ITN can be a challenge for radiologists and clinicians. This study showed that ITN represented almost one third of thyroid nodules biopsied with ultrasound-guided FNA and that use of guidelines could reduce the numbers of workups by one third. Implementation of such guidelines could result in enormous health care savings if multiplied over time and across other institutions in the United States.

The SRU recommendations were developed by an expert group of radiologists to assist physicians in deciding which nodules seen at ultrasound should be biopsied with FNA. One of the concerns of the society was the unknown effect of the guidelines on the workup rate of thyroid nodules and the possibility that following the guidelines could increase the rates of FNA and subsequent surgery. Our study showed that use of the SRU recommendations reduced by 30% the rate of workup of ITN seen initially at ultrasound compared with current practice in which these criteria are not used. Ahn et al. [18] also studied the SRU recommendations, applying the criteria to all nodules biopsied with ultrasound guidance. For nodules 10 mm or larger the reduction in workup would have been 40% (396 SRU-negative nodules among 996 total nodules). They found the sensitivity and specificity of the SRU criteria were 72% and 42% (28% false-negative rate). When the SRU recommendations are applied to our cohort of 47 ITN detected with ultrasound, we found a similar sensitivity of 75% but even lower specificity (30%). This result is likely due to differences in patient populations. Our study included only asymptomatic nodules (incidental) rather than all thyroid nodules biopsied by FNA, and so we had a larger number of benign cases.

The three-tiered system for selecting thyroid nodules for workup based on CT and MRI findings was first described in a review by Hoang et al. [10] but has since been evaluated in a retrospective study that compared the three-tiered system with a 10-mm or greater size cutoff for ITN detected at neck CT [7]. When the investigators applied the three-tiered system to nodules 10 mm or larger, there was a 46% reduction in the workup rate. In our study the reduction in workup rate for ITN detected at CT, MRI, or PET/CT was 35% with the three-tiered system compared with current practice of no guidelines. When we excluded the PET cases and only considered ITN detected at CT or MRI, the reduction in workup rate was 42%, similar to the results of Nguyen et al. [7]. This is because most of the nodules detected incidentally at PET were metabolically active and met the criteria for workup. Our study confirmed that metabolic activity at PET/CT is an indication for workup of an ITN (category 1 of three-tiered system). Two of eight (25%) nodules with FDG avidity at PET were malignant. Our results are consistent with those of other studies that have shown higher rates of malignancy (27–47%) of nodules detected at PET/CT, particularly those with focal hypermetabolic activity [19–21].

Performance of the classification criteria is strongly influenced by the prevalence of malignancy. The true prevalence of malignancy in an ITN is difficult to determine accurately in retrospective studies because not all ITN are biopsied and the malignancy rate depends on the study population. We found that the malignancy rate in 113 low-risk ITN was 6%, lower than that in other studies. Nam-Goong et al. [22] found a malignancy rate of 12% in nonpalpable nodules biopsied with FNA, but only 10% of their cohort had truly incidental nodules and not selected nodules for which the patient was referred for biopsy from other primary care clinics. The only U.S. study, to our knowledge, evaluating the rate of malignancy in ITN biopsied with FNA showed a malignancy rate of 17% [23]. However, that was a small retrospective study of 21 ITN compared with 90 thyroid nodules not incidentally detected. The authors could not explain why the latter group had a much lower malignancy rate of 3%.

In acknowledging the wide range of the reported rates of malignancy of ITN, it is important to recognize that the true prevalence of malignancy in ITN seen at imaging is many times lower. For each ITN in our study, many more ITN were not biopsied because they might not have been reported at imaging, the patient did not undergo sonographic workup, or the patient underwent diagnostic ultrasound but not biopsy. This has been a limitation in previous retrospective studies that attempted to identify the malignancy rate in ITN seen at consecutive CT and MRI examinations. In two studies ITN were found in 16–18% of imaging examinations, but FNA was performed on only 10% of ITN [4, 7]. Thus the true prevalence of malignancy in ITN seen at imaging is unknown but is certainly a small fraction of our malignancy rate of 6%.

The greatest drive to work up nodules is the fear of missing malignancy. In this study, one case of cancer was missed. It was a 14-mm papillary carcinoma that did not meet the SRU criteria for FNA. According to the 2009 SEER database [9], the 10-year survival rate for a 14-mm localized tumor is greater than 99.6%. A contentious question for future research is whether treatment of small papillary cancers improves survival [24]. An observational trial that provides a perspective on untreated small cancers followed 340 patients with papillary microcarcinomas who were selected to not receive treatment [25]. After 10 years, new nodal metastases were found in 3% of cases, and there had been no cancer deaths. The opinion that most patients die with, rather than of, thyroid cancer is also supported by a postmortem study that showed incidental small foci of thyroid cancer in 36% of patients who died of other causes [26]. On the basis of this body of evidence, the intentions of both the SRU recommendations and the three-tiered system are not to diagnose all cancers but to diagnose cancers that have reached clinical significance while avoiding unnecessary tests and surgery on patients with benign nodules.

There were several limitations to this study. First, it was a retrospective study with a small number of malignant tumors. We would expect more malignancies to be missed with the SRU recommendations and the three-tiered system when a larger number of ITN are evaluated. It is important to note, however, that the false-negative rate is low relative to the percentage reduction in benign nodules worked up. Second, this study was conducted over a limited period of 12 months at a single institution, and the result may not be generalizable to past and future referral patterns. Review of institutional data on ultrasound-guided neck FNA in the preceding 9 years shows that our patient population represents the expected volume of FNAs in the radiology department. Third, FNAs performed in the radiology department do not include all thyroid FNAs at our institution. FNA of thyroid nodules is performed under ultrasound guidance by surgeons, endocrinologists, and pathologists, and these were not included. The differences between nonradiology FNAs and our study cohort are unknown. The nodules biopsied by clinicians and pathologists might have a higher proportion of SRU-negative or three-tiered criteria–negative nodules because of more liberal biopsy practices due to convenience of biopsy in a clinic. Or the opposite could occur because clinic biopsies may be performed on larger, less technically challenging thyroid nodules. Finally, we applied the criteria on the basis of the imaging findings and clinical information available. There might have been other clinical information not available in the medical records that put the patients at higher risk of malignancy and warranted biopsy. The guidelines are intended for patients without a high-risk clinical history.

The implication of this study is that simple criteria could be introduced to reduce the rate of unnecessary biopsy of benign nodules and thus reduce the rising financial and nonmonetary costs associated with workup and subsequent surgery. However, this study represents the biopsy practices in only one academic radiology practice. Practices may differ at other institutions depending on how thyroid nodules are reported at imaging and the practice patterns of the referring clinicians. To understand the effect of implementing guidelines nationally, it would be useful to evaluate the criteria across different practice types or assess practice by performing a survey of how ITN are reported by radiologists. If reduction in unnecessary workup is possible at other institutions, these criteria could also be used to develop evidence-based practice guidelines for ITN. The challenge beyond creating guidelines is having them adopted by the clinicians who refer patients for biopsy. There are other existing guidelines for nodules seen at ultrasound, such as those of the National Comprehensive Cancer Network and the American Thyroid Association [27, 28]. The National Comprehensive Cancer Network guidelines are similar to the SRU recommendations in use of a 1.5-cm cutoff for nodules without other suspicious findings. We agree that the larger size cutoff of 1.5 cm used in the SRU criteria and the three-tiered system will help to reduce unnecessary workups.

This study showed that applying the SRU recommendations and the three-tiered system to ITN can reduce the workup rate almost one third compared with current practice without specific guidelines. The cost of reducing workup would be one missed papillary carcinoma. This malignant tumor and all other incidental thyroid malignancies in this study were low stage without regional or distant metastatic disease.