National Cancer Institute NCI Cancer Bulletin: A Trusted Source for Cancer Research News
September 20, 2011 • Volume 8 / Number 18

A Closer Look

Communications series icon
This article is part of a series of stories related to cancer communications. You can read more articles in the series here.

Unpublished Phase III Cancer Trials: Eliminating the Negative?

Stack of peer-reviewed medical journalsMany phase III clinical trials with negative or inconclusive results are never published in peer-reviewed journals, contributing to publication bias.

Phase III clinical trials are the "gold standard" in oncology research, providing the highest level of evidence to support the effectiveness of cancer treatments. These trials usually involve hundreds of patients at multiple sites, cost millions of dollars, and take years to complete. Most important, the published results of a phase III trial can substantially change the standard of care for a particular tumor type or stage of disease.

The American Society of Clinical Oncology (ASCO) annual meeting, the largest meeting of its kind for oncologists, often provides a "first look" at potentially practice-changing clinical trial results, which are subsequently fleshed out in reports published in peer-reviewed journals.

A new study finds, however, that results of nearly one in ten phase III trials presented in abstract form at the ASCO annual meeting remained unpublished more than 6 years later. Most of these unpublished trials, involving more than 23,000 cancer patients, had negative or inconclusive results—that is, they failed to show that the experimental treatment under study was superior to the standard or control treatment.

Publication Bias

Previous studies have shown that clinical trials with positive results—trials that find the treatment being tested to be superior to the standard or control treatment—are more likely to be published than trials with negative or inconclusive results, said Dr. Monika Krzyzanowska of Princess Margaret Hospital in Toronto, senior author of the study published in the Journal of Clinical Oncology in August. Such selective publication of positive research findings is known as publication bias.

"When trials with negative results aren't published, the benefits of treatment may be overestimated and patients may be harmed because of inappropriate treatment decisions," said Dr. Krzyzanowska.

In a previous study, Dr. Krzyzanowska and her co-authors found that 26 percent of the 510 phase III trials presented as abstracts at the ASCO annual meeting between 1989 and 1998 were not published in full within 5 years after presentation. In the current study, the researchers expanded their analysis to include 709 phase III trials presented at the ASCO annual meeting between 1989 and 2003.

"Negative results can be just as important as positive results," said Dr. Jeff Abrams, director of NCI's Cancer Therapy Evaluation Program. "When negative findings don't become part of the scientific literature, other researchers may go down the wrong path, spending time on duplicative studies because they don't realize that approach was tried and was unsuccessful."

Important Questions

Most of the unpublished trials were of therapies for breast, gastrointestinal, hematologic, or lung cancer. A majority addressed an important clinical question and might have affected clinical practice had they been published in a timely fashion, according to oncologists specializing in those cancers, whom the investigators asked to evaluate the trial abstracts.

It's a disservice to science not to publish studies that are inconclusive or negative.

—Dr. Jeff Abrams

The specialists' opinions are subjective, acknowledged Dr. Krzyzanowska. "It is challenging to retroactively assess the potential clinical impact of trials that in some cases were performed 15 or more years ago," she said.

Investigators involved in the unpublished trials most frequently cited a lack of time, funds, or other resources as the reason for not publishing the results. Some said a report of the trial had been submitted for publication but rejected.

Nonpublication of clinical trial results "breaks an implicit contract" between investigators and patients, who are motivated to enroll in a trial at least in part by the belief that their participation will advance scientific knowledge. It also breaches trust between investigators and funding agencies, which may adversely affect funding for future clinical trials, Dr. Krzyzanowska and her colleagues wrote.

"The term 'negative' is a psychological barrier to getting results published," said Dr. Bjorn R. Olsen of Harvard Medical School, who is editor-in-chief of the Journal of Negative Results in Biomedicine, an open-access, peer-reviewed online journal. "Journal editors and reviewers are looking for results that they believe will move the field forward." Papers reporting negative findings may be rejected for publication because reviewers or editors attribute the findings to weaknesses in the study design or execution, he added.

Trial Registration

Since 2007, all clinical trials initiated at one or more sites in the United States (except for phase I trials) by law must be registered in ClinicalTrials.gov, a publicly available database. Since 2008, trial sponsors must submit a summary of the trial's results to ClinicalTrials.gov within a year of the trial's completion.

These requirements will help ensure that, even if a trial's results are not published in a peer-reviewed journal, a summary of the findings will be available to researchers, patients, and the public, explained Dr. Abrams.

A journal dedicated to publishing negative results specifically in oncology might also be part of the solution, according to Dr. Krzyzanowska. Dr. Abrams, however, questioned whether a journal focused on negative results would attract readers. A better approach, he suggested, might be to encourage medical journals to publish brief reports of negative trials. This would offer an incentive to investigators who would likely not be motivated to spend time writing full-length reports of negative trials that they believe are unlikely to be published. These brief reports would provide more information about the trials than is available in an abstract.

"It's a disservice to science not to publish studies that are inconclusive or negative," said Dr. Abrams.

Eleanor Mayfield

Delayed Publication: How Long is Too Long?

In addition to trials that remained unpublished 6 years after being presented as abstracts at the ASCO annual meeting, Dr. Krzyzanowska and her colleagues found another 13 percent of trials were published after a delay of 5 years or more from the time of their presentation.

The impact trial results have on practice depends on their being published in a timely fashion, she explained.

"A 250-word abstract touches only briefly on a trial's key results—and those results may evolve between presentation of the abstract and publication of the final manuscript," said Dr. Krzyzanowska. "Many clinicians and funding agencies wait to see the published paper, which not only provides more detailed information about the study but also has been rigorously peer-reviewed."

When an abstract presents interim study results, delayed publication is understandable because time is needed for the data to mature, Dr. Krzyzanowska commented. "For mature studies, most people would agree that, if your goal is to influence practice, a delay of 5 years or more from presentation of an abstract to publication of the full results is too long."