Thyroid nodules come to clinical attention when noted by the patient, as an incidental finding during routine physical examination, or during a radiologic imaging, such as carotid doppler, neck computed tomography (CT), or positron emission tomography (PET). Their clinical importance is primarily related to the need to exclude thyroid cancer, which accounts for 4.0% to 6.5% of all thyroid nodules [1,2].

Thyroid incidentalomas are nonpalpable thyroid nodules defined as newly identified thyroid lesions encountered during imaging procedures. Nonpalpable nodules have approximately the same risk of malignancy as palpable nodules. In some circumstances, especially nodules discovered by PET/CT, the risk of malignancy may be higher [3,4].

As a molecular imaging modality, 2-deoxy-2-¹⁸F-fluoro-D-glucose (¹⁸F-FDG) PET/CT can detect a wide variety of tumor sites. Abnormal incidental findings, including thyroid lesions, are not uncommon because of the nonspecific physiologic properties of FDG, as well as the range of the scan. As the use of PET/CT becomes more common, the discovery of thyroid incidentalomas with focal FDG uptake will increase [5].

Focal thyroid incidentalomas detected by ¹⁸F-FDG PET/CT have been reported in 1% to 4% of known cancer patients and the normal healthy population, with positive predictive values for underlying thyroid malignancy of 27.8% to 74% [4,6,7,8,9]. However, the prevalence of malignancy in thyroid nodules detected incidentally by ¹⁸F-FDG PET/CT has not been fully characterized, and several issues remain. For example, it is controversial whether the semiquantified index maximal standardized uptake value (SUV_max) can differentiate between malignant and benign thyroid nodules [10,11,12,13].

In this study, we performed a retrospective review of our institutional experience of FDG-avid thyroid incidentaloma in healthy subjects undergoing cancer screening and patients with nonthyroid cancer. The risk of malignancy in FDG-avid thyroid incidentaloma and its association with SUV_max in ¹⁸F-FDG PET/CT were evaluated to define the predictor variables in assessing the risk of malignancy.

Of the 2,584 subjects who underwent ¹⁸F-FDG PET/CT (319 for cancer screening, 2,265 for metastatic evaluation), 52 (2.0%) were identified as having focal FDG uptake in the thyroid (Fig. 2). Among these 52 patients, further evaluation was not performed on 16 subjects because of refusal of additional workup, loss to clinical follow-up, or extensive disease of another underlying primary malignancy. These cases were excluded. Finally, 36 patients (12 males and 24 females; mean age, 63.4±10.9 years; range, 40 to 80 years; 28 patients with suspected or known cancer, eight healthy subjects who underwent ¹⁸F-FDG PET/CT for cancer screening) who underwent FNAC for further diagnostic evaluastion were included in the analysis. All subjects who underwent further histologic evaluation, except for one, also had a thyroid US scan and had their tumor size determined by measurement of the longest diameter.

Cytologic diagnosis was obtained by fine needle aspiration cytology (FNAC). Among the 13 subjects with suspected malignancy based on FNAC, 11 underwent surgical resection. One subject who did not undergo surgical resection had suspected metastatic squamous carcinoma from known esophageal cancer for which surgery was not indicated, and another subject refused to undergo surgery. The subjects who underwent surgery were proven to have malignant disease, which in all 11 subjects was papillary carcinoma. Meanwhile, among the 23 subjects with suspected benign disease based on FNAC, seven subjects underwent surgical resection (follicular neoplasm or suspicious for a malignancy on thyroid US finding) and two lesions were proven to be malignant (one follicular carcinoma and one papillary carcinoma) (Fig. 1). The remaining five lesions were nodular hyperplasias. The subject shown to have follicular carcinoma by surgical resection had been thought to have a follicular neoplasm based on FNAC. The subject shown to have papillary carcinoma had had suspected nodular hyperplasia based on FNAB; however, thyroid US made us suspect malignancy and a BRAF gene mutation test was positive. Therefore, surgical resection was performed and this subject was finally proven to have a malignancy.

Of the 23 subjects with suspected non-malignant lesions, 11 were shown by FNAB to have follicular neoplasms; follicular carcinoma could not be excluded. However, surgical resection was not performed in all of these cases; therefore, the final diagnosis was not confirmed. The 12 subjects without follicular neoplasms had benign lesions: one benign follicular nodule, 10 nodular hyperplasias, and one case of thyroiditis (Table 1).

The positive predictive value of malignancy in FDG-avid thyroid incidentaloma was 41.7% (15 of 36 subjects). The prevalence of malignancy was 0.58% (15 of 2,584 subjects) and was higher in the cancer screening group than in patients with suspected or known cancer (0.94% vs. 0.53%).Comparisons of PET/CT findings between benign and malignant thyroid lesions detected incidentally on PET/CT scans are shown in Table 2. Age and sex had no relevance to malignancy. Median SUV_max for FDG-avid thyroid incidentalomas was 3.4 (IQR, 2.8 to 5.4). Median SUV_max for benign and malignant lesions was 2.8 (IQR, 2.6 to 4.0) and 4.7 (IQR, 3.4 to 6.0), respectively, which suggests a trend for FDG-avid thyroid incidentalomas with higher SUV_max values to be malignant. The difference was statistically significant in the current study (P=0.001; Mann-Whitney U test). The median tumor size (longest diameter in thyroid US) for FDG-avid thyroid incidentalomas was 1.5 cm (IQR, 1.1 to 2.3). The median tumor size for benign and malignant lesions was 1.8 cm (IQR, 1.1 to 2.6) and 1.2 cm (IQR, 1.0 to 1.5), respectively, and there was no statistical significance (P=0.158; Mann-Whitney U test). To determine the correlation between SUV_max and tumor size, Spearman's correlation coefficient was calculated. There was no correlation between SUV_max and tumor size (Spearman correlation coefficient=0.052; P=0.767).

Thyroid carcinoma accounts for 1% of all malignant tumors. It is the most frequent endocrine cancer and is usually characterized by a favorable prognosis and long-term survival [14,15]. About 5% of thyroid nodules are thyroid cancer. Thyroid incidentalomas have become increasingly common with the development and more frequent use of highly sensitive imaging modalities [16]. The more widespread use of ¹⁸F-FDG PET/CT for cancer staging, restaging, and treatment response monitoring has increased the incidence of thyroid incidentalomas identified by ¹⁸F- FDG PET/CT [17].

Prevalence rates of thyroid incidentalomas detected by ¹⁸F-FDG PET/CT have been reported, and results show that thyroid incidentalomas are relatively frequent, with a rate of 0.2% to 8.9% [10,18,19,20]. However, the prevalence of malignant thyroid nodules detected incidentally by ¹⁸F-FDG PET/CT has not yet been fully characterized and varies widely (range, 10.3% to 80.0%) [6,7,8,11,21,22,23,24,25]. At our institution, the prevalence of FDG-avid thyroid incidentalomas and the risk of malignancy of thyroid incidentalomas were 2.0% and 41.7%, respectively. These data are similar to those from previous studies and slightly higher than previous Korean values. The very high prevalence of malignancy justifies further work-up, such as US and FNAB [8,20,21].

Because of the high cost of ¹⁸F-FDG PET/CT, it is generally not used for cancer screening in asymptomatic healthy subjects. Therefore, the prevalence of incidentalomas in these subjects is unclear [10]. In contrast, a higher prevalence in cancer screening subjects than in patients with suspected or known cancer has been reported in some studies (Kang et al. [7], 3.0% vs. 1.9%; Yang et al. [10], 3.1% vs. 2.3%). In our study, ¹⁸F-FDG PET/CT detected 12 unexpected thyroid malignancies in 28 patients with suspected or known cancer and three thyroid malignancies in eight healthy subjects who underwent ¹⁸F-FDG PET/CT for cancer screening. The prevalence of malignancy in FDG-avid thyroid incidentalomas was similar in patients with suspected or known cancer (42.0%) and in cancer screening patients (37.5%) in our study. However, other studies suggest that PET/CT is useful in detecting unexpected second primary cancers [26].

SUV_max is a semiquantitative parameter that reflects metabolic activity, but it is not a specific marker of malignancies. Many thresholds have been proposed to distinguish benign from malignant lesions, but no safe cutoff has been identified. In fact, approximately half of studies report a statistically significant difference between SUV_max for benign lesions and the value for malignant ones, whereas the other half show the opposite [18,21,27]. In this study, median SUV_max was significantly higher in malignant lesions than in benign ones (4.7 [IQR, 3.4 to 6.0] vs. 2.8 [IQR, 2.6 to 4.0], P=0.001). This study supports the view that SUV_max is a helpful diagnostic tool to discriminate benign from malignant nodules.

At our institution, all subjects who underwent further histologic evaluation, except for one, also underwent thyroid US and had their tumor size determined by measurement of the longest diameter. Tumor size is known as an important factor that affects FDG uptake. When tracer uptake in small tumors is measured, large biases can be introduced by the partial-volume effect. If a tumor is small, FDG uptake can be underestimated because of the partial-volume effect [28]. In this study, Spearman's correlation coefficient was calculated to determine the correlation between SUV_max and tumor size. There was no correlation between SUV_max and tumor size (Spearman correlation coefficient=0.052; P=0.767).

Our study has several limitations that must be taken into account. First, follicular neoplasms of the thyroid are defined as follicular carcinomas, follicular adenomas, and nodules of goiters. Surgical resection was not performed in all of these cases; therefore, the final diagnosis was not confirmed. So, some cases of follicular neoplasms may be thyroid malignancies, and the prevalence of malignancy in thyroid incidentaloma may have been underestimated. Second, SUV_mean was not evaluated. Third, because our study population was a cohort of patients cared for in a single center, the results might have been affected by selection bias. Nonetheless, this single-center study had the advantage of having a high degree of consistency regarding laboratory and imaging data and histological diagnosis.

In conclusion, thyroid incidentalomas found on ¹⁸F-FDG PET/CT in patients with suspected or known cancer and healthy cancer screening subjects have a high risk of malignancy, with a positive predictive value of 41.7% for underlying thyroid malignancy. There was a trend for FDG-avid thyroid incidentalomas with higher SUV_max to be malignant. Given the risk of malignancy, patients with FDG-avid thyroid incidentaloma should receive a tissue diagnosis and proper management.