Background Multiparametric magnetic resonance imaging (MRI), with or withoutMRI-targeted biopsy, is an alternative test to systematic transrectal ultrasonography-guided biopsy in men suspected of having prostate cancer. At present, evidence on which test to use is insufficient to inform detailed evidence-based decision-making. Objectives To determine the diagnostic accuracy of the index tests MRI only, MRI-targeted biopsy, the MRI pathway (MRI with or without MRI-targeted biopsy) and systematic biopsy as compared to template-guided biopsy as the reference standard in detecting clinically significant prostate cancer as the target condition, defined as International Society of Urological Pathology (ISUP) grade 2 or higher. Secondary target conditions were the detection of grade 1 and grade 3 or higher-grade prostate cancer, and a potential change in the number of biopsy procedures. Search methods We performed a comprehensive systematic literature search up to 31 July 2018. We searched CENTRAL, MEDLINE, Embase, eight other databases and one trials register. Selection criteria We considered for inclusion any cross-sectional study if it investigated one or more index tests verified by the reference standard, or if it investigated the agreement between the MRI pathway and systematic biopsy, both performed in the same men. We included only studies on men who were biopsy naive or who previously had a negative biopsy (or a mix of both). Studies involving MRI had to report on both MRI-positive and MRI-negative men. All studies had to report on the primary target condition. Data collection and analysis Two reviewers independently extracted data and assessed the risk of bias using the QUADAS-2 tool. To estimate test accuracy, we calculated sensitivity and specificity using the bivariate model. To estimate agreement between theMRI pathway and systematic biopsy, we synthesised detection ratios by performing random-effects meta-analyses. To estimate the proportions of participants with prostate cancer detected by only one of the index tests, we used random-effects multinomial or binary logistic regression models. For the main comparisions, we assessed the certainty of evidence using GRADE. Main results The test accuracy analyses included 18 studies overall. MRI compared to template-guided biopsy: Based on a pooled sensitivity of 0.91 (95% confidence interval (CI): 0.83 to 0.95; 12 studies; low certainty of evidence) and a pooled specificity of 0.37 (95% CI: 0.29 to 0.46; 12 studies; low certainty of evidence) using a baseline prevalence of 30%, MRI may result in 273 (95% CI: 249 to 285) true positives, 441 false positives (95% CI: 378 to 497), 259 true negatives (95% CI: 203 to 322) and 27 (95% CI: 15 to 51) false negatives per 1000 men. We downgraded the certainty of evidence for study limitations and inconsistency. MRI-targeted biopsy compared to template-guided biopsy: Based on a pooled sensitivity of 0.80 (95% CI: 0.69 to 0.87; 8 studies; low certainty of evidence) and a pooled specificity of 0.94 (95% CI: 0.90 to 0.97; 8 studies; low certainty of evidence) using a baseline prevalence of 30%, MRI-targeted biopsy may result in 240 (95% CI: 207 to 261) true positives, 42 (95% CI: 21 to 70) false positives, 658 (95% CI: 630 to 679) true negatives and 60 (95% CI: 39 to 93) false negatives per 1000 men. We downgraded the certainty of evidence for study limitations and inconsistency. The MRI pathway compared to template-guided biopsy: Based on a pooled sensitivity of 0.72 (95% CI: 0.60 to 0.82; 8 studies; low certainty of evidence) and a pooled specificity of 0.96 (95% CI: 0.94 to 0.98; 8 studies; low certainty of evidence) using a baseline prevalence of 30%, the MRI pathway may result in 216 (95% CI: 180 to 246) true positives, 28 (95% CI: 14 to 42) false positives, 672 (95% CI: 658 to 686) true negatives and 84 (95% CI: 54 to 120) false negatives per 1000 men. We downgraded the certainty of evidence for study limitations, inconsistency and imprecision. Systemic biopsy compared to template-guided biopsy: Based on a pooled sensitivity of 0.63 (95% CI: 0.19 to 0.93; 4 studies; low certainty of evidence) and a pooled specificity of 1.00 (95% CI: 0.91 to 1.00; 4 studies; low certainty of evidence) using a baseline prevalence of 30%, systematic biopsy may result in 189 (95% CI: 57 to 279) true positives, 0 (95% CI: 0 to 63) false positives, 700 (95% CI: 637 to 700) true negatives and 111 (95% CI: 21 to 243) false negatives per 1000 men. We downgraded the certainty of evidence for study limitations and inconsistency. Agreement analyses: In a mixed population of both biopsy-naive and prior-negative biopsy men comparing the MRI pathway to systematic biopsy, we found a pooled detection ratio of 1.12 (95% CI: 1.02 to 1.23; 25 studies). We found pooled detection ratios of 1.44 (95% CI 1.19 to 1.75; 10 studies) in prior-negative biopsy men and 1.05 (95% CI: 0.95 to 1.16; 20 studies) in biopsy-naive men. Authors' conclusions Among the diagnostic strategies considered, the MRI pathway has the most favourable diagnostic accuracy in clinically significant prostate cancer detection. Compared to systematic biopsy, it increases the number of significant cancer detected while reducing the number of insignificant cancer diagnosed. The certainty in our findings was reduced by study limitations, specifically issues surrounding selection bias, as well as inconsistency. Based on these findings, further improvement of prostate cancer diagnostic pathways should be pursued.
