Ethical considerations for artificial intelligence in dermatology: a scoping review.

The field of dermatology is experiencing the rapid deployment of artificial intelligence (AI), from mobile applications (apps) for skin cancer detection to large language models like ChatGPT that can answer generalist or specialist questions about skin diagnoses. With these new applications, ethical concerns have emerged. In this scoping review, we aimed to identify the applications of AI to the field of dermatology and to understand their ethical implications. We used a multifaceted search approach, searching PubMed, MEDLINE, Cochrane Library and Google Scholar for primary literature, following the PRISMA Extension for Scoping Reviews guidance. Our advanced query included terms related to dermatology, AI and ethical considerations. Our search yielded 202 papers. After initial screening, 68 studies were included. Thirty-two were related to clinical image analysis and raised ethical concerns for misdiagnosis, data security, privacy violations and replacement of dermatologist jobs. Seventeen discussed limited skin of colour representation in datasets leading to potential misdiagnosis in the general population. Nine articles about teledermatology raised ethical concerns, including the exacerbation of health disparities, lack of standardized regulations, informed consent for AI use and privacy challenges. Seven addressed inaccuracies in the responses of large language models. Seven examined attitudes toward and trust in AI, with most patients requesting supplemental assessment by a physician to ensure reliability and accountability. Benefits of AI integration into clinical practice include increased patient access, improved clinical decision-making, efficiency and many others. However, safeguards must be put in place to ensure the ethical application of AI.

The use of artificial intelligence (AI) in dermatology is rapidly increasing, with applications in dermatopathology, medical dermatology, cutaneous surgery, microscopy/spectroscopy and the identification of prognostic biomarkers (characteristics that provide information on likely patient health outcomes). However, with the rise of AI in dermatology, ethical concerns have emerged. We reviewed the existing literature to identify applications of AI in the field of dermatology and understand the ethical implications. Our search initially identified 202 papers, and after we went through them (screening), 68 were included in our review. We found that ethical concerns are related to the use of AI in the areas of clinical image analysis, teledermatology, natural language processing models, privacy, skin of colour representation, and patient and provider attitudes toward AI. We identified nine ethical principles to facilitate the safe use of AI in dermatology. These ethical principles include fairness, inclusivity, transparency, accountability, security, privacy, reliability, informed consent and conflict of interest. Although there are many benefits of integrating AI into clinical practice, our findings highlight how safeguards must be put in place to reduce rising ethical concerns.

Part 2: Management of Intraoperative and Perioperative Bleeding.

With an increase in number of patients on antithrombotic therapies, management of bleeding during dermatologic surgery is increasingly important. As described in Part 1, perioperative discontinuation of antithrombotic therapies may increase the risk of embolic events thus the risks and benefits must be weighed carefully when deciding whether to continue or suspend therapy. However, continuing oral anticoagulants may result in increased intraoperative and postoperative bleeding. Here we describe various methods to effectively achieve hemostasis which include: 1) mechanical methods to compress the vasculature 2) pharmacologic agents that induce vasoconstriction 3) physiologic agents that augment clot formation and 4) physical agents that promote platelet aggregation.

Part 1: Management of Antithrombotic Medications in Dermatologic Surgery.

Perioperative management of antithrombotic agents requires practical and medical considerations. Discontinuing antithrombotic therapies increases the risk of thrombotic adverse events including cerebrovascular accidents, myocardial infarction, pulmonary embolism, deep vein thrombosis, and retinal artery occlusion. Conversely, continuation of antithrombotic therapy during surgical procedures has associated bleeding risks. Currently, no guidelines exist regarding management of antithrombotic agents in the perioperative period for cutaneous surgeries and practice differs by surgeon. Here, we review the data on antithrombotic medications in patients undergoing cutaneous surgery including medication-specific surgical and postoperative bleeding risk if the medications are continued, and thromboembolic risk if the medications are interrupted. Specifically, we focus on vitamin K antagonist (VKA) (warfarin), direct-acting oral anticoagulants (DOAC) (rivaroxaban, apixaban, edoxaban, dabigatran), antiplatelet medications (aspirin, clopidogrel, prasugrel, ticagrelor, dipyridamole), unfractionated heparin, low molecular weight heparin (enoxaparin and dalteparin), fondaparinux, bruton tyrosine kinase inhibitors (BTKi) (ibrutinib, acalabrutinib), and dietary supplements (i.e., garlic, ginger, gingko).

Criteria2Query 3.0: Leveraging generative large language models for clinical trial eligibility query generation.

OBJECTIVE: Automated identification of eligible patients is a bottleneck of clinical research. We propose Criteria2Query (C2Q) 3.0, a system that leverages GPT-4 for the semi-automatic transformation of clinical trial eligibility criteria text into executable clinical database queries. MATERIALS AND METHODS: C2Q 3.0 integrated three GPT-4 prompts for concept extraction, SQL query generation, and reasoning. Each prompt was designed and evaluated separately. The concept extraction prompt was benchmarked against manual annotations from 20 clinical trials by two evaluators, who later also measured SQL generation accuracy and identified errors in GPT-generated SQL queries from 5 clinical trials. The reasoning prompt was assessed by three evaluators on four metrics: readability, correctness, coherence, and usefulness, using corrected SQL queries and an open-ended feedback questionnaire. RESULTS: Out of 518 concepts from 20 clinical trials, GPT-4 achieved an F1-score of 0.891 in concept extraction. For SQL generation, 29 errors spanning seven categories were detected, with logic errors being the most common (n = 10; 34.48 %). Reasoning evaluations yielded a high coherence rating, with the mean score being 4.70 but relatively lower readability, with a mean of 3.95. Mean scores of correctness and usefulness were identified as 3.97 and 4.37, respectively. CONCLUSION: GPT-4 significantly improves the accuracy of extracting clinical trial eligibility criteria concepts in C2Q 3.0. Continued research is warranted to ensure the reliability of large language models.

Superficial Atypical Lipomatous Tumor With Pleomorphic Features: Case Report and Discussion of the Literature.

ABSTRACT: Among liposarcomas, well-differentiated liposarcoma and dedifferentiated liposarcoma are the most common. The majority of these tumors are found in deep retroperitoneum or extremities. When found outside the retroperitoneum, these adipose-derived tumors are known as atypical lipomatous tumors (ALT). Superficial ALT are particularly rare; thus, little is known about their clinical presentation, genomic status, and management. Here, we present the case of a 54-year-old man with an intermittently bothersome, slowly growing mass on his left upper back for over 2 years, which was incidentally diagnosed as ALT. This patient's ALT, however, showed a profound degree of pleomorphism with MDM2 and control centromere 12 (CEP12) coamplification and negative CD34 and S100 and RB1 expression, unlike most other ALT described in the literature. This case report details the diagnostic workup and histopathological findings for adipose tumors and summarizes the different subtypes, including atypical spindle cell/pleomorphic lipomatous tumor, pleomorphic liposarcoma, and spindle cell/pleomorphic lipoma, with brief discussion on management.

Extragenital lichen sclerosus et atrophicus-morphea overlap as an initial presentation of genital lichen sclerosus.

Lichen sclerosus et atrophicus (LSA) is a chronic inflammatory disorder, most often characterized by atrophic skin plaques located on female genitalia. Infrequently, LSA may present extragenitally; however, much is unknown about the temporal relationship between genital and extragenital LSA. Morphea, also known as localized scleroderma, is a rare inflammatory skin condition characterized by sclerotic plaques. Investigators debate whether LSA and morphea exist on the same spectrum of disease, with LSA representing a superficial variant of morphea involving genitalia, or if they are distinct but coincidental entities. Although researchers have described LSA and morphea occurring in different locations on the same patient, few reports describe LSA and morphea occurring in the same lesion and in the inguinal folds. Herein, we report a case of a 62-year-old woman with extragenital LSA-morphea overlap in the inguinal folds, who three months later developed genital LSA. Extragenital LSA-morphea in the same plaque, with no signs of genital lesions on initial exam, with later development of genital LSA, is especially uncommon. The temporal progression of extragenital LSA-morphea overlap to genital LSA over a three-month period is an important contribution to the literature, as the temporal relationship between extragenital and genital LSA is not previously discussed.

Genomic analysis reveals HDAC1 regulates clinically relevant transcriptional programs in Pancreatic cancer.

Novel strategies are needed to combat multidrug resistance in pancreatic ductal adenocarcinoma (PDAC). We applied genomic approaches to understand mechanisms of resistance in order to better inform treatment and precision medicine. Altered function of chromatin remodeling complexes contribute to chemoresistance. Our study generates and analyzes genomic and biochemical data from PDAC cells overexpressing HDAC1, a histone deacetylase involved in several chromatin remodeling complexes. We characterized the impact of overexpression on drug response, gene expression, HDAC1 binding, and chromatin structure using RNA-sequencing and ChIP-sequencing for HDAC1 and H3K27 acetylation. Integrative genomic analysis shows that HDAC1 overexpression promotes activation of key resistance pathways including epithelial to mesenchymal transition, cell cycle, and apoptosis through global chromatin remodeling. Target genes are similarly altered in patient tissues and show correlation with patient survival. We also demonstrate that direct targets of HDAC1 that also show altered chromatin are enriched near genes associated with altered GTPase activity. HDAC1 target genes identified using in vitro methods and observed in patient tissues were used to develop a clinically relevant nine-transcript signature associated with patient prognosis. Integration of multiple genomic and biochemical data types enables understanding of multidrug resistance and tumorigenesis in PDAC, a disease in desperate need of novel treatment strategies.

Transcriptomic and functional analysis of ANGPTL4 overexpression in pancreatic cancer nominates targets that reverse chemoresistance.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) is one of the deadliest cancers based on five-year survival rates. Genes contributing to chemoresistance represent novel therapeutic targets that can improve treatment response. Increased expression of ANGPTL4 in tumors correlates with poor outcomes in pancreatic cancer. METHODS: We used statistical analysis of publicly available gene expression data (TCGA-PAAD) to test whether expression of ANGPTL4 and its downstream targets, ITGB4 and APOL1, were correlated with patient survival. We measured the impact of ANGPTL4 overexpression in a common pancreatic cancer cell line, MIA PaCa-2 cells, using CRISPRa for overexpression and DsiRNA for knockdown. We characterized global gene expression changes associated with high levels of ANGPTL4 and response to gemcitabine treatment using RNA-sequencing. Gemcitabine dose response curves were calculated on modified cell lines by measuring cell viability with CellTiter-Glo (Promega). Impacts on cell migration were measured using a time course scratch assay. RESULTS: We show that ANGPTL4 overexpression leads to in vitro resistance to gemcitabine and reduced survival times in patients. Overexpression of ANGPTL4 induces transcriptional signatures of tumor invasion and metastasis, proliferation and differentiation, and inhibition of apoptosis. Analyses revealed an overlapping signature of genes associated with both ANGPTL4 activation and gemcitabine response. Increased expression of the genes in this signature in patient PDAC tissues was significantly associated with shorter patient survival. We identified 42 genes that were both co-regulated with ANGPTL4 and were responsive to gemcitabine treatment. ITGB4 and APOL1 were among these genes. Knockdown of either of these genes in cell lines overexpressing ANGPTL4 reversed the observed gemcitabine resistance and inhibited cellular migration associated with epithelial to mesenchymal transition (EMT) and ANGPTL4 overexpression. CONCLUSIONS: These data suggest that ANGPTL4 promotes EMT and regulates the genes APOL1 and ITGB4. Importantly, we show that inhibition of both targets reverses chemoresistance and decreases migratory potential. Our findings have revealed a novel pathway regulating tumor response to treatment and suggest relevant therapeutic targets in pancreatic cancer.

Pooled CRISPR screening in pancreatic cancer cells implicates co-repressor complexes as a cause of multiple drug resistance via regulation of epithelial-to-mesenchymal transition.

BACKGROUND: Pancreatic ductal adenocarcinoma (PDAC) patients suffer poor outcomes, including a five-year survival of below 10%. Poor outcomes result in part from therapeutic resistance that limits the impact of cytotoxic first-line therapy. Novel therapeutic approaches are needed, but currently no targeted therapies exist to treat PDAC. METHODS: To assess cellular resistance mechanisms common to four cytotoxic chemotherapies (gemcitabine, 5-fluorouracil, irinotecan, and oxaliplatin) used to treat PDAC patients, we performed four genome-wide CRISPR activation (CRISPRact) and CRISPR knock-out (CRISPRko) screens in two common PDAC cell lines (Panc-1 and BxPC3). We used pathway analysis to identify gene sets enriched among our hits and conducted RNA-sequencing and chromatin immunoprecipitation-sequencing (ChIP-seq) to characterize top hits from our screen. We used scratch assays to assess changes in cellular migration with HDAC1 overexpression. RESULTS: Our data revealed activation of ABCG2, a well-described efflux pump, as the most consistent mediator of resistance in each of our screens. CRISPR-mediated activation of genes involved in transcriptional co-repressor complexes also conferred resistance to multiple drugs. Expression of many of these genes, including HDAC1, is associated with reduced survival in PDAC patients. Up-regulation of HDAC1 in vitro increased promoter occupancy and expression of several genes involved in the epithelial-to-mesenchymal transition (EMT). These cells also displayed phenotypic changes in cellular migration consistent with activation of the EMT pathway. The expression changes resulting from HDAC1 activation were also observed with activation of several other co-repressor complex members. Finally, we developed a publicly available analysis tool, PancDS, which integrates gene expression profiles with our screen results to predict drug sensitivity in resected PDAC tumors and cell lines. CONCLUSION: Our results provide a comprehensive resource for identifying cellular mechanisms of drug resistance in PDAC, mechanistically implicate HDAC1, and co-repressor complex members broadly, in multi-drug resistance, and provide an analytical tool for predicting treatment response in PDAC tumors and cell lines.

Chromodomain Helicase DNA-Binding Protein 7 Is Suppressed in the Perinecrotic/Ischemic Microenvironment and Is a Novel Regulator of Glioblastoma Angiogenesis.

Tumorigenic and non-neoplastic tissue injury occurs via the ischemic microenvironment defined by low oxygen, pH, and nutrients due to blood supply malfunction. Ischemic conditions exist within regions of pseudopalisading necrosis, a pathological hallmark of glioblastoma (GBM), the most common primary malignant brain tumor in adults. To recapitulate the physiologic microenvironment found in GBM tumors and tissue injury, we developed an in vitro ischemic model and identified chromodomain helicase DNA-binding protein 7 (CHD7) as a novel ischemia-regulated gene. Point mutations in the CHD7 gene are causal in CHARGE syndrome (a developmental disorder causing coloboma, heart defects, atresia choanae, retardation of growth, and genital and ear anomalies) and interrupt the epigenetic functions of CHD7 in regulating neural stem cell maintenance and development. Using our ischemic system, we observed microenvironment-mediated decreases in CHD7 expression in brain tumor-initiating cells and neural stem cells. Validating our approach, CHD7 was suppressed in the perinecrotic niche of GBM patient and xenograft sections, and an interrogation of patient gene expression datasets determined correlations of low CHD7 with increasing glioma grade and worse patient outcomes. Segregation of GBM by molecular subtype revealed a novel observation that CHD7 expression is elevated in proneural versus mesenchymal GBM. Genetic targeting of CHD7 and subsequent gene ontology analysis of RNA sequencing data indicated angiogenesis as a primary biological function affected by CHD7 expression changes. We validated this finding in tube-formation assays and vessel formation in orthotopic GBM models. Together, our data provide further understanding of molecular responses to ischemia and a novel function of CHD7 in regulating angiogenesis in both neoplastic and non-neoplastic systems. Stem Cells 2019;37:453-462.

R2DGC: threshold-free peak alignment and identification for 2D gas chromatography-mass spectrometry in R.

Summary: Comprehensive 2D gas chromatography-mass spectrometry is a powerful method for analyzing complex mixtures of volatile compounds, but produces a large amount of raw data that requires downstream processing to align signals of interest (peaks) across multiple samples and match peak characteristics to reference standard libraries prior to downstream statistical analysis. Very few existing tools address this aspect of analysis and those that do have shortfalls in usability or performance. We have developed an R package that implements retention time and mass spectra similarity threshold-free alignments, seamlessly integrates retention time standards for universally reproducible alignments, performs common ion filtering and provides compatibility with multiple peak quantification methods. We demonstrate that our package's performance compares favorably to existing tools on a controlled mix of metabolite standards separated under variable chromatography conditions and data generated from cell lines. Availability and implementation: R2DGC can be downloaded at https://github.com/rramaker/R2DGC or installed via the Comprehensive R Archive Network (CRAN). Contact: sjcooper@hudsonalpha.org. Supplementary information: Supplementary data are available at Bioinformatics online.

Assessing outcomes of topical 5-fluorouracil as primary and adjuvant therapy for squamous cell carcinoma in-situ.

Cutaneous squamous cell carcinoma in-situ (SCCis) is an intraepithelial tumor with a good prognosis. Standard treatment includes both surgical and non-surgical interventions. We determined the clearance rate for SCCis and residual SCCis identified on frozen section during Mohs micrographic surgery (MMS) after treatment with topical fluorouracil 5% cream (5-FU). All MMS cases were initiated for biopsy-proven invasive squamous cell carcinoma (SCC). A retrospective chart review was conducted from January 2017-February 2024 at Columbia University Irving Medical Center (CUIMC) to identify patients with SCCis who were treated with topical 5-FU as primary therapy or adjuvant therapy (AT) for residual SCCis post-MMS for invasive SCC. 41 patients were included (80% males, 70.1 ± 11.8 years). The average follow-up time for the primary therapy group was 25.4 ± 12.8 months, and for the post-MMS AT group 22.5 ± 11.1 months. In the group treated with topical 5-FU as primary therapy (n = 28), 27 patients (96.43%, 95% confidence interval: 81.65-99.91%) achieved complete clearance. One patient had recurrence at 8 months post-treatment. Of the patients in the post-MMS adjuvant treatment group (n = 13), 12 (92.3% clearance, 95% confidence interval 63.97-99.81%) achieved complete clearance. One patient had recurrence at 8 months post-treatment. This study found that topical 5-FU cream is effective as both primary therapy for SCCis and as adjuvant therapy for residual SCCis following MMS of invasive SCC.

Artificial intelligence for nonmelanoma skin cancer.

Nonmelanoma skin cancers (NMSCs) are among the top five most common cancers globally. NMSC is an area with great potential for novel application of diagnostic tools including artificial intelligence (AI). In this scoping review, we aimed to describe the applications of AI in the diagnosis and treatment of NMSC. Twenty-nine publications described AI applications to dermatopathology including lesion classification and margin assessment. Twenty-five publications discussed AI use in clinical image analysis, showing that algorithms are not superior to dermatologists and may rely on unbalanced, nonrepresentative, and nontransparent training data sets. Sixteen publications described the use of AI in cutaneous surgery for NMSC including use in margin assessment during excisions and Mohs surgery, as well as predicting procedural complexity. Eleven publications discussed spectroscopy, confocal microscopy, thermography, and the AI algorithms that analyze and interpret their data. Ten publications pertained to AI applications for the discovery and use of NMSC biomarkers. Eight publications discussed the use of smartphones and AI, specifically how they enable clinicians and patients to have increased access to instant dermatologic assessments but with varying accuracies. Five publications discussed large language models and NMSC, including how they may facilitate or hinder patient education and medical decision-making. Three publications pertaining to the skin of color and AI for NMSC discussed concerns regarding limited diverse data sets for the training of convolutional neural networks. AI demonstrates tremendous potential to improve diagnosis, patient and clinician education, and management of NMSC. Despite excitement regarding AI, data sets are often not transparently reported, may include low-quality images, and may not include diverse skin types, limiting generalizability. AI may serve as a tool to increase access to dermatology services for patients in rural areas and save health care dollars. These benefits can only be achieved, however, with consideration of potential ethical costs.

Cutaneous lymphoproliferative disorders after COVID-19 vaccination: clinical presentation, histopathology, and outcomes.

Individual reports described lymphoproliferative disorders (LPDs) after COVID-19 vaccination; however, the relationship between cases is unexamined. We aim to determine if there are cases of cutaneous LPDs associated with COVID-19 vaccination and their outcomes. We present a review of world literature, vaccine registries, and two unreported cases of LPDs after COVID-19 vaccination. Review of the medical literature, VAERS, and our two cases reveal predominance of Pfizer-BioNTech vaccine, younger patients, and males. All cases resulted in favorable outcomes. Approximately 84% of cases demonstrated CD30+ positivity in their skin biopsies, suggesting that an antigenic trigger may lead to a type IV adaptive immune response, with clonal expansion of CD30+ T-cells and subsequent oncogenic mutational hits eventuating in transient LPDs. LPDs after COVID-19 vaccination appear in the context of the same vaccines (proportionally to their global market shares), share clinical and pathological findings, and have indolent, self-limited character.

Exploring cutaneous lymphoproliferative disorders in the wake of COVID-19 vaccination.

Background: Individual reports have described lymphoproliferative disorders (LPDs) and cutaneous lymphomas emerging after administration of the COVID-19 vaccine; however, the relationship between reactions and vaccine types has not yet been examined. Objective: Determine if there are cases of cutaneous LPDs associated with certain COVID-19 vaccines and their outcomes. Methods: We analysed PubMed, the Vaccine Adverse Events Reporting System (VAERS), and our database for instances of biopsy-proven LPDs following COVID-19 vaccines. Results: Fifty cases of biopsy-proven LPDs arising after COVID-19 vaccination were found: 37 from medical literature, 11 from VAERS and two from our institution. Geographical distribution revealed the most cases in the United States, Italy, and Greece, with single cases in Spain, Colombia, Canada, Japan, and Romania. The average age of patients was 53; with a slight male predominance (male-to-female ratio of 1.5:1). The Pfizer-BioNTech vaccine was associated with LPDs in 36/50 (72%) cases, aligning with its 70% share of the global vaccine market. Histopathology revealed CD30+ in 80% of cases. The most prevalent form of LPD was lymphomatoid papulosis (LyP, 30%). All reported cases produced favourable outcomes (either complete or near-complete remission). Therapeutic approaches ranged from observation to treatment with steroids, methotrexate, or excision. Conclusion: LPDs after COVID-19 vaccination appear in the context of the same vaccines (proportionally to their global market shares), share clinical and pathological findings, and have indolent, self-limited character.

From Molecules to Microbes: Tracing Cutaneous T-Cell Lymphoma Pathogenesis through Malignant Inflammation.

The etiology of CTCL is a subject of extensive investigation. Researchers have explored links between CTCL and environmental chemical exposures, such as aromatic hydrocarbons (eg, pesticides and benzene), as well as infectious factors, including various viruses (eg, human T-lymphotropic virus [HTLV]-I and HTLV-II) and bacteria (eg, Staphylococcus aureus). There has been growing emphasis on the role of malignant inflammation in CTCL development. In this review, we synthesize studies of environmental and infectious exposures, along with research on the aryl hydrocarbon receptor and the involvement of pathogens in disease etiology, providing insight into the pathogenesis of CTCL.