Accuracy and Reproducibility of Histologic Features Predictive of BRAF V600E Mutation in Papillary Thyroid Carcinoma
Renu K Virk, Alexander Finkelstein, Avinash Prasad, Pei Hui, David Chhieng, Constantine Theoharis, JoAnna Gibson, Sanziana Roman, Manju Prasad. Yale School of Medicine, New Haven, CT; New York University, New York, NY; Hartford Hospital, Hartford, CT
Background: We proposed recently that papillary thyroid carcinomas (PTCs) with BRAF V600E mutation are morphologically distinct. Here we investigate the accuracy and interobserver reproducibility of a defined set of histological criteria in predicting BRAF V600E mutation.
Design: We created a training set of 5 PTCs with and 5 without BRAF V600E mutation. The former were classic, tall cell or subcapsular sclerosing variants, and showed well-developed nuclear features of PTC, tall or polygonal cells with moderate to abundant eosinophilic cytoplasm (plump pink cells), stromal fibrosis/sclerosis/desmoplasia, infiltrative tumor borders and psammoma bodies. The latter in general, were follicular variants with subtle nuclear features of PTC, and lacked most or all of the above mentioned histologic features of mutation positive tumors. After self-learning on the training set, two pathologists predicted the presence or absence of BRAF V600E mutation in 30 PTCs (test set) using the morphologic criteria learnt from the training set. The predictions were evaluated against BRAF V600E mutational analysis by single strand conformation polymorphism on tumor DNA.
Results: Table 1 shows the sensitivity, specificity, accuracy, and positive and negative predictive values of the histologic criteria for predicting BRAF V600E mutation by each pathologists.
|Pathologist 1||Pathologist 1|
|Sensitivity||15/15 (100%)||14/15 (93%)|
|Specificity||12/15 (80%)||12/15 (80%)|
|Accuracy||27/30 (90%)||26/30 (87%)|
|PPV||15/18 (83%)||14/17 (82%)|
|NPV||12/15 (80%)||12/16 (75%)|