Lois B. Travis, Rochelle E. Curtis, Bengt Glimelius, Eric J. Holowaty, Flora E. Van Leeuwen, Charles F. Lynch, Anton Hagenbeek, Marilyn Stovall, Peter M. Banks, Johanna Adami, Mary K. Gospodarowicz, Sholom Wacholder, Peter D. Inskip, Margaret A. Tucker, John D. Boice

Cyclophosphamide is an established bladder carcinogen, but few studies have examined the relationship between dose and effect. The largest analysis to date included only seven cases of bladder cancer. No investigation has estimated the risk of kidney cancer. The purpose of this study was to quantify the risk of bladder and kidney cancer following cyclophosphamide therapy. Within a cohort of 6171 two-year survivors of non-Hodgkin's lymphoma (NHL), 48 patients with secondary cancer of the urinary tract were identified and matched to 136 control subjects with NHL who did not develop a second malignancy. Detailed information on chemotherapeutic drugs and cumulative dose received was collected for all subjects. Radiation dose to the target organ was estimated from individual radiotherapy records. Evaluations of the risk of second cancer as a result of treatment with cyclophosphamide alone, radiation alone, or both therapies were made relative to those patients who were exposed to neither treatment modality. A significant 4.5-fold risk of bladder cancer (95% confidence interval [CI] = 1.5-13.6) followed therapy with cyclophosphamide, and risk was dependent upon cumulative dose. Among patients who received a total amount of cyclophosphamide of less than 20 g, a nonsignificant 2.4-fold risk of bladder cancer was apparent. Significantly elevated sixfold (95% CI = 1.3-29) and 14.5-fold (95% CI = 2.3-94) risks of bladder malignancy followed cumulative doses of 20-49 g and 50 g or more, respectively (P value for trend = .004). Radiotherapy given without cyclophosphamide was associated with a nonsignificant increased risk of bladder malignancy. Excess bladder cancer risk following treatment with both radiotherapy and cyclophosphamide was as expected if individual risks were summed. Neither radiotherapy nor cyclophosphamide was associated with excesses of kidney cancer. Cyclophosphamide-related bladder cancer is dose dependent. For patients given cumulative doses between 20 and 49 g, the absolute risk of bladder cancer is on the order of three excess cancers per 100 NHL patients after 15 years of follow-up. At cumulative doses of 50 g or more, the excess risk increases to approximately seven excess bladder cancers per 100 NHL patients. The strong dose-response relationship and high absolute risk of bladder cancer underscore the importance of limiting the cumulative dose of cyclophosphamide to what is required to achieve therapeutic end points. The risk of secondary bladder malignancy and other late sequelae of therapy must be carefully weighted against the curative gains provided by cyclophosphamide. Moreover, long-term side effects of therapy that might be acceptable in cancer treatment may need to be re-evaluated for patients with non-neoplastic disorders.