Thorough Expert Witness Report Explains Risk of Developing PPH While Taking Fen Phen – Part Two of Two

Here is the second and final part of the expert witness report filed by Lemuel A. Moye in Longs v. Wyeth. 

United States District Court, N.D. Ohio.

                               Ramonia LONGS, et al.,

                                         v.

                                    WYETH, et al.

                                   No. 103CV02042.

                                   March 31, 2005.

101 Combined Endpoints in Khan: For Khan, mitral valvulopathy was just one
component of the combined endpoint for cardiac valvulopathy. It is no surprise
that the high odds ratio seen for the combined endpoint could not be reproduced
for mitral valvulopathy if the number of events was small. It would be
unreasonable to conclude that there is no relationship between anorexigen use and
mitral valvulopathy solely because a high odds ratio could not be reproduced for
this low prevalence component of the endpoint. In general, the practice is to
extend the findings for the combined endpoint to each of its components as long as
there is no discordance in the components (i.e. the OR for mitral valvulopathy
suggests a strong protective effect of anorexigens). If this device were not
admissible, it would be impossible to study links between exposure and diseases of
very low prevalence, since the large number of patients required for the studies
would preclude any meaningful analysis.

102 The Epidemiological Study of Jick Jick[27] conducted a study, which confirmed
the findings of Khan, His confirmatory study was a population-based follow-up
study, which included a nested case-control analysis diet drug exposure, to
evaluate the relationship between fenfluramine exposure and the occurrence of
cardiac valvulopathy. In lick's study, out of 9,281 patients who were unexposed to
the fenfluramines, not a single patient had been diagnosed with cardiac valve
disease. Of the 8,903 patients who had been exposed to the fenfluramines, 11 had
cardiac valve disease (cumulative incidence rate 14.2 95% confidence interval 7.8
to 26.2). Some have somewhat naively criticized Jick for not reporting an odds
ratio for this overall effect However, they misunderstand that the odds ratio is
mathematically impossible to compute because no one in the unexposed group had the
disease. Cardiac valvulopathy only occurred for patients taking the fenfluramines.
Criticism of Jick for not computing an odds ratio in this circumstance is like
criticizing a baseball player for not stealing any bases when, in fact, that
player hits only home runs.

103 Underestimating the Odds Ratio. This point is worthy of further elaboration.
Jick's original manuscript of does not compute odds ratios with 95% confidence
intervals for the association between the fenfluramines (i.e. fenfluramine and
dexfenfluramine) with heart valve regurgitation. It may be instructive to examine
the reason for this omission.

104 Odds Ratio:In the odds ratio computation, consider that N1 patients are
exposed to the fenfluramines, and n1 of these exposed patients are found to have
cardiac valve regurgitation. Similarly, let there be N2 patients who are unexposed
to the fenfluramines, and n2 of them are determined to have cardiac valve
regurgitation. The crude odds ratio is

TABLE

105 If we try to apply this formula to lick, we see from the data of Table 3, page
722 that there were 11 patients with cardiac valve regurgitation out of 8903
patients exposed to the fenfluramines. There were 0 patients with cardiac valve
regurgitation out of 9281 untreated patients. Applying these results to equation
(0.1) we see that

TABLE

106 Odds Ratio and Strength of Association: The strength of most epidemiologic
relationships is weak enough that formula (0.1) can be used. Not so with Jick. In
his manuscript, the number of patients in the unexposed group with cardiac valve
regurgitation, n2, is zero. If this is placed into the formula for the crude odds
ratio above, the odds ratio becomes infinity. This is an extremely powerful
result. All of the disease occurred in the group with the exposure to the
fenfluramines. This is among the strongest in the universe of findings in
epidemiologic studies.

107 Odds Ratio and Jick: As an example of a computation from the Jick manuscript,
we may consider the hypothetical circumstance where one patient out of the 9281
patients unexposed to the fenfluramines is found to have cardiac valve
regurgitation. From Table 3 of Jick's manuscript (page 722), we know that there
are 11 patients who had cardiac valve regurgitation out of 8903 patients exposed
to the fenfluramines. Having weakened Jick's findings by increasing n2 from 0 to
1, we may now apply formula (0.1) to compute

TABLE

108 The 95% confidence interval for this estimate can be computed from Kleinbaum,
Kupper, and Morgenstern[32], formula 15.10 as

TABLE

which, when applied to the data for the fenfluramines, reveals a 95% confidence
interval of 1.48 to 89.10.

109 Caveats to Odds Ratio Computation There are three caveats to these
computations. a. The above computation is hypothetical, and although the data on
which the computations are based have appeared in the peer reviewed literature,
the computations contained herein have not gone through the peer reviewed process.
b. The computations are crude, that is to say, they have not been adjusted for
concomitant variables which may themselves be associated with the occurrence of
cardiac valve regurgitation. c. The exposure to fenfluramine or dexfenfluramine in
the computations above cannot be considered as exposure to these agents to the
exclusion of other agents.

110 The Epidemiological Study of Weissman Weissman[33] reported the results of a
randomized, double-blind, plalecbo-controlled study of dexfenfluramine, modified
to include echocardiographic examinations of 1,072 (out of the 1,212 original
study participants) overweight patients within a median of one month after the
discontinuation of treatment. This clinical trial is an example of an experiment
that does not answer the crucial question of exposure-disease because of important
design flaws, demonstrating that the presence of randomization is not sufficient
to cover other methodological issues and difficulties.

111 Prospective Statements: It was pointed out in this affidavit that it is
important to state the research questions before the study is carried out to
ensure that study will be able to answer the research questions. This was not the
case in Weissman's study, and the absence of these principle applications
represents a methodological weakness this study cannot overcome. The particular
mechanism by which this weakness operates in this study is readily apparent
because Dr. Weissman never intended, at the inception of the study, to obtain
echocardiograms on any of his patients. The number of echocardiograms he happened
to obtain is far below the number required to draw a reasonable conclusion from
the study. Since there was no echocardiogram performed at the beginning of the
study to exclude these patients with baseline valvulopathy, patients were admitted
to the trial with valvulopathy that could not be attributed to fenfluramine, and
their inclusion confuses any conclusion, which one might draw. In addition, the
patients randomized into his study did not know they would need to have
echocardiography done, and drifted away or refused the request for the
echocardiography at the conclusion of the experiment. Additionally, the sample
size was too small to be able to detect a clinically meaningful relationship
between exposure and valvulopathy. Most importantly, and critical to the study's
interpretation, every patient in the study did not undergo a test for cardiac
valvulopathy at the trial's end. This is because the trial was not designed to
measure cardiac valvular disease, but was diverted to this purpose in midstream.
Since the study was not designed to evaluate the relation between dexfenfluramine
and valvulopathy and since it was underpowered, the results are neither reassuring
nor negative, merely inconclusive. Criticisms of Weissman's article on this very
point have been published.[34J Unfortunately, the Weissman manuscript demonstrates
that expensive, state of the art clinical trials, such as expensive state of the
art new cars, can be wrecked by sudden sharp and unexpected changes in direction
while under way.

112 Conclusions from Connolly, Khan, Jick, and Weissman: This review of the issue
of strength of association between fenfluramine and valvulopathy must weigh the
totality of the evidence from the available studies. Consideration of the
possibility of a causal relationship between anorexigen exposure and valvulopathy
most prudently contains an evaluation of this evidence measured against the
standard of the epidemiological causality tenets, which now follows.

113 Strength of Association: The large odds ratios from each of the Khan and Jick
studies support the presence of an association between fenfluramine and
dexfenfluramine use and cardiac valvulopathy in the population at risk. The
findings of the case reports of Connolly26 and the results of each of the case
control studies of Khan and Jick all, independently, identify a relationship
between fenfluramine exposure and valvulopathy. Even the Weissman29 study, despite
its methodological weaknesses, had an upper bound on its confidence interval
consistent with a tripling of the risk of valvulopathy. It is noteworthy that the
New England Journal of Medicine demanded that Weissman recalculate the odds
ratios, this time including all levels of exposures.35 This recalculation showed a
greater (in fact, statistically significant) odds ratio between exposure to
fenfluramines and valvulopathy. This recalculated odds ratio appears in the
Weissman manuscript.

114 Age Specific Prevalence: As pointed out by Connolly valvulopathy is rare in
patients younger than 50 years of age. Additionally, the type seen in the patients
who had to have their heart valves replaced is known to have only two previous
causes, ergot exposure36 and carcinoid tumor37. Furthermore, in each of the case
control studies of Khan and Jick, 27 careful work insured that exposure to
fenfluramine preceded the occurrence of valvulopathy. In the Jick study, the
strength of association was proportional to the duration of exposure to
anorexigen. The weaker exposure-disease association in Weissman's study is
consistent with a dose-esponse relation between anorexigens and valvulopathy. The
mean exposure was less than 3 months in the study by Weissman et a, as compared
with 4 to 20 months in the study by Khan et al. and 0 to 10 or more months in the
study by Jick et al. Since the exposure levels in the studies by Khan et al. and
Jick et al. reflect the actual patterns of use of these anorexigens, those two
studies are more representative of actual experience.

115 Serotonin: As expressed by Connolly26, fenfluramine is known to influence
serotonin levels. Previous models (ergot toxicity, carcinoid syndrome) have
demonstrated that elevated serotonin levels are associated with valvulopathy. The
findings in Connolly, Khan, Jick, and Weissman27 are concordant with those of
ergot toxicity and carcinoid syndrome. The analogy of ergot poisoning and
valvulopathy and the analogy of carcinoid poisoning and valvulopathy substantially
increase the tenability of the hypothesis that fenfluramine, through serotonin
level alterations, can induce valvulopathy. The totality of evidence from the case
series studies and the epidemiological studies, provide substantial documentation
for an association between fenfluramine and valvulopathy. Furthermore, a reasoned
review of the established and standard tenets of causality provides important
information demonstrating this relationship is causal.

116 The Epidemiological Study of Burger

The manuscript of Burger[38] describes observations on a cohort of individuals who
were followed for up to 30 months to observe the prevalence of valvular heart
disease. This research effort was not designed to evaluate the relationship
between use of fenfluramines and cardiac valve disease. Instead, these patients
were part of a population of 591 subjects who were participating in a four-year
prospective, nonrandomized open label (i.e. unblinded) study of combination
therapy with phentermine and fenfluramine to sustain weight loss. The findings
were that, using the FDA criteria, 1.3% of patients had mitral regurgitation, and
6.9% of patients had aortic regurgitation. Since these prevalence rates are
approximately those identified in the Framingham study, Burger concluded that the
prevalence of valvular regurgitation is comparable to those not taking these
anorectic agents.

117 The Epidemiological Study of Schiller

Building on this manuscript, Schiller[39, in an editorial appearing in the same
journal, propounded that the association between the use of fenfluramines and
cardiac valve disease has been overstated. Unfortunately, Burger has missed a key
point that stems from the fact that his study was inappropriately designed for
this work. Because his study was not designed to perform an echocardiographic
assessment of cardiac valves, no plans were made to collect this information on
every subject who entered the research effort. In this work, there were 591
patients who were accepted into the study. Of these 591 subjects, only 226
returned to have their hearts evaluated. Most of the cohort (365 patients) did not
return for evaluation of their hearts. Since no findings were available for the
missing 365 patients, it is likely that many of these missing patients would have
had valvulopathy, invalidating Burger's estimates of mitral regurgitation
prevalence and aortic regurgitation prevalence. Burger's work, unfortunately,
sheds no new light on the progressive nature of fenflurarne-induced valvular heart
disease. Also, Burger did not appropriately adjust for either the differences in
age or the differences in gender distribution between his cohort and his
comparison group, the Framingham data. This is a critical weakness because
valvulopathy prevalence increases with age, and is more prevalent in males than
females.[40] This absence of adjustment overestimates the prevalence of
valvulopathy in the Framningham group and distorts Burger's comparisons.
Unfortunately, the editorialist Schiller completely misses each of these two
points. On page 1161, Schiller states, "It would seem, then, that as studies have
become more scientifically rigorous, the role of fenphen in valve disease appears
to be approaching the vanishing point." This is an extremely unfortunate choice of
words to describe the fatally flawed study of Burger, and is an inadequate summary
of our understanding of the fenfluramine-valvulopathy relationship. Burger's
research effort is a fine example of what can go awry in science when the
scientist strays from the tenet of "first say what you will do, then do what you
said". If Burger had designed the study to look for echocardiographic evidence of
cardiac valve disease, he would have planned to follow everyone. Since he did not,
most of the cohort escaped the echocardiographic evaluation and the study cannot
be interpreted. Schiller's editorial must be discounted simply because Schiller
chose to base his reasoning on the weak foundation of a flawed research effort. A
follow-up article by Burger41 unfortunately suffers from several of the same
methodologic weaknesses of the first Burger manuscript, although Dr. Burger wisely
chose to refraim from a direct comparison of the event rate of valulopathy in his
incomplete cohort to that of Framingham.

118 The Epidemiological Study of Shively

The manuscript of Shively et. al. [42] describes a cross-sectional study in which
patients are simultaneously queried about their history of fenfluramine use and
undergo echocardiography. Three echocardiographers interpreted echocardiograms
independently. Shively found that F.D.A. grade regurgitation was greater in the
fenfluramine-exposed patients than in those unexposed with an odds ratio of 3.82
(p = 0,01). There was a greater degree of mild aortic regurgitation in the
fenfluramine-exposed group, and no difference in the prevalence of mitral valve
disease between the fenfluramine exposed and non-fenfluramine exposed patients.
Two comments are in order to guide the interpretation of this work: First, with no
baseline echocardiogram before exposure, patients were admitted to this study with
valvulopathy unassociated with fenfluramine exposure. Including these patients as
Shively did, underestimates the odds ratio. As with the findings of Weissman,
Shively's finding tends to minimize the strength of the association between
fenfiuramine and cardiac valve disease. Secondly, it is very unlikely, that, given
the prevalence of mitral valve disease is so low in the population, Shively would
have found an association between fenfluramine and mitral valve disease in the
small sample of 172 exposed and 172 controls. If anything, Shively's findings
strengthen our believe that fenfluramines cause heart valve disease.

119 The Epidemiological Study of Wadden

In this study[43], 20 patients out of a total of 26 patients had been randomly
assigned to receive medication combined with either group or individual lifestyle
modifications. Before treatment, none of these patients had valvulopathy on
examination by their primary care doctor (however, no baseline echocardiograms
were obtained). The findings that five of these patients had aortic regurgitation
of mild or greater degree is further suggestive of the relationship between
fenfluramine exposure and valvulopathy.

120 The Epidemiological Study of Ryan Ryan et. al. publishef a recent cross
sectional study examining the relationship between fenfluramine exposure and
valvulopathy.[44] These researchers obtained serial echocardiographic and clinical
evaluation of the valvular regurgitation before, during and after
fenfluramine/dexfenfluramine and mazindol/phentermine. These researchers noted
that there was significantly greater risk for developing valvulopathy for those
who took medication for longer than six months. Significantly, some patients who
had normal echocardiograms at baseline developed cardiac valvular regurgitation
after exposure to these anorexigens.

121 The Epidemiologic Study of Jollis-Ryan The study of Jollis-Ryan[45]
demonstrates the ability of a study to identify an effect that it was designed to
detect. This study examined 1864 patients; 672 of these patients were control
patients never exposed to fenfluramine, and 1163 patients were exposed to
fenfluramine. This study was designed to determine a duration-disease relationship
between fenfluramine use and valvulopathy. In fact, this study was designed to
detect an increased prevalence of valvular heart disease with greater duration of
fenfluramine exposure and in fact, the study did detect this effect. However, the
sutdy was not designed to determine if a short duration of exposure to
fenfluramine is associated with valvulopathy. The study did not detect such a
relationship. The interpretation of this last evaluation however, like every
sample-based study result, must be viewed through the prism of the study's
methodology. This study had only 25 patients exposed in the short duration group.
The power of an analysis to detect a two fold increase in the prevalence of
valvulopathy associated with fenflurmine based on 25 patients in the exposed group
and 672 patients in the control group is 15%. With this low power, the Jollis
study tells us nothing of value about the relationship between short dose
fenfluramine use and valvular heart disease. Specifically, even though the Jollis
study demonstrates increased danger of valvular heart disease associated with an
increase in duration of fenfluramine, it does not demonstrate that the fenflurmine
is not dangerous at the low duration. For each of the other durations, the
underpowered moniker applies.

122 The Epidemiological Study of Kancherla Using a cross-sectional research
design, Kancherla et al[46] evaluated 200 patients who had been exposed to diet
drugs and requested an evaluation of their heart with specific attention to their
heart valves. These patients were not selected randomly, nor was the drug therapy
randomly allocated. The mean duration of exposure to the diet drugs was less than
one year. Using the F.D.A. criteria of valvulopathy, 5% of the cohort had
developed mitral regurgitation, and 12% of the population had developed aortic
regurgitation. The prevalence was also elevated for those patients exposed to
fenfluramine-phentermine combinations (5% for mitral regurgitation and 9% for
aortic regurgitation). In the dexfenfluramine group, 9% of patients developed
aortic regurgitation (no patient had mitral regurgitation in the small sample of
patients taking dexfenfluramine). Because of the nature of cross sectional studies
(lack of random allocation of therapy, and the lack of assurance that the exposure
to fenfluramine predates the development of cardiac valvulopathy), the Kancherla
study, by itself, does not demonstrate causality. However, when considered with
the wealth of information available from other studies, this study does add to the
body of evidence pointing to the fenfluramine-valvulopathy association.

123 A resolution of the issue of causation between the fenfluramines and heart
valve disease does not depend on one study and one odds ratio. The weight of the
evidence in favor of a causal relationship between the fenfluramines and
valvulopathy is clear when one considers the Bradford Hill criteria.

124 In 1999 following the initial simultaneous release of studies by Khan, Jick,
and Weissman, there has been additional literature addressing the association
between the use of fenfluramines and cardiac valve disease. These studies meet the
criteria of peer review publication, and require careful review.

                         Progression of heart valve disease

125 Progression vs. Latency Before continuing, we must differentiate the process
of disease progression from disease latency. Each begins with an exposure. In the
case of progression, detectable tissue damage is sustained, and this damage
increases over time. Examples of disease progression include the KW lesion of
diabetes mellitus where the kidneys progressively deteriorate in the diabetic
environment; hypertensive heart disease, in which case the heart increases in size
over years, commonly resulting in heart failure; and nephrosclerosis or kidney
failure after years of elevated blood pressure. Nutritional diseases e.g. scurvy,
beriberi, and pellagra produce progressive deterioration in multiple organ system.
Latency may be defined as the development of an exposure linked undetected
condition long after the exposure has ceased.

126 Latency There are many examples of latency in pathophysiology. Among the most
infamous are the late cardiovascular effects of syphilis. Syphilis starts as a
local genital infection. Left untreated, these lesions resolve, and the patient
commonly believes that, in the absence of any obvious manifestations of the
illness, that they are "cured". However, the long-term consequences are profound.
Unbeknownst to the patient, lesions (termed "gummas") form in several organ
systems of the body. These lesions are not nests of infected organisms but
aberrant, distorted system responses to the infection that occurred years ago. The
principle latent cardiovascular component is luetic aortitis, a condition that
frequently leads to aortic aneurism and death. This latent lesion is first
recognized ten to thirty years after the initial infection, a time span
representing a long latency period. Patients who are exposed to syphilis in their
twenties are usually over fifty years old when the signs of aortitis first
develop. It should be noted that, while heart values are not commonly affected by
syphilis, syphilis has been known to produce separation of the aortic valve
commissures and aortic regurgitation. This aortic regurgitation is a latent
consequence of syphilitic infection in the remote past Tabes dorsalis is yet
another latent clinical manifestation of early syphilis, producing spinal cord
lesions thirty years after the initial infection.

127 There are other, more prominent causes of latent mitral valve disease. For
example, rheumatic heart disease is a latent consequence of a bacterial infection
that commonly occurs in childhood and adolescence. The infection resolves quickly;
however, years later, latent heart valve disease, including mitral regurgitation,
mitral stenosis, aortic regurgitation, and aortic stenosis appear. These lesions
can lead to heart failure, the requirement of valve replacement, and death.

128 Finally, systemic lupus erethematosis (SLE) can produce severe illness at its
onset but, for reasons that remain unclear goes into extended remission. However
over many years, sterile vegetations (termed verruca) develop on the heart valves.
These verruca can cause valve disease (termed Libman-Sachs endocardopathy) and
resultant embolic phenomena years after the first diagnosis of SLE and its
remission.

                  Valvulopathy Progression: Basis in the Literature

129 A fine summary of the understanding of mitral valvular heart disease is
provided by Enriquez-Sarano.[47] The author states that there is a trend for
progression of valvular lesions, which leads to worsening heart function and
clinical complications. In fact, it is the recognition of this natural consequence
of acquired (as opposed to congenital) valvular heart disease that is the
rationale for the requirement of regular monitoring of patients with this disease,
In his study, Dr. Enriquez-Sarano demonstrated that left untreated, patients with
mitral regurgitation often progress. It is significant to note that Dr.
Enriquez-Sarano observed that some patients underwent regression of their mitral
valve regurgitation. However, when observed, this was seen to be a function of not
a return of the valve structure to normal, but with a resultant change in blood
pressure. There was no observed regression of the disease process itself. Although
there is important variability in the disease course of patients with mitral
regurgitation, progression is a natural consequence of the lesion. This is a
critical point. Regression refers to the regression of symptoms - it does not
refer to the reversion of the appearance of the heart valve from its diseased to
its pre disease, normal state.

130 The weight of the evidence is that acquired valvular heart disease is
progressive. With this progression of acquired mitral and aortic valvular heart
disease as a backdrop, it becomes important to assess the information concerning
the progression of fenfluramine-induced changes in valvular heart disease. There
must be firm, scientifically rigorous information to demonstrate that fenfluramine
induced valvular heart disease is not progressive, distinguishing it from all
other acquired valvular heart disease. Without this firm proof, we must conclude,
that fenfluramine induced valvular heart disease, like other acquired valvular
heart disease, is progressive, will cause symptoms, and lead to disability,
hospitalization and death.

131 The execution of such a progression study would be complicated, requiring
careful attention to detail. Before exposure, patients would receive an
echocardiogram to remove those patients who had a preexisting valvulopathy.
Patients without valvulopathy who were randomly assigned to either the
fenfluramines or placebo would be would be evaluated after exposure to determine
the incidence rate of fenfluramine valvulopathy. Those patients unfortunate enough
to develop valvulopathy would then be followed for an additional period of time
during which serial echocardiograms would be obtained to measure the extend of
progression of the valvular lesions. The size of the studies would be immense
because of the relatively low annual incidence of aortic and mitral valve
regurgitation (using the F.D.A. criteria). In addition, patients who develop
valvulopathy would need to be followed for years perhaps as many as five to ten
years) to document the occurrence of any progression, and patient follow-up losses
must be minimized to the extend that patient losses would not effect the overall
conclusion of the study. No such study has been carried out. Even in a study such
as Jick, which measured over 9000 patients, the definition of valvulopathy
provided a much lower prevalence rate, requiring a much greater sample size.
Neither the Khan nor Jick studies were designed to follow patients longitudinally,
which would be a requirement in a progression study.[FNj]

      FNj. Random assignment would be required to insure that any differences in
      valvulopathy rates at the end of the study could be attributed to
      fenfluramine exposure.

132 Acquired vaivular heart disease is not static but progressive. It has been
shown that patients with mild or moderate aortic regurgitation slowly progress to
the symptomatic phase48. As pointed our earlier in this section, Reimold[49] notes
that the severity of aortic regurgitation is dependent on the size of the
regurgitant changed surface of the valve. However the changes in the lesions over
time, and the changes in the function of the heart, depend on clinical factors,
which depend of personal characteristics[FNk] of the patients and therapy
decisions.

      FNk. These would include illnesses e.g. diabetes meilitus, hypertension,
      congestive heart failure, and habitis such as cigarette smoking.

133 Absence of Lesion Healing:The natural history of valve lesions does not
include lesion healing. Patients can have their regurgitant symptoms get better or
worse, but this is unrelated to the lesion size. Changes in regression are often
due to coexisting disease. Patients with mild aortic valve disease, in the absence
of co morbidity, may be free of symptoms for several years. However, other
patients, with coexistent diabetes mellitus, or hypertension, or a history of
cigarette smoking may have their disease progress more rapidly. The presence of
comorbidity is a factor in determining disease progression.

134 Progression and Therapy: Another component of the determinant of progression
is therapy. Patients who have the weakening or destabilization of their heart
function treated aggressively may experience an alleviation of the symptoms of
their regurgitant lesion. The requirement of therapy may lead some to call this
change progression (i.e. heart disease which has progressed to the point of
therapy). Others, who see that the patient's symptoms have improved deem this
circumstance as evidence of disease regression. The use of the term progression or
regression do not indicate the state of the heart lesion but instead refer to the
treatment and clinical symptoms and signs of the disease. A similar circumstance
arises with surgery. Valvular disease can progress to the point where the valve
needs to be replaced surgically. The determination of when to replace the valve is
a clinical decision, but the clinical decision can induce an artificial answer
about disease progression. Traditionally surgeons have waited until the patient's
clinical disease has progressed to the point of destabilization of the heart. This
would make it appear that it has taken the disease a long time to progress.
However, other surgeons, recognizing that the postponement of surgery until the
patient has severe symptoms may decrease the likelihood that the patient will
survive surgery. These surgeons will operate on their patients earlier in the
disease course. This may be seen as early progression, but in fact is just an
artifact of the surgeons' decision process. Thus, the progression and regression
definitions are confounded with patient characteristics, and clinical decisions.
However, one fact penetrates this mist of definitions - the heart valve lesion,
once present, does not disappear. The heart valve does not heal.

135 Interpreting studies with control groups In clinical research, there are three
statistical errors that can lead to difficulty in interpreting research. Not
unlike the fruit from a poisonous tree, the venom that a severe protocol violation
introduces into a study corrupts the rest of the research effort, making it
impossible to understand the implications of the study. Two other errors deal with
the notion of sampling error. Each one (type I error and the type II error)[FNl]
need to be minimized. The population produces many samples, some of which will
mirror efficacy and some of which will not. In this circumstance, the issue is one
of minimizing the probability that the population will produce a sample that
appears to suggest efficacy. However, for the type II error, we shift the
paradigm, assuming not as before that there was no efficacy in the population, but
instead, assuming that there is efficacy of the treatment in the population. If
this treatment effect is present in the population, we wish the sample to reflect
it. The probability that the sample does not reflect this positive population
finding is termed the probability of type II error, or the beta error. It has
become common for investigators to refer to the beta error subtracted from one.
This new quantity is called the power of the experiment. The power of an
experiment is the probability that the population in which the treatment is
effective produces a sample that mirrors this effect, i.e. that a population with
embedded efficacy produces a sample with embedded efficacy. Unlike alpha, power is
to be maximized. The greater the power, the more confidence we have that the
population In which there is an exposure-disease relationship has not misled us.

      FNl. Type I and type II errors are statistical errors, which measure the
      extent to which the population has misled the investigator by producing an
      unrepresentative sample. If the researcher's sample has produced a
      relationship between fenfluramine exposure and valvular heart disease, the
      type I error measures how likely is it that the population contains no such
      relationship, but produced a sample that did. The type II error measures the
      reverse error, i.e. the likelihood that a population in which there is an
      association between the exposure and disease produces a sample in which
      there is no relationship.

                              Sample Size Calculations

136 The construction of the minimum sample size allows researchers to draw
conclusions about their results assuming that they have adequate protection from
these errors. Since the researcher does not know whether their study will be
positive or negative, they must have adequate insurance that the result of the
study (be it positive or negative) will be generalizable to the population at
large, i.e. they must have an adequate sample size. The third quantity that is
factored into the sample size computation is the number of cases (patients with
the disease) that are anticipated to occur. It is perhaps a truism to say that
there must be enough cases in each of the exposed and unexposed groups in order to
say with reliability that these clinical events occur at different rates within
these two groups. Finally, the difference between the ent rates plays an important
role in the sample size computation. These factors are bound up in the sample size
formula:

TABLE

N = number of placebo patients number of active group patients a Type[FNm] I error
<<beta>> = Type II error Zc = the cth percentile from the standard normal
probability distribution Pe = prevalence of valvular heart disease in the exposed
group of patients Pu = prevalence[FNn] of valvular heart disease in the unexposed
group of patients

      FNm. The greater the difference you are trying to detect, the smaller the
      sample needed to detect it.

      FNn. Type I error is 0.05 in a two tailed testing setting, type II error is
      0.20, and the prevalence ratio is 2.0.

137 By making some reasonable statistical assumptions the relationship between the
sample size and the prevalence of valvulopathy in the unexposed group can be
computed and displayed (Figure 2).

TABLE

138 The salient message of figure 2 is that the lower the prevalence of the
disease in the population, the greater the sample size needed to detect a
difference in the prevalence between groups. One way to understand this
relationship is to recognize that the difference between the exposed and unexposed
groups is hypothesized to be the number of clinical events. A required number of
clinical events must be garnered. The greater the prevalence of the clinical
endpoint, the more frequently the clinical endpoint occurs, the fewer the number
of patients required to obtain that number of clinical endpoints. If the endpoint
is rare, (i.e. low prevalence), more patients must be observed in order to obtain
enough patients with the clinical endpoint in order to draw a conclusion about the
relative occurrence of these clinical endpoints between the two groups.

139 The prevalence-sample size relationship is central to consideration of the
results of studies purporting to examine the relationship between fenfluramine use
and each of aortic and mitral valvulopathy. Since (FDA-defined) aortic
valvulopathy occurs with a greater frequency than (FDA-defined) mitral
valvulopathy, the sample size needed to detect a difference between exposed and
unexposed groups with aortic valvulopathy is smaller than that required for mitral
valvulopathy. Thus, strictly from several fundamental issues of research design
and statistics, we would anticipate that studies designed to identify the effect
of the fenfluramines' use on the prevalence of aortic valvulopathy would not find
that same effect size in patients with the rarer FDA mitral valvulopathy because
the studies would have inadequate sample size[FNo].

      FNo. Another way to state this is that these studies with inadequate sample
      size to detect mitral regurgitation are inadequately powered to detect an
      effect of the fenfiuramines' use on mitral valvulopathy.

140 Thus from the fundamental principles of good research design, the questions
which must be addressed for each study to be evaluated are 1) have there been
severe protocol violations which corrupt the measures of statistical error and 2)
what is the probability of a type I errorp if the study finds a relationship
between fenfluramine use and valvular heart disease or 3) what is the power if the
study finds no relationship between fenfluramines and valvular heart disease.
Studies with either protocol corruption, or no protocol corruption but large [FNp]
values for "positive" studies or low power for "negative" studies cannot be
integrated into our fund of knowledge (i.e. we can draw a useful conclusion about
the findings in the population from the findings in the sample) and must have
their specific results discarded.

      FNp. This is the probability that the population that holds no relationship
      between the exposure and the disease produces which contains a spurious
      relationship.

141 Study Power The power of a study is computed as:

TABLE

142 So, for a given exposed group prevalence pe, unexposed group prevalence pu,
and number of subjects involved in the research N, the power of the analysis can
be computed. Power is only useful when the researchers claim there is no
relationship between exposure and the disease in the sample because, in this
circumstance we need to know the likelihood that a population with the
relationship would have produced a sample without the relationship. If this would
be a likely occurrence (i.e. the power is low), we cannot believe the
investigators when they claim to extend the negative findings from the sample to
the populations. The best we can say is that the study is "uninformative".

143 We may now apply these principles to the computation of the prevalence of
mitral valvulopathy in patient's exposed to the fenfluramines.

144 The Epidemiological Study of Gardin[50] Dr. Gardin et, al. describes a study
to evaluate the cardiovascular status and the prevalence of valvular abnormalities
as assessed by clinical cardiovascular parameters and echocardiography in patients
treated for obesity with dexfenfluramine (Redux) and phentermine/fenfluramine. The
study had primary echocardiographic endpoints including mild or greater aortic
regurgitation and moderate or greater mitral regurgitation based on FDA criteria.
The study evaluated 1473 participants, and identified that prevalence rates of
aortic regurgitation were significantly increased in anorexigen-treated patients
with 8.9% in the dexfenfluramine group and 13.7% in the phentermine/fenfluramine
group, and 4.1% in the untreated group. The authors identified 4.9% of the
dexfenfluramine treated group as having mitral regurtation by the FDA criteria,
5.1% in the phentermine/fenfluramine treated patients, and 3.2% in the untreated
group. The prevalence ratio for mitral valvulopathy is 1.53 for the
dexfenfluramine treated group, and 1.59 for the phenterine/fenfluramine treated
group. The authors concluded that there is an association between fenflurine use
and aortic regurgitation but not between fenfiuramine use and

145 However, before we can accept that these conclusions apply to the population
at large, we must ask first, have there been protocol violations severe enough to
corrupt the study. The answer here is no. Assuming that the report is complete,
with all inclusion/exclusion criteria reported[FNq], then the rates that are
reported have no inherent inaccuracies due protocol violations. The authors claim
that there is an increased risk of aortic regurgitation with exposure to the
fenfluramines (p &lt;0.001)q , and the low type I error tells us that we can with
confidence extend this finding to the community at large.

      FNq. Statistical error increases with the number of hypothesis tests carried
      out. Correction for the two tests carried out here does not alter the
      authors' conclusions.

146 Turning to the mitral valve, the researchers claim that the prevalence of FDA
mitral valvulopathy is no greater for patients exposed to fenfluramines then those
who were not exposed. However, before we can extend this "negative" result to the
population, we must ask "How likely is it that the fenfluramine-mitral
valvulopathy relationship is real in the population, but the population fooled the
investigator by producing a sample which did not contain this same relationship?"
This is the type II error, and is 1- power. We can easily compute the power for
this evaluation.

TABLE

147 This result states that it is very likely (67%) that a population in which
there is a relationship between fenfluramine and FDA criteria mitral valvulopathy
would produce a sample with results as demonstrated by Gardin[FNr] . Therefore we
cannot believe that the results in the sample regarding mitral valve disease are
generalizable to the population. The sample size needed to detect mitral
valvulopathy in the population and produce a meaningful comparison of the
prevalence of mitral valvulopathy in the population as compared with the sample is
2331 This was a study funded by AHP.

      FNr. The sample size to detect an odds ratio of 2.0 with 80% power and a FDA
      criteria mitral valve prevalence rate in the population is 1554 divided
      between each of the exposed and unexposed group. However, in Gardin, there
      were three groups (dexfenflurarnine, phentermine/fenfluramine and
      untreated). Thus the comparable sample size is (1554/2)*3=2331.

148 The general notion that the absence of evidence of an effect denote evidence
of the absence of effect is a fallacy (termed for convenience here the "absent
evidence" fallacy) in sample based research. The research must be specifically
designed to identify the effect, before the lack of the effect establishes that
the effect is not present. These design considerations includes many facets, one
of them being that their must be enough subjects in the population. We therefore
cannot believe that the relationship is not present merely because the studies did
not find a relationship they were not designed to detect. I have completed a
preliminary computation of the power of a pilot study as a function of efficacy
detection and of a (the type I error). The power is computed as:

TABLE

where n = of unexposed patients a = Type I error Zc = the cth percentile from the
standard normal probability distribution p1 event rate in the unexposed group p2 =
event rate in the exposed group = p1- p2 k = ratio of exposed to unexposed sample
sizes

Preliminary Power of Hypotheses tested in the Gardin Study (Two- sided <<alpha>> =
0.05 to detect a relative risk of 2.0)

Hypothesis Test            Power

Aortic Regurgitation            98%

Mitral Regurgitation            60%

3-6 month dexfen exposure compare to &lt; 3 months             4%

6-2 months dexfen exposure compare to &lt;3 months             4%

12-18 months dexfen exposure compare to &lt; 3 months             4%

3-6 months fenphen exposure compare to &lt;3 months             15%

6-12 months fenphen exposure compare to&lt;3 months             17%

12-18 months fenphen exposure compare to &lt;3 months             15%

&gt; 18 months fenphen exposure compare to &lt; 3 months             15%

149 There were no cases among the group compared to dexfenfluramine for less than
three months. In order to be able to compute the power in this circumstance, I
used an event rate of 0.001, which allowed the power to be approximated for this
group.

150 Clearly, the only hypothesis test for which there was adequate power was the
test of aortic regurgitaion. All other examinations are predictable underpowered.
It is difficult to see how the authors could claim that an examination of mitral
regurgitation was a "primary endpoint" given the lower power of this examination.
A prospective statement that mitral regurgitation was a primary endpoint must be
buttressed by the assurrance of adequate power for the primary endpoint. In fact
the statement about primary endpoints which is the introduction of the statistical
methodology section of this paper could be reasonable interpreted as an assertion
that there is only one primary endpoint for the study, which is the combined
presence of aortic or mitral valvulopathy. If that were the case, the overall
result would apply to each of aortic and mitral evaluations, e.g. in Khan.

151 It should be pointed out that Gardin carried out a second study51 that
evaluated the whether the initial echocardiographic changes in the original cohort
of patients that was assayed in the first study changed over time. Gardin found no
change in the occurence of cardiac valulpathy over the one year followup period in
his study. However, the number of patients who actually had changes were small.
Unfortunately, since only 1142 of 1466 patients returned from the study, we are
forced to wonder what the results would have been if 324 patients who were
supposed to undergo followup echocardiographic analysis in this study had actually
been included in this research effort. The absence of this required data makes
this work uninterpretable. Just as a contractor asked to build a home is sharply
criticized for putting a leaky roof in place, regardless of how solid the home's
foundation is that he laid, Gardin's partial effort must be sharply criticized,
because he did not accomplish what he set out to do.

                        The Epidemiological Study of Hensrud

152 In 1999, Hensrud published a manuscript[52] that reviewed the experience of
patients randomized into a randomized clinical trial. Although this study was
conducted in the upper Midwest, it is important to note that the patients
considered in this manuscript are not the patients of Heidi Connolly's 1997 New
England Journal of Medicine. The patients on whom the Hensrud report is based had
their clinical trial discontinued when fenfluramine was withdrawn from the market.
Only thirty one patients were randomized in this study, and of this small number,
only eighteen patients received two echocardiograms, the minimum number of
measurements on which to assess whether a change took place in the cardiac valve
over time. These lethal weaknesses commonly result from redirecting a clinical
trial designed for one purpose to another. In addition, all patients who were
identified to have regurgitation improved over time, not just those patients
exposed to the fenfluramines. Unfortunately, follow-up losses cripple this study's
ability to say anything constructive about the fenfluramine - valvulopathy
relationship.

153 The Epidemiological Study of "Weissman Il-III" Neal Weissman provided a view
of the additional experience of patients in his original study.

154 In this follow-up study[53] , he states that the small increase in prevalence
of minor degrees of aortic regurgitation and mitral regurgitation in patients
treated with dexfenfluramine previously reported is no longer present three to
five months after discontinuation of use of medications. There are two
difficulties with this conclusion, both of which can be traced to the fact that
Dr. Weissman did not design this study to look at the relationship between
fenfluramine and cardiac valvulopathy. The first is that Dr. Weissman, much like
Dr. Burger, did not obtain echocardiograms on every patient. His initial study
randomized 1250 subjects. In that initial study, 1073 patients had
echocardiograms, 1250-1073 = 177 patients did not have echocardiograms for unknown
reasons. The absence of these echocardiograms makes his determination of no
importance difference in prevalence of heart valve disease difficult to defend
because the missing observations on these 177 patients could have a profound
impact on his findings. This problem is further compounded in the follow-up study,
in which an addition 132 patients did not return for echocardiographic evaluation
after the three to five month follow period. If a small number of these patients
randomized to receive the fenfluramines developed significant mitral
regurgitation, for example, Weissman's findings of no difference in the mitral
valvulopathy prevalence rates would be instantly overturned. Because he did not
follow all of his patients, and he did not obtain the endpoint measure on all of
his patients, Weissman's follow-up study makes no contribution to the fund of
knowledge on the long-term prevalence of fenfluramine caused valvulopathy. In
addition, the absence of a baseline echocardiogram means that patients with
preexisting valvulopathy were admitted to the study, a procedural weakness that
underestimates the relative prevalence of heart valve disease. This continued
weakness in Weissman's research lead to continued underestimation of the relative
occurrence of fenfluramine caused cardiac valvulopathy. This was a study funded by
AHP. These chronic weaknesses in the Weissman methodology were transmitted to and
through a third manuscript [54] which suggested that prolonged followup with the
same cohort demonstrated little evidence of progression. Unfortuanately, the
continued attrition in the patients followed over time makes it very unlikely
that, if there was a two fold increase in the use of cardiac valulopathy with the
fenfluramines that this third Weissman manuscript would have identifed it.
Therefore this manucript's thesis that there is no relationship between the
fenfluramines and cardiac valvulopathy must be rejected.

                        The Epidemiological Study of Shively

155 The manuscript of Shively et. al.[55] describes a cross sectional study in
which patients are simultaneously queried about their history of fenfluramine use
and undergo the procedure called echocardiography. In this study, three
echocardiographers interpreted echocardiograms independently. Shively found that
F.D.A. grade regurgitation was greater in the fenfluramine patients than in those
unexposed with an odds ratio of 3.82 (p = 0.01). There was a greater degree of
mild aortic regurgitation in the fenfluramine exposed group and no difference in
the prevalence of mitral valve disease between the fenfluramine exposed and
non-fenfluramine exposed patients. Two comments are in order to guide the
interpretation of this work. First, with no baseline echocardiogram before
exposure, patients were admitted to this study with valvulopathy unassociated with
fenfluramine exposure. Including these patients as Shively did, underestimates the
odds ratio. As with the findings of Weissman, Shively's finding tends to minimize
the strength of the association between fenfluramine and cardiac valve disease.
Secondly, it is very unlikely, that, given the prevalence of mitral valve disease
is so low in the population, Shively would have found an association between
fenfluramine and mitral valve disease in the small sample of 172 exposed and 172
controls. If anything, Shively's findings strengthen our believe that
fenfluramines cause heart valve disease. This was a study funded by Interneuron.

156 The Epidemiologic Study of Davidoff Davidoff et aL[56] carried out an
evaluation of the effect of fenfluarmine use in patients who had been recruited
into another study to assess the effect of weight loss on women who smoked
cigarettes. In this second study, 619 of 720 patients underwent echocardiographic
examination in order to determine if the use of the fenfluramines were associated
with valvular heart disease. No association was found by Davidoff, even though
there were slighly more patients in the fenfluramine group that had FDA criteria
aortic regurgitation and FDA criteria mitral valve regurgitation. The short
duration of fenfluramine exposure (&lt; 3 months) adds to the difficulty of
drawing a useful conclusion from this study. However, another consideration of
greater importance is the incomplete follow-up of patients in this trial. It is
impossible to draw any reliable conclusion from this study since the near
equivalence of valvulopathy prevalence in the fenfluramine and control groups can
quite easily and quite radically be altered by the findings in the 101 patients
whose results should have been included but were unavailable. Because Davidoff's
group did not obtain echocardiographic results on all patients who were in the
fenfluramine and the control group, the result of the study is uninterpretable.
Like the swimmer who drowns because he only traversed 75% of the lake's width
before insurmountable fatique took over, the partial finding of Davidoff are
obtained with great effort, but unfortunately teaches us little about the effect
of fenfluramine because they are incomplete.

157 The Epidemiologic Study of Mast Mast[57] et al. studied 50 patients who they
believed had been exposed to the fenfluramines and who had two echocardiograms. In
this retrospective cohort study, they found very few patients had a valular heart
lesion progress, and a greater number of patients experiencing regression. Ht
obvious difficulty with drawing any conclusion from this study is that the
selection criteria (i.e., the requirement of two echocardiograms available in the
past) eliminates from consideration patients who may have been exposed to the
fenfluramines, had a first echocardiogram and then subsequently deteriorated and
were unable to have a second echocardiogram obtained. This criticism, in
combination with the exceedingly small number of patients who participated in this
study precludes the integration of its findings into our state of knowledge about
the relationship between fenfluramine exposure cardiac valvulopathy.

        Literature Review of Progression of Acquired Cardiac Valvular Disease

158 Aortic Valve Lesions Acquired valvular heart disease is not static but
progressive. It has been shown that patients with mild or moderate aortic
regurgitation slowly progress to the symptomatic phase. Reimold58 notes that the
severity of aortic regurgitation is dependent on the size of the regurgitant
changed surface of the valve. Dr. Reimold examined 59 patients with chronic aortic
regurgitation and followed these patients for a median of 38 months, assessing the
serial changes in aortic regurgitant severity by the use of echocardiography. In
this study of patients with clinically significant aortic regurgitation, there was
progression of the disease process over time. Once signs of left ventricular
failure become apparent, the disease deteriorates, with a life expectancy of only
seven years after onset of significant symptoms. In some patients, surgical
intervention is warranted even in the absence of significant symptoms.[59]
Padial[60], studying 127 patients with aortic regurgitation of all ranges of
severity, concluded that aortic regurgitation is a progressive disease, not only
in patients with severe aortic regurgitation, but also in patients with mild and
moderate regurgitation.

159 Mitral Valve Lesions A fine summary of the understanding of mitral valvular
heart disease is provided by Enriquez-Sarano. The author states that there is a
trend for progression of valvular lesions, which leads to worsening heart function
and clinical complications. In fact, it is the recognition of this natural
consequence of acquired (as opposed to congenital) valvular heart disease that is
the rationale for the requirement of regular monitoring of patients with this
disease. In his study, Dr. Enriquez-Sarano demonstrated that, left untreated,
patients with mitral regurgitation often progress. It is significant to note that
Dr. Enriquez-Sarano observed that some patients underwent regression of their
mitral valve regurgitation. However, when observed, this was seen to be a function
of not a return of the valve structure to normal, but of a resultant change in
blood pressure. There was no observed regression of the disease process itself.
Although there is important variability in the disease course of patients with
mitral regurgitation, progression is a natural consequence of the lesion.

160 The weight of the evidence is that acquired valvular heart disease is most
likely progressive. With this progression of acquired mitral and aortic valvular
heart disease as a backdrop, it becomes important to assess the information
concerning the progression of fenfluramine-induced changes in valvular heart
disease. There must be firm, scientifically rigorous information to demonstrate
that fenfluramine induced valvular heart disease is not progressive,
distinguishing it from all other acquired valvular heart disease, Without this
firm proof, we must conclude, that fenfluramine induced valvular heart disease,
like other acquired valvular heart disease, is progressive, will cause symptoms,
and lead to disability, hospitalization and death.

161 Complicated Design of the Desired Study The execution of such a progression
study would be complicated and would be planned with careful attention to detail.
Patients without valvulopathy who were randomly assigned to either the
fenfluramines or placebo, would be evaluated after exposure to determine the
incidence rate of fenfluramine valvulopathy. Those patients unfortunate enough to
develop valvulopathy would then be followed for an additional period of time
during which serial echocardiograms would be obtained to measure the extent of
progression of the valvular lesions. The size of the studies would be immense
because of the relatively low annual incidence of aortic and mitral valve
regurgitation (using the F.D.A. criteria). In addition, patients who develop
valvulopathy would need to be followed for years (perhaps as many as five to 10
years) to document the occurrence of any progression. Follow-up losses must be
minimized to the extent that patient losses would not effect the overall
conclusion of the study. No such study has been carried out. Even in a study such
as Jick27, which measured more than 9,000 patients, the definition of valvulopathy
provided a much lower prevalence rate, requiring a much greater sample size.
Neither the Khan28 or Jick studies were designed to follow patients
longitudinally, which would be a requirement in a progression study.

162 Conclusions on progression of heart valve disease There is no reliable
evidence that fenfluramine caused valvulopathy is benign and harmless. In the
absence of this information, cardiac valvulopathy induced by the fenfluramines
must be treated as any other acquired progressive and serious valvular heart
disease.

                 Fenfluramines - Valvulopathy causation: Conclusions

163 There is overwhelming scientific evidence to support my opinion that
fenfluramine and dexfenfluramine are causally related to valvulopathy: These
studies were properly designed and executed by qualified scientists. These studies
follow accepted methodology and rely upon objective interpretation of scientific
and medical data. These theories have been subjected to peer-review on countless
occasions including but not limited to the attached articles in Exhibit A. Random
and systematic errors have been appropriately controlled for by generally accepted
epidemiological methods. These theories are generally accepted as valid in the
scientific community and are based upon reliable studies. These theories have been
advanced in scientific journals and seminars and not just used for litigation
purposes.

                        Food and Drug Administration (F.D.A.)

164 FDA Expertise: I have attended more than 20 meetings of the Cardio-Renal
Advisory Committee to the Food and Drug Administration (F.D.A.), and I have served
for four years on the Cardio-Renal Advisory Committee, served as an ad hoc member
of other FDA Advisory Committees, and currently serve on the FDA Pharmacy Sciences
Advisory Committee. These positions have provided me the opportunity to observe
and take part in the interplay that characterizes the relationship between the
F.D.A. and the drug sponsors.

165 As an actively serving member of this advisory committee, I have published two
articles on the deliberations of the Cardio-Renal Advisory Committee to the F.D.A.
which have appeared in the peer reviewed, public literature Panel61,62.

166 My service on the F.D.A. advisory committees, as well as my service to drug
manufacturers, has given me specialized knowledge and experience concerning the
F.D.A. policies, procedures and regulations as well as the corresponding duties of
a reasonable and prudent drug manufacturer.

167 In addition, I have argued before the F.D.A. on behalf of sponsors on numerous
occasions. In my service on behalf of sponsors to the F.D.A., and my service to
the F.D.A. at advisory committee meetings, I have developed expertise, knowledge,
training and experience in F.D.A. matters.

168 Role of F.D.A. The Federal Food and Drug Administration (F. D. A.) is an
organization charged by the public through Congress to insure the effectiveness
and safety of foods and compounds produced for public consumption. Over the course
of this century, the powers of the F. D. A. have expanded commensurate with their
public responsibilities in a growing and increasingly complex food and drug
environment.

169 The responsibilities of the F. D. A. have often placed it in perceived
opposition to the pharmaceutical industries. The F.D.A. regrets this comparison
and has devoted substantial effort to overcome it. The F.D.A. has encouraged the
view that the relationship between it and the pharmaceutical industry s a
collaborative one, and has suggested that the efforts of both the F.D.A. and the
pharmaceutical industry be viewed as one of joint partnership.

170 Modernization Act: This emphasis can been seen in the F.D.A. Modernization
Act, which provided a set of streamlined procedures for the sponsor and F.D.A. to
follow, helping to insure a prompt and efficient review of a new drug application
(NDA). These procedures have substantially reduced the time it took the F.D.A. to
complete the review of a sponsor's application for a new drug approval. This
pivotal move to streamline the administrative review process was a direct response
to complaints from both the pharmaceutical industry, and, to some extent, from
segments of the public that the F.D.A. was not responsive enough to the need for
the swift introduction of safe and effective medications. The F.D.A. correctly
recognized that the timely delivery of these new medications was essential for
maintenance of the public health.

171 Collaboration: In order for the F.D.A.-pharmaceutical industry collaborative
process to work, the F.D.A. and the Sponsor must work together to demonstrate the
efficacy and safety of the compounds. Contrary to public perception, the F.D.A is
not a central laboratory. It generally does not carry out independent testing of
the products submitted to it by the pharmaceutical industry for approval. The
F.D.A. does not independently execute clinical trials to study in an independent
manner the safety and effectiveness of new compounds. The F.D.A. instead receives
the crucial information about the effects of a drug from outside sources. The
F.D.A. has developed skill in identifying relevant information for the effects of
a drug. Several of these sources include the general medical literature, and
research reports from academic investigators, and research study results which are
sponsored by other institutions e.g. the National Institutes of Health, (NIH).
However, the primary source of information concerning the safety and efficacy of
the compound is the sponsor itself. It is the sponsor's task to provide the F.D.A.
with the majority of the information concerning the effects of its compound in
animals (when appropriate) and in humans. The F.D.A. reviews, scrutinizes and
reanalyzes the information it receives, but it relies on its partner, the sponsor,
to provide the data on which its decision ultimately rests.

172 It is therefore essential that the sponsor and the F.D.A. have a relationship,
which at its foundation is truthful, honest, open, characterized by a timely and
accurate communication process. The pharmaceutical company must provide accurate
information describing the pharmacology (including the mechanism of the compound's
action), the effectiveness of the compound (known as the efficacy), and the risk
afforded by the compound (the safety). The F.D.A. must respond to this information
in a timely manner, either approving the compound in a reasonable period of time,
or disapproving the compound. When the compound is disapproved, the reasons for
the disapproval must be clear, and the procedure to appeal the decision must be
available. One of the mechanisms of appeal is the convening of an advisory
committee.

173 Advisory Committees:The collection of advisory committees to the F.D.A. is an
official body with codified rules in the Code of Federal Regulations (C.F.R.). It
consists of experts in the field in which the drug is expected to be effective. It
also includes experts from the disciplines of epidemiology and biostatistics, and,
in some instances, consumer advocates and representatives from the pharmaceutical
industry itself. The F.D.A. and the sponsor each provide written information to
the advisory committees, and each make presentations before the committee in a
session that is in general open to the public. The advisory committee provides
guidance to the F.D.A. by responding to specific questions posed to it by the
F.D.A. The F.D.A. considers both the content of the advisory committee's
deliberations and committee's answers to the F.D.A. posed questions very
seriously. However, the advisory committee can only render its most informed and
effective judgments when it is also the beneficiary of open, honest, accurate and
timely reporting from both the F.D.A. and the sponsor. To this end, the sponsor
spends months preparing briefing packets for the advisory committee, and weeks in
intense practice sessions preparing for the meeting.

174 Decision Processes: The decision process of the F.D.A. and its external
advisory committees is built on a foundation of the open, accurate and timely
dissemination of scientific evidence concerning the mechanism of drug action, the
drug's effectiveness, and the drug's safety profile. When this foundation is
solid, the process will swiftly identify and rapidly approve drugs that are safe,
effective, and meet a medical need. During this process, the public is protected
from compounds that either have no efficacy or are unsafe, these inferior products
being removed from further consideration. The maintenance and promotion of the
public health depend on the integrity of this system If either the sponsor or the
F.D.A. engage in dishonest or deceptive practices, the process begins to erode,
and the public suffers. This occurs either through the delay in the approval of
safe and effective compounds that meet a medical need, or through the publics
direct exposure to ineffective and dangerous compounds. The F.D.A. cannot make
timely, informed decisions when the sponsor denies them information.

175 Material Information: Embedded in this body of procedures are the channels
through which accurate, relevant, and material information describing the drugs
actions is promptly communicated to practicing physicians and through the
physicians to their patients. After the F.D.A. approves the drug and the drug is
advertised by the sponsor to the medical community, it is the sponsor's obligation
to closely monitor the use of the drug for the appearance of unanticipated adverse
events.

176 Adverse Events: An adverse event is an undesirable effect produced by the
drug. Many adverse events are identified during the early testing phase of the
drug. These adverse effects are included in the drug label. The drug label is the
official description of the drug, which appears, with all the labels of all
approved drugs in the Physicians' Desk Reference (PDR). This label describes the
effect of the drug, the population in which the drug is effective, the adverse
effects of the drug, and the general precautions reasonable and prudent physicians
should take in prescribing the drugs. Physicians and pharmacists rely on these
labels in making decisions involving the use of these drugs in their patients, and
they require that the information in these drug labels be accurate and up to date.
The sponsor owns the drug label. It has the responsibility for ensuring that the
information in the label is accurate and up to date. Although there are specific
regulations in the C.F.R., which govern the content of the label, responsibility
for guaranteeing that the information in the drug label is accurate resides with
the sponsor.

177 Labeling. Both physicians and patients have the right to draw independent
conclusions concerning the risks and benefits of a therapy. Their best and
clearest recommendations come from these considerations. Review of the Pondimin
and Redux labels reveals several understatements about the label. The federal
regulations require that the labeling shall contain a summary of the essential
scientific information needed for the safe and effective use of the drug. The
essential information was lacking in the warning labels for the fenfluramines,

178 Drug labels are to be designed for the safe and effective use of the drug.
There is no way to write an adequate label about a drug that has minimal efficacy
and causes serious, life threatening side effects because such a compound cannot
be used safely and effectively. By definition an adequate label cannot be written
for Redux since the drug cannot be used safely and effectively. The particular
defects in the Redux label are as follows.

179 An adequate label must contain complete statements about the effectiveness of
the compound. A complete description of the efficacy of Redux is not in the
Pondimin and Redux package inserts. The fenfiuramines were introduced as
pharmacological measures to reduce obesity. It is accepted that fenfluramine use
is associated with some weight loss (Weintraub). In the treatment of chronic
obesity, however, weight loss must be sustained. The researcher Guy-Grand studied
the effect of chronic administration of dexfenfluramine in 822 obese subjects from
24 centers in nine European countries. Although 339 patients withdrew from this
study and follow-up was less than one year for many, some estimates of the pattern
of weight loss of patients taking fenfluramine versus placebo were available. The
weight loss was not progressive after six months of dexfenfluramine
administration. The effect of withdrawal and re-introduction of dexfenfluramine in
the treatment of obesity was studied by DitschuneiL In this study, 25 patients who
had completed the INDEX trial (Guy-Grand et al.) were reexamined two months after
withdrawal of the study medication. The observed weight gain after withdrawal of
study medication was higher in patients treated beforehand with dexfenfluramine
(5.7 +- 2.64 lbs.) than in patients who had received the placebo (1.76 +- 2.20
lbs.). Thus, the "benefit" of Redux was a minimal, short lived weight loss that
did not persist with use of the drug for a year, and that was followed by rebound
weight gain after the drug was discontinued. Weight loss produced by Redux is
small, transient. Rebound weight loss occurs when the medication was discontinued.
This is important information that should have been shared with physicians and
patients in the 1997 Redux label. However, this statement was missing. It is
critical to note that there have never been any fenfluramine or dexfenfluramine
studies proving a reduction in long-term morbidity or mortality. This statement
was also absent from the label.

180 Deficient Warnings: The 1997 Redux warning is also deficient in its
description of the side effects of the drug. One of the important uses of the
black box label is to warn physicians above and beyond the statement in the
warning section of the label that a medications risks in general exceeds its
benefits and that the drug should only be used with exceeding care and in very
special circumstances.

181 The absence of a black box on the 1997 and earlier warning labels about the
hazards of Redux was an important omission for which Wyeth is responsible.

182 By 1997 Wyeth knew that the active ingredient in Redux was associated with
primary pulmonary hypertension and valvular heart disease. In addition, there were
no statements about the risk of therapy for primary pulmonary hypertension in
patients who used the therapy for less than three months. Available evidence in
1997 was that the use of Redux increased the risk of primary pulmonary
hypertension to 6.3 times greater than the background rate of this deadly disease,
regardless of duration (short term vs. long term) or timing of use (recent use vs.
remote use). There was no such statement in the 1997 Redux label.

183 Furthermore, not a single label for Redux and Pondimin state that the
fenfluramines cause primary pulmonary hypertension. Specifically, Wyeth knew of
the causal link between the fenfluramines and PPH and did not tell the FDA,
physicians, or the public of this discovery, allowing the fenfluramines to remain
on the market for eleven years after the discovery of this causal link. This
finding is buttressed by the absence of a statement in all of the Pondimin and
Redux warning labels concerning the assessment of the causal link between the
fenfluramines and PPH. In 1986, Servier produced a causality assessment of the
relationship between Redux and PPH. In that formal report, they concluded that
there were cases in which Redux was, by their own assessment, the cause of PPH.
This is a far stronger statement than the assertion that there is an association
between the fenfluramines and PPH. The finding of a causal link between Redux and
PPH was information that was critical to the risk benefit assessments of health
care providers and physicians, and its omission from the label was a glaring
omission. Furthermore, this information was not shared with the FDA Advisory
Committee in September of 1995, a pivotal FDA meeting in which, in view of the
public, the FDA discusses with experts the salient issues concerning the risks and
benefits of the fenfluramines.

184 Wyeth Owns their Drug Labels: It is important to note that Wyeth is not
required to have the F.D.A's approval to strengthen the label. Specifically, CFR
314.70(c)(2)(i) directly empowers Wyeth to strengthen the contraindication
section, the warning section, the adverse event section and/or the precaution
section without the prior approval of the F.D.A. As a result of the lassitude of
WYETH's labeling update procedures, WYETH provided incomplete accurate information
in its 1997 Redux label.

185 Since adverse events can occur in the population after the drug has been
approved, the public safety requires a mechanism be in place to quickly identify
and swiftly report the findings to the medical community and the F.D.A. Although
both the F.D.A. and the sponsor share a role in this vigilance monitoring, it is
the sponsor who bears primary responsibility for identifying these adverse events.
Just as the F.D.A. does not have the resources available to carry out its own pre
drug approval clinical tests, its ability to monitor the occurrence of adverse
events in large populations is sharply circumscribed. Again, the sponsor bears the
burden of monitoring and identifying the appearance of these adverse events. The
sponsor's role is simple - to report the adverse events. Although the sponsor may
which to try to explain the occurrence of the adverse effect, its major role, like
that of a sentry, is to be vigilant and quickly report the occurrence of the
unexpected. The fine differentiation between whether the adverse event is
merely[FNs] associated with the drug or caused by drug5 comes later. The first
task is to report the adverse event to the F.D.A. and the medical community. Wyeth
had opportunities that arose as early as 1991 to take action in 1) relabeling
fenfluramine and 2) informing the medical community and the population at large of
a possible VHD danger presented by fenfluramine because of early evidence
reflected in adverse drug event reports of an association between valvular heart
disease and fenfluramine use, but chose not to do this Additionally, in 1994-1995
Wyeth again had the opportunity to to take action in 1) relabeling fenfluramine
and 2) informing the medical community and the population at large of a possible
VHD danger presented by fenfluramine because of ADE based evidence of an
association between valvular heart disease and fenfluramine use, but chose not to
do this. The occurrence These were missed opportunities to respond to new
information, signalling a new but easily anticipated risk factor of fenfluramine
use.

      FNs. A drug is associated with an adverse event if the use of the drug and
      the adverse event occur simultaneously. A drug causes the adverse event if
      the presence of the drug excites the production of the adverse event .e.g.
      the use of thalidomide and the occurrence of phocomelia (limb
      foreshortening).

186 Wyeth CFR Violations: In the case of Wyeth and fenfluramines, Wyeth committed
several important violations of federal code in the communication with the F.D.A.
and with the label. The authorities and Wyeth understood the importance of honest,
accurate, and timely communication with the F.D.A. However, Wyeth's action
involving their responsibilities to the patients who take their products displays
a consistent pattern of regulation violation.

187 The vigilance requirement: Immediate upon the assumption by Wyeth of the
product Pondimin (fenfluramine), Wyeth was required to collect and manage
information concerning the adverse effects experiences reported to them by health
care providers. This is a serious responsibility for even the safest of drugs.
Health care providers rely on the manufacturers to provide prompt reports
describing any new drug findings, which will effect an assessment of the
effectiveness and safety of that compound. It is the responsibility of the drug
manufacturers to look out for the occurrence of side effects of their drugs and to
report them promptly, both to the medical community and to the F.D.A. In the case
of fenfluramine, there was readily available information that should have
indicated to Wyeth the need for especially perspicacious vigilance. Specifically,
the appearance of an epidemic of pulmonary hypertension in Europe following the
use of the amphetamine aminorex fumarate in Austria, Switzerland and Germany from
1968-1972, provided adequate evidence that anorexigens offered substantial safety
risk to the populations exposed to them. This epidemic should have raised the
specter of the occurrence of pulmonary hypertension associated with fenfluramine,
itself a halogenated amphetamine.63 in addition, a harbinger of future safety
problems with the fenfluramines was contained in the Brenot64 article in 1993 that
identified cases of pulmonary hypertension recognized during or after the
ingestion of the fenfluramines. Each of these occurrences served as warnings that
Wyeth should be extremely sensitive to the occurrence of pulmonary hypertension in
patients who were exposed to the fenfluramines warnings that Wyeth did not heed.
However, rather than diligently search for such cases, and promptly and accurately
report them both to the F.D.A. and to the medical community, Wyeth's reporting
system to the F.D.A. and the medical community was slow, unresponsive and laggard.
In addition, the medical community must have up-to-date information from the
sponsor on the occurrence of adverse events.

188 Wyeth's negligent performance in updating the medical community was
inexcusable. The F.D.A. itself specifically states that the labeling requirements
in the CFR do not prohibit a drug sponsor from warning health care professionals
whenever possibly harmful adverse effects associated with the use of the drug are
discovered (note causality is not required for this warning). Wyeth did not
include a warning to physicians through the PDR label of the dangerous association
between the fenfluramines and valvular heart disease until after the 1997
appearance of the Connolly article in the New England Journal of Medicine,, a full
three years after the appearance of such cases in its own database. In addition
the FDA grants the sponsor authority to directly notify the medical community
through the issuance of letters directed to health care professionals. Wyeth did
tell the Endocrinologic and Metabolic Drugs Advisory Committee to the F.D.A. in
September 1995 about the association of the fenfluramines and valvular heart
disease. For three years after the appearance of cases associating the
fenfluramines and valvular heart disease, Wyeth never notified physicians of this
dangerous association.

189 A second violation of the usual and customary standard of honest, open,
accurate, and prompt reporting of information was Wyeth's delay in moving to
update the label for a description of the compounds association with each of
primary pulmonary hypertension and valvular heart disease. Given the availability
of information concerning the link between anorexigen use and pulmonary
hypertension as provided by the European aminorex experience and the report of the
Brenot findings, Wyeth, with almost 40 cases of pulmonary hypertension reported to
it in patients in the summer of 1994, had the responsibility to warn the F.D.A. of
this finding openly, accurately, honestly, and promptly. They also had a grave
responsibility to warn physicians prescribing this compound of the association
between the use of fenfluramines and the occurrence of pulmonary hypertension and
primary pulmonary hypertension immediately. Patients commonly rely on their
physicians for an interpretation of the benefits and risks of taking medications.
These physicians, in turn, rely on the accuracy of the label, whose responsibility
for accuracy resides with both the Sponsor and F.D.A. Given the information known
to the Sponsor concerning the association between the use of the fenfluramines and
primary pulmonary hypertension in the summer of 1994, taken against the backdrop
of the European aminorex experience and the findings of the Brenot manuscript, the
delay in the update of the Pondimin label for valvular heart disease, primary
pulmonary hypertension was a clear dereliction.

190 Willful negligence: The label of a drug is the descriptor of the drugs
indications and problems and is the balance scale on which health care
professionals weigh the risks versus the benefits of the drug. The fact that the
F.D.A. insisted that Wyeth meet minimal standards in the information conveyed in
the label must not be allowed to hide the fact that it is the sponsor's label The
sponsor owns its drug label. The sponsor can exceed the minimal F.D.A. standards
for the label contents, and, in fact, the F.D.A. recognizes that the sponsor is in
the best position to exceed these standards for the public good since it is often
the sponsor who first receives information about the side effects associated with
their drug. The F.D.A. therefore grants the sponsor the authority to strengthen a
label in the contraindications, warnings, adverse events, and precautions segments
of the label. The F.D.A also grants the sponsor the authority to remove
unrealistic claims of efficacy from the label without prior notification. In fact,
the F.D.A. even grants the sponsor the right to place a black box (which is a
statement in the label which describes a special problem that leads to death or
serious injury which is offset and highlighted from the remainder of a label by
being placed in a black box) on the sponsor's label. Even though the F.D.A. most
often must insist on this black box label, the F.D.A. will not block the effort of
a sponsor which, through its own assessment of the risk associated with the
compound, chooses to place this prominent warning on their own product of their
own accord. Not only did Wyeth choose not to place such a black box on Pondimin or
on Redux; Wyeth fought the F.D.A.'s attempt to so label these products from 1995
through 1997.

191 The Advisory Committee Meeting for Redux: The information concerning the
association between pulmonary hypertension and Pondimin would have been
particularly helpful to the F.D.A. advisory committee in its review of the
information concerning the approval of Redux (dexfenfluramine). Redux had already
been disapproved by the F.D.A. in 1994. The pharmaceutical companies (Wyeth and
Servier) exercised their right to have a hearing before the Endocrinologic and
Metabolic Drugs Advisory Committee to the F.D.A. Unfortunately, they did not
exercise their responsibility to provide honest, accurate, and open information to
the advisory committee. During this public meeting, the advisory committee members
recognized the importance of a possible link between pulmonary hypertension and
Redux. They also understood the strong chemical link between Pondimin, which had
been on the market for more than 10 years, and Redux. Both drugs had the same
active ingredient. According to the transcript of the September 28, 1995 Advisory
Committee meeting, at least one advisory committee member believed that the
experience of Pondimin was relevant and would contribute important information
useful in assessing the association between the fenfluramines and pulmonary
hypertension. The sponsors had this information that the committee requested.
Wyeth did not report it to the committee in response to verbal questions asked by
advisory committee members, when in fact; Wyeth was in possession of this relevant
information.

192 Furthermore, Wyeth did not inform the Advisory Committee that the causality
assessment conducted by Servier concluded that cases of primary pulmonary
hypertension were likely to be caused by Redux,

193 Unfortunately, this consistent absence of vigilance, dearth of accurate and
prompt reporting to the F.D.A., and paucity of open communication with the F.D.A.
Advisory committee led to the continued, falsely labeled availability of Pondimin,
and the subsequent approval of Redux (dexfenfluramine). The availability of each
of these medications to the public has led to the occurrence of primary pulmonary
hypertension and heart valve disease that is directly attributable to their
exposure to the fenfluramines.

194 Regulation Sec. 201.56 (b): General requirements on content and format of
labeling for human prescription drugs. Prescription drug labeling described in
Sec. 201.100(d) shall contain the information in the format required by Sec.
201.57 and shall meet the following general requirements:(a) The labeling shall
contain a summary of the essential scientific information needed for the safe and
effective use of the drug. (b) The labeling shall be informative and accurate and
neither promotional in tone nor false or misleading in any particular.

195 Wyeth violated this statute by 1) not adding a statement to the Pondimin label
in the 1995 Physicians Desk Reference concerning the association between Pondimin
and heart disease 2) not updating the Pondimin label for 1995 when new cases of
pulmonary hypertension were reported 3) not advising physicians of the association
between Pondimin (fenfluramine) and primary pulmonary hypertension, an often fatal
disease. It is important to note that Wyeth is not required to have the F.D.A.'s
approval to strengthen the label. Specifically, CFR 314.70(c)(2)(i) directly
empower Wyeth to strengthen the contraindication section, the warning section, the
adverse event section and/or the precaution section without the prior approval of
the F.D.A. As a result of the lassitude of Wyeth's labeling update procedures,
Wyeth provided inaccurate information in its 1995 Pondimin label. Page 76 of 88

196 Regulation Sec. 201.57(d): Specific requirements on content and format of
abeling for human prescription drugs. Each section heading listed in Sec.
201.56(d), if not omitted under Sec. 201.56(d)(3), shall contain the following
information in the following order(d) Contraindications. Under this section
heading, the labeling shall describe those situations in which the drug should not
be used because the risk of use clearly outweighs any possible benefit. These
situations include administration of the drug to patients known to have a
hypersensitivity to it; use of the drug in patients who, because of their
particular age, sex, concomitant therapy, disease state, or other condition, have
a substantial risk of being harmed by it; or continued use of the drug in the face
of an unacceptably hazardous adverse reaction. Known hazards and not theoretical
possibilities shall be listed, e.g., if hypersensitivity to the drug has not been
demonstrated, it should not be listed as a contraindication. If no
contraindications are known, this section of the labeling shall state "none
known."

197 In fact, even though the association of Pondimin with valvular heart disease
was known in 1994, Wyeth chose not to describe the patients in whom the risk of
valvular heart disease associated with Pondimin use would outweigh the benefits.

198 Regulation Sec 201.57(e): Specific requirements on content and format of
labeling for human prescription drugs. Each section heading listed in Sec.
201.56(d), if not omitted under Sec. 201.56(d)(3), shall contain the following
information in the following order: (e) Warnings. Under this section heading, the
labeling shall describe serious adverse reactions and potential safety hazards,
limitations in use imposed by them, and steps that should be taken if they occur.
The labeling shall be revised to include a warning as soon as there is reasonable
evidence of an association of a serious hazard with a drug; a causal relationship
need not have been proved. A specific warning relating to a use not provided for
under the "Indications and Usage section of the labeling may be required by the
Food and Drug Administration if the drug is commonly prescribed for a disease or
condition, and there is lack of substantial evidence of effectiveness for that
disease or condition, and such usage is associated with serious risk or hazard.
Special problems, particularly those that may lead to death or serious injury may
be required by the Food and Drug Administration to be placed in a prominently
displayed box. The boxed warning ordinarily shall be based on clinical data, but
serious animal toxicity may also be the basis of a boxed warning in the absence of
clinical data. If a boxed warning is required, its location will be specified by
the Food and Drug Administration. The frequency of these serious adverse reactions
and, if known, the approximate mortality and morbidity rates for patients
sustaining the reaction, which are important to safe and effective use of the
drug, shall be expressed as provided under the "Adverse Reactions" section of the
labeling.

199 Wyeth violated this regulation by choosing not to warn health care providers
about the relationship between Pondimin and cardiac valvulopathy, a relationship
that was readily apparent to Wyeth in 1994. Knowledge of this association should
have been communicated with health care providers in 1994, and should also have
led to a strongly worded warning in the 1995 PDR about the association between
Pondimin and valvular heart disease. Furthermore, Wyeth violated this regulation
by not moving aggressively to warn health care providers of the association
between Pondimin and primary pulmonary hypertension. These warnings should have
included, but not limited to official communications with physicians beginning in
the summer of 1994 about the occurrence of 37 cases of pulmonary hypertension,
reported to Wyeth in June 1994. These warnings would also include strengthening
the label statement for Pondimin in 1995, including the new information about 37
cases of pulmonary hypertension.

200 Regulation Sec. 201.57(f)(1): Specific requirements on content and format of
labeling for human prescription drugs. Each section heading listed in Sec.
201.56(d), if not omitted under Sec. 201.56(d)(3), shall contain the following
information: (f) Precautions. Under this section heading, the labeling shall
contain the following subsections as appropriate for the drug: (1) General. This
subsection of the labeling shall contain information regarding any special care to
be exercised by the practitioner for safe and effective use of the drug, e.g.,
precautions not required under any other specific section or subsection of the
labeling.

201 Wyeth violated this regulation by choosing not to place a statement in its
precautions section on the Pondimin label describing the association between
valvular heart disease and Pondimin use, a relationship that was discovered in
1994. A precaution consisting of the requirement that patients be monitored for
valvular heart disease should have included in the 1995 PDR.

202 Regulation Sec. 201.57(g)(1): Specific requirements on content and format of
labeling for human prescription drugs. Each section heading listed in Sec.
201.56(d), if not omitted under Sec. 201.56(d)(3), shall contain the following
information: (g) Adverse Reactions. An adverse reaction is an undesirable effect,
reasonably associated with the use of the drug that may occur as part of the
pharmacological action of the drug or may be unpredictable in its occurrence. (1)
This section of the labeling shall list the adverse reactions that occur with the
drug and with drugs in the same pharmacologically active and chemically related
class, if applicable.

203 Wyeth violated this regulation by choosing not to place a statement in the
Redux label in 1996 which described the adverse event experience for Pondimin.
Since the active ingredient in Pondimin is the same active ingredient in Redux,
Wyeth should have placed a statement in the 1996 label for Redux describing the
association between Pondimin and valvular heart. This 1996 label for Redux should
also have included a statement describing the relationship between Pondimin and
primary pulmonary hypertension. Analogously, a statement describing the
relationship between Redux and primary pulmonary hypertension should have been
placed in the 1996 label for Pondimin.

204 Regulation Sec. 314.70(c)(2)(i): Supplements and other changes to an approved
application. (c) Supplements for changes that may be made before FDA approval. An
applicant shall submit a supplement at the time the applicant makes any kind of
change listed below in the conditions in an approved application, unless the
change is made to comply with an official compendium. A supplement under this
paragraph is required to give a full explanation of the basis for the change,
identify the date on which the change is made, and, if the change concerns
labeling, include 12 copies of final printed labeling. The applicant shall
promptly revise all promotional labeling and drug advertising to make it
consistent with any change in the labeling. The supplement and its mailing cover
should be plainly marked: "Special Supplement--Changes Being Effected" (2) Changes
labeling to accomplish any of the following: (i) To add or strengthen a
contraindication, warning, precaution, or adverse reaction;

205 Wyeth was obligated to change the label to strengthen the contraindication,
warning, precaution and adverse event section of the label for Pondimin in 1995,
1996 and 1997 to include the information about the relationship between Pondimin
and valvular heart disease. Wyeth was obligated to change the label to strengthen
the contraindication, warning, precaution, and adverse event section of the Redux
label in 1996, and 1997 to add the information about the relationship between
Pondimin and valvular heart disease. In fact, Wyeth had specific information about
the association between valvular heart disease and the use of fenphen and did not
respond appropriately to this important signal. Additionally, in the 1980's

206 Regulation Sec. 14.171: Utilization of an advisory committee on the initiative
of FDA. (f) Presentation of all relevant information about the matter will be made
in open session unless it relates to an IND the existence of which has not
previously been disclosed to the public as defined in Sec. 20.81 or is otherwise
prohibited from public disclosure under part 20 and the regulations referenced
therein. Sections 314.430 and 601.51 determine whether, and the extent to which,
relevant information may be made available for public disclosure, summarized and
discussed in open session but not otherwise made available for public disclosure,
or not in any way discussed or disclosed in open session or otherwise disclosed to
the public.

207 Wyeth violated this regulation by not providing relevant information about the
adverse event experience of Pondimin when Dr. New specifically requested this
verbally at the September 28, 1995, advisory committee for the approval of Redux.
Although the answer to the Advisory Committee's question was in Wyeth's possession
at the time of the meeting, Wyeth did not provide the information when requested.
It is useful to note that Wyeth was a sponsor of Redux at the September 18, 1995,
advisory committee meetings considering Redux according to definition set out by
section 310.3(j) as follows.

208 Sec. 310.3: Definitions and interpretations. (j) The term sponsor means the
person or agency that assumes responsibility for an investigation of a new drug,
including responsibility for compliance with applicable provisions of the act and
regulations. The "sponsor" may be an individual, partnership, corporation, or
Government agency and may be a manufacturer, scientific institution, or an
investigator regularly and lawfully engaged in the investigation of new drugs.

209 Wyeth was required to provide the required information verbally requested by
Dr. New and violated regulations and the standard of care of a reasonable and
prudent drug company by not providing it.

210 Conclusions: To a reasonable degree of scientific and medical certainty, the
fenfluramines cause primary pulmonary hypertension. The understanding of the
relationship between fenfluramine exposure and disease is clear from a review of
the relevant scientific information. Epidemiologists develop our understanding of
the relationship between exposure to a risk factor and disease through the
application of scientific principles as described by Friedman, Bradford Hill,
Salim Yusuf, and others. These principles are the basis by which we assess whether
a risk factor causes a disease (i.e. excites the production of a disease) or is
merely associated with the disease (the risk factor and the disease happen to
occur simultaneously). The results of these studies appear in the literature.

I certify that the foregoing statements made by me are true and correct. Executed
this 31 day of March 2005 at Houston, Texas.