Screening high-risk populations for thyroid cancer

Background. Children treated with radiotherapy to the neck or exposed to environmental radiation are at risk for developing thyroid cancer later in life. The best method for screening these high-risk patients is unclear. We systematically reviewed evidence on the accuracy of ultrasound and palpation to detect thyroid nodules and of fine needle aspiration (FNA), a confirmatory test, to diagnose thyroid cancer. Procedure. We searched the MEDLINE database for papers published since 1966, using the MeSH term thyroid neoplasms and terms related to diagnostic test performance. To supplement our MEDLINE searches, we searched reference lists from recent reviews and articles recommended by thyroid cancer experts. We recorded the tests used, the gold standard determination of disease, the test performance results, and the presence of biases that could affect the reported results. We also abstracted the number of patients who underwent surgery and the final diagnoses. We created two decision models: one for screening 10,000 medically irradiated patients, and one for screening 10,000 environmentally irradiated patients. Results. Using ultrasound as the gold standard determination of the presence of a nodule, the sensitivity of palpation for all sized nodules was 10-41 percent, indicating that a high proportion of nodules detected by ultrasound are too small to be palpated. Sensitivity of palpation increased with nodule size. The specificity of palpation ranged from 95 to 100%. In studies from referral centers, the reported sensitivity and specificity of FNA were 71-95 and 52-99%, respectively. However, most authors excluded the proportion of patients (6-33%) who had inadequate or nondiagnostic FNA results when calculating sensitivity and specificity, even though 6-100% of these patients went on to have a diagnostic lobectomy. When each study was reanalyzed so that patients with nondiagnostic FNA results who went directly to surgery were reclassified as positive tests, sensitivity increased slightly, but specificity dropped by 4-20 percentage points per study. The decision model for screening 10,000 medically irradiated patients revealed that if ultrasound were used as an initial screen, 2,741 patients would have nodules at least 1 cm in size; assuming no patients with smaller nodules had surgery, 1,964 patients would have surgery; 275 patients would have a diagnosis of thyroid cancer. Screening with ultrasound as an initial test would detect an additional 150 cases of thyroid cancer compared to those screened with palpation. However, an additional 1,689 patients would have surgery for nonmalignant nodules (compared to 480 patients with nonmalignant nodules screened with palpation). The yield for screening 10,000 environmentally irradiated patients was several times smaller than for screening 10,000 medically irradiated patients. If 10,000 environmentally irradiated patients were screened initially with ultrasound, approximately 708 patients would have nodules at least 1 cm in size; 89 patients would have surgery; and 38 patients would be diagnosed with thyroid cancer. Conclusions. Regardless of type of exposure, testing initially with ultrasound detects several times more cases of thyroid cancer than palpation. However, when ultrasound is the initial test, many more patients also have surgery for nonmalignant nodules. Screening with palpation is not very reassuring, particularly to medically irradiated patients with negative tests, since almost half (46%) of these patients may have undetected nodules.