Saisindhu Narala1, Shirley Q Li2, Natasha K Klimas1, Anisha B Patel1,3. 1. Department of Dermatology, University of Texas Medical School at Houston, Houston, Texas, USA. 2. Baylor College of Medicine, Houston, Texas, USA. 3. Department of Dermatology, University of Texas MD Anderson Cancer Center, Houston, Texas, USA.
Abstract
BACKGROUND: The distinction between chondrodermatitis nodularis helicis (CNH) and hyperplastic actinic keratosis (HAK) on the ear can pose a diagnostic challenge. We aimed to identify histopathological characteristics that could distinguish between CNH and HAK on routine sections using penalized least absolute shrinkage and selection operator (LASSO) logistic regression analysis. METHODS: Cases of CNH (n = 80) and HAK (n = 28) were analyzed for selected histopathological characteristics. Fisher's exact test and LASSO regression were performed. RESULTS: In univariate analyses, the following were significantly associated with CNH: ulceration, acanthosis, granular layer in the majority of the lesion, hypergranulosis at the periphery of the lesion, hyperkeratosis at the periphery of the lesion, hyperparakeratosis at the periphery of the lesion, fibrosis, increased blood vessels, vertically oriented blood vessels, and fibrin. A LASSO model excluding atypia found that fibrin, fibrosis, presence of granular layer, ulceration, and vertically oriented blood vessels were most predictive of CNH. Keratinized strap cells were not a significant predictor. CONCLUSION: We have identified features that may aid in differentiating these entities and demonstrated that a LASSO regression model can identify predictors that may improve diagnostic accuracy. Our results indicate that the highest diagnostic accuracy in this dilemma is dependent on obtaining biopsy specimens with visible dermis.
BACKGROUND: The distinction between chondrodermatitis nodularis helicis (CNH) and hyperplastic actinic keratosis (HAK) on the ear can pose a diagnostic challenge. We aimed to identify histopathological characteristics that could distinguish between CNH and HAK on routine sections using penalized least absolute shrinkage and selection operator (LASSO) logistic regression analysis. METHODS: Cases of CNH (n = 80) and HAK (n = 28) were analyzed for selected histopathological characteristics. Fisher's exact test and LASSO regression were performed. RESULTS: In univariate analyses, the following were significantly associated with CNH: ulceration, acanthosis, granular layer in the majority of the lesion, hypergranulosis at the periphery of the lesion, hyperkeratosis at the periphery of the lesion, hyperparakeratosis at the periphery of the lesion, fibrosis, increased blood vessels, vertically oriented blood vessels, and fibrin. A LASSO model excluding atypia found that fibrin, fibrosis, presence of granular layer, ulceration, and vertically oriented blood vessels were most predictive of CNH. Keratinized strap cells were not a significant predictor. CONCLUSION: We have identified features that may aid in differentiating these entities and demonstrated that a LASSO regression model can identify predictors that may improve diagnostic accuracy. Our results indicate that the highest diagnostic accuracy in this dilemma is dependent on obtaining biopsy specimens with visible dermis.