Introduction/Objective
Seminal vesicle invasion (SVI) is a significant prognostic factor in prostate cancer (PCa) associated with biochemical recurrence and metastasis [1]. This systematic review aims to determine the diagnostic accuracy, sensitivity, and specificity of magnetic resonance imaging (MRI) for predicting SVI in PCa.
Methods
Literature for this review was sourced from three databases: MEDLINE, EMBASE, and CENTRAL, from inception to February 2023. Search terms were used for PCa, SVI and MRI. Articles were included by two independent reviewers based on predetermined criteria. This included studies allowing calculation of sensitivity and specificity for MRI predicting SVI and studies comparing MRI to histopathology while excluding studies with non-conventional MRI. Conflict was settled by a third reviewer and articles were assessed for quality using QUADAS-2. Meta-analysis was performed using a random-effects model to evaluate the pooled data of the studies and subgroup analysis of patient factors.
Results
A total of 56 studies were included from 1968 articles after removal of 376 duplicates, exclusion of 1259 articles based on title and 277 articles excluded based on full text (Figure 1). This yielded a total cohort of 8596 patients and dates ranged from 1987 to 2023. Meta-analysis showed pooled sensitivity and specificity of 0.535 (95% CI 0.464-0.606) and 0.958 (95% CI 0.949 – 0.967) respectively. Pooled positive predictive value (PPV) and negative predictive value (NPV) were 0.695 (95% CI 0.635-0.755) and 0.918 (95% CI 0.898-0.937) respectively. Pooled accuracy for MRI prediction of SVI was 0.883 (95% CI 0.8605 – 0.9063). (Table 1) (Figure 2)
Conclusion
This systematic review shows a high specificity and NPV as well as moderate sensitivity and PPV for MRI in the prediction of SVI. Studies mentioned limitations with regards to cohort size, potential reader error, and changes in imaging protocol during the study. Further large-scale studies could be beneficial in providing guidance for use of MRI in PCa.