Comment Hyman Attachment 2 8-27-13

APPENDIX 1

1. SIRTURO (bedactuiline)
This December 28, 2012 approval for treating multi-drug resistant tuberculosis
(MDR-TB) was based on a surrogate of time to sputum culture conversion.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
"Overall mortality still exceeds 10%, with a range of 8 to 21% for patients
enrolled into good treatment programs." (Medical Review, Dec. 26, 2012, at p.
22).
b. Rarity of the Condition
FDA granted Sirturo orphan drug designation on January 10, 2005.
Furthermore, in FDA's determination that the time to sputum culture
conversion is an acceptable surrogate on which to base accelerated approval, it
appears that FDA may have taken into account specifically the rarity of MDRTB in this country in that FDA acknowledged that: "In the United States, the
total number of MDR-TB cases has fluctuated from 88 to 132 cases [since]
1993, with 88 cases reported in 2010." (Medical Review at p. 22).
c. Lack of Available Therapy
"Treatment of MDR-TB is more complex (than treating drug-susceptible
TB or DS-TB) and prolonged and typically has a favorable outcome rate [of
only] 41-70%. Cases of MDR-TB are currently treated with at least five
second-line anti-TB drugs for an extended period of time that may last up to
two years . . . The challenges of the treatment of MDR-TB include toxicities of
the drugs, decreased potency, cost (50-200 times more expensive than DS-TB)
and the need for possible hospitalization." (Medical Review at p. 22).
d. Use of External Expertise
FDA did turn for external expertise to the Anti-Infective Drug Advisory
Committee, which on June 3, 2009 "voted 18 to 1, recommending that sputum
culture conversion . . . could be used as a surrogate . . . therefore, the
committee recommended that approval of an antimycobacterial drug could be
done under Subpart H regulations (Accelerated Approval) using sputum
culture conversion as a surrogate endpoint. Further, traditional endpoints used
to evaluate treatment response such as relapse, failure and mortality should still
be used for traditional approval." (Medical Review at p. 28).

A2

Part 2. Understanding of the Disease Process

In this case, the pathophysiology of MDR-TB is well understood.
Part 3. Understanding of the Relationship Between Sputum Culture Conversion
and Relapse, Long-Term Response and Mortality

Epidemiologic evidence exists that supports the relationship between sputum
culture conversion and clinical outcome, in particular, mortality. See Shama D. Ahuja et
al., "Multidrug Resistant Pulmonary Tuberculosis Treatment Regimens and Patient
Outcomes: An Individual Patient Data Meta-analysis of 9,153 Patients," 9(8) PLOS
Medicine e 1001300 (2012).
Part 4. Clinical Evidence of Sirturo's Effect on Sputum Culture Conversion and on
Relapse and Mortality

The FDA Medical Reviewer noted the existence of the epidemiological evidence,
but stressed that the clinical evidence provided by the sponsor both on the surrogate and
on traditional endpoints of clinical benefit, especially mortality, would be "most
persuasive." In this case, the Medical Reviewer listed these traditional endpoints as
relapse, long-term response, and mortality. (Medical Review at p. 16).
There were two Phase 2 clinical trials that comprised the clinical evidence for this
drug on the surrogate and on clinical benefit, but only one of which was considered to be
the single, pivotal trial: Study C208 Stage 2. Study C208 Stage 2 was a randomized,
double-blinded, placebo-controlled trial with a 24-week treatment period in which both
the drug and "placebo" arms received an optimized background regimen. (Statistical
Review, July 26, 2012, at p. 6).
a. Sputum Culture Conversion
The primary endpoint, which was the surrogate endpoint, of the time to sputum
culture conversion was highly statistically significant (p-value of 0.0005) (N=160
randomized, with 67 and 66 subjects in the drug and placebo arms in the mITT analysis,
respectively). Sputum culture conversion at week 24 was a key secondary endpoint (as
well as another supportive measure of the surrogate endpoint of sputum culture
conversion) and it too was statistically significant (p-value = 0.014) with 78% and 58% of
drug and placebo arm subjects, respectively, achieving sputum culture conversion at week
24. (Statistical Review at p. 6). "Lastly culture conversions data after all patients
completed 72 weeks in the study showed a statistically significant but diminishing
improvement in the time to sputum culture conversion for [Sirturo-]tested patients
compared to placebo-tested patients." (Medical Review at p. 44).
b. Relapse and Mortality

A3

Relapse is a "traditional" measure of clinical benefit. The Medical Reviewer notes
that in "the mITT population, five subjects (7.6%) in the [drug] group and eight subjects
(12.4%) in the placebo group experienced relapse . . . [However,] the subjects in the
placebo group appear to take a longer time from culture conversion to relapse than those
in the [drug] group." (Medical Review at pp. 59-60). Therefore, the Medical Reviewer
conducted an alternative analysis and in this analysis, "the two treatment arms become
more comparable with respect to relapse with 5 relapses on [drug] and 4 on placebo."
(Medical Review at p. 60).
Survival is the most objective and clinically meaningful benefit in MDR-TB. In
the pivotal study, 9 of 79 in the drug arm died (11.4%) compared to 2 of 81 (2.5%) in the
placebo arm. (Medical Review at p. 70). Both placebo subjects died of TB as did 5 of the
9 subjects in the drug group. (Medical Review at p. 70). Signals of QT prolongation and
serum transaminase elevation, with one death due to liver injury in the drug arm were also
observed. (Medical Review at pp. 70-71).
In the "summary and conclusions" section of the statistical review FDA observed:
"There was a statistically significant increase in mortality in the [drug] group. Despite the
observed treatment benefit in time to culture conversion, it did not lead to a benefit in
patient survival. This was a major concern for efficacy and safety." (Statistical Review at
p. 60).
The relationship between the traditional clinical endpoints of relapse and survival
and the surrogate endpoint of sputum culture conversion were not robust in this case. In
fact, the clinical evidence on survival was actually and strongly in the wrong numerical
direction.' Notwithstanding this, the FDA appears to have, as noted in its draft Guidance,
relied in part on the "external expertise" of the June 2009 Anti-Infective Drug Advisory
Committee as well as took "into account" these three factors that were listed in FDASIA:
(1) the "severity" of the disease; (2) the "rarity" of the disease; and (3) the "lack of
alternative treatments." (See, e.g., Medical Review at top of p. 59).`
This is the reason for the commentators scoring clinical evidence on the actual
clinical benefit as -1 on a scale of 0 to 3. The scale was set up under the
assumption that, at worst, there would be an absence of any clinical evidence of
benefit, or if clinical evidence, then not even any "lean" in favor of the
investigational treatment, which then would have been rated as "0."
2

In addition to Dr. Porcalla's medical review reaching this conclusion, every other
review unanimously supported a recommendation for approval. For instance, the
statistical review by Dr. Lit Higgins concluded: "The efficacy in terms of a
surrogate endpoint, sputum culture conversion, was supported by the pivotal study
C208 and supportive study C209. There was a significantly elevated mortality
risk in the [Sirturo] group. This should be considered in an approval decision and
use of this regimen." The reviews of the Cross-Discipline Team Leader, Dr.
Navarro (December 21, 2012), the Deputy Division Director, Dr. Laessig
(December 27, 2012) and the Office Director, Dr. Cox (December 28, 2012) all
A4

2. FERIUPROX (deferiprone)
FDA approved Ferriprox on October 14, 2011 as "an iron chelator . . . for the
treatment of patients with transfusional iron overload due to thalassemia syndrome when
current chelator therapy is inadequate." Ferriprox was approved on the basis of its
showing on an unvalidated surrogate, serum ferritin.
Part 1. Factors Weighing into FDA Determination
a. Severity of the Condition
Persons with certain inherited anemias, especially sickle cell anemia and
thalassemia, require frequent red blood cell (RBC) transfusions because they are unable
to manufacture hemoglobin. Each unit of packed RBCs contains 200 mg of iron, which
is an extreme excess of iron as compared with the dietary intake of 1 mg of iron
necessary to maintain normal total body iron stores in healthy individuals. Without a
way for the body to excrete excess iron, persons receiving these regular transfusions of
RBCs build up massive iron overload which leads to morbidity and often eventually
death due to cardiac damage. (Medical Review #1, Sept. 20, 2011, pp. 1-2).
b. Rarity of the Condition

FDA designated Ferriprox as an orphan drug on December 21, 2001.
c. Lack of Available Therapy

At the time of Ferriprox's approval, there were two other approved therapies for
iron overload due to transfusions: Desferal (deferoxamine) and Exjade (deferasirox).
Ferriprox was given fast track designation in January 2004, before Exjade was approved.
Exjade, an orally active iron chelator, was approved in 2005. In January 2004, Desferal
was the only available therapy and requires continuous infusion over many hours, every
day.
The sponsor first submitted its NDA seeking an indication for "all transfusiondependent anemias for whom the use of other iron chelators has been considered
inappropriate." A complete response letter was issued in November 2009 and a
resubmission was made in April 2011 for essentially the same second-line use. However,
recognized the robust finding on the surrogate endpoint of sputum culture
conversion and recommended approval despite serious consideration of the
clinical safety results, especially the survival results in the pivotal study. This
unanimity of support for a Subpart H approval decision within the entirety of the
internal FDA expert review team was not always observed in the other 18 Subpart
H precedents.
A5

the data submitted were almost exclusively from thalassemia patients and FDA's October
2011 approval is for "patients with transfusional iron overload due to thalassemia
syndromes when direct chelator therapy is inadequate." For this specific use, there is a
lack of available therapy.
d. External Expertise
FDA appears to have given consideration to two types of external expertise. First,
FDA seems to have given some weight to the "expertise" of clinical practice that uses
serum ferritin to monitor the patient's iron status. While serum ferritin is a non-specific
endpoint for which FDA noted that "the relationship between the serum ferritin and
clinical outcome is not well-established" (Medical Review #2, Sept. 16, 2011, p. 34),
FDA nevertheless appears to give serum ferritin some weight because serum ferritin is "a
commonly used parameter for following body iron burden in patients undergoing red
blood cell transfusions," (Medical Review #1, at p. 12), and because "in clinical practice,
measurements of serum ferritin and [liver iron concentration] have been the generally
accepted methods of evaluation of the efficacy of therapy in persons with iron overload."
(Medical Review #3, November 20, 2009, p. 5).
Second, the Oncology Drugs Advismy Committee recommended Ferriprox for
approval on September 14, 2011 by a vote of 10 to 2 for treating patients in whom current
chelator therapy is inadequate.
Part 2. Understanding of the Disease Process

In this case, the pathophysiology by which iron overload leads to deposition of
iron in tissues and leads to iron-catalyzed peroxidation of membrane lipids which then
leads to morbidity and death due to cardiac damage is well-known. (Medical Review #1
at p. 1).
Part 3. Understanding of the Relationship Between the Effect on Serum Ferritin and
Cardiotoxicity and Death

The mechanism of the drug's action is well-known, that is, binding to iron in a 3:1
complex which is excreted in the urine, and the reduction in iron in these persons is
needed to avoid iron overload morbidities. (Medical Review #1 at p. 2). However,
serum ferritin is non-specific and "changes in serum ferritin are difficult to interpret
because serum ferritin is subject to variations induced by a number of mechanisms that
are unrelated to total body iron." (Medical Review #4, Oct. 19, 2009, p. 15). Most of
all, "the relationship between the serum ferritin and clinical outcome is not wellestablished." (Medical Review #1 at p. 34).
This part was scored a 2 on a scale of 0 to 3 mainly on the basis of the biologic
plausibility that this drug, due to its mechanism, would reduce iron stores,
notwithstanding the weakness of serum ferritin itself as a surrogate, due to its lack of
A6

specificity as a measure of iron stores. The non-specificity of serum ferritin and the lack
of understanding of the relationship between the surrogate and outcomes led to a score of
2 instead of 3. 3

Part 4. Clinical Evidence of Ferriprox's Effect on Serum Ferritin and on Outcome
It is of value here to note that FDA rejected the original NDA submitted in 2009
for Ferriprox because the "primary efficacy endpoint of the single major controlled trial .
. . was the change in cardiac MRI T2* which was said to measure iron content within the
heart. FDA stated that this endpoint was a surrogate endpoint and there were no data to
support the incremental changes in the values as predictive of clinical benefit." (Medical
Review #1 at p. 10) (emphasis added). Moreover, "secondary endpoints [of serum
ferritin and liver iron concentration] also were not consistently corroborative of the
primary endpoint [MRI T2*] results." (Medical Review #1 at p. 5). Overall, "the study
did not find a significant correlation between change in cardiac MRI T2* and measures of
cardiac function and there were no differences between treatments in change in liver iron
concentration (LIC)." (Medical Review #1 at p. 2). The statistical review observed that
"the patients in this study were not followed for clinical outcome and therefore, this study
was not designed to obtain internal validation of MRI T2* change as a surrogate for any
clinical outcome indicative of reduced cardiac iron." 4 (Statistical Review, March 24,
2009, p. 7).
"Although the data from this study provided statistically significant evidence . . .
in MRI T2* . . . this study was not designed to and therefore, does not provide evidence
that change in MRI T2* is reasonably likely to predict clinical benefit due to lack of longterm follow-up of these patients." (Statistical Review #2, Nov. 22, 2009, p. 3).
In response to the FDA's rejection of the original NDA, the sponsor "conducted an
analysis of a subpopulation of patients drawn from the previously conducted studies and
defined as being inadequately treated with current chelator therapy." (Medical Review
#1 at p. 10). In this analysis, approximately 50% met the primary efficacy endpoint of
having a 20% or greater decline in serum ferritin. Of additional importance, the sponsordefined "success rate" in this same analysis was 42% for liver iron concentrate (LIC).
(Medical Review #1 at pp. 7-8). FDA noted that "change in LIC using liver biopsy has
3

Others may score this differently, perhaps even only a "1" given the nonspecificity of serum ferritin and lack of well-established relationship between
surrogate and outcomes.

4

Note that FDA states that this study could provide both evidence of the effect of
the drug on an unvalidated surrogate and at the same time, in the same study,
evidence of the effect of the drug on clinical outcome, thereby "validating" that
surrogate.
A7

generally been considered to be the standard measure of efficacy in response to iron
chelation therapy." (Medical Review #4 at p. 15).
Overall, FDA first rejected the original NDA on grounds that the primary endpoint
of the key pivotal study, MRI T2* changes, was not sufficiently correlated with any
clinical outcome to warrant being the basis for even an accelerated approval,
notwithstanding the disease being severe, rare, and without adequate therapy. However,
FDA approved a second resubmission that was based on an analysis of a commonly used
measure in clinical practice of patients with transfusion-related iron overload, serum
ferritin, which itself was supported internally by a positive finding in the same population
on liver iron concentration which is the "standard measure of efficacy in response iron
chelator therapy."
FDA's actions on Ferriprox illustrate both the fatal flaws in a clinical program
attempting to rely upon a surrogate (MRI T2*), and the factors to be considered and the
clinical evidence that were found by FDA to be of sufficient merit to allow FDA, as a
matter of its judgment, to conclude that serum ferritin is reasonably likely to predict
clinical benefit, even without any clinical trial results on any cardiac outcomes such as
heart failure or mortality and notwithstanding an FDA acknowledgement that serum
ferritin is a non-specific measure. However, FDA's Subpart H approval here was based
clinically on the corroboration of the serum ferritin results by the liver iron concentrate
results and bolstered by the known mechanistic action of the drug (i.e., that by its
mechanism of action the drug leads to iron excretion in the urine).
Overall, the clinical evidence of the surrogate was scored a full 4 out of a possible
4 due to the strength of evidence on serum ferritin which itself was buttressed by the
clinical findings on LIC. However, since there was no clinical evidence on any ultimate
clinical outcome, the score for clinical evidence of outcome benefit is zero.

A8

3. MAKENA (hydroxyprozesterone caproate)

FDA's February 3, 2011 approval of Makena to reduce the risk of preterm birth
(P1I3) was based on a surrogate of reducing preterm birth as defined as those births
occurring at less than 37 weeks of gestation. "Preterm birth <37 weeks gestation was a
surrogate 5 for pregnancy outcome (neonatal/infant morbidity and mortality)." (Medical
Review, Feb. 3, 2011, p. 14).
Part 1. Regulatory Factors Weighing into FDA Determination

a. Severity of the Condition
The risks of miscarriage, stillbirths, and neonatal mortality are associated with
delivery prior to full-term gestation, as well as neonatal morbidities and adverse maternal
outcomes as well.
b. Rarity of the Condition
Makena was designated as an orphan drug on January 25, 2007.
c. Lack of Available Therapy
"Currently there is no drug product approved in the United States to reduce the
risk of preterm birth; however, [the active ingredient in Makena] is compounded by
pharmacists and is used widely for this indication in women at high risk." (Medical
Review at p. 11). In 1956, FDA had approved an NDA for Delalutin, which had the same
active ingredient as Makena, for treating pregnant women for "habitual and recurrent
abortion, threatened abortion." (Medical Review at p. 12). In 2000, FDA withdrew the
approval of Delalutin at the request of the NDA sponsor because it no longer marketed
Delalutin. In a June 25, 2010 Federal Register notice, FDA announced its determination
that Delalutin was not withdrawn from marketing for safety or efficacy reasons.
d. Use of External Expertise
With Makena, FDA relied upon two forms of external expertise and FDA reached
its "informed judgment" that the surrogate endpoint of preterm birth less than 37 weeks
was reasonably likely to predict clinical benefit, that is, pregnancy outcome or neonatal
infant and maternal morbidity and mortality. These two forms of external advice are
summarized in the Medical Review: (1) 2006 Advisory Committee, and (2) subsequent
scientific papers published in the literature.
While FDA Medical and Statistical Reviews refer to PTB <37 weeks as a
"surrogate," preterm birth is a clinical event and, therefore, in the terminology of
the Draft Guidance, PTB <37 weeks is an "intermediate clinical endpoint."
A9

i. "The surrogate endpoints of reductions of PTB (preterm birth) at <35 and
<32 weeks were thought by the Advisory Committee to predict a reduction
in neonatal mortality and morbidity. At the time of the Advisory
Committee in 2006, the endpoint PTB at <37 weeks was not believed to be
an adequate surrogate for neonatal outcome." 6 (Medical Review at p. 6).
ii. "The Applicant submitted a single phase 3 clinical trial which demonstrated
a statistically strong (p<.001) reduction in the incidence of preterm births
prior to 37 weeks gestation, the protocol pre-specified primary endpoint.
There is recent evidence that 'late preterm births' (births between 34" and
366/7), which comprise 71.3% of all preterm births, are increasing, and
suffer greater neonatal and childhood morbidity and mortality than
previously thought [5 papers are cited that published between the time of
the 2006 Advisory Committee and the Medical Review]. These data
indicate that 'preterm birth prior to 37 weeks' is a surrogate endpoint that is
reasonably likely to predict clinical benefit." (Medical Review at p. 5).
Part 2. Understanding of the Disease Process

Here the disease process is complex and has multiple pathophysiologic pathways
and therefore, this mitigates against reliance upon any surrogate. The biological means by
which the gestational process progresses to premature delivery is complex and
multifaceted. Therefore, the surrogate endpoint of PTB<37 weeks is likely more
analogous to the PSA example than the enzyme replacement example in the Draft
Guidance (see Draft Guidance at p. 19, lines 634-648) in that PTB<37 weeks is not on the
pathophysiological causal pathway and is not the biologic mechanism that causes the
neonatal mortality and morbidity, even though, like PSA, it is correlated with increased
risk.
Part 3. Understanding of the Relationship Between PTB and Pregnancy Outcomes

a. Epidemiological Evidence
The epidemiologic evidence is strong with Makena. The 2006 Advisory
Committee assessed the epidemiologic evidence supporting the relationship between PTB
and pregnancy outcomes and found that this evidence was strong enough to support the
endpoints of PTB<32 weeks and PTB<35 weeks as surrogate endpoints but not PTB<37
weeks. However, additional evidence published subsequent to the 2006 Advisory
Committee permitted the Medical Officer, Dr. Barbara Wesley, to conclude that PTB<37
6

"The Committee stated that a reduction of preterm birth <37 weeks was not an
adequate surrogate (Yes: 5; No: 16) but that reductions in preterm birth <35 weeks
(Yes: 13; No: 8) and <32 weeks (Yes: 20; No: 1) were adequate surrogates."
(Medical Review #2, Jan. 23, 2009, p. 7).
A10

weeks was also a reliable, consistent and acceptable surrogate endpoint. 7 (Medical
Review at p. 5).
b. Effect of Drugs in the Same or Closely Related Pharmacologic Class to Affect
Pregnancy Outcomes
Since there are no drugs in any pharmacologic class approved for reducing the risk
of PTB, there are no analogous therapies here on which to draw support directly for
reducing the risk of PTB. However, other progesterones including the active ingredient
in Makena have been approved for aiding in assisted reproductive technologies and other
conditions supporting the maintenance of pregnancy.
Part 4. Clinical Evidence of the Makena's Effect on PTB < 37 Weeks and on
Pregnancy Outcomes
a. PTB <37 Week

The surrogate of PI B <37 weeks was highly statistically significant (p<0.001).
b. Pregnancy Outcomes

"The proportion of babies with at least one event on the [secondary] composite
index of neonatal morbidity and mortality was lower in the [Makena] group (11.9%,
35/295 infants) than in the vehicle group (17.2%, 26/151 infants) but the between-group
differences was not statistically significant (nominal p-value of 0.1194)." (Medical
7

It is also likely that the Advisory Committee was opining on PTB <32 weeks, PTB
<35 weeks and PTB <37 weeks as validated surrogates which would have
qualified Makena for traditional approval, not Subpart H approval. Outside of
AIDS and cancer, FDA has not often asked Advisory Committees to opine on
whether clinical evidence on a particular endpoint would qualify a therapy for
Subpart H approval. For example, note that the August 5, 2013 Cardiorenal
Advisory Committee, addressing the approvability of tolvaptan, a vasopressin V2
receptor antagonist, was not asked whether total kidney volume would qualify as
an unvalidated surrogate that may support a Subpart H approval if the Advisory
Committee found that total kidney volume is "reasonably likely to predict clinical
benefit," which, in this case, clinical benefit would likely be end-stage renal
disease and/or clinically meaningful outcomes such as significant worsening of
renal function or kidney pain. However, there are exceptions outside of AIDS and
cancer. For instance, the Oncology Drugs Advisory Committee (ODAC) was
asked whether FAP was an adequate "unvalidated" surrogate, that is, to qualify
Celebrex (Precedent #12) for Subpart H approval. But even this case was before
ODAC and while FAP is not cancer, the ultimate clinical benefit was prevention
of colon cancer so even this "exception" is not fully outside of AIDS and cancer.
Al 1

Review #1 at p. 6). "Approximately 6.5% of the women in each treatment group
experienced a fetal or neonatal deaths . . . The results . . . show that despite the treatment
groups having about the same rate of fetal and neonatal deaths, the losses occur earlier
among [Makenal women." (Statistical Review #2, Oct. 19, 2006, p. 20).
This impact on fetal or neonatal deaths was stated another way by the Medical
Reviewer: "There was a trend toward an increased risk of miscarriage and stillbirths in
the [Makena] treatment arm and a trend toward a decrease in neonatal death, with no
overall net survival benefit." (Medical Review #1 at p. 7) (emphasis is original).
Overall, the secondary endpoint of a composite measure of neonatal morbidity/mortality
leaned in favor of the Makena group while the separate analysis of neonatal morality
showed essentially no numerical difference and had a nominal p-value of 0.6887
(Medical Review #1 at p. 7). The clinical evidence for the ultimate clinical benefits in
the single pivotal trial was not strong.

Al2

4. PROMACTA (eltrombouae)
FDA approved Promacta on November 20, 2008 on "short-term platelet count
response" as a surrogate marker for longer platelet count responses (platelet counts are
recognized as acceptable measures of clinical benefit for patients with ITP [idiopathic
thrombocytopenic purpura])." (Medical Review #1, Nov. 4, 2008, p. 3). The two clinical
trials of Promacta administered drugs over 6 weeks or less (this is the meaning of "short
term" in the Reviewer's statement above). Had the Promacta trials studied and
established the drug's effect on platelet counts out to 6 months, this approval would have
been a traditional approval and not one under Subpart H.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
Chronic ITP is a serious medical condition. (Medical Review #1 at p. 3). The
frequency of death from hemorrhage in patients with platelet counts below 30,000/mcl is
estimated to be between 1.6 and 3.9% per patient year. (Medical Review #2, Sept. 12,
2008, p. 17).
b. Rarity of the Condition
FDA designated Promacta as an orphan drug on March 4, 2008.
c. Lack of Available Therapy
"[Promacta] approval would provide a meaningful therapeutic benefit to patients
over existing treatments because of its minimal risk for immunogenicity (based upon [its]
small molecule characteristics). The labeling for romiplostine, the only currently
marketed TPO receptor agonist, includes information regarding the risks for
immunogenicity. These risks are not applicable to [Promacta]." (Medical Review #1 at
P. 3)d. Use of External Expertise
In the medical and statistical reviews, the commentators found no evidence of any
reliance on special government employees (SGEs), an Advisory Committee for Promacta,
or specific published literature.
Part 2. Understanding of the Disease Process
"The clinical hallmark of the disease is an increased tendency to bleed." (Medical
Review #2 at p. 17). Furthermore, the relationship of platelet count to bleeding is wellestablished: "Patients with platelet counts between 30,000/mcl and 10,000/ma are
generally considered treatment candidates due to slightly increased risk of spontaneous
bleeding or increased risk of bleeding due to trauma." (Medical Review #2 at p. 17).
A 13

Part 3. Understanding of the Relationship Between the Drug's Effect on ShortTerm Platelet Counts and Increased Risk of Bleeding

There was no epidemiological evidence cited in the FDA review documents to
support the surrogate - which is "short term" (that is, six weeks) increase in platelet count
- as reasonably likely to predict long-term, chronic increase in platelet count - which is
generally established in six month trials or generally on increased risk of bleeding. While
there was no evidence to support the use of this surrogate, there was a therapy approved
from the same pharmacologic class but based on an endpoint of six-month duration.
Earlier in 2008 (the year FDA approved Promacta), FDA had approved romiplostim, a
biological product that is a member of the same pharmacologic class - thrompoietin
(TPO) receptor agonists - and this approval for the same indication (that is, to treat ITP)
was a traditional approval based on two clinical trials, each of six-months duration.
Part 4. Clinical Evidence of Promacta's Shorter-term (Surrogate) Effect and
Long-Term Effect on Platelets and/or Bleeding

Both Promacta pivotal studies showed a robust short-term (surrogate) effect on
platelets (p<0.001) (Statistical Review, Apr. 29, 2008, pp. 19 and 27).
As for clinical evidence that the FDA had at the time of the approval that
Promacta's "short-term" (six weeks) impact on platelet counts would predict either
clinical benefit of long-term impact on platelet counts or on bleeding, there was mixed
evidence.
As supportive evidence that the platelets produced by Promacta behaved in a
physiologically "normal" way, the Sponsor had conducted "an exploratory clinical study
that demonstrated [that Promacta] prompted platelet count increases in healthy subjects.
These drug-stimulated platelets had in vitro platelet function characteristics typical of
platelets. Hence, this study supported the generally accepted use of platelet counts as an
'accepted' measure of clinical benefit for clinical studies of TPO receptor agonists among
patients with chronic ITP." (Medical Review #1 at pp. 2-3).
As Promacta was only administered for six weeks (or less) in the two pivotal
trials, there is no clinical evidence as to the impact long-term on platelet counts if
Promacta was administered chronically (for which a trial of six-months duration would
have been relied upon). Furthermore, of some concern, "discontinuation of [Promacta] at
the end of the study resulted in an unacceptable amount of serious hemorrhage."
(Medical Review #1 at p. 3). Also, the statistical reviewer observed that within two
weeks after the subjects on drug were off treatment there was a return to placebo levels of
platelet counts. (Statistical Review at pp. 27-28).
As for bleeding events, there was a numerical lean in favor of Promacta but in
neither trial was this statistically significant with p-values of 0.121 and 0.088 for the
Al4

between-group difference on bleeding events in the two pivotal trials. (Statistical Review
at pp. 8-9).

A15

5. EXJADE (deferasirox)
The FDA approval of Exjade for treating "chronic iron overload due to blood
transfusions" on November 2, 2005 was based on a surrogate endpoint of improvement in
liver iron concentration (LIC).
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition

"Chronic iron overload due to requisite blood transfusion is a serious and lifethreatening condition." (Medical Review #1, Nov. 2, 2005, p. 2).
b. Rarity of the Condition

Exjade was granted orphan drug designation on November 21, 2002.
c. Lack of Available Therapy

At the time FDA was reviewing the Exjade NDA, the Medical Team Leader, Dr.
Dwaine Rieves, stated: "Deferoxamine, the only available therapy for this condition,
presents unique compliance and infectious risks due to the need for prolonged
administration of the drug. [Exjade] is an orally administered drug that provides a
meaningful therapeutic benefit over the existing therapy." (Medical Review #1 at p. 2).
d. External Expertise

FDA sought the advice of the Blood Products Advisory Committee (BPAC) and at
its September 29, 2005 meeting, the BPAC found that "the applicant [had] provided
substantial evidence of the effectiveness of [Exjade] in the reduction of liver iron
concentration, an outcome indicative of a clinical benefit . . . The sponsor's major clinical
evidence of [Exjade] effectiveness . . . is based upon alterations in liver iron content, an
endpoint the BPAC discussants regarded as a measure of clinical benefit. In this context,
the endpoint is not regarded as a surrogate endpoint rather as an endpoint other than
survival or irreversible morbidity 8, as cited in the Subpart H regulations." (Medical
Review #1 at p. 2).
Part 2. Understanding of the Disease Process
8

An "intermediate clinical endpoint" (rather than a surrogate) is the term used in
the Draft Guidelines for this kind of endpoint; however, during the later FDA
approval of Ferriprox, the FDA Reviewers refer to both serum ferritin and LIC as
"surrogates," and in an earlier medical review of Exjade, FDA refers to LIC in this
pivotal trial as a "surrogate" (see Medical Review #2 at p. 38). Therefore, this
analysis will refer to LIC as a surrogate and not as an intermediate clinical benefit.

Al6

See Item 2 under Ferriprox.
Part 3. Understanding of the Relationship Between LIC and Cardiac Outcomes,
Including Mortality

"Although accepted by the Division as a clinically meaningful endpoint, the
primary endpoint [of LIC] is technically a surrogate endpoint since it does not necessarily
address clinically significant morbidity or mortality. The main mortality on P-thallasemia
is due to cardiac dysfimction whose etiology in P-thallasemia is probably multifactorial.
Nonetheless, most of the literature in 13-thallasemia has used LIC as a marker for
morbidity for other organ involvement and as a surrogate for mortality. There is some
information, however, that LIC does not completely correlate to the extent of cardiac
hemosiderosis, the primary cause of mortality. Obviously, repetitive biopsy of the
myocardium to measure iron concentration in the heart is not acceptable." (Medical
Review #2, Oct. 10, 2005, p. 38).
As for understanding the relationship between drugs in the same pharmacologic
class as LIC, the single pivotal trial for Exjade was a noninferiority study design which
used as its active comparator, deferoxamine, and therefore, FDA had evidence from a
within-study comparison of the only other member of the same or closely related class on
the surrogate endpoint of LIC.
Part 4. Clinical Evidence of Exjade's Effect on LIC and/on Cardiac Outsomes
including Mortality

FDA, in its review of this NDA, noted that LIC as "the primary endpoint is
acceptable and it was agreed to by the Division in the Special Protocol Assessment. It
should be remembered, however, the LIC is a surrogate marker and that the effects of
Exjade on morbidity/mortality, which are the truly important clinical endpoints, are not
likely to be demonstrated in this short trial." (Medical Review #2 at p. 31).
Rather than bolster LIC results by seeing trends on irreversible morbidity and
mortality in this "short" trial, FDA looked to find support from other critical surrogate
markers such as serum ferritin. 9
As for LIC, the protocol had specified that "non-inferiority of [Exjade] to
[deferoxamine] was to be established if the two sided 95% confidence interval of the
difference in success rate between the two groups was above -15%. The basis for the
choice of this [non-inferiority] margin was unclear in the submission. Notably, FDA had
9

The authors must inform the reader that this trial was a year-long trial and
therefore by many would not be considered "short:" however, even a year long
study is too "short" to see effects on mortality and irreversible morbidity.

A 17

questioned the meaningfulness of this margin during the study's protocol review. 1°
(Medical Review #1 at p. 4).
The primary efficacy result was a point estimate difference of -13.5%, with a
lower 95% confidence interval of -21.6% (or, in other words, the margin defining success
of the trial was not met). About this, the FDA concluded: "Given that the original basis
of the non-inferiority margin was poorly substantiated, little clinical meaningfulness
could be assigned to failure to achieve the primary endpoint. The primary endpoint data
did establish that both [Exjade and deferoxamine] lowered LIC over a 12 month period of
time, a time period during which subjects would have been expected to have increases in
LIC due to continuing blood transfusions. This observation provides evidence of a
treatment effect for [Exjade]." (Medical Review #1 at p. 5).
With respect to serum ferritin, FDA concluded that "[s]erum ferritin values
declined in a dose-related manner for subjects receiving [Exjade], a pattern similar to that
for subjects receiving [deferoxamine]." (Medical Review #1 at p. 5).

10

Query, though, how FDA nevertheless had accepted the design of this pivotal
study under an SPA.
Al8

6. LEVAQUIN (levolloxacin)

FDA approved Levaquin for post-exposure prevention of inhalational anthrax on
November 11, 2004. Much of what the Agency had learned from its Subpart H approval
of Cipro for inhalational anthrax in August 30, 2000 was used to create a draft guidance.
"FDA Draft Guidance for Industry: Inhalational Anthrax (Post-Exposure) - Developing
Antimicrobial Drugs" ("Anthrax Draft Guidance") (March 2002). FDA then relied on its
Anthrax Draft Guidance when it approved Levaquin in 2004. (Statistical Review, Nov.
15, 2007, p. 1).
As for Cipro, there was a two part or "compound" surrogate for this approval in
that FDA concluded: (1) that "[m]ortality due to anthrax for animals that received a 30
day regimen of oral Levaquin beginning 24 hrs post exposure was significantly lower
(1/10), compared to the placebo group (9/10) [p=0.0011]," and (2) "mean plasma
concentrations of Levaquin associated with a statistically significant improvement in
survival over placebo in rhesus monkey model of inhalational anthrax are reached or
exceeded in adult . . . [human] patients receiving the recommended oral and intravenous
dosage regimens." (Levaquin Package Insert).
Part 1. Regulatory Factors Weighing into FDA Determination

a. Severity and Rarity of the Condition
"Mortality for established [inhalational anthrax] even after treatment was 80-100%
in the 20th century." (Anthrax Draft Guidance at p. 3). In addition, "inhalational anthrax
is extremely rare. There have been only approximately 20 cases in the United States in
the past 100 years . . . For these two reasons, the rarity of disease and the extremely high
mortality rate, a clinical study is not feasible." (Cipro, Statistical Review, Aug. 16, 2000,
p. 1).
b. Rarity of the Condition
Although the prevalence of inhalational anthrax is sufficiently low, the Sponsor
did not seek orphan drug designation.
c. Lack of Available Therapy
At the time of Levaquin's approval, Cipro was indicated specifically for postexposure prophylaxis for disease caused by inhaled B. anthracis, and, although
doxycycline and penicillin G procraine products were not specifically indicated for postexposure prophylaxis for disease caused by inhaled B. anthracis, FDA "had published a
notice in the Federal Register (66 Fed. Reg. 55679) that clarified the dosing regiments for
[those drugs] in the management of patients with inhalational anthrax." (Anthrax Draft
Guidance at p. 5).
d. External Expertise
Al9

Although no advisory committee was convened specifically for Levaquin, FDA
had sought "input from the Anti-Infective Advisory Committee [and determined that] the
use of the rhesus (macaque) monkey disease and treatment model for inhalational anthrax
(post-exposure) provides convincing evidence of efficacy for regulatory evidence."
(Inhalational Anthrax Draft Guidance at p. 4).
Part 2. Understanding of the Disease Process

Before the approval of Cipro in 2000, four years before the approval of Levaquin,
FDA had stated that "[t]he inhalational form of the disease, which affects the mediastinal
lymph nodes, other organs of the reticuloendothelial system and the central nervous
system, is considered the most likely clinical entity resulting from the intentional use of
an aerosolized preparation of the spores of B. anthracis." (Cipro Medical Review, Aug.
31, 2000, p. 2).
Part 3. Understanding of the Relationship between the Monkey Data and Human
Mortality and Part 4. Clinical Evidence

FDA's draft guidance document on the development of treatment for postinhalational anthrax exposure stated that, "a non-human primate model that models the
drug disposition in humans [was] considered an adequate surrogate for human disease
and objective endpoints such as mortality, time of death relative to antimicrobial use,
pathology, and bacteremia in the macaque." (Statistical Review at p. 1).
Thus, two findings formed the basis of FDA's Subpart H approval of Levaquin for
inhalational anthrax:
First, "[s]urvival was significantly better (p=0.0011, two-sided Fishers exact test)
and time to death was significantly longer (p<0.0001, log rank test) [in macaques] in the
levofloxacin group compared to the placebo group." (Statistical Review at p. 1). Also,
Levaquin had a numerical advantage with 90% (9/10) of the macaques surviving,
compared to 80% (8/10) in the ciprofloxacin group, and only 10% (1/10) in the placebo
group. (Statistical Review at p. 1).
Second, as for comparative monkey/human exposure levels, the "mean plasma
concentrations [and mean steady state AUC0.24] associated with a statistically significant
improvement in survival over placebo in the rhesus monkey model of inhalational
anthrax are reached or exceeded in adult . . . patients receiving the recommended oral and
intravenous dosage regimens." (Levaquin Package Insert).
As for understanding the relationship of drugs in the same or closely related class
on the compound surrogate, see above discussion under 1.c. regarding other drugs
including Cipro for anthrax.
Monkey survival data was one part of this unusual compound surrogate, see the
discussion above. However, there were "complete pharmacokinetic data on the drug in
A20

human volunteers . . . and pharmacokinetic data in the rhesus monkey in the efficacy
study of inhalational anthrax [is used] to demonstate that the desired systemic exposure
achieved in humans after the anticipated dosage regimen can actually be achieved and is
effective in the animal model in preventing inhalational anthrax infection and consequent
mortality." (Anthrax Draft Guidance at p. 10).

A21

7. TYSABRI (natalizumab)
FDA approved Tysabri on November 23, 2004 for treating relapsing-remitting
multiple sclerosis (RRMS), relying upon the reduction in MS relapse rates at one year as
the surrogate endpoint. Applying the terms of the Draft Guidance, this would be an
intermediate clinical endpoint that would be reasonably likely to predict the benefit at
two years. All previous MS therapies were approved on the basis of two-year relapse
rate reduction and "the clinical meaningfulness of a decrease in the relapse rate through
only one year is uncertain." (Medical Review, Nov. 23, 2004, P. 6).
Part 1. Regulatory Factors Weighing into FDA Determination

a. Severity of the Condition
Relapsing-remitting multiple sclerosis is a serious, life-threatening condition.
b. Rarity of the Condition
While Tysabri was not designated an orphan drug for RRMS, the statutory
threshold for qualifying as an orphan drug was, in part, set in the 1984 amendment to the
Orphan Drug Act specifically to include all of multiple sclerosis as an orphan disease, not
just the subset of RRMS. This was because in considering how to amend the original
1983 Orphan Drug Act to make it less difficult to garner orphan drug designation, key
Senators caucused with the National Organization for Rare Diseases (NORD) and
mutually determined to establish that the maximum number of Americans with a
condition which would still qualify as an "orphan" would be 200,000. This number was
chosen, specifically, to make sure that MS would be an "orphan" disease and in 1984
there were just under 200,000 Americans diagnosed with MS. However, soon after FDA
approved the first therapy for multiple sclerosis (Betaseron in August 1993, which was
also the first non-AIDS Subpart H approval), the number of Americans diagnosed with
multiple sclerosis dramatically increased. So, while Tysabri was never designated as an
orphan drug for RRMS, the commentators, fully cognizant of the intent of the 1984
orphan drug amendment, view Tysabri as, nevertheless, falling within the "penumbra" of
orphan drug status and score Tysabri a "1" on rarity.
c. Lack of Available Therapy
"Accelerated approval requires that the new drug provide evidence of the potential
to address an unmet medical need. Many MS patients continue to have exacerbations
while taking one of the available first-line MS therapies. None of the currently available
therapies have proven efficacy when used as add-on therapy. [One of the two pivotal
Tysabri studies] provides evidence that [Tysabri] is effective as add-on therapy for
subjects who continue to have relapses while on a first-time therapy (Avonex).
Therefore, [Tysabri] has the potential to address an unmet medical need. (Medical
Review at p. 6).
A22

d. External Expertise
FDA did not rely on an advisory committee during its initial review of Tysabri.
However, Tysabri was withdrawn from the market by the manufacturer in February 2005
after three patients developed progressive multifocal leukoencephalopathy (PML).
Subsequently, FDA convened an Advisory Committee to consider the reintroduction of
Tysabri in March 2006. Furthermore, FDA had convened and considered the input from
several earlier advisory committees on other multiple sclerosis therapies.
Part 2. Understanding of the Disease Process

"Multiple Sclerosis is a chronic, inflammatory, possibly autoimmune,
demyelinating disease of the central nervous system." (Medical Review at p. 11). Note
that the FDA review status that multiple sclerosis may be "possible autoimmune." Given
that Tysabri's mechanism of an action is as an immunomodulator, having a more
definitive view of the causative role of autoimmunity in the pathophysiology of this
disease would have been more compelling.
Part 3. Understanding of the Relationship Between the One-Year Relapse Rate and
Two-Year Relapse Rate

The effect of [Tysabri] on relapse rate in [the pivotal study on Tysabri's use as
first-line therapy] was approximately twice the effect observed with current first-line
drugs for this indication. Such comparisons of different agents across studies are
problematic . . . However, the magnitude of [Tysabri's] effect is sufficient that the effect
at one year is reasonably likely to predict a clinical benefit at two years." (Medical
Review at p. 102).
As for understanding the relationship of drugs in the same or closely-related
pharmacologic class on rate and extent of exacerbations at one year of treatment as .
predictive of their two year effectiveness, at the time of Tysabri's approval, there were
four other approved immunemodulators approved for treatment of MS: Betaseron,
Avonex, Rebif and Copaxone. While each of these was approved on the basis of twoyear studies of impact on reducing rate and extent of MS exacerbations, their impacts
after one year of therapy, while generally more modest than at the end of two years, were
predictive of their two year results.
Part 4. Clinical Evidence on One-Year and Two-Year Relapse Rates

"For other MS products, FDA has required two-year data . . . A salutary effect on
relapse rate at one year is not a validated surrogate for benefit at two years. However, the
apparent treatment effect of [Tysabri] with respect to relapse rate at one year is
unprecedented in the MS field, and its magnitude is reasonably likely to predict clinically
meaningful effectiveness at two years. If, in fact, the benefit on clinical relapses is shown
to be durable through two years, the product may be substantially more efficacious than
A23

currently approved MS therapies. It is possible, however, that the magnitude of
[Tysabri's] effect on relapse rate, when assessed through one year, may substantially
overestimate [Tysabri's] benefit on relapse rate through two years . . . In particular, the
treatment effect appears to wane with the development of [anti-Tysabri] antibodies,
which may increase in time." (Medical Review at p. 53) (emphasis added).

A24

8. LUVERIS (lutreuinalfo)
On October 8, 2004, FDA approved Luveris for stimulating follicular development
in infertile hypogonadotropic hypogonadal women with profound LH deficiency (LH <
1.2). "The Division Director further concluded that in this orphan population of women
with severe LH deficiency (LH < 1.2), the surrogate endpoint of follicular development
(as defined by the Sponsor) was reasonably likely to predict clinical benefit [with respect
to pregnancy] . . . (Medical Review #1, Oct. 6, 2004, p. 2).
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
The inability to ovulate due to profound luteinizing hormone (LH) deficiency
includes, among other serious consequences, the inability to become pregnant. "The
Director believes that infertility in the context of hypogonadotropic hypogonadism and
profound LH deficiency is a serious condition with very limited options for pregnancy."
(Medical Review #2, Oct. 6, 2004, p. 7).
b. Rarity of the Condition
Luveris was granted orphan drug designation by FDA on October 7, 1994.
c. Lack of Available Therapy
"Luveris would be the only LH-alone product . . . on the U.S. market. There are
no approved drug products that have the indication of treatment of infertility in women
with hypogonadotropic hypogonadism." (Medical Review #3, Sept. 28, 2004, p. 17).
d. External Expertise
The Reproductive Health Advisory Committee considered Luveris on September
30, 2003. "After hearing presentations from experts in Reproductive Endocrinology on
the subject of female hypogonadotropic hypogonadism . . . the Committee voted 15 to 0
that the Sponsor's data did not demonstrate efficacy for Luveris in ovulation induction
when the primary endpoint was ovulation rate. The Committee voted 8 to 7 that the
Sponsor's data demonstrated efficacy for Luveris in ovulation induction when the
primary endpoint was follicular development. Finally, the Committee voted 11 to 3 . . .
that the Sponsor's data demonstrated efficacy for Luveris for follicular development
when the primary endpoint was follicular development." (Medical Review #1 at p. 2)
(emphasis in original).
Part 2. Understanding of the Disease Process
The FDA's medical review suggests that the disease process is complex and
multifactorial: "the role of LH in hypogonadal female infertility patients is clouded by
the spectrum of clinical disorders that cause hypogonadotropic hypogonadism with the
A25

differing patterns of gonadotropin secretion may further confound clinical outcome
results." (Medical Review #3 at p. 19).
Part 3. Understanding of the Relationship Between Follicular Development and
Fertility

"The Division believed that although both follicular development and ovulation
are surrogates for pregnancy (the clinically meaningful outcome), ovulation is more
temporally proximate to pregnancy and therefore more appropriate as a surrogate."
(Medical Review #2 at p. 5). Nevertheless, follicular development is on the causal
pathway, as is ovulation. However, there was no epidemiological evidence cited in the
FDA review documents linking follicular development to pregnancy.
As for understanding the relationship of drugs in the same or closely-related
pharmacologic class on follicular development: "Recognition of the therapeutic potential
of gonadotropins began in the 1950's with the extraction and purification of human
menopausal gonadotropins (both follicle stimulating hormone and luteinizing hormone)
from both human pituitaries and urine sources. Successful clinical pregnancies resulting
from the use of these human-derived gonadotropins were first reported in the 1960's. In
the 1990's cells that are capable of producing biologically-active LH in culture produced
LH. This recombinant derived LH is from in vitro cultured cells." (Medical Review #3
at p. 17).
Part 4. Clinical Evidence of Luveris on Follicular Development and Fertility

"The primary efficacy parameter for both Studies 6905 and 6253 was follicular
development as defmed by three co-primary endpoints (follicle size as measured by
ultrasound, pre-ovulatory serum estradiol levels and mid-luteal progesterone levels). The
Sponsor's analysis demonstrated that in Study 6253, 75 IU of Luveris was numerically
better than 25 R1 of Luveris or placebo for follicular development in women with LH
<1.2 IU/L." (Medical Review #1 at p. 3). "The Division's analysis of Study 6905
demonstrated . . . the placebo was as efficacious as 75 IU of Luveris. Therefore, in the
opinion of the Division, Luveris was not demonstrated to be effective." (Medical Review
#1 at p. 3).
Therefore, the Sponsor planned and conducted a third study, Study 21008, with
follicular development as the Sponsor's prespecified primary endpoint despite the
Division's recommendations that ovulation rate be the primary endpoint. The Sponsor's
"evaluable patient analysis of Study 21008 demonstrated that 67% of patients receiving
75 IU of Luveris achieved follicular development compared to 20% of patients receiving
placebo." (Medical Review #1 at p. 4). "The Director [Dr. Shames] concluded that the
results from Studies 21008 and 6253 provide substantial evidence that Luveris 75 TU,
when administered concomitantly with FSH, induces follicular development in this
population of infertile women. These studies, however, do not demonstrate a positive
effect on clinical pregnancy, etc. Study 21415 evaluated titrable FSH dosing with the
A26

dose of Luveris fixed at 75 IU and demonstrated a 36% clinical pregnancy rate after one
cycle. While reassuring, this finding is not definitive because there was no placebo
comparator group in Study 21415, and the finding has not been replicated in a second
trial." (Medical Review #1 at p. 7). Study 21415 also reported follicular development
rates of 63% "in all cycles combined." (Medical Review #3 at pp. 29-30). Therefore, in
Study 21415, there was within-study clinical evidence both on follicular development,
the surrogate, as well as on pregnancy, the ultimate clinical outcome.

A27

9. FABRAZYME (agalsidose beta)

FDA approved Fabrazyme on April 23, 2003 to treat Fabry's disease. This
approval was based on a surrogate endpoint of near-elimination of all accumulation of
enzyme in renal capillary endothelium, one type of vascular endothelium.
Part 1. Regulatory Factors Weighing into FDA Determination

a. Severity of Condition and Lack of Available Therapy
"[W]ith age, the principal manifestations of concern in Fabry's disease are in the
kidney, heart, and brain. Renal disease is manifested by proteinuria, hypertension, and
progressive azotemia; the principal cause of death in Fabry's disease in the past was renal
failure . . . The median age of death for homozygous males is 50 years." (Medical
Review #1, Apr. 21, 2003, p. 4).
b. Rarity of the Condition
Fabrazyme was designated an orphan drug on January 19, 1988.
c. Lack of Available Therapy
"There is no specific treatment for Fabry's disease." (Medical Review #1 at p. 4).
d. Use of External Expertise
"Vessels (capillaries in this case) that are essentially near-normal in appearance
that may well lead to an altered development of vascular occlusion, and thus to an
alteration in expression of the clinical impairments of the disease. The [January 20031
Advisory Committee has also supported this assessment of the potential impact of nearabsence of capillary accumulation, as well as concurring that the evidence submitted by
[the Sponsor has] demonstrated this effect on capillary endothelium." (Medical Review
#2, Apr. 23, 2003, p. 3).
Part 2. Understanding of the Disease Process

"The underlying basis of Fabry disease is well understood; it is an X-linked
enzyme deficiency leading to a lipid storage disorder. Lipid storage occurs in a wide
variety of cell types, and consequently there are a wide variety of signs and symptoms
from different organ systems . . . However, [there] is widespread belief that a number of
the organ injury manifestations are related to vascular injury. It is believed that while this
may not be the sole pathologic process, progressive substrate accumulation within
vascular wails will ultimately lead to local vessel occlusion, with organ impairment as a
consequence." (Medical Review #2 at p. 3).

A28

Part 3. Understanding of the Relationship Between Near-Elimination of Substrate
in the Renal Capillary Endothelium and the Outcomes of Fabry's Disease including
Renal Failure and Mortality
"Vascular injury does appear to be an important mechanism of promoting the
progressive organism impairment, and substrate accumulation within vascular walls is the
basis for this. The exact (quantitative) relationship between the amount of substrate
accumulation and the degree or rate of vascular ischemia is unknown and not addressed
in any information presented by [the Sponsor]. It is unknown if reducing substrate
accumulation by half might show vascular injury by half, or if there is a threshold effect,
wherein some specific amount of accumulation will invariably lead to vascular occlusion
and thus no change in the clinical expression of the disease. However, by focusing upon
a near-elimination of all accumulation within a specific cell type [the Sponsor's] data
appear to overcome these concerns." (Medical Review #2 at p. 3).
"Following FDA requests to [the Sponsor], additional data were submitted which
demonstrated that while not all cell types show a marked decrease in substrate
accumulation (e.g., renal podocytes, with a limited degree of reduction in substrate
accumulation) there are a variety of cell types with moderate and several that show
marked reduction in substrate accumulation." (Medical Review #2 at p. 1).
As for understanding the relationship of drugs in the same or closely related
pharmacologic class on near-elimination of substrate in specific cell types and Falmy's
disease, there were no other drugs approved at that time, and there was only one other
drug with controlled clinical studies in Fabry's disease, Replagal.
Part 4. Clinical Evidence on Substrate Reduction in Certain Cell Types and
Fabry's Disease Outcomes
a. Substrate Reduction
The primary endpoint in the 58 patient, placebo-controlled randomized trial was
clearance (that is, elimination) of kidney intestinal capillary endothelium GL-3 inclusions
(or substrate). While none of the 29 placebo subjects achieved a score of "zero" GL-3
inclusions over the 5 month duration of the trial, 20 of the 29 Fabrazyme subjects
"cleared" all substrate (p<0.001) (Medical Review #1 at p. 30).

b. Clinical Outcomes
"The clinical trials failed to show clinical benefit on a wide range of tests of
neurologic, renal, and cardiac function. This finding weakens confidence in the clinical
importance of the reduction of kidney interstitial capillary endothelial cell GL-3 [enzyme
substrate] that constituted the primary endpoint of the pivotal trail." (Medical Review #1
at p. 74).
A29

In the pivotal study, there was only one secondary endpoint that assessed a clinical
outcome, and that was pain. In the five ways in which pain was assessed, the placebo
group outperformed the treated group in 4 of the 5 measures of pain. (Medical Review
#1 at pp. 35-36). There were tertiary endpoints that assessed clinical outcomes and in
eight of these, there were no numerical between-group differences, and in one measure of
neuropathy, the placebo group fared somewhat better and in two measures (symptom-free
days and episode-free days), the Fabrazyrne group fared somewhat better. Of interest,
renal function was assessed by Inulin-GFR and by serum cystatin-C and on both of these
measures of renal function, there were essentially no numerical differences between
placebo and Fabrazyme groups. Among "other" endpoints, there were ophthalmic
assessments, and "the ophthalmological findings, like the tertiary endpoints, did not show
a clinical change effected by the product." (Medical Review #1 at pp. 39-42).

A30

10. REMODULIN (trenostinil)
The May 21, 2002 approval of Remodulin for treating pulmonary hypertension
(now referred to as pulmonary arterial hypertension or PAH) was based on an
intermediate clinical endpoint of 6-minute walk (6MW) test, a measure of exercise
capacity that is a clinical endpoint, but not the ultimate clinical outcome of this serious
disease.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition

PAH is a serious, life-threatening condition.
b. Rarity of the Condition

FDA designated Remodulin for PAH an orphan drug on June 4, 1997.
c. Lack of Available Therapy

The only other therapy approved before Remodulin was Flolan, whose labeling
states that "8 of 40 patients receiving standard therapy alone died, whereas none of the 41
patients receiving Flolan died (p=0.003)." (Medical Review, Mar. 28, 2001, p. 55). This
same Medical Review states also that Flolan's "use is difficult and inconvenient. The
infusion of Flolan requires the insertion of an indwelling central catheter with the . . .
subsequent risk of catheter infection . . . Any inadvertent interruption of the infusion is
potentially life-threatening." (Medical Review at p. 55).
d. External Expertise

The Cardiovascular and Renal Drugs Advisory Committee, on August 9, 2001,
voted 6 to 3 in favor of approving Remodulin.
Part 2. Understanding of the Disease
The pathophysiology of PAH is well-understood.
Part 3. Understanding of Relationship Between 6MW Results and Clinical
Worsening of PAH
Exercise capacity as measured by the 6MW test was judged by FDA as reasonably
likely to predict clinical benefit which was determined to be clinical worsening of PAH
symptoms. Confirmation of FDA's decision to rely upon the 6MW test results as
predictive of clinical benefit was later seen in that this same measure, 6MW, was the
basis for the approval of several subsequent PAH therapies, especially after this
Sponsor's successful completion of its Phase 4 confirmatory trial established
Remodulin's effect on preventing clinical worsening (p<0.001). The Sponsor's Phase 4
A31

trial results on clinical worsening demonstrated the positive predictive value of the 6MW
test results with Remodulin.

Part 4. Clinical Evidence on 6MW and on Clinical Worsening or Mortality
The primary endpoint of the pivotal trials was "change in [6 minute]
walking distance from baseline at the end of week 12 . . . The database was to be
considered demonstrating a benefit for [Remodulin] if either both studies where by
themselves significant at the p<0.049 or if one study was significant (P<0.049) and the
pooled studies had a p-value of less than 0.01 . . . Neither of the studies demonstrated a pvalue of <0.049 (p=0.06 for both studies), although the pooled studies demonstrated an
overall p-value of <0.01 (p = 0.006 for the pooled studies)." (Medical Review at p. 10).
In the pivotal [Remodulin] studies, the drug demonstrated no mortality benefit. (Medical
Review at p. 14).

A32

11. CIPRO (ciprofloxacin hydrochloride)

On August 30, 2000, FDA approved a supplemental NDA for Cipro for
prophylaxis after exposure to inhalational anthrax. There was a two-part or "compound"
surrogate for this approval in that FDA concluded: (1) that Cipro reduced "the rate of
death due to anthrax over control in the macaque monkey model," (Statistical Review,
Aug. 16, 2000, p. 3), and (2) "that [Cipro] serum concentrations achievable in human
populations reach or exceed those associated with improved survival in animals exposed
to aerosol challenge with spores of B. anthracis [in that] serum concentrations in both
human and animal populations consistently exceed the MIC 90 of the causative
organism."11 (Medical Review, August 31, 2000, 12 p. 34).
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition

"The mortality rate of inhalational anthrax is as high as 80-100% . . ." (Statistical
Review, August 16, 2000, p. 1).
b. Rarity of Condition

"[I]nhalational anthrax is extremely rare. There have been only approximately 20
cases in the United States in the past 100 years . . . For these two reasons, the rarity of
disease and the extremely high mortality rate, a clinical study is not feasible." (Statistical
Review at p. 1). Although the prevalence of inhalation of anthrax is sufficiently low, the
sponsor did not seek orphan drug designation.
c. Lack of Available Therapy

"There are drugs with currently approved labeling by FDA for disease associated
with B. anthracis. Labels for penicillin, tetracycline, doxycycline, and minocycline
products list B. anthracis among the organisms susceptible to these agents. None of these
agents is indicated specifically for post-exposure prophylaxis for disease caused by
inhaled B. anthracis" (Medical Review at p. 2).
d. External Expertise

The Anti-Infective Drug Products Advisory Committee on July 28, 2000
unanimously voted "yes" to the question: "Do the data presented support the safety and
11

Obviously, there was no requirement for a Phase 4 confirmatory story and the
commentators hope there is never any open-label uncontrolled anecdotal evidence
obtained.

12

The Medical Review was completed, signed and dated the day after the approval.
A33

efficacy of [Cipro] for post-exposure prophylaxis of inhalational anthrax?" (Medical
Review at p. 33).
Part 2. Understanding of the Disease Process

"The inhalational form of the disease, which affects the mediastinal lymph nodes,
other organs of the reticuloendothelial system and the central nervous system, is
considered the most likely clinical entity resulting from the intentional use of an
aerosolized preparation of the spores of B. anthracis." (Medical Review at p. 2).

Part 3. Understanding of the Relationship Between the Monkey Studies and Human
Mortality and Part 4. Clinical Evidence

First, "the p-value comparing the death rate of [Cipro] to that of control is highly
significant (p=0.0011) showing that the treatment with [Cipro] significantly reduces the
rate of death due to anthrax over control in the macaque monkey model." (Statistical
Review at p. 3).
Second, as for comparative monkey/human exposure levels, the data
"demonstrates that [Cipro] peak and trough serum concentrations achieved in the Rhesus
monkey are reached or exceeded in human populations receiving the doses recommended
for the post-exposure inhalational anthrax. Peak and trough concentrations reported in
both monkey and human populations are shown to consistently exceed 0.06 mcg/ml, the
value of the MIC90 for B. anthracis." (Medical Review at p. 10).
As for understanding the relationship of drugs in the same or closely related class
on the compound surrogate, see above discussion under 1.c. regarding other drugs
approved for anthrax, but note that none had evidence that assessed their utility
specifically against post-exposure inhalational anthrax.
Monkey survival data was one part of this unusual compared surrogate, see the
discussion above. Moreover, the Medical Reviewer stated: "There have been no
prospective studies performed that link clinical outcome to drug exposure for infection
with B. anthracis. However, in general, when there is a demonstrated relationship
between plasma concentrations of drug and response, pharmacokinetic data may be used
as one way to relate dose and possible outcome." (Medical Review at p. 14).

A34

12. CELEBREX (celecoxib)
FDA's December 23, 1999 approval of a supplemental NDA for Celebrex to
reduce the risk of colorectal cancer in patients with familial adenomatous polyposis
(FAP) was based on a surrogate endpoint which was reduction in colorectal polyps.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
"The average life expectancy for patients with untreated FAP has been estimated
to be 42 years." (Medical Review, Dec. 22, 1999, P. 25).
b. Rarity of the Condition
"The frequency of the FAP gene has been estimated on the basis of disease
prevalence to be 1 in 5,000 to 1 in 7,500." (Medical Review at p. 22). Although the
prevalence of FAP is sufficiently low, the Sponsor did not seek orphan drug designation.
c. Lack of Available Therapy
"Surgical therapy is the only acceptable option for patients with FAP after colonic
polyps have been detected." (Medical Review at p. 26).
d. Use of External Expertise
Here are the recommendations of the Oncologic Drugs Advisory Committee that
met on December 14, 1999:

i.

Do you believe that a reduction in colorectal polyps count in FAP
patients in focal areas of some magnitude is "reasonably likely" to
predict benefit?
Yes: 13

Abstain: 2

No: 0

Do you believe that the observed reduction (about 25% at 6 months) is
likely to predict benefit in FAP patients?
Yes: 12

Abstain: 3

No: 0

iii. Do you recommend approval of Celebrex under the accelerated
approval rule for treatment of FAP?
Yes: 14

No: 0

Abstain: 1

(Medical Review at pp. 76-77).
Part 2. Understanding of the Disease Process on Polyp Counts on Colon Cancer
A35

"FAP is characterized by the presence of hundreds to thousands of colorectal
adenomatous polyps and the inevitable development of colon cancer . . . The disease
results from germ line mutations of the APC gene . . . The APC gene is thus believed to
be a tumor suppressor gene." (Medical Review at pp. 22-23). "A significant body of
evidence suggests that cellular expression of COX-2 is prominent in several types of
tumors, including colon . . . as well as pre-cancerous changes such as Barrett's
esophagus, the adenomatous polyp and actinic keratosis." (Medical Review at p. 15).
Part 3. Understanding of the Relationship Between Reducing Polyp Counts and
Colon Cancer

"Celebrex was evaluated in two models of colon cancer. The MM mouse model
represents a genetic model of human FAP . . . Adenomas and adenocarcinomas of the
colon can be chemically induced in rats by administration of azoxymethane." Celebrex
was shown to prevent or inhibit colorectal tumor development in both of these animal
models. (Medical Review at pp. 16-17).
As for understanding the relationship of drugs in the same or a closely-related
class on FAP polyp counts, "studies have shown that Sulindac, one of the non-selective
NSAIDs, induces apoptosis . . . Recent study of COX-2 inhibitors showed that inhibition
of COX-2 produced sequential increases in arachidonic acid and ceramide, the latter a
potent stimulant of apoptosis. Furthermore, in vitro evidence exists that angiogenesis is
regulated by COX-2 expression in colon cancer cells. Therefore, another mechanism by
which tumor growth may be inhibited by COX-2 inhibitor is through blockade of
angiogenesis and tumor vascularization." (Medical Review at pp. 15-16).
Part 4. Clinical Evidence on Polyp Counts and on Colon Cancer

"A single, randomized, double-blind, placebo-controlled study has been submitted.
A total of 83 patients received treatment with either placebo, Celebrex 100mg BID, or
Celebrex 400mg BID for 6 months (with a 1:2:2 randomization) . . . The mean reduction
in colorectal polyps count was 28% on the Celebrex 400mg BID arm, 15% on the
Celebrex 100mg BID arm and 5% on placebo. Only treatment with Celebrex 400mg BID
was associated with a statistically superior mean reduction in polyp counts, with
p=0.003." (Medical Review at pp. 1-2). In a six-month study there were, as expected, no
cases of colon cancer in any arm of the trial.

A36

13. SYNERCID (dalfonristin/quinupristin)
The FDA approval of Synercid on September 21, 1999 was for treating patients
with vancomycin-resistant Enterococcus faecium (VREF) and was based on a surrogate
showing of clearance of the VREF bacteremia.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
"The mortality rates in both [pivotal] studies [was] approximately 50%."
(Statistical Review, Mar. 5, 1998, p. 17).
b. Rarity of the Condition
The Sponsor has no intention of developing Synercid for this use, but a "rise in the
United States in both the number of nosocomial infections due to E. faecium and in the
proportion of strains of this pathogen found to be vancomycin-resistant, led to increasing
requests for the emergency use of Synercid." (Medical Review, Aug. 21, 1998, P. 2).
Synercid appears not to have been granted orphan drug designation. Given the Sponsor's
reluctance to submit an NDA for this use, the Sponsor likely never had applied for
designation, even though the condition was rare.
c. Lack of Available Therapy
Those patients that enrolled in the two pivotal trials were only those "infected with
VREF who did not have any other therapeutic option." (Statistical Review at p. 2).

d. External Expertise
On February 19, 1998, the Anti-Infective Drugs Advisory Committee voted 9 to 1
in favor of approval of Synercid for VREF.
Part 2. Understanding of the Disease Process
The understanding of the pathophysiology of infections with vancomycin-resistant
strains of Enterococcus faecium is well-known.
Part 3. Understanding of the Relationship Between Clearance of the VREF
Bacteremia and Mortality (and Other IDSA/FDA Guideline Clinically Meaningful
Endpoints)
"The VREF literature is clear that VREF bacteremia . . . should be treated and that
clearance of VREF from the bloodstream can be seen as beneficial to the patient . .
There is consensus that bacteremia should be treated. Thus, while clearance of
bacteremia is not a clinical benefit by itself, it can be seen as likely to predict clinical
A37

benefit. Thus, it is proposed that the clearance of VREF bacteremia be viewed as a
surrogate endpoint likely to predict clinical resolution of infection." (Medical Review at
p. 32).
Part 4. Clinical Evidence on VREF Bacteremia Clearance and Mortality
FDA concluded that the four emergency use VREF studies did not provide
evidence of an improvement in mortality or resolution of infection due to a host of issues.
None of these four studies had a concurrent control and, while FDA had advised that the
lack of concurrent control would be acceptable because it would be unethical to include a
placebo arm, the FDA had stipulated that then the studies either: (1) had to show a
"dramatic improvement in overall mortality as compared to a historical perspective"
(Medical Review at p. 30) and these studies did not (these four studies had mortality rates
of 48.8%, 49.5%, 53.8% and 54.0% compared to the VREF literature reporting "allcause" mortality rates in the range of 30% to 70%) (Medical Review at p. 18), or (2) had
to have a historical control and this was not established (Medical Review at pp. 18-19).
While two of the four studies, according to the FDA Medical Reviewer,
established clearance of VREF bacteremia, only 18% of the patients in these emergency
use studies were "evaluable" due primarily to missing data, and there was a low response
rate as well. (Medical Review at pp. 19-20, 29-32). In addition, "in the unevaluable
patients who died on therapy but with negative blood cultures, there is 'apparent'
clearance of the organism." (Medical Review at p. 32).

A38

14. REMICADE (infliximab)
The August 24, 1998 FDA approval of Remicade to treat patients with Crohn's
disease was based on an intermediate clinical endpoint of a clinical response defined as a
reduction in the Crohn's Disease Activity Index (CDAI) of at least 70 points at the 4week evaluation.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
"The prognosis for Crohn's disease is generally unfavorable . . . The mortality rate
increases with the duration of disease and most likely ranges from 5% to 10%. Most
deaths occur from peritonitis and sepsis." (Medical Review, July 10, 1998, p. 4).
b. Rarity of the Condition
"In . . . the United States, the prevalence is estimated at 20 to 40 per 100,000."
(Medical Review at p. 3). Remicade was designated as an orphan drug on November 14,
1985.
c. Lack of Available Therapy
The FDA Medical Review surveys all the therapies being used and at the time, no
robustly effective therapies were available. "Because its cause is unknown, medical
management of the disease is largely empirical and is designed to reduce inflammation."
(Medical Review at p. 5).
d. External Expertise
On May 28, 1998, the Anti-Infective and Gastrointestinal Drug Advisory
Committees voted unanimously in favor of approval for both: treatment of patients with
moderate-severe inflammatory disease refractory to conventional therapy, and treatment
of patients with fistulizing Crohn's disease for the reduction in the number of draining
enterocutaneous fistula(s).
Part 2. Understanding of the Disease Process
"Crohn's disease most likely represents a heterogeneous group of disorders. After
much effort that has focused on the identification of a specific pathogenic cause, it is
being recognized that disease manifestations could result from a combination of any, or
all of, a number of factors." (Medical Review at p. 2).
Part 3. Understanding of the Predictive Potential of a 70 Point Change in CDAI at
Week 4 on Crohn's Disease

A39

"Pathologic review of biopsy.. . . often can aid in . . . measurement of extent and
severity of disease. Pathologically, Crohn's disease is described as a transmural disease
with focal or microscopic skip areas of inflammation in the lamina propria. The degree
of inflammation in the most heavily involved area often is an accurate assessment of the
severity of disease . . . Disease activity indices are used to objectively measure the
activity of disease for judgment of response in clinical trials. The [CDAI] was
developed . .. [in] 1979. . . to objectively assess response to therapy. . . . Although
imperfect and cumbersome, e.g., requirement of recording of symptoms for 7 days and
for hematocrits, the CDAI remains the most commonly [used] index." (Medical Review
at p. 4).
As for understanding the relationship between drugs in the same pharmacologic
class, Remicade is a chimeric monoclonal antibody to Tumor Necrosis Factor (TNF). As
such, Remicade was the first of this kind in a new class of immunomodulatory drugs.
Other immunomodulatory drugs, including azathioprine, mercaptopurene, cyclosporine,
and methotrexate were accepted for use for long-term treatment of some Crohn's
patients. "The mechanism of action of these drugs may involve inhibition of lymphocyte
function, primarily that of T cells." (Medical Review at p. 5). As such, they have a
different mechanism of action than Remicade.
Part 4. Clinical Evidence on CDAI and on Long-Term Clinical Benefit

Study T16, a placebo-controlled, dose-ranging (n=108) study, "was designed as a
Phase 2 trial to determine an effective dose in the acute treatment of patients with active
Crohn's disease not responding to immunosuppressant therapy and to explore
maintenance therapy with a single dose in patients who responded initially. This clinical
trial became the pivotal trial for licensure of [Remicade] for this indication." (Medical
Review at p. 10). "65.1% of the [Remicade] treated patients achieved a clinical response
(> 70-point reduction from baseline in the CDAI) at the week 4 evaluation compared to
16.7% of the placebo patients (p<0.00l) . . . There was no apparent relationship between
[Remicade dose] [5mg/kg, 10mg/kg, 20mg/kg] and the proportion of patients
responsding; the highest clinical response was observed in the 5mg/kg dose group
(81.5%; p<0.001 vs placebo)." (Medical Review at p. 19).
"Summary Conclusions on the Review of the Safety and Efficacy Data. "The
Sponsor has presented phase 2 clinical data results to support licensing of a potent, novel
immunomodulating agent for the management of patients with Crohn's disease, a chronic
debilitating disease . . . The number of patients with moderate to severe disease who have
received the proposed dose of 5mg/kg . . . is very low (n=28) and no patients have
received chronic retreatment with 5mg/kg every 8 weeks as proposed in the original
submission. The effects of a single dose [last] approximately 12-16 weeks, compatible
with the half-life of the compound. For patients with fistula, although the majority of
patients experienced stoppage of drainage in two weeks, there are no data on internal
healing of the fistula canal. Once [Remicade] was stopped the effect of therapy was lost.
A40

In summary, there are inadequate data to support the long-term benefit of [Remicade] in
patients with either fistulizing on moderate/severe disease." (Medical Review at p. 81).
From the conclusions of the Medical and Statistical Reviews, there appear to have
been some concerns among FDA Reviewers as to the appropriateness of the short-term
(CDAI improvement after 4 weeks) surrogate endpoint as being adequate to predict longterm benefit in a chronic disease. The conclusion of the Statistician on Study T16 in
moderate to severe Crohn's disease patients was redacted from the publicly available
version of the Statistician's Review. However, there was a second Phase 2 study in
patients with Crohn's disease with fistula, Study T20, which is referred to in the
conclusions of the Medical Review. From the information in the Statistician's Review of
Study T16 that was made publicly available, it would seem that the Statistician's
conclusions with respect to Study T20 may have closely paralleled those for Study T16.
With respect to Study T20, here are the Statistician's conclusions: "Although the
differences in response rates between the placebo group and the [Remicade]-treated
groups were statistically significant, questions remain about the durability of response.
Patients received doses at weeks 2, 4, and 6, but this dosing strategy should be thought of
as one-time dosing. After 6 months of follow-up, the drug effect had disappeared and the
proportion of responding patients in the placebo arm was similar to the proportions in the
treatment arms. The data suggest, therefore, that although this agent has an initial
beneficial effect on Crohn's disease, a single set of doses is unlikely to provide durable
benefit in this chronic disease. There are no data to assess chronic use of [Remicade] for
this indication. There is no information regarding the formation of neutralizing
antibodies (HACA) with repeated dosing and how this may affect the efficacy of this
product. There is also no safety data to allay concerns of a possible increase in
malignancies or serious infections. The Agency should carefully weigh the observed
early benefits seen with this product against the paucity of information regarding the
safety and efficacy of repeated use for this chronic indication." (Statistical Review, Aug.
5, 1998, p. 13).

A41

15. PRIFTIN (rifapentine)

On June 22, 1998 FDA approved Priftin for treating pulmonary tuberculosis (TB)
and this approval was based on a surrogate of a 6-month relapse rate as contrasted with
the standard 2-year relapse rate information for a traditional approval.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition

"[TB] is the leading infectious cause of morbidity and mortality worldwide."
(Medical Review, June 19, 1998, p. 5).
b. Rarity of the Condition

"In 1990, there were 25,701 new cases of TB reported in the [U.S1" (Medical
Review at p. 5). Priftin was designated as an orphan drug on June 9, 1995.
c. Lack of Available Therapy

"During development of rifapentine for TB, the applicant was encouraged to
submit 6 month follow-up data from one study, under the accelerated approval
regulations (21 CFR 314 Subpart H). There is a need for new anti-tuberculosis
medications, and for medications which will potentially increase the adherence to dosing
thereby decreasing the potential for the development of resistant organisms. It was
anticipated that rifapentine would be such an agent. Six-month relapse data would serve
as a surrogate for two-year relapse data predictive of long term clinical benefit."
(Medical Review at p. 8). FDA had previously approved rifampin for use in treating TB.
d. Use of External Expertise

At the Anti-Viral Advisory Committee Hearing on May 5, 1998, "the committee
voted to recommend approval of [Priftin] for the treatment of pulmonary tuberculosis,
with only one dissenting vote." (Medical Review at p. 61).
Part 2. Understanding of the Issues

In this case, the pathophysiology of TB is well-understood.
Part 3. Understanding the Predictive Potential of a Six-Month Relapse Rate on Two
Year Relapse Rate and on Mortality

The Medical Review stated that: "It is expected that the majority of relapses will
occur by 6 months of follow-up, however, the 'gold standard' is 2-year relapse rate."
(Medical Review at p. 19). However, the pattern of relapses for [Piftin] does not appear
to reflect the same showing of relapses in the latter half of six-month follow-up that was
seen for rifampin in the pivotal study. See discussion of results under Section 4.
A42

Part 4. Clinical Evidence on Six-Month and Two-Year Relapse Rates

The single pivotal trial was an open-label, randomized, two-arm parallel, rifampincontrolled trial with 570 patients in the modified ITT analysis. "The primary efficacy
endpoint for this accelerated approval review was treatment outcome at the end of 12
months (6 months of active treatment + 6 months of follow-up). This was a binary
variable with success defined as achieving a negative sputum culture during active
treatment and sustaining it to the end of [6] months of follow-up." (Statistical Review,
July 27, 1998, p. 3).
"There is essential equivalence for [negative sputum culture] rates at the end of
[the 6-month active treatment] between the rifampin [83% negative sputum cultures] and
[Priftin] [88%] arms." (Medical Review at p. 39). However, "[t]here is a statistically
significant difference between the treatment arms for relapse . . . The risk is 5% for
rifampin . . . and 11% for [Priftin]." (Medical Review at p. 40). The Statistical and
Medical Reviews agree that while 10 of the 11 relapses on rifampin occurred within the
first 6 months of follow-up, 7 relapses occurred in the [Priftin] arm at time points
between 6 and 12 months of follow-up. (Note: While the endpoint was at 6 months of
follow-up, almost all subjects had had 12 months of follow-up, so FDA analyzed the 12
months of follow-up data as well and noted that the Priftin arm continued to experience
sizable numbers of relapses beyond the first 6 months of follow-up, which was much
different than the pattern of relapses observed for rifampin).
Despite the above discrepancy between the rifampin and Priftin arms in relapse
rate beyond 6 months, the FDA reviewers seemed (as well as the Advisory Committee
members) to believe that this may reflect lack of optimized dosing of Priftin, rather than a
lack of confidence in the prognostic surrogate of 6-month relapse rate predicting 2-year
relapse rate, and eventually, survival. However, at the time of approval there appear to
be no clinical evidence of Priftin on two year relapse rate or on mortality.

A43

16. SULFAMYLON (mafenide acetate)
FDA approved Sulfamylon on June 5, 1998 "to control bacterial infection when
used under moist dressings over meshed autografts on excised burn wounds." The
approval was based on an intermediate clinical endpoint of evidence derived from
patients who were burned over up to 20% of their total body surface area (TBSA) with a
Phase 4 commitment to conduct a confirmatory trial in patients with 20% to 60% TBSA
thermal injuries.

Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition
The Medical Review commenting on the results of the single pivotal trial (done
exclusively in children) observed the following: "It is remarkable that so many of these
severely burned children survived to leave the hospital . . . It is not unexpected that
survival rates fall as TBSA burned increases." (Medical Review, Sept. 23, 1997, p. 17).
Large [TBSA] burns are serious and life-threatening." (Medical Review at p. 49).
b. Rarity of the Condition
The number of persons in the country in need of such care is small, thankfully,
very small. FDA designated Sulfamylon as an orphan drug for this use for two different
sponsors at separate times: on August 29, 1985 and on July 18, 1990. (Medical Review
at p. 3).
c. Lack of Available Therapy
"There is no existing approved treatment for these burn patients who require
excision and meshed autografts." (Medical Review at p. 50).
d. External Expertise
"Sulfamylon [was] discussed by the FDA Anti-Infective Drug Products Advisory
Committee [on July 24, 1996]. The Committee concluded that since topical antimicrobial
solutions had evolved to a standard of care [(SOC)] over the last 20 years, a placebocontrolled study would be unethical." (Medical Review at p. 3).

Part 2. Understanding of the Disease Process
"There is adequate evidence available in the literature to establish that wounds,
including burn wounds, may be expected to progress satisfactorily if the microbial load
present is reduced to less than 10 5 organisms per gram of tissue . . . it may be said that if a
topical antimicrobial is successful in maintaining low bacterial levels on a newly placed
skin graft until the graft is adequately vascularized, the antimicrobial has contributed to
take of the graft." (Medical Review at p. 42).
A44

Part 3. Understanding of the Relationship Between the Treatment Failures in those
with <20% TBSA Burned and Treatment Failures in those with >20% TBSA
Burned
"The applicants have been reluctant to use a vehicle control on the grounds that
failure to treat a burn patient with a [TBSA] bum of larger than 10-20% would be
unethical." (Medical Review at p. 4). This was supported by the deliberations of the
Advisory Committee. Therefore, while the single pivotal trial enrolled all patients with
bums, regardless of how extensively the body was burned, there was "no protocolspecified assignment of patients to treatment with [either Sulfamylon or standard of care
(SOC)]. This was a medical decision, made by the attending physician . . . The reviewers
separated the results into patient groups by TBSA burned. All patients who had burns
covering more than 40% TBSA were treated with [Sulfamylon] . . . It is impossible to
assess the effect of [Sulfamylon] in this group. In the 20-40% TBSA burn group, there
were a few patients who received [SOC] but . . . the contribution of [Sulfamylon] is
difficult to quantify. However, there [were] sufficient [SOC] patients in the 0-20%
TBSA burn group to permit comparison of the two treatment regimens." (Medical
Review at p. 48).
As for understanding the relationship between drugs in the same pharmacologic
class as Sulfamylon, "Sulfamylon for 5% Topical Solution" is the drug product that was
the subject of this NDA. However, "Sulfamylon cream is currently approved for use in
the treatment of second and third degree burns and the proposed indication for the
Sulfamylon 5% solution is related. (Medical Review at p. 49). "Because of the pain
caused by the cream, burn physicians began to make a 5% solution using mafenide
acetate power in the mid-1970s . . . and the 5% solution has become the standard of use
in some bum units for maintaining skin grafts in the period between graft placement and
graft take." (Medical Review at p. 4).
Part 4. Clinical Evidence on Those <20% and Those >20% TBSA Burned
The single pivotal efficacy study was an unblinded, retrospective, nonrandomized, parallel group study with an active control of Standard of Care (SOC) and
was conducted at a single site and with a single investigator: Dr. Glenn Warden at
Shriner's Burn Institute in Cincinnati, Ohio.
In this study, among the 229 procedures in persons with less than 20% TBSA
burned, there were 19 (19%) who were "treatment failures" in those treated with
Sulfamylon compared to 33 (26%) who failed on SOC. However, those treated with
Sulfamylon had more serious bums, that is, third-degree bums (6.5% vs. 3.3% SOC), a
higher percentage of the body surface area burned (10.6% vs. 7.0% SOC), and fewer with
only less serious burns, that is, those with second-degree bums only (4.4% vs. 17.3%
SOC).

A45

"In her review, the reviewing Statistician, Dr. YuIan Li, reached the following
conclusion: Based on the Cincinnati study, the applicant has demonstrated that the use of
[Sulfamylon] is associated with the decreasing of treatment failure in the subgroup of
patients with 0-20% TBSA burns after separately adjusting for etiology and degree of
burn. However, it is unknown whether . . . treatment failure reflects the benefit of
[Sulfamylon] due to non-random treatment assignment and investigator knowledge of
treatment at the time treatment failure was assessed." (Medical Review at p. 6).
While there appears to be no disagreement in any FDA review as to the
intermediate clinical endpoint of effect in those with less than 20% TBSA burned as
"reasonably likely to predict benefit" in those with burns over more than 20% TBSA;
there were concerns expressed, especially by the Statistician, as to the strength of the
efficacy evidence for the findings in those with less than 20% TBSA burned: 3

13

While scored as a "1," the strength of clinical evidence on the surrogate here with
Sulfamylon could reasonably be scored as either "1" or "zero," and the same may
be said of the strength of clinical evidence for the surrogate in Synercid, Precedent
#13.
A46

17. PROAMATINE (midodrine hydrochloride)

FDA approved Proamatine for treating "symptomatic orthostatic hypotension" on
September 6, 1996 on the basis of "increases in 1-minute standing systolic blood
pressure, a surrogate marker likely to correspond to a clinical benefit" (as stated in FDAapproved labeling)."
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Condition and Lack of Alternative Therapy

Although the review documents for Proamatine are not publicly available on
FDA's website, the Agency's Subpart H approval of Proamatine has to mean that FDA
assessed the condition as rather serious and lacking available therapy.
b. Rarity of the Condition

Proamatine was designated as an orphan drug on June 21, 1985.
c. External Expertise

There is no evidence from documents currently available, including approved
labeling and trade press, whether FDA sought the advice of an Advisory Committee.
Therefore, we scored this as a "zero."
Part 2. Understanding of the Disease

For FDA to have approved Proamatine on the basis of a change in 1-minute
systolic blood pressure suggests that FDA must have considered that there was a sound
understanding of the pathophysiology of the disease.
Part 3. Understanding of the Relationship Between Change in 1-Minute Systolic
Blood Pressure and the Ability to Perform Life Activities

Since there are no other drugs in any class approved for this condition, FDA could
not have relied upon their effects on this disease. However, many drugs are approved on
changes in blood pressure as a validated surrogate based upon both robust epidemiology
and multiple interventions affecting serious cardiovascular outcomes such as MACE and
I-, DA may have relied upon this strong association for support of the power of a change in
1-minute systolic blood pressure in this disease to predict clinical benefit in this disease.
Part 4. Clinical Evidence on 1-Minute Systolic Blood Pressure and Clinical
Outcome
14

All of the formation in this analysis is drawn from the FDA approved labeling as
no Medical or Statistical Reviews from FDA were publicly available.
A47

"Midodrine has been studied in 3 principal controlled trials, one of 3-weeks
duration and two of 1-to-2 days duration. All studies were randomized, double-blind and
parallel-design trials in patients with orthostatic hypertension of any etiology and supineto-standing fall of systolic blood pressure of at least 15 mmHg accompanied by at least
moderate dizziness/lightheadedness . . . In the 3-week study in 170 patients . . . , the
midodrine-treated patients . . . had significantly higher (by about 20 mmHg) 1-minute
standing systolic pressure 1 hour after dosing . . . for all 3 weeks. After week 1,
midodrine-treated patients had small improvements in
dizziness/lightheadedness/unsteadiness scores and global evaluations, but these effects
were made difficult to interpret by a high early drop-out rate (about 25% vs. 5% on
placebo). Supine and sitting blood pressure rose 16/8 and 20/10 mmHg, respectively, on
average. In the 2-day study, after open-label midodrine, known midodrine responders
received midodrine 10 mg or placebo at 0, 3, and 6 hours. One-minute standing systolic
blood pressures were increased 1 hour after each dose by about 15 mmHg and 3 hours
after each dose by about 12 mmHg; 3-minute standing pressures were increased also at 1,
but not 3, hours after dosing. There were increases in standing time seen intermittently 1
hour after dosing, but not at 3 hours. In the 1-day, dose-response trial, single does of 0,
2.5, 10, and 20 mg of midodrine were given to 25 patients. The 10- and 20-mg doses
produced an increase in standing 1-minute systolic pressure of about 30 mmHg at 1 hour;
the increase was sustained in part for 2 hours after 10 mg and 4 hours after 20 mg.
Supine systolic pressure was =200 mmHg in 22% of patients on 10 mg and 45% of
patients on 20 mg; elevated pressures often lasted 6 hours or more." (Midodrine Package
Insert).

A48

18. BIAXIN (clarithromvcin)
FDA approved Biaxin on December 23, 1993 for treating disseminated
mycobacterial infections due to mycobacterium avium complex (MAC) on the basis of a
showing of Biaxin's effect on the surrogate of decreases in MAC bacteria.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity of the Disease, Rarity, and Lack of Alternative Therapy
The pivotal studies were conducted in persons with CDC-defmed AIDS and CD 4
counts <100 cells/pLL, and median survival time in the one trial that was randomized and
blinded was 249 days and 215 days for the two dose groups reported in the approved
labeling. 15
While Biaxin was not designated as an orphan drug for this use, this condition was
not prevalent and the absence of orphan drug status is likely due to the FDA approval of
Biaxin for many other prevalent diseases (such that orphan drug exclusivity would have
had substantially diminished, if any, value).
b. External Expertise
On May 11, 1993, the Antiviral Drugs Advisory Committee provided insight on
the approvability of Biaxin for treatment of MAC. 16
Part 2. Understanding of the Disease
The pathophysiology of MAC in immune-compromised AIDS patients was likely
understood relatively well for the extent of time that the condition had been known.
Part 3. Understanding of the Relationship Between Reducing MAC Bacteremia and
Clinical Outcomes
The general axiomatic principles of infectious disease likely guided and
illuminated FDA's interpretation of the prognostic value of reducing MAC bacteremia on
achieving negative cultures and clinical benefit. Other antibiotic regimens had shown
some value as well.

15

There were no FDA medical or statistical reviews publicly available and nearly all
information is from the FDA approved labeling.

16

Based on public documents currently available, it is unclear what the outcome of
this Advisory Committee was.
A49

Part 4. Clinical Evidence on Reducing MAC Bacteremia and Clinical Outcomes
Including Mortality
Of the 3 studies conducted from May 1991 to March 1992, Study 500 was the only
one to be blinded and randomized (dose comparison trial of 3 different doses of Biaxin).
Study 500 showed a reduction in MAC bacteremia with the lowest dose having the
smallest decrease in colony-forming units (CFUs). There was seemingly no survival
benefit as the FDA-approved labeling reported that: "The median survival times for these
[Biaxin] dosages were similar to recent historical controls with MAC when treated with
combination therapies." However, there was some evidence of improvement in other
signs and symptoms of MAC infection including night sweats, fever, and weight loss.

A50

19. BETASERON (interferon beta-lb)

FDA approved Betaseron as the first therapy to treat multiple sclerosis (MS) on
July 23, 1993 on the basis of a showing on both rate and extent of exacerbations and on
improvement in MRI-measured lesion area.
Part 1. Regulatory Factors Weighing into FDA Determination
a. Severity. Rarity, and Lack of Available Therapy

MS is a serious disease for which, prior to Betaseron, there was no FDA approved
treatment. Betaseron was designated as an orphan drug on November 17, 1988.
b. External Expertise

The FDA Peripheral and Central Nervous System Advisory Committee on March
19, 1993 voted 7-2 to recommend approval of Betaseron.
Part 2. Understanding of the Disease

The pathophysiology of multiple sclerosis was known to a fair degree at the time
of the conduct of the pivotal trial which permitted the Sponsor in collaboration with the
lead FDA CBER official, Dr. Woodcock, and her office to have general agreement on coprimary endpoints of clinical utility related to exacerbations, as well as somatic measures
of the putatively key causal biologic marker, MRI lesion volume.
Part 3. Understanding of the Relationship Between Mill Lesion Volume and
Multiple Sclerosis

"It was also clear that the Committee as a whole placed great weight on the MRI
findings in their deliberations. Specifically, although the clinical benefit, as measured by
the proportion of exacerbation-free patients and exacerbation frequency, was considered
real and of value clinically, the Committee considered the size of the treatment effect to
be relatively small.
However, it was obvious that great emphasis was placed on the MRI findings.
Specifically, the Committee appeared convinced by the firm's presentation that the drug
had an important effect on the underlying pathology as measured by total lesion area as
seen on MRI. The statistically significant decrease in the total lesion area in the high
dose group as compared to placebo patients over the course of the study that the sponsor
claimed was demonstrated was interpreted by the Committee, in my view, as powerful
support for the conclusion that the drug was having an important effect on the underlying
disease process. While the Committee stopped short of declaring that the data proved
the drug had an effect on the progression of the disease, I believe it is fair to characterize
their view with a quote, made at the meeting, by Dr. McFarland, who said at one point,
that, while the sponsor had not proved that the drug had an effect on the course of the
A51

disease, 'I would be amazed if it didn't change the course of disease.' A number of
Committee members explicitly referred to Dr. McFarland's comments in this regard
when explaining their votes." (Memo of Dr. Katz, May 28, 1993, at pp. 356-357)
(emphasis is original).
"That is, it appears clear that the Committee felt that the MM results not only were
consistent with the clinical benefit observed (i.e., the changes seen corresponded to the
exacerbation rate data at a given point in time), but that they could be relied upon to
accurately 'predict' patients' future courses. In other words, the MRI data were
considered, for all intents and purposes, as a surrogate marker for disease." (Memo of
Dr. Katz at p. 359)
"If the lesions detected on MR' are taken to be a better index of the 'activity' of
the pathologic process than are clinical manifestations of MS, (a not unreasonable
possibility given the knowledge that lesions detected on MRI may be unaccompanied by
clinical signs/symptoms when they occur in so-called 'silent' regions of the CNS) and if
the rate of clinical progression of MS (in the sense of increasing physical disability) is a
positive function of the activity of that athologic process, it follows logically that any
drug suppressing this activity" must' 17 have some beneficial effect on the progression of
MS (as manifest by increasing physical disability). Although the clinical evidence
collected 18 in Study TB01-35(6/8)86 does not provide convincing affirmative support for
17

"Must" appears in quotations as a reminder of prior occasions in the history of
therapeutics where perfectly logical extrapolations based on beliefs about the
pathophysiology of a disease and the postulated mechanism of a drug's action
have led experts to reach totally incorrect conclusions about the promise of a
particular drug (e.g., CAST: the suppression of ventricular ectopy 'must' save
lives.) [Footnote is part of quotation]

18

In their report of the study, the sponsor asserts that the correlation between EDSS
disability scores and MRI lesion areas detected at both baseline (r=0.169) and at
the end of year two (r=0.2) establishes that MR1 'burden' predicts disability
(EDSS score). Although these statements are correct in a statistical sense, the
correlation does not tell us what we really seek to learn: whether a treatment
reducing the extent of MRI area increase over time will reduce the extent of
clinical worsening, as judged by EDSS, over the same interval or in a future one.
[Footnote is quotation from the memo cited.]

A52

this hypothesis, that does not necessarily undercut its appeal or its psychological impact
on those asked to render an opinion about the 'therapeutic potential' of Betaseron.
During the PCNS meeting, the sponsor's representatives, several members of the
Committee and, in particular, Dr. Henry McFarland, who was attending the meeting as
the Agency's expert consultant on neuro-imaging and MS, espoused the hypothesis just
described. Although virtually all proponents of this hypothesis acknowledged that the
link between MRI lesion frequency/intensity/area and subsequent outcome (progression
in level of physical disability) in MS was not proven, almost all affirmed that they would
be very surprised if the link was not eventually demonstrated. Thus, for many experts,
the number and area of lesions detected on MRI are tantamount to a 'surrogate' endpoint
that predicts disease progression in MS." (Memo of Dr. Leber, May 28, 1993, pp. 340341) (emphasis in original).
"In the Betaseron data there is a second kind of replication, the MR1 results, which
are more or less persuasive, depending on one's beliefs. At a minimum, as Dr. Leber
says, these data are an independent measurement that supports the clinical finding, a kind
of 'within-study' replication. At best, they are evidence of an effect far more important
than the modest effect on exacerbations. We certainly are not qualified to choose
between these interpretations, but our advisors seem to believe the latter, even though all
would agree that, strictly, the correlation of improved clinical outcome and improved
MR1 has not been made.
It would be possible, we believe, to grant approval under the accelerated approval
regulations, which allow this procedure where a surrogate or clinical, but non-ultimate
endpoint is the basis for approval. (Memo of Dr. Temple to Dr. Woodcock, June 3, 1993,
pp. 329-330) (emphasis in original)
Part 4. Clinical Evidence on MRI Lesion Volume and on Reduction in
Exacerbations of MS
"The trial was designated as a randomized, double-blind, and placebo- controlled
study to evaluate the safety and efficacy of Betaseron in the treatment of patients with
relapsing-remitting MS . . . The protocols proposed that the primary efficacy evaluations
will be based on reduction in frequency of exacerbations per subject and proportion of
exacerbation-free subjects." (Statistical Review, March 1, 1993, at p.1).
"The proportions of exacerbation-free subjects in the three arms of the study are given in
Table 1. If we consider all reported exacerbations, 18 of the 112 placebo patients
(16.1%) and 36 of the 115 45 m111 Betaseron patients (31.3%) were exacerbation-free.
This difference was significant at p=0.008." (Statistical Review at p. 3).
"The second primary endpoint, prospectively specified in the protocol, was the
frequency of exacerbation per subject . . . If we consider the outcomes in all six
categories of exacerbations (i.e., 0, 1, 2, 3, 4, and 5+) then the probability of better
A53

response on Betaseron therapy is 63%. It is significantly different (p=0.0004) from
50%." (Statistical Review at pp. 5-6).
As for the MRI lesion volume results, depending upon the analysis used by the
FDA reviewer, Dr. Jay Siegel, the p-value for the comparison between Betaseron and
placebo arms ranges from a p-value of 0.03 to a p-value of 0.001. (Memo of Dr. Siegel,
June 24, 1993, p. 1)

A54