[431] Use of HMGA2 Overexpression by RT-PCR Performed on Suspicious Cytology Smears to Preoperatively Separate Benign from Malignant Thyroid Nodules

PJ Lappinga, NS Kip, RV Lloyd, L Jin, MR Henry, A Nassar. Mayo Clinic, Rochester, MN

Background: Follicular lesions continue to be a problem in thyroid cytopathology. No specific cytomorphologic features allow for distinction of benign from malignant follicular lesions. Approximately 40% of nodules that are diagnosed as “suspicious for follicular neoplasm” are resected. Up to 80% of indeterminate follicular nodules prove to be benign. Presently, there are no reliable ancillary tests to improve the preoperative detection of malignant follicular nodules. We sought to evaluate the performance of HMGA2 overexpression by RT-PCR performed on cells obtained from suspicious cytologic smears to distinguish benign from malignant follicular thyroid nodules.
Design: A search was performed to identify patients who had undergone thyroid FNA and subsequent thyroid nodule resection at our institution from 2001 to 2007. Only cases diagnosed as “suspicious” were included. Cases with 200 or more cells were included for HMGA2 testing. Resection specimen slides from the malignant cases were evaluated for architecture (solid, trabecular or insular), invasiveness (minimally vs. widely), extrathyroidal extension, necrosis, mitoses per 10 HPF, and presence of blood vessel invasion. HMGA2 expression was detected by one-step real-time qRT-PCR. HMGA2 mRNA expression was normalized to phosphoglycerate kinase mRNA expression and HMGA2 expression was expressed as relative fold change compared to HMGA2 expression in a thyroid cancer cell line (TPC-1).
Results: The following 164 patients were included: goiter 3 (1.8%), adenomatous nodule 15 (9.1%), follicular adenoma 60 (36.6%), Hurthle cell adenoma 27 (16.5%), follicular carcinoma 14 (8.5%), Hurthle cell carcinoma 11 (6.7%), papillary carcinoma (PTC) 5 (3%) and follicular variant of PTC 29 (17.7%). Using an HMGA2 expression ratio of ≥ 5.9 as a positive cutoff, the test had the following performance overall: sensitivity 71%, specificity 97%, PPV 94% and NPV 84%. HMGA2 overexpression had low sensitivity for Hurthle cell carcinoma (33%). When only non-Hurthle carcinomas including papillary and follicular carcinoma were considered, the sensitivity and specificity were 80% and 98% respectively.
Conclusions: HMGA2 overexpression can be performed on cells scraped from cytologic smears and it has high sensitivity and specificity for detecting malignant nodules in non-Hurthle cell carcinomas. HMGA2 overexpression analysis in conjunction with conventional cytology may markedly improve the ability to predict malignancy in follicular thyroid nodules.
Category: Cytopathology

Monday, March 22, 2010 9:30 AM

Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 50, Monday Morning


Close Window