PD-1-based immune checkpoint inhibition (ICI) is the major backbone of current melanoma therapy. Tumor PD-L1 expression represents one of few biomarkers predicting ICI therapy outcome. The objective of the present study was to systematically investigate whether the type of tumor tissue examined for PD-L1 expression has an impact on the correlation with ICI therapy outcome. Pre-treatment tumor tissue was collected within the prospective DeCOG cohort study ADOREG/TRIM (CA209-578; NCT05750511) between February 2014 and May 2020 from 448 consecutive patients who received PD-1-based ICI for non-resectable metastatic melanoma. The primary study endpoint was best overall response (BOR), secondary endpoints were progression-free (PFS) and overall survival (OS). All endpoints were correlated with tumor PD-L1 expression (quantified with clone 28-8; cutoff ≥5%) and stratified by tissue type. Tumor PD-L1 was determined in 95 primary tumors (PT; 36.8% positivity), 153 skin/subcutaneous (34.0% positivity), 115 lymph node (LN; 50.4% positivity), and 85 organ (40.8% positivity) metastases. Tumor PD-L1 correlated with BOR if determined in LN (OR = 0.319; 95% CI = 0.138-0.762; P = 0.010), but not in skin/subcutaneous metastases (OR = 0.656; 95% CI = 0.311-1.341; P = 0.26). PD-L1 positivity determined on LN metastases was associated with favorable survival (PFS, HR = 0.490; 95% CI = 0.310-0.775; P = 0.002; OS, HR = 0.519; 95% CI = 0.307-0.880; P = 0.014). PD-L1 positivity determined in PT (PFS, HR = 0.757; 95% CI = 0.467-1.226; P = 0.27; OS; HR = 0.528; 95% CI = 0.305-0.913; P = 0.032) was correlated with survival to a lesser extent. No relevant survival differences were detected by PD-L1 determined in skin/subcutaneous metastases (PFS, HR = 0.825; 95% CI = 0.555-1.226; P = 0.35; OS, HR = 1.083; 95% CI = 0.698-1.681; P = 0.72). For PD-1-based immunotherapy in melanoma, tumor PD-L1 determined in LN metastases was stronger correlated with therapy outcome than that assessed in PT or organ metastases. PD-L1 determined in skin/subcutaneous metastases showed no outcome correlation and therefore should be used with caution for clinical decision making. Bristol-Myers Squibb (ADOREG/TRIM, NCT05750511); German Research Foundation (DFG; Clinician Scientist Program UMEA); Else Kröner-Fresenius-Stiftung (EKFS; Medical Scientist Academy UMESciA).

Background: Artificial intelligence (AI)–enabled decision support systems are critical tools in medical practice; however, their reliability is not absolute, necessitating human oversight for final decision-making. Human reliance on such systems can vary, influenced by factors such as individual psychological factors and physician experience. Objective: This study aimed to explore the psychological factors influencing subjective trust and reliance on medical AI’s advice, specifically examining relative AI reliance and relative self-reliance to assess the appropriateness of reliance. Methods: A survey was conducted with 223 dermatologists, which included lesion image classification tasks and validated questionnaires assessing subjective trust, propensity to trust technology, affinity for technology interaction, control beliefs, need for cognition, as well as queries on medical experience and decision confidence. Results: A 2-tailed t test revealed that participants’ accuracy improved significantly with AI support (t222=−3.3; P<.001; Cohen d=4.5), but only by an average of 1% (1/100). Reliance on AI was stronger for correct advice than for incorrect advice (t222=4.2; P<.001; Cohen d=0.1). Notably, participants demonstrated a mean relative AI reliance of 10.04% (139/1384) and a relative self-reliance of 85.6% (487/569), indicating a high level of self-reliance but a low level of AI reliance. Propensity to trust technology influenced AI reliance, mediated by trust (indirect effect=0.024, 95% CI 0.008-0.042; P<.001), and medical experience negatively predicted AI reliance (indirect effect=–0.001, 95% CI –0.002 to −0.001; P<.001). Conclusions: The findings highlight the need to design AI support systems in a way that assists less experienced users with a high propensity to trust technology to identify potential AI errors, while encouraging experienced physicians to actively engage with system recommendations and potentially reassess initial decisions.

Digital histopathology poses several challenges such as label noise, class imbalance, limited availability of labelled data, and several latent biases to deep learning, negatively influencing transparency, reproducibility, and classification performance. In particular, biases are well known to cause poor generalization. Proposed tools from explainable artificial intelligence (XAI), bias detection, and bias discovery suffer from technical challenges, complexity, unintuitive usage, inherent biases, or a semantic gap. A promising XAI method, not studied in the context of digital histopathology is automated concept-based explanation (ACE). It automatically extracts visual concepts from image data. Our objective is to evaluate ACE’s technical validity following design science principals and to compare it to Guided Gradient-weighted Class Activation Mapping (Grad-CAM), a conventional pixel-wise explanation method. To that extent, we created and studied five convolutional neural networks (CNNs) in four different skin cancer settings. Our results demonstrate that ACE is a valid tool for gaining insights into the decision process of histopathological CNNs that can go beyond explanations from the control method. ACE validly visualized a class sampling ratio bias, measurement bias, sampling bias, and class-correlated bias. Furthermore, the complementary use with Guided Grad-CAM offers several benefits. Finally, we propose practical solutions for several technical challenges. In contradiction to results from the literature, we noticed lower intuitiveness in some dermatopathology scenarios as compared to concept-based explanations on real-world images.

Adjuvant treatment of melanoma patients with immune-checkpoint inhibition (ICI) and targeted therapy (TT) significantly improved recurrence-free survival. This study investigates the real-world situation of 904 patients from 13 German skin cancer centers with an indication for adjuvant treatment since the approval of adjuvant ICI and TT. From adjusted log-binomial regression models, we estimated relative risks for associations between various influence factors and treatment decisions (adjuvant therapy yes/no, TT vs. ICI in BRAF mutant patients). Of these patients, 76.9% (95% CI 74–80) opted for a systemic adjuvant treatment. The probability of starting an adjuvant treatment was 26% lower in patients >65 years (RR 0.74, 95% CI 68–80). The most common reasons against adjuvant treatment given by patients were age (29.4%, 95% CI 24–38), and fear of adverse events (21.1%, 95% CI 16–28) and impaired quality of life (11.9%, 95% CI 7–16). Of all BRAF-mutated patients who opted for adjuvant treatment, 52.9% (95% CI 47–59) decided for ICI. Treatment decision for TT or ICI was barely associated with age, gender and tumor stage, but with comorbidities and affiliated center. Shortly after their approval, adjuvant treatments have been well accepted by physicians and patients. Age plays a decisive role in the decision for adjuvant treatment, while pre-existing autoimmune disease and regional differences influence the choice between TT or ICI.