Validating the Interval to Biochemical Failure for the Identification of Potentially Lethal Prostate Cancer

Purpose To validate the interval to biochemical failure (IBF) as a prognostic factor at the time of biochemical failure for prostate cancer mortality (PCM) following radiotherapy (RT). Patients and Methods From a collaborative data set of men with clinically localized prostate cancer treated with RT from four institutions in three countries, we identified 1,722 men with biochemical failure (BF; prostate-specific antigen nadir + 2 ng/mL). The IBF was defined as the time interval from completion of treatment to the date of BF. The primary outcome measure was discriminatory power in the form of the concordance index (c-index). Results Seventeen percent of men had an IBF <= 18 months. Median potential follow-up beyond the time of BF was 67 months. There were 290 deaths from prostate cancer. The IBF was the most discriminating individual prognostic factor overall, with a sensitivity of IBF <= 18 months to predict PCM within 10 years of 48.4% (95% CI, 43.3% to 54.1%); the specificity was 86.1% (95% CI, 84.5% to 87.7%), equating to a c-index of 0.611 (95% CI, 0.578 to 0.647). The 5-year cumulative incidence of PCM for IBF more than 18 months versus IBF <= 18 months was 9.4% (95% CI, 7.7% to 11.5%) versus 26.3% (95% CI, 21.2% to 31.8%); corresponding 10-year estimates were 26.2% (95% CI, 21.5% to 30.8%) versus 55.9% (95% CI, 48.9% to 63.0%), respectively (P < .001 for both). IBF exhibited minimal change in performance across various follow-up durations. Conclusion IBF is the single most robust prognostic factor for PCM following RT without androgen deprivation therapy. This external validation demonstrates that patients and clinicians can use this information to make decisions about subsequent treatments. J Clin Oncol 30:1857-1863. (c) 2012 by American Society of Clinical Oncology