163 Diagnostic utility of molecular and imaging biomarkers 2 Separate meta-analysis of the five miRInform® thyroid studies showed a slightly better performance of the commercial test than the other methods, with a pooled 58.1% sensitivity (95% CI: 50.7%- 65.2%), 95.2% specificity (95% CI: 89.2%-98.0%), 12.21 positive LR (95% CI: 5.38-27.75) and 0.44 negative LR (95% CI: 0.37-0.52) at a prevalence of malignancy of 23.6% (278/1,179) (Table 28, Figure 33 and Figure 34). For a given prevalence of malignancy of 15%, 25% or 40%, these results correspond to an estimated PPV and NPV of 68.3% (95% CI: 48.7%-83.0%) and 92.8% (95% CI: 91.6%-93.8%), 80.3% (95% CI: 64.2%-90.2%) and 87.2% (95% CI: 85.2%-89.0%), or 89.1% (95% CI: 78.2%-94.9%) and 77.3% (95% CI: 74.2%-80.1%), respectively. 7-gene GMP in Bethesda III nodules Three studies reported results separately for Bethesda III nodules. They all used the miRInform® thyroid test [69, 97, 114]. A total of 336 Bethesda III nodules were included, with a malignancy rate of 15% (49/336). Only 33 (9.8%) nodules had a positive GMP test result. Histology was available in 86.7% (288/332) of nodules with a conclusive index test, including 31 (94%) of the GMP-positive nodules. Estimated pooled sensitivity and specificity using a random effects model are 35% (95% CI: 0%-71%) and 98% (95% CI: 96%-100%), respectively (Table 29, Figure 35). Figure 34. SROC curve of miRInform® analysis Summary receiver operating characteristic plot showing sensitivity versus 1-specificity of miRInform® analysis in indeterminate thyroid nodules with available histopathology. 0.67 (95% CI: 0.630.71). AUC, area under the curve; HSROC, hierarchical summary receiver operating characteristic.
RkJQdWJsaXNoZXIy MTk4NDMw