206 chapter 2 Table 47. Estimated pooled test performance of CK-19 ICC in indeterminate thyroid nodules ni test + test - test ? nH (%) TP FP FN TN malignant (TP+FN/ni) Khurana 2003 19 7 12 19 (100%) 7 0 1 11 44% Saggiorato 2005 125 62 63 125 (100%) 57 5 18 45 60% Lacoste-Collin 2014 31 10 21 31 (100%) 5 5 1 20 19% POOLED RESULTS 175 79 96 0 175 (100%) 69 10 20 76 50.9% FN: false negative. FP: false positive. LR+: positive likelihood ratio. LR-: negative likelihood ratio. nH: number of indeterminate thyroid nodules with conclusive index test and histopathological follow-up. ni: number of indeterminate thyroid nodules. test+: number of positive index tests. test-: number of negative index tests. test?: number of unknown or nondiagnostic index tests. TN: true negative. TP: true positive. test was considered positive if galectin-3, HBME-1 or both were positive; alternatively, the test was only considered positive if there was simultaneous positivity for both stains. The first method had an excellent sensitivity and true benign results (dual negative stain) in 41.4% of patients (109/263). The estimated pooled sensitivity, specificity, positive LR and negative LR are 96.6% (95% CI: 90.8%-98.8%), 78.6% (95% CI: 67.9%-86.4%), 4.51 (95% CI: 2.92-6.97) and 0.04 (95% CI: 0.01-0.12), respectively. The AUC is 0.97 (95% CI: 0.95-0.98) (Table 48, Figure 60, Figure 61, Figure 62). For a given prevalence of malignancy of 15%, 25% or 40%, these results correspond to an estimated PPV and NPV of 44.3% (95% CI: 34.0%-55.2%) and 99.3% (95% CI: 97.9%-99.7%), 60.0% (95% CI: 49.3%-69.9%) and 98.6% (95% CI: 96.1%-99.5%), or 75.0% (95% CI: 66.0%-82.3%) and 97.2% (95% CI: 92.5%-99.0%), respectively. The alternative method of double-positivity had a very high specificity and positive LR, and a true malignant call rate of 35% (92/263). The estimated pooled sensitivity, specificity, positive LR and negative LR are 76.7% (95% CI: 62.2%-86.8%), 94.3% (95% CI: 79.4%-98.6%), 13.47 (95% CI: 3.6050.42) and 0.25 (95% CI: 0.15-0.41), respectively. The AUC is 0.89 (95% CI: 0.86-0.91) (Table 49, Figure 63, Figure 64 and Figure 65)). For a given prevalence of malignancy of 15%, 25% or 40%, these results correspond to an estimated PPV and NPV of 70.4% (95% CI: 38.9%-89.9%) and 95.8% (95% CI: 93.3%-97.4%), 81.8% (95% CI: 54.6%-94.4%) and 92.4% (95% CI: 88.0%-95.3%), or 90.0% (95% CI: 70.6%-97.1%) and 85.9% (95% CI: 78.6%-91.0%), respectively. Combining these modes of interpretation would provide 76.4% (201/263) of the population with a conclusive benign or malignant diagnosis. An inconclusive ICC result (positive for one of the stains) would remain in 24% (62/263) of patients at a 53% (33/62) post-test probability of malignancy, for which hemithyroidectomy would be indicated. Highly promising, additional large prospective trials are warranted to validate this diagnostic approach.
RkJQdWJsaXNoZXIy MTk4NDMw