Investigators Point to Unethical Reporting in Gabapentin Studies

Allison Gandey

November 20, 2009

November 20, 2009 — Internal company documents obtained during litigation suggest that Pfizer influenced outcomes in published trials of gabapentin, report researchers.

"There was a failure to publish everything that was known," Kay Dickersin, PhD, from the Center for Clinical Trials, Johns Hopkins Bloomberg School of Public Health in Baltimore, Maryland, said in an interview. "In some cases, predetermined outcomes in the original protocol were substituted for different ones depicting a more favorable result."

The anticonvulsant gabapentin, marketed under the brand name Neurontin, is reportedly one of Pfizer's best-selling drugs. It is listed among the 50 most-prescribed drugs in the United States. Analysts suggest the anticonvulsant made this list because off-label prescriptions account for a large portion of sales.

There was a failure to publish everything that was known.

Pfizer has been heavily criticized for off-label marketing of the drug. In 2004, the company plead guilty to criminal fraud and paid close to $430 million to settle charges that included defrauding Medicaid over illegal promotion of gabapentin. Further legal action is pending.

Using the documents made available by this case, investigators found inconsistencies between the original protocol and published results. "We are concerned that the reporting practices observed in our analysis do not meet the ethical standards for clinical research or maintain the integrity of scientific knowledge," write the researchers, led by Swaroop Vedula, MD, also from the Center for Clinical Trials, Johns Hopkins Bloomberg School of Public Health.

The report appears in the November 12 issue of the New England Journal of Medicine. It states that of the 21 studies initiated, only 12 were published, and those suffered from selective outcome reporting. "Fair and honest treatment of patients enrolled in clinical trials of any kind requires full, open, and unbiased reporting," write the researchers.

Among Top 50 Most Prescribed Drugs

In a statement emailed to Medscape Neurology, the company said, "The suggestion that Pfizer attempted to mislead the medical community about the effectiveness of gabapentin for certain off-label conditions is untrue."

Pfizer and others criticize the authors for working with plaintiffs' lawyers in litigation concerning gabapentin. Critics say because of these relationships, the work is tainted. "We believe the review suffers from significant bias, insufficient data, poor methodology, and cannot pass the threshold of credible scientific research," Pfizer said.

The problem is that the company documents are not publicly available. Researchers not involved in litigation do not have access to these records to conduct independent studies.

This is a wake-up call for the scientific community.

Dr. Dickersin said she asked that she be allowed to use the documents for research purposes as part of her agreement to testify. Dr. Dickersin then donated the funds she received from the proceedings to Johns Hopkins to support academic scholarship in the area of reporting biases.

During an interview, Dr. Dickersin said that her group's findings underwent rigorous peer review at the New England Journal of Medicine.

"This is a wake-up call for the scientific community," said Karen Johnston, MD, from the University of Virginia in Charlottesville. Dr. Johnston, a member of the American Academy of Neurology, added, "This isn't about Pfizer, 1 drug, or individual investigators. There is a larger problem and these findings are generalizable."

The study is the second this month to report that trial outcomes often change between original design and publication.

Problems in Publications

Reporting in the Annals of Family Medicine, investigators found that the primary outcomes of nearly a third of randomized trials — as published in 5 top journals — differed from those previously indicated in clinical trial registry listings (Ewart R, et al. Ann Fam Med. 2009;7:542-546).

The group, led by Robert Ewart, MD, from the Southern Illinois University School of Medicine in Springfield, calls for full disclosure and discussion to improve transparency in the performance and reporting of trials.

"When investigators find it necessary to make legitimate changes in a trial's protocol while studies are ongoing, these changes should be clearly stated in protocol amendments, clinical trial registry records, and the statistical analysis plan," Dr. Dickersin said.

"The published report should include a clear statement explaining the reason for the changes and their timing, potential biases that might have been introduced as a result, and the reasons for excluding outcomes from the analysis," she noted.

Dr. Dickersin suggests such transparency was lacking in all 12 of the published studies on gabapentin for off-label use. The investigators acknowledge that there are often good reasons to make changes to a trial's protocol. "There are some excellent investigators on these studies," Dr. Dickersin said. "Some may have done the right thing by making the changes, but the problem is with the lack of transparency in the publication."

Table. Overview of the Findings by Vedula and Colleagues From the Gabapentin Studies

Lead Investigator Publication Complaint P Value
Torsten Gordh, MD, Uppsala University Hospital, Sweden Pain. 2008;138:255-266. Introduced new primary outcome in the publication. Relegated 1 or more protocol-specified primary outcomes to a secondary outcome in publication. P value in research report same as in publication.
Eduard Vieta, MD, University of Barcelona, Spain J Clin Psychiatry. 2006;67:473-477. Introduced new primary outcome in the publication. Did not describe protocol-specified primary outcomes. P value for primary outcome significant in publication but not in initial research report.
Augusto Caraceni, MD, National Cancer Institute of Milan, Italy J Clin Oncol. 2004;22:2909-2917. Introduced new primary outcome in the publication. Did not describe protocol-specified primary outcomes. P value for protocol-defined primary outcome in research report described as not significant. Different P value published.
Francisco Gómez-Pérez, MD, National Institute of Nutrition of Salvador Zubiran, Mexico City Br J Diabetes Vasc Dis. 2004;4:173-178. Did not describe a change in primary outcomes between protocol and publication. P value for primary outcome significant in publication but differed in initial research report.
Miroslav Backonja, MD, University of Wisconsin, Madison Clin Ther. 2003;25:81-104. Did not describe a change in primary outcomes between protocol and publication. P value for primary outcome in publication reported as not statistically significant.
Michael Serpell, MD, Gartnavel General Hospital, Glasgow, UK Pain. 2002;99:557-566. Did not describe a change in primary outcomes between protocol and publication. P value in research report same as in publication.
Po Wang, MD, Stanford University, California Bipolar Disord. 2002;4:296-301. Relegated 1 or more protocol-specified primary outcomes to a secondary outcome in publication. P value for primary outcome in publication was .0001
Ninan Mathew, MD, Houston Headache Clinic, Texas Headache. 2001;41:119-128. Introduced new primary outcome in the publication. Did not describe protocol-specified primary outcomes. P value for primary outcome significant in publication but not in initial research report.
Atul Pande, MD, Parke-Davis Pharmaceuticals, Ann Arbor, Michigan Bipolar Disord. 2000;2:249-255. Did not distinguish between primary and secondary outcomes in the publication even though they were specified in the protocol. P value for protocol-defined primary outcome in research report defined as not significant. Different P value published.
Kenneth Gorson, MD, St. Elizabeth’s Medical Center, Boston, Massachusetts J Neurol Neurosurg Psychiatry. 1999;66:251-252. Introduced new primary outcome in the publication. Did not describe protocol-specified primary outcomes. Not published in full.
Miroslav Backonja, MD, University of Wisconsin, Madison JAMA. 1998;280:1831-1836. Did not describe a change in primary outcomes between protocol and publication. P value for primary outcome significant in publication, but differed in initial research report.
Peter Wessely, MD, University of Vienna, Austria Cephalalgia. 1987;6:477-478. Introduced new primary outcome in the publication. Did not distinguish between primary and secondary outcomes in the publication even though they were specified in the protocol. Not published in full.

 

During an interview, Karl Kieburtz, MD, from the University of Rochester in New York, emphasized the enormous responsibility and heavy obligation all researchers involved in clinical trials carry. "We must conform to the rules very exactly and work competently and truthfully," he said.

Dr. Kieburtz, also a member of the American Academy of Neurology, has worked on more than 50 clinical trials with about half funded by the National Institutes of Health and the rest by the private sector.

"Every sponsor — whether government or industry — is disappointed if study results are different from what was anticipated," he said. "But I can tell you that I have never been pressured to behave unscientifically. I have been encouraged to spin a secondary outcome, but we have to resist any sponsor in such instances," he said.

"I don't know what the situation was in the studies listed in this paper and I don't know if what the investigators say happened actually did," Dr. Kieburtz said. "Researchers may well have changed the primary outcome before unblinding — in which case it would have been very appropriate to do so."

Lead investigators Drs. Torsten Gordh, Miroslav Backonja, Ninan Mathew, and Po Wang did not respond to Medscape Neurology's request for an interview to address these questions.

"Bias happens in research," Dr. Johnston said. "As a research community, we need to take steps to minimize it." She suggests that individual investigators could benefit from support from an independent oversight committee.

Independent Oversight Committee

"The solutions to these problems lie in preplanning and prespecifying," Dr. Johnston said. "It could be easy for investigators in the middle of the trial to lose focus, and I think they could really benefit from the added support of having people help them oversee the trial. Sponsors could provide the funding to hire this independent group to help supervise."

Dr. Johnston added, "We have to go back to the basics and remember why we do research in the first place. There are lots of reasons that drive people, but at the end of the day, it is our responsibility to uncover the truth."

Dr. Kieburtz emphasized that investigators should first and foremost rely on the Consolidated Standards of Reporting Trials (CONSORT) guidelines. Reviewers and editors also play a role, he noted, and should provide oversight by referring back to the clinical trials registry they lobbied for.

Dr. Dickersin argues that existing registration systems need to go further and should include registration of the full study protocol and amendments. She is also concerned about the 2005 Cochrane systematic review that concluded that gabapentin is effective on the basis of published findings. She said, "It should now be updated with unpublished information made available through litigation."

Dr. Vedula reports receiving fees from plaintiffs' lawyers in litigation concerning gabapentin. Dr. Dickersin reports serving as an expert witness.

N Engl J Med. 2009;361:1963-1971.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....