Prognostic value of MRI-derived features for biochemical recurrence after curative radiotherapy in prostate cancer.

BACKGROUND

To evaluate whether pretreatment multiparametric MRI features can predict biochemical recurrence in patients with localized prostate cancer treated with radiation therapy of curative intent.

METHODS

This retrospective single-institution study analyzed patients with prostate cancer who underwent primary radiotherapy between January 2015 and December 2019. All patients underwent pretreatment multiparametric MRI, with or without endorectal coil. Clinical, pathological and MRI features were assessed, with emphasis on qualitative and quantitative imaging characteristics. Univariate binary logistic regression identified significant predictors of recurrence. Variables showing significance (p < 0.05) were included in forward stepwise logistic regression analysis. The primary endpoint was biochemical recurrence, defined by Phoenix criteria.

RESULTS

The study included 54 patients (mean age 76.57 ± 7.77 years), with 10 patients (18.5%) experiencing biochemical recurrence at a median follow-up of 74.4 months. In univariate analysis, clinical T-stage (OR: 2.24, 95% CI: 1.11-4.56), ISUP grade (OR: 1.86, 95% CI: 1.03-3.37) and NCCN risk groups (OR: 2.55, 95% CI: 1.16-5.62) were significant predictors. Among MRI parameters, total tumor volume (OR: 4.28, 95% CI: 1.59-11.57), extraprostatic extension (OR: 8.57, 95% CI: 1.61-45.73), neurovascular bundle invasion (NBI) (OR: 14.78, 95% CI: 2.97-73.44), and seminal vesicle invasion (OR: 15.00, 95% CI: 2.94-76.56) showed strong associations with recurrence. In stepwise multivariate analysis, NBI emerged as the sole independent predictor (OR: 3.16, 95% CI: 1.59-6.28, p = 0.001), with 86.4% specificity and 92.7% negative predictive value.

CONCLUSION

NBI on pretreatment MRI independently predicts biochemical recurrence in patients with localized prostate cancer treated with radiation therapy, highlighting the need to refine risk stratification by incorporating imaging criteria.