Reply to M. Kumar et al

Abstract

We would like to thank Kumar et al for their thoughtful reflection on the analysis we published in 2008. In that study, we reported a significant increase in the risk of prostate-specific antigen (PSA) failure by using a Cox regression multivariable analysis in men with clinically localized prostate cancer who underwent less than 6 compared with 6 months of the antiandrogen therapy, flutamide, in addition to 6 months of a luteinizing hormone–releasing hormone (LHRH) agonist and 70 Gy radiation therapy (RT) in a randomized study comparing that treatment to 70 Gy of RT alone. Two possible reasons exist for the difference in the results that Trans-Tasman Radiation Oncology Group (TROG) investigators found and ours. First, in the multivariable model, the TROG group included as covariates both months of flutamide and treatment (RT and 6 months of hormonal therapy [HT] versus RT only). By doing this, the information regarding HT is included twice; once in the continuous covariate, months of flutamide, and again in the categorical covariate, RT and 6 months of HT. This can lead to a loss of significance in one or both of these terms. Moreover, because these two terms are not independent, a single term—such as months of flutamide—should be considered, for which they could include men receiving RT only as the baseline group and then have RT and all 6 months of combined HT and RT and 6 months of the LHRH agonist but less then 6 months of flutamide in the model. Second, when comparing the TROG study to men in the U.S. study, the U.S. study enrolled men with less-advanced prostate cancer. Specifically, in the TROG 9601 study, approximately 75% of men had clinical category T2c to T4 disease, whereas these men were excluded from the U.S. study, in which only men with clinical category T1b to T2b disease were enrolled. The more advanced nature of the prostate cancers treated in the TROG study is additionally supported by a median PSA of approximately 15 ng/mL compared with approximately 11 ng/mL in the U.S. study. Given these differences in patient selection, a hypothesis to explain the discordant results in addition to the statistics could be that men with clinically localized compard with locally advanced prostate cancer are less likely to harbor hormone-refractory disease and, therefore, are more likely to benefit from more complete hormonal blockade. For now, however, this remains a conjecture. Therefore, we agree that our analysis as well as the post–random assignment analysis from the TROG 9601 study provided in their letter is hypothesis generating. Prospective validation of the association between the use of combined HT and a reduction in the risk of PSA recurrence in men with clinically localized prostate cancer is needed, as stated in the final sentence of our study.