Background Oral cavity squamous cell carcinoma (OSCC) is a global health burden, where negative margins are essential for reducing recurrence and improving survival. Intraoperative frozen-section analysis is limited by time, sampling error, and interpretive variability, underscoring the need for more reliable margin assessment. Reflectance confocal microscopy (RCM) enables real-time, in vivo high-resolution imaging, but accuracy depends on expert interpretation. This study evaluated the diagnostic performance of an artificial intelligence (AI)-driven model for RCM in OSCC, aiming to develop a point-of-care platform for intraoperative use. Methods Patients with biopsy-confirmed OSCC underwent in vivo RCM imaging using a handheld intraoral probe before biopsy. Histopathology was the reference standard. A deep learning model was developed with the Google Cloud Vertex AI Automated Machine Learning (AutoML) Vision platform and trained on 4,090 annotated RCM images (1,998 benign, 2,092 malignant). Performance was compared with blinded expert pathologist and RCM readers. Results The AI model achieved an area under the precision-recall curve (AUC-PR) of 0.99 and an area under the receiver operating characteristic curve (AUC-ROC) of 0.99, with sensitivity 98.09%, specificity 95.00%, accuracy 96.58%, positive predictive value (PPV) 95.35%, and negative predictive value (NPV) 97.94%. Expert readers showed sensitivity 90.00%, specificity 98.30%, accuracy 94.15%, PPV 88.20%, and NPV 96.60%. Inter-reader agreement was 95.00% for benign and 81.70% for malignant cases. Conclusions AI-driven RCM interpretation provides an accurate, rapid, noninvasive approach for OSCC diagnosis and intraoperative margin assessment. It outperformed expert readers and can reduce reliance on frozen-section analysis, streamline workflows, and improve outcomes.

12066 Background: Pruritus is a debilitating symptom in oncologic patients, sometimes leading to interruption of cancer therapy. Gabapentinoids such as gabapentin and pregabalin are effective for the treatment of uremic and neuropathic pruritus. We report on the safety and efficacy of gabapentinoids for oncologic pruritus defined as any pruritus originating from malignancy, oncologic therapies, or other cancer-associated toxicities, such as cutaneous graft-versus-host disease (GVHD). Methods: In this single-center retrospective cohort study conducted at the Memorial Sloan Kettering Cancer Center between 4/1/2019 and 8/1/2024, 224 patients who were prescribed gabapentinoids for oncologic pruritus were included. Patients taking gabapentinoids for indications other than pruritus were excluded. The primary efficacy endpoint was ≥ 1-grade improvement for pruritus severity on the Common Terminology Criteria for Adverse Events (CTCAE) v5.0 scale. Results: The cohort was predominantly male (54%) and White (67%), with Black (14.3%) and Asian (12.5%) patients also represented; the mean age was 63.2 ± 15 years. Gabapentinoids were most prescribed for drug-induced causes (58.9%) and malignancy-related pruritus (14.3%). Of the drug-induced causes, immune checkpoint inhibitors (40.2%), monoclonal antibodies (12.9%; including mogamulizumab, brentuximab vedotin, and enfortumab vedotin), and tyrosine kinase inhibitors (11.4%) were the most common triggers. Pregabalin was more commonly prescribed compared to gabapentin (80.8% vs. 19.2%). Ninety percent of patients (n=202) experienced a ≥1-CTCAE grade improvement at a median total daily dosage of 50 mg (IQR, 25 mg) for pregabalin and 300 mg (IQR, 50 mg) for gabapentin. Median time for patient-reported improvement was 18.5 days and there was no significant difference in gabapentin response between pruritus etiologies. The most common adverse event was sedation (8.5%), with 3 patients discontinuing gabapentinoids due to fatigue. Conclusions: This study is the first to report that gabapentinoids are a safe and effective way to treat oncologic pruritus stemming from malignancy, cancer therapies, or cutaneous GVHD. Prospective trials would further establish gabapentinoids’ safety and efficacy profile in the oncologic population. Outcomes & adverse events of patients on gabapentinoids. Variable Category/Statistic N (%) CTCAE Grade Change Did Not Improve (0 Grade) 22 (9.8%) Improved 1 Grade 132 (58.9%) Improved ≥ 2 Grade 70 (31.3%) Antineoplastic Therapy Interrupted Due to Oncologic Pruritus Yes 26 (11.7%) No 157 (70.4%) Not Applicable 40 (17.9%) Adverse Events* Sedation 19 (8.5%) Leg Swelling 2 (0.9%) Other 9 (4.0%) *The remaining 194 (86.6%) patients did not report any adverse events related to the gabapentinoids.

e24199 Background: Radiation therapy is a cornerstone of treatment for early-stage breast cancer; however, up to one third of patients develop grade 2–3 acute radiation dermatitis (ARD), which is associated with pain, pruritus, and impaired quality of life. Therapeutic options for established grade 2-3 ARD are limited, and not all patients respond adequately to topical corticosteroids. Oleogel-S10 (Filsuvez) is a topical betulin-based gel approved by the U.S. Food and Drug Administration for dystrophic and junctional epidermolysis bullosa that promotes keratinocyte migration and epidermal barrier regeneration. Prior phase III studies have demonstrated accelerated wound healing with Oleogel-S10. We hypothesized that Oleogel-S10 could facilitate healing of high-grade ARD following breast irradiation. Methods: This IRB-approved (Memorial Sloan Kettering Cancer Center IRB #21-091), single-center, phase II, double-blind, randomized controlled trial enrolled women ≥18 years who developed grade 2–3 ARD during conventionally fractionated whole-breast radiation therapy. Patients were randomized 1:1 to Oleogel-S10 or placebo and instructed to apply standard-of-care triamcinolone 0.1% cream daily with study gel nightly for three weeks. The primary endpoint was percent change in ARD body surface area from baseline to day 14, assessed using novel Canfield three-dimensional (3D) imaging. Secondary endpoints included ARD resolution, pigmentation changes, patient-reported outcomes (PRO-CTCAE), and adverse events. Results: Nineteen patients were enrolled, of whom 11 completed the study (six placebo, five Oleogel-S10); attrition rates were similar between groups. No significant difference was observed between groups in percent change in ARD surface area at day 14. Hyperpigmentation at follow-up was lower in the Oleogel-S10 arm, although this difference did not reach statistical significance. Skin pain severity was lower in the Oleogel-S10 group by CTCAE and PRO-CTCAE assessments, but this difference was not statistically significant. Adverse event rates were similar between groups, with no grade 4 or 5 adverse events reported. Conclusions: Although Oleogel-S10 did not improve the primary endpoint of wound surface area as calculated by a novel 3D imaging tool compared with standard-of-care triamcinolone, treatment was associated with a trend toward improvement in skin pain and hyperpigmentation. These findings may represent early signals of therapeutic activity in ARD, particularly for patient-centered outcomes. Interpretation is limited by early study closure related to COVID-19–associated recruitment challenges and small sample size. Larger, adequately powered studies are warranted to further evaluate the role of Oleogel-S10 in the management of grade 2-3 ARD. Clinical trial information: NCT05190770 .

BACKGROUND:Artificial intelligence (AI)-based predictive systems generate heat maps that highlight informative regions as proxy explanations for diagnostic predictions. Comparing AI- and dermatologist-derived heat maps may help determine whether AI allocation aligns with clinically relevant features-an essential step toward improving interpretability. OBJECTIVE:To compare dermatologists and AI-generated heat maps of dermoscopic images. METHODS:Four dermatologists, blinded to diagnoses, inspected 120 dermoscopic images in randomized order. Their eye movements were tracked to generate heat maps. The dataset included melanomas, basal cell carcinomas, squamous cell carcinomas, nevi, benign keratoses, and vascular lesions in equal numbers. For the same images, class activation maps were produced using the DEXI algorithm (Dermoscopy EXplainable Intelligence, Canfield Scientific's Vectra software system). Overlap was assessed using pixel-wise rank correlation. Inter-dermatologist correlation served as the upper reference, and median null correlations between DEXI and non-homologous dermatologist maps defined the lower reference. RESULTS:Median pixel-wise correlation was ρ = 0.540 for dermatologist-DEXI, ρ = 0.591 for inter-dermatologist, and ρ = 0.434 for null comparisons. LIMITATION:Small sample sizes for each skin lesion type and the absence of lesion size data. CONCLUSIONS:Dermatologists and DEXI heat maps showed substantial overlap, suggesting shared diagnostic anchors, underscoring the model's potential for interpretability.

OBJECTIVES:The primary objective was to evaluate the long-term recurrence rate following RCM-OCT-guided laser ablation of basal cell carcinomas (BCCs). The secondary objective was to assess patient-reported outcomes, including willingness to undergo the treatment again, scar acceptability, and adverse effects. METHODS:A prospective cohort study evaluated 104 patients with 148 BCCs (2017-2021). Low risk BCC subtypes and with tumor depth of ≤ 300 µm were determined using RCM-OCT, and eligible lesions underwent same-day Er:YAG laser ablation with immediate post-treatment imaging to confirm clearance. Patients underwent clinical and RCM-OCT or RCM surveillance at 6 and 12 months and annually thereafter to assess recurrence. A patient-reported questionnaire was conducted in 2023. RESULTS:Fifty-eight lesions (39.2%) in 39 patients met the criteria for laser ablation. Most lesions were located on the trunk (29/58, 50.0%), with median tumor depth and lesion size of 200 µm and 9.8 mm, respectively. Immediate post-ablation RCM-OCT demonstrated imaging-defined complete clearance after a single session in 53/58 lesions (91.4%). Residual tumors (5/58, 8.6%) were successfully treated during the same visit. Fifty-four of 58 lesions (93.1%) had at least 6 months of follow-up (median 32 months; min/max 8-70). One recurrence occurred at 53 months and was successfully retreated, corresponding to 1.04 recurrences per 100 person-years of follow-up. The patient-reported questionnaire response rate was 79.5% (31/39). Overall, 96.8% (30/31) reported acceptable scar appearance, 83.9% (26/31) were likely to choose laser ablation again (Score 1-2 of 5), and 9.7% (3/31) reported mild, transient side effects. CONCLUSIONS:RCM-OCT-guided Er:YAG laser ablation offers a "one-stop" treatment approach for selected low-risk BCCs, demonstrating low long-term recurrence rates and high patient satisfaction.

12156 Background: Biologic agents are increasingly used to treat immune-related cutaneous adverse events (ircAEs). However, limited data exist regarding immune checkpoint inhibitor (ICI) rechallenge and ICI efficacy following biologic therapy for ircAE management. Methods: We conducted a retrospective cohort study of patients enrolled under U01AR077511-01 (PIs Leung, Kern, Lacouture) treated with ICIs who developed dermatologist-confirmed grade ≥2 ircAEs. Demographics, ircAE treatments, and dermatologic and oncologic outcomes were collected via chart review and compared using t-tests and chi-square tests. Results: We identified 159 patients who developed grade ≥2 ircAEs, including 80 treated with biologic agents. Mean age was 66.8 years; 40% were female, 79% White, and 78% had stage IV disease. The most common tumor types were kidney (20%), lung (16%), and melanoma (12%). Common ICI regimens included ICI monotherapy (46%), ICI–ICI combination (21%), and ICI+chemotherapy (19%). Pruritus (23%), maculopapular rash (21%), and eczema (19%) were the most common ircAEs. Most patients received topical steroids (88%), while systemic steroids were used sparsely (13%). Biologics were used in ~50% of patients; the most common agents were dupilumab (43%), omalizumab (33%), and ustekinumab (11%). Median biologic duration was 16 weeks, with 25% receiving therapy for >40 weeks. ICI rechallenge occurred in 60% of biologic-treated versus 67% of non-biologic patients (p=0.35). Rechallenge rates were 67% vs 65% for grade 2 (p=0.85) and 44% vs 74% for grade 3 ircAEs (p=0.05) (biologic vs non-biologic, respectively). Antitumor outcomes were similar between biologics and non-biologic groups: 31% vs 28% achieved partial/complete response, 43% had stable disease, and 26% vs 29% had progressive disease, respectively. Conclusions: In this oncodermatology-managed cohort, ircAEs were primarily treated with topical corticosteroids and targeted biologics, with infrequent systemic steroid use. Biologic therapy enabled ICI rechallenge in most patients, and oncologic outcomes were comparable to those not receiving biologics. Overall, these findings support biologics as a steroid-sparing option for moderate-to-severe ircAEs that may allow continued immunotherapy without apparent loss of antitumor efficacy, warranting prospective validation. Clinical characteristics. Variable Overall n=159 No Biologics n=79 Biologics n=80 Age at ICI (mean) 66.8 64.3 69.4 Tumor Types Kidney/Lung/Melanoma (%) 20/16/12 22/17/9 18/15/15 Line of treatment 1L/2L/2+L (%) 52/25/23 49/27/24 54/24/22 ircAEs: Pruritus/MPR/Eczema/Others (%) 23/21/19/37 22/33/17/28 25/9/21/45 ircAE Grade 2/Grade 3 (%) 74/26 76/24 71/29 Dupilumab/Omalizumab/Ustekinumab/Others (%) 33/33/11/23 N/A 33/33/11/23 Systemic Steroids - Yes (%) 13 8 18 ICI Rechallenge - Yes (%) 64 67 60 Response CR-PR/SD/PD (%) 30/43/27 31/43/26 28/43/29

BACKGROUND:CD19 CAR T-cell therapy is a significant advance in B-NHL and ALL. This study describes the incidence, onset, and factors of dermatologic adverse events (dAE) post-therapy. METHODS:A retrospective analysis at Memorial Sloan Kettering Cancer Center (4/2013-8/2020) identified 193 patients undergoing CD19 CAR T-cell therapy. We aimed to characterize dAEs post CAR T-cell therapy including the 100-day cumulative incidence, time to dAE onset, and associations with cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS). RESULTS:Eighty-two patients experienced 94 dAEs within the first 100 days (incidence: 0.42 (95% CI: 0.36-0.51) post CAR T-cell therapy. Common dAEs were rash (30.9%, n = 29) including maculopapular rash, inflammatory papules, and local erythema; infection (23.4%, n = 22) including cellulitis and folliculitis; and xerosis (16.0%, n = 15). Specific early onset dAEs included rash and chemotherapy-related events, eg, alopecia, mucositis (median 12- and 17-days postinfusion, respectively). Thrombocytopenic purpura and xerosis presented later (median 22-and 25-days). CRS and dAEs occurred in 33.5% of patients, with CRS preceding dAEs in 86% of cases. Among 15% with ICANS and dAEs, ICANS was antecedent in 67%. CONCLUSION:dAEs following CAR T-cell therapy are common but mostly low-grade and often manifest within the initial month postinfusion.

This case series examines the effectiveness of dupilumab as a steroid-sparing therapy among patients with antibody-drug conjugate–related cutaneous toxicities.

INTRODUCTION:Atypical network is a dermoscopic criterion that helps in the diagnosis of melanoma. Despite its importance, the interpretation of atypical networks varies widely among experts. OBJECTIVE:This study examined the impact of viewing the whole lesion versus viewing foci of pigment network (i.e., snippets) in isolation from within the lesion on expert classification of pigment network in dermoscopic images. METHOD:Six dermoscopy experts, blinded to the diagnosis, each evaluated a total of 92 images (80 nevi and 12 melanomas) for the presence of typical versus atypical pigment network. While 57% of images had consistent classification of the network between whole lesion and snippets, 43% shifted the network classification between the snippet to the whole lesion view. Melanomas were more prone than nevi to intra-rater discrepancy between whole lesion and snippets (54.2% vs. 41.7%; odds ratio (OR): 1.65; 95% confidence interval (CI): 1.11-2.47). The inter-observer agreement was higher for the snippet view (65.22%) than for the whole lesion view (55%). RESULTS:These findings suggest that both the objective morphology of the pigment network and the subjective interpretation of the network in context with other features within the lesion influence expert classification of pigment network. CONCLUSION:Factors such as the variability in the distribution, thickness, and color of network lines, overall pattern, and other dermoscopic structures likely contributed to the classification changes.

BACKGROUND:Ultraviolet-induced fluorescent dermoscopy (UVFD) is a novel dermoscopic technique that has shown potential to improve diagnostic accuracy for select conditions in small case series. However, its performance has not yet been assessed in a large cohort of cutaneous tumors. OBJECTIVE:To evaluate the role of UVFD in cutaneous tumors. METHODS:This retrospective, multicenter study included consecutive lesions assessed at 3 tertiary care clinics in Chile, France, and Poland. Standard polarized dermoscopic images and UVFD images were independently reviewed by 2 expert dermoscopists, with consensus evaluation. Established dermoscopic criteria, as well as border definition and the presence of "ochre color" under UVFD, were specifically assessed. RESULTS:Five hundred fifty-one lesions were included. The most frequent diagnosis was basal cell carcinoma (28.2%), followed by melanocytic nevus (18.6%) and melanoma (12.8%). Keratin, comedo-like openings, and fissures and ridges were better seen under UVFD. The border definition of lesions was also better seen under UVFD. Ochre color was seen under UVFD mostly in invasive melanomas. LIMITATIONS:Retrospective study. Some diagnoses were underrepresented. CONCLUSION:UVFD enhances specific keratin-related dermoscopic structures and colors. It also provides for better border delineation and may help in identifying invasive melanomas via the presence of ochre color.

BACKGROUND:Immune checkpoint inhibitors (ICIs) have revolutionized cancer therapy but are associated with treatment-limiting immune-related cutaneous adverse events (irCAEs). Immune checkpoint inhibitor-related bullous pemphigoid (irBP), a severe, blistering irCAE occurs in 0.3%-1.5% of patients receiving ICI therapy. While systemic steroids can be effective, they are associated with significant toxicity and may mitigate -immunotherapy antitumor efficacy. Consequently, steroid-sparing therapies are needed. Dupilumab, an IL-4 and IL-13 receptor antagonist, has demonstrated efficacy in non-ICI-related BP and appears promising for managing irBP. METHODS:We conducted a retrospective review of patients treated with dupilumab for irBP from April 2020 to April 2024. Clinical data, outcomes, and adverse events were assessed. Inhibitor-related bullous pemphigoid response was categorized as complete response (CR), partial response (PR), or no response (NR). RESULTS:In all, 17 patients (59% male, 82% non-Hispanic White; mean age 72.7 years) developed irBP while receiving PD-1/PDL-1 inhibitors. Sixteen patients (94%) received dupilumab for active irBP and one (6%) for prevention of recurrence. Dupilumab achieved CR of irBP for 12 patients (75%) and PR for 2 (12%) patients with active irBP. Ten (62%) achieved CR with dupilumab systemic monotherapy. Median time to first response was 19.5 days (range = 3-50). Most patients with CR (58%) failed prior oral corticosteroid therapy. The patient treated prophylactically experienced no irBP recurrence. Dupilumab was well-tolerated, with no adverse events. CONCLUSIONS:Dupilumab is a promising steroid-sparing option for irBP, achieving initial response in under 20 days for most cases. Dupilumab is a valuable tool to manage this challenging irCAE while minimizing risk related to systemic steroid treatment.

6082 Background: Oral cavity squamous cell carcinoma (SCC) remains a common malignancy in the head and neck region, with challenges in tumor resection and recurrence prevention. Traditional methods like frozen-section analysis are limited by time delays, sampling errors, and tissue distortion. Reflectance Confocal Microscopy (RCM) provides a noninvasive alternative for real-time, high-resolution imaging, but interpreting RCM images accurately requires expert knowledge. Integrating artificial intelligence (AI) could improve the accuracy and reliability of RCM image interpretation for diagnosing SCC and assessing surgical margins. The integration of machine learning and artificial intelligence (AI) has the potential to enhance the accuracy and reliability of RCM image interpretation, providing a more efficient tool for diagnosing oral cavity SCC and assessing surgical margins in real-time during surgery. Methods: We developed an AI model using Google Cloud’s AutoML platform to classify RCM images for diagnosing oral cavity SCC and evaluating tumor margins. The dataset comprised 4,090 RCM images from 83 patients, including 1,998 images of benign tissue and 2,092 images of malignant tissue. The dataset was divided into training (80%), validation (10%), and test (10%) sets. A single-label classification approach was employed to differentiate benign and malignant tissue. Model performance was evaluated using sensitivity, specificity, accuracy, F1 score, and negative predictive value. Results: The AI model achieved an area under the curve (AUC) of 0.99, sensitivity of 98.09%, specificity of 95.00%, accuracy of 96.58%, and an F1 score of 96.70%. In comparison, expert human readers in our prior study achieved accuracies of 90.91% for normal tissue and 81.7% for tumor detection, highlighting the accuracy of the AI model's diagnostic performance. Conclusions: The combination of RCM imaging with AI-powered analysis provides an accurate, noninvasive method for real-time diagnosis and surgical margin assessment in oral cavity SCC. The AI-driven model has excellent sensitivity, specificity, and overall accuracy, offering a potentially efficient and reliable modality for the real-time evaluation of digital RCM images. This approach can reduce the time required for intraoperative margin assessment, minimize patient anesthesia time, and overcome challenges related to conventional histopathology, ultimately improving surgical outcomes in patients with oral cavity SCC.