Absolute Effect of Prostate Cancer Screening: Balance of Benefits and Harms by Center within the European Randomized Study of Prostate Cancer Screening

• Published in: Clinical cancer research 2016-01, Vol.22 (1), p.243-249
• Main Authors: Auvinen, Anssi, Moss, Sue M, Tammela, Teuvo L J, Taari, Kimmo, Roobol, Monique J, Schröder, Fritz H, Bangma, Chris H, Carlsson, Sigrid, Aus, Gunnar, Zappa, Marco, Puliti, Donella, Denis, Louis J, Nelen, Vera, Kwiatkowski, Maciej, Randazzo, Marco, Paez, Alvaro, Lujan, Marcos, Hugosson, Jonas
• Format: Article
• Language: English
• Subjects: Aged; Biomarkers, Tumor; Early Detection of Cancer - adverse effects; Early Detection of Cancer - methods; Europe; Humans; Incidence; Male; Mass Screening - adverse effects; Mass Screening - methods; Middle Aged; Mortality; Multicenter Studies as Topic; Population Surveillance; Prostate-Specific Antigen - blood; Prostatic Neoplasms - diagnosis; Prostatic Neoplasms - epidemiology; Randomized Controlled Trials as Topic; Urologi och njurmedicin; Urology and Nephrology
• ISSN: 1078-0432; 1557-3265
• DOI: 10.1158/1078-0432.CCR-15-0941

The balance of benefits and harms in prostate cancer screening has not been sufficiently characterized. We related indicators of mortality reduction and overdetection by center within the European Randomized Study of Prostate Cancer Screening (ERSPC). We analyzed the absolute mortality reduction expressed as number needed to invite (NNI = 1/absolute risk reduction; indicating how many men had to be randomized to screening arm to avert a prostate cancer death) for screening and the absolute excess of prostate cancer detection as number needed for overdetection (NNO = 1/absolute excess incidence; indicating the number of men invited per additional prostate cancer case), and compared their relationship across the seven ERSPC centers. Both absolute mortality reduction (NNI) and absolute overdetection (NNO) varied widely between the centers: NNI, 200-7,000 and NNO, 16-69. Extent of overdiagnosis and mortality reduction was closely associated [correlation coefficient, r = 0.76; weighted linear regression coefficient, β = 33; 95% confidence interval (CI), 5-62; R(2) = 0.72]. For an averted prostate cancer death at 13 years of follow-up, 12 to 36 excess cases had to be detected in various centers. The differences between the ERSPC centers likely reflect variations in prostate cancer incidence and mortality, as well as in screening protocol and performance. The strong interrelation between the benefits and harms suggests that efforts to maximize the mortality effect are bound to increase overdiagnosis and might be improved by focusing on high-risk populations. The optimal balance between screening intensity and risk of overdiagnosis remains unclear.