SPECIAL COMMUNICATION SHOULD WE TREAT LOCALIZED PROSTATE CANCER? AN OPINION MAN1 MENON, M.D., B.G. PARULKAR, M.D., AND STEPHEN BAKER, M.Sc.P.H. T he last decade has seen an exponential rise in the number of patients diagnosed with carci- noma of the prostate. Currently, it ranks first among the tumors in adult men, with an annual incidence of 200,000 and a mortality of about 38,000. A recent study has shown that the mor- tality rate in a population with prostate cancer is 2.25fold that of an age-matched population with- out prostate cancer.l In the United States, a new case of prostate cancer is diagnosed every 3 min- utes, and 1 patient dies of prostate cancer ap- proximately every 15 minutes. One of the basic tenets of cancer treatment is to detect and treat tumors early. Solid tumors that are confined to the organ of origin are best treated by removal of tumor or the organ of origin. Organ- confined prostate cancer can be detected quite eas- ily using a screening blood test-prostate-specific antigen (PSA). Indeed, according to Gann et al.,’ PSA has the highest validity of any circulating screening marker discovered thus far. Conven- tional opinion is that clinically localized prostate cancer can be treated successfully with either ra- diation or surgery, whereas metastatic cancer is in- curable. Yet, there is no consensus among health care providers as to whether one should attempt to diagnose and treat prostate cancer in its early stages. This article attempts to examine the rea- sons for this controversy and to derive some con- clusions based on our analysis of the literature. The lack of consensus about whether localized prostate cancer should be treated is a result of the peculiar biology of prostate cancer. The incidence of microscopic prostate cancer detected at autop- sies in adult men over age 50 years is around 40%. Yet the lifetime risk of clinical cancer in American men is 10%. This means that a significant per- centage of men may die with prostate cancer rather than of it. The median age at the time of detec- tion of prostate cancer is 72 years, and the median From the Division of Urological and Transplantation Surgery and Department of Academic Computing, University of Massachusetts Medical Center, Worcester, Massachusetts Reprint requests: Mani Menon, M.D., Division of Urological and Transplantation Surgery, University of Massachusetts Medical Center, 5.5 Lake Avenue North, Worcester, MA 01655 Submitted: March 13, 199.5, accepted (with revisions): June 14,199.5 UROLOGY@ 46 (51, 1995 age at death due to prostate cancer in white Amer- ican men is 77 years. Since the life expectancy of American white men is 74.2 years, it may be ar- gued that the average man has effectively outlived his expected age of survival at the time of diag- nosis. This has led to the famous aphorism by Whitmore: “Is cure possible in those for whom it is necessary and is cure necessary in those for whom it is possible?” Localized prostate cancer gives rise to symp- toms only rarely; thus, benefit from treatment must be measured in terms of improvement in survival or a decrease in metastatic rates. It may be argued that therapy is effective only if it im- proves the crude survival of the population treated. Thus, if 100 patients with prostate cancer are treated with radiation and 50 are alive after 10 years, the crude survival will be 50% at 10 years. However, some of the patients who died will have died of causes other than prostate cancer. An older, sicker patient population will have a poorer crude survival than a younger, fitter patient population, irrespective of the disease or treatment. Because crude survival takes into account death from dis- eases other than prostate cancer, a treatment modality that cures prostate cancer may not im- prove overall survival. Thus, many investigators do not use crude survival as the endpoint by which they estimate the benefits of therapy. Nonetheless, to the individual patient, it may appear to be of no special benefit to be cured of prostate cancer only to die from other diseases. Although it is en- tirely possible that patients may prefer death from stroke or cardiac disease to death from prostate cancer, to our knowledge no study on prostate cancer or any other cancer has directly addressed patient preferences in these matters. Cancer-specific or disease-specific survival at- tempts to correct for the deaths due to comorbid factors. Deaths due to intercurrent illnesses are treated as censored cases, as are patients still sur- viving at the end of the study or cases lost to follow-up. Modern statistical methods for sur- vival analysis such as Kaplan-Meier product limit analysis or Cox proportionally hazards analysis easily accommodate censored observations con- tributing information on survival until the time of censoring. Thus, this type of analysis corrects to 607