BETHESDA, Md--Ending a clinical trial of a promising intervention may be just as difficult and delicate as starting one. The stopping point, Gilbert S. Omenn, MD, PhD, said at the American Society of Preventive Oncology meeting, applies to both the desired effect (or efficacy) and adverse effects (toxicity).

Investigators must be clear about the primary endpoint of the trial, but at the same time make provisions for interim analyses at one-third and two-thirds of the projected weighted endpoints, said Dr. Omenn, executive vice president for medical affairs, University of Michigan.

Dr. Omenn headed the CARET study, which tested a high-risk population of 18,314 smokers and asbestos workers to see if 30 mg of beta-carotene and 25,000 IU of vitamin A given orally each day would reduce the risk of lung cancer.

The researchers monitored lung cancer risk factors in the study population; potential side effects (for the participants’ safety); and actual lung cancer incidence. They also assayed the conditional power of the study to see if there was sufficient power to obtain a valid conclusion.

In the meantime, the parallel ATBC study in Finland, which examined the effects on smokers of beta-carotene and vitamin E, reported to Dr. Omenn that subjects in the intervention arm were developing lung cancer at a higher rate than those on placebo.

For the first 12 to 18 months of the trial, the CARET investigators found no difference between the study arms--not in lung cancer incidence, cardiovascular disease, or total mortality. However, after that point, they found that CARET participants taking the vitamins had a relative risk of 1.28 for lung cancer, compared with the placebo arm. Six cases of lung cancer per 1,000 subjects were seen in the treatment arm vs `five cases per 1,000 subjects in the placebo arm.

Dr. Omenn and his colleagues had to figure out how to deal with this finding as well as the "unexpected and unexplained" news from Finland. They prepared scripts of answers to likely questions for the staff at the six study sites. All participants were notified and told they could stop taking the medications immediately, if they wished, but only 606 participants out of 18,000 became inactive. The NCI told the study leaders that the decision to stop was theirs to make, and they voted unanimously to stop the trial.

Prepare a Close-Out Plan

Dr. Omenn suggested a number of ways to prepare for the possibility of a study having adverse effects. Plan ahead. Don’t think of the study in isolation. Review all possible scenarios, even ones that are barely imaginable. "You can’t just walk away from a study," he said. "You need a close-out plan for every study, whether the news is good or bad."

Plans for informing the media and explaining the meaning of the outcome are critical, he said. The investigators should also make plans for grief and loss counseling for participants and staff, and keep them informed for as long as possible. Finally, he said, the researchers should have a plan for data security--how to maintain data and for how long.