Evolution of thyroid fine-needle aspiration biopsy performance over 15 years: implications for risk stratification in clinical practice.
We aimed to evaluate temporal changes in the diagnostic performance of thyroid fine-needle aspiration biopsy (FNAB) over a 15-year period in a single-center, surgically confirmed cohort.
We retrospectively analyzed thyroid FNABs performed between 2009 and 2023, dividing the data into two intervals (Group 1: 2009-2014; Group 2: 2015-2023). All specimens were interpreted by the same two experienced cytopathologists. Sensitivity, specificity, and Youden's J were calculated at four diagnostic thresholds (≥AUS/FLUS, ≥FN/SFN, ≥SM, and malignant only), and level-specific likelihood ratios (LRs) were compared between cohorts. Overall diagnostic discrimination was assessed with receiver operating characteristic (ROC) curves and compared by DeLong's test.
The study included 1,241 nodules; after excluding non-diagnostic cytology, 601 nodules in Group 1 and 553 in Group 2 were available for diagnostic performance analysis. At the ≥AUS/FLUS cutoff, sensitivity increased from 85.6% to 94.8%, while specificity declined from 58.4% to 33.0% (both p<0.01). Specificity at higher thresholds remained similar or increased in Group 2. The area under the ROC curve (AUC) was 0.799 for Group 1 and 0.831 for Group 2 (p = 0.24). Of the five Bethesda LRs, only that for AUS/FLUS changed significantly (1.24 vs. 0.37; p<0.01).
The overall discriminative ability of the Bethesda System remained stable over 15 years. The decrease in the AUS/FLUS LR reflects a spectrum effect driven by more selective surgical referral patterns and should be interpreted as a change in cohort composition rather than diminished test performance.
We retrospectively analyzed thyroid FNABs performed between 2009 and 2023, dividing the data into two intervals (Group 1: 2009-2014; Group 2: 2015-2023). All specimens were interpreted by the same two experienced cytopathologists. Sensitivity, specificity, and Youden's J were calculated at four diagnostic thresholds (≥AUS/FLUS, ≥FN/SFN, ≥SM, and malignant only), and level-specific likelihood ratios (LRs) were compared between cohorts. Overall diagnostic discrimination was assessed with receiver operating characteristic (ROC) curves and compared by DeLong's test.
The study included 1,241 nodules; after excluding non-diagnostic cytology, 601 nodules in Group 1 and 553 in Group 2 were available for diagnostic performance analysis. At the ≥AUS/FLUS cutoff, sensitivity increased from 85.6% to 94.8%, while specificity declined from 58.4% to 33.0% (both p<0.01). Specificity at higher thresholds remained similar or increased in Group 2. The area under the ROC curve (AUC) was 0.799 for Group 1 and 0.831 for Group 2 (p = 0.24). Of the five Bethesda LRs, only that for AUS/FLUS changed significantly (1.24 vs. 0.37; p<0.01).
The overall discriminative ability of the Bethesda System remained stable over 15 years. The decrease in the AUS/FLUS LR reflects a spectrum effect driven by more selective surgical referral patterns and should be interpreted as a change in cohort composition rather than diminished test performance.
Authors
Doğruel Doğruel, Karaoğlu Karaoğlu, Yılmaz Yılmaz, Aydemir Aydemir, Çalış Çalış, Arıcı Arıcı, Ocak Ocak, Gürer Gürer, Sarı Sarı
View on Pubmed