Multimodal Nomogram Evaluates Clinically Significant Prostate Cancer Risk

Researchers have developed and externally validated an interpretable multimodal nomogram that combines routinely available magnetic resonance imaging findings and prostate-specific antigen biomarkers to assess individualized risk of clinically significant prostate cancer, defined as Grade Group 2 or higher. In a retrospective multicenter investigation spanning three clinical centers in China, the multivariable logistic regression model integrated patient age, PI-RADS score, free-to-total PSA ratio, and PSA density to distinguish clinically significant disease from lower-grade or benign conditions.

The clinical motivation outlined in the study focuses on the ongoing diagnostic challenge of accurately identifying Grade Group 2 or higher prostate cancer. Prostate-specific antigen testing alone lacks specificity, whereas relying exclusively on magnetic resonance imaging can lead to unnecessary prostate biopsies. To address these distinct limitations, the research team designed a predictive framework intended to improve risk discrimination while potentially sparing patients from invasive biopsy procedures.

The study cohorts comprised a primary development cohort of 260 patients from Center 1, located at The Third Affiliated Hospital of Shenzhen University, alongside two external validation cohorts comprising 91 patients from Center 2 at Shandong Cancer Hospital and Institute and 29 patients from Center 3 at the Cancer Hospital and Shenzhen Hospital of the Chinese Academy of Medical Sciences. Within the development cohort, 112 patients, representing 43.1 percent of the group, were diagnosed with Grade Group 2 or higher disease.

In internal evaluations, the multimodal model demonstrated an apparent area under the receiver operating characteristic curve of 0.872, with a 95 percent confidence interval of 0.823 to 0.914, and an optimism-corrected area under the curve of 0.864 following bootstrap internal validation. This predictive discrimination significantly exceeded that of PSA density evaluated alone, which recorded an area under the curve of 0.807, representing a statistically significant difference of 0.065 with a 95 percent confidence interval of 0.012 to 0.120 and a P value of 0.013. Goodness of fit and model calibration in the development cohort were reported as satisfactory, yielding a Hosmer-Lemeshow test P value of 0.797.

Reclassification and clinical utility analyses further favored the integrated nomogram over comparator diagnostic strategies. The model achieved a continuous net reclassification improvement of 0.789 and an integrated discrimination improvement of 0.206. Decision curve analysis indicated greater net benefit across evaluated threshold probabilities between 20 percent and 30 percent. Specifically, at a decision threshold of 25 percent, the nomogram avoided 17 biopsies per 100 patients without leading to any additional missed Grade Group 2 or higher cancers when compared against a PSA density threshold of 0.15 ng/mL/cm³.

External validation conducted without refitting the model yielded an area under the curve of 0.832 with a 95 percent confidence interval of 0.733 to 0.917 in Center 2, and an area under the curve of 0.880 with a 95 percent confidence interval of 0.690 to 1.000 in Center 3. While the authors concluded that the model demonstrates favorable discrimination, satisfactory development calibration, and potential clinical utility for predicting Grade Group 2 or higher prostate cancer, they noted that prospective validation across representative non-Chinese populations and diverse healthcare systems remains necessary prior to broader clinical adoption.

Subject of Research: Multimodal risk assessment model for clinically significant prostate cancer

Article Title: Development and external multicenter validation of a multimodal nomogram integrating MRI findings and PSA biomarkers for individualized risk assessment of clinically significant prostate cancer

Article References: Zhang, K., Wang, M., Zhang, J., Yuan, Y., Chen, Z., Teng, S., Zhang, S., Lin, M., Liu, H., & Wang, J. (2026). Development and external multicenter validation of a multimodal nomogram integrating MRI findings and PSA biomarkers for individualized risk assessment of clinically significant prostate cancer. Frontiers in Bioengineering and Biotechnology, 14, Article 1956142. https://doi.org/10.3389/fbioe.2026.1956142

Image Credits: AI Generated

DOI: 10.3389/fbioe.2026.1956142

Keywords: prostate cancer, magnetic resonance imaging, PSA density, PI-RADS, nomogram, multicenter validation, biopsy avoidance, Development, external, multicenter, validation, multimodal