Background: The purpose of this study was to reanalyze modern trials and use meta-analysis to determine how well frozen section gender, age, and tumor size could differentiate follicular adenoma from follicular carcinoma.
Method: Inclusion criteria were studies where patients had a permanent pathologic diagnosis of follicular adenoma or follicular carcinoma and underwent frozen section or had clinical features recorded. Data were pooled, and the random effects model of meta-analysis was used. A probability value of less than.05 was considered significant.
Results: Nineteen studies were included (n = 3486 patients). Frozen section was evaluated in 11 studies (n = 2204 patients). Frozen section had an 87% sensitivity, a 48% specificity, a 92% and 35% positive and negative predictive value, respectively, an 82% accuracy, an odds ratio of 0.181, a 95% confidence interval (CI) of 0.07 to 0.49, and a probability value of.001. Clinical features were evaluated in 10 studies (n = 1954 patients). Of the patients with follicular carcinoma, 27.5% were male compared with patients with follicular adenoma, of whom 17.7% were male (P <.01; odds ratio, 2.17; CI 1.3-3.6; P =.003). Of the patients with follicular carcinoma, 52.2% were older than 50 years (52.2%) compared with patients with follicular adenoma, of whom 28.5% were older than 50 years (P <.001). Of patients with follicular carcinoma, 36.8% had tumors larger than 3 to 5 cm compared with patients with follicular adenoma, of whom 14.7% had tumors larger than 3 to 5 cm (P <.001; odds ratio, 3.99; CI 1.5-10.8; P =.006).
Conclusions: Meta-analysis suggests that frozen section is not a specific test and cannot be used to confidently rule out follicular carcinoma. Male gender and large tumor size are significantly associated with carcinoma.