European Journal of Cancer

This study compared machine-learning models for predicting recurrence-free survival (RFS), disease-specific survival (DSS), and overall survival (OS) using clinicopathologic data from 1,621 stage I/II primary cutaneous melanoma patients. Our time-to-event models achieved concordance indices of 0.829 for RFS, 0.812 for DSS, and 0.778 for OS. Tumor thickness and mitotic rate were the most important predictors for RFS. Charlson comorbidity score and insurance type were critical for DSS and OS.

Distant melanoma metastasis at the time of diagnosis is uncommon, but has major implications for patient prognosis and treatment selection. However, few tools can reliably predict the risk of distant metastasis at initial presentation. Here, we developed and evaluated machine learning models to predict distant melanoma metastasis using routinely captured clinicopathologic and demographic variables across all histologic subtypes. Using the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) program from 2010-2022, we identified adults aged 20 to 90 years with melanoma as the first and only primary malignancy (n=51,285). Explainable Boosting Machine achieved a strong balance of discrimination and precision (AUROC = 0.947, AUPRC = 0.610, Precision = 0.793, Brier = 0.015). At 90% sensitivity, specificity was 0.843 with consistent performance across cross-validation folds. Clinicopathologic variables, including T stage, Breslow thickness, ulceration, and mitotic activity, contributed the largest share of predictive signal across descriptive, regression-based, and SHAP analyses, with smaller contributions from demographic factors. Decision curve analysis supported clinical utility, showing a net reduction of 88.3 per 100 patients and a standardized net benefit of 0.541. This model could be used to identify patients at sufficiently elevated risk to justify staging PET/CT despite otherwise localized clinical presentation. Cost-consequence analysis further showed that imaging true- and false-positive patients at 85% to 95% sensitivity threshold nearly doubled downstream imaging cost. We deployed the final model as an online calculator to support exploration of individualized risk estimates (https://melanoma-calculator.streamlit.app/).

Background: The tumor microenvironment (TME) plays a critical role in cancer progression and treatment response. Stromal components, including cancer-associated fibroblasts (CAFs), extracellular matrix (ECM), and angiogenesis, contribute to tumor aggressiveness. However, a comprehensive stromal activity score integrating multiple stromal dimensions for pan-cancer prognosis prediction is lacking. Methods: We developed a Stromal Activity Score (SAS) integrating five stromal dimensions: CAF signature (12 genes), ECM remodeling (15 genes), TGF-{beta} signaling (13 genes), angiogenesis (12 genes), and complement activation (11 genes). SAS was calculated using single-sample Gene Set Enrichment Analysis (ssGSEA) on TCGA pan-cancer data comprising 1,303 samples across 12 cancer types. Prognostic value was evaluated using Kaplan-Meier analysis and Cox regression. Immunotherapy response prediction was validated in two independent cohorts (IMvigor210, n=88; Liu2019, n=105). Results: Pan-cancer Cox regression demonstrated a significant association between SAS and overall survival (HR = 1.165, 95% CI: 1.065-1.275, P = 0.001). Per-cancer analysis identified BRCA (HR = 1.942, P = 0.022), STAD (HR = 1.684, P = 0.024), and LUSC (HR = 1.552, P = 0.038) as significant, though none survived FDR correction. SAS correlated strongly with ESTIMATE Stromal Score (Spearman {rho} = 0.835) and moderately with Immune Score ({rho} = 0.396). Immunotherapy validation showed consistent trends (IMvigor210: AUC = 0.602; Liu2019: AUC = 0.617). Time-dependent ROC analysis showed 1-year AUC = 0.596, 3-year = 0.579, 5-year = 0.559. Leave-one-out analysis identified angiogenesis removal as enhancing prognostic signal (HR = 3.737, P = 0.0002). Three distinct TME subtypes were identified with differential SAS profiles. Conclusions: SAS is a novel pan-cancer stromal activity score that captures TME biology with strong construct validity. Its clinical utility as a standalone biomarker remains modest, but it may complement existing immunotherapy biomarkers.

Backgroundfollowing a first cutaneous squamous cell carcinoma (CSCC), one-third of patients develop new primaries, escalating their risk of metastasis and poor outcomes. However, current follow-up strategies are not risk-stratified, representing a critical gap in patient management. Objectiveto develop and validate a prognostic model to quantify individualized absolute risk of a first metachronous CSCC after an index tumor, accurately accounting for the high competing risk of mortality in this typically elderly population. Methodswe conducted a nationwide, population-based cohort study of 11,737 patients with a first histologically confirmed CSCC (Netherlands Cancer Registry, 2007-2008) with up to 10 years of follow-up. Data on subsequent tumors was retrieved via linkage to the Automated National Pathological Anatomy Archive (Palga). A Fine-Gray competing-risk model was developed using routinely available clinical and pathological predictors (age, sex, hematologic malignancy, basal cell carcinoma (BCC) and actinic keratosis (AK) history, presence of synchronous CSCC, primary tumor location, and differentiation). Model performance was assessed 10-fold cross-validation, quantifying discrimination (time-dependent C-index) and calibration. Resultsduring follow-up, 3,288 (28%) developed a first metachronous CSCC. The model identified key predictors: markers of cumulative UV-exposure (included AK history, [≥]5 prior BCCs), and immunosuppression (chronic lymphocytic leukaemia/small lymphocytic leukaemia). Male sex, presence of synchronous CSCC at baseline were also associated with higher risk. While discrimination was modest (cross-validated 5-year C-index: 0.64), the model demonstrated excellent calibration. Conclusionsthis competing-risk model provides individualized, well-calibrated absolute risk estimates for a first metachronous CSCC. Based on routinely available clinical features, it offers insight into how established predictors shape risk in this high-susceptibility population. External validation and the identification of novel predictors are necessary to further refine the model and support personalized dermatologic care.

Background: Cases requiring 13 or more tissue sections in Mohs micrographic surgery (MMS) demand extended operative time, additional resources, and often specialised closure techniques. Pre-operative identification of such cases would improve surgical scheduling, resource allocation, and patient counselling. We aimed to develop and validate a machine learning prediction tool using pre-operative clinical features to identify cases likely to require13 sections. Objectives: To develop and validate machine learning models for predicting which Mohs procedures will require 13 sections, using pre-operative clinical features, and to identify key predictive factors. Methods: We analysed 408 consecutive Mohs procedures with 16 pre-operative clinical variables. Thirty machine learning algorithms were evaluated, including ensemble methods (Stacking, Voting), gradient boosting (XGBoost, LightGBM, CatBoost), neural networks (3-7 layers), support vector machines, and traditional classifiers. Model performance was assessed using 5-fold stratified cross-validation and independent test set evaluation. Feature importance was determined using SHAP (SHapley Additive exPlanations) analysis. Results: The stacking ensemble achieved the highest cross-validation AUC of 0.891 (95% CI: 0.849-0.934) and test AUC of 0.884. Tumour area (cm2), calculated using the ellipse formula to approximate clinical tumour morphology, emerged as the strongest predictor (SHAP importance: 0.141), followed by tumour size dimensions (0.086 and 0.068), aggressive histopathology (0.046), and recurrence status (0.035). Wide neural network architectures (5-layer) outperformed deeper configurations (7-layer). The model demonstrated 70.7% high-confidence predictions with uncertainty <15%. Conclusions: Machine learning models using pre-operative clinical features can accurately predict which Mohs procedures will require 13 or more sections. The stacking ensemble approach provides robust predictions suitable for clinical decision support. External validation in multi-centre cohorts with diverse patient populations and practice patterns is warranted to assess model generalisability.

Abstract: Background: More than half of metastatic melanomas arise from patients initially diagnosed with early-stage melanoma. Objective biomarkers are needed to better identify high-risk patients. Objective: To evaluate the prognostic value of multiple histopathological characteristics in predicting distant metastasis risk, in early-stage melanoma. Methods: Using data from discovery set (n=442) and a population-based validation cohort (n=306, sampled from 5,815 patients) of the Dutch Early-Stage Melanoma (D-ESMEL) study, we investigated 14 histopathological characteristics of melanoma and their tumor micro-environment (TME) in an unprecedented integration, by expert pathologist scoring and automated quantitative measurements derived from a validated automated segmentation. Results: Increased immune infiltrates (40% in cases vs. 50% in controls) were associated with lower risk of metastasis. Automated immune cell density was predictive in both the discovery set and the validation cohort, outperforming the manual pathological tumor infiltrating lymphocytes. The remaining histopathological features, including mitotic activity, did not retain independent value after controlling for current staging variables. Limitations: TME evaluation in standard Hematoxylin-Eosin slides. Conclusion: TME reaction is an important determinant of melanoma progression. The automated quantification of immune cell density appears to be a biomarker for distant metastasis risk. Further investigation into specific immune cell subtypes is required to facilitate clinical integration.

ObjectivesTo identify existing DNA methylation-based prognostic biomarkers and prediction models for colorectal cancer (CRC) prognosis and to validate them in a large external cohort. DesignSystematic review and external validation study. Data sourceSystematic search in PubMed and Web of Science until October 2022 to identify epigenome-wide studies reporting methylation at CpG sites (CpGs) associated with survival among CRC patients. Validation data were drawn from the 2310 CRC patients of the DACHS study recruited from 22 hospitals in the Rhine-Neckar region in the southwest of Germany. Main outcome measuresOverall survival (OS) in CRC patients. ResultsWe identified 200 unique CpGs and 10 CpG-based prognostic models derived from 15 studies. In the external validation analysis, 1252 of 2310 patients died during follow-up (median 10.4 years). Thirty-nine CpGs (20%) and five prognostic models (50%) were independently associated with overall survival after adjustment for clinical variables. The five models had unsatisfactory discrimination ability, with area under the receiver operating characteristic curves at five years ranging from 0.54 to 0.60. The calibration accuracy of the five models using recalibrated baseline survival was also poor, and no relevant added prognostic value to traditional clinical variables was observed. Based on the Prediction Model Risk of Bias Assessment Tool, all models were rated as high risk of bias. ConclusionsOnly 20% of published CpGs associated with survival in CRC patients could be externally validated. So far derived published CpG-based prognostic models for CRC do not seem to be useful for clinical practice. Summary boxO_ST_ABSWhat is already known on this topicC_ST_ABSO_LISeveral studies have suggested that DNA methylation biomarkers could have the potential to improve prognostic accuracy for patients with colorectal cancer (CRC), but these studies mostly did not include large-scale external validation C_LIO_LIMany CpG sites associated with CRC prognosis and prognostic models based on these CpGs have been proposed C_LIO_LIAn independent study to validate these biomarkers and prediction models is essential for assessing their utility in clinical practice, but has not yet performed C_LI What this study addsO_LIThis external validation study verified the prognostic relevance of a fraction of existing DNA methylation-based prognostic biomarkers for CRC C_LIO_LIPublished CpG-based prognostic models all performed poorly in our external validation and were rated as at high risk of bias, so they do not seem to be useful for clinical practice C_LI

ImportanceEasy and accessible biomarkers to predict response to immune checkpoint inhibition (ICI)-treated melanoma are limited. ObjectiveTo evaluate artificial intelligence (AI) detected tumor-infiltrating lymphocytes (TILs) on pretreatment melanoma metastases as a biomarker for response and survival in ICI-treated patients. DesignMulticenter cohort study including patients with advanced melanoma treated with first-line anti-PD1 {+/-} anti-CTLA4 between 2016 and 2023. Median follow-up was 36.3 months. Setting11 melanoma treatment centers in the Netherlands. Participants1,202 patients with advanced cutaneous melanoma. ExposureAll patients received first-line anti-PD1 {+/-} anti-CTLA4. Main Outcome(s) and Measure(s)The percentage of TILs inside manually annotated tumor area in H&E stained pretreatment metastases was determined using the Hover-NeXt model trained and evaluated on an independent melanoma dataset containing 166,718 pathologist-verified manually annotated cells. The primary outcome was objective response rate (ORR); secondary outcomes were progression-free survival (PFS) and overall survival (OS). Correlation with manual TILs, scored according to the guidelines stated by the immune-oncology working group, was evaluated with Spearman correlation coefficients. Logistic regression and Cox proportional regression were conducted, adjusted for age, sex, disease stage, ICI type, BRAF status, brain metastases, LDH level, and performance status. ResultsMetastatic melanoma specimens were available for 1,202 patients, of whom 423 received combination therapy. Median TIL percentage was 9.9% (range 0.3% - 69.4%). A 10% increase in TILs was associated with increased ORR (adjusted OR 1.40 [95% 1.23-1.59]), PFS (adjusted HR 0.85 [95% CI 0.79 - 0.92]) and OS (adjusted HR 0.83 [95% CI 0.76 - 0.91]. Results were consistent for both patients treated with anti-PD1 monotherapy and combination treatment with anti-PD1 plus anti-CTLA4. When comparing manual TIL scoring with AI-detected TILs, associations with response and survival were consistently stronger for AI-detected TILs. Conclusions and RelevanceIn patients with advanced melanoma, higher levels of AI-detected TILs on pre-treatment H&E slides were independently associated with improved ICI response and survival. Given the accessibility of TIL scoring on routine histology, TILs may serve as a predictive biomarker for ICI outcomes. To facilitate broader validation, the Hover-NeXt architecture and model weights are publicly available. Key pointsQuestion: What is the predictive value of artificial intelligence-detected tumor-infiltrating lymphocytes (TILs) for clinical outcomes in patients with advanced melanoma receiving first-line immune checkpoint inhibition? Findings: In this multicenter cohort of 1202 patients, TILs in pretreatment metastases were quantified using a melanoma-specific publicly available AI model trained on an independent dataset. A 10% increase was associated with response (aOR 1.40 [95% CI 1.23-1.59]), progression-free survival (aHR 0.85 [95% CI 0.79-0.92]), and overall survival (aHR 0.83 [95% CI 0.76-0.91]). Associations were independent of clinical predictors. Meaning: AI-detected TILs in pretreatment melanoma metastases independently correlate with response and survival.

Structured AbstractO_ST_ABSImportanceC_ST_ABSThe genetic architecture of disease risk may not be independent of social determinants. This can be leveraged to investigate for causality. ObjectiveTo investigate for a possible causal relationship between socioeconomic status (SES) and melanoma in situ (MIS) based on their genetic architectures. DesignGenetic correlation study SettingMulticenter and population-based data sources. ParticipantsThe Ingold et al., 2024 GWAS summary statistic dataset is derived from 3,564 MIS cases, 10,552 invasive melanoma (MM) cases, and 1,022,070 melanoma-free controls. Individuals with both MIS and MM were only included as MM cases. The Kweon et al., GWAS summary statistic dataset on income, a proxy for SES, is derived from an effective sample size of 668,288 individuals. Main Outcome(s) and Measure(s)Genetic correlation between MIS, MM, and SES. ResultsWe obtained European genomic ancestries-based GWAS summary statistics for MIS (3,564 cases and 1,022,070 melanoma-free controls), MM (10,552 cases and 1,022,070 melanoma-free controls), and SES (668,288 individuals). We identify a positive and significant genetic correlation between MIS and SES (0.14, 95% CI 0.06 to 0.21; p = 5.55 x 10-04) but not MM and SES (0.05, 95% CI -0.01 to 0.11; p = 0.11). The genetic architecture of MIS subtracting that of MM (MIS-MM) remained positively and significantly correlated with the genetic architecture of SES (0.23, 95% CI 0.07 to 0.39; p = 4.18 x 10-03). In contrast, the genetic architecture of MM subtracting that of MIS (MM-MIS) is negatively correlated with the genetic architecture of SES (-0.24, 95% CI -0.42 to -0.06; p = 8.01 x 10-03). Finally, taking a Mendelian Randomization approach, we identify consistent evidence for a causal pathway between MIS and SES but not MM and SES. Conclusions and RelevanceThe genetic architecture of SES correlates with that of MIS but not MM. There is also evidence for a causal link between SES and MIS. Both these findings support an overdiagnosis of MIS. Importantly, our results demonstrate genetic risk scores for disease are not inherently independent of social determinants of diagnosis. Clinical application of genetics-based risk stratification without consideration of social determinants may have limited utility. Key PointsO_ST_ABSQuestionC_ST_ABSIs socioeconomic status genetically correlated with melanoma in situ and is there any evidence for a causal relationship? FindingsIn this genetic correlation study based on data from 3,564 melanoma in situ cases, 10,552 invasive melanoma cases, 1,022,070 melanoma-free controls, and 668,288 individuals with income data; we identify a positive and significant genetic correlation between socioeconomic status and melanoma in situ but not invasive melanoma. We also identify Mendelian Randomization-based evidence for a causal relationship between socioeconomic status and melanoma in situ but not invasive melanoma MeaningThe genetic architecture of disease risk is not independent of social determinants of diagnosis. Clinical application of genetics-based risk stratification without consideration of social determinants may have limited utility.

BackgroundNumerous studies have shown that elderly women with endometrial cancer (EC) have a higher risk of recurrence and cancer-related death. It is, however, unclear whether aging is a causal prognostic factor, or whether other risk factors become increasingly common with age. We address to this with a unique multi-method study design using state of the art statistical and causal inference techniques on datasets of three large randomised trials. MethodsData of 1801 women participating in the randomised PORTEC-1, -2 and -3 trials were used for statistical analyses and causal inference. The cohort included 714 patients with intermediate-risk EC, 427 high-intermediate risk EC patients and 660 high-risk EC patients. Associations of age with clinicopathological and molecular features were analysed using non-parametric tests. Multivariable competing risk analyses were performed to determine the independent prognostic value of age. To analyse age as a causal prognostic variable a deep learning Causal Inference model called AutoCI was used. FindingsMedian follow-up was 12{middle dot}3 years for PORTEC-1, 10{middle dot}5 years for PORTEC-2 and 6{middle dot}1 years for PORTEC-3. Both overall recurrence and EC-specific deaths significantly increased with age. Moreover, elderly women had a higher incidence of deep myometrial invasion, serous tumour histology and p53abn tumours. Age was an independent risk factor for both overall recurrence (HR 1{middle dot}02 per year, 95%CI 1{middle dot}01-1{middle dot}04; p=0{middle dot}0012) and EC-specific death (HR 1{middle dot}03 per year, 95%CI 1{middle dot}01-1{middle dot}05; p=0{middle dot}0012), and was identified as a significant causal variable. InterpretationThis study shows that advanced age is associated with more aggressive tumour features, and independently and causally related to worse oncological outcomes. Therefore, treatment for endometrial cancer in elderly women should not be de-escalated based on their age alone. FundingThe PORTEC-1, -2 and -3 trials and the associated translational studies are supported by the Dutch Cancer Society.

Cancer survivors are at an increased risk of cardiovascular disease (CVD) compared to the general population. Here, we evaluated the impact of somatic mosaic chromosomal alterations (mCAs) on death of CVD causes, coronary artery disease (CAD) causes, from cancer, and of any cause in patients with a cancer diagnosis within the UK Biobank (n=48 919). mCAs were derived from DNA genotyping array intensity data and long-range chromosomal phase inference from participants. Overall, 10 070 individuals (20.6%) carried [&ge;]1 mCA clone. In adjusted analyses, mCA was associated with an increased risk of death of CAD causes (hazard ratio [HR]: 1.37, 95% confidence interval [CI]: 1.09-1.71, P=0.006), from cancer (HR: 1.06, 95% CI: 1.00-1.11, P=0.041), and death of any cause (HR: 1.07, 95% CI: 1.02-1.12, P=0.005). Among cancer survivors, carriers of any mCA are at an increased risk of death of CAD causes and of any cause as compared to non-carriers.

Genomic testing has transformed treatment decisions for hormone receptor-positive, HER2-negative (HR+/HER2-) early breast cancer; however, it remains inaccessible to many patients worldwide due to high costs and logistical barriers. Here, we developed an artificial intelligence (AI) model using a multimodal deep learning approach that estimates Oncotype DX 21-gene recurrence scores (RS) from routine histopathology images and clinicopathologic variables, including age at diagnosis, tumor size, and receptor status. Using a foundation model pre-trained on 171,189 histopathological slides, we fine-tuned and validated our AI model on the TAILORx randomized trial (n=8,284). Among 2,407 patients in the TAILORx validation, the model classifies 45.6% of patients as low-risk, 42.4% as intermediate risk, and 12.0% as high-risk. For predicting high genomic risk disease (RS[&ge;]26), occurring in 15.9% in the TAILORx validation set, the model achieves AUC=0.898. Patient stratification by our model shows strong prognostic value across multiple clinical endpoints, including recurrence-free interval, distant recurrence-free interval, and disease-free survival. Importantly, chemotherapy benefit is demonstrated for premenopausal patients classified by our model as high AI risk and chemotherapy benefit is ruled out for postmenopausal patients classified as low AI risk. External validation across six independent cohorts (n=5,497 patients) demonstrates robust generalization of the AI model for prognostication and prediction of RS. Notably, in postmenopausal patients, the AI model reclassifies approximately 30% of clinically high-risk cases, defined by the MINDACT criteria, as low-risk. These findings demonstrate that artificial intelligence applied to standard histopathology can be a valuable tool for chemotherapy decision-making in HR+/HER2- early breast cancer. This approach can help reduce unnecessary chemotherapy and extend precision medicine, particularly in resource-limited settings, where genomic testing is not widely accessible.

BackgroundImmune checkpoint inhibitors (ICI) have improved outcomes in Merkel cell carcinoma (MCC). Population analyses suggest improved survival following the 2017 approval of ICI, but registry data lack treatment-level information including type of systemic therapy and initiation timepoint to directly estimate the benefit attributable to immunotherapy. This study compared Merkel Cell Carcinoma-specific survival between patients treated with first-line ICI versus cytotoxic chemotherapy. MethodsPatients were identified from the Seattle Merkel Cell Carcinoma Registry. Among 1,517 patients with MCC, 463 received first-line systemic therapy with either ICI or chemotherapy. Propensity scores were estimated using logistic regression including AJCC 8th stage, age, sex, MCPyV status, and immunosuppression. One-to-one nearest-neighbor matching produced balanced cohorts of 133 ICI-treated and 133 chemotherapy-treated patients. Merkel Cell Carcinoma-specific survival from therapy initiation was analyzed using Kaplan-Meier and Cox proportional hazards models with follow-up administratively censored at five years. ResultsBaseline clinical characteristics were comparable between matched cohorts. ICI therapy was associated with significantly improved Merkel Cell Carcinoma-specific survival compared with chemotherapy (log-rank p<0.0001). Five-year Merkel Cell Carcinoma-specific survival was 56.8% (95% CI 46.8-65.6) for ICI versus 23.9% (95% CI 16.9-31.6) for chemotherapy. In multivariable stage-stratified Cox analysis, ICI remained independently associated with improved Merkel Cell Carcinoma-specific survival (HR 0.32, 95% CI 0.21-0.50; p<0.0001), while immunosuppression was associated with worse Merkel Cell Carcinoma-specific survival (HR 2.03, 95% CI 1.10-3.74; p=0.0228). ConclusionsICI therapy was associated with substantially improved MCC-specific survival compared with chemotherapy.

BackgroundMelanoma is a lethal form of skin cancer with a high propensity for metastasizing, making early detection crucial. This study aims to develop a machine learning model using electronic health record data to identify patients at high risk of developing melanoma to prioritize them for dermatology screening. MethodsThis retrospective study included patients diagnosed with melanoma (cases), as well as matched patients without melanoma (controls), from Massachusetts General Hospital (MGH), Brigham and Womens Hospital (BWH), Dana-Farber Cancer Institute (DFCI), and other hospital centers within the Research Patient Data Registry at Mass General Brigham healthcare system between 1992 and 2022. Patient demographics, family history, diagnoses, medications, procedures, laboratory tests, reasons for visits, and allergy data six months prior to the date of first melanoma diagnosis or date of censoring were extracted. A machine learning framework for health outcomes (MLHO) was utilized to build the model. Performance was evaluated using five-fold cross-validation of the MGH cohort (internal validation) and by using the MGH cohort for model training and the non-MGH cohort for independent testing (external validation). The Area Under the Receiver Operating Characteristic Curve (AUC-ROC) and the Area Under the Precision-Recall Curve (AUC-PR), along with 95% Confidence Intervals (CIs), were computed. ResultsThis study identified 10,778 patients with melanoma and 10,778 matched patients without melanoma, including 8,944 from MGH and 1,834 from non-MGH hospitals in each cohort, both with an average follow-up duration of 9 years. In the internal and external validations, the model achieved AUC-ROC values of 0.826 (95% CI: 0.819-0.832) and 0.823 (95% CI: 0.809-0.837) and AUC-PR scores of 0.841 (95% CI: 0.834-0.848) and 0.822 (95% CI: 0.806-0.839), respectively. Important risk features included a family history of melanoma, a family history of skin cancer, and a prior diagnosis of benign neoplasm of skin. Conversely, medical examination without abnormal findings was identified as a protective feature. ConclusionsMachine learning techniques and electronic health records can be effectively used to predict melanoma risk, potentially aiding in identifying high-risk patients and enabling individualized screening strategies for melanoma.

BackgroundArtificial intelligence (AI)-based mobile health (mHealth) smartphone apps for skin cancer detection are increasingly available to the general population, but their impact on care is unclear. MethodsThe SPOT study is an investigator-initiated and -designed, unblinded, randomized controlled trial. Participants from a Dutch non-profit health insurance living in and around region Rotterdam the Netherlands, were recruited between August and December 2021. Participants were randomly assigned (3:2) to either free access to an AI-based mHealth app for skin cancer risk detection or care-as-usual. The primary endpoint was the difference in healthcare consumption for (pre)malignant and benign skin lesions at 12-months follow-up in the intention-to-treat population. Secondary endpoints included differences in the proportion of surgical interventions, overall use of dermatological care, and costs. FindingsAmong the 19,009 participants, the incidence of claims for (pre)malignant skin lesions was 2{middle dot}8-fold higher than among non-responders. Within the group of study participants, the skin cancer incidence was higher among the intervention group compared to the control group at 12 months follow-up (2{middle dot}7% (n=305) vs. 2{middle dot}3% (n=171); risk difference (RD) 0{middle dot}4% (95% confidence interval (CI) -0{middle dot}07 to 0{middle dot}85), p = 0{middle dot}10), though this difference was not statistically significant. Furthermore, participants in the intervention group had significantly more claims for benign skin lesions (3{middle dot}9% (n=443) vs. 2{middle dot}6% (n=198), RD 1{middle dot}3 (95% CI 0{middle dot}7 to 1{middle dot}7), p < 0{middle dot}001), underwent more surgical interventions, and had higher mean costs per participant ({euro}63 (95% CI 58 to -67), vs. {euro}47 (41 to -52); p<0{middle dot}001) compared to controls. InterpretationIn the first 12 months of this study, access to an AI-based mHealth app for skin cancer risk detection showed a modest trend toward a higher rate of skin cancer detection compared to care-as-usual. However, it also resulted in significantly more dermatological care for benign skin lesions. FundingDSW and SkinVision(R) Research in context Evidence before this studyPrior to the start of this study, we conducted a PubMed search for articles published between January 1, 2011, and December 31, 2021, using the search terms artificial intelligence AND skin cancer. This search resulted in 809 articles which were screened for relevance. We also included the results of one prospective validation study for which we had conducted the analyses ourselves, but which had not yet been published at that time. Several commercial companies have implemented such algorithms in smartphone-based mobile health (mHealth) applications, making them available to the general public. A systematic review and meta-analysis reported that AI-based apps assessing skin cancer risk from macroscopic images achieved varying sensitivity and specificity, depending on the algorithm and the type of skin cancers detected. Four studies had prospectively validated the specific mHealth app investigated in this study against histopathology, with reported sensitivities ranging from 57% to 87% and specificities from 27% to 83%. Despite promising indications, no randomised controlled trials have yet evaluated the effectiveness of AI-based mHealth apps for skin cancer screening in the general population. Added value of this studyThe SPOT study is, to our knowledge, the first randomised controlled trial to investigate how implementing an AI-based skin cancer risk detection app in the general population affects skin cancer detection and healthcare use for benign skin lesions. We found a modestly higher skin cancer incidence amongst those who were offered to use the app, though this difference was not statistically significant. However we also found those who were offered to use the app had a significantly larger increase in healthcare visits and procedures for benign skin lesions. Suggesting that implementation in the general population may involve a possible trade-off between increased skin cancer detection and unnecessary care due to overdiagnosis. Implications of all the available evidenceEven though research in a sterile setting shows potential for implementation of AI-based mHealth apps, the results from this study suggests that nationwide implementation of an mHealth with its current accuracy is not the most optimal strategy. Targeted implementation in higher-risk populations may offer a more favourable balance between benefits and harms.

ObjectiveThis study explored the association between low-dose computed tomography (LDCT)-derived body composition and melanoma incidence risk. MethodsLDCT scans from the Pittsburgh Lung Screening Study (n=3,422, 22 follow-up years) were analyzed. Body composition features were segmented and quantified from baseline scans using in-house artificial intelligence algorithms. Features were selected before modeling. Fine-Gray subdistribution hazard models assessed the association between body composition and melanoma incidence. Model performance was evaluated using time-dependent area under the curve (AUC). Restricted mean survival time (RMST) compared melanoma-free survival across BMI and body composition groups at 5, 10, and 15 years. Participants were stratified into risk groups, with risk estimated at each time point. Sex-specific analyses were conducted separately. Statistical significance was defined as p<0.05. ResultsAmong 3,422 participants, 80 developed melanoma (43 males, 37 females). In the overall model, visceral adipose tissue (VAT) volume (hazard ratio [HR]=1.27), skeletal muscle (SM) density (HR=0.81), and bone density (HR=1.33) were included, achieving a 21-year AUC of 0.68 (95% CI: 0.65-0.70). The male-specific model included only SM density (HR=0.74; AUC=0.67, 95% CI: 0.65-0.68). The female-specific model (AUC=0.68, 95% CI: 0.65-0.71) included VAT volume (HR=1.47), intramuscular adipose tissue (IMAT) ratio (HR=0.67), and bone density (HR=1.75). Higher VAT, IMAT volume, and lower SM density showed shorter melanoma-free survival and stratified risk better than BMI. Males exhibited higher estimated risk than females. ConclusionLDCT-derived body composition metrics may provide incidental insights into melanoma risk during lung cancer screening, though their predictive utility remains limited and warrants further investigation. Key PointsO_ST_ABSQuestionC_ST_ABSTo investigate the association between CT-derived three-dimensional (3D) body composition and the risk of developing melanoma. FindingsCT-derived body composition was associated with melanoma incidence. Males demonstrated higher estimated risk than females over both short- and long-term follow-up periods. Clinical RelevanceGiven melanomas high mortality and the limited effectiveness of current screening programs, these findings highlight the potential of leveraging routinely acquired lung cancer screening CT scans to enhance melanoma risk assessment.

BackgroundThe Immunoscore(R) is a validated tool for predicting colorectal cancer (CRC) prognosis, yet its adoption is impeded by complex commercial software and patient reimbursement challenges. Utilizing open-source methods, this study aimed to explore whether an immune cell score can be facilitated by focusing on single T-cell markers, to provide a simplified prognostic model in non-metastatic CRC. MethodsA multicentric prospective cohort study was conducted in non-metastatic CRC patients who underwent curative surgical resection. CD3+ and CD8+ tumor infiltrating lymphocytes (TILs) were quantified in both invasive margin (IM) and tumor core (TC) using QuPath. A composite score, termed immune cell score, mirroring the methods employed for the Immunoscore(R), was calculated based on the TIL densities (CD3-IM, CD8-IM, CD3-TC, CD8-TC]. We used a split sample approach (70:30) to estimate adjusted hazard ratios of cancer-specific survival (CSS) in a training and a validation set. Classification and regression tree analysis (CART) was performed to select the most prognostic TIL. The model incorporating the CART-selected TIL was compared to a two-tiered immune cell score model for overall performance (Brier score) and discrimination (concordance probability estimate, CPE). ResultsDuring a median follow-up time of 9.0 years, among 1260 patients, there were 203 CRC specific deaths. CART-selected CD8-IM was the most prognostic TIL at a cut-off of 231 cells/mm2. Patients with CD8-IMHi had better CSS than CD8-IMLow in both training (HR 0.58, 95% CI 0.40-0.84) and validation sets (HR 0.35, 95% CI 0.21-0.60). Brier scores of CD-8IM and immune cell score survival models were comparable in both training and validation cohort, whereas the survival discrimination of CD8-IM slightly outperformed the immune cell score in the validation set (CPE: CD8-IM 0.748, IS 0.730). ConclusionA single TIL marker, specifically CD8-IM, provided prognostic information comparable to the immune cell score. Simplified and cost-effective TIL assessments could enhance their bench to bedside translation and may guide adjuvant therapy in early-stage CRC.

Successful treatment of solid cancers relies on complete surgical excision of the tumor either for definitive treatment or before adjuvant therapy. Radial sectioning of the resected tumor and surrounding tissue is the most common form of intra-operative and post-operative margin assessment. However, this technique samples only a tiny fraction of the available tissue and therefore may result in incomplete excision of the tumor, increasing the risk of recurrence and distant metastasis and decreasing survival. Repeat procedures, chemotherapy, and other resulting treatments pose significant morbidity, mortality, and fiscal costs for our healthcare system. Mohs Micrographic Surgery (MMS) is used for the removal of basal cell and squamous cell carcinoma utilizing frozen sections for real-time margin assessment while assessing 100% of the peripheral and deep margins, resulting in a recurrence rate of less than one percent. Real-time assessment in many tumor types is constrained by tissue size and complexity and the time to process tissue and evaluate slides while a patient is under general anesthesia. In this study, we developed an artificial intelligence (AI) platform, ArcticAI, which augments the surgical workflow to improve efficiency by reducing rate-limiting steps in tissue preprocessing and histological assessment through automated mapping and orientation of tumor to the surgical specimen. Using basal cell carcinoma (BCC) as a model system, the results demonstrate that ArcticAI can provide effective grossing recommendations, accurately identify tumor on histological sections, map tumor back onto the surgical resection map, and automate pathology report generation resulting in seamless communication between the surgical pathology laboratory and surgeon. AI-augmented-surgical excision workflows may make real-time margin assessment for the excision of more complex and challenging tumor types more accessible, leading to more streamlined and accurate tumor removal while increasing healthcare delivery efficiency.

Electronic medical records (EMR) represent a rich informatics resource that remains largely unexploited for improving healthcare outcomes. Here we report a systematic text mining analysis of EMR correspondence for 4791 cancer patients treated between 2001 and 2017. Meaningful groups of text descriptors correlating with poor survival outcomes were systematically identified, and applying machine learning analysis to clinical text accurately predicted cancer patient survival at selected timepoints up to 12 months. In a validation cohort of 726 patients, inclusion of EMR descriptors to machine learning models outperformed the predictivity of conventional clinical symptom scores by 4.9% (p = 0.001). These results prove that labour-intensive EMR data collection can be repurposed to add clinical value. Extension of this approach to a broader spectrum of digital health data should transform the real-time utility of such latent informatics resources, enabling healthcare systems to be more adaptive and responsive to patient circumstances.

ImportanceMany patients will develop more than one skin cancer, however most research to date has examined only case status. ObjectiveDescribe the frequency and timing of the treatment of multiple skin cancers in individual patients over time DesignLongitudinal claims and electronic health record-based cohort study SettingVanderbilt University Medical Center database called the Synthetic Derivative, VA, Medicare, Optum Clinformatics(R) Data Mart Database, IBM Marketscan ParticipantsAll patients with a Current Procedural Terminology code for the surgical management of a skin cancer in each of five cohorts. ExposuresNone. Main Outcomes and MeasuresThe number of CPT codes for skin cancer treatment in each individual occurring on the same day as an ICD code for skin cancer over time ResultsOur cohort included 5,508,374 patients and 13,102,123 total skin cancers treated. Conclusions and RelevanceNearly half of patients treated for skin cancer were treated for more than one skin cancer. Patients who have not developed a second skin cancer by 2 years after the first are unlikely to develop multiple skin cancers within the following 5 years. Better data formatting will allow for improved granularity in identifying individuals at high risk for multiple skin cancers and those unlikely to benefit from continued annual surveillance. Resource planning should take into account not just the number of skin cancer cases, but the individual burden of disease. Key pointsQuestion: How many skin cancer patients are treated for more than one skin cancer and how soon after the first skin cancer do they occur? Findings: 43% of patients were treated for more than one skin cancer, the majority of which occurred within two years after the initial skin cancer. Just 3% of patients were treated for 10 or more skin cancers, but these patients accounted for 22% of all of the skin cancer treatments in the cohort Meaning: Nearly half of all skin cancer patients were treated for multiple skin cancers, while those without a second skin cancer after two years were less likely to be treated for a subsequent skin cancer within the next five years.