119 Diagnostic utility of molecular and imaging biomarkers 2 RAS point mutation Our systematic literature search yielded 20 studies that investigated the various RAS mutations in a total of 3,112 indeterminate thyroid nodules [60, 67, 69, 75-77, 87, 88, 93, 97-100, 102, 107, 109, 114, 118, 128, 351] (Table 8). Most studies investigated HRAS, KRAS and NRAS mutations; one only tested for KRAS and NRAS [88]. Multiple studies describe that they only tested for RAS mutations when BRAF mutation analysis was negative [77, 98, 128]. Sequential testing is an accepted approach with regard to BRAF and RAS, since the mutations are mostly presumed mutually exclusive. Nineteen studies combined RAS mutation analysis with the assessment for other mutations or included this mutation in a multi-gene mutation panel [60, 67, 69, 75-77, 87, 88, 93, 97-100, 102, 107, 109, 114, 118, 351]. In our pooled data from 20 studies, a RAS mutation was found in 11.2% (350/3,112) of the indeterminate nodules. Mutation analysis yielded a nondiagnostic result in 2.4% (76/3,112); these cases were excluded from further analysis. Meta-analysis was performed of the 2,212 indeterminate nodules (72.9%, 2,212/3,036) for which a conclusive RAS mutation analysis and histopathological correlation were available. This included 83.7% (293/350) of the RAS mutation-positive and 71.4% (1,919/2,686) of the RAS mutation-negative nodules. Approximately one third (34.9%, 212/608) of all malignancies were RAS mutation positive, including 21 of 124 reported PTC (17%), 113 of 262 FVPTC (43%), 26 of 81 FTC (32%), 2 of 18 FTC-OV (11%), 1 of 3 MTC (33%), 1 of 2 mPTC, and 6 of 9 other types of thyroid malignancies (67%) (three oncocytic PTC and FVPTC [75], one tumour with a combination of PTC and FTC [76], one poorly differentiated carcinoma [351], and one (follicular) tumour of unknown malignant potential (FT-UMP) [109]). Five studies did not correlate the presence of RAS mutation to specified histopathological results: 42 of the 109 (39%) of the unspecified thyroid carcinoma in these studies were RAS mutation positive (Table 9) [60, 93, 98, 107, 118]. The frequency of RAS mutations in FVPTC was significantly higher than in PTC (Pearson χ2, p<0.0001) but not than in FTC (Pearson χ2, p=0.08). Eighty-one histopathologically benign nodules carried a RAS mutation and were considered false positive, including 44 follicular adenoma (54%), 4 Hürthle cell adenoma (5%), 14 hyperplastic nodules (17%), one colloid nodule (1%) and 18 unspecified benign lesions (22%). In individual studies, the sensitivity and specificity of RAS mutation analysis ranged from 0% to 77% and from 75% to 100%, respectively [77, 98, 128]. Visual analysis of the forest plots suggests between study variance; I2 is 84.3% for sensitivity and 75.9% for specificity. Estimated pooled sensitivity, specificity, positive and negative LR are 27.3% (95% CI: 18.3%-38.7%), 94.4% (95% CI: 91.5%-96.3%), 4.87 (95% CI: 2.67-8.90), and 0.77 (95% CI: 0.66-0.89), respectively (Table 10, Figure 12). The AUC is 0.83 (95% CI: 0.79-0.86) (Figure 13). For a given prevalence of malignancy of 15%, 25% or 40%, these results correspond to an estimated PPV and NPV of 46.2% (95% CI: 32.0%-61.1%) and 88.0% (95% CI: 86.5%-89.5%), 61.9% (95% CI: 47.1%-74.8%) and 79.6% (95% CI: 77.1%-81.9%), or 76.5% (95% CI: 64.0%-85.6%) and 66.1% (95% CI: 62.7%-69.3%), respectively (Figure 14).
RkJQdWJsaXNoZXIy MTk4NDMw