Multiplanar versus axial measurement of prostate lesion size: agreement and impact on PI-RADS v2.1 categorization and histopathologic correlation

Abstract

Purpose

To evaluate whether measurement of prostate lesion size on sagittal or coronal imaging planes influences PI-RADS v2.1 categorization and histopathologic correlation compared with axial measurement.

Methods

In this single-center retrospective study, 261 prostate lesions categorized as PI-RADS 4 or 5 underwent subsequent targeted MRI/TRUS fusion biopsy or prostatectomy. Lesion size was measured as maximal axial diameter (MAD) and maximal coronal or sagittal diameter (MCSD). Lesions were subsequently categorized according to PI-RADS v2.1 using these measurements. The agreement between MAD and MCSD was assessed using a Bland–Altman analysis. Measurement reliability and interobserver agreement were evaluated using intraclass correlation coefficients (ICC). Kappa statistics were used to assess agreement in PI-RADS categorization for clinically significant prostate cancer.

Result

Bland–Altman analysis demonstrated minimal bias and strong agreement between MAD and MCSD measurements for the whole prostate (mean difference 0.28 mm; 95% limits of agreement − 2.62 to 3.19 mm), with similar results in both the peripheral and transition zones. Intraclass correlation analysis showed excellent reliability between measurements (ICC = 0.93). Agreement in PI-RADS categorization between MAD and MCSD was almost perfect (κ = 0.931). Reclassification from PI-RADS 4 to 5 occurred in 1.74% of lesions when using MCSD, and all reclassified lesions demonstrated clinically significant prostate cancer on histopathology.

Conclusion

MAD and MCSD measurements of prostate lesion size demonstrate excellent agreement and highly concordant PI-RADS categorization. Although reclassification is uncommon, evaluation of the sagittal or coronal planes may provide additional value for selected lesions with craniocaudal elongation, where the maximal tumor extent may not be fully represented on axial images.