Prostate cancer screening still leans heavily on prostate-specific antigen, or PSA, a blood protein that flags trouble but often cries wolf. A systematic review now published in the African Journal of Urology pulls together sixteen studies of emerging serum and urine protein biomarkers and concludes that several of them measurably outperform PSA-based models at telling aggressive cancers apart from benign disease. The work, led by Okorie Nnaemeka with Nur Aishah Che Roos and Armania Nurdin, was registered in PROSPERO and conducted under PRISMA guidelines, and its authors argue that these markers could complement rather than replace PSA testing.
The scale of the problem the review addresses is considerable. Prostate cancer is among the most frequently diagnosed malignancies in men worldwide, with roughly 1.4 million new cases and 375,000 deaths reported annually. PSA screening has high sensitivity but low specificity, meaning it detects many cancers but also rises in benign prostatic hyperplasia and prostatitis. That weakness has driven substantial overdiagnosis of indolent disease and unnecessary biopsies, which is precisely the gap the reviewed biomarkers are meant to close.
The search covered PubMed, Scopus and ScienceDirect for studies published between January 2006 and December 2025. From 1,543 records, the reviewers removed 215 duplicates, screened 1,328 titles and abstracts, excluded 1,307, and assessed 21 full texts before settling on sixteen eligible cohort, case-control and cross-sectional studies. Two reviewers screened and extracted data independently, with a third resolving disagreements, and study quality was graded with the QUADAS-2 tool. Most studies showed moderate-to-low overall risk of bias, although two raised concerns in patient selection and six in flow and timing, mainly through incomplete reporting of diagnostic datasets.
Among PSA derivatives, the Prostate Health Index, or PHI, emerged as a consistent performer. Friedersdorff and colleagues found PHI outperformed percent free PSA in men with low testosterone, reaching an AUC of 0.73, and combining PHI with bioavailable testosterone pushed the model to 0.76, then to 0.84 with a testosterone cutoff of 8 nmol/L. Cantiello and colleagues showed that adding PHI to a clinical model significantly improved prediction of aggressive features including tumour volume above 0.5 mL, seminal vesicle invasion and pathologic Gleason score of 7 or higher. Garrido and colleagues reported that PSA density, PHI and PHI density all achieved higher AUCs than PSA alone for detecting cancer and clinically significant cancer, with density-based indices adding roughly seven percent predictive improvement over a PSA baseline model.
Outside the PSA family, the serum PSP94-to-PSA ratio stood out. Mhatre and colleagues measured a roughly 14.4-fold reduction in that ratio in prostate cancer and a 3.2-fold reduction in benign prostatic hyperplasia relative to healthy controls, and the ratio discriminated cancer from benign disease with an AUC of 0.859, sensitivity of 90.9 percent and specificity of 70.5 percent. Nam and colleagues found serum PSP94 about 1.3-fold lower in cancer patients than in healthy men, and combining PSP94 with free testosterone raised the AUC from 0.66 to 0.74 in a low-risk cohort of 649 men. Notably, PSP94, unlike PSA and free testosterone, distinguished aggressive from non-aggressive disease at cutoffs preserving 85 and 90 percent sensitivity.
Extracellular vesicle markers added another layer. Logozzi and colleagues reported plasma exosomal PSA levels 4.5 to 5.0-fold higher in cancer than in benign disease and controls, though the review cautions that exosomal PSA findings rest on small samples without external validation. Chung and colleagues found serum EV-associated EDIL3, also called Del-1, elevated 1.8-fold in cancer versus benign disease, and in the PSA gray zone of 3 to 8 ng/mL, 88.9 percent of cancer patients showed Del-1 values at or above 0.120 against 11.1 percent of benign cases, with specificity of 93.5 percent. Krishn and colleagues identified the EV-associated αvβ3 integrin as a tumour-enriched marker, although that study did not formally report sensitivity or AUC.
Urine-based markers broadened the non-invasive toolkit. Schostak and colleagues showed urinary ANXA3 achieved AUCs of 0.687 overall, 0.725 in the 2 to 6 ng/mL PSA subgroup and 0.755 in the 4 to 10 ng/mL gray zone, and combining ANXA3 with free PSA lifted the gray-zone AUC to 0.832. De Souza and colleagues validated uromodulin peptide panels with sensitivity of 78.79 to 84.85 percent and specificity of 76 to 88 percent for separating cancer from benign disease, with some peptide combinations reaching AUCs of 0.847 to 0.852 against PSA’s 0.598. Shrivastava and colleagues reported a two-fold reduction in post-DRE urinary MSMB in cancer, with the post-DRE PSA/MSMB ratio yielding an AUC of 0.75.
For prognostic stratification, Yamamichi and colleagues identified the GDF15 propeptide as a robust marker of bone metastasis in castration-resistant disease, achieving an AUC of 0.92 across two randomised cohorts, well above PSA, bone-specific alkaline phosphatase and lactate dehydrogenase. Shore and colleagues built the MiCheck panel, combining PSA with IL-7, leptin, VEGF and glypican-1, which reached an AUC of 0.83 for distinguishing aggressive from non-aggressive cancer and, by the authors’ modelling, could avoid up to 30 percent of unnecessary biopsies while delaying diagnosis of only 6.8 percent of clinically significant cancers. The review notes some multivariable combinations achieved AUCs as high as 0.979.
The authors are careful about limits. Heterogeneity in study design, biospecimen handling and assay platforms prevented meta-analysis, so findings were synthesised narratively. Many candidate markers come from small, single-centre cohorts without independent external validation, and no biomarker showed consistent population- or ethnicity-specific utility across the reviewed geographic spread. Metabolic markers such as 25-hydroxyvitamin D, which predicted benign prostatic hyperplasia with an AUC of 0.74 in a Chinese cohort, and oxidized LDL, which differentiated cancer from benign disease with an AUC of 0.766 in a Ghanaian cohort, illustrate both the breadth of candidates and the fragility of single-population findings.
The bottom line is measured. No emerging biomarker is yet ready to replace PSA in routine practice, and none has been conclusively validated as a standalone test across ethnicities and regions. But the review argues that serum, urine and extracellular vesicle protein markers, particularly MSMB/PSP94, the GDF15 propeptide, ANXA3 and uromodulin peptides, show real potential as adjuncts that could raise specificity, cut unnecessary biopsies and sharpen risk stratification. The most promising path forward, the authors conclude, lies in validated multi-marker panels combining molecular signals with existing clinical parameters, backed by standardised assays and large-scale prospective multicentre validation before clinical adoption.
Subject of Research: Emerging serum and urine protein biomarkers for non-invasive prostate cancer detection and risk stratification
Article Title: Emerging serum and urine biomarkers for non-invasive detection, risk stratification and prognostic assessment of prostate cancer: a systematic review
Article References: Nnaemeka, O., Che Roos, N. A., & Nurdin, A. (2026). Emerging serum and urine biomarkers for non-invasive detection, risk stratification and prognostic assessment of prostate cancer: a systematic review. African Journal of Urology, 32(1), Article 57. https://doi.org/10.1186/s12301-026-00605-2
Image Credits: AI Generated
DOI: 10.1186/s12301-026-00605-2
Keywords: prostate cancer, biomarkers, PSA, systematic review, Prostate Health Index, PSP94, ANXA3, uromodulin peptides, GDF15, extracellular vesicles, risk stratification, liquid biopsy
