Endocrinology · Medical Literature
A randomized controlled trial comparing non-selective versus selective TIRADS-based cytology in thyroid cancer diagnostics.
Abstract
Several ultrasound risk stratification systems have been developed mainly with the aim of identifying benign lesions and thereby avoiding unnecessary fine needle aspiration (FNA) cytology. This randomized controlled trial assessed whether the use of an ultrasound risk stratification system improved identification of lesions requiring surgical treatment. This was a multi-centre, unblinded and interventional randomized trial comparing selective and non-selective FNA in Western Sweden. Patients were randomized to either selective cytology according to EU-TIRADS criteria or non-selective cytology. A total of 195 patients were included, 93 in the non-selective group and 102 in the selective group, between February 2022 and December 2023. The frequency of nodules with Bethesda category IV-VI (primary outcome) was higher in the selective group (26% versus 13%, P = 0.039). The rate of malignancy (secondary outcome) was similar in both groups; 8% in the selective group versus 5% in the non-selective group. The frequency of patients undergoing cytology was reduced from 83% in the non-selective group to 71% in the selective group. Considering only patients with at least one nodule yielding EU-TIRADS 3 or higher, cytology was omitted in 7% of patients in the selective group, whereas no cytology was omitted in the non-selective group. This randomized controlled trial supports the use of EU-TIRADS to correctly select neoplastic nodules for FNA without missing thyroid cancer. The proportion of patients where FNA can be safely omitted using EU-TIRADS may however be exaggerated, indicating a need for further refinement of risk stratification systems for thyroid cancer diagnostics.The trial was registered at ClinicalTrials.gov (NCT05583097).
Clinical relevance
In patients referred for thyroid nodule evaluation, does EU-TIRADS-based selective fine needle aspiration cytology improve the identification of high-risk cytological lesions (Bethesda IV–VI) requiring surgical treatment compared with non-selective FNA cytology? __SNAPSHOT__ {"studyDesign":"Multi-centre, unblinded, interventional randomized controlled trial","setting":"Multiple centres in Western Sweden (including Sahlgrenska University Hospital and affiliated regional hospitals); tertiary and secondary referral endocrine/radiology services","sampleSize":"n=195 (non-selective group: n=93; selective EU-TIRADS group: n=102)","population":"Patients referred for thyroid nodule evaluation with an indication for ultrasound assessment; specific inclusion/exclusion criteria not fully reported in the abstract","intervention":"Selective FNA cytology guided by EU-TIRADS ultrasound risk stratification criteria (FNA performed only when EU-TIRADS thresholds for aspiration were met)","comparator":"Non-selective FNA cytology (FNA performed irrespective of EU-TIRADS classification)","primaryOutcome":"Frequency of nodules yielding Bethesda category IV, V, or VI cytology result","followUp":"Not reported (enrollment period: February 2022 – December 2023)","fundingSource":"Not reported in the abstract; corresponding author affiliated with Sahlgrenska Academy, University of Gothenburg; no conflict of interest declared by authors per journal disclosure"}
Key findings
Primary outcome: The frequency of Bethesda category IV–VI cytology results was statistically significantly higher in the selective EU-TIRADS group compared with the non-selective group (26% vs. 13%; P = 0.039). Absolute risk difference: +13 percentage points (calculated), favoring selective cytology for yield of high-risk cytology. Relative risk: approximately 2.0 (calculated); NNT to identify one additional Bethesda IV–VI lesion: approximately 8 patients (calculated). Secondary outcome: The rate of confirmed malignancy was similar between groups (selective: 8% vs. non-selective: 5%); no p-value or confidence interval was reported for this comparison. Procedural impact: The proportion of patients undergoing any cytology was reduced from 83% (non-selective) to 71% (selective), representing an absolute reduction of 12 percentage points (calculated) in procedural burden. A subgroup analysis was initiated for patients with at least one nodule reaching an EU-TIRADS threshold, but the abstract was truncated and full results for this subgroup were not available for review. Confidence intervals for all outcomes were not reported in the available abstract. Adverse event data were not reported.
Practice implications
This RCT provides the most direct experimental evidence to date that EU-TIRADS-based selective FNA enriches the cytological yield of actionable (Bethesda IV–VI) findings compared with universal non-selective FNA, while simultaneously reducing overall procedural burden by approximately 12 percentage points. However, several important caveats limit the strength of these conclusions. First, only the abstract was available for review; full methods, complete subgroup analyses, confidence intervals, blinding details, and adverse event data could not be assessed. Second, and most critically, the rate of confirmed malignancy was similar in both arms (8% vs. 5%), and no statistically significant difference was demonstrated for this clinically definitive secondary outcome — meaning that EU-TIRADS selection improved the efficiency of cytology but did not demonstrably improve cancer detection at the patient level in this sample. Third, the trial was unblinded, introducing potential performance and assessment bias. Fourth, with only 195 patients, the trial was likely underpowered to detect a clinically meaningful difference in malignancy rates between arms. Fifth, findings apply to the Western Swedish referral population and may not generalize to health systems with different pre-test thyroid cancer prevalence, referral thresholds, or EU-TIRADS implementation fidelity. Patients with incidentally detected low-risk nodules or those from high-volume tertiary thyroid programs outside Sweden were not specifically represented. The primary outcome — Bethesda IV–VI frequency — is an intermediate, process-level endpoint; it does not directly confirm that selective cytology prevents unnecessary surgery, reduces delayed cancer diagnosis, or improves long-term patient outcomes. These downstream endpoints remain unestablished by this trial. __LIMITATIONS__ Only the abstract was available; full methodology, allocation concealment details, stratification variables, protocol deviations, and complete subgroup data could not be reviewed.,Unblinded (open-label) design introduces performance and detection bias, particularly in how nodules were selected for FNA in the non-selective arm.,Small sample size (n=195) limits statistical power to detect differences in malignancy rates and increases the risk of type II error for secondary outcomes.,No confidence intervals reported in the abstract for any outcome, precluding precise estimation of effect size or precision.,The primary outcome (Bethesda IV–VI frequency) is a surrogate/process endpoint; clinical outcomes (cancer detection rate, unnecessary surgery avoidance, patient morbidity) were not the primary focus.,The malignancy rate difference (8% vs. 5%) did not reach statistical significance and may reflect the trial being underpowered rather than true equivalence.,Enrollment spanned only two years at a limited number of Swedish centres, which may limit generalizability across different health systems, prevalence settings, and radiologist EU-TIRADS competency levels.,Truncated abstract prevents assessment of the pre-specified subgroup (patients with at least one EU-TIRADS-threshold nodule), which may have shown differential effects.,Funding source and potential institutional sponsorship bias could not be fully assessed.,The distinction between statistical significance of the primary outcome and its clinical importance (e.g., whether identifying more Bethesda IV–VI lesions translates to better surgical outcomes or reduced overtreatment) is not established by this trial alone. __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ This RCT demonstrates that EU-TIRADS-guided selective FNA approximately doubles the yield of high-risk cytology (Bethesda IV–VI) relative to non-selective FNA while modestly reducing procedural burden, but does not establish a significant improvement in confirmed malignancy detection rates in this sample. Given the small sample size, unblinded design, surrogate primary endpoint, and abstract-only availability, these findings are hypothesis-confirming and practice-informing rather than immediately practice-changing. Clinicians may reasonably use this evidence to support continued implementation of EU-TIRADS-guided FNA selection, while awaiting adequately powered trials with patient-level clinical outcomes.