Josiah Brown Poster Abstract


Maxwell M. Wang
Michael Yeh, MD
Katrina Beckett, MD; Michael Douek, MD; Rinat Masamed, MD; Chi-Hong Tseng, PhD; Masha Livhits, MD; Michael Yeh, MD
Intersection between ultrasonography and molecular testing in the workup of indeterminate thyroid nodules

BACKGROUND: Thyroid nodules are common but only 5% are malignant. Thyroid biopsy yields an indeterminate result in 20% of cases. Non-invasive methods to evaluate the risk of malignancy, including ultrasound (US) characteristics and molecular gene profile testing, help to avoid unnecessary diagnostic surgery. Our study tests the hypothesis that US characteristics can enhance the diagnostic performance of molecular testing for indeterminate thyroid nodules.

METHODS: Four UCLA radiologists performed blinded retrospective US reads on thyroid nodules with an indeterminate biopsy result between 2012-2016. All nodules with indeterminate cytology reflexively underwent molecular testing with Afirma Gene Expression Classifier (GEC). Each nodule received paired reads, which assigned an American Thyroid Association (ATA) and American College of Radiology’s Thyroid Imaging Reporting and Data System (TI-RADS) classification of sonographic suspicion of malignancy. Retrospective chart review was conducted to determine the patient and nodule characteristics and results of biopsy and molecular testing. Outcomes of surgically resected nodules were determined via pathology reports. Nodules not resected were considered benign if they met strict criteria for stability on follow-up over one year after initial biopsy.

RESULTS: 144 indeterminate thyroid nodules (mean size of 2.5 cm) from 135 patients (83.7% female, mean age 50.2 years) were analyzed. Of these, 92 nodules were assigned the same ATA US classification from both radiologists (63.9% convergence): high suspicion 9.8%, intermediate suspicion 33.7%, low suspicion 54.3%, and very low suspicion 2.2%. The rate of suspicious GEC result was 88.9% for the high suspicion category, 54.8% for intermediate suspicion, 56.0% for low suspicion, and 50.0% for very low suspicion. While the positive predictive value (PPV) of GEC was 51.9% for the overall cohort, the PPV increased to 62.5% among high suspicion nodules, compared to 41.2% and 57.1% among intermediate and low suspicion nodules, respectively. In overall diagnostic performance, GEC had high sensitivity and negative predictive value (NPV) (97.8% and 97.9%, respectively), regardless of ATA classification. ATA had significantly higher specificity (p < 0.0001) than GEC when high suspicion ATA classification was considered a positive result for suspected malignancy (93.7% and 46.5%). Integrating US with molecular testing and considering a dual suspicious finding (high suspicion ATA with suspicious GEC) a positive result enhanced the specificity and PPV, but reduced the sensitivity and NPV of GEC on its own (specificity: 95.2% and 46.5%, PPV: 62.5% and 45.4%, sensitivity: 17.2% and 97.8%, NPV: 71.4% and 97.9%).

CONCLUSION: Molecular testing is effective for thyroid malignancy screening and ruling out of malignancy in indeterminate nodules, given its high sensitivity and NPV. We found suspicious US findings were highly specific, indicating US may be valuable as a malignancy confirmatory tool targeted for higher suspicion indeterminate nodules – for example, complex nodules in patients with notable family history or history of ionizing radiation. The rate of suspicious GEC result was meaningfully higher in ATA high suspicion nodules compared to in lower suspicion categories, suggesting molecular testing may have lower clinical utility in nodules with suspicious US findings. Our current data did not compellingly demonstrate that US findings affect GEC PPV; additional data acquired in the next phase of this study will determine whether high suspicion US findings enhance molecular testing’s diagnostic performance. Integrating US with molecular testing was a highly specific diagnostic test, leading us to hypothesize that a dual suspicious finding (high suspicion US and suspicious molecular test result) may reduce patient hesitation or reservations about surgery. We ultimately conclude that by elucidating suspicious features correlated with malignancy, US may be a valuable adjunct to molecular testing in the workup of indeterminate thyroid nodules.

KEY WORDS: Thyroid cancer, indeterminate thyroid nodules, ultrasound, molecular testing