Table 2.
MRI prediction of reclassification on active surveillance
| Author | Baseline MRI likelihood ratio | Surveillance MRI likelihood ratio | ||||||
|---|---|---|---|---|---|---|---|---|
| Amin [22] | PPV | 38% | LR + | 2.83 [1.03–7.78] | PPV | 31% | LR + | 5.33 [2.62–11] |
| NPV | 84% | LR − | 0.83 [0.65–1.06] | NPV | 90% | LR − | 0.45 [0.26–0.78] | |
| Chesnut [27] | PPV | NA | LR + | NA | PPV | 41% | LR + | 1.41 [1.21–1.64] |
| NPV | LR − | NPV | 85% | LR − | 0.24 [0.10–0.57] | |||
| Gallagher [24] | PPV | NA | LR + | NA | PPV | 23% | LR + | 1.68 [1.41–2.00] |
| NPV | LR − | NPV | 98% | LR − | 0.10 [0.01–0.70] | |||
| Jayadevan [14] | PPV | NA | LR + | NA | PPV | 11% | LR + | 0.86 [0.67–1.11] |
| NPV | LR − | NPV | 83% | LR − | 1.33 [0.88–2.00] | |||
| Nougaret [34] | PPV | NA | LR + | NA | PPV | 68% | LR + | 5.33 [4.03–7.03] |
| NPV | LR − | 95% | LR − | 0.13 [0.08–0.23] | ||||
| Osses [26] | PPV | NA | LR + | NA | PPV | 48% | LR + | 1.97 [1.48–2.64] |
| NPV | LR − | NPV | 90% | LR − | 0.25 [0.11–0.58] | |||
| Pepe [32] | PPV | NA | LR + | NA | PPV | 54% | LR + | 4.8 [1.88–12.00] |
| NPV | LR − | NPV | 91% | LR − | 0.39 [0.15–0.98] | |||
| Kornberg [21] | PPV | 41% | LR + | 1.29 [1.15–1.45] | PPV | NA | LR + | NA |
| NPV | 85% | LR − | 0.34 [0.19–0.62] | NPV | LR − | |||
All studies enrolled GG1 prostate cancer on active surveillance enrollment; 2 × 2 table values were calculated utilizing ≥ GG2 as the definition of reclassification. GRADE calculated according to template available at: https://www.gradeworkinggroup.org/. LR + = Positive likelihood ratio, LR + = True positive rate/false positive rate = Sensitivity/(1 − Specificity). LR − = Negative likelihood ratio, LR − = False negative rate/True negative rate = (1 − Sensitivity)/Specificity. Likelihood ratios of surveillance MRI serve to modify initial risk assessment of baseline MRI to give a probability of prostate cancer [28]
PPV positive predictive value, NPV negative predictive value