US Senate Committee Report January 2010
| COMMITTEE ON FINANCE |
||
| MAX BAUCUS, Montana, Chairman |
||
| JOHN D. ROCKEFELLER IV, West Virginia |
CHUCK GRASSLEY, Iowa |
|
| KENT CONRAD, North Dakota |
ORRIN G. HATCH, Utah |
|
| JEFF BINGAMAN, New Mexico |
OLYMPIA J. SNOWE, Maine |
|
| JOHN F. KERRY, Massachusetts |
JON KYL, Arizona |
|
| BLANCHE L. LINCOLN, Arkansas |
JIM BUNNING, Kentucky |
|
| RON WYDEN, Oregon |
IKE CRAPO, Idaho |
|
| CHARLES E. SCHUMER, New York |
PAT ROBERTS, Kansas |
|
| DEBBIE STABENOW, Michigan |
JOHN ENSIGN, Nevada |
|
| MARIA CANTWELL, Washington |
MICHAEL B. ENZI, Wyoming |
|
| BILL NELSON, Florida |
JOHN CORNYN, Texas |
|
| ROBERT MENENDEZ, New Jersey |
||
| THOMAS R. CARPER, Delaware |
||
| RUSSELL SULLIVAN, Staff Director |
||
| KOLAN DAVIS, Republican Staff Director and Chief Counsel |
||
(II)
| CONTENTS |
|
Page |
|
Executive Summary ……………………………………………………………………………………. 1 Introduction ……………………………………………………………………………………………….. 1 Response to the Nissen Study …………………………………………………………………….. 4 Leaked Manuscript and a Scrambled Defense ……………………………………………… 5 The RECORD Trial as a Marketing Tool for Competition …………………………….. 11 Concerns About Avandia Raised Prior to 2007 …………………………………………….. 13 Conclusion …………………………………………………………………………………………………. 14 Appendix I: Visual Timeline of Public and Internal Information …………………… 17 Appendix II: Timeline …………………………………………………………………………………. 19 Appendix III: Relevant Definitions ……………………………………………………………… 21 Appendix IV: Documents Not Publicly Available ………………………………………….. 25 |
(III)
EXECUTIVE
SUMMARY
This
staff report
was developed over the last 2 years by U.S. Senate
Committee on Finance investigators who reviewed over 250,000 pages
of documents provided by GlaxoSmithKline (GSK/the Company), the
Food and Drug Administration (FDA), the University of North
Carolina, and others. Committee investigators also conducted
numerous interviews and phone calls with GSK, the FDA, and
anonymous whistleblowers.
Committee
staff began this investigation in May 2007 after a study was
published in the
New England Journal of Medicine, showing
a link between the diabetes drug Avandia (rosiglitazone) and
heart attacks. However, the reviewed evidence suggests that GSK
knew for several years prior to this study that there were possible
cardiac risks associated with Avandia. As a result, it can be argued
that GSK had a duty to warn patients and the FDA of the Company’s
concerns. Instead, GSK executives attempted to intimidate
independent physicians, focused on strategies to minimize or
misrepresent
findings that Avandia may increase cardiovascular risk,
and sought ways to downplay findings that a competing drug might
reduce cardiovascular risk.
When
an independent scientist sought to publish a study in 2007 pointing
out the cardiovascular risk of Avandia, GSK acquired a leaked copy of
that study from one of its consultants prior to the study
being published. The company’s own experts analyzed the study,
found it to be statistically reliable, and then attacked the
soundness of that study in press releases and public comments. GSK
also sought to counter the study’s findings by quickly releasing
preliminary results from its own study on Avandia, even though
the company’s internal communications established that its study
was not primarily designed to answer questions about cardiovascular
risk.
INTRODUCTION
For
the past 4
years, the staff of the Senate Committee on Finance
(Committee) has been examining allegations that pharmaceutical
companies attempt to manipulate science to improve the marketability
of drugs, potentially at the expense of public safety. These
allegations include intimidating scientists, ghostwriting studies for
academic researchers, suppressing studies that may show
that a drug could be dangerous, and selecting data to publish results
that favor one product over another.
In
November 2007, the Committee reported on the intimidation of
Dr. John Buse, a professor of medicine at the University of
(1)
2
North
Carolina (UNC) who specializes in diabetes.1
Based partly on internal documents from GSK, the Committee reported
on what appeared
to be an orchestrated plan by GSK to stifle the opinion of
Dr. Buse in 1999. At that time, Dr. Buse argued at several medical
conferences and in letters to the FDA that GSK’s diabetes drug
Avandia
may cause cardiovascular problems.2
According
to GSK emails made available to the Committee, GSK executives
labeled Dr. Buse a ”renegade” and silenced his concerns about
Avandia by complaining to his superiors at UNC and threatening a
lawsuit. The call to Dr. Buse’s superiors was made by Dr. Tachi
Yamada, then GSK’s head of research. In discussions with Committee
investigators, Dr. Yamada denied that his call was meant
to intimidate Dr. Buse. Instead, Dr. Yamada argued that he had made
the call to determine if Dr. Buse was making legitimate statements
or if he was possibly on the payroll of a GSK rival.
Dr.
Yamada also made a call to the University of Pennsylvania (Penn)
regarding two physicians who were about to publish a case study that
Avandia may have caused liver problems in one of their patients.3
Committee investigators contacted the two Penn physicians. Both
physicians chose to remain anonymous because of concerns
about possible retaliation by pharmaceutical companies.4
In
hindsight, both physicians agree that Avandia probably does not
cause liver problems. However, in 1999 Avandia was a new drug and the
two physicians wanted to publish a report on their patient who had
liver failure while on Avandia. Both physicians also said that the
calls placed by GSK officials, including Dr. Yamada, were highly
unprofessional and had a chilling effect on their
professional activity.5
Commenting
on the calls by GSK, one of the two physicians told Committee
investigators, ”It was really ridiculous. It was a case report
and I had no intention of bringing down GSK. I just wanted people
to know.” The physician added, ”It left a really bad taste in my
mouth. After that happened, I said that I would never work for
a
drug company.”
6
Also
commenting on the calls from GSK, the other physician told Committee
investigators, ”I have never encountered anything like this
in my career. I don’t even know how [GSK] knew that we were
publishing. It’s the kind of thing you imagine happening on TV.”
7
1
Committee Staff Report to the Chairman and Ranking Member, Committee
on Finance, United
States Senate, November 2007, ”The Intimidation of Dr. John Buse and
the Diabetes
Drug
Avandia.”
2
Id.
3
According to GSK internal emails, Dr. Yamada placed a call to senior
officials at the University
of Pennsylvania
Medical School after receiving the following email on August 4, 1999,
from
a
GSK executive:
Tachi,
I need you to
place another call to your contacts at U Penn. The situation is that
Dr.
NAME REDACTED is apparently on the Takeda speaker’s circuit. He is
reported to be speaking about the case and implicating Avandia.
Obviously, this is not in anyone’s best interest.
The
following day,
Dr. Yamada responded:
What
exactly do you
want me [sic] ask for? Obviously, we are not going to be able to pre-
vent
Dr. NAME REDACTED from speaking on behalf of Takeda. I would be happy
to speak with
either NAME REDACTED (Dept. Chair) or NAME REDACTED (Hepatology
Chief) but we
need to be clear on the message we want to send.
4
Staff interviews, December 2007. 5
Id. 6
Id. 7
Id.
3
In
an interview with
Committee investigators, Dr. Yamada stated
that he had no intention of intimidating the two physicians at Penn,
and that he had merely placed the call because he was concerned
that Avandia may cause liver problems.
In
a December 2007 floor speech, Senator Grassley revealed that Dr.
Steve Haffner, a professor of medicine at the University of Texas
Health Sciences Center, San Antonio, and a consultant for GSK, leaked
to GSK the draft of a study critical of Avandia that was to appear in
the
New England Journal of Medicine (NEJM).8
Dr.
Haffner was entrusted with a confidential copy of the manuscript
draft because he was peer-reviewing the study for the NEJM.
The study’s lead author, Dr. Steven Nissen, professor of cardiology
at the Cleveland Clinic, found that Avandia was associated
with a 43-percent increased risk of heart attacks, one of the main
health outcomes physicians hoped to avoid by treating diabetic
patients with medication.9
According
to documents produced by GSK, the leaked manuscript was
widely disseminated within the Company, and allowed GSK to launch
a public relations plan to protect Avandia, a multi-billion dollar
product.10
The Committee staff reviewed documents showing that
over 40 executives at GSK received and/or learned of the results
in the leaked study, including then CEO Dr. Jean-Pierre Garnier;
head of research, Dr. Moncef Slaoui; Vice President of Corporate
Media Relations, Nancy Pekarek; and GSK Senior Advisor,
Sir Colin Dollery.11
Before
Dr. Nissen’s study on Avandia was published, GSK’s statistical
experts were examining the study for potential flaws. In addition,
GSK officials were drafting ”key messages” to undermine the
main conclusion of the Nissen study. GSK had already published
several large trials on Avandia (rosiglitazone) including studies
named ADOPT and DREAM. After Nissen’s study was published,
GSK began publicly referencing those trials, as well as another
trial called RECORD, in what appeared to be an effort to further
repudiate any link between Avandia and heart attacks. RECORD
is a study GSK had been conducting for several years. GSK later
published the interim results of the RECORD trial in what
appeared to be an attempt to cast doubt on Nissen’s results.
However,
internal GSK emails indicate that GSK executives, not the study’s
independent steering committee, made the final decision to publish
the RECORD trial results. Further, based on a review of emails, it
can be argued that the authors of the RECORD trial
appeared more concerned about countering claims that Avandia
may be associated with heart attacks, than in trying to understand
the underlying science. While circulating a draft of a manuscript
on the RECORD trial, one of the authors wrote to his
8
Stephanie Saul, ”Doctor accused of leak to drug maker,”
The New York Times,
January 30, 2008.
9
Steven E. Nissen et. al. ”Effect of Rosiglitazone on the Risk of
Myocardial Infarction and Death
from Cardiovascular Causes” the
New England Journal of Medicine,
May 21, 2007.
10In
2006, global sales of Avandia reached nearly $3.4 billion. Citation:
Gardiner Harris, ”Report
Backs Up Warnings About Drug Avandia,”
The New York Times,
July 27, 2007.
11
Letter from Daniel F. Donovan III, counsel to GSK, to Senator
Grassley, dated May 23, 2008.
4
colleagues,
”[W]hat’s to stop [Nissen] adding the events from
RECORD
to his meta-analysis and re-enforcing his view?”
12
Further,
after the authors of the RECORD study submitted their paper
to the
NEJM,
one of the peer reviewers and several of the NEJM
editors replied, ”an explanation for the continued use of
[Avandia]
is needed in this manuscript.”
13
Committee
investigators also learned that GSK was aware since at least 2004
that the RECORD trial was statistically inadequate, or
”underpowered”
14
to answer questions regarding cardiovascular safety. Such
”inconclusive” results could be favorable to GSK and the
marketing strategy for Avandia. Further, experts were advising GSK
since 2004 about the possible biological mechanisms related to
why Avandia may cause an increased risk for heart attacks. However,
GSK appeared eager to design studies to prove that Avandia
was safer than its competitor ACTOS (pioglitazone), which is
manufactured by Takeda.
At
a July 30, 2007, safety panel on Avandia, Food and Drug Ad-
ministration (FDA) scientists presented an analysis estimating that
Avandia
use was associated with approximately 83,000 excess heart
attacks since the drug came on the market.15
Had GSK considered
Avandia’s potential increased cardiovascular risk more seriously
when the issue was first raised in 1999 by Dr. Buse, as well as
by some of their own consultants in later years, some of these heart
attacks may have been avoided.
RESPONSE
TO THE
NISSEN STUDY
In
March 2007, GSK
held a meeting with company officials and
academic
advisors to discuss several studies on Avandia and its cardiac
risks and benefits.16
Several presentations were made about
studies on Avandia’s possible cardiac risk. During the discussion
of a GSK meta-analysis (integrated study) and a study GSK
commissioned
by Ingenix, GSK noted that the academic advisors
stated
the
following:
Dr.
NAME REDACTED
commented that the [cardiovascular]
effect seen in the Integrated Clinical Trials Analyses
with rosiglitazone was small but real, and that it is counter
to the proposed [cardiovascular] benefits associated with Avandia.
Dr. NAME REDACTED agreed, noted that
all data point to rosiglitazone having a hazard ratio greater than
unity. . . . Dr. NAME REDACTED summarized
the discussion on the Integrated Clinical Trials data by
stating that rosiglitazone causes weight gain and
edema,
leading to a greater number of events.17
12
Email 13
Letter
1,
2007.
from
John McMurray
to Nigel Jones et al., dated May 29, 2007.
from
the
New England Journal of Medicine
to Philip H. Home, M.D., dated June
14
A study is underpowered if it does not meet the statistical
requirements to adequately measure
a medical outcome or study endpoint.
15
FDA, ”Assessment of the cardiovascular risks and health benefits of
rosiglitazone,” presented
July 30, 2007. Estimate presented publicly at FDA advisory committee
meeting.
16
Internal GSK report, ”GSK Diabetes Franchise Cardiology Advisory
Board,” meeting held March 1-2, 2007, report dated March 16, 2007.
17
Id.
5
Moreover,
during the discussion of the DREAM
18
trial, a cardiologist
from Stanford
stated:
[T]he
diabetes
prevention afforded by rosiglitazone was
very
impressive, but there was no cardioprotective benefit. He
then asked what the point of diabetes prevention is
if there is no cardiovascular benefit.19
[Emphasis
added]
When
discussing ADOPT,20
the academic advisors concluded
that,
”The data in ADOPT and DREAM as well as in the CV Clinical Trials
are consistent in indicating a signal for heart failure and ischemic
events.” According to GSK interal documents, GSK’s experts
were discussing problems with DREAM as early as 2006.21
Around
this same time, Dr. Steven Nissen began studying the potential
cardiac risks of Avandia, by reviewing data found in previously
published studies. He placed several requests to GSK asking
for patient level data on several studies published about Avandia.
However, GSK would provide the requested data only if Dr. Nissen
agreed to use a GSK statistician for the analysis.22
Dr. Nissen refused to use the Company’s statistician, citing a need
to
maintain
independence.23
On
May 2, 2007, Dr. Nissen submitted an analysis of 42 published
and unpublished clinical trials on Avandia to the
NEJM
for peer review and publication.
NEJM
then sent confidential copies of the study to several independent
experts, including Dr. Steve Haffner, to peer review the Nissen
study. According to
NEJM,
peer reviewers must acknowledge in writing that the material they
are reviewing is confidential, not to be shared with others, and is
to be destroyed or returned to the medical journal after a review is
com-
pleted.24
However,
the very next day, May 3, 2007, Dr. Haffner faxed Dr. Nissen’s
unpublished study to a GSK executive. Dr. Haffner wrote
”confidential”
on the fax cover sheet and checked a box marked ”urgent.”
25
LEAKED
MANUSCRIPT
AND A SCRAMBLED DEFENSE
One
day after
receiving the unpublished study from Dr. Haffner,
GSK
produced a
detailed, 8-page analysis of Dr. Nissen’s paper,
18
DREAM is an acronym for ”The Diabetes Reduction Assessment with
Ramipril and Rosiglitazone
Medication.” DREAM is an international, multi-center, randomized,
double-blind diabetes
trial involving 5,269 patients from 21 countries. The DREAM study was
conducted by the Population Health Research Institute and published
in the middle of 2006.
19
Internal GSK report, ”GSK Diabetes Franchise Cardiology Advisory
Board,” meeting held March 1-2, 2007, report dated March 16, 2007.
20
ADOPT is an acronym for ”A Diabetes Outcome Progression Trial.”
ADOPT is a randomized,
double-blind, parallel-group study conducted on
∼3,600
drug-naive patients designed to
measure
the efficacy
of rosiglitazone in controlling the glycemic levels of Type 2
diabetes patients.
21
Internal GSK slide show, ”DREAM: Results of the Rampiril Arm,”
undated but several slides
state ”updated Sept 6/06.” One particular slide titled, ”DREAM vs.
Previous Trials,” notes that
DREAM ”was low power to detect differences in CVD events (short
duration, low risk participants).” The summary and conclusions
slide on DREAM finds that the study had ”too few
events
to draw any
conclusion re the effect on other CV events or death.”
22
Internal GSK emails, dated May 3, 2007. ”I have made oit [sic] clear
in my letter of Feb 26 [to Dr. Nissen] that analyses should be
conducted by GSK personnel pursuant to prospectively
agreed
analyses plan.”
23
Multiple staff discussions with Dr. Steven Nissen, from June 2007 to
the present. 24
Email from
NEJM
editor to Committee staff, dated December 18, 2007. 25
Steven Haffner, fax to Alex Cobitz, dated May 3, 2007.
6
weeks
before the paper’s public release.26
The GSK statistician attempted
to find deficiencies in Nissen’s meta-analysis but noted, ”The
selection of trials therefore appears to be thorough, though others
more familiar with the trials can comment more knowledgeably.”
27
The
GSK statistician also performed a regression analysis
28
on each
study that Dr. Nissen used in his meta-analysis to see if the effects
of myocardial infarction and/or cardiovascular death would still
appear. The statistician stated, ”These results are very similar to
the conclusion from the [Nissen] paper using the Peto method.29
As
such there is no statistical reason for disregarding the findings
as
presented.”
30
The
GSK statistical analysis was circulated to senior executives within
GSK. These executives then discussed several large trials, such
as RECORD, DREAM and ADOPT that GSK could use to combat
Dr. Nissen’s analysis. RECORD was an ongoing trial that had
not been published. On the other hand, DREAM and ADOPT were
published and were included in Dr. Nissen’s analysis. GSK, as well as
the FDA, had also performed their own meta-analyses.
Both
meta-analyses were consistent with Dr. Nissen’s results.31
On
May 8, 2007, Dr. Moncef Slaoui, head of research at GSK, wrote an
email to several company executives.32
Commenting on the
meta-analyses,
he wrote:
—FDA,
Nissen and GSK all come to comparable conclusions
regarding
increased risk for ischemic events, ranging
from
30 percent to
43 percent!
—FDA
and Nissen (but no final data from GSK [to] date)
reach
the conclusion
of an [hazard ratio] for death (CHF
+
IHD) of 1.72 or 1.75!
33
Dr.
Slaoui also
noted in this email that a GSK commissioned
study
by Ingenix did not find any significant problems with rosiglitazone.
Ingenix had performed an epidemiological study of Avandia.
While medical experts place greater importance on a clinical
trial over an epidemiological study, Dr. Slaoui sought to highlight
the Ingenix results. He also expressed concern that a beneficial
effect was observed (6 to 16 percent) in the PROactive
34
study
of ACTOS in high-risk cardiovascular disease patients.35
26
Internal GSK document, ”Report on the article by SE Nissen & K
Wolski ‘Effect of rosiglitazone on the risk of myocardial
infarction and cardiovascular death.’ ” Research Statistic Unit,
GSK, DRAFT May 4, 2007.
27
Id.
28
A statistical method that allows data to be simultaneously adjusted
for differences in the
distribution
of a wide variety of measured risk factors that may exist between
patients in a study
treated with one therapy compared to those treated with another or
with placebo.
29
Peto Method is a widely used way of combining odds ratios in
meta-analysis.
30
Internal GSK document, ”Report on the article by SE Nissen & K
Wolski ‘Effect of rosiglitazone
on the risk of myocardial infarction and cardiovascular death.’ ”
Research Statistic Unit,
GSK, DRAFT May 4, 2007.
31
Internal GSK email from Moncef Slaoui to multiple GSK executives,
dated May 8, 2007. 32
Id. 33
Id.
34
PROactive—”PROspective
PioglitAzone Clinical Trial In MacroVascular Events Study.” The
PROactive
Study was initiated as a randomized, double-blind, placebo-controlled
cardiovascular outcome study to determine the effects of pioglitazone
on reducing the risk of a wide variety of cardiovascular events as
well as to determine its ability to control blood glucose levels of
patients with Type 2 Diabetes. The study was commissioned by Takeda
pharmaceuticals, a company
that competes directly with GSK and produces a similar diabetes
medication called ACTOS.
35
Internal GSK email from Moncef Slaoui to multiple GSK executives,
dated May 8, 2007.
7
Dr.
Slaoui asked,
”How can we reinforce the value of the
[Ingenix]
study? The FDA criticizes the fact that we excluded cases of
sudden cardiac death.”
36
He then asked his team to strategize
further
on the
issue:
[W]hat
studies could
we offer the FDA to further assess
the
contradictory data between the integrated study and the
two others? can we expand Record? Propose something else
(very high risk patients? ok? ethical?), compare to
Actos
for superiority on some end points?
37
By
May 9, 2007, GSK
began drafting ”key messages” to counteract
the findings of the Nissen study.38
In an email, GSK’s Vice President
for Corporate Media Relations noted, ”The Nissen analysis
is one way of looking at the data, but it doesn’t reflect all we know
about the safety of this medicine. . . . [W]e are not seeing a
proven link between Avandia and increased cardiovascular
deaths.
. . .”
39
On
May 9, 2007, Sir Colin Dollery, a senior consultant to GSK, laid
out many of the problems with Avandia in an email to Dr.
Slaoui
and others.
He wrote:
To
a great extent,
the numbers are the numbers, the [Nissen]
analysis is very similar to our own. . . . We cannot undermine
the numbers but I think they can be explained so
we
must concentrate on effective risk management.40
Later
in the email,
Sir Dollery noted that the PROactive study
on
ACTOS
(pioglitazone) is undermining Avandia (rosiglitazone).
He
wrote:
The
main argument
here lies in that pioglitazone [ACTOS]
causes
a small reduction of LDL [Low-Density Lipoprotein] and rosiglitazone
causes a small elevation. . . . [W]e should search
for evidence that the use of statins in diabetics generally and
with rosiglitazone in particular has risen steeply
over the time the thiazolidenediones have been on the market. We can
then argue that any problem that existed with
LDL is now controlled or controllable. It would also be
worth obtaining the evidence that the use of antihypertensives
in diabetics has also been increasing rap-
idly.41
On
fluid retention
and links with cardiovascular disease, Sir
Dollery
mentioned a
possible mechanism to explain how Avandia
may
cause heart
attacks. He wrote:
If
[fluid retention
is] substantial in patients with an impaired
myocardium it can lead to [cardiac heart failure] and to cardiac
ischemia by decreasing myocardial efficiency
in the face of existing coronary disease. . . . If there is
criticism of GSK it might be that we were a bit slow off
36
Id. 37
Id.
38
Internal GSK email from VP Corporate Media Relations, US
GlaxoSmithKline, dated May
9,
2007.
39
Id.
40
Internal GSK email from Colin Dollery to Moncef Slaoui and other GSK
officials, dated May
9,
2007.
41
Id.
8
the
[mark] in making
firm recommendations about the use
of
diurectics . . .42
and recognizing that the sodium retention
is mediated via distal renal tubular ENaC.43
On
May 21, 2007,
NEJM
published online Dr. Nissen’s metaanalysis
that found a link between Avandia and heart attacks. That
same day, GSK responded, ”GSK strongly disagrees with the
conclusions
reached in the
NEJM
article, which are based on incomplete evidence and a methodology
that the author admits has significant
limitations.”
44
Instead, GSK highlighted the results of company sponsored trials like
RECORD as ”the most scientifically rigorous way to examine the
safety and benefits of a medicine.”
45
In
a subsequent letter to
The Lancet,
GSK maintained that the RECORD
trial is ”compelling evidence” for the safety of Avandia.46
On
May 23, 2007, a GSK official emailed members of the RECORD
steering committee, the group of independent academics overseeing the
study, to alert them of a teleconference to be held the
following day.47
GSK officials also emailed internal talking points
to help guide their discussion with the steering committee. However,
it appears that prior to receiving input from the steering committee,
GSK had already decided to publish the RECORD results.
Later that same day, a GSK official wrote, ”. . . we’ve decided
to
disclose the results. . . .”
48
The
following day, GSK officials discussed potential problems if the
academics on the RECORD steering committee raised concerns about
publishing the interim results of the RECORD trial.49
In an
email,
one GSK
official wrote:
[I]f
the Steering Committee [SC] are reluctant to publish—
Frank
and I will argue the case that there is a balance to be drawn between
very negative press coverage and specific
reassurance for the patients in the study. However if the
SC believe that publishing interim data will fatally damage their
ability to bring the study to a completion—
Frank and I will bring that opinion with reasons back to GSK, before
pursuing the line—that
a decision has been
made—live
with it.50
42
Diuretics are blood pressure medications that cause the body to
excrete water and sodium (salt).
43
Internal GSK email from Colin Dollery to Moncef Slaoui and other GSK
officials, dated May 9,
2007.
44
GlaxoSmithKline press release, ”GlaxoSmithKline responds to NEJM
article on Avandia,” published online May 21, 2007.
45
Id.
46
Ronald Krall M.D., Chief Medical Officer, GlaxoSmithKline,
”Cardiovascular Safety of
Rosiglitazone,”
The Lancet,
letter published online May 30, 2007. ”The most compelling evidence
comes
from RECORD (Rosiglitazone Evaluated for Cardiac Outcomes and
Regulation of glycaemia in Diabetes), an open-label, 6-year,
cardiovascular outcomes trial (with prospectively defined
cardiovascular
endpoints) in 4458 patients that started in 2000.”
47
Email to GSK officials and RECORD steering committee, dated May 23,
2007. 48
Internal GSK email, dated May 23, 2007. 49
Internal GSK email, dated May 24, 2007. 50
Id.
9
A
few hours after
this email, the acting chair of the RECORD
steering
committee, contacted the
NEJM
to inquire about publishing the interim results.51
The editor of the
NEJM
responded
that
the journal would be interested in publishing the study.52
By
May 29, 2007, several authors of the RECORD study began passing
around a manuscript, discussing the results, and offering suggestions
for improvement. The third author on the RECORD study wrote, ”We do
not find more myocardial infarctions with rosiglitazone treatment,
but again there is a tendency supporting the Nissen argument. It is
important to stress that it does not affect
cardiovascular death.”
53
That
same day, a
senior author of the RECORD study, wrote:
There
are several
striking issues:
(1)
The HR ratio
(and 95 percent CI) for MI in RECORD
is
not inconsistent with Nissen’s—and
he had more events; what’s to stop him adding the events from RECORD
to his meta-analysis and re-enforcing his view? . . .
(2)
Same is for CV
death, although the number of events
in
RECORD and in the
meta-analysis are similar and at
least
in RECORD the
HR is in the other direction!
(3)
Manuscript looks
to downplay the 239 percent INCREASE
in HF. I have
taken the liberty of doing some rewording.54
Once
a study is
submitted to a journal, the journal editors then
send
the article to several experts for peer-review. After the review, the
editors send the peer-review comments back to the author. On June
1, 2007, the RECORD authors received a reply from
NEJM regarding
their earlier submitted manuscript. The
NEJM
editors summarized
the issues presented by all 8 peer reviewers, many of
whom
were highly critical of the study in their reply.55
Reviewer
A, along with other reviewers, asked that the authors ”modify the
language in multiple locations in the manuscript to tone
down your conclusions.”
56
The editor also noted, ”[I]n the opinion
of all the readers, the data that you present are completely
compatible with the results of the meta-analysis by Nissen and the
meta-analysis for myocardial ischemic events posted on the GSK
Web
site.”
57
Regarding
the comments of Reviewer B, the editors wrote that for
myocardial infarction the ”estimates in the RECORD trial and the
Nissen meta-analysis” overlap in their confidence intervals, meaning
that they found a similar trend for heart attacks.58
They continued,
”The editors feel strongly that your data do not support
51
Email from Acting Chair of the RECORD trial to Editor at
NEJM,
dated May 24, 2007. The
Acting Chair wrote, ”We the Steering Committee of the RECORD Study
would like to submit a brief report of the current interim findings
of this ongoing trial concerning the key cardiovascular
outcomes.”
52
Email from Editor at
NEJM
to the acting chair of RECORD trial, dated May 24, 2007. 53
Email between members of the RECORD trial, dated May 29, 2007. 54
Email between members of the RECORD trial, dated May 29, 2007.
55
Letter from the
New England Journal of Medicine
to Philip H. Home, M.D. dated June 1,
2007.
56
Id. 57
Id. 58
Id.
10
the
statement that
the RECORD results for MI contradict the Nissen
meta-analysis; this statement must be removed or modified.”59
Reviewer
C noted that the RECORD trial is not blinded,60
and pointed
out ”the serious problem of the low event rate, especially for
MI events, in this study.”
61
He continued to ask, ”Do you have an
explanation for the very low event rate?” This reviewer also noted
the ”need to greatly tone down your language to reflect the
substantial
level of uncertainty in the data.”
62
Reviewer
D
questioned the need for keeping rosiglitazone on the market. ”The
editors also agree that an explanation for the continued
use of rosiglitazone is needed in this manuscript.”
63
The
NEJM
published the interim analysis of the RECORD study on
July 5, 2007. The GSK study authors concluded that the data was
”insufficient” to find a link between Avandia and heart attacks.64
However,
an editorial by the
NEJM
questioned the RECORD study, as well as several of GSK’s studies of
Avandia such as DREAM and ADOPT. The authors of the editorial wrote,
”The DREAM
trial and ADOPT focused largely on marketing questions and
failed to address questions of myocardial infarction-related risk or
benefit directly.” In addition, the editorial noted that the RECORD
trial had ”several weaknesses in design and conduct” including
a lack of blinding when treatment was assigned. The authors
also pointed out that events of myocardial infarction would have been
a preferred clinical endpoint for the study. Studies are normally
designed to evaluate certain clinical endpoints or disease symptoms
such as heart attack, tumor size, or depression. The authors also
added that the RECORD study was not powered (or designed)
to detect a myocardial infarction as an endpoint.65
On
June 6, 2007, the House of Representatives Committee on Oversight
and Government Reform held a hearing on Avandia. Despite
mounting criticism of the RECORD trial, Dr. Slaoui again highlighted
the study in his sworn testimony. ”I will say that we found
the RECORD data which we published yesterday in the
New England Journal of Medicine
very reassuring, recognizing that it
is
interim and therefore not fully conclusive.”
66
That
same day, GSK dismissed the idea that Dr. Nissen’s study spurred
the publication of the RECORD interim results. Instead, the
Company placed blame on the media. In talking points created for
its sales force, GSK stated, ”Because of the widespread media
59
Id.
60
A blinded study is a study done in such a way that the patients or
subjects do not know
what
treatment they are receiving to ensure that the results are not
affected by a bias on the part
of patients, doctors, or the sponsors who are paying for the study.
61
Letter from the
New England Journal of Medicine
to Philip H. Home, M.D. dated June 1, 2007.
62
Id. 63
Id.
64
Philip D. Home et al., ”Rosiglitazone Evaluated for Cardivascular
Outcomes—An
Interim
Analysis,”
the
New England Journal of Medicine,
July 5, 2007. The study authors concluded, ”Rosiglitazone
was associated with an increased risk of heart failure. The data were
insufficient to
determine whether the drug was associated with an increase in the
risk of myocardial infarction.”
65Bruce
M Psaty, et al. ”The Record on Rosiglitazone and the Risk of
Myocardial Infarction,” the
New England Journal of Medicine,
July 5, 2007.
66
House of Representatives, Committee on Oversight and Government
Reform, ”Hearing on FDA’s
Role on the Evaluation of Avandia’s Safety,” June 6, 2007,
preliminary transcript, page 168 (see
also http://oversight.house.gov/documents/20071114160344.pdf
).
11
coverage
of the
NEJM
[Nissen] meta-analysis and the confusion it
has
created, the RECORD Steering Committee decided it was important
to publish the interim analysis in the interests of patient
safety.”
67
Regarding
its
competitor Takeda, which sells ACTOS, GSK advised its sales force
if asked questions about the PROactive study:
Please
do not
discuss Actos or the Proactive study with
your
physicians. For questions regarding Actos or the Proactive study,
healthcare providers should contact Takeda.
GSK’s focus is on Avandia. Communicate the key points
from the interim analysis of RECORD to your physicians.68
THE
RECORD TRIAL
AS A MARKETING
TOOL
FOR
COMPETITION
Despite
attempts to
highlight the RECORD study, it appears
that
GSK knew for years that the study was ”underpowered,” i.e., the
study did not provide sufficient data to test for cardiovascular
safety. And executives appeared more concerned about designing a
study
to limit competition from ACTOS. Such evidence can be found in a GSK
slide presentation, emails, and other documents created
in 2004 to 2006.
For
instance, in an undated slide show, apparently created in 2004,
GSK noted that RECORD does not have sufficient ”power.”
69 The
slide presentation also noted that GSK was trying to create studies
to counter the PROactive study on ACTOS that Takeda
planned
to release.70
Slide
number 6 titled, ”PROactive: Potential Impact,” noted that GSK’s
challenge was to ”maintain share in growing market over
next
2-3 years.”
71
Slide
number 8
reads:
Situation
Summary:
• We
have a gap
—In
2005 Actos will have some [cardiovascular] outcome data
• To
keep our share of the growing class
—Additive
benefit to RECORD of non-inferiority result
• However
this gap may be permanent
—RECORD
has a lower event rate than expected
PROPOSAL
Fill
this gap with
an outcome study reporting in 2007
Slide
number 10
compared the potential impact of a new GSK
study
to counter the marketing danger of PROactive and the potential
impact on sales in UK pounds in 2010. The slide reads: ”Timely CV
Outcomes data would more than fill the RECORD ‘potential gap’
and would have twice the impact on our sales than PRO-
67
GSK’s RECORD
Study Questions, dated June 6, 2007, for GSK Internal Use Only. 68
Id.
69
GSK Internal
Slide Show, ”European Commercial Need for a Post-ACS Study
Proposal,” undated.
70
Id. 71
Id.
12
active.”
72
The final slide pointed out that GSK should do a ”kick off
study only after review of results from Proactive in Sept 2005
and
assessing benefits/risks.”
73
A
second instance is found in a June 2005 email where GSK executives
discussed the need for a study to counter PROactive. In the
email, a GSK official wrote, ”Clearly no patients will be re-
cruited until [we] have made a decision based on the go-no go cri-
teria
from the PROactive data. However, there is a great deal of
EU
commercial push to initiate this study in 2005.”
74
A
third case is found in an internal GSK document outlining an upcoming
meeting for December 2004. Several points were discussed about
RECORD and PROactive. Regarding RECORD, the document
noted that RECORD has ”low events rates.” This means that the study
did not have the statistical ”power” to give sufficient
cardiovascular event data. The document also stated, ”PROactive
results
to be coming soon—need
to be able to respond to a variety of
different outcomes. Communications plan in place for various
possible
outcomes of PROactive.”
75
A
fourth instance is
found in a briefing document for a June 2005 meeting on Avandia’s
cardiovascular plan. The document
notes
several ”important limitations of RECORD.”
76
—the
study will not be available until 2009
—the
current observed rate for the primary endpoint is very
much
lower (approximately 3.5 percent per annum) than that anticipated
in the original protocol (11 percent per annum).77
A
fifth case is
found in another GSK email. On July 26, 2005,
GSK
officials began
emailing each other about potential problems with RECORD and how the
PROactive study by Takeda on ACTOS
will
create problems
for Avandia. One official wrote:
Ron
Krall [then GSK
Chief Medical Officer] has asked
Lawson
[unknown GSK executive] to provide an urgent update
to David Stout [then GSK President of Global Pharmaceutical
Operations] regarding RECORD. In particular
he has asked for our ”intent to manage information flow
in Europe to manage the competitive situation.” Clearly
we can provide a summary of the communications around
PROactive but I wonder if you could put a few sentences
together regarding the communications piece
around
RECORD.78
A
sixth incident is
documented in July 2005, when GSK officials
continued
expressing concerns about cardiovascular problems with Avandia and
potential problems arising from the PROactive study which
focused on positive findings with ACTOS. GSK held a meeting
on July 18, 2005 to discuss the need for a study to compete
72
Id. 73
Id.
74
Internal 75
Internal 76
Internal
GSK
email, dated
June 16, 2005.
GSK
document,
untitled, unknown date.
GSK
document,
”Briefing Document for 27 June 2005 PMB Avandia Cardiovascular
Modeling
Plan.”
77
Id.
78
Internal GSK email, dated July 26, 2005.
13
with
PROactive.79
The briefing document from this meeting discussed
the
”European Commercial Need” for a study:
A
recently completed
evidence gap analysis completed by
the
Metabolic Centre of Excellence has identified the need for
the rapid generation of clinical endpoint data to support
the superiority of rosiglitazone [Avandia] for the prevention of
future cardiovascular clinical events in patients with
[type 2 diabetes mellitus]. Publication of the PROactive
data may result in important commercial disadvantage in Europe. We
therefore have the opportunity to start a
CV outcomes study with the aim of getting superiority
data
in 2007.80
The
document also
noted that GSK’s studies provided insufficient
data
on
cardiovascular outcomes:
The
primary endpoint
in RECORD is powered for noninferiority
and taking into account the low observed event rate, it is unlikely
that this study will demonstrate any potential for [Avandia]
combination to be superior in terms of
the primary endpoint compared to SU+MET combination
therapy. DREAM and ADOPT are collecting CV safety data,
but these are low risk populations and it is unlikely that [Avandia]
will be superior to controls for the prevention
of CV events.81
CONCERNS
ABOUT
AVANDIA RAISED PRIOR TO 2007
In
June 2004, GSK’s
leader for a cardiac safety study called the
”Avandia
211 Cardiac Heart Failure Study”
82
reported on a meeting
with a consulting academic. The academic was the chairman of the
independent clinical endpoint committee for the Avandia 211 study.83
The study leader’s report of the academic consultant’s
feedback
on Avandia
211 follows:
With
regard to CV
mortality and morbidity data, [the academic
consultant] said that the results were ‘almost identical’ to the
results he had seen from a previous glitazone study
as a member of the DSMB with increased CV events,
hospitalizations, and ischaemic events. [The academic consultant]
said that he felt this was a class effect as a result of reduced
oxygen carrying capacity as a result
of
haemodilution to fluid-retention.84
The
report of the
Avandia 211 meeting noted that the academic
consultant
said he would not stop prescribing Avandia, as the study
was too small, and that he ”would continue to use [Avandia]
79
Internal GSK document, ”MDC Briefing Document: Ad-hoc meeting 18th
July 2005.
AVD104821:
rosiglitazone in post-acs patients.”
80
Id. 81
Id.
82
Internal GSK slide show titled ”Avandia 211 CHF study: Senior Review
of Additional Analysis,”
undated.
83
Internal GSK report, ”Avandia 211 CHF study—Review
of Study Results, Feedback from Professor
NAME REDACTED,” dated June 3, 2004.
84
Id.
14
as
a second or third
line therapy whilst taking appropriate precautions.”
85
Later
that month,
several GSK representatives met with the advisory
board for study protocol 211.86
The meeting notes state:
There
was
disappointment verbalized about the morbidity
and
mortality table that showed that there were ten ischemia-related
adverse events in the rosiglitazone group versus
five events in the placebo group. . . . Dr. NAME REDACTED
found [it] unusual that there was an increase in edema
and cardiac events despite the fact that there was significant
improvement in glycemic control in the rosiglitazone
arm of the trial. He thought the glycemic control and
pleitrophic [sic] effects of rosiglitazone would have predicted
a different outcome than what was observed.87
In
late 2005, GSK
published a draft retrospective analysis of cardiovascular
events in Avandia clinical trials discussing the underlying cause
for the increase in ischemia.88
In a section of the analysis
that examined myocardial ischemia, the authors mention a ”hypothesis
that small degrees of fluid retention may be an important
contributor to the development of worsening myocardial ischemia
in high risk patients.”
89
After
GSK reviewed the evidence found in this analysis, it appears
that the Company was aware of the potential cardiovascular risks
associated with Avandia in late 2004 or early 2005. In 2005, GSK
commissioned an ”observational” trial study that was conducted
in two parts: the first part in 2005 and the second in 2006. The
results of these studies support the further investigation of the
cardiovascular
risks associated with Avandia.
The
first study included 11,586 subjects randomly placed in clinical
trials before September 20, 2004. The analysis of the trials was
completed during the fall of 2005, giving a hazard ratio for myocar-
dial
ischemia of 1.29, meaning that rosiglitazone increased the risk of
heart-related ischemia by 29 percent. This number was statistically
significant.
GSK’s
second observational study involved analyzing 14,237 patients
by the summer of 2006. The results found a hazard ratio of 1.31,
meaning that Avandia increased the risk of myocardial ischemia
by 31 percent.90
CONCLUSION
In
preparing this
report, Committee investigators reviewed over
250,000
pages of
documents provided by GSK, the FDA, the University of North
Carolina, and others. Anonymous whistleblowers who contacted Senator
Grassley’s investigators provided hundreds of other pages. For well
over a year, Committee investigators also
85
Id.
86
GSK Internal Meeting Minutes, ”Summary of the feedback from the
Advisory Board Meeting
held on June
23rd, 2004, the Philadelphia Airport Marriot to discuss Study
Protocol 211.”
87
Id.
88
Internal GSK document titled, ”Rosiglitazone: Further Interim
Results from Retrospective
Analysis
of
Cardiovascular Events in Clinical Trials DRAFT,” undated.
89
Id.
90
GlaxoSmithKline, Studies ZM2005/00181/01 and
HM2006/00497/00/WEUSRTP866;
http://
ctr.gsk.co.uk/Summary/rosiglitazone/studylist.asp.
15
conducted
numerous
interviews and phone calls with GSK, the
FDA
and anonymous
whistleblowers.
The
totality of evidence suggests that GSK was aware of the possible
cardiac risks associated with Avandia years before such evidence
became public. Several years prior to Nissen’s study, it can be
argued that GSK was on notice that Avandia may have problems. Based
on this knowledge, GSK had a duty to sufficiently warn
patients and the FDA of its concerns in a timely manner. Instead,
GSK executives intimidated independent physicians, focused on
strategies to minimize findings that Avandia may increase cardiovascular
risk, and sought ways to downplay findings that the rival
drug ACTOS (pioglitazone) might reduce cardiovascular risk.
In
recent years, pharmaceutical companies have committed acts that
forced them to pay the largest criminal fines in American history.91
In cases involving Pfizer, Eli Lilly, Bristol Myers Squibb and
four other drug companies, these fines and penalties have totaled
over $7 billion since May 2004.92
In particular, Pfizer has been fined multiple times in the past 6
years for illegal off-label promotion
of their drugs. In its latest plea agreement, which took place last
September, Pfizer paid $2.3 billion in fines and penalties for
off-label promotion of Bextra. This settlement was the largest
criminal fine in U.S. history.93
Such an environment requires diligent
oversight by the FDA to protect the citizens of this country and
to ensure the safety of American medicine.
91
David Evans, ”Pfizer Broke the Law by Promoting Drugs for Unapproved
Uses,”
Bloomberg, November
9, 2009.
92
Id. 93
Id.
The full 342 pages can be read via this link: http://finance.senate.gov/press/Gpress/2010/prg022010a.pdf.
Reprint of web pages are only allowed with explicit permission. Please request our permission by emailing us with a complete description of the intended use.