Federated Learning for Decentralized Artificial Intelligence in Melanoma Diagnostics.

Importance: The development of artificial intelligence (AI)-based melanoma classifiers typically calls for large, centralized datasets, requiring hospitals to give away their patient data, which raises serious privacy concerns. To address this concern, decentralized federated learning has been proposed, where classifier development is distributed across hospitals. Objective: To investigate whether a more privacy-preserving federated learning approach can achieve comparable diagnostic performance to a classical centralized (ie, single-model) and ensemble learning approach for AI-based melanoma diagnostics. Design, Setting, and Participants: This multicentric, single-arm diagnostic study developed a federated model for melanoma-nevus classification using histopathological whole-slide images prospectively acquired at 6 German university hospitals between April 2021 and February 2023 and benchmarked it using both a holdout and an external test dataset. Data analysis was performed from February to April 2023. Exposures: All whole-slide images were retrospectively analyzed by an AI-based classifier without influencing routine clinical care. Main Outcomes and Measures: The area under the receiver operating characteristic curve (AUROC) served as the primary end point for evaluating the diagnostic performance. Secondary end points included balanced accuracy, sensitivity, and specificity. Results: The study included 1025 whole-slide images of clinically melanoma-suspicious skin lesions from 923 patients, consisting of 388 histopathologically confirmed invasive melanomas and 637 nevi. The median (range) age at diagnosis was 58 (18-95) years for the training set, 57 (18-93) years for the holdout test dataset, and 61 (18-95) years for the external test dataset; the median (range) Breslow thickness was 0.70 (0.10-34.00) mm, 0.70 (0.20-14.40) mm, and 0.80 (0.30-20.00) mm, respectively. The federated approach (0.8579; 95% CI, 0.7693-0.9299) performed significantly worse than the classical centralized approach (0.9024; 95% CI, 0.8379-0.9565) in terms of AUROC on a holdout test dataset (pairwise Wilcoxon signed-rank, P < .001) but performed significantly better (0.9126; 95% CI, 0.8810-0.9412) than the classical centralized approach (0.9045; 95% CI, 0.8701-0.9331) on an external test dataset (pairwise Wilcoxon signed-rank, P < .001). Notably, the federated approach performed significantly worse than the ensemble approach on both the holdout (0.8867; 95% CI, 0.8103-0.9481) and external test dataset (0.9227; 95% CI, 0.8941-0.9479). Conclusions and Relevance: The findings of this diagnostic study suggest that federated learning is a viable approach for the binary classification of invasive melanomas and nevi on a clinically representative distributed dataset. Federated learning can improve privacy protection in AI-based melanoma diagnostics while simultaneously promoting collaboration across institutions and countries. Moreover, it may have the potential to be extended to other image classification tasks in digital cancer histopathology and beyond.

BRAF and MEK inhibitor combinations induce potent molecular and immunological effects in NRAS-mutant melanoma cells: Insights into mode of action and resistance mechanisms.

About 25% of melanoma harbor activating NRAS mutations, which are associated with aggressive disease therefore requiring a rapid antitumor intervention. However, no efficient targeted therapy options are currently available for patients with NRAS-mutant melanoma. MEK inhibitors (MEKi) appear to display a moderate antitumor activity and also immunological effects in NRAS-mutant melanoma, providing an ideal backbone for combination treatments. In our study, the MEKi binimetinib, cobimetinib and trametinib combined with the BRAF inhibitors (BRAFi) encorafenib, vemurafenib and dabrafenib were investigated for their ability to inhibit proliferation, induce apoptosis and alter the expression of immune modulatory molecules in sensitive NRAS-mutant melanoma cells using two- and three-dimensional cell culture models as well as RNA sequencing analyses. Furthermore, NRAS-mutant melanoma cells resistant to the three BRAFi/MEKi combinations were established to characterize the mechanisms contributing to their resistance. All BRAFi induced a stress response in the sensitive NRAS-mutant melanoma cells thereby significantly enhancing the antiproliferative and proapoptotic activity of the MEKi analyzed. Furthermore, BRAFi/MEKi combinations upregulated immune relevant molecules, such as ICOS-L, components of antigen-presenting machinery and the "don't eat me signal" molecule CD47 in the melanoma cells. The BRAFi/MEKi-resistant, NRAS-mutant melanoma cells counteracted the molecular and immunological effects of BRAFi/MEKi by upregulating downstream mitogen-activated protein kinase pathway molecules, inhibiting apoptosis and promoting immune escape mechanisms. Together, our study reveals potent molecular and immunological effects of BRAFi/MEKi in sensitive NRAS-mutant melanoma cells that may be exploited in new combinational treatment strategies for patients with NRAS-mutant melanoma.

Successful Combination of Olaparib and 225Ac-Dotatate in a Patient with Neuroendocrine Tumor G3 and BRCA Mutation.

Based on the results of the NETTER-1 trial, peptide receptor radionuclide therapy with Lutetium-177 (177Lu) - DOTATATE is authorized for the treatment of neuroendocrine tumors (NET) grade 1 (G1) and grade 2 (G2) of the intestine. After the failure of 177Lu-DOTATATE therapy, targeted alpha-particle therapy (TAT) may be a possible treatment option. Here, we present a patient with cancer of unknown primary NET G2 later G3. The patient was referred to our hospital with urosepsis due to a second-degree urinary retention. After stent insertion, a contrast-enhanced computed tomography revealed a huge pelvic tumor without metastases. Initially, the patient had undergone surgical treatment. Later the patient developed liver metastasis and was treated by 177Lu-DOTATATE therapy and four lines of systemic therapy. A disease progression was observed and with the knowledge of a germline BRCA1 mutation, the patient was treated with TAT (Actinium-225 [225Ac]-DOTATATE) combined with olaparib. The patient achieved a significant treatment response for 12 months indicating that a combination therapy with an alpha emitter and olaparib demands further investigations in clinical trials.

Model soups improve performance of dermoscopic skin cancer classifiers.

BACKGROUND: Image-based cancer classifiers suffer from a variety of problems which negatively affect their performance. For example, variation in image brightness or different cameras can already suffice to diminish performance. Ensemble solutions, where multiple model predictions are combined into one, can improve these problems. However, ensembles are computationally intensive and less transparent to practitioners than single model solutions. Constructing model soups, by averaging the weights of multiple models into a single model, could circumvent these limitations while still improving performance. OBJECTIVE: To investigate the performance of model soups for a dermoscopic melanoma-nevus skin cancer classification task with respect to (1) generalisation to images from other clinics, (2) robustness against small image changes and (3) calibration such that the confidences correspond closely to the actual predictive uncertainties. METHODS: We construct model soups by fine-tuning pre-trained models on seven different image resolutions and subsequently averaging their weights. Performance is evaluated on a multi-source dataset including holdout and external components. RESULTS: We find that model soups improve generalisation and calibration on the external component while maintaining performance on the holdout component. For robustness, we observe performance improvements for pertubated test images, while the performance on corrupted test images remains on par. CONCLUSIONS: Overall, souping for skin cancer classifiers has a positive effect on generalisation, robustness and calibration. It is easy for practitioners to implement and by combining multiple models into a single model, complexity is reduced. This could be an important factor in achieving clinical applicability, as less complexity generally means more transparency.

Hyper-IgE and Carcinoma in CADINS Disease.

Background: Atopic dermatitis (AD) affects up to 25% of children and 10% of adults in Western countries. When severe or recurrent infections and exceedingly elevated serum IgE levels occur in AD patients, an inborn error of immunity (IEI) may be suspected. The International Union of Immunological Societies classification lists variants in different genes responsible for so-called Hyper-IgE syndromes. Diagnosing an underlying IEI may influence treatment strategies. Methods: Clinical and diagnostic workup of family members are presented including a detailed immunological description and histology of the carcinoma. Functional testing of the novel variant in CARD11 underlying 'CARD11-associated atopy with dominant interference of NF-kB signaling' (CADINS) was performed. Results: We report on an 18-year-old patient with a long-standing history of infections, accompanied by hypogammaglobulinemia, intermittent agranulocytosis, atopy, eosinophilia and colitis. The working diagnosis of common variable immunodeficiency was revised when a novel heterozygous CARD11 variant [c.223C>T; p.(Arg75Trp)] was identified. Functional studies confirmed this variant to have a dominant negative (DN) effect, as previously described in patients with CADINS. Five other family members were affected by severe atopy associated with the above variant, but not hypogammaglobulinemia. Malignancies occurred in two generations: an HPV-positive squamous cell carcinoma and a cutaneous T-cell lymphoma. So far, one patient is under treatment with dupilumab, which has shown marked benefit in controlling severe eczema. Conclusion: The phenotypic spectrum associated with heterozygous CARD11 DN mutations is broad. Partial T-cell deficiency, diminished IFN-γ cytokine and increased IL-4 production, were identified as disease-causing mechanisms. Malignant disease associated with germline CARD11 DN variants has only been reported sporadically. HPV vaccination in teenage years, and cytology screening analogous with routine cervical swabs may be recommended. Treatment with dupilumab, a monoclonal antibody blocking interleukin-4- and interleukin-13 signaling, may be of benefit in controlling severe and extended AD for some patients as reported for STAT3 loss-of-function.

Explainable artificial intelligence in skin cancer recognition: A systematic review.

BACKGROUND: Due to their ability to solve complex problems, deep neural networks (DNNs) are becoming increasingly popular in medical applications. However, decision-making by such algorithms is essentially a black-box process that renders it difficult for physicians to judge whether the decisions are reliable. The use of explainable artificial intelligence (XAI) is often suggested as a solution to this problem. We investigate how XAI is used for skin cancer detection: how is it used during the development of new DNNs? What kinds of visualisations are commonly used? Are there systematic evaluations of XAI with dermatologists or dermatopathologists? METHODS: Google Scholar, PubMed, IEEE Explore, Science Direct and Scopus were searched for peer-reviewed studies published between January 2017 and October 2021 applying XAI to dermatological images: the search terms histopathological image, whole-slide image, clinical image, dermoscopic image, skin, dermatology, explainable, interpretable and XAI were used in various combinations. Only studies concerned with skin cancer were included. RESULTS: 37 publications fulfilled our inclusion criteria. Most studies (19/37) simply applied existing XAI methods to their classifier to interpret its decision-making. Some studies (4/37) proposed new XAI methods or improved upon existing techniques. 14/37 studies addressed specific questions such as bias detection and impact of XAI on man-machine-interactions. However, only three of them evaluated the performance and confidence of humans using CAD systems with XAI. CONCLUSION: XAI is commonly applied during the development of DNNs for skin cancer detection. However, a systematic and rigorous evaluation of its usefulness in this scenario is lacking.

Whole-exome sequencing in eccrine porocarcinoma indicates promising therapeutic strategies.

Malignant sweat gland tumours are rare, with the most common form being Eccrine porocarcinoma (EP). To investigate the mutational landscape of EP, we performed whole-exome sequencing (WES) on 14 formalin-fixed paraffin-embedded samples of matched primary EP and healthy surrounding tissue. Mutational profiling revealed a high overall median mutation rate. This was attributed to signatures of mutational processes related to ultraviolet (UV) exposure, APOBEC enzyme dysregulation, and defective homologous double-strand break repair. All of these processes cause genomic instability and are implicated in carcinogenesis. Recurrent driving somatic alterations were detected in the EP candidate drivers TP53, FAT2, CACNA1S, and KMT2D. The analyses also identified copy number alterations and recurrent gains and losses in several chromosomal regions including that containing BRCA2, as well as deleterious alterations in multiple HRR components. In accordance with this reduced or even a complete loss of BRCA2 protein expression was detected in 50% of the investigated EP tumours. Our results implicate crucial oncogenic driver pathways and suggest that defective homologous double-strand break repair and the p53 pathway are involved in EP aetiology. Targeting of the p53 axis and PARP inhibition, and/or immunotherapy may represent promising treatment strategies.

Skin cancer classification via convolutional neural networks: systematic review of studies involving human experts.

BACKGROUND: Multiple studies have compared the performance of artificial intelligence (AI)-based models for automated skin cancer classification to human experts, thus setting the cornerstone for a successful translation of AI-based tools into clinicopathological practice. OBJECTIVE: The objective of the study was to systematically analyse the current state of research on reader studies involving melanoma and to assess their potential clinical relevance by evaluating three main aspects: test set characteristics (holdout/out-of-distribution data set, composition), test setting (experimental/clinical, inclusion of metadata) and representativeness of participating clinicians. METHODS: PubMed, Medline and ScienceDirect were screened for peer-reviewed studies published between 2017 and 2021 and dealing with AI-based skin cancer classification involving melanoma. The search terms skin cancer classification, deep learning, convolutional neural network (CNN), melanoma (detection), digital biomarkers, histopathology and whole slide imaging were combined. Based on the search results, only studies that considered direct comparison of AI results with clinicians and had a diagnostic classification as their main objective were included. RESULTS: A total of 19 reader studies fulfilled the inclusion criteria. Of these, 11 CNN-based approaches addressed the classification of dermoscopic images; 6 concentrated on the classification of clinical images, whereas 2 dermatopathological studies utilised digitised histopathological whole slide images. CONCLUSIONS: All 19 included studies demonstrated superior or at least equivalent performance of CNN-based classifiers compared with clinicians. However, almost all studies were conducted in highly artificial settings based exclusively on single images of the suspicious lesions. Moreover, test sets mainly consisted of holdout images and did not represent the full range of patient populations and melanoma subtypes encountered in clinical practice.

Elevated expression of prostate cancer-associated genes is linked to down-regulation of microRNAs.

BACKGROUND: Recent evidence suggests that the prostate cancer (PCa)-specific up-regulation of certain genes such as AMACR, EZH2, PSGR, PSMA and TRPM8 could be associated with an aberrant expression of non-coding microRNAs (miRNA). METHODS: In silico analyses were used to search for miRNAs being putative regulators of PCa-associated genes. The expression of nine selected miRNAs (hsa-miR-101, -138, -186, -224, -26a, -26b, -374a, -410, -660) as well as of the aforementioned PCa-associated genes was analyzed by quantitative PCR using 50 malignant (Tu) and matched non-malignant (Tf) tissue samples from prostatectomy specimens as well as 30 samples from patients with benign prostatic hyperplasia (BPH). Then, correlations between paired miRNA and target gene expression levels were analyzed. Furthermore, the effect of exogenously administered miR-26a on selected target genes was determined by quantitative PCR and Western Blot in various PCa cell lines. A luciferase reporter assay was used for target validation. RESULTS: The expression of all selected miRNAs was decreased in PCa tissue samples compared to either control group (Tu vs Tf: -1.35 to -5.61-fold; Tu vs BPH: -1.17 to -5.49-fold). The down-regulation of most miRNAs inversely correlated with an up-regulation of their putative target genes with Spearman correlation coefficients ranging from -0.107 to -0.551. MiR-186 showed a significantly diminished expression in patients with non-organ confined PCa and initial metastases. Furthermore, over-expression of miR-26a reduced the mRNA and protein expression of its potential target gene AMACR in vitro. Using the luciferase reporter assay AMACR was validated as new target for miR-26a. CONCLUSIONS: The findings of this study indicate that the expression of specific miRNAs is decreased in PCa and inversely correlates with the up-regulation of their putative target genes. Consequently, miRNAs could contribute to oncogenesis and progression of PCa via an altered miRNA-target gene-interaction.