A morphological classification for vocal fold leukoplakia

Abstract Introduction: There is still no general method for discriminating between benign and malignant leukoplakia and identifying vocal fold leukoplakia. Objective: To evaluate the reliability of a morphological classification and the correlation between morphological types and pathological grades of vocal fold leukoplakia. Methods: A total of 375 patients with vocal fold leukoplakia between 2009 and 2015 were retrospectively reviewed. Two observers divided the vocal fold leukoplakia into flat and smooth, elevated and smooth, and rough type on the basis of morphological appearance. The inter-observer reliability was evaluated and the results of classification from both observers were compared with final pathological grades. Clinical characteristics between low risk and high risk group were also analyzed. Results: The percentage inter-observer agreement of the morphological classification was 78.7% (κ = 0.615, p < 0.001). In the results from both observers, the morphological types were significantly correlated with the pathological grades (p1 < 0.001, p2 < 0.001, Kruskal-Wallis test; r1 = 0.646, p1 < 0.001, r2 = 0.539, p2 < 0.001, Spearman Correlation Analysis). Multivariate analysis showed patient's age (p = 0.018), the size of lesion (p < 0.001), and morphological type (p < 0.001) were significantly different between low risk group and high risk group. Combined receiver operating characteristic curve analysis of significant parameters revealed an area under the receiver operating characteristic curve of 0.863 (95% CI 0.823-0.903, p < 0.001). Conclusions: The proposed morphological classification of vocal fold leukoplakia was consistent between observers and morphological types correlated with pathological grades. Patient's age, the size of lesion, and morphological type might enable risk stratification and provide treatment guidelines for vocal fold leukoplakia.