Development and Assessment of an Artificial Intelligence-Based Tool for Skin Condition Diagnosis by Primary Care Physicians and Nurse Practitioners in Teledermatology Practices.

Importance: Most dermatologic cases are initially evaluated by nondermatologists such as primary care physicians (PCPs) or nurse practitioners (NPs). Objective: To evaluate an artificial intelligence (AI)-based tool that assists with diagnoses of dermatologic conditions. Design, Setting, and Participants: This multiple-reader, multiple-case diagnostic study developed an AI-based tool and evaluated its utility. Primary care physicians and NPs retrospectively reviewed an enriched set of cases representing 120 different skin conditions. Randomization was used to ensure each clinician reviewed each case either with or without AI assistance; each clinician alternated between batches of 50 cases in each modality. The reviews occurred from February 21 to April 28, 2020. Data were analyzed from May 26, 2020, to January 27, 2021. Exposures: An AI-based assistive tool for interpreting clinical images and associated medical history. Main Outcomes and Measures: The primary analysis evaluated agreement with reference diagnoses provided by a panel of 3 dermatologists for PCPs and NPs. Secondary analyses included diagnostic accuracy for biopsy-confirmed cases, biopsy and referral rates, review time, and diagnostic confidence. Results: Forty board-certified clinicians, including 20 PCPs (14 women [70.0%]; mean experience, 11.3 [range, 2-32] years) and 20 NPs (18 women [90.0%]; mean experience, 13.1 [range, 2-34] years) reviewed 1048 retrospective cases (672 female [64.2%]; median age, 43 [interquartile range, 30-56] years; 41 920 total reviews) from a teledermatology practice serving 11 sites and provided 0 to 5 differential diagnoses per case (mean [SD], 1.6 [0.7]). The PCPs were located across 12 states, and the NPs practiced in primary care without physician supervision across 9 states. The NPs had a mean of 13.1 (range, 2-34) years of experience and practiced in primary care without physician supervision across 9 states. Artificial intelligence assistance was significantly associated with higher agreement with reference diagnoses. For PCPs, the increase in diagnostic agreement was 10% (95% CI, 8%-11%; P < .001), from 48% to 58%; for NPs, the increase was 12% (95% CI, 10%-14%; P < .001), from 46% to 58%. In secondary analyses, agreement with biopsy-obtained diagnosis categories of maglignant, precancerous, or benign increased by 3% (95% CI, -1% to 7%) for PCPs and by 8% (95% CI, 3%-13%) for NPs. Rates of desire for biopsies decreased by 1% (95% CI, 0-3%) for PCPs and 2% (95% CI, 1%-3%) for NPs; the rate of desire for referrals decreased by 3% (95% CI, 1%-4%) for PCPs and NPs. Diagnostic agreement on cases not indicated for a dermatologist referral increased by 10% (95% CI, 8%-12%) for PCPs and 12% (95% CI, 10%-14%) for NPs, and median review time increased slightly by 5 (95% CI, 0-8) seconds for PCPs and 7 (95% CI, 5-10) seconds for NPs per case. Conclusions and Relevance: Artificial intelligence assistance was associated with improved diagnoses by PCPs and NPs for 1 in every 8 to 10 cases, indicating potential for improving the quality of dermatologic care.

A deep learning system for differential diagnosis of skin diseases.

Skin conditions affect 1.9 billion people. Because of a shortage of dermatologists, most cases are seen instead by general practitioners with lower diagnostic accuracy. We present a deep learning system (DLS) to provide a differential diagnosis of skin conditions using 16,114 de-identified cases (photographs and clinical data) from a teledermatology practice serving 17 sites. The DLS distinguishes between 26 common skin conditions, representing 80% of cases seen in primary care, while also providing a secondary prediction covering 419 skin conditions. On 963 validation cases, where a rotating panel of three board-certified dermatologists defined the reference standard, the DLS was non-inferior to six other dermatologists and superior to six primary care physicians (PCPs) and six nurse practitioners (NPs) (top-1 accuracy: 0.66 DLS, 0.63 dermatologists, 0.44 PCPs and 0.40 NPs). These results highlight the potential of the DLS to assist general practitioners in diagnosing skin conditions.

Metastatic basal cell carcinoma of the posterior neck: case report and review of the literature.

Although primary basal cell carcinoma (BCC) represents an extremely common malignancy, metastases derived from BCC are exceedingly rare. The prognosis for metastatic BCC is poor, and little consensus exists regarding predictive factors or optimal treatment strategies. Here, we present the case of a 63-year-old man with BCC of the neck who subsequently developed multiple metastases to subcutaneous tissue, lymph nodes, and the parotid gland. Risk factors and clinical features of metastatic BCC are reviewed, as is the relationship of histopathologic subtype to metastatic behavior. Current chemotherapeutic and targeted therapies also are discussed in the context of recent advances in molecular biology.

Expression of the embryonic stem cell transcription factor SOX2 in human skin: relevance to melanocyte and merkel cell biology.

SOX2 is a gene located on chromosome 3q26.33 that encodes a transcription factor important to maintenance of embryonic neural crest stem cell pluripotency. We have identified rare SOX2-immunoreactive cells in normal human skin at or near the established stem cell niches. Three subsets of SOX2-positive cells were defined in these regions: those expressing only SOX2 and those that co-expressed SOX2 and either CK20 or microphthalmia-associated transcription factor, which are consistent with dichotomous differentiation of SOX2-expressing precursors along neuroendocrine (Merkel cell) or melanocytic lines, respectively. Examination of Merkel cell carcinomas confirmed nuclear SOX2 expression in this tumor type. In human patient melanoma, strong nuclear expression of SOX2 was noted in a subset of tumors, and the ability to detect SOX2 in lesional cells significantly correlated with primary tumor thickness in a survey cohort. To assess the potential role of SOX2 in melanoma growth, an in vivo tumorigenesis assay was used. Whereas SOX2 knockdown failed to influence proliferation of cultured melanoma cells in vitro, tumor xenografts generated with the SOX2-knockdown cell line showed significant decrease in mean tumor volume as compared with controls. In aggregate, these findings suggest that SOX2 is a novel biomarker for subpopulations of normal skin cells that reside in established stem cell niches and that might relate to Merkel cell and melanocyte ontogeny and tumorigenesis.

Imiquimod enhances IFN-gamma production and effector function of T cells infiltrating human squamous cell carcinomas of the skin.

Squamous cell carcinomas (SCCs) are sun-induced skin cancers that are particularly numerous and aggressive in patients taking T-cell immunosuppressant medications. Imiquimod is a topical immune response modifier and Toll-like receptor 7 (TLR7) agonist that induces the immunological destruction of SCC and other skin cancers. TLR7 activation by imiquimod has pleiotropic effects on innate immune cells, but its effects on T cells remain largely uncharacterized. Because tumor destruction and formation of immunological memory are ultimately T-cell-mediated effects, we studied the effects of imiquimod therapy on effector T cells infiltrating human SCC. SCC treated with imiquimod before excision contained dense T-cell infiltrates associated with tumor cell apoptosis and histological evidence of tumor regression. Effector T cells from treated SCC produced more IFN-gamma, granzyme, and perforin and less IL-10 and transforming growth factor-beta (TGF-beta) than T cells from untreated tumors. Treatment of normal human skin with imiquimod induced activation of resident T cells and reduced IL-10 production but had no effect on IFN-gamma, perforin, or granzyme, suggesting that these latter effects arise from the recruitment of distinct populations of T cells into tumors. Thus, imiquimod stimulates tumor destruction by recruiting cutaneous effector T cells from blood and by inhibiting tonic anti-inflammatory signals within the tumor.

Human squamous cell carcinomas evade the immune response by down-regulation of vascular E-selectin and recruitment of regulatory T cells.

Squamous cell carcinomas (SCCs) of the skin are sun-induced skin cancers that are particularly numerous in patients on T cell immunosuppression. We found that blood vessels in SCCs did not express E-selectin, and tumors contained few cutaneous lymphocyte antigen (CLA)(+) T cells, the cell type thought to provide cutaneous immunosurveillance. Tumors treated with the Toll-like receptor (TLR)7 agonist imiquimod before excision showed induction of E-selectin on tumor vessels, recruitment of CLA(+) CD8(+) T cells, and histological evidence of tumor regression. SCCs treated in vitro with imiquimod also expressed vascular E-selectin. Approximately 50% of the T cells infiltrating untreated SCCs were FOXP3(+) regulatory T (T reg) cells. Imiquimod-treated tumors contained a decreased percentage of T reg cells, and these cells produced less FOXP3, interleukin (IL)-10, and transforming growth factor (TGF)-beta. Treatment of T reg cells in vitro with imiquimod inhibited their suppressive activity and reduced FOXP3, CD39, CD73, IL-10, and TGF-beta by indirect mechanisms. In vivo and in vitro treatment with imiquimod also induced IL-6 production by effector T cells. In summary, we find that SCCs evade the immune response at least in part by down-regulating vascular E-selectin and recruiting T reg cells. TLR7 agonists neutralized both of these strategies, supporting their use in SCCs and other tumors with similar immune defects.

Structural basis for removal of adenine mispaired with 8-oxoguanine by MutY adenine DNA glycosylase.

The genomes of aerobic organisms suffer chronic oxidation of guanine to the genotoxic product 8-oxoguanine (oxoG). Replicative DNA polymerases misread oxoG residues and insert adenine instead of cytosine opposite the oxidized base. Both bases in the resulting A*oxoG mispair are mutagenic lesions, and both must undergo base-specific replacement to restore the original C*G pair. Doing so represents a formidable challenge to the DNA repair machinery, because adenine makes up roughly 25% of the bases in most genomes. The evolutionarily conserved enzyme adenine DNA glycosylase (called MutY in bacteria and hMYH in humans) initiates repair of A*oxoG to C*G by removing the inappropriately paired adenine base from the DNA backbone. A central issue concerning MutY function is the mechanism by which A*oxoG mispairs are targeted among the vast excess of A*T pairs. Here we report the use of disulphide crosslinking to obtain high-resolution crystal structures of MutY-DNA lesion-recognition complexes. These structures reveal the basis for recognizing both lesions in the A*oxoG pair and for catalysing removal of the adenine base.