Citations

e24089 Background: Pancreatic cancer is associated with high symptom burden and frequent acute care utilization. Although early palliative care improves quality of life in advanced malignancies, real-world data describing its timing and downstream impact in pancreatic cancer remain limited. Methods: We conducted a retrospective cohort study using Epic Cosmos (2015–2025). Adults (≥18 years) with pancreatic cancer were identified using ICD-10 code C25.*. Early palliative care was defined as a documented palliative care encounter (ICD-10 Z51.5) within 0–2 months of cancer diagnosis. Patients were categorized as early palliative care versus not-early palliative care (late or no palliative care). A landmark approach was applied, with outcomes assessed beginning 2 months after diagnosis to mitigate immortal time bias. Outcomes included opioid use, antiemetic use, occupational/physical therapy (OT/PT) utilization, emergency department (ED) encounters resulting in inpatient admission, and inpatient length of stay. Results: Among 4,641 patients with pancreatic cancer, 464 (10.0%) received early palliative care, 365 (7.9%) received palliative care later, and 3,812 (82.1%) had no documented palliative care encounter. Compared with patients without early palliative care, early palliative care recipients demonstrated higher utilization of symptom-directed interventions, including opioids (40.6% vs 22.5%), antiemetics (37.2% vs 20.3%), and occupational/physical therapy services (6.2% vs 3.2%). Early palliative care was associated with a higher proportion of emergency department encounters resulting in inpatient admission (52.5% vs 23.6%), consistent with earlier consolidation of symptom-driven care. Mean inpatient length of stay was similar between groups (6.0 vs 6.2 days). Overall survival appeared comparable between groups; however, survival estimates in EHR-based analyses are limited by residual confounding and timing constraints. Conclusions: In a large national EHR cohort, early palliative care within two months of pancreatic cancer diagnosis was infrequently used but was associated with greater engagement in symptom-directed medications and supportive services, as well as a distinct downstream pattern of acute care utilization. These findings suggest that earlier integration of palliative care may improve coordination of symptom-focused care and represent an actionable quality-of-care target in pancreatic cancer.

e24088 Background: Chemotherapy-induced nausea and vomiting (CINV) significantly affects treatment tolerance and quality of life. ASCO guidelines recommend a NK-1 receptor antagonist with a 5-HT3 receptor antagonist and corticosteroid for patients receiving highly emetogenic chemotherapy (HEC). Olanzapine has emerged as an alternative antiemetic, with evidence suggesting noninferiority in triple regimens. Given its favorable cost profile, we performed a pooled analysis comparing olanzapine- versus aprepitant-based triple antiemetic regimens for CINV prevention. Methods: This PRISMA-compliant review was prospectively registered in PROSPERO (CRD420261284983). Five databases were systematically searched to identify studies comparing olanzapine (OLZ) versus aprepitant (APR) based triple antiemetic regimens. The primary endpoint was complete response (CR) for prophylaxis of chemotherapy-induced nausea and vomiting (CINV) in adult patients receiving HEC. Statistical analyses were performed in R (version 2025.05.0+496; Posit). Results: 13 studies were included in the final synthesis, comprising 1,557 patients (OLZ: n = 779; APR: n = 788). CR was comparable between groups (RR 1.02, 95% CI 0.95–1.09; I² = 55.8%). Exclusion of Mehta et al. (2017) reduced heterogeneity to zero (I² = 0%). In the acute phase ( < 24 h), CR was also similar between OLZ and APR (RR 1.09, 95% CI 0.98–1.20; I² = 82.7%); however, OLZ demonstrated superior CR in the 10-mg dose subgroup (RR 1.08, 95% CI 1.03–1.14; I² = 74%). For delayed-phase CR (24-120 h), the pooled effect estimate was RR 1.11 (95% CI 0.98–1.27; I² = 84.3%). Control of nausea was significantly improved in the OLZ group compared with APR (RR 1.19, 95% CI 1.01–1.40; I² = 72.8%), particularly in anthracycline–cyclophosphamide–based regimens (RR 1.30, 95% CI 1.09–1.54; I² = 62.9%) and the 10-mg OLZ dose subgroup (RR 1.22, 95% CI 1.01–1.49; I² = 75.2%). Nausea control was also superior in both the acute phase (RR 1.13, 95% CI 1.01–1.27; I² = 76.5%) and the delayed phase (RR 1.31, 95% CI 1.05–1.65; I² = 93.3%). Control of vomiting was comparable between OLZ and APR (RR 0.99, 95% CI 0.92–1.07; I² = 0%), with no significant differences observed in the acute ( < 24 h) phase (RR 1.03, 95% CI 0.97–1.10; I² = 49%) or the delayed (24–120 h) phase (RR 1.00, 95% CI 0.97–1.04; I² = 0%). Conclusions: OLZ-based triple antiemetic regimens provide comparable CR and vomiting control to APR, with superior nausea control, particularly in anthracycline–cyclophosphamide regimens and at the 10-mg dose.

e24087 Background: Radiotherapy-induced nausea and vomiting (RINV) negatively impacts adherence and quality of life. Multiple antiemetic strategies exist, but comparative effectiveness and ranking across regimens remain unclear, particularly regarding olanzapine-containing combinations. Methods: We conducted a systematic review and network meta-analysis of randomized and comparative studies evaluating antiemetic prophylaxis for RINV. Databases included MEDLINE, Embase, CENTRAL, and clinical trial registries. Interventions were categorized into clinically interpretable nodes (e.g., 5-HT3 antagonist ± dexamethasone, NK1-based combinations, olanzapine-containing regimens). The primary outcome was complete response (no emesis, no rescue therapy) during acute and delayed phases. Secondary outcomes included no nausea, severity, sedation, and discontinuation. Random-effects network meta-analysis was performed with treatment ranking using SUCRA probabilities. Results: Twenty-two studies encompassing 3,145 patients were included. Olanzapine-containing regimens ranked highest for complete response during both acute (SUCRA 92%) and delayed phases (SUCRA 89%), outperforming standard 5-HT3 + dexamethasone (SUCRA 64%). NK1-based triple therapy showed intermediate efficacy (SUCRA 78%). Olanzapine regimens had modest increases in sedation (RR 1.14, 95% CI 1.01–1.29) but no significant differences in discontinuation rates. Subgroup analyses indicated consistent benefits across radiation sites, fractionation schedules, and emetogenic risk. Conclusions: Olanzapine-based antiemetic regimens provide superior prevention of RINV compared with standard therapies, with acceptable tolerability across radiotherapy settings. Clinical Takeaway: Incorporating olanzapine into prophylactic antiemetic strategies can optimize symptom control, improve patient adherence, and enhance quality of life during radiotherapy.

e24014 Background: Cardio-oncology traditionally focuses on cardiotoxicity prevention and acute cardiovascular-events. However, the functional, vocational, and mental health challenges that define daily survivorship remain largely hidden, representing a critical gap in holistic care. Methods: Utilizing the 2024 National Health Interview Survey (NHIS), we analyzed a nationally representative cohort of U.S. adults (≥18 years) across four groups: (1) Controls (no cancer or CVD), (2) CVD only, (3) cancer only, and (4) cardio-oncology. We evaluated five patient-centered metrics reflecting quality of life: functional and work limitations, assistive equipment use, personal assistance needs, and overall disability. Multivariable regression models, using Taylor series linearization for complex survey weighting, were adjusted for age, sex, race, hypertension, diabetes, hyperlipidemia, smoking, psychological distress (K6 score), and insurance status. Results: Among 32,599 adults (Weighted N = 259.7M), 2.9% lived with cardio-oncology multi-morbidity (weighted N = 5.34M; mean age 73y). After multivariable adjustment, functional limitations were highest in the cardio-oncology (63.6%, OR 2.34 [95%CI 1.89-2.88]), exceeding both CVD only (OR 1.99) and cancer only (OR = 1.26) groups. This population was also at the highest risk of disability (OR 2.74) and work limitations (31.7%, OR = 2.53). One in five cardio-oncology survivors reported severe mobility impairment (difficulty walking/steps), with a need for assistive equipment (OR = 2.86) and personal assistance (OR = 1.38). Conclusions: Cardio-oncology survivors face a functional and vocational burden significantly exceeding that of cancer or CVD alone. Care models must expand beyond event prevention to integrate routine functional screening and early rehabilitation to preserve quality of life in this high-risk population. Adjusted odds ratios by disease phenotype. Outcome CVD Only (vs. Control) Cancer Only (vs. Control) Cardio-Oncology (vs. Control) OR (95% CI) OR (95% CI) OR (95% CI) Functional limitation 1.99 (1.75–2.28) 1.26 (1.14–1.39) 2.34 (1.89–2.88) Uses special equipment 2.56 (2.20–2.98) 1.26 (1.08–1.46) 2.86 (2.32–3.53) Needs assistance 1.51 (1.17-1.94) 0.84 (0.64-1.12) 1.38 (0.99-1.92) Work limitation 2.12 (1.88-2.39) 1.19 (1.07-1.33) 2.53 (2.11-3.03) Disability 2.20 (1.90–2.56) 1.11 (0.96–1.29) 2.74 (2.26–3.32) Note: Models adjusted for age, sex, race, hypertension, diabetes, hyperlipidemia, smoking status, psychological distress, and insurance type. Reference group = Adults with no history of cancer or CVD. CVD = Cardiovascular disease; OR = Odds Ratio; CI = Confidence Interval.

e23554 Background: Radiographic imaging remains the standard of care for disease surveillance and response assessment in soft tissue sarcoma (STS). Circulating tumor DNA (ctDNA) has emerged as a potential biomarker in solid tumors; however, its feasibility and detectability in STS remain incompletely defined due to histologic heterogeneity and variable tumor DNA shedding. We evaluated ctDNA detectability and its relationship to radiographic disease status in the most common STS subtypes: undifferentiated pleomorphic sarcoma (UPS) and leiomyosarcoma (LMS). Methods: We retrospectively analyzed patients with UPS or LMS who underwent serial ctDNA testing using a personalized assay (Signatera) alongside routine radiographic surveillance between 2022 and 2025. Any detectable mean tumor molecules (MTM) >0 was considered positive. ctDNA results were paired with the closest imaging within ±30 days and assessed using radiology report–based determinations of disease status. Exploratory sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were estimated per paired assessment using radiographic progression or recurrence as the reference standard. Results: 13 patients were included (7 female, 6 male), with a median age of 62 years (range 28–75); 7 patients (54%) had LMS and 6 (46%) had UPS. A total of 59 ctDNA–imaging paired assessments were analyzed (median 4 per patient). Overall ctDNA detectability was low and histology-dependent. Baseline ctDNA was detectable in 2 of 13 patients (15%), both with LMS. During longitudinal follow-up, ctDNA remained persistently undetectable in 10 of 13 patients (77%), including all UPS patients and 4 of 7 LMS patients. Detectable ctDNA occurred almost exclusively in LMS and was often observed in association with radiographic disease progression. Across all paired assessments, 46 of 59 (78%) were concordant. Discordance included false-negative ctDNA (negative ctDNA with radiographic disease/progression; n=9) and transient ctDNA positivity without immediate radiographic progression (n=4). Where longitudinal follow-up was available, subsequent ctDNA or imaging assessments frequently demonstrated eventual alignment; lead-time analysis for transient positives was limited by sample size. Using radiographic progression as the reference standard, exploratory per-pair test characteristics demonstrated a sensitivity of approximately 64%, specificity of 88%, PPV of 80%, and NPV of 77%. Conclusions: In this real-world cohort, ctDNA detectability differed by histology and was observed predominantly in LMS. When detectable, ctDNA trends often aligned with radiographic disease progression or recurrence, supporting a potential complementary role in disease monitoring for selected patients. These findings require confirmation in larger, prospective studies.

e23510 Background: There is minimal data comparing the prognostic models (PM) developed to predict recurrence of Gastrointestinal stromal tumors (GIST). Current PM include tumor size, location, and mitotic index (MI) but lack genotypic data. This analysis compares the performance of the National Institute of Health, Armed Forces Institute of Pathology, and Memorial Sloan Kettering Cancer Center models for predicting tumor recurrence with and without adjustment for c-kit Exon 11 mutation status (MS). Methods: This is an IRB-approved single-institution retrospective analysis. Adults with GIST mentioned in their medical records from 2014 – 2025 at the authors’ institution were identified. 330 patients with documented biopsy-proven GIST were included. Data was collected by manual review. PM were evaluated using multivariate analysis, adjusting for surgical resection, systemic therapy, and c-Kit exon 11 MS. Predictive performance was measured by area under the curve (AUC) and receiver operating characteristics (ROC) analysis. Results: Mean age at diagnosis was 64.6 years (SD =13.1 years). 46.7 % of patients were male. Primary GIST was most common in the stomach (63.9%) and small bowel (23.6%). Primary tumors were an average of 6.5 cm (SD 5.6 cm) with a MI of 6.6/50 HFU (SD = 13.4 50/HFU). Patients received resection (80.9%) and first-line systemic therapy with imatinib (43.3%). Recurrence occurred in 17.9% of patients. 19.4% of patients completed genetic testing, of which 51.6% were c-Kit exon 11 mutation positive. The predictive performance of all PM was not statistically different with (p value = 0.16) or without (p value = 0.18) adjustment for c-Kit exon 11 MS (Table 1). However, adjusting for c-Kit exon 11 MS improved AUC for all PM. Conclusions: Adjusting PM for c-KIT exon 11 status improved the predictive performance of all PM, although the results were not significant. This study found no significant difference in the performance of PM that predict GIST recurrence, which aligns with the one other study on this topic. Lack of significance may be due to a small sample size, which also limited the multivariate analysis. Including genotypes in PM that predict GIST recurrence should be further explored with larger multi-center datasets that can adjust for multiple mutations. Multivariate analysis comparing prognostic models for gastrointestinal stromal tumors at predicting recurrence with and without adjustment for c-KIT exon 11 genotypes using receiver operating characteristic. Without c-Kit exon 11 AUC (95% Confidence Interval) P value NIH 0.86 (0.81,0.90) 0.18 AFIP 0.87 (0.82,0.91) MSKCC 0.89 (0.84,0.94) With c-Kit exon 11 genotype NIH 0.80 (0.67,0.93) 0.16 AFIP 0.84 (0.73,0.95) MSKCC 0.90 (0.80,0.99) NIH: National Institute of Health, AFIP: Armed Forces Institute of Pathology, MSKCC: Memorial Sloan Kettering Cancer Centre. AUC: Area under the curve.

e23419 Background: Current risk assessment for cancer-associated thrombosis remain suboptimal in NSCLC, likely due to unaccounted tumor-specific biology. Although ALK and ROS1-positive NSCLC share closely related structure and elevated thrombotic risk, real-world data on thromboembolic outcomes remain limited. Methods: This multicenter retrospective study was conducted using TriNetX, a global federated health research network;14 of 113 healthcare organizations contributed to analysis. Adults (≥18 years) with NSCLC ROS1 or ALK from 2012-2025 were included. Index event was defined as the first documented NSCLC diagnosis. Outcomes included cumulative incidence, Kaplan–Meier survival analysis (KMS), and event recurrence for arterial thromboembolism (ATE), deep vein thrombosis (DVT), and pulmonary embolism (PE). Analyses were conducted pre and post-propensity score matching (PSM) adjusted for age and sex. Subgroup analyses evaluated outcomes at 1, 3, 6, and ≥12 months. Results: 3,310 NSCLC patients were identified (ALK n=3,155; ROS1 n=155). Mean age was 73 years, with 52.8% female and 46.2% male; 72.5% White, 12.7% Black, 9.5% Asian; South (49%), Northeast (29%). ALK NSCLC were older than those with ROS1 (mean 68.9 vs 64.6 years, p<0.0001), with similar sex distribution. ROS1 had higher proportion of White (85% vs 72%), while ALK included more Asian and Black. Median follow-up prior to matching was 482 days for ROS1 (post-PSM:482) and 546 days for ALK (post-PSM:601) Incidence proportions in ROS1 NSCLC were 12.3% for ATE, 9.7% for DVT, and 12.9% for PE (vs. ALK: 22.6% for ATE, 12.5% for DVT, 11.8% for PE) For ATE, prior to PSM, ROS1 NSCLC demonstrated a significantly lower risk compared with ALK (16.6% vs 25.0%), with an absolute risk difference of −8.4% (95% CI −14.7 to −2.2; p=0.02) and relative risk of 0.66 (95% CI 0.46–0.96). KMS showed longer ATE-free survival in ROS1 compared to ALK (48.2% vs 44.5%, p=0.01; HR 0.60, 95% CI 0.40–0.91, p=0.04). After PSM, overall ATE risk difference did not reach statistical significance, although ROS1 continued to demonstrate lower risk (16.6% vs 25.8%; risk difference −9.2%, p=0.06). Notably, in the subgroup analyses, this significant risk difference emerged ≥12 months, with improved late ATE-free survival in ROS1(91.1% vs 80.3%, p=0.01). For VTE and PE, no significant differences in risk, survival analyses or recurrence, were observed between ROS1 and ALK before or after PSM (DVT risk: 9.2% vs 9.7% pre-PSM; 9.2% vs 10.1% post-PSM; PE: 11.2% vs 9.9% pre-PSM; 11.2% vs 8.6% post-PSM; all p>0.05). Conclusions: Arterial thromboembolism differed by oncogenic subtype, with higher ATE risk in ALK compared with ROS1-positive NSCLC, particularly ≥12-months. In contrast, no significant difference in venous thromboembolism and pulmonary embolism were observed; further validation in prospective studies is warranted.

e23276 Background: More than 50,000 patients with cancers of all stages were enrolled in the E2C2 trial, which captured patient-reported data across 6 symptom domains (SPPADE): sleep, pain, impaired physical function, anxiety, depression, and energy deficit or fatigue. This trial offers a unique opportunity to evaluate the evolution of symptom burden as prostate cancer (PCa) advances and the association of symptom burden with prognosis in the various disease states. Methods: The cohort was derived by cross-referencing participants in the E2C2 study with the Mayo Clinic Cancer Registry for more granular data on disease state and treatment characteristics, focusing on PCa. Symptoms were patient-reported on a 0-10 numerical rating scale, categorized as none/mild (0-3), moderate (4-6), and severe (7-10). We compared the prevalence of symptom burden across disease states, using the first survey completed by patients, in 3 groups: (1) non-metastatic (localized, locally advanced, and biochemically recurrent), (2) metastatic hormone sensitive (mHSPC), and (3) castration resistant (CRPC), using chi square tests. We then used log-rank tests to compare survival distributions based on symptom severity within each disease state. Results: A total of 1,011 patients were included. The distribution of moderate to severe symptoms by disease state is shown in Table 1. Fatigue, impaired physical function, and pain were more prevalent in patients with advanced disease, whereas the prevalence of depression and anxiety was not statistically different by disease state. Greater fatigue and physical impairment were consistently associated with poorer OS across all disease states, including patients with non-metastatic disease (p < 0.001). Pain was most strongly associated with inferior outcomes in the more advanced disease states of mHSPC (p < 0.001) and CRPC (p < 0.001). Anxiety (p < 0.001) and depression (p = 0.001) were only prognostic in CRPC patients. Conclusions: SPPADE symptoms are common in PCa. Physical symptoms are greater in more advanced disease, yet the high burden of unmet psychologic needs is similar across the disease states. Notably, severe patient-reported fatigue and physical dysfunction are strongly prognostic even in early disease, suggesting they are markers of physiologic vulnerability. Structured symptom monitoring may improve the identification of high-risk patients. Clinical trial information: NCT03892967 . Moderate to severe symptoms by disease state. Non-Metastatic (N= 243) mHSPC (N= 479) CRPC (N= 289) P-value Pain, n (%) 51 (21.3) 121 (25.5) 100 (35.1) 0.001 Fatigue, n (%) 92 (38.3) 230 (48.6) 163 (57.2) <0.001 Impaired Physical Function, n (%) 79 (32.5) 180 (37.7) 146 (50.7) <0.001 Sleep Disturbance, n (%) 72 (30.1) 167 (35.2) 104 (36.1) 0.3 Depression, n (%) 47 (19.6) 92 (19.4) 63 (22.0) 0.67 Anxiety, n (%) 56 (23.4) 93 (19.6) 68 (23.9) 0.3

e23274 Background: Treatment-related symptoms substantially impair quality of life (QoL) and drive unplanned acute care use in oncology. Although patient-reported digital symptom monitoring is widely studied, outcomes remain inconsistent. We examined how intervention design influences clinical impact. Methods: We conducted a scoping systematic review of prospective randomized trials and pragmatic real-world studies evaluating patient-reported digital symptom monitoring or management interventions in adult oncology populations. Eligible studies reported symptom burden, QoL and/or acute care utilization. Data were abstracted on cancer type, digital modality, triage approach, and direction of outcomes. Protocols and qualitative studies were reviewed for context but excluded from outcome synthesis. Results: Across studies involving solid tumors and hematologic malignancies, interventions used web-based portals, mobile applications or interactive voice response systems. A consistent pattern emerged across the evidence base: interventions that paired real-time symptom reporting with structured, rule-based triage and proactive clinician response demonstrated improvements in symptom burden and QoL across multiple settings. In contrast, passive monitoring or self-management alone showed limited or no benefit despite similar reporting intensity. Effects on acute care utilization were heterogeneous, with reductions most often seen in pragmatic, workflow-integrated implementations. Conclusions: The clinical value of patient-reported digital symptom monitoring lies not in symptom collection itself, but in how reported data are reviewed and acted upon by clinicians. Monitoring alone is insufficient. Implementation-ready models that embed actionable triage and clinician response workflows are essential to improve patient outcomes and reduce avoidable utilization. Intervention approaches and clinical outcomes of digital symptom monitoring in oncology. Intervention Approach Digital Modality Actionable Triage / Clinician Response Symptom Burden Quality of Life Acute Care Use Monitoring with proactive clinician response Portal / App / IVR Yes (rule-based, clinician-reviewed) Improved Improved Reduced / Mixed Monitoring with automated alerts only Portal / App Partial Mixed Mixed NR Passive monitoring or self-management App / Portal No No benefit No benefit NR Pragmatic, workflow-integrated models Mixed Yes + staffing workflows Improved Improved Reduced Heterogeneous or non-integrated implementations Mixed Variable Mixed Mixed Mixed Abbreviations: IVR = Interactive Voice Response; QoL = Quality of Life; NR = Not Reported.

e23223 Background: Early-stage lung cancer is potentially curable with surgery, yet emergency or nonelective presentation represents a care-delivery phenotype reflecting failures in early diagnosis or access to care rather than tumor biology alone. The national impact of emergency presentation among surgically treated, presumed early-stage lung cancer patients is poorly defined. This study evaluates emergency versus elective presentation as a care-delivery phenotype and its association with perioperative morbidity, failure-to-rescue, and inpatient outcomes. Methods: A survey-weighted analysis of the National Inpatient Sample (2016–2023) was conducted. Adult lung cancer hospitalizations were identified using ICD-10-CM C34* in any diagnosis position. A presumed early-stage surgical cohort was defined by curative-intent lung resection without metastatic disease, malignant pleural effusion, or palliative care coding. The primary exposure was nonelective versus elective admission. Outcomes included in-hospital mortality; ICU-level care, defined by invasive mechanical ventilation, shock, or respiratory failure; major complications; failure-to-rescue; length of stay; and hospitalization cost and charges. Multivariable survey-weighted models adjusted for patient demographics, payer, neighborhood income, hospital characteristics, and year. Results: In the weighted cohort of 422,925 hospitalizations, nonelective admission occurred in 27.3%. After multivariable adjustment, nonelective admission was independently associated with higher odds of in-hospital mortality (adjusted odds ratio [aOR] 2.59, 95% CI 2.22–3.03), ICU-level care (aOR 4.53, 95% CI 4.32–4.76), and major complications (aOR 5.00, 95% CI 4.78–5.23), as well as longer length of stay (+2.42 days). Notably, failure-to-rescue among patients with major complications did not differ by admission type (aOR 0.90, 95% CI 0.77–1.05), indicating that excess mortality associated with nonelective presentation is driven by presentation severity and care pathway rather than differences in rescue quality. In unadjusted analyses, nonelective presentation was associated with higher mortality (2.21% vs 0.68%), greater ICU-level care use (33.8% vs 8.9%), more frequent major complications (45.0% vs 12.4%), and longer length of stay (7.84 vs 4.99 days) compared with elective admissions. Conclusions: In a procedure-defined, presumed early-stage lung cancer surgical cohort, nonelective admission identifies a common care-delivery failure phenotype associated with higher perioperative morbidity, ICU-level care, in-hospital mortality, and longer hospitalization. Emergency presentation represents a systems-level vulnerability in curative-intent lung cancer care and serves as a scalable and actionable marker for quality improvement and disparity-focused analyses.