The following document was filed in the Louisiana lawsuit of Tarver v. Wyeth. It discusses some concerns that the FDA had with metoclopramide and movement disorders.
Regarding the safety of MT100, my main concern relates to the extrapyramidal
adverse events known to be caused by metoclopramide.[FN3] Reglan labeling states
that extrapyramidal symptoms, mainly acute dystonic reactions, occur in
approximately 0.2% of patients receiving oral metoclopramide at doses between 30
to 40 mg per day. Dr. Prohaska notes that during the clinical development program
for MT100, there were two subjects reporting acute dystonic reactions, which
demonstrates the potential for extrapyramidal adverse events even at the dose (16
mg) of metoclopramide contained in MT 00. Akathisia has also been reported with
metoclopramide in the literature,[FN4] and metoclopramide-induced akathisia can
even apparently lead to suicide attempt, as reported in the literature.[FN5]
United States District Court, W.D. Louisiana.
TARVER, et al,
v.
WYETH INC et al.
No. 04CV02036.
2004.
(Report or Affidavit of Eric P. Bastings, M.D.)
MEMORANDUM
DEPARTMENT OF HEALTH & HUMAN SERVICES
Public Health Service
Food and Drug Administration
Date: May 18, 2004
From: Eric Bastings, MD.
To: Russell Katz, MD
Subject: 21-645 Myzan (MT100)
This submission is a new drug application (NDA) for the use of MYZAN (previously
MT100) for the intermittent treatment of acute migraine with and without an aura
in adults. MYZAN is a combination product consisting of a metoclopramide
hydrochloride monohydrate (16 mg/tablet) shell, and naproxen sodium (500
mg/tablet) central core.
For this NDA, Josephine Jee provides the chemistry review, Dr. Kathleen Haberny
provides the pharmacology/toxicology review, Dr. Kofi A. Kumi provides the
clinical pharmacology review, Dr. Yeh-Fong Chen provides the statistical review
and Dr. Kevin Prohaska provides the clinical review. Jerry Phillips provides a
consultative review of the drug name, carton, and container for the Division of
Medication Errors and Technical Support. Jeanine Best from ODS provides a review
of the patient insert labeling. Dr. Ni Khin provides the DSI review.
Chemistry, Manufacturing and Controls (CMC)
Josephine Jee recommends approval from a CMC perspective.
Pharmacology/Toxicology
As noted by Dr. Haberny and by Dr. Barry Rosloff (Pharm/Tox team leader), Pozen
was required to conduct a two year rat carcinogenicity study. MT 100 caused
increased incidences of mammary tumors, adrenocortical tumors, pheochromocytomas,
and pancreatic islet cell tumors. Pozen postulates that all of these tumors result
from the increased levels of prolactin known to be produced by metoclopramide, and
Dr. Rosloff agrees that this is presumably the case, although there are inadequate
data to fully support this assertion.
Even if prolactin as a mechanism for carcinogenicity is accepted, there is no
evidence that it would not be operative in humans. Metoclopramide is known to
increase prolactin at therapeutic doses in humans, and is even cited as the
galactologue of choice in the literature.[FN1] If prolactin is indeed the cause of
the tumors, the increased prolactin level reported with therapeutic doses of
metoclopramide in the human is not a comforting observation. Dr. Rosloff notes
that most dopamine-blocking antipsychotic drugs produce mammary tumors in rodents
but as a class are not thought to produce mammary tumors in humans. However, Dr.
Rosloff cites a more recent study which suggested a small but significant risk of
breast cancer with dopamine antagonists in the human.[FN2]
FN1. Gabay MP. Galactogogues: medications that induce lactation. J Hum Lact.
2002; Aug; 18(3): 274-9.
FN2. Wang PS, et al. Dopamine antagonists and the development of breast
cancer. J. Arch Gen Psychiatry. 2002 Dec; 59(12): 1147-54.
Assuming a non-genotoxic mechanism, Dr Rosloff believes that there is somewhat of
a safety margin with respect to expected human plasma levels of metoclopramide
(except for female mammary tumors, where there was no clear NOEL established). On
the other hand, the proposed human administration is meant to be intermittent,
which lowers the risk for a non-genotoxic carcinogen (with the caveat that some
patients will probably use the drug chronically even if labeling recommends
intermittent use). Also, Dr. Haberny and Dr. Rosloff remind of the context of
other drugs used in migraine, several of which have carcinogenicity findings
described in the labeling.
Clinical Pharmacology and Biopharmaceutics
Dr. Kumi recommends approval. The pharmacokinetics of naproxen and metoclopramide
were considered bioequivalent following administration of the MT 100 formulation
versus the individual naproxen and metoclopramide components. Metoclopramide
pharmacokinetics were unaffected by a migraine attack, but naproxen AUC and Cmax
were slightly reduced (10- 15%). There was a slight food effect for naproxen and
metoclopramide pharmacokinetics, with Tmax delayed by approximately 1 hour for
both drugs, and naproxen Cmax reduced by approximately 24% by administration with
food. In patients with moderate hepatic impairment, metoclopramide
pharmacokinetics were similar to those of healthy subjects and total naproxen AUC
increased by 27%. However, the AUC and Cmax of unbound naproxen increased by
greater than 2- fold in subjects with hepatic impairment compared to healthy
subjects. Because of increased exposure to naproxen, particularly unbound
naproxen, Dr. Kumi recommends that MT 100 should be used with caution in patients
with hepatic impairment. Dr. Kumi also recommends that MT 100 is not administered
to patients with creatinine clearance below 40 mL/min (this is based on
metoclopramide labeling recommendations; no study was conducted in patients with
renal impairment). OCPB made several labeling, dissolution and specifications
recommendations. However, since my final recommendation is a non approval action,
these recommendations will not be implemented in this review cycle.
Clinical and Biostatistics
Dr. Kevin Prohaska and Dr. Yeh-Fong Chen recommend non-approval. According to the
combination policy [CFR 300.50 (a)], two or more drugs may be combined in a single
dosage form when each component makes a contribution to the claimed effects and
the dosage of each component is such that the combination is safe and effective
for a significant patient population requiring such concurrent therapy as defined
in the labeling for the drug. Pozen has not met the combination policy requirement
for MT 00, and has not established that both active drug components make a
contribution to the claimed effects. Special cases of this general rule are where
a component is added to enhance the safety or effectiveness of the principal
active component, or to minimize the potential for abuse of the principal active
component. MT100 does not correspond to either of these special cases.
Two factorial studies explored the contribution of individual components to the
safety and efficacy of the combination: Study MT100-301 and -304. I concur that
both studies demonstrate a clear benefit of the combination over metoclopramide
for the sustained headache pain response, but both studies fail to demonstrate a
benefit of the combination over naproxen. The primary endpoint for the two
factorial studies was sustained headache response (defined as moderate to severe
headache pain at baseline going to no or mild pain at 2 hours and did not relapse
or require rescue medication up to hour 24). The following table (copied from
table 3 of Dr. Prohaska review, page 10) summarizes the primary endpoint results
for the factorial studies:
TABLE
Both studies failed to reach statistical significance for the primary endpoint
analysis in the comparison of MT100 and naproxen. It is important to briefly
discuss here the differences between the analysis conducted by the sponsor and
these conducted by FDA.
For Study 301, Pozen tried to use a post-hoc "refinement analysis" (ordered
logistic regression with baseline pain and investigator site as the covariates)
instead of the a priori designated logistic regression. The use of the "refined"
analysis led to a statistically significant difference between MT 00 and naproxen
(p=0.025). In a 3/27/00 telecon with Pozen (minutes of which are filed in DFS and
were sent to the sponsor), the division notified Pozen that FDA does not agree
that Study 301 is positive on its primary endpoint, "sustained response" because
Pozen performed a "refinement" to the analysis which was not prespecified, and was
performed after the data were unblinded. Pozen proposed using an "almost positive"
Study 301 and a positive second study to demonstrate superiority of MT100 over its
components and then using other trials to demonstrate efficacy over placebo. The
division agreed to this approach but also reminded Pozen that the division would
be looking at improvements in the secondary outcome measures of nausea,
photophobia, and phonophobia in order to establish efficacy.
For Study 304, as discussed by Dr. Chen on page 16 of her review, and according to
the study report, Pozen used ordered logistic regression (with baseline pain and
investigator site as covariates) to test MT100 versus naproxen, and MT100 versus
metoclopramide. Using that analysis, MT 00 was significantly better than both
naproxen (p=0.03) and metoclopramide (p<0.001). However, Dr Chen (and Dr.
Prohaska) found that ordered logistic regression was not the sponsor's protocol
specified primary method. The protocol specified method was the extended Mantel
Haenszel statistic with score of 0, 1, and 2 for the three ordered categories (no
sustained pain relief, sustained pain relief, and sustained pain free), and using
a model that controls for center, baseline pain and gender. Pozen was informed of
this discrepancy during the NDA review cycle, and asked to reanalyze the data
according to the pre-stated analysis plan. Pozen's reanalysis showed a slightly
higher p- value for comparison of MT100 to naproxen (p=0.038 versus p=0.030).
Pozen concluded that the interpretation of the statistical significance of the
results remains the same. As discussed by Dr. Chen on page 22 of her review, Pozen
performed this analysis using a SAS macro written by Koch. Dr. Chen requested that
Pozen submits to the division the program and the SAS macro, and then evaluated
Pozen's analysis results. Dr. Chen found that by using the SAS macro written by
Koch, the p- value for the comparison between MT100 and naproxen is 0.063 and not
0.038. Dr. Chen noted that Pozen mistakenly used equal weight for all strata in
their analysis, instead of a weight that is comparable to the stratum's proportion
of patients in the trial.
To further evaluate the study results, Dr Chen also analyzed the data by
stratifying the center factor only, which is the only factor usually stratified.
This led to a non significant difference between MT100 and naproxen (p=0.09), both
with the Koch's SAS macro or with Dr. Chen's SAS program. In addition to the lack
of a statistically significant difference between MT100 and naproxen in the
factorial studies, the treatment effect size (for sustained relief) of MT100 over
naproxen is clinically marginal (4-6%). The trend for statistical significance
despite the small treatment effect size is probably explained by the very large
sample size of the factorial studies (i.e. n=2627 in Study 304), much larger than
for typical migraine studies. In that sense, the studies (and especially Study
304) were apparently overpowered. As discussed by Dr. Prohaska on page 10 of his
review, the lack of benefit of MT100 over naproxen is further supported by the
fact that MT100 was not statistically different from naproxen for all key symptoms
at the 2-hour endpoints typically used in migraine trials, and for sustained
nausea-free, photophobia-free and phonophobia-free (as detailed in the following
table, copied from page 11 of Dr. Prohaska review).
TABLE
The lack of a demonstrated benefit of the combination over its components is
sufficient, in my opinion, to justify a non approvable action. There are however
additional efficacy and safety issues, as discussed below. The sponsor conducted
three key efficacy studies (MT100-306, -308 and -303) comparing MT100 to placebo
in all three studies, and to Imitrex 50 mg as an active comparator in two studies
(MT100-306 and -308).
Study 306 can be considered as an almost positive study. For the comparison to
placebo (primary comparison), MT100 was statistically superior to placebo for the
key symptoms of pain response, nausea, and photophobia at 2 hours. However, the
comparison did not reach statistical significance for phonophobia, but trended
strongly (p=0.06 by the sponsor analysis, and p=0.052 by FDA analysis). I believe
that this study could be used to support a migraine claim for MT100 if a second
study showed efficacy on all key symptoms of migraine at 2 hours (with the
remaining obligation to address the combination rule).
Study 308 is, in my opinion, a negative study. In Study 308, the primary
comparison was for equivalence to sumatriptan. As discussed by Dr. Chen in page
4041 of her review, the division did not agree with the analysis plan and informed
Pozen of that disagreement long before the NDA was filed (in a 6/8/01 email).
Specifically, the division identified three problems: the sponsor did not plan to
compare MT100 with the entire approved therapeutic range for Imitrex, the analysis
plan failed to define what margin of difference would be acceptable to determine
equivalence, and the comparison did not include the key migraine associated
symptoms. Pozen proposed a margin for non inferiority, which the division did not
agree to, and did not address the other two issues. Therefore, as concluded by Dr.
Chen, the non- inferiority comparison is not acceptable. For that reason, I will
not discuss the comparison to sumatriptan further, and I will only consider the
comparison to placebo.
In the comparison of MT100 to placebo, the p- values are below 0.05 for migraine
pain (p=0.001) and photophobia (p=0.044), but above 0.05 for nausea and
phonophobia. Dr Chen expresses some additional concerns for the comparison of
MT100 to placebo, because this was a secondary comparison, and because the study
was overpowered. In my opinion, since the division rejected the primary comparison
to sumatriptan before seeing any result of the study, it is reasonable to consider
the comparison to placebo as the primary comparison. For pain relief, the
treatment effect size of 12% is in the lower end of effect sizes observed for
acute migraine treatment, but appears large enough to overcome the concern for a
significance difference mostly supported by overpowering the study. For
photophobia however, the effect size is marginal (8%). In addition, since all key
endpoints did not win against placebo, I believe that a correction for multiple
comparisons must be used to interpret the individual key associated symptoms
analysis. I believe that the significance (p=0.044) of the comparison of
photophobia between MT100 and placebo would not be sustained after correction for
multiplicity. Overall, I believe that this trial shows a superiority of MT100 over
placebo for migraine pain only, which comes short of an acute migraine treatment
claim.
The third pivotal efficacy trial, Study MT100-303, was also a negative study. For
this study, the primary comparison was between MT100 and placebo, and used
sustained pain relief instead of the typical 2-hour endpoints. Pozen can be
commended for randomizing non responders at 2 hours to a second dose of MT100 or
placebo, which was a secondary comparison. According to the study report, ordered
logistic regression analysis of the sustained pain response data demonstrated that
an initial dose of MT100 was significantly better than placebo (p= 0.048). Dr.
Chen however reanalyzed the data and got a p-value of 0.062. According to the
sponsor's second and last protocol amendment, the 2-hour sustained response data
were to be analyzed by ordered logistic regression controlling for center,
baseline severity and gender. Dr. Chen (see page 49 of her review) found that
Pozen's p- value (0.048) was actually obtained from an ordered logistic regression
model that did not include any covariates, although the study report mentioned the
baseline pain and investigator site as covariate. In addition, according to the
last version of the study protocol, there were two primary endpoints: the 2- hour
sustained response for subjects initially randomized to MT100 or placebo and the
4- hour sustained response for MT100 non- responders randomized to a second dose
of MT100 or placebo. Pozen, however, did not specify that the 4- hour sustained
response was the second primary endpoint in the study report, and did not propose
any multiple comparison method for preserving the type I error rate. This
observation strengthens the above conclusion that this study was negative. Also,
there was no evidence to support the efficacy of a second dose of MT100 in
non-responders.
Regarding the safety of MT100, my main concern relates to the extrapyramidal
adverse events known to be caused by metoclopramide.[FN3] Reglan labeling states
that extrapyramidal symptoms, mainly acute dystonic reactions, occur in
approximately 0.2% of patients receiving oral metoclopramide at doses between 30
to 40 mg per day. Dr. Prohaska notes that during the clinical development program
for MT100, there were two subjects reporting acute dystonic reactions, which
demonstrates the potential for extrapyramidal adverse events even at the dose (16
mg) of metoclopramide contained in MT 00. Akathisia has also been reported with
metoclopramide in the literature,[FN4] and metoclopramide-induced akathisia can
even apparently lead to suicide attempt, as reported in the literature.[FN5]
FN3. In a web search, I also found that metoclopramide manufacturers are
apparently being sued for the tardive dyskinesia imputed to long term use of
Reglan (see http://druginte.com/reglan tardive dyskinesia.htm; http://
www.reglan-lawsuit.com).
FN4. Akagi H, Kumar TM. Lesson of the week: Akathisia: overlooked at a cost.
BMJ. 2002 22; 324(7352): 1506-7.
FN5. Chow LY, Cung D, Leung V, Leung TF, Leung CM. Suicide attempt due to
metoclopramide-induced akathisia. Int J Clin Pract 1997; 51: 330-331
Reglan labeling also states that tardive dyskinesia, a syndrome consisting of
potentially irreversible, involuntary, dyskinetic movements may develop in
patients treated with metoclopramide. The label states that the risk increases
with the duration of treatment and the cumulative dose. The literature supports
that metoclopramide-induced tardive dyskinesia can be permanent.[FN6] That type of
side effect can only be expected to be observed when large populations are exposed
to the drug for a sufficient period of time, and not during the drug development
program such as the one for MT 00. In that sense, the absence of cases of tardive
dyskinesia in the MT 00 database is not indicative that this adverse reaction will
not occur. Indeed, in an analysis of 67 cases of metoclopramide associated tardive
dyskinesia, the mean length of treatment with metoclopramide before the onset of
symptoms was 20 months.[FN7] In that study, the symptoms were still present 6
months or more after discontinuation of metoclopramide in most patients on whom
long-term follow-up was provided. Tardive tremor due to metoclopramide was also
recently described.[FN8] There is also a warning in Reglan labeling about a risk
for depression, suicidal ideation and suicide. Finally, the label reports rare
cases of neuroleptic malignant syndrome, which is potentially fatal.
FN6. Jeste DV, Caligiuri MP. Tardive dyskinesia. Schizophr Bull. 1993;
19(2):303-15.
FN7. Sewell DD, Jeste DV. Metoclopramide-associated tardive dyskinesia. An
analysis of 67 cases. Arch Fam Med. 1992 Nov; 1(2): 271-8.
FN8. Tarsy D et al. Mov Disord 2002. 17 (3): 620-621.
In light of the concern with the CNS toxicity of metoclopramide, FDA has approved
metoclopramide for short-term therapy (4 to 12 weeks) of gastroesophageal reflux,
and only when conservative treatment fails. However, it is well known that
labeling restrictions are not always followed by patients and practitioners, and
that chronic use may occur despite labeling recommendations. Indeed, a published
paper reports that 32% of metoclopramide users in a sample of elderly ambulatory
patients used metoclopramide for longer than a year.[FN9] The exact incidence of
metoclopramide-induced tardive dyskinesia remains unclear, but well known movement
disorder experts have stated that to prevent persistent and disabling movement
disorders, long-term use of metoclopramide should be avoided.[FN10] Two factors
mitigating the risk of tardive dyskinesia in the case of MT100 are the recommended
chronic/intermittent use and the relatively low dose of metoclopramide. However,
given the concern for permanent CNS toxicity, a robust clinical benefit is needed
to justify the risks of MT100, especially for a migraine indication.
FN9. Stewart RB, Cerda JJ, Moore MT, Hale WE. Metoclopramide: an analysis of
inappropriate long-term use in the elderly. Ann Pharmacother. 1992; 26
(7-8): 977-9.
FN10. Miller LG, Jankovic J. Metoclopramide-induced movement disorders.
Clinical findings with a review of the literature. Arch Intern Med. 1989
Nov; 149(11): 2486-92.
DSI
Overall, data quality appears acceptable. Four sites were inspected. Dr. Ni notes
that various minor violations were observed in site (b) (4) [ ] I don't think
these violations had any signil??(b) (4) [ ] ??udy. The most worrisome problem
occurred in site [ ] In addition to some minor violations, there were
discrepancies between data listings and data reported in diaries/electronic CRF
(data were entered correctly in the electronic CRF from diaries) in most patients
for the pain (b) (4) [ ] various time points. The issue was however resolved.
Following a [ ] Pozen identified the origin of the discrepancies as of an error in
the production of the data listings generated for purposes of the FDA audit. Pozen
also stated that the SAS transport files (SAS data sets) exactly matched the data
in the electronic CRFs and that NDA data listings were correct (I do not believe
that th(b) (4) [ ] ??ificant deficiencies were noted in site [ ] Finally, for site
(b) (4) [ ] ?? Pozen terminated the study based on findings from monitoring visits
that the study coordinator changed the dates of administration of study drugs to
reflect a date which would allow the follow- up visit to fall within the protocol
required one to three day window. There were also issues with drug accountability
records in this site. Dr. Prohaska reviewed the results with and without the data
obtained from (b) (4) [ ] site and found no difference.
Drug name - Labeling
Kimberly Culley, from the Division of Medical Errors and Technical Services
(DMETS) reviewed the drug name, carton, and container. DMETS does not recommend
the use of the proprietary name, Myzan. The primary concerns relate to look-alike
and sound-alike confusion with Zyban. Additionally, DMETS reviewed the container
labels, carton and insert labeling from a safety perspective. DMETS has identified
several areas of possible improvement which I will not discuss further in this
review cycle, because I am recommending a non approval action. Jeanine Best
(Division of Surveillance, Research, and Communication Support) reviewed the
patient insert. Ms. Best made a number of recommendations for changes: she
simplified the wording, made it consistent with the PI, removed unnecessary
information, and put it in the DSRC recommended format. Since I am recommending a
non-approval action, I did not implement the changes in this review cycle.
Reviewers of various disciplines made labeling recommendations. Since I am
recommending a non-approval action, I did not implement these labeling
recommendations. They can be found in the discipline reviews.
Conclusions and Recommendation
I recommend a non-approvable action. There are a number of issues with this
product.
First, Pozen has not met the combination policy requirement for MT100, and has not
established that both active drug components make a contribution to the claimed
effects of the product. In addition, the (non statistically significant) effect
size of the combination over naproxen is clinically marginal (4-6% for sustained
pain relief, no benefit for 2-hour outcomes typically used in migraine trials, and
no sustained effect on nausea, photophobia and phonophobia).
Second, Pozen has not established the efficacy of MT100 in the acute treatment of
migraine (only efficacy in treating headache pain was established). I note however
that Pozen has an almost positive study for that effect (Study 306).
Third, there are safety concerns associated with the chronic use of
metoclopramide. Metoclopramide is well known to induce not only acute dystonias,
but also tardive dyskinesia, which can be permanent in some cases. Current
labeling of metoclopramide and many papers in the literature recommend against a
chronic use of the drug. Even though chronic use can be discouraged by labeling,
some degree of chronic use is expected with patients or practitioners non
compliant to the labeling.
Fourth, there are concerns about carcinogenicity associated with chronic use of
metoclopramide. While there are a number of factors to consider in the
interpretation of the carcinogenicity findings, this remains an important
observation to integrate in the risk/benefit analysis of the product.
To support approval of MT 00, I suggest the sponsor provides the following:
1. One positive factorial study with a clinically significant effect size. It will
be important that this study is not overpowered.
2. One positive study showing effect on all migraine key symptoms at 2 hours. Of
note, the factorial and the efficacy study could be combined into one single
study, with the proper analysis plan.
3. A justification of the risk/benefit of MT100 in light of the extrapyramidal
side effects associated to metoclopramide use, and to the carcinogenicity seen in
the rat. In that sense, the clinical effect must be robust. In my opinion, even if
statistical significance had been reached in all factorial and efficacy studies
reported in this NDA, the effect sizes observed do not outweigh the safety
concerns.
TARVER, et al, v. WYETH INC et al.
