| Presentation: |
S017 |
| Topic: |
Clinical - Thyroid / Parathyroid |
| Type: |
Oral Presentation |
| Date: |
Thursday, April 11, 2013 |
| Session: |
08:00 AM - 09:00 AM Thyroid / General |
| Authors: |
Allen S Ho, MD, Evan Sarti, DO, Hangjun Wang, MD, Kunal S Jain, MD, Oscar Lin, MD, Iain J Nixon, MD, Ashok R Shaha, MD, Jatin P Shah, MD, R. Michael Tuttle, MD, Ronald Ghossein, MD, Richard J Wong, MD, Luc G.T. Morris, MD, MSc |
| Institution(s): |
Memorial Sloan-Kettering Cancer Center |
Background: The Bethesda System for reporting thyroid cytopathology is the current standard for interpreting thyroid fine needle aspiration (FNA) specimens. The Atypia of Undetermined Significance (AUS) category has an implied risk of malignancy between 5-15%, but the true incidence remains unknown. Our objective was to evaluate the prognostic role of substratifying the AUS diagnosis, while analyzing its true malignancy rate in surgical specimens.
Study Design: Retrospective cohort analysis at a comprehensive cancer center.
Methods: The management of 577 thyroid nodule patients presenting with AUS between 2007-2011 was evaluated. Patients with pre-existing or concurrent thyroid malignancies were excluded. Clinicopathologic predictors of surgery were assessed using multivariable logistic regression. Among patients undergoing surgery for diagnostic or therapeutic purposes, FNA-biopsied nodules were correlated with surgical pathology to determine the incidence of malignancy.
The AUS diagnosis was further substratified into five categories: AUS-NOS; AUS-favor benign; AUS-cannot exclude papillary thyroid carcinoma (PTC); AUS-cannot exclude Hurthle cell neoplasm; and AUS-cannot exclude follicular neoplasm. Cytologic diagnoses were correlated with findings in the thyroidectomy specimens to determine benign (goiter and lymphocytic thyroiditis) and neoplastic (benign and malignant) findings.
Results: Of all thyroid FNAs, 8.0% were classified as AUS, matching the recommended Bethesda rate. Of patients with AUS on initial FNA, 63.6% (367/577) underwent immediate surgery, 16.5% (95/577) had repeat FNA, and 16.6% (96/577) were observed. Repeat FNA revealed AUS in 38.9% (37/95), malignancy in 4.2% (4/95), and was benign in 43.2% (41/95). Of those with consecutive AUS diagnoses who went on to surgery, 36.8% (7/19) were identified with malignancy. In multivariable analysis, the only predictor of immediate surgery was calcifications on ultrasound (OR=3.8, p=0.02). Among all AUS nodules, the malignancy rate was 22.1% (95% confidence interval (CI), 18.7-25.6). Among AUS nodules triaged to surgery, the malignancy rate was 31.0% (95% CI, 27.2-34.9). There were separate incidental thyroid cancers in 14.2% of patients, increasing the total malignancy rate in resected specimens to 45.2%.
Upon substratification of the AUS category, the malignancy rate was significantly higher if a specific neoplasm could not be ruled out. Conversely, an AUS-favor benign diagnosis conferred a higher rate of non-malignant findings (Table 1).
Conclusions: Malignancy rates in AUS are higher than typically estimated, with 22.1-31.0% of AUS nodules found to be malignant and an additional 14.2% of glands harboring incidental cancers. Substratification also appears to improve prediction of neoplasm and selection for surgery. Emerging molecular assays may further help to triage AUS nodules to observation or surgery.
Table 1: Histologic correlation in each AUS subcategory
AUS cases with surgery
|
Benign |
Malignant |
| AUS-NOS (n=218) |
41.3% (95% CI, 36.0-46.6) |
36.2% (95% CI, 29.4-42.5) |
| AUS-favor benign (n=13) |
76.9% (95% CI, 69.2-84.6) |
7.7% (95% CI, 1.9-36.0) |
| AUS-cannot exclude PTC (n=28) |
28.6% (95% CI, 9.4-47.8) |
53.6% (95% CI, 33.9-72.1) |
| AUS-cannot exclude Hurthle cell neoplasm (n=44) |
34.1% (95% CI, 20.5-47.7) |
29.5% (95% CI, 16.8-45.2) |
| AUS-cannot exclude follicular neoplasm (n=78) |
24.4% (95% CI,11.9-36.9) |
56.4% (95% CI, 44.7-67.6) |