The validation results for APhiGT algorithm for clarification of prostate cancer staging before treatment (first step)

• Main Authors: N. S. Sergeeva, T. E. Skachkova, N. V. Marshutina, K. M. Nushko, I. M. Shevchuk, M. R. Nazirov, S. A. Pirogov, E. F. Yurkov, V. G. Gitis, B. Ya. Alekseev, A. D. Kaprin
• Format: Article
• Language: Russian
• Published: ABV-press 2019-07-01
• Series: Onkourologiâ
• Subjects: prostate cancer; staging before treatment; aphigt algorithm; validation
• Online Access: https://oncourology.abvpress.ru/oncur/article/view/948

Background. We have previously described an algorithm APhiGT (Age, Prostate Health index, Gleason score, TNM stage) for staging of prostate cancer (PC) before treatment. The algorithm was developed by logistic regression on an educational selection (ES) of 337 PC cases. The algorithm includes data about the age of patients, the levels of total prostate-specific antigen (PSA), free PSA, [-2]proPSA and the ranked data of the Gleason score (by biopsy results) and T (by TNM).Objective. Validation of APhiGT on the validation selection (VS) of 83 PC cases was carried out in this work.Materials and methods. ROC analysis was performed in ES and VS.Results and сonclusion. It is established that area under the curve (AUC), characterizing the ability to divide clinically significant subgroups of patients (Gleason score <7 vs. Gleason score ≥7, рТ2 vs. рТ3, localized indolent PC vs. localized aggressive PC) for APhiGT both in ES and VS was significantly higher than AUC for total PSA, %[-2]proPSA in free PSA and prostate health index. At the same time, in all clinical subgroups of patients AUC for VS was lower than AUC for ES, which may be due to a significantly smaller size of VS compared to ES.