Ultrasound Risk Scores Track Cytology Results in Low-Risk Thyroid Nodules

Thyroid nodules are among the most common findings in clinical medicine, and high-resolution ultrasound has made them harder to escape noticing. By some estimates, nodules are detectable in 60 to 70 percent of adults, yet only a small fraction, typically between 5 and 15 percent, turn out to be malignant. The challenge for clinicians is therefore not finding nodules but deciding which of them deserve a needle. A new prospective study from Sulaymaniyah Teaching Hospital in Iraq now offers a detailed look at how two widely used low-risk ultrasound categories perform when put to the test of actual biopsy cytology, and the results are both reassuring and carefully qualified.

Twenty-seven patients fell into the TIRADS II group and sixty-three into TIRADS III.

The headline finding concerns benign cytology. Among TIRADS II nodules, 88.9 percent yielded Bethesda II results, the category that denotes benign findings, compared with 54.0 percent of TIRADS III nodules. That difference was statistically significant, with a Fisher exact test P value of 0.002. In other words, nodules that looked entirely benign on ultrasound, purely cystic or spongiform lesions in ACR terminology, were far more likely to produce reassuring cytology than their mildly suspicious counterparts. The complete two-by-six distribution of TIRADS categories against all six Bethesda categories, however, showed only a trend that did not reach the conventional significance threshold, with a Pearson chi-square of 10.67 on five degrees of freedom and a P value of 0.058.

The picture at the high-risk end was more nuanced. Bethesda V-VI cytology, meaning suspicious for malignancy or malignant, occurred in 3.7 percent of TIRADS II nodules and 17.5 percent of TIRADS III nodules. Although the proportion was nearly five times higher in the mildly suspicious group, the difference did not reach statistical significance, with a Fisher exact P value of 0.099. The authors attribute this partly to the small numbers involved: only twelve Bethesda V-VI outcomes existed in the entire cohort, eleven of them in the TIRADS III group. With events that scarce, even a visually striking difference in proportions cannot be distinguished reliably from sampling noise.

The resulting sensitivity was 91.7 percent, with a 95 percent confidence interval stretching from 61.5 to 99.8 percent, meaning that nearly all high-risk cytology results occurred in TIRADS III nodules. But specificity was only 33.3 percent, the positive predictive value just 17.5 percent, and overall accuracy 41.1 percent. The negative predictive value was the standout figure at 96.3 percent, with a confidence interval of 81.0 to 99.9 percent, suggesting that a TIRADS II classification among these biopsied nodules was highly likely to exclude high-risk cytology. The authors stress that the wide intervals, particularly for sensitivity and negative predictive value, reflect the small number of endpoint events.

An important interpretive caveat runs through the entire paper. Bethesda V denotes suspicious for malignancy, and Bethesda III and IV are indeterminate categories whose definitive resolution may require repeat cytology, molecular testing, surgery, or follow-up. The study therefore cannot report true malignancy rates, only the distribution of cytological categories, and the authors explicitly warn against reading their numbers as evidence for omitting or mandating biopsy outside established size and clinical criteria.

Selection dynamics also shape what the numbers mean. ACR-TIRADS generally does not recommend fine-needle aspiration for TIRADS II nodules, so the twenty-seven biopsied TIRADS II nodules in this study represent a clinically selected population; the specific non-imaging reasons that prompted their biopsy were not systematically categorized. This verification bias means the observed cytological rates in the TIRADS II group may overestimate the frequency of atypical or high-risk findings among all TIRADS II nodules. Similarly, the tertiary referral setting preferentially selected patients with additional clinical concerns, and institutional biopsy practices, patient or clinician preference, and unmeasured risk factors may have enriched the sample for higher-risk cytology. The authors note that regional epidemiological differences are possible but that their design cannot separate them from referral and selection effects.

Serum TSH, free T3, and free T4 were measured, and thyroid functional status was categorized, though no significant association emerged between thyroid function and cytological results.

The authors position their contribution as regional prospective external validation rather than a new diagnostic algorithm. Prospective data from Iraq and the Kurdistan Region had been limited, and low-risk TIRADS II and III nodules are underrepresented in biopsy-based series precisely because TIRADS II nodules rarely meet ACR biopsy criteria. Their results align with prior validation work: previous studies have reported strong concordance between low-risk ultrasound categories and benign cytology, systematic reviews and multi-institutional analyses have shown very low malignancy risk in lower ACR-TIRADS categories, and European risk-stratification systems support the broader principle that structured sonographic classification can reduce unnecessary biopsies. The observed 17.5 percent rate of high-risk cytology in TIRADS III nodules sits higher than some published low-suspicion series but remains within the broad range reported across ultrasound risk-stratification studies, though it rests on just eleven events.

The study’s limitations are stated plainly. It was single-center and relatively small, with wide confidence intervals and too few events for reliable adjusted modeling. Only patients who underwent biopsy were included, creating selection and verification bias that is most acute for TIRADS II nodules. Cytology could not be validated against a definitive histopathological reference standard, no longitudinal follow-up was available to confirm the long-term behavior of benign results, and a complete patient-level distribution of individual ultrasound features was not available for separate analysis. The restricted spectrum, containing only two adjacent low-risk categories, also means the diagnostic-performance indices describe a binary comparison within a narrow range rather than the performance of the full ACR-TIRADS system across its five tiers. The authors call for larger multicenter studies that enroll a broader spectrum of TIRADS categories, prospectively document biopsy indications, include independent reader agreement, report individual sonographic feature distributions, and obtain surgical histopathology or structured follow-up. Until such data exist, they conclude, management decisions should remain anchored to established guideline criteria and clinical context rather than to this cohort alone, even as the findings reinforce the value of standardized ultrasound risk stratification as a complement to cytological assessment.

Subject of Research: Correlation between ACR-TIRADS ultrasound categories and Bethesda fine-needle aspiration cytology results in low-risk thyroid nodules

Article Title: Correlation between ultrasound findings of thyroid imaging reporting and data system II and III thyroid nodules and fine-needle aspiration results

Article References: Roshen, T., & Mohammed, N. A. (2026). Correlation between ultrasound findings of thyroid imaging reporting and data system II and III thyroid nodules and fine-needle aspiration results. Egyptian Journal of Radiology and Nuclear Medicine, 57(1), Article 224. https://doi.org/10.1186/s43055-026-01887-w

Image Credits: AI Generated

DOI: 10.1186/s43055-026-01887-w

Keywords: thyroid nodules, ACR-TIRADS, fine-needle aspiration, Bethesda System, ultrasound, risk stratification, cytology, Iraq, prospective study, diagnostic accuracy, thyroid cancer, Correlation