Research Shows Poor Ability to Determine Which Adverse Events Are Caused By Which Drug in Cancer Treatment Trials

Monday, June 05, 2006

ATLANTA — Researchers in the North Central Cancer Treatment Group (NCCTG) have found that physician assessment of adverse events regarding their relationship to cancer treatment is significantly flawed. Mayo Clinic's Shauna Hillman presented these findings today at the 2006 American Society of Clinical Oncology (ASCO) annual meeting.

When reviewing over 1,000 adverse event attributions in patients taking a placebo, the researchers found that almost half were incorrectly attributed.

"Determining whether an adverse event is related to a drug or some other condition takes considerable time and effort," says Hillman, cancer researcher at Mayo Clinic and lead author of the study. "This data is routinely collected, yet our experience suggests it might not be reliable, so we decided to scientifically review the process." The Common Toxicity Criteria (CTC) Version 2.0 was implemented in March 1998 by the National Cancer Institute (NCI) and requires collection and reporting of adverse events within a treatment study, and assigning them a relationship to the treatment or some other factor.

NCI defines adverse events as unexpected medical problems that occur during drug treatment or other therapies. They do not have to be caused by the drug or therapy, and can be mild, moderate or severe.

Hillman and her fellow researchers looked at patients in the placebo arm of a Phase III clinical trial NCCTG (NCCTG 97-24-51), which was comparing an experimental treatment — carboxyamido-triazole (CAI) — versus a placebo for stage III or IV non-small cell lung cancer. The placebo arm was examined because all adverse events are truly "unrelated" to treatment.

For the 84 patients in the placebo arm, more than 1,000 adverse events were reported. Physicians reported the events as being "not related" (not related or unlikely) or "related" (possible, probable or definite). Despite all patients being on the placebo, 47 percent of events were reported as at least "possibly related" to treatment. Physicians' preconceptions also influenced event attribution. Expectations that the treatment would cause problems resulted in events occurring in healthier patients and events typically associated with lung cancers being more likely to be reported as related.

Not only were nearly half of the attributions incorrect, but the researchers also found that when patients experienced the same event more than once, 36 percent of the time the attribution category changed one or more times.

Because of the significant number of misattributions and inconsistency in attribution reporting, the researchers report limited confidence in the value or reliability of this type of data, and recommend revising the CTC requirements.

Other Mayo Clinic researchers participating in this study included Sumithra Mandrekar, Ph.D.; Brian Bot; Edith Perez, M.D.; Alex Adjei, M.D., Ph.D.; Jan Buckner, M.D.; and Daniel Sargent, Ph.D. They collaborated with John Kugler, M.D., from Oncology Hematology Associates, Peoria, Ill.

NCCTG, a national clinical research group sponsored by the National Cancer Institute, consists of a network of cancer specialists at community clinics, hospitals and medical centers in the United States, Canada and Mexico. NCCTG works with Mayo Clinic to bring the latest cancer research to the community. Find out more about NCCTG and available clinical trials or learn about the CTC.

###

To obtain the latest news releases from Mayo Clinic, go to www.mayoclinic.org/news. MayoClinic.com is available as a resource for your health stories.