Immunohistochemistry in Dermatopathology: A Retrospective Study of the Most Frequently Used Antibodies

Evaluating the diagnostic performance of a large language model-powered chatbot for providing immunohistochemistry recommendations in dermatopathology

BACKGROUND

Large language model (LLM)-powered chatbots such as ChatGPT have numerous applications. However, their effectiveness in dermatopathology has not been formally evaluated. Dermatopathological cases often require immunohistochemical workup. Here, we evaluate the performance of a chatbot in providing diagnostically useful information on immunohistochemistry relating to dermatological diseases.

METHODS

We queried a commonly used chatbot for the immunophenotypes of 51 cutaneous diseases, including a diverse variety of epidermal, adnexal, hematolymphoid, and soft tissue entities. We requested it to provide references for each diagnosis. All tests were repeated, compiled, quantified, and then compared with established literature standards.

RESULTS

Clustering analysis demonstrated that recommendations correlated with tumor type, suggesting chatbots can supply appropriate panels. However, a significant portion of recommendations were factually incorrect (13.9%). Citations were rarely clinically useful (24.5%). Many were confabulated (27.2%). Prompt responses for cutaneous adnexal lesions tended to be less accurate while literature references were less useful. Reference retrieval performance was associated with the number of PubMed entries per entity.

CONCLUSIONS

This foundational study suggests that LLM-powered chatbots may be useful for generating immunohistochemical panels for dermatologic diagnoses. However, specific performance capabilities and biases must be considered. In addition, extreme caution is advised regarding the tendencies to fabricate material. Future models intentionally fine-tuned to augment diagnostic medicine may prove to be valuable.

Incorporation of dermoscopy improves inter-observer agreement among dermatopathologists in histologic assessment of melanocytic neoplasms

Histopathologic assessment of melanocytic neoplasms is the current gold standard of diagnosis. However, there are well recognized limitations including inter-observer diagnostic discordance. This study aimed to determine if integrating dermoscopy with histopathology of melanocytic neoplasms impacts diagnosis and improves inter-observer agreement. We conducted a prospective cohort study in a pigmented lesion clinic. Consecutive melanocytic lesions were identified for biopsy based on atypical gross or dermoscopic features. Standardized immunohistochemistry and levels were ordered on each specimen. The cases were randomized. Three dermatopathologists blinded to the clinical impression assessed each lesion. The cases were then re-randomized and re-assessed with addition of gross clinical and dermoscopic images. Inter-rater reliability (IRR) using Fleiss' kappa statistic revealed an increase from 0.447 without to 0.496 with dermoscopy amongst all dermatopathologists. The kappa increased from 0.495 before to 0.511 with dermoscopy in separating high-grade atypia or melanoma from moderate atypia or less. In 16 of 136 cases, at least 2 of 3 dermatopathologists favored a diagnosis of melanoma only after dermoscopy. In total, the consensus grade of atypia changed in 24.3% (33/ 136) of cases thereby representing changes to excisional margins and patient follow up. This study is limited by the cohort size. Dermoscopy significantly impacts diagnosis and improves identification of early melanomas in high risk populations and improves inter-observer agreement.

Dermatologic Lesions Submitted to an Oral and Maxillofacial Pathology Biopsy Service: An Analysis of 2487 Cases

Skin lesions are often submitted to oral and maxillofacial pathology practices. The purpose of this study is to evaluate the frequency, distribution, variability, and composition of dermatologic lesions within a large oral and maxillofacial pathology biopsy service. An IRB-approved retrospective search of skin lesions diagnosed at University of Florida oral and maxillofacial pathology biopsy service between 1994 and 2015 was performed. 2487 cases were included in the study. Gender was reported in 2466 cases, of which 59% were male and 41% female. Age was provided in 2367 cases and ranged from 2 weeks to 96 years with an average of 55 years. Location was indicated in 2473 cases. Lips were the most common (41.7%), followed by face (25.3%), neck (7.4%), nose (6.5%), periorbital (5.3%), ear (4.1%), and scalp (3.8%). Of the 2487 cases, five diagnoses (actinic keratosis/cheilitis, intradermal nevus, epidermal inclusion cyst, seborrheic keratosis, and basal cell carcinoma) constituted 84.4% of the cases. 69 of 2487 cases (2.8%) resulted in dermatopathologic consultation prior to final reporting. Skin lesions accounted for ~ 1.0% of all lesions submitted to an oral and maxillofacial pathology biopsy service. This study found a large variation in the dermatologic lesions submitted to an oral pathology biopsy service. Although most were routine in complexity, dermatopathology consultation was an important tool in the diagnosis of the more challenging cases. This study may help pathologists gain a better understanding of the frequency and variability of dermatologic lesions submitted to an oral and maxillofacial pathology biopsy service and promote more interdisciplinary consultation within the field. This study evaluated the incidence and scope of dermatologic lesions submitted to a large oral and maxillofacial pathology biopsy service over a long time period. A wide scope of lesions was found, and dermatopathology consultation was important to quality assurance.

Immunostain use in the diagnosis of melanomas referred to a tertiary medical center: a 15-year retrospective review (2001-2015)

BACKGROUND

Little is known regarding the clinical practice of immunohistochemistry in the diagnosis of melanoma. We aimed to assess the incidence of immunostain usage by referring pathologists and dermatopathologists in melanoma cases sent for consultative review. As a secondary objective, associations between immunostain use and specific melanoma characteristics were also evaluated.

METHODS

This is a retrospective review of consultation reports of referred melanomas at a tertiary academic center in New York, NY from 2001 to 2015. Univariate regression analysis was performed on melanomas with accompanying immunostains and on characteristics such as Breslow's depth, location, prognostic factors and morphologic subtypes. Associations between immunostain usage and these characteristics were analyzed using Fisher's exact test.

RESULTS

Immunostain use significantly increased over the study period (p < 0.001) and was more likely to be associated with melanomas that were thicker [odds ratio (OR) = 2.5; 1.7-3.6]; located on the head and neck (OR = 1.6; 1.4-1.9) or acral sites (OR = 1.5; 1.1-2.0); had ulceration (OR = 2.1; 1.6-2.8), dermal mitoses (OR = 1.3; 1.1-1.5), or perineural invasion (OR = 3.6; 2.0-6.5); or were of desmoplastic (OR = 7.4; 4.5-12), amelanotic (OR = 7.1; 3.6-14), or nevoid subtypes (OR = 4.0; 1.7-8.9).

CONCLUSIONS

Immunostain use in the diagnosis of melanoma has increased significantly in the past 15 years for reasons that remain unclear.