Randomized trials of high-dose chemotherapy in breast cancer: fraud, the press and the data (or lessons learned in medical policy governing clinical research).

High dose therapy for breast cancer remains controversial. Of the 15 randomized trials of high dose therapy in breast cancer reported to date, two South African studies have been discredited leaving 13 remaining studies. Mortality was consistently low, in the 0 to 2.5% range, except for the BCNU containing American Intergroup study, which had a 7.4% toxic mortality rate. Seven of the remaining 13 studies randomized fewer than 200 patients. Three of these small studies have significant differences in disease free survival, and a fourth study has a trend in favor of high dose therapy. The other three small studies cannot exclude a survival difference of 20%. Of the 6 remaining moderately large trials of 219 to 885 randomized patients, 5 are adjuvant studies and one included patients with metastatic disease. Of the five adjuvant trials, four have significant differences in relapse rate favoring the high dose arm, and the remaining study has a trend (with a high dose sequential single agent design rather than combination therapy as in the other studies). A planned subset analysis of the first 284 patients in the largest study funded by the Dutch insurance industry showed a significant advantage for high dose therapy. Given the 2-year median time to relapse and an addition 2-year median to death after relapse, the follow up for survival of 3-5 years on these studies is still short. In the only moderately sized metastatic trial from the National Cancer Institute of Canada with a very short median follow-up of 19 months, a significant difference in disease free survival has emerged, with no difference in survival.