Longitudinal kinetics of human epididymis protein 4 in recurrent advanced ovarian carcinoma treated with platinum-based chemotherapy: mathematical modeling and prognostic implications

OBJECTIVE: Evidence on the clinical utility of baseline human epididymis protein 4 levels and their kinetics in recurrent ovarian cancer remains limited. This study evaluated longitudinal human epididymis protein 4 dynamics in real-world patients treated with platinum-based chemotherapy, aiming to determine the prognostic significance of human epididymis protein 4 kinetic parameters compared with CA125 response and established prognostic factors for progression-free and overall survival. METHODS: This retrospective analysis included 220 patients with recurrent epithelial ovarian cancer treated with second-line platinum-based chemotherapy between 2000 and 2020. Four mathematical models were tested to describe human epididymis protein 4 kinetics, with model selection based on goodness of fit, visual predictive checks, and prognostic performance. The prognostic value of the most informative human epididymis protein 4 kinetic parameter (the modeled residual human epididymis protein 4 level), was assessed in univariable and multi-variable analyses. RESULTS: Among the 93 assessable patients, no clear associations were observed between baseline human epididymis protein 4 levels and disease characteristics. Longitudinal human epididymis protein 4 kinetics differed from those of CA125 and showed an initial decline followed by a plateau at approximately 40 days. The best-fitting model was a mono-exponential decline incorporating the modeled residual human epididymis protein 4 parameter, which was subsequently evaluated as a prognostic factor. Baseline human epididymis protein 4 levels lacked prognostic significance for progression-free or overall survival. In contrast, a higher modeled residual human epididymis protein 4 level was significantly associated with shorter progression-free and overall survival in univariable analyses (residual human epididymis protein 4 ≥median vs