completed, show a
reasonable relationship between changes in the biomarker and changes in clinical
endpoints. This may rarely occur, but if adequate and reasonable parallels for another rare
disease with similar mechanisms exist, these data may be useful. This is not to suggest that
the use or failure of biomarkers in complex multi-genic common diseases should
necessarily be applicable to results in diseases with far more specific and clear underlying
pathophysiology. One example of is the use of plasma ammonia level to approve several
drugs that reduce ammonia in patients affected by defects in the urea cycle.
CLINICAL TRIAL DESIGN CONSIDERATIONS

In many rare diseases, traditional randomized concurrent controlled studies may be difficult to
conduct, both because of the rarity of the disease and the ethical issues associated with the study of
some devastating rare diseases. Alternative study designs may be deemed acceptable, but the
sponsor will need to provide sufficient information on safety and efficacy to provide reasonable
assurance that the criteria for AA have been satisfied.
Some key considerations for the study design and analysis are as follows:

22

A. Pivotal Study Trial Design: Concepts for studies using a primary surrogate marker
endpoint


Randomized, concurrent controlled studies are preferred when feasible and
appropriate. The use of placebo control groups is the gold standard for quantitating the
magnitude of the treatment effect when no other treatment exists, and should be
implemented except in exceptional circumstances. In some situations, blinding a treatment
is impossible or it may be unethical either due to the type intervention itself, or the
withholding of potential other care with irreversible devastating consequences cannot be
justified. When placebo control groups are not feasible or ethical, historical control data
can be considered with a single-arm treatment study (e.g. Myozyme for Pompe) or an
unblinded parallel control group when blinding is difficult. These study design alternatives
have been used in oncology more frequently, but there are situations in which good
objective data can be obtained to support the assessment of efficacy. In this open-label
setting, the appropriate blinding of the analysis of the biomarker as the surrogate primary
endpoint may be appropriate to assure an objective assessment of efficacy can be made.



An adequate assessment of safety is still required. Given the extraordinarily small
patient populations for some disorders, it is important to assure the best possible safety
assessment by understanding potential adverse safety pathophysiologies that may be
operative based on the drug’s mechanism of action and the underlying disease state. AA
via a biomarker may allow for smaller clinical studies, but the study size of the exposed
safety population should also be considered in the context of the expected population
exposure and not just in terms of a study size sufficient to demonstrate efficacy. When the
number of existing patients is small, limiting the safety dataset in a study, a longer period
of observation can provide some additional support for safety. Therefore, it is advisable to
continue treatment of early treated subjects in extension studies to provide this additional
safety information.



Off-target adverse activities need to be evaluated for the drug. Studies using in vitro,
toxicologic, or model pharmacologic studies that suggest the existence of any significant
alternative pathways or adverse physiologies not reflected in response to the drug or in the
specific biomarker proposed for use as a surrogate, needs to be evaluated in the studies. It
is important to assess whether these independent adverse effects via another mechanism on
the patient which might lead to harm that outweighs the expected benefit. A drug that
might be predictive of clinical benefit through one pathway could still have a negative
benefit-risk ratio if an alternative unmeasured adverse action occurs that has a greater
impact on risk than the primary benefit. If the alternative adverse pathway can be measured
and followed in clinical studies, then the confounding effect on benefit may be quantified
during the clinical program and included in the benefit-risk analysis.



AA does not require internal validation of the biomarker. The concept of AA is the
acceptance of a biomarker as a primary surrogate endpoint without the requirement for
positive clinical outcome data within the same study. It is still advisable to measure
23

clinical endpoints in these studies to begin the assessment of the surrogate and to provide
supportive information about treatment and safety. Observations of trends in clinical
endpoint data can be supportive, but for some diseases that have prolonged courses, there
can be no expectations of supporting clinical endpoint data. In addition to the primary
surrogate endpoint, other biomarker and clinical physiology endpoints can help substantiate
that the expected biology is occurring.


Alternative study designs are potentially acceptable, but require careful planning.
o Open-label studies using no parallel control groups or conducting within-patient
comparisons, have substantial limitations on the interpretation of efficacy and safety. If
used, the design and conduct should consider sufficient safeguards and assessments in
place to ensure that a reasonable assessment of drug-related safety effects can be made
and efficacy assessments are objective. The interpretative value can be improved by
conducting blinded readings of data when possible in this setting.
o Alternative designs including placebo-run in periods, randomized withdrawal or
other alternative within patient controlled designs. Studies based on alternative
controlled designs or the N of 1 concepts of within-patient evaluations under conditions
of treatment or placebo or treatment withdrawal can be powerful in assessing efficacy
and safety, despite a small cohort of patients.
o Historically controlled studies can be difficult to interpret and require extensive
planning and rigorous execution to be accepted. Crucial aspects include controlling for
severity or ascertainment of the subjects in the control versus the treated group,
differences in ancillary medical care, and the differences that may occur simply as a
result of participating in a study. To be successful, the natural history control group
data must be robust with adequate assessment of the disease to verify comparability
with the treated patient group. This assessment could, to extent possible, include
genotypes, phenotypes, age of onset, degree of medical care utilization, physiologic
assessments of disease severity and carefully recorded data on ancillary treatments
applied. Matched control groups or other designs as close as possible to the
contemporaneous population would be preferred in order to minimize the effect of
changes in medical care or attitudes about treatment that might have an effect on
outcomes.



Sufficient long-term data on safety and efficacy is required. Given the smaller
populations of patients affected by rare diseases, longer-term data is critical in verifying the
durable effect on a particular biomarker, the safety of repeated treatments, and the potential
evolution of adverse responses or other physiologies to have occurred and their effects fully
realized. For acute diseases, this is not a relevant consideration.

24

B.

Post-marketing confirmatory studies in rare disease indications

The approval of a therapeutic via AA requires that additional data be collected in the postmarketing setting to support the efficacy and safety of the drug. In rare diseases, post-marketing
studies can be difficult to conduct, especially if the confirmatory study is expected to be placebocontrolled. Important clinical outcomes may require a long period of observation, making placebocontrolled studies impractical. Careful planning and effective designs to support the clinical
meaningfulness of the data are required.
Recommended steps and considerations for post-marketing confirmatory studies for rare disease
indications include:


A confirmatory study plan should be discussed early in the development when the AA
pathway is being considered. To assess the complete development pathway regarding the
provision of adequate safety and efficacy information, the context for post-marketing
studies and the type of data to support efficacy and safety should be carefully considered.



Placebo versus non-placebo controlled confirmatory studies. While placebo-controlled
studies provide the greatest assurance of efficacy, they may not be appropriate, ethical, or
feasible given the long timeframes for disease progression, the frequency of events in some
diseases, or the ability to recruit patients with severe and life-threatening diseases. If
placebo-controlled studies are possible, they should be conducted early in the launch of the
program and could include periods of placebo treatment for patients prior to being
converted to the marketed product. This would allow for the collection of more clinical
data on shorter-term outcomes. If placebo-controlled studies are impractical due to the
timeframe or nature of the clinical benefit, an effective plan on how to obtain objective
data, or sufficiently robust long-term data to support efficacy on hard clinical endpoints, is
important to the AA pathway. For diseases with long-term neurologic progression,
adequate untreated control subject data with appropriate data on patient comparability and
phenotype/genotype are important.



The size, length, and scope of the confirmatory program should be sufficient to
compensate for the limitations of the patient population. For small patient populations,
the post-marketing program should be designed to allow for the participation of enough
patients to verify the clinical outcome. For rare diseases, this may require longer-term
observation to help support the magnitude of the clinical impact. In studies without parallel
control groups, the magnitude or nature of the clinical benefit might need to be sufficient to
overcome the potential for bias in assessments. When patient numbers are extremely
limited, the inclusion of a large fraction of the patients may be required and can be
balanced with a design that allows the major clinical assessments to be collected at
intervals during efficient clinical disease surveys of treated patients. This can eliminate the
need for all of the patients to participate in an ongoing clinical study, which may be
impractical or impossible to conduct. The collection of high quality outcome data from
previously untreated patients may be a critical element in supporting the determination of
25

efficacy in long, open-label post-marketing programs, and this possibility should be
discussed with the FDA as part of the AA discussion.
VIII. EXAMPLES OF SUCCESSFUL USE OF BIOMARKER ENDPOINTS DURING
DEVELOPMENT OF RARE DISEASE DRUGS
A number of drugs have been approved for the treatment of rare diseases using biomarker-based
primary endpoints. In most cases, the standard approval pathway was used and involved some
degree of FDA flexibility. 8 AA was used in some cases. While these examples provide support for
the types of information that has been successful in achieving approval, they may not necessarily
reflect the full range of information needed to successfully develop a rare disease therapeutic.
A.

Ammonia in urea cycle defects: glycerol phenylbutyrate

The urea cycle defects cause a block in the process that disposes of ammonia as urea, and results in
elevation of toxic ammonia levels. The defect is directly in the pathophysiologic process that
creates the biomarker ammonia, and ammonia is intrinsically toxic in excess. A series of drugs that
divert ammonia via glycine or glutamine depletion have been approved using ammonia levels as
the primary indicator. The most recent example is glycerol phenylbutyrate for urea cycle disorders
that was approved with a randomized double cross-over clinical study comparing it with the
approved original phenylbutyrate. The control over ammonia over a 24-hour period was compared
with the active control treatment. Given the history of approvals for drugs intended for urea cycle
defects using ammonia control, and the use of ammonia control in other diseases such as liver
diseases, there has been a precedent for ammonia control as a biomarker endpoint.
B.

Phenylalanine for phenylketonuria (PKU): Sapropterin dihydrochloride

Phenylalanine is increased in large excess in patients due to defects in the phenylalanine
hydroxylase enzyme. This enzyme is primarily responsible for initiating the oxidative degradation
of phenylalanine, and without it the phenylalanine level rises many-fold above normal.
Phenylalanine has been shown to be directly toxic to neurons and has been shown to predict IQ
outcome in multiple clinical studies of another therapy, dietary restriction of phenylalanine. The
use of phenylalanine blood level was accepted as a primary endpoint in an 89-patient randomized,
placebo-controlled clinical study in the sapropterin dihydrochloride development program for the
PKU program. Although the mechanism of action of sapropterin was different from the compared
diet therapy used to qualify the biomarker endpoint, its mechanism was demonstrated by labeling
studies to involve the restoration of the normal oxidative metabolic pathway.
C.

GL3 storage granules in the vascular endothelial cells for Fabry disease: Agalsidase beta

Fabry disease is a lysosomal storage disorder caused by a defect in the alpha-galactosidase gene
and results in storage in many cell types. The disease has a pronounced vascular phenotype with
disease most commonly in the kidney, heart and brain. Storage within the endothelium is directly
responsible for these vascular problems. In the development of an enzyme replacement therapy for
Fabry, it was shown that the enzyme can clear the storage and return the endothelial cells to near26

normal if not normal status in terms of GL3 granules using renal biopsies and a scoring system.
This pathologic endpoint was used in the approval of the enzyme therapy agalsidase beta in a 58patient randomized placebo-controlled study. The challenge was that biopsy data can be quite
variable in sampling and the scoring can be subjective, so extensive work on multiple biopsies and
scoring systems and adjudication of results was needed to develop and gain agreement on the
biopsy and the analysis of the pathology. The confirmatory study for this approval had some
complications and though the result is debated, agalsidase beta did appear to reduce the major
event rate of Fabry disease as expected.
D.

Hemoglobin and platelet count for Gaucher disease: Alglucerase

Patients with Gaucher have lysosomal storage in the macrophages which leads to a large spleen
and sequestration of red cells and platelets. Anemia and thrombocytopenia can be severe and be
associated with bleeding problems. Alglucerase was studied in a 12-patient single-arm, open-label
study and shown to improve hemoglobin and platelet counts, as well as reduce spleen and liver
size. Although the magnitude of the changes had not been shown to be clinically meaningful
specifically in these disease patients, it was assumed based on general medical experience that low
hemoglobin and low platelets are problems and that magnitude of the resolution of these problems
observed should be beneficial the patient.
E.

Alpha-1-antitrypsin level for Alpha-1-antitrypsin deficiency disease

Patients with alpha-1-antitrypsin deficiency disease have excessive protease action that results in
pulmonary disease like emphysema over many years, and can also be associated with liver disease.
Blood-derived replacement therapy was successfully approved by demonstrating the reasonable
restoration of blood levels of the protease inhibitor, although no direct proof of inhibiting proteases
at the tissue level was demonstrated. An open-label study of one form alpha-1- proteinase inhibitor
(Prolastin) was studied in 19 patients over 24 weeks and shown to achieve a serum level exceeding
80 mg/dl and bronchoalveolar lavage demonstrated that the level in the plasma compartment was
reaching the alveolar space.
F.

Deferasirox for reduction in iron overload in beta-thalassemia

Deferasirox was approved using a liver biopsy measure of iron as a primary biomarker endpoint
for the reduction in iron overload derived from transfusion therapy in the red cell disease, betathalassemia. In this program, a randomized, open-label study comparing standard therapy with
deferoxamine compared the iron content in a liver biopsy at 12 months to baseline content. Liver
iron content is a measure of total iron load and the drug’s action is the direct removal of iron via
the urine. The biomarker is in the liver, which is an important target organ and therefore an
appropriate tissue compartment for measurement. The precision of biopsy methods can be
challenging in general, and this study randomized a total of 586 patients to achieve their
demonstration of efficacy over 4 dose levels.

27

IX.

CONCLUSIONS AND RECOMMENDATIONS

The effective utilization of the AA pathway for rare diseases will require development and use of a
scientifically sound framework of data for qualifying biomarker endpoints allowing the practical
use of biomarkers as a measure of efficacy. A scientific framework with defined sets of supporting
data should allow the beginning of a more structured approach to qualifying biomarkers for use in
pivotal studies of rare disease treatments and ensure a wider array of important considerations are
included in this process. The proposed data that help qualify a biomarker will cover the disease,
the drug, the biomarker, preclinical data, and clinical survey or natural history data. It is extremely
important to recognize that clinical outcome data for a novel biomarker is rarely available or
plausible and therefore a systematic process that builds support for the predictive value of a
surrogate using data that is available will allow more investment innovative treatments for rare
diseases.
This proposed scientific framework is a first step and will need further evolution and development
going forward with experience. Regardless, the judgment and insight of experts is needed to assess
the scientific support for a biomarker, to weight the importance of the various results and make a
structured decision regarding qualification. The evaluation process should also consider the
benefit-risk assessment for that disease as a critical factor to managing the qualification process.
With a better defined process, there should be more opportunities to advance therapies into
development.
A disease survey early in development can establish the factors that make AA more critical to the
development of a disease and provides insight on the endpoints and benefit-risk assessment for a
disease. This information can be helpful in understanding the degree of flexibility appropriate for
a rare disease and the relative impact of the disease relative to potential treatment. Qualification of
a biomarker for accelerated approval should include the considerations from the severity and rarity
of the disease being studied.
To make the qualification of a biomarker most useful and to enable early investment in the
development of treatment, a Biomarker Qualification Request process should be available at the
pre-IND stage for a treatment intended to treat a rare disease. For this pre-IND Biomarker
Qualification Request, the sponsor should provide a briefing book containing a disease survey, the
analyses of a disease/drug/biomarker set by the proposed criteria and verified using preclinical
models, as well as any clinical survey/natural history data on the biomarker. The review and
approval of a potential biomarker endpoint at the pre-IND stage of development before the
investment in drug manufacturing and clinical studies will help support the early investment in the
most rare and difficult diseases. If this can be achieved, then greater investment in developing
treatments in rare diseases, especially with small populations and complex disease manifestations,
will occur and new treatments will finally be developed for so many more untreated rare diseases
using the AA pathway.

28

X.

REFERENCES

1

National Institutes of Health. National Center for Advancing Translational Sciences, FAQ About Rare Diseases.
(Accessed 22 April 2013).
2
U.S. Food and Drug Administration. Search Results for Orphan Drug Designations and Approvals.
(Accessed 22 April 2013).
3
Kathleen Sebelius. (2011). Building a strong foundation for a new century of treatments and cures. Health Affairs
30(1):81-83.
4
21 U.S.C. §356 – Fast Track Products
5
Ibid.
6
Government Accounting Office. FDA Needs to Enhance Its Oversight of Drugs Approved on the Basis of Surrogate
Endpoints. GAO-09-866, Washington, DC, October 26, 2009.
7
Brigitta E. Miyamoto and Emil D. Kakkis. (2011). The potential investment impact of improved access to accelerated
approval on the development of treatments for low prevalence rare diseases. Orphanet Journal of Rare Diseases 6:49.
8
Quantum of Effectiveness Evidence in FDA’s Approval of Orphan Drugs Cataloguing FDA’s Flexibility in
Regulating Therapies for Persons with Rare Disorders F. J. Sasinowski, M.S., M.P.H., J.D., 2011, DIA J. 46 (2), p
238.
9
Thomas R. Fleming and D. L. DeMets. (1996). “Surrogate End Points in Clinical Trials: Are We Being Misled?”
Annals of Internal. Medicine 125:605-613.
10
Institute of Medicine: Board on Health Sciences Policy. Rare Diseases and Orphan Products: Accelerating Research
and Development. Washington, DC: The National Academies Press, 2011
11
FDASIA P.L. 112-144, July 9, 2012
12
Ibid.
13
Ibid.
14
Brigitta E. Miyamoto and Emil D. Kakkis. (2011). The potential investment impact of improved access to
accelerated approval on the development of treatments for low prevalence rare diseases. Orphanet Journal of Rare
Diseases 6:49.
15
Institute of Medicine: Board on Health Sciences Policy. Rare Diseases and Orphan Products: Accelerating Research
and Development. Washington, DC: The National Academies Press, 2011
16
Institute of Medicine: Committee on Qualification of Biomarkers and Surrogate Endpoints in Chronic Disease,
"Perspectives on Biomarker and Surrogate Endpoint Evaluation: Discussion Forum Summary. Washington, DC: The
National Academies Press, 2011.
17
Institute of Medicine: Committee on Qualifications of Biomarkers and Surrogate Endpoints in Chronic Disease,
Evaluation of Biomarkers and Surrogate Endpoints in Chronic Disease. Washington, DC: The National Academies
Press, 2010.
18
President’s Council of Advisors on Science and Technology. Report to the President on Compelling Innovation in
Drug Discovery, Development and Evaluation. Executive Office, Washington, DC, September 2012.
19
Thomas R. Fleming. (2005). Surrogate Endpoints And FDA’s Accelerated Approval Process. Health Affairs
24(1):67-68.
20
Ibid.

29