Introduction: Because thyroid nodules are common and the vast majority are benign, the authors investigated whether ultrasound characteristics could be used to reduce the number of unnecessary biopsies.
Methods: The authors examined medical records from 8,806 patients who underwent 11,618 thyroid ultrasound (US) examinations from January 2000 to March 2005. Of this group, 105 patients were diagnosed with thyroid cancer.
Results: Thyroid nodules >5 mm were found in almost all patients with thyroid cancer (96.9%). Thyroid nodules also were found in over half (56.4%) of patients without thyroid cancer. The following 3 ultrasound characteristics of thyroid nodules were significant predictors of thyroid cancer:
Further analysis showed that requiring 2 of the 3 US characteristics to be present to prompt biopsy would have a sensitivity of 0.52, a false-positive rate of 0.07, and the risk of cancer in patients with a suspicious ultrasound would be 6.2%. The number of biopsies needed per cancer diagnosis would be 16. In addition, the risk for cancer would be 2 per 1,000 patients with 0 of these characteristics, 18 per 1,000 patients with 1 characteristic, 62 per 1,000 patients with 2 characteristics, and 960 per 1,000 patients with all 3 characteristics.
Using this requirement of at least 2 abnormal characteristics to prompt biopsy would reduce the number of unnecessary biopsies by 90% with a low risk of cancer in patients who do not undergo biopsy (ie, 5 cancers [0.5%] per 1,000 US examinations), the authors found.
Conclusion: The ultrasound criteria identified in this study may help determine when a thyroid nodule should be biopsied and help reduce the number of unnecessary procedures.
Smith-Bindman R, Lebda P, Feldstein VA, et al. Risk of thyroid cancer based on thyroid ultrasound imaging characteristics: results of a population-based study. JAMA Intern Med. 2013;173(19):1788-1796.
Commentary by Priyathama Vellanki MD
Dr. Priyathama Vellanki is an Assistant Professor, Division of Endocrinology, Metabolism and Lipids at Emory University School of Medicine in Atlanta, GA.
The main reason to work-up thyroid nodules is to detect thyroid cancer. The 2009 American Thyroid Association (ATA) guidelines suggest that nodule biopsy be based on size and ultrasound (US) features.1 However, most thyroid nodules are benign. Therefore, there are a large number of thyroid nodules that potentially do not require biopsy.
The 2011 Korean consensus statement on thyroid nodules suggests US-based management of thyroid nodules.2 The consensus statement recommends biopsy of a nodule when suspicious features such as microcalcification, hypoechogenicity or increased vascularity are present. Furthermore, ultrasound correlations to features of malignancy are based on studies when a biopsy was performed and possibly introduced an ascertainment bias.
The study by Smith-Bindman et al, therefore, sought to overcome this ascertainment bias by performing a population-based study that correlates US features with malignancy. This large retrospective study was performed at a single center (University of California, San Francisco) and used information linked from the California Cancer Registry to identify the patients with thyroid cancer.
Over a 5-year period (2000-2005), the authors examined the records of 8806 patients who underwent 11,618 ultrasound evaluations. Of these, only 105 patients had thyroid cancer. The authors were then able to retrieve the US records of 96 of these patients and identified that the features correlated with cancer. The control group was comprised of 428 nodules from patients without thyroid cancer and 87 benign nodules in patients who had thyroid cancer.
The most important predictors of malignancy were the presence of microcalcifications, nodule size >2 cm, and solid composition. The study found the likelihood of cancer was increased 7-fold if microcalcifications were present and 30% less likely if there were no microclaficiations. In addition, the study reported the accuracy of a cancer diagnosis if 1, 2 or 3 predictors were used to decide which nodules required biopsy.
The authors concluded sensitivity and positive predictive value (PPV) for cancer would be best when 2 predictors were used. The number needed to detect thyroid cancer would be 56 biopsies using only one predictor. When 2 predictors were used, biopsies needed to detect thyroid cancers decreased to 16, and the number of missed cancers was 2 to 5 per 1000 patients.
Overall, the authors concluded that using 2 ultrasound criteria when making a decision to biopsy would reduce the number of biopsies by at least 90%. The main strength of this study is that it is a large population-based study. The authors did not have the same ascertainment bias as in studies that correlated US features with malignancy where all patients underwent biopsies. The Smith-Bindman study also had a control group, which included nodules from patients without cancer and benign nodules from patients with cancer for comparison.
The weakness in this study is that it does not address the reasons patients received a thyroid US, which may introduce an ascertainment bias. This study is retrospective and therefore lacks data on behavior of nodules that were deemed benign. In order to be inclusive, the authors did include data from 2 years after the last patient's US so as not to miss a potential thyroid cancer. Nevertheless, this study demonstrated the criteria for biopsy can (1) be further refined by ultrasound characteristics and (2) reduce the number of biopsies performed.
1. Cooper DS, Doherty GM, Haugen BR, Kloos RT, et al. American Thyroid Association Guidelines Taskforce on Thyroid Nodules and Differentiated Thyroid Cancer. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. Nov 2009;19(11):1167-1214.
2. Moon WJ, Baek JH, Jung SL, et al. Ultrasonography and the ultrasound-based management of thyroid nodules: consensus statement and recommendations. Korean Journal of Radiology. Jan-Feb 2011;12(1):1-14.