Retrospective analysis of serum testosterone levels by LC-MS/MS in chemically castrated prostate cancer patients: Biological variation and analytical performance specifications

•An LC-MS/MS assay for the quantitation of castration levels of testosterone was validated.•Within- and between patient variation was estimated using 474 samples from 72 patients.•Biological variation was estimated at 6.1%, 24.6% and 60.3% for CVA, CVI and CVG, respectively.•Desirable APS were 12.3% for imprecision, 16.3% for bias and 26.4% for total error. A sensitive liquid chromatography tandem-mass spectrometry (LC-MS/MS) method was used to monitor serum testosterone levels in castrated prostate cancer patients. We subsequently performed an observational and retrospective study to estimate the within- and between-subject biological variation of these patients. In total, 474 samples from 72 prostate cancer patients in the Netherlands receiving either chemical castration (CAS) or castration plus enzalutamide (ENZA) treatment were selected for data analysis. ANOVA was performed to estimate analytical variation (CVA) and within-patient variation (CVI). A nested ANOVA was applied to estimate between-patient variation (CVG). From these data, the reference change value (RCV) and analytical performance specifications (APS) were calculated. Testosterone levels were significantly higher in the ENZA group (0.318 vs. 0.191 nmol/L, p