Second primary cancers after prostatic irradiation: Ensuring an appropriate analysis

Moon et al. published a Surveillance, Epidemiology, and End Results (SEER) 2002-based analysis of 140,767 men, alive at 5 years after diagnosis of prostate cancer. This analysis associated external beam irradiation with subsequent cancers.1 We recently published a SEER 2004 analysis that showed no increased hazard for rectal cancer after prostatic irradiation, and we suggested that selection bias could explain previously described differences.2 We report here on Cox proportional hazard analyses of 200,525 SEER 2006 cases with prostatic cancer and at >5 years of follow-up, where we used covariates of attained age,2 stage, tumor grade, and therapy. The hazard ratios for second cancer sites after irradiation were: rectum, 1.13 (95% confidence interval [CI], 0.98–1.31); urinary bladder, 1.23 (1.15–1.32), transverse colon, 1.14 (0.89–1.45), brain, 1.22 (0.89–1.68), lung, 1.15 (1.08–1.24); and stomach, 1.05 (0.88–1.26). When 520,708 SEER 2006 prostatic cancer cases, with any length of follow-up, were similarly analyzed, none of these sites showed increased risk. The use of attained age, as well as Cox modeling, seems a more appropriate strategy than the analysis presented by Moon.1 Moon's analysis did not incorporate an appropriate measure of duration of follow-up. Because malignancy risk increases with age, an older patient followed for a number of years accumulates a greater risk than a younger patient followed for the same period. Failure to account for factors like this can lead to an overstatement of the attributable risk in older subgroups, like men irradiated for prostatic cancer. Moon's finding of increased second cancer risk in men treated with transurethral resection of prostate (TURP) alone may be another example of this.1 Exclusion of cases with fewer than 5 years of follow-up is based on the assumptions that radiation-induced cancer has a long latency and that cancers arising shortly after radiotherapy are not radiation-related.1 Indeed, we found no evidence for a latency effect.2 Men with prostatic cancer are interested in whether they have an increased risk of cancer subsequent to irradiation, regardless of the cancer cause. Studies that could influence treatment decisions should, therefore, analyze all cancers. Moreover, efforts should be taken to ensure an appropriate analysis. Wayne Kendal MD, PhD*, Libni Eapen MD*, Garth Nicholas MD*, * The Ottawa Hospital, Regional Cancer Centre, Ottawa, Ontario, Canada.