NEGATIVE TRIAL DO NOT ALWAYS PUBLISH

July 13, 2011 — A "substantial" number of phase 3 clinical trials of systemic cancer treatments with potential influence on clinical practice remain unpublished after 6.5 years or more, according to a new report from Canadian researchers.
The investigators also found that "many" other trials are published after a delay of 5 years or more. Their paper was published online July 11 in the Journal of Clinical Oncology.
The findings have potential ramifications in oncology practice, say the authors.
"Nonpublication of clinical trials predominantly reporting negative results can lead to overestimation of treatment benefits and may adversely influence clinical practice," write the authors, led by Vincent C. Tam, MD, from the University of Toronto, in Ontario.
The researchers considered phase 3 trials of systemic cancer treatments that were presented at the annual meeting of the American Society of Clinical Oncology (ASCO) from 1989 to 2003.
A total of 709 phase 3 trials were identified, and 66 (9.3%) of those remain unpublished at a minimum follow-up of 6.5 years. In addition, 94 (13%) were published 5 years or more after their initial presentation. Thus, nearly a quarter of all the trials were either unpublished or their publication was greatly delayed.
Perhaps not surprisingly, 71% of these clinical trials reported negative results.
There is an ethical issue at stake in all of this, suggest the authors. "Nonpublication also breaks an implicit contract that investigators make with study participants," they write.
Dr. Tam and colleagues explain further: "Approximately 23,770 patients participated in the unpublished trials listed in our compendium — an astounding number of patients with cancer who participated in trials after being informed that the results would contribute to public knowledge, and might improve cancer treatment."
Examples of Unpublished Trials
The clinical impact of unpublished trials is mostly unknowable, the authors acknowledge. "It is difficult to quantify the extent to which trials in our cohort could have impacted clinical practice," they write. Nevertheless, they had the unpublished trials analyzed by 2 disease-site-specific oncology experts for each of the 4 most common tumor sites. One unnamed expert was from Princess Margaret Hospital and the other was from Sunnybrook Odette Cancer Centre, both in Toronto, which is where the authors are based.
For each abstract, the experts were asked whether they believed the clinical trial addressed an important question and to rate the potential impact of nonpublication on clinical practice.
None of the 66 unpublished trial results were judged to have "critical" impact, but 32 might have had some impact on clinical practice if their results had been published shortly after presentation, report the authors.
The majority of the unpublished trials belonged to the following tumor-site groups: breast (23 trials), gastrointestinal (14), hematologic (9), and lung (8).
The unpublished breast cancer trials included 1 that compares 2 common chemotherapy combinations; it was reported at the 1991 ASCO annual meeting. The trial showed that, in women with positive axillary nodes, adjuvant fluorouracil, doxorubicin, and cyclophosphamide led to survival rates similar to those seen with cyclophosphamide, methotrexate, and fluorouracil in the adjuvant setting.
The unpublished lung cancer trial with the "highest potential impact" was the Radiation Therapy Oncology Group 9410, say the authors. The trial, reported in 2000, compared concurrent cisplatin, vinblastine, and radiation with sequential chemotherapy and radiation in patients with unresected stage III nonsmall-cell lung cancer. The abstract reported no statistically significant improvement in overall survival with either treatment. However, there was a trend toward improved survival with the concurrent treatment.
One of the concerns about negative trials not being published is that the medical literature overestimates treatment effects, say the authors. One paper has shown as much, they say, citing a review of oncology metaanalyses that only included published papers (J Clin Oncol.1986;4:1529-1541). This "publication bias" toward positive clinical trial results is supported by many studies, write Dr. Tam and colleagues.
They propose a number of ideas to remedy the problem of unpublished major clinical trials.
The potential solutions include "funding organizations mandating publication as a condition of funding, research ethics committees mandating publication as a condition of ethics approval, mandatory prospective registration of clinical trials in a database, medical journals allowing for short reports of negative trials, and/or inclusion of unpublished study results in reviews and metaanalyses."
J Clin Oncol. Published online July 11, 2011. Abstract