Surgery Versus Radiotherapy for Localised Prostate Cancer
Total Page:16
File Type:pdf, Size:1020Kb
PROSTATE diSEASE 35 Surgery versus radiotherapy for localised prostate cancer PRASANNA SOORIAKUMARAN Even though there have Study sHR (95% CI) sHR (95% CI) been no controlled trials Risk group 1 All 1.91 (1.16–3.14) comparing the effectiveness Age <64 1.92 (0.96–3.82) of radical prostatectomy and Age ≥65 1.87 (0.94–3.69) Charlson comorbidity index score 0 1.67 (0.93–2.99) radiotherapy for treatment of Charlson comorbidity index score ≥1 2.91 (0.88–9.59) localised prostate cancer, there Risk group 2 is substantial observational All 1.77 (1.37–2.29) Age <64 1.95 (1.31–2.91) evidence in favour of surgery. Age ≥65 1.61 (1.16–2.25) Charlson comorbidity index score 0 1.91 (1.42–2.57) Charlson comorbidity index score ≥1 1.36 (0.81–2.30) rostate cancer is the commonest non- dermatological cancer and the second Risk group 3 P All 1.50 (1.19–1.88) leading cause of cancer death in western Age <64 1.78 (1.26–2.51) 1 men. The overwhelming majority of new Age ≥65 1.24 (0.92–1.68) cases are clinically localised, and the most Charlson comorbidity index score 0 1.59 (1.21–2.07) frequent methods of treatment are with Charlson comorbidity index score ≥1 1.30 (0.83–2.03) 2 radical prostatectomy or radiotherapy. Non-metastatic (risk groups 1–3) 1.76 (1.49–2.08) Risk group 4 TRIAL EVIDENCE LACKING All 0.76 (0.49–1.19) Despite millions of men having undergone Age <64 1.08 (0.57–2.03) Age ≥65 0.58 (0.33–1.01) these prostate cancer treatments, no Charlson comorbidity index score 0 0.81 (0.46–1.43) randomised controlled trial evidence Charlson comorbidity index score ≥1 0.65 (0.31–1.35) currently exists to compare their effectiveness. The only such trial underway, 0.25 0.5 124 the ProtecT study, will not report until 2016 Favours radical and may have limited generalisability due to Favours radiotherapy prostatectom y differences in the randomised population and the prostate cancer population at large. 3 Figure 1. Forest plot showing propensity score-adjusted subdistribution hazard ratios for Also, only a small number of the entire radiotherapy versus radical prostatectomy for prostate cancer mortality stratified by risk study cohort has undergone randomisation group (1 = low risk; 2 = intermediate risk; 3 = high risk; 4 = advanced prostate cancer) and (500 cases per treatment arm), and around substratified by age and Charlson score 11 85% of the cases are low or intermediate risk; hence, follow-up to 10 years may not treatment options, and thus is more likely be long enough to find survival differences to show up any comparative effectiveness Prasanna Sooriakumaran, MD, PhD, in this study. differences, but unfortunately will not FRCS(Urol), FEBU, Senior Clinical mature for the next 10 years (O. Akre, Researcher, University of Oxford; A soon-to-be-set-up study, the Scandinavian personal communication). Hence, as prostate Honorary Consultant Urological Surgeon, Prostate Cancer Group (SPCG)-15 trial, is cancer clinicians, we are left with only Oxford University Hospitals NHS Trust randomising high-risk men to different observational data currently to guide us TRENDS IN UROLOGY & MEN’S HEALTH NOVEMBER/DECEMBER 2014 www.trendsinmenshealth.com PROSTATE diSEASE 36 when we counsel patients with localised A similar, but larger, study using the same Charlson comorbidity index, and cancer- prostate cancer. primary dataset, on 67 087 men with related events during follow-up. Hence, localised prostate cancer, stratified survival PCBaSe is a truly population-based dataset OBSERVATIONAL DATA outcome based not only on treatment but and with near-complete data on almost all There is a large amount of such also on life expectancy. 9 It found that in variables of interest. observational data, but unfortunately men with an estimated life expectancy it is of variable quality. Much of the greater than 10 years, surgery was Methods data examines surrogate oncological associated with improved survival over We examined prostate cancer mortality as endpoints such as biochemical recurrence, radiotherapy for all localised disease our primary outcome, and compared an outcome with differing definitions stages. Interestingly, survival benefits subjects who had been managed with between treatment modalities and with a were not consistent for those with shorter radical prostatectomy as their initial variable progression to clinically significant life expectancies. treatment with those who had undergone oncological outcomes such as death. 4 primary radiotherapy. 11 A total of 34 515 Only few observational studies remain One problem with these US datasets is men made this study cohort, with a if we therefore consider survival as that they are not comprehensive, in that follow-up to 15 years (median 5.37 years). the outcome of primary interest when they capture only a proportion of the Subjects were classified by clinical risk (low, comparing treatments. prostate cancer population. For example, intermediate, high), age (<65, 65 or more) the SEER registry represents only 14% of and Charlson score (0, 1 or more). Data from an observational study of more the US population before 2000 and 26% than 404 000 men treated at over 1000 thereafter, and the Medicare insurance- To visualise cause-specific mortality, community hospitals in 44 states of the linked programme captures only those cumulative incidence curves were plotted USA found that surgery had superior aged over 65 years. Another problem is for the treatment groups, and differences prostate cancer and other-cause mortality that these datasets have a lot of missing in cancer-specific and other-cause outcomes for men aged under 80 with data in their captured variables, and mortality were investigated using low- and intermediate-risk disease. 5 This do not include many important covariates subdistribution hazard ratios corrected study used the Nationwide Inpatient that can influence outcome, such as for competing risks by Fine and Gray Sample of the United States as its dataset, comorbidities. proportional hazards regression. To deal and others have found similar benefits for with differences in baseline characteristics surgery over radiotherapy using other SWEDISH STUDY between the treatment groups, we US datasets, such as SEER (Surveillance, We therefore wanted to interrogate this produced both traditional multivariable Epidemiology and End Results)/Medicare, question of comparative oncological model-adjusted and propensity score- and CapSUrE (Cancer of the prostate effectiveness in our most common cancer, adjusted estimates of subdistribution Strategic Urologic research Endeavor). 6,7 using a higher-quality dataset. In Sweden, hazard ratios. Propensity scores were every person is assigned a national identity calculated using logistic regression, with Recently, the Prostate Cancer Outcomes number that tracks them throughout life, treatment group as the outcome variable Study, based on SEER data, reported and thus the National Prostate Cancer and all adjustment covariates as predictors. on 1655 men with localised prostate Registry (NPCR) of Sweden has been shown We also used propensity scores for cancer. 8 The majority of the subjects to cover 98% of all cases diagnosed since matching, which was carried out within were diagnosed in the PSA era as a 1998 (and with coverage from 1996 to 1997 each risk group. As well as all the above, result of screening and thus the cohort limited to certain geographical regions). we performed a sensitivity analysis to look was relatively contemporary. Another for residual confounding, analysis by year advantage of this study was the use of The NPCR has been linked to eight other of surgery to examine what effect, if any, different, robust statistical techniques, national registries, including the Swedish higher radiotherapy dosing with time including using propensity scores cancer register and the cause-of-death would have had, and an inverse probability as covariates in survival analyses, register, to compose the PCBaSe Sweden. 10 of treatment weights analysis. for stratification, for matched-pair PCBaSe has virtually complete data on year analyses, and in an inverse probability of of diagnosis, age, clinical (TNM) stage, Results treatment weights analysis. The authors tumour grade, serum PSA, planned primary At baseline, radiotherapy patients generally found consistent benefits for radical treatment within 6 months of diagnosis, had worse patient-tumour characteristics; prostatectomy over radiotherapy for county of residence, marital status, specifically, they had a higher proportion overall and prostate cancer mortality. educational level, socioeconomic status, of high clinical stage and grade disease, www.trendsinmenshealth.com TRENDS IN UROLOGY & MEN’S HEALTH NOVEMBER/DECEMBER 2014 PROSTATE diSEASE 37 Recent evidence from a large Canadian KEY POINTS population-based study has shown that • Conventional treatments for localised prostate cancer include radical men who received radiotherapy had a prostatectomy and radiotherapy higher incidence of hospital admissions, rectal or anal procedures, open surgical • No randomised trial data that compare surgery and radiotherapy are procedures, and secondary cancers at currently available 5 years post-treatment than did those who • Low-risk prostate cancer, especially in older and/or unfit men, is highly underwent radical prostatectomy. 13 Hence, unlikely to result in death regardless of treatment choice radical prostatectomy may be superior over radiotherapy for localised prostate cancer, • There is a large body of high- and intermediate-quality observational