Context: Controversy exists about whether thyroid fine-needle aspirates (FNAs) should be processed with conventional smears or liquid-based preparations (LBPs).
Objective: To compare the performance of conventional smears to LBPs for thyroid FNA slides circulated in the College of American Pathologists Interlaboratory Comparison Program in Non-Gynecologic Cytology.
Design: Participant responses for thyroid FNA slides were compared with the reference diagnosis at the level of 3 general diagnostic categories: negative, suspicious (which included only follicular and Hürthle cell neoplasm), and malignant. For specific reference diagnoses of benign/goiter and papillary thyroid carcinoma, the participants' specific diagnoses were analyzed and poorly performing slides were rereviewed.
Results: The 47, 076 thyroid FNA slide responses, between 2001 and 2009, included 44, 478 responses (94%) for conventional smears and 2598 responses (6%) for LBPs. For the general reference category negative, participant responses were discrepant in 14.9% of conventional smears compared with 5.9% for LBPs (P < .001). The specific reference diagnosis of benign/goiter was misdiagnosed as a follicular neoplasm in 7.8% of conventional smears, compared with 1.3% of LBP. For the general reference category of malignant, participant responses were discrepant in 7.3% of conventional smears compared with 14.7% of LBPs (P < .001). The specific reference diagnosis of papillary thyroid carcinoma was misdiagnosed as benign/goiter in 7.2% of LBPs, compared with 4.8% of conventional smears (p <.001).
Conclusions: LBPs performed worse than conventional smears for cases with a reference diagnosis of papillary thyroid carcinoma. However, LBPs performed better than conventional smears for cases with a benign reference diagnosis. Specific features in thyroid FNAs that may improve the diagnostic accuracy of LBPs and conventional smears are described.