Radiologists agree on thyroid features using TI-RADS, with less-experienced experts leading the way

Radiologists largely agree on nodule characteristics when using the thyroid imaging reporting and data system, according to a new study, albeit with a slight experiential divide.

In fact, novice imaging experts agreed on a larger number of nodule characteristics, compared to their more experienced counterparts, according to a Feb. 25 study published in the American Journal of Roentgenology. These results suggest that younger practitioners may rely more heavily on reporting criteria, but that overall decision-making isn’t greatly affected.

“Although interreader concordance was higher among less experienced readers with respect to certain nodule features and categories, there was no substantial difference in the final TI-RADS level or recommendation for fine-needle aspiration on the basis of reader experience,” Ryan Chung, MD, with NYU Langone Health’s Department of Radiology, and colleagues wrote.

This is a good sign, given that up to 68% of adults have at least one nodule, and a majority of such findings are benign.

TI-RADS, which is relatively new, was developed to standardize thyroid nodule reporting and select only those findings that should be referred for biopsy based on “reasonable likelihood of biologically significant malignancy.” There are few studies out there, however, that have tested the reproducibility of the data system for achieving this goal, Chung et al. noted.

To address this knowledge gap, the researchers had three experienced and three less-experienced radiologists assess 150 thyroid nodules using TI-RADS.

Overall, the six experts showed “generally moderate” agreement for a range of TI-RADS categories, features and recommendations. And in terms of experience, early-career readers agreed on more characteristics, compared to later-stage radiologists. Specifically, that applied to margins, comet tail artifacts, punctate echogenic foci and the overall echogenic foci category.

“We postulate that this may be because of stricter adherence by the less experienced readers to the TI-RADS lexicon and guidelines owing to less influence from experience,” Chung and co-authors wrote. “That is, the experienced readers were perhaps subconsciously influenced by prior experience when evaluating these sonographic features to form their impressions.”

On the flip side, reproducibility was lowest for features such as echogenicity, margins and echogenic foci, the authors wrote. Future research should incorporate this data to help improve TI-RADS. The researchers suggested that directly educating readers on these items and clarifying, revising and expanding the definitions and criteria on these specific features may be effective approaches.

“The overall TI-RADS level and recommendation for FNA were unaffected, supporting the robustness of the TI-RADS lexicon and its continued use in practice,” the authors wrote. “Areas of identified relatively low interreader concordance could be the focus of targeted reader education and clarification in future TI-RADS updates to further standardize interpretations.”