Citations

Background Challenges in predicting which neoantigens mediate tumor rejection limit the efficacy of neoantigen vaccines to treat cancers, especially for cancers with a high mutational burden like cutaneous squamous cell carcinoma (cSCC). Only a small portion of neoantigens prioritized by current methods elicit effective T cell responses, demonstrating the critical need for improved criteria for the prediction of tumor-rejecting neoantigens. Methods Publicly available human cSCC datasets were used to assess the frequency of shared mutations between patients. A transplantable ultraviolet light-induced mouse model of cSCC was generated. The mutational signature and driver mutations in the mouse model were compared with human tumors. Neoantigens were prioritized in the mouse model, and tumor-rejecting neoantigens were identified through (enzyme-linked immunosorbent spot (ELISpot) and in vivo prophylactic vaccination. Binding of the neoantigens and corresponding wild-type peptides to major histocompatibility complex (MHC) class I was determined. Structural modeling of peptide:MHC complexes was performed to assess for changes in structural characteristics of the neoantigens relative to the wild-type peptides. Results A minority of human cSCC tumors shared neoantigens. The mouse cSCC model recapitulated the mutational signature and driver mutations found in human disease and was constrained by CD8 T cells. Two MHC class I neoantigens were identified in the mouse model that constrained cSCC growth. One tumor-rejecting neoantigen exhibited improved MHC binding, and the other had increased solvent accessibility of the mutated residue, compared with wild-type. Across known neoantigens that do not impact MHC binding, increased exposure of the mutated residue distinguished tumor-rejecting from non-immunogenic neoantigens. Conclusions Given the paucity of shared mutations, this work supports the need for personalized neoantigen vaccines in cSCC. To facilitate further discovery, we provide a clinically relevant mouse cSCC model with two defined neoantigens that mediate tumor rejection. Structural changes in the exposure of features that promote T cell receptor recognition defined tumor-rejecting neoantigens. Incorporation of structural modeling to predict changes in T cell receptor accessibility is anticipated to improve the selection of neoantigens for inclusion in personalized cancer vaccines.

Evaluation of T-cell receptor (TCR) β-chain constant region 1 (TRBC1) expression is an alternative T-cell clonality assessment method. However, evaluation of TRBC1 immunohistochemistry (IHC) for distinguishing mycosis fungoides (MF)/Sezary syndrome (SS) from reactive inflammatory infiltrates in skin is incomplete. We evaluated the utility of a novel dual TRBC1-CD3 IHC stain in skin biopsies with MF/SS (n=40), reactive (n=24), or atypical T-cell infiltrates indeterminant for MF/SS (n=9). Twenty of 24 reactive cases showed clear polytypic TRBC1 expression (median percent TRBC1 positivity among CD3-positive T cells [%TRBC1+] 50% in dermis, 42.5% in epidermis among all cases), while 34/40 MF/SS were clearly monotypic (%TRBC1+ either ≤20% or ≥85%) (sensitivity 85.0%, specificity 91.7%, positive predictive value 94.4%, negative predictive value 78.6%). Discordance between TRBC1 expression and diagnosis was associated with few neoplastic/monotypic T cells and/or lack of physical separation between neoplastic and non-neoplastic populations. Among patch/plaque MF/SS, TRBC1 showed similar diagnostic sensitivity (80.0% vs. 86.7%) and high categorical correlation (monotypic vs. not monotypic, 80%) with TCR gene rearrangement results. TRBC1 interpretation was reproducible, with ≥2/3 and 3/3 pathologists rendering identical interpretations in 100% and 86% of cases, respectively, after independent review and 100% agreement following collective review. Digital image analysis confirmed visual %TRBC1+ accuracy ( r =0.9749, P <0.0001). On the basis of these results, we propose %TRBC1+ cutoffs of <25% or >75% for establishing T-cell monotypia in skin. Considering such thresholds, TRBC1-CD3 evaluation facilitated diagnostic refinement in 7/9 (78%) atypical cases. Overall, TRBC1-CD3 IHC clearly and rapidly aids diagnosis of cutaneous T-cell proliferations.

ABSTRACT Importance Risk stratification of cutaneous squamous cell carcinoma (CSCC) is central to effective management. Despite advancements in the 8th edition of the American Joint Committee on Cancer (AJCC8) staging system, the distinctiveness of T‐stages remains limited. Objective To evaluate the effectiveness of the Brigham and Women's Hospital (BWH) and the AJCC8 staging systems in predicting poor outcomes in head and neck (HN) CSCC. Design A retrospective, multinational cohort study of CSCCs diagnosed between January 10, 1991, and December 31, 2023. Setting Twelve centers across the United States (10), Spain (1), and Brazil (1). Participants Patients with invasive CSCC who underwent curative‐intent surgical treatment. Exclusions included cases of lip CSCC, cases with prior HN cancer with associated regional disease, patients with a history of chemotherapy/radiotherapy for other HN neoplasms, and recurrent primary tumors. Exposure Tumors were staged according to both the AJCC8 TNM staging system and the BWH tumor classification. Main Outcomes and Measures Local recurrence (LR), nodal recurrence (NR), distant recurrence (DR), and disease‐specific death (DSD). Results A total of 9852 excised tumors from 3168 patients were included. The 2 systems had comparable monotonicity and homogeneity. Significant differences could be observed in 5‐year cumulative incidence for DSD in both BWH and AJCC8, and also for LR in BWH. Higher T‐stages exhibited similar curves regarding NR and DR for both AJCC8 and BWH staging systems. Overall, we observed high specificity and NPV, low sensitivity and PPV, and moderately high c‐indices for both the BWH and AJCC8 staging systems in predicting the main outcomes for HNCSCC. Conclusion and Relevance Current AJCC8 and BWH staging systems can accurately predict survival in HNCSCC, although there are still important characteristics to be addressed in future staging systems for better stratification according to the other main outcomes.

BACKGROUND There is a scarcity of artificial intelligence models trained on frozen pathology. One way to expand the clinical utility of models trained on permanent pathology is by applying them to frozen sections and fine-tune based on weaknesses. OBJECTIVE To qualitatively evaluate a deep learning model trained on permanent pathology to classify squamous cell carcinoma on Mohs surgery frozen sections to learn model shortcomings and inform retraining and fine-tuning. MATERIALS AND METHODS The authors trained a model for classification of tumor on 746 skin biopsy slides and tested it on 15 Mohs surgery frozen sections. The authors estimated performance metrics and compared the regions of interest generated by the model with the original H&E slides. RESULTS The model achieved an AUC-ROC of 0.985 and 0.796 for tumor classification in permanent pathology and in frozen sections, respectively. Regions of interest for frozen sections with scarce tumor areas were inaccurate, focusing on normal tissue for slides classified as false negative, or highlighting structures different from tumor (e.g., inflammation, muscle, and nerves) for slides classified as true positive. CONCLUSION Deep anatomical structures more commonly present in Mohs frozen pathology might represent data out-of-distribution for models trained on permanent pathology, potentially leading to inadequate model outputs.

ImportanceThere exists substantial heterogeneity in outcomes within T stages for patients with cutaneous squamous cell carcinoma (cSCC).ObjectiveTo determine whether a customized generative pretrained transformer model, trained on a comprehensive dataset with more than 1 trillion parameters and equipped with relevant focused context and retrieval augmented generation (RAG), could excel in aggregating and interpreting vast quantities of data to develop a novel class-based risk stratification system that outperforms the current standards.Design, Setting, and ParticipantsTo build the RAG knowledge base, a systematic review of the literature was conducted that addressed risk factors for poor outcomes in cSCC. Using the RAG-enabled generative pretrained transformer (GPT) model, we developed a novel class-based risk stratification system that assigned point values for risk factors, culminating in a GPT-based prognostication system called the artificial intelligence–derived risk score (AIRIS). The system’s performance was validated on a combined prospective and retrospective cohort of 2379 primary cSCC tumors (1996-2023) with at least 36 months of follow-up, against Brigham and Women’s Hospital (BWH) and American Joint Committee on Cancer Staging Manual, eighth edition (AJCC8) systems in stratifying risk for locoregional recurrence (LR), nodal metastasis (NM), distant metastasis (DM), and disease-specific death (DSD).Main Outcomes and MeasuresPerformance metrics evaluated included distinctiveness, homogeneity, and monotonicity, as defined by the AJCC8, as well as sensitivity, specificity, positive predictive value, negative predictive value, accuracy, the area under the receiver operating characteristic curve, and concordance.ResultsThe median age at diagnosis was 73 (IQR, 64-81) years, with 38.5% female patients and 61.5% male patients. The AIRIS prognostication system demonstrated superior sensitivity across all outcomes (LR, 49.1%; NM, 73.7%; DM, 82.5%; and DSD, 72.2%) and the highest area under the receiver operating characteristic curve values (LR, 0.69; NM, 0.81; DM, 0.85; and DSD, 0.80), indicating significantly enhanced discriminative capability compared with the BWH and AJCC8 systems. While all systems were comparably distinctive, the AIRIS prognostication system consistently demonstrated the lowest proportion of tumors exhibiting poor outcomes in low-risk categories, suggesting its improved homogeneity and monotonicity.Conclusions and RelevanceThe results of this diagnostic study suggest that the AIRIS system outperforms the existing BWH and AJCC8 prognostication systems, potentially providing a more effective tool for predicting poor outcomes in cSCC. This study illustrates the potential of large language models in refining prognostic tools, offering implications for treating patients with cancer.

Background: Hidradenitis suppurativa (HS) is a chronic, recurrent inflammatory condition and is associated with significant psychosocial impacts on patient quality of life. Objective: This study aimed to characterize the utilization of patient-reported outcomes (PROs) in HS clinical trials and their concordance with trial primary endpoints. Methods: A systematic review of clinical trials was performed using the publicly available U.S. National Library of Medicine (ClinicalTrials.gov) in June 2023 utilizing the search terms hidradenitis, hidradenitis suppurativa, and suppurativa hidradenitis. Study title, start year, trial status, intervention, design, location, and primary and secondary outcomes were collected. To assess for concordance of patient and provider-reported outcomes, we identified published placebo-controlled trials that included Dermatology Life Quality Index (DLQI) and/or a numeric rating scale (NRS) for pain, the 2 most utilized PROs. Results: One hundred sixty-four HS clinical trials were identified, of which 115 were interventional studies. A total of 65.2% ( n = 107) of HS trials included at least one PRO. Pain NRS (42.7%, n = 70) and DLQI (42.1%, n = 69) were the most frequently used PRO instruments. The use of PROs in clinical trials increased over time, with 50% of trials between 2020 and 2023 utilizing PROs. Of 11 published HS trials, 82% ( n = 9) trials showed concordance of provider-assessed and PROs in at least one treatment arm. Limitations: Significance and standard deviations of PROs were rarely reported, preventing the calculation of significance and is a limitation of this review. Conclusion: While lesion counts provide single snapshots of disease, PROs can capture perceptions of lesions not actively present. PROs are frequently concordant with both achievement and nonachievement of primary endpoints in clinical trials.