Citations

6008 Background: ATC patients present with advanced, often surgically unresectable disease with historically dismal prognosis. Median PFS and OS of DT without pembrolizumab were 6.7 and 13.5 months, respectively, in the ROAR trial. Neoadjuvant DTP achieved locoregional control without radical surgery in a retrospective series of BRAF m-ATC patients, providing rationale to evaluate efficacy and safety in a phase 2 prospective trial. Methods: In this single-arm multicenter phase 2 trial, patients with BRAF m-ATC stage IVB/IVC were enrolled in 5 US centers. Following a 3-6 week run-in with D (150 mg BID) and T (2 mg daily), P (200 mg Q3W) was added, with restaging every three (21-day) cycles. Post-operatively, patients continued P (or DTP) with radiotherapy or transitioned directly to adjuvant DTP (up to 26 cycles). Primary endpoints included R0/R1 resection rate (historically 5%) and overall survival (OS). Secondary endpoints included RECIST 1.1 response after DTP and progression-free survival (PFS). Results: Between 9/2021-1/2025, 42 patients were enrolled; 36 are included in the current analysis (3 not evaluable, 3 pending surgery) (Table). Patients received median 4 (range: 2-7) neoadjuvant DTP cycles, with 26 (72%) achieving radiographic PR/CR. 30 patients (83%) had surgery after neoadjuvant DTP, achieving R0/R1 resection in 29/30 (97%). Mean surgical morbidity score (0-4 scale, 4=unresectable) improved from 3.3 to 1.6 after DTP (p<0.01). Complete ATC pathologic response occurred in 20/30 patients (67%), while 10/30 (33%) had residual ATC in the surgical specimen. Postoperatively, 11/30 (37%) received adjuvant neck radiation, and 28/36 (78%) completed a median of 11 (range: 1-26) adjuvant DTP cycles. With median follow up 18 months, 15/36 (42%) patients died. Median OS was 20 months (95% CI: 12.6-NR); 1- and 2-year OS were 71% and 48%. Complete pathologic responders had better 2-year OS than those with residual ATC (69% vs. 22%, p=0.02). Median PFS was 13.9 months (95% CI, 7.5-NR); 1- and 2-year PFS were 57% and 36%. Grade 5 adverse events occurred in 8 patients (22%), including one possible (duodenal perforation), one probable (kidney injury with sepsis), and 6 unlikely/unrelated treatment-related deaths. Conclusions: Neoadjuvant DTP enables surgical resection in BRAF m-ATC compared with historical controls, and leads to improved PFS and OS. This approach should now be considered a standard of care for BRAF m-ATC. Clinical trial information: NCT04675710 . Total N=36 Age (y), median (range) 67 (46-86) Stage IVB/IVC, n (%) 15 (42%)/21 (58%) Best RECIST response neoadjuvant phase, n (%) CR 2 (6%) PR 24 (67%) SD 6 (17%) PD 4 (11%) Percent change target lesion diameter, mean (95% CI) -44% (-54%, -34%) Surgical morbidity score change, mean (95% CI) -1.7 (-2.1, -1.3)

11106 Background: Patients with cancer take different types of medication (e.g., enteral, parenteral), with varying treatment schedules, in various settings. They are also sometimes instructed to stop medication in response to toxicity. To measure medication nonadherence in this heterogenous population, we modified and pilot-tested a self-report measure originally developed to assess nonadherence to daily oral medication, the Domains of Subjective Extent of Nonadherence (DOSE-Nonadherence), in adults with cancer. Methods: Participants were ≥21 years, on active cancer treatment at Memorial Sloan Kettering, and spoke English and/or Spanish. A branching logic was added to allow participants to select the setting where they take cancer medication: 1) only at home, 2) only in clinic, or 3) partly at home/partly in clinic. Based on administration setting, participants reported nonadherence to medications their clinician expected them to take over a given reference period (1 week for home and 1 month for clinic medications). Participants then reported on setting-relevant reasons for nonadherence. Survey instructions/items were refined in an iterative process using feedback from a patient investigator committee and participant cognitive interviews. During interviews, we asked participants about the interpretation of instructions/items, recall period, and comprehensiveness of reasons for nonadherence, regardless of their own adherence. After initial testing in English, a Spanish version was developed through transcreation. Adherence was dichotomized for analysis (complete adherence vs. any nonadherence). For all participants who received clinic medications, concordance between chart abstracted and self-reported adherence was evaluated. Results: Of 55 participants who completed the surveys and interviews (67% in English; 33% in Spanish), 87% were female and 44% identified as Latino. The majority (73%) had breast cancer; 60% had metastatic disease. 12 of 45 who received medication in clinic reported nonadherence as did 8 of 35 who took medication at home. Participants felt able to respond accurately to both recall periods. All reasons for nonadherence were perceived as relevant to themselves or other patients. There was 89% concordance between self-reported and chart abstracted adherence for clinic-administered medication. To improve concordance, changes were made to the formatting and instructions. Of 12 participants who completed the final version of the survey for clinic-administered medication, there was 100% concordance between self-reported and chart abstracted nonadherence. Conclusions: Our results support the reliability and validity of modified DOSE-Nonadherence for cancer patients. This instrument can be used to better identify nonadherence to cancer medications in a population of adult patients with cancer receiving various treatment regimens.

e16479 Background: Recent genomic data has demonstrated SBA to be a molecularly unique entity when compared to colon or gastric cancers. The ongoing SWOG 1922 randomized clinical trial is comparing FOLFIRI to ramucirumab/paclitaxel as the second-line treatment for metastatic SBA. We sought to further define the activity of taxane-based therapy in the largest collection of SBA patients treated with this therapy. Methods: A retrospective review of SBA patients treated with taxane-based chemotherapy from 1994 to 2024 at MD Anderson Cancer Center was performed. All patients had pathologic confirmation, received > 1 cycle of therapy, and had documented tumor response evaluation. Response was assessed by the treating physician based on the imaging studies and categorized as response, stable disease or progression. Patient demographics, clinical parameters, molecular data, and taxane treatment was reviewed. Kaplan-Meier survival analysis was performed to calculate time to progression (TTP) and overall survival (OS). Results: 70 SBA patients treated with taxane-based therapy were identified. The median age at diagnosis was 57 years (range 33-80 years), majority being male (59%) with primary tumor located at duodenum (44%), Jejunum (34%), Ileum (16%). The most common presenting metastatic site was peritoneum (39%), followed by liver (31%). The genomic status revealed mutations in the following genes: 62% TP53, 47% KRAS, 24% SMAD4, and 15% APC. Taxane treatment was single agent (nab-paclitaxel, paclitaxel, or docetaxel) in 29%, gemcitabine and taxane combination in 61% and other taxane-based combinations in 10%. Therapy was given in the metastatic setting as 2nd-line in 40%, 3rd-line in 33% and 4th-line or greater in 20%. Response status was response in 24%, stable disease in 26%, and progression in 50%. Median TTP was 3.1 months (95%CI: 2.0-4.2 months), and median OS was 8.7 months (95%CI: 7.4-10.1 months). Median TTP was similar between single agent (2.6 months) and gemcitabine-based combination (2.5 months) (p =0.75). Efficacy did not differ by line of therapy nor primary tumor site. Efficacy differed by TP53 mutation status: TP53 mutated vs wildtype (response 21% vs 43%, p = 0.02; median TTP 2.4m vs 4.9m, p = 0.013; median OS 7.3m vs 10.6m, p = 0.004). Conclusions: This retrospective study of 70 SBA patients confirms the activity of taxane-based therapy demonstrating a response rate of 24% and median TTP of 3.1 months. Though limited by sample size, TP53 mutations may represent a negative predictive marker for taxane-based therapy in SBA patients. These findings support the ongoing SWOG 1922 trial investigating taxane-based therapy as a second line for metastatic SBA.

AbstractObjectiveEpilepsy is a debilitating disorder affecting more than 50 million people worldwide, and one third of patients continue to have seizures despite maximal medical management. If patients' seizures localize to a discrete brain region, termed a seizure onset zone, resection may be curative. Localization is often confirmed with stereotactic electroencephalography; however, this may require patients to stay in the hospital for weeks to capture spontaneous seizures. Automated localization of seizure onset zones could therefore improve presurgical evaluation and decrease morbidity.MethodsUsing more than 1 000 000 interictal stereotactic electroencephalography segments collected from 78 patients, we performed five‐fold cross‐validation and testing on a multichannel, multiscale, one‐dimensional convolutional neural network to classify seizure onset zones.ResultsAcross held‐out test sets, our models achieved a seizure onset zone classification sensitivity of .702 (95% confidence interval [CI] = .549–.805), specificity of .741 (95% CI = .652–.835), and accuracy of .738 (95% CI = .687–.795), which was significantly better than models trained on random labels. The models performed well across the entire brain, with top five region performance demonstrating accuracies between 70.0% and 88.4%. When split by outcomes, the models performed significantly better on patients with favorable Engel outcomes after resection or who were responsive neurostimulation responders. Finally, SHAP (Shapley Additive Explanation) value analysis on median‐normalized input data assigned consistently high feature importance to interictal spikes and large deflections, whereas similar analyses on histogram‐equalized data revealed differences in feature importance assignments to low‐amplitude segments.SignificanceThis work serves as evidence that deep learning on brief interictal intracranial data can classify seizure onset zones across the brain. Furthermore, our findings corroborate current understandings of interictal epileptiform discharges and may help uncover novel interictal morphologies. Clinical application of our models may reduce dependence on recorded seizures for localization and shorten presurgical evaluation time for drug‐resistant epilepsy patients, reducing patient morbidity and hospital costs.

Objective Raynaud phenomenon (RP) is often the initial clinical manifestation of systemic sclerosis (SSc), but some patients develop other manifestations first. To help elucidate the diversity of SSc presentation in its early stages, we describe the initial clinical manifestations and antinuclear antibody (ANA) profiles of patients in two early SSc cohorts. Methods All patient data in the Genetics vs Environment in Scleroderma Outcomes Study (GENISOS) and Collaborative National Quality and Efficacy Registry (CONQUER) cohorts were reviewed. Both studies enrolled patients within five years of the first non‐RP symptom. Results In GENISOS and CONQUER, respectively, 194 (44.2%) of 439 and 292 (31.1%) of 938 patients had a non‐RP initial symptom, most commonly puffy fingers/hands. Black patients had a non‐RP symptom before RP more commonly than patients in other race and ethnicity categories. Non‐RP first patients were more likely than RP first patients to have diffuse cutaneous involvement and joint contractures at enrollment and had a higher prevalence of RNA polymerase III antibody positivity. Conclusion In two large US cohorts, >30% of patients began to manifest SSc with puffy fingers/hands or other symptoms, without the “warning sign” of RP as their initial symptom. These patients presented with more severe skin and musculoskeletal disease on average, highlighting the importance of early recognition. The most common autoantibody associated with this presentation was RNA polymerase III. These results should be considered in efforts to recognize SSc in its earliest stages. Puffy fingers/hands, even in the absence of RP, should prompt consideration of early SSc and testing for ANA and SSc‐associated autoantibodies, including RNA polymerase III.

ABSTRACTBackgroundDeep learning (DL) models for accurate renal tumor characterization may benefit from multi‐center datasets for improved generalizability; however, data‐sharing constraints necessitate privacy‐preserving solutions like federated learning (FL).PurposeTo assess the performance and reliability of FL for renal tumor segmentation and classification in multi‐institutional MRI datasets.Study TypeRetrospective multi‐center study.PopulationA total of 987 patients (403 female) from six hospitals were included for analysis. 73% (723/987) had malignant renal tumors, primarily clear cell carcinoma (n = 509). Patients were split into training (n = 785), validation (n = 104), and test (n = 99) sets, stratified across three simulated institutions.Field Strength/SequenceMRI was performed at 1.5 T and 3 T using T2‐weighted imaging (T2WI) and contrast‐enhanced T1‐weighted imaging (CE‐T1WI) sequences.AssessmentFL and non‐FL approaches used nnU‐Net for tumor segmentation and ResNet for its classification. FL‐trained models across three simulated institutional clients with central weight aggregation, while the non‐FL approach used centralized training on the full dataset.Statistical TestsSegmentation was evaluated using Dice coefficients, and classification between malignant and benign lesions was assessed using accuracy, sensitivity, specificity, and area under the curves (AUCs). FL and non‐FL performance was compared using the Wilcoxon test for segmentation Dice and Delong's test for AUC (p < 0.05).ResultsNo significant difference was observed between FL and non‐FL models in segmentation (Dice: 0.43 vs. 0.45, p = 0.202) or classification (AUC: 0.69 vs. 0.64, p = 0.959) on the test set. For classification, no significant difference was observed between the models in accuracy (p = 0.912), sensitivity (p = 0.862), or specificity (p = 0.847) on the test set.Data ConclusionFL demonstrated comparable performance to non‐FL approaches in renal tumor segmentation and classification, supporting its potential as a privacy‐preserving alternative for multi‐institutional DL models.Evidence Level4.Technical EfficacyStage 2.

Physicians-in-training experience high rates of mental health, burnout, and well-being concerns, yet face many barriers to accessing care. Burnout is associated with poor self-rated health, sleep disturbances, and other health risks. Unfortunately, residents often face barriers to participating in health-promoting activities due to lack of time, perceived stigma, confidentiality concerns, and low awareness of available services. To improve access to health resources, and specifically mental health support, we developed a pilot opt-out well-being check-in program, offered via videoconferencing, at a large academic medical center. Internal medicine and neurology residents ( N = 93) were protected from clinical duties to meet individually with a psychologist. The 45-minute session consisted of a brief assessment, identification of stressors, discussion of well-being goals, and referral to further services as indicated. The study goals were to establish the feasibility of implementing a program-wide intervention, assess resident satisfaction with the intervention, and reduce barriers to mental health and well-being resources. The study authors developed a brief post-session satisfaction survey. Analysis included participation rates and responses to satisfaction survey questions. 64.5% of residents attended the check-in session. Among the 57% who responded to a post-intervention survey, over 90% agreed that the session was valuable, convenient, and recommended program continuation. This health promotion program successfully offered tailored mental health and well-being support within demanding residency training structures. Opt-out programs that minimize barriers to participation are a valuable part of multifaceted workplace well-being initiatives.

Abstract Background Primary aldosteronism (PA), a primary adrenal disorder leading to excessive aldosterone production by one or both adrenal glands, is a common cause of hypertension. It is associated with an increased risk of cardiovascular complications compared with primary hypertension. Despite effective methods for diagnosing and treating PA, it remains markedly underdiagnosed and undertreated. Objective To develop an updated guideline that provides a practical, clinical approach to identifying and managing PA to improve diagnosis rates and encourage targeted treatment. Methods The Guideline Development Panel (GDP), composed of a multidisciplinary panel of clinical experts and experts in systemic review methodology, used the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach to define 10 questions related to the diagnosis and treatment of PA. Systematic reviews were conducted for each question. The GDP used the GRADE Evidence to Decision (EtD) framework to consider contextual factors, such as stakeholder values and preferences, costs and required resources, cost-effectiveness, acceptability, feasibility, and the potential impact on health equity. Results We suggest that all individuals with hypertension be screened for PA by measuring aldosterone and renin and determining the aldosterone to renin ratio, and that subsequent clinical care be guided by the results. We suggest that individuals with PA receive PA-specific therapy, either medical or surgical. In individuals who screen positive for PA, we suggest (1) commencement of PA-specific medical therapy in individuals who do not desire or are not candidates for surgery and in situations where the probability of lateralizing PA (excess aldosterone produced by one adrenal) is low based on screening results; and (2) aldosterone suppression testing in situations when screening results indicate an intermediate probability for lateralizing PA and individualized decision making confirms a desire to pursue eligibility for surgical therapy. In those who test positive by aldosterone suppression testing, and in those in whom screening results show a high probability of lateralizing PA (obviating the need for aldosterone suppression testing), we suggest adrenal lateralization with computed tomography scanning and adrenal venous sampling prior to deciding the treatment approach (medical vs surgical). In all individuals with PA and an adrenal adenoma, we suggest performing a 1-mg overnight dexamethasone suppression test. We suggest the use of mineralocorticoid receptor antagonists (MRAs) over epithelial sodium-channel (ENaC) inhibitors in the medical treatment of PA. We suggest the use of spironolactone over other MRAs, given its lower cost and greater availability; however, all MRAs, when titrated to equivalent potencies, are anticipated to have similar efficacy in treating PA. Thus, MRAs with greater mineralocorticoid receptor specificity and fewer androgen/progesterone receptor-mediated side effects may be preferred in some situations. In individuals receiving MRA therapy, we suggest monitoring renin and, in those whose hypertension remains uncontrolled and renin is suppressed, titrating the MRA to increase renin. Conclusion These recommendations provide a practical framework for the diagnosis and treatment of PA. They are based on currently available literature and take into consideration outcomes that are important to key stakeholders. The goal is to increase identification of individuals with PA and, by initiating PA-specific medical or surgical therapy, improve blood pressure control and reduce PA-associated adverse cardiovascular events. The guidelines also highlight important knowledge gaps in PA diagnosis and management.

Abstract The 21st Century Cures Act Final Rule (Final Rule) increases patient access to their health records but raises concerns about distress and misunderstanding of automatically released results and documentation. Little is known about genetic counselors' (GCs') experiences with the Final Rule. In Fall 2023, we conducted a cross‐sectional survey of US GCs about the Final Rule's perceived impact on practice and assessed if being the ordering versus non‐ordering provider for genetic testing affected perceptions of Final Rule‐related changes. GCs ( n = 102) reported demographic and workplace characteristics, institutional policy changes, workflow changes, and perceived patient harms and benefits due to the Final Rule. To compare ordering and non‐ordering providers, we conducted Fisher's exact tests for categorical variables and trend tests for ordinal variables. Open‐response questions elicited examples of positive and negative patient impacts, the effects of patient characteristics, and practice changes in response to the Final Rule. Twenty‐seven GCs were ordering providers and 66 were non‐ordering providers. Relative to ordering providers, non‐ordering providers expressed stronger agreement with statements indicating concern about the emotional impact on patients reviewing results or notes without support ( p = 0.002, 33% vs. 51% strongly agree), patients misunderstanding or misinterpreting results ( p = 0.025, 44% vs. 70% strongly agree), and patients contacting the inappropriate party to discuss results ( p = 0.023, 19% vs. 48% strongly agree). Agreement with the statement “patients have more knowledge/context/questions at our disclosure session due to previous results review” also differed by ordering provider status ( p = 0.007). In open responses, GCs expressed concerns about patients' strong emotional reactions, patients misinterpreting results, workflow disruptions, and widening health disparities. Benefits included patients' ability to be reassured, informed, or empowered earlier; ease of sharing health information; more efficient workflow due to the automatic release of results. These results emphasize the importance of clear communication within health systems and between patients and providers.