Both single-institutional retrospective series and a multiinstitutional observational study1,2 find a significant association between a reduced risk of death and treatment of nodepositiveprostatecancerusing both external-beam radiation treatment (EBRT) of the prostate and pelvic lymph nodes (LNs) and androgen deprivation therapy (ADT) compared with ADT alone. However, whether this association is causal remains unanswered and requires testing in aprospective randomized trial. In this issue of JAMA Oncology, James and colleagues3 use data from the control arm in the Systemic Therapy in Advancing or Metastatic Prostate Cancer: Evaluation of Drug Efficacy (STAMPEDE) Trial to investigate this issue. Specifically, they perform a Cox regressionmultivariable analysis4 evaluating the riskof failure-free survivalwhere failure is defined as prostate-specific antigen (PSA) level; local, regional, or distant treatment failure; or death from prostate cancer, adjusting forhighest initialGleason score category (≤7, ≥8, unknown), log-transformed pre-ADT PSA level (continuous), age (<60, 60-64, 65-69, ≥70years), andWorldHealthOrganization performance status (0 vs 1 or 2) at randomization. Among 177 men with node-positive prostate cancer, 97 were planned for EBRT and 68 of these, plus 10men not originally slated for EBRT,were recordedashaving receivedEBRT to the prostate with or without pelvic LNs. After a median follow-up of 17months in the overall study cohort, 20menwith node-positiveprostate cancer experienceda failure event.The authors report that both planned and delivered prostatewith orwithoutpelvicEBRTwasassociatedwith reducedriskof failure, with adjusted hazard ratios of 0.48 (95% CI, 0.29-0.79) and 0.35 (95% CI, 0.19-0.65), respectively. There are several reasons to interpret these results with caution. First, as the authors note, EBRT use was prescribed by physician choice and therefore nonrandomized. In addition, the short median follow-up of 17 months in a study in which at least 2 years of ADTwas required and a primary end point (ie, failure-free survival) that is likely driven by PSArelated failureeventsat this shortmedian follow-up timemake it difficult topredictwhat effect a reduction inmostlyPSA failure–free survivalwill haveonoverall survival.Moreover, given the lowevent rate (20of 177 [11.3%]), theauthorswerenot able to analyze overall survival as an end point or adjust for all knownprostate cancer prognostic factors in theirmodel, such as tumor category, tertiary grade 5 inmenwith Gleason score 7prostate cancer, andpercent positive biopsies. Also,with40 deaths in the overall study cohort of which 9 could not be attributed to prostate cancer, a competing risk5 and not Cox regression analysis4 would have beenmore appropriate to analyze the end point of time to first failure or prostate cancer death. Finally, treatment use varied, in that somemen could have received irreversible and life-long testosterone suppression via bilateral orchiectomy as comparedwith 2 years of reversible ADT using a luteinizing hormone–releasing hormoneagonist.Moreover, inmen inwhomEBRTwasdelivered, the prostate was always treated whereas the pelvic LNs were only treated in a subset and thedose fractionation scheme for EBRTdeliveryvaried.Both thevariations inEBRTandADTcan affect time to first failure. Ideally, to account for these variations in treatment an adjustment in the model using a treatment propensity score would have been used, but the short median follow-up and consequently a low event rate would likelynotpermit the inclusionof thetreatmentpropensityscore and other known prostate cancer prognostic factors without running the risk of overfitting the model. Given these limitations, the conclusion that the addition of prostate and pelvic EBRT to ADT in the treatment of nodepositiveprostate cancer reduces the riskof failure,whileprobably true, cannot be rigorously concluded from the present analysisbecauseof theshortmedian follow-up, lowevent rate, nonrandomized data, and lack of adjustment for treatment variation and some known prostate cancer prognostic factors. Moreover, to affect clinical practice, at a minimum one should provide evidence in amodel powered for survival and adjusted for age, comorbidity, known prostate cancer prognostic factors, and type and duration of ADT that treatment using both pelvic and prostate RT andADT as comparedwith Related article page 348 Failure-Free Survival and Nonmetastatic Prostate Cancer Radiotherapy Original Investigation Research
Read full abstract