The Case for Randomized Trials in Cancer Treatment: Title and subTitle BreakNew Is Not Always Better

Given a choice, most patients diagnosed with cancer would prefer the most recently discovered treatments, prompting them to seek out the latest innovation, even if it includes relatively untried medications that have not been thoroughly evaluated.^{1 - 2} In the mid 1950s, pioneering cancer researchers, anticipating the need for clinical trials, requested funds from the US Congress for the development of the first clinical research cooperative groups.³ A small network of institutions was formed to test new anticancer agents arising from the National Cancer Institute’s Drug Development Program. Eleven of the original groups remain and 3 are dedicated to specific areas of therapy such as the Radiation Therapy Oncology Group (RTOG), the American College of Surgeons, and the American College of Radiology Imaging Network. Nineteen hundred institutions are involved, with more than 60% of patients who participate in publicly or privately funded clinical cancer trials enrolled through the cooperative groups; more than 20 000 patients take part annually.^{3 - 4}

Clinical trials determine whether a new treatment is superior to the current standard therapy. Some trials are designed to establish the equivalence of new treatment regimens, others to compare the incidence of adverse events. A third group of trials evaluate the possibility of minimizing therapy. Prevention of incident or recurrent disease is an increasingly important outcome.

For study approval, such trials undergo a rigorous process and need approval by their own trial committees and the National Cancer Institute. Through this mechanism, some of the major advances in cancer therapy have been made, including less extensive surgery for breast and colorectal cancer^{5 - 6} ; adjuvant chemotherapy with improved overall survival for breast and colon cancer^{7 - 9} ; treatment of leukemia,¹⁰ non-Hodgkin lymphoma,¹¹ and Hodgkin disease¹² ; and prevention of breast and prostate cancer.^{3 ,13 - 14} Trials have also seen the introduction of quality of life and economic evaluations, including Markov modeling.¹⁵ In a review of the National Institutes of Health database (clinicaltrials.gov) nearly one fourth of the 11 918 active trials involve cancer.¹⁶

The article in this issue of JAMA by Soares et al¹⁷ examines whether new treatments in radiation oncology are better than standard treatment by analyzing data from more than 12 000 patients from 59 RTOG trials. The report card for the RTOG was excellent: 95% of the studies were published, the general quality of the trials was high, and the choice of the comparator for the innovation intervention group was good. Investigators designing these trials would not have been able to predict what intervention was best, despite any prejudice at the outset. In fact, new treatments were no more likely than the standard treatment to be successful. The authors of this study have made a strong argument for the system and conclude that the rationale for randomized clinical trials works. An example of the importance of not assuming that new treatments are better than the old standard was the use of beta-carotene to prevent lung cancer among male smokers.¹⁸ Initially designed to show the benefit of vitamins, the trial was halted because of excess mortality in the treated group.

In a related article by the same authors published in 2004,¹⁹ the investigators reported on the methodological quality of the same 59 RTOG trials that were assessed in this issue of JAMA. They found that, although the published reports left out serious methodological issues, the studies themselves were of high quality and followed the actual and more inclusive original study protocols.^{19 - 20}

In the present study,¹⁷ the investigators evaluate additional factors that influence trial results, namely publication bias, design, and choice of treatments in each study group. Reviewing the results of these long and expensive studies at first glance is discouraging in that even though the comparators seemed equivalent at the initiation of the study, few showed an advantage for the “innovative” treatment group. In fact, the meta-analyses conducted failed to show any significant advantage either in survival or all other outcomes between innovative and standard treatments in the entire cohort of studies. Yet it is only after a trial is completed that physicians and patients know which treatment is best. In 1994, in a highly publicized national high-priority intergroup trial, it was shown that cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) chemotherapy for high-grade non-Hodgkin lymphoma had less toxicity and the same survival outcome as the treatments in the 2 innovative groups that had been shown in phase 2 trials to be nearly twice as effective as CHOP.¹¹ Without this study, the current standard of care might have consisted of more toxic, expensive, and complicated treatment programs.¹¹

More than a decade ago Hellman and Hellman²¹ argued that randomized clinical trials carry a great risk of violating the relationship physicians have with their patients. If a patient knew in advance or believed that an outcome would show one treatment to be better than another, they argued, the patient should be provided with that information. The results of RTOG studies make a strong argument against this. Equipoise proved to be present throughout the trials and the superiority of the innovative or the standard treatment was not known until study completion.¹⁷

The studies chosen by RTOG from 1968-2002 seem to be stacked against the innovative treatment groups. The diseases addressed in these trials included the most resilient cancers: head and neck, lung, central nervous system, and metastatic disease.¹⁷ The patients most likely to enroll were only those with tumors that were very advanced (not surgical candidates) and potentially resistant to radiation therapy and cure.

Fifty-two trials showed no difference between the standard and innovative treatment groups¹⁷ despite the high expectations of the investigators for success of the new treatments. Only 7 trials of the 59 comparisons demonstrated statistically significant differences, with 6 favoring the new intervention.

The results of this analysis study suggest that quality of life may be a more appropriate primary outcome than survival. Although Soares et al¹⁷ did not find use of primary quality-of-life outcomes in any of the phase 3 studies that they reviewed, quality of life has been or is now being measured in more than 45 RTOG trials (D. W. Bruner, PhD, oral communication, January 2005).¹⁵ Quality of life includes the harms of treatment, such as lymphedema, radiation fibrosis, neurological damage, nausea, and weight loss. It is likely that future RTOG trials will routinely include quality of life since few treatments against such a resistant group of cancers will be curative.¹⁵

A problem with many large cooperative group randomized trials is that the studies take too long to complete due to poor accrual. In this group of 56 studies, the average time to complete publication was 10.7 years, with a range from 5 to 22 years.¹⁷ Because of this delay diagnostic imaging, laboratory testing, and staging criteria can change and treatments can become obsolete. With this in mind, it appears that a better reporting time should be less than 5 years.

Presently approximately fewer than 3% of adult patients with cancer enroll through the Clinical Trials Cooperative Group mechanism; for children with cancer the proportion is more than 50%.^{16 ,22} Because of this disparity a major effort is under way to publicize trials and encourage enrollment of adults, including minority patients who have been historically underrepresented.²³

Whenever possible, barriers to recruitment need to be removed. Payment for treatments by insurance companies should help, as would the use of culturally and linguistically fluent recruiters. In addition, outreach would improve community understanding of the importance of research. These efforts are even more important for prevention trials, in which large numbers of healthy participants need to be enrolled. In the initial National Surgical Adjuvant Breast and Bowel Project tamoxifen breast cancer prevention trial, fewer than 4%¹³ of the women were minorities and for Study of Tamoxifen and Raloxifene (STAR) it was slightly higher, but still only 7%.²⁴ Although minorities enrolled in trials seem to have the same outcomes, they may enter with more advanced disease because of delayed access to health care and the enrollment criteria may be too restrictive. Similar limitations also apply to elderly patients.

The timely completion of high-quality clinical trials is crucial for the more than 1.3 million US residents who will be diagnosed with cancer this year²⁵ and who need to be provided with evidence-based treatments they can trust. It is a priority to know that the current system of clinical trials research through the cooperative groups is healthy. Soares and colleagues¹⁷ have gone a long way in providing this reassurance.