Comment on: A randomized controlled trial comparing non-selective versus selective TIRADS-based cytology in thyroid cancer diagnostics

Correspondence to:​ Yangxian Xu (email: xuyangxian@163.com)

Department of Gastrointestinal Surgery

Longhua Hospital

Shanghai University of Traditional Chinese Medicine

Shanghai 200032

China

_____

BJS, https://doi.org/10.1093/bjs/znag076, published 20 June 2026

_____

We read with interest the randomized clinical trial (RCT) by Dahlberg et al.1. While this is a welcome contribution as the first randomized evidence for EU‑TIRADS, two methodological aspects deserve closer scrutiny.

First, the primary endpoint pools Bethesda IV with V–VI. Supplementary Table 1 shows that the higher rate of Bethesda IV–VI in the selective arm is largely attributable to Bethesda IV (14% vs 4%), whereas the difference in Bethesda V–VI combined is modest (5.6% vs 2.7%)1. According to the 2023 Bethesda System, category IV (follicular neoplasm) carries a malignancy risk of 30% (range 23–34%), but its recommended management—molecular testing ± diagnostic lobectomy—is fundamentally different from that of category V (74%, range 67–83%) and VI (97%, range 97–100%), which warrant direct surgery2. Merging these categories blurs clinically relevant differences in both risk and management. Using histologically confirmed malignancy as the primary endpoint would better capture the clinical benefit of selective FNA.

Second, 11% of nodules in the selective arm (versus 5% in controls) did not undergo FNA (Supplementary Table 1), leaving their malignant status unknown. Because the decision to withhold FNA was determined by the allocated strategy, this represents non‑ignorable missing data. The authors’ calculation of malignancy rates implicitly assumes these patients are cancer‑free, which weakens the claim of comparable oncological safety. Standard methodology recommends sensitivity analyses to assess whether the main finding is robust to alternative plausible assumptions about the malignant status of these non‑biopsied nodules.

These points do not detract from the value of this first RCT, but addressing them would strengthen the conclusions.