178 chapter 2 GEC in Bethesda IV nodules In eight studies, 210 Bethesda IV nodules were reported [164, 167-171, 180, 182]. There were 28% (59/210) malignancies in the Bethesda IV group: 10 PTC, 13 FVPTC, five FTC, seven FTC-OV, one MTC and 23 malignancies of unspecified subtype. Remarkably, the malignancy rate was lower than in Bethesda III nodules, presumably due to higher numbers of selectively withheld surgery for Bethesda III nodules that were presumed benign. Estimated pooled sensitivity, specificity, positive and negative LR are 93.6% (95% CI: 82.0%-97.9%), 16.2% (95% CI: 6.8%-34.0%), 1.12 (95% CI: 0.94-1.33) and 0.39 (95% CI: 0.10-1.49), respectively. The AUC is 0.92 (95% CI: 0.89-0.94) (Table 37 and Figure 43 and 44). Best-case and worst-case scenarios As stated before, the high proportion of missing histopathology for the GEC is a major problem in the interpretation of its performance. Interaction between the index and the reference test likely introduced bias. The missing histology mainly comprised patients with a negative GEC result, likely resulting in an overestimation of its sensitivity (i.e., by missing at least some malignancies in the many unoperated GEC-negative nodules) and an underestimation of its specificity (i.e., relatively more GECpositive nodules with benign histology (false-positives) were operated on than GEC-negative nodules with benign histology (true-negatives)). The trend that studies with higher surgical rates for GECnegative nodules showed more moderate results supports these hypotheses [164, 168, 180]. On the other hand, NPV may also be underestimated if suspicious clinical or radiological signs influenced surgical decision-making despite a negative GEC result, pursuing surgery for high-risk patients. This would result in relatively more false-negative results. The exact opposite holds true for clinically benign, GEC-positive nodules that did not undergo surgery, leading to overestimation of the PPV. Yet all Figure 42. SROC curve of the Afirma® GEC – Bethesda III Summary receiver operating characteristic plot showing sensitivity versus 1-specificity of the Afirma® GEC in Bethesda III thyroid nodules with available histopathology. AUC = 0.82 (95% CI: 0.78-0.85). AUC, area under the curve; HSROC, hierarchical summary receiver operating characteristic.
RkJQdWJsaXNoZXIy MTk4NDMw