Diagnostic Performance of a Combined KLK2, miR-21, PSA, CRP, Albumin, AGR, and GGT Biomarker Panel for Distinguishing Prostate Cancer from Benign Prostatic Hyperplasia in Nigerian Men
Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background: Prostate-specific antigen (PSA) has limited specificity for distinguishing prostate cancer from benign prostatic hyperplasia (BPH) in African populations, where high baseline PSA and a high background prevalence of BPH inflate false-positive results. Whether a combined panel of PSA, Kallikrein-related peptidase 2 (KLK2), miR-21, C-reactive protein (CRP), albumin, the albumin-to-globulin ratio (AGR), and gamma-glutamyl transferase (GGT) improves on PSA alone for distinguishing malignant from non-malignant prostatic disease in Nigerian men has not been established. Methods: In this cross-sectional case-control study, 90 men (30 histologically confirmed prostate cancer cases, 30 BPH cases, and 30 healthy controls) were recruited from four hospitals in Abuja, Nigeria. Serum PSA, KLK2, CRP, albumin, and GGT were measured by immunoassay, circulating miR-21 by quantitative PCR, and AGR calculated from albumin and globulin. Receiver operating characteristic (ROC) curve analysis established the diagnostic accuracy of each biomarker and of five incremental combined models against a non-malignant reference group (BPH and healthy controls combined), with the DeLong test used to compare correlated areas under the curve (AUC). Firth's penalized logistic regression identified independent predictors of cancer status, and five-fold cross-validation assessed model stability. Results: PSA, KLK2, and CRP each differed significantly across every pairwise group comparison, including BPH versus cancer (all p<0.01). PSA alone achieved the highest diagnostic accuracy (AUC 0.990; sensitivity 93.3%, specificity 100.0%), exceeding KLK2 (AUC 0.913), CRP (AUC 0.868), GGT (AUC 0.721), miR-21 (AUC 0.688), and AGR (AUC 0.502). Each of PSA, KLK2, miR-21, CRP, and GGT was independently associated with cancer status on univariate analysis (all p<0.05), but none retained significance after mutual adjustment in a multivariable model, reflecting strong intercorrelation among PSA, KLK2, and CRP (Spearman rho 0.78-0.81). Combining KLK2, miR-21, CRP, and the biochemical markers with PSA did not improve diagnostic accuracy over PSA alone, and the full combined panel achieved a significantly lower AUC than PSA alone (DeLong test, p=0.029). A reduced five-variable model (PSA, KLK2, miR-21, CRP, age) nonetheless achieved strong internal discrimination (AUC 0.901) and was translated into a nomogram. Conclusions: In this late-presenting Nigerian cohort, PSA alone achieved near-ceiling diagnostic accuracy for distinguishing prostate cancer from non-malignant prostatic disease, and combining it with KLK2, miR-21, CRP, and biochemical markers did not improve, and numerically reduced, accuracy. KLK2 and CRP nonetheless carry genuine, independently significant diagnostic information and distinguished cancer from BPH on direct pairwise comparison, and a reduced composite model built entirely on locally available assays may hold feasibility-stage value for earlier-presenting or grey-zone patients, pending external validation.
Source: Research Square
Language: English
Subfield: Urology
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7204544950