Proper management of subcentimeter thyroid nodules remains a great challenge for both clinicians and patients. Extensive research on ultrasonography, which is a safe and inexpensive tool for identifying thyroid nodules, is therefore important. While the 2015 ATA guidelines estimated a 70–90% malignancy risk for thyroid nodules > 1 cm in size that appear highly suspicious on US, data for subcentimeter nodules are still lacking. In our sample, 59.9% of highly suspicious thyroid nodules in the subcentimeter range were classified as malignant on postoperative pathological examination. We found that the characteristic US finding of taller-than-wide dimensions allows for the best prediction of malignancy; similar to the results of Frates et al. (2005) and Lee et al. (2011), when this feature was present, the risk of having a malignant tumour was approximately nine times greater than when it was absent. Other studies have also suggested that a taller-than-wide shape on US predicts malignancy with the highest specificity (Ito et al. 2007; Ahn et al. 2010). Specifically, the anteroposterior diameter of the thyroid cancer nodule is usually larger than the transverse diameter, which may be related to the fact that tumour cells in the anteroposterior direction are in the division phase, while those in other directions are relatively static (Moon et al. 2011).

The sensitivity of microcalcification for subcentimeter malignant nodule detection in our sample was low; a previous study reported a sensitivity of 15.2% (Popowicz et al. 2009). Furthermore, we did not observe rim calcifications with ETE or noticeable soft-tissue components, which indicates that these two US features are rarely expressed in subcentimeter thyroid nodules. Earlier studies have investigated risk factors for poor prognoses in patients with papillary thyroid microcarcinoma (PTMC) and have reported that ETE, capsular invasion, and neck lymph node metastasis are independently associated with distant metastasis and nodal recurrence (Mercante et al. 2009; Pisanu et al. 2015). However, the treating physician does not have easy access to such results when assessing patients for subcentimeter nodules before an intervention, which can lead to a poor prognosis. In addition, PTMC is no longer recognized by the WHO 5th edition as an independent histological subtype. This was also the case in this study, where we observed no capsular invasion or ETE and identified only 3 patients with suspicious neck lymph nodes before preoperative US, but lymph node metastases were observed in 57 of 333 malignant nodules by postoperative pathology confirmation. However, we also found that both the characteristic US findings of a taller-than-wide shape and age younger than 45 years were significant risk factors for subcentimeter malignant thyroid nodules with lymph node metastasis and the invasive biological behaviour of malignant tumours. These results are consistent with those of Baier et al. (2009) and Rago et al. (2010).

The presence of highly suspicious US findings helps determine which thyroid nodules require FNA biopsy and reduces the likelihood of over- or misdiagnosis. According to the usual guidelines, FNA should be performed when nodules are > 1 cm in size and have highly suspicious features. However, for thyroid nodules less than 1 cm, existing guidelines generally recommend AS. However, there is no clear documentation for follow-up management. Thus, there is a gap in the literature regarding the appropriate treatment for subcentimeter nodules. In other words, the available guidelines provide clear recommendations regarding thyroid biopsy, and they do not offer a clear and suitable management approach for patients who do not undergo a biopsy. Some studies found a significantly greater frequency of AS in patients with benign biopsy results (83%) than in those without suspicious nodule cytopathology (36%) (Genere et al. 2020). Interestingly, more than 80% of nodules with benign results were still recommended for continued AS. This may stem from the hesitation of clinicians to dismiss the suspicious appearance of a nodule and diagnose it as cytologically benign due to the high probability of malignancy (Brito et al. 2014). The use of AS also reflects a lack of confidence in being able to rule out malignancies. In addition, patients can feel psychologically burdened by concerns about tumours remaining untreated or metastasizing (Yeh et al. 2015) and might want to take future quality of life outcomes into account when making treatment decisions. This study clearly shows that many patients with subcentimeter thyroid nodules that appear highly suspicious on US choose to undergo surgery in our region as well as in China in general. Before performing FNA in such patients, the clinician needs to determine which nodules are most at risk of malignancy.

The results of this study suggest that subcentimeter thyroid nodules that appear highly suspicious on US, particularly those with taller-than-wide dimensions, the most suspicious ultrasound feature, may be recommended for FNA; however, regarding the choice between AS and surgery, because of this patient population’s greater risk (approximately nine times) of having a malignant tumour, the patient’s preference needs to be taken into account. The diagnostic accuracy of FNA in our sample was greater than expected, with high sensitivity and specificity and a 100% PPV. FNA is therefore a simple, accurate, and reliable preoperative diagnostic method for detecting thyroid nodules in the subcentimeter range that appear highly suspicious on US. It can also help in diagnosis and has proven clinical value. If a patient is younger than 45 years and has a nodule with a diameter < 1 cm and a taller-than-wide shape, AS should not be considered for management. Instead, FNAs or surgical interventions could be selected according to the patient's preference, and patients experiencing fear or anxiety related to the progression of their disease should be offered additional support.

We found that one specific malignant US feature alone cannot be used to determine the malignancy risk of a thyroid nodule; this must be done based on a combination of multiple signs. An increase in the number of suspicious features thus suggests a subsequent increase in the risk of malignancy. Because of the limited number of patients with two or more suspicious features on US, we did not conduct further research, which may have reduced the statistical power for identifying differences. The risk of malignancy increases with the accumulation of highly suspicious US features, reaching 97.9% if all highly suspicious US features are present (Jin Young et al. 2011).

The current study has the following advantages. (1) There was a large sample size. (2) Compared with those in other studies, all subcentimeter nodules with suspicious features within our cohort were diagnosed by pathological examination, with relatively few nodules undergoing puncture cytopathology and postoperative pathology. (3) Our separate analysis of the correlations between single US features and pathological properties of subcentimeter thyroid nodules may provide useful reference data for clinical management.

A limitation of this study is its single-centre retrospective observational cohort design. We excluded patients who had not undergone surgery or who were potentially missing many highly suspicious nodule US features that were not confirmed pathologically but were benign. Additionally, some enrolled patients exhibited two to three highly suspicious US features, thus increasing the likelihood of malignancy. This may have resulted in a bias towards the ‘A’ malignancy rate of 59.9% in this study. Furthermore, we were not able to comprehensively correct for some potentially confounding factors. For example, the patient's family history, occupational factors (i.e., radiation exposure), body mass index, and eating habits (i.e., excessive iodine intake) have been associated with an increased risk of thyroid cancer (Babić Leko et al. 2023). Prospective studies are warranted to further verify the results of this study. Health-related quality-of-life questionnaires, the Thyroid Cancer-Specific Quality of Life questionnaire, and the Fear of Disease Progression Questionnaire may also be useful in determining a patient's status when deciding between surgical and nonsurgical treatment. The presence of features in each of the categories in Table 5 was associated with a greater risk of malignancy (ROM) than expected according to the 2023 Bethesda System for Reporting Thyroid Cytopathology. We selected to puncture subcentimeter nodules with highly suspicious US features, which may have led to a greater proportion of postoperative pathological malignancies. In addition, the limited number of patients who underwent FNA may bias the results, and in our future research, more subcentimeter nodules need to be punctured before surgery.