Faculty Bibliography

IMPORTANCE/UNASSIGNED:There exists substantial heterogeneity in outcomes within T stages for patients with cutaneous squamous cell carcinoma (cSCC). OBJECTIVE/UNASSIGNED:To 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 PARTICIPANTS/UNASSIGNED:To 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 MEASURES/UNASSIGNED:Performance 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. RESULTS/UNASSIGNED:The 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 RELEVANCE/UNASSIGNED:The 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:Lymphovascular invasion (LVI) is regarded as a high-risk feature of cutaneous squamous cell carcinoma (CSCC) but is currently absent from CSCC staging systems. OBJECTIVE:To assess whether LVI serves as an independent predictor of major poor outcomes in CSCC. METHODS:Twelve centers contributed to a multinational CSCC database. Clinical and pathologic risk factors, treatment, and patient outcomes were retrospectively collected. CSCCs were stratified based on LVI status. Tumors that developed major poor outcomes defined as nodal metastasis, in-transit metastasis, distant metastasis, and disease-specific death were identified. RESULTS:A total of 23,166 CSCCs were identified, 179 were LVI+ tumors (0.8%). LVI+ tumors had a higher cumulative incidence of major poor outcomes than those without LVI (33.5% vs. 3.2% at 3 years; overall cumulative incidence function p < 0.001). In an adjusted analysis, LVI+ tumors had an 82% increase in the rate of developing major poor outcomes when compared to LVI- tumors (subdistribution hazard ratio (SHR) = 1.82; p = 0.002). Notably, LVI+ low-stage BWH tumors (T1 or T2a) had a greater cumulative incidence of major poor outcomes compared to LVI- BWH low-stage tumors (20.7% vs. 1.61% at 3 years, overall cumulative incidence function p < 0.001). LIMITATIONS/CONCLUSIONS:Retrospective study design CONCLUSION: The presence of LVI in CSCC is a high-risk feature that is an independent predictor of metastasis and disease-specific death in both low and high BWH stage tumors.

IMPORTANCE/UNASSIGNED:Cutaneous squamous cell carcinoma (CSCC) risk stratification is central to management, and physicians rely on tumor staging systems to estimate risk. The Brigham and Women's Hospital (BWH) T staging system predicts risk based on 4 tumor risk factors (RFs). However, stage is not precisely associated with the number of RFs, as BWH stage T2b includes CSCCs with 2 and 3 RFs. OBJECTIVE/UNASSIGNED:To determine how RF number is associated with the risk of recurrence, metastasis, and disease-related death. DESIGN, SETTING, AND PARTICIPANTS/UNASSIGNED:This retrospective multination cohort study of CSCCs diagnosed between October 1, 1991, and July 19, 2023, was conducted at 12 centers in the US (10), Spain (1), and Brazil (1). Invasive CSCCs with confirmed negative margins longer than 14 days were included. Tumors were excluded if they were metastatic at presentation or received adjuvant therapy. Data were analyzed from October 2023 to August 2024. INTERVENTIONS OR EXPOSURES/UNASSIGNED:CSCCs were stratified by the number of the following RFs (0, 1, 2, 3, or 4): a diameter of 2 cm or larger, poorly differentiated histology, tumor extension beyond subcutaneous fat, and large caliber nerve invasion. MAIN OUTCOMES AND MEASURES/UNASSIGNED:Five-year cumulative incidences of local recurrence, nodal metastasis, distant metastasis, and disease-specific death. RESULTS/UNASSIGNED:A total of 16 844 CSCCs were included (5978 female individuals [35.5%]; median [IQR] age, 73.9 [65.7-81.8] years), with 0 (12 657 [75.1%]), 1 (2892 [17.2%]), 2 (1015 [6.0%]), 3 ( 225 [1.3%]) or 4 (55 [0.3%]) RFs. Median (IQ) follow up time was 33.6 (14.5-60.3) months. For local recurrence, the risk increased as the number of RF increased from 0 (1.7%; 95% CI, 1.5%-2.0%) to 1 (5.0%; 95% CI, 4.1%-5.9%) to 2 (8.8%; 95% CI, 7.0%-11.0%) to 3 (16.0%; 95% CI, 11.0%-22.0%) to 4 (33.0%; 95% CI, 19.0%-47.0%; P < .001 for between-group differences). This increase was also observed for nodal metastasis (0.6% [95% CI, 0.4%-0.7%] vs 3.6% [95% CI, 2.9%-4.4%] vs 11.0% [95% CI, 9.2%-13.0%] vs 20.0% [95% CI, 15.0%-26.0%] vs 28.0% [95% CI, 15.0%-42.0%], respectively; P < .001), distant metastasis (0.2% [95% CI, 0.1%-0.3%] vs 1.1% [95% CI, 0.7%-1.6%] vs 2.3% [95% CI, 1.4%-3.4%] vs 7.9% [95% CI, 4.6%-12.0%] vs 8.4% [95% CI, 2.6%-19.0%], respectively; P < .001), and disease-specific death (0.3% [95% CI, 0.2%-0.4%] vs 1.9% [95% CI, 1.4%-2.7%] vs 5.4% [95% CI, 4.0%-7.0%] vs 11.0% [95% CI, 6.7%-16.0%] vs 25% [95% CI, 12%-39%], respectively; P < .001). CSCCs with 3 RFs had higher cumulative incidences of local recurrence (1.6-fold), nodal metastasis (1.9-fold), distant metastasis (4.3-fold), and disease-specific death (1.9-fold) compared with those with 2 RFs. CONCLUSIONS AND RELEVANCE/UNASSIGNED:The results of this cohort study suggest that the number of RFs is an indicator of risk, and among BWH T2b tumors, those with 3 RFs represent a higher risk subset.

BACKGROUND:Cutaneous squamous cell carcinoma (CSCC) is a prevalent disease for which improved risk stratification strategies are needed. OBJECTIVE:To develop a novel prognostic model (herein "riSCC") for CSCC and compare riSCC performance to Brigham and Women's Hospital and American Joint Committee on Cancer Staging eighth edition T staging systems. METHODS:Retrospective 12-center, multinational cohort study of CSCCs from 1991 to 2023. Clinical and pathologic risk factors, treatments, and outcomes were collected. Fine-Gray model was employed for each outcome with inverse probability of treatment weighting. A final model was trained for prospective use and estimation of hazard ratios. RESULTS:Twenty-three thousand one hundred sixty-six localized CSCC tumors were included. riSCC prognostic model performed superiorly to American Joint Committee on Cancer eighth edition and Brigham and Women's Hospital T staging for all outcomes. At 5 years, the C-index for riSCC ranged from 0.74 for local recurrence to 0.87 for disease specific death. LIMITATIONS/CONCLUSIONS:Retrospective study design. CONCLUSION/CONCLUSIONS:riSCC prognostic model offers fine-grained risk estimates and improved stratification for important CSCC outcomes compared to T staging systems.

BACKGROUND:While Brigham and Women's Hospital (BWH) T1 cutaneous squamous cell carcinomas (CSCCs) are overall low risk, a small subset develop poor outcomes. OBJECTIVE:To evaluate the impact of minor risk factors on poor outcomes in T1 tumors. METHODS:Data was collected retrospectively from 11 centers. Univariable and multivariable regression analyses were performed evaluating the impact of minor risk factors (moderate differentiation, diameter 1-2 centimeters, fat invasion, and small-caliber perineural invasion [PNI]) on poor outcomes. Cumulative incidence function (CIF) plots were created for time to poor outcomes by number of minor risk factors. RESULTS:15,481 BWH T1 tumors were included, of which 90 (0.58%) developed major poor outcomes and 332 (2.1%) developed any poor outcome. Minor risk factors that were significant on multivariable analysis included moderate differentiation, diameter, and subcutaneous fat invasion. CIF plots demonstrated an increased risk of poor outcomes with presence of multiple minor risk factors; the risk of metastasis and major poor outcomes exceeded 5% in tumors with 3 minor risk factors. LIMITATIONS/CONCLUSIONS:Retrospective design, limited number of major poor outcomes. CONCLUSION/CONCLUSIONS:T1 tumors with multiple minor risk factors may be eligible for closer surveillance. Future staging systems should consider incorporating both major and minor risk factors.

Primary cutaneous squamous cell carcinoma (cSCC) is responsible for ~10,000 deaths annually in the United States. Stratification of risk of poor outcome at initial biopsy would significantly impact clinical decision-making during the initial post operative period where intervention has been shown to be most effective. Using whole-slide images (WSI) from 163 patients from 3 institutions, we developed a self supervised deep-learning model to predict poor outcomes in cSCC patients from histopathological features at initial diagnosis, and validated it using WSI from 563 patients, collected from two other academic institutions. For disease-free survival prediction, the model attained a concordance index of 0.73 in the development cohort and 0.84 in the Mayo cohort. The model's interpretability revealed that features like poor differentiation and deep invasion were strongly associated with poor prognosis. Furthermore, the model is effective in stratifying risk among BWH T2a and AJCC T2, known for outcome heterogeneity.

BACKGROUND:Impaired immunity may drive the increased incidence and aggression of cutaneous squamous cell carcinoma (cSCC) in patients with hematologic malignancy; however, precise mechanisms and prognostic biomarkers remain undefined. CD73 maintains elevated immunosuppressive adenosine levels and is associated with poor prognosis in several tumor microenvironments. OBJECTIVE:Identify poor outcome biomarkers in patients with cSCC and hematologic malignancy. MATERIALS AND METHODS/METHODS:Differentially expressed genes in tumors from patients with hematologic malignancy experiencing good (n = 8) versus poor (n = 7) outcomes were identified by NanoString analysis. Results were validated at the protein level using CD73 immunohistochemistry in cSCC patients with (n = 38) and without (n = 29) hematologic malignancy. RESULTS:Forty-eight genes were differentially expressed in tumors from patients with hematologic malignancy experiencing good versus poor outcomes. CD73 gene expression was >2-fold higher in patients with poor versus good outcomes or normal skin. Significantly increased CD73 protein levels were observed in cSCC tumors with poor versus good outcomes from patients with hematologic malignancies (p < .01), whereas no differences were noted in tumors with poor versus good outcomes from patients without hematologic malignancies (p = .49). CONCLUSION/CONCLUSIONS:CD73 is highly expressed in poor prognosis cSCC from patients with hematologic malignancy and may represent a useful biomarker and potential therapeutic target.