Time to Nadir PSA: Of Popes and PSA--The Immortality Bias

The objective of the study was to investigate prostate-specific antigen (PSA) nadir (nPSA) and time to nPSA (TnPSA) as prognostic variables for outcomes after definitive high-dose (>75 Gy) external beam radiation therapy (RT) without androgen deprivation therapy while correcting for immortal-time bias.

nPSA and TnPSA were available for 410 patients. nPSA and TnPSA's impact on freedom from biochemical failure, freedom from metastasis, and prostate cancer-specific survival was assessed on univariate and multivariate analysis using Kaplan-Meier and Cox proportional hazards regression. Outcomes were also evaluated relative to the time since achieving nPSA and not relative to the time of RT, given the intrinsic time bias in TnPSA.

Median nPSA was 0.7 ng/mL (interquartile range: 0.4 to 1.1), with a median TnPSA of 25.0 months (IQR: 15.0 to 40.0). On univariate analysis both nPSA and TnPSA were predictive of all endpoints: freedom from biochemical failure, freedom from metastasis, and prostate cancer-specific survival, as categorical (all P<0.0001) and continuous (all P<0.01) variables. However, after adjusting for immortal-time bias the benefit of long TnPSA was mostly lost. On Cox proportional hazards, a TnPSA<12 months did have worse prognosis for biochemical failure and distant metastasis as compared with longer TnPSA, but for those who achieved nadir >12 months from the time of RT the TnPSA was no longer prognostic.

For dose-escalated RT a lower nPSA is prognostic, but the benefit of a long TnPSA is largely an immortal-time bias and that a longer TnPSA is not in and of itself a significant favorable factor except as compared with those with the shortest TnPSA of <12 months.