Track: Clinical Studies

Abstract:

This study evaluated the diagnostic accuracy and interobserver reproducibility of 12 histopathologic clues to dermatophytosis. We conducted a cross-sectional analysis of 97 consecutive H&E-stained skin biopsy slides in which PAS staining was ordered due to clinical suspicion of dermatophytosis. A positive PAS stain for fungal organisms served as the reference gold standard. Interobserver agreement for individual histopathologic clues ranged from slight (e.g., compact red corneum and papillary edema; Cohen’s kappa < 0.2) to substantial (pustules; Cohen’s kappa = 0.61). Among all features, the presence of structures suspicious for dermatophytes was the only significant predictor of PAS positivity on multiple logistic regression (overall accuracy = 0.979, P < 0.001). Correlation analysis with hierarchical clustering grouped the features into three categories: (1) corneal changes associated with dermatophytes, (2) features related to spongiosis, and (3) granulocytic inflammation. Random forest analysis indicated that granulocytic inflammation was the least important category in predicting PAS positivity. In machine learning models with 5-fold cross-validation, logistic regression using only the presence of suspected fungal hyphae as a predictor outperformed more complex approaches, including random forest, k-nearest neighbors, latent factor analysis, and support vector machines. Using penalized maximum likelihood, the estimated probability of a positive PAS result was 0.93 when possible hyphae were identified and 0.03 when they were not. These findings highlight that the presence of possible fungal hyphae in the stratum corneum on H&E-stained sections is the most reliable histopathologic clue to dermatophytosis.

Tatiana Movchan

Affiliation

Refine Results

(109) Diagnostic Accuracy of Histopathologic Clues to Tinea in a Cross-Validated Machine Learning Study