INTRODUCTION:Treatment according to guidelines for second-line and beyond (2L+) therapy of diffuse large B-cell lymphoma (DLBCL) differentiate between transplant-eligible and transplant-ineligible patients. Transplant-eligible patients typically receive salvage immunochemotherapy followed by high-dose chemotherapy and autologous stem cell transplantation (HD-ASCT) or CAR-T cell therapy. Although literature suggests a 50-50 distribution between transplant-eligible and transplant-ineligible patients, real-world evidence from Germany is lacking. This study aimed to estimate the proportion of transplant-eligible and transplant-ineligible patients using intended for HD-ASCT national inpatient billing data to evaluate the validity of the commonly assumed distribution. METHODS:Inpatient billing data from all German hospitals in 2024, provided by the Institute for the Hospital Remuneration System (InEK), were analyzed. Cases with the main diagnosis "C83.3" (DLBCL) defined the base population. Applying guideline-based first-line cure rates (60% [minimum]-70% [maximum]), the 2L population size was estimated. Transplant-eligible cases were identified via OPS codes for salvage immunochemotherapies (R-ICE, R-DHAP, R-GDP). The transplant-eligible proportion was calculated by dividing the number of transplant-eligible cases by the estimated 2L population. Sensitivity analyses were performed using data from 2022 and 2023, validating the 2L relapse/refractoriness rate assumption against German real-world claims data, and assessing potential impact of polatuzumab vedotin-related coding bias. RESULTS:The base population included 32,851 DLBCL cases. Based on cure rate assumptions, the estimated 2L population size ranged from 9,855 (minimum) to 13,140 (maximum). A total of 7,523 cases were identified to have received salvage immunochemotherapy based on coding criteria. This corresponds to a transplant-eligible proportion of patients intended for HD-ASCT of 57% (minimum) and 76% (maximum). Sensitivity analyses yielded a transplant-eligible proportion range of 40% to 54% (2022) and 52% to 69% (2023), 52% when applying the relapse/refractoriness rate from German real-world claims data, and 34% to 46% when excluding polatuzumab vedotin-related cases. CONCLUSION:This analysis introduces a method for approximating estimating the distribution of transplant eligibility among DLBCL patients intended for HD-ASCT in the 2L setting using real-world inpatient billing data. The results question the widely assumed balance between transplant-eligible and -ineligible groups, indicating a possible tendency toward a slightly higher proportion of transplant-eligible cases which may be clinically eligible for HD-ASCT.
Introduction Fear of Cancer Recurrence (FCR) is a prevalent and distressing issue for a significant proportion of cancer patients across various disciplines, yet it remains one of the most frequently unaddressed psychosocial needs. A key barrier to identifying affected individuals is the lack of brief screening tools suitable for clinical settings. The objective of this research is to validate the one-item FCR screening tool (FCR-1r) in a German cohort of urologic cancer patients. Validating such an ultra-brief measure may help address this clinically relevant gap and serve as a viable, clinically applicable alternative to lengthy questionnaires. Methods Statistical analyses were conducted using R software. Criterion validity was assessed against the CARQ-4. Convergent validity was tested against the FoP-Q-SF and GAD-7. Discriminant validity was tested against the HLS-EU-Q16. Validity was assessed using chi-square tests with phi coefficients, accuracy measures, negative and positive predictive values, sensitivity, specificity and odds ratios. Results The final analysis sample consisted of 236 participants with complete data on all study measures. The FCR-1r demonstrated excellent criterion validity with respect to the CARQ-4 (χ²(1) = 101.97, p < 0.001, φ = 0.66), with an overall accuracy of 84% (95% CI: 79-89%). Sensitivity was 84%, and specificity was 85%. The analysis of convergent validity revealed a moderate association with both the FoP-Q-SF (χ²(1) = 20.25, p < 0.001, φ = 0.29; accuracy 62% (95% CI: 56-68%)) and the Generalised Anxiety Disorder-7 (GAD-7) (χ²(1) = 11.82, p < 0.001, φ = 0.22; accuracy 64% (95% CI: 58-70%)). The analysis of discriminant validity showed no significant association between the FCR-1r and health literacy (HLS-EU-Q16): χ²(1) = 3.44, p = 0.064, φ = 0.12. Conclusion The German version of the FCR-1r demonstrates good psychometric properties, confirming its validity as a tool for assessing fear of cancer recurrence in clinical practice.
The expanding availability of multi-omics profiling and advances in artificial intelli-gence (AI) and machine learning (ML) are changing precision oncology. Molecular testing strategies such as longitudinal liquid biopsy assessment, germline variant analysis, and the evaluation of tumor-infiltrating clonal hematopoiesis (TI-CH) are gain-ing attention, given their potential to inform treatment decisions from surveillance through therapy selection. In 2025, during a NORD (Northern Oncology, Research and Development) workshop on "Precision Oncology", stakeholders from three Northern German university hospitals exchanged insights on the impact of tumor heterogeneity and the tumor microenviron-ment, approaches for multi-layer data integration, and strategies to accelerate transla-tion using innovative clinical trial designs. Building on this discussion, participants evaluated how the integration of multi-omics data, AI and ML into molecular tumor board (MTB) workflows could enhance decision-making, as well as the challenges as-sociated with implementing these tools in routine clinical care. The workshop thereby not only identified key opportunities and obstacles but also formulated recommenda-tions on how these tools can be stepwise integrated into existing workflows to generate the evidence required for their implementation in clinical practice. This review connects expert perspectives with targeted literature to outline how these emerging methods advance precision oncology. It further identifies the technical, struc-tural, and clinical requirements for their implementation in clinical use, and highlights how they can improve the quality, speed and accessibility of MTB recommendations.
e15577 Background: Somatic copy number alterations (CNAs) drive chromosomal instability in colorectal cancer (CRC). We aimed to determine the prognostic value of CNA burden and elucidate its underlying molecular adaptations, focusing on the discovered RNA-protein decoupling of metallothionein (MT1E). Methods: Genome-wide CNA burden was analyzed in 78 CRC patients (Taipei Veterans General Hospital) via targeted NGS and validated in 437 patients (TCGA-COAD). Multi-omic integration screened for critical dysregulated targets in high-CNA tumors. Findings were validated by examining MT1E protein expression using immunohistochemistry and digital pathology in an independent cohort (n = 189). Results: High CNA burden was mutually exclusive with MSI-high status and consistently predicted inferior overall survival in both cohorts (TVGH: HR = 1.80, p = 0.019; TCGA: HR = 1.71, p = 0.009). Notably, integrative analysis revealed a striking discordance regarding MT1E: while high-CNA tumors exhibited significant downregulation of MT1E mRNA (logFC = −1.13; FDR = 4.23×10⁻¹²), protein-level analysis demonstrated that high MT1E protein expression was paradoxically associated with aggressive disease and worse survival (p = 0.028). This suggests that post-transcriptional stabilization of MT1E acts as a critical stress-adaptive mechanism specific to genomically unstable tumors. Conclusions: High CNA burden serves as a robust prognostic biomarker in CRC, identifying a high-risk subgroup characterized by stress-adaptive MT1E protein stabilization despite transcriptional repression. These findings expose a potential proteostatic vulnerability in chromosomal-unstable CRC that could be exploited therapeutically.
e13138 Background: HER2-low/ultralow metastatic breast cancer (mBC) is now clinically actionable with HER2-directed antibody–drug conjugates, but expanding HER2 classification beyond a binary model has increased diagnostic complexity and variability in practice. A prior quality improvement (QI) initiative identified diagnostic and therapeutic gaps among community oncologists, including ambiguous results, inconsistent HER2 testing and reporting, limited multidisciplinary communication, and uncertainty in applying HER2 results to treatment decisions. Building on these findings, we surveyed pathologists nationwide to characterize real-world HER2 testing and reporting practices and identify diagnostic and care coordination barriers between specialties. Methods: From May–June 2025, 100 pathologists completed surveys assessing HER2 testing workflows, interpretation, and classification challenges, including barriers to coordinating with oncologists, with key items tethered to the prior oncology survey. Results: The majority of respondents reported practicing in community-based hospitals and clinics (58%). Sixty-three percent of pathologists identified risk of HER2 misclassification as the greatest challenge in identifying HER2-low/ultralow mBC, compared with 21% of oncologists surveyed in the prior QI initiative, whose top challenge was interpreting ambiguous or borderline results. Similar proportions of pathologists and oncologists reported difficulty interpreting ambiguous or borderline results (53% vs 51%) and limited familiarity with the clinical relevance of lower HER2 expression (28% vs 24%), while inconsistencies in testing and reporting were cited by 55% and 24%, respectively. Pre-analytic and analytic barriers included tumor heterogeneity (52%), suboptimal sample conditions (43%), and inadequate clinical information (39%). Evolving HER2 classification was the most common reporting challenge (76%), and only 47% of pathologists felt comfortable distinguishing HER2 IHC 0 from IHC 1+. Fewer than half of pathologists and oncologists reported that HER2-low status was routinely specified in pathology reports 41% and 37%, respectively. Collaboration with oncologists is hindered by understanding of HER2 testing guidelines (58%), pressure to alter or expedite reporting (28%), and communication through intermediaries (26%). To improve identification, nearly two-thirds of pathologists favored workflows to better differentiate IHC 0 versus 1+, and 58% supported standardizing definitions and documentation. Conclusions: Persistent diagnostic variability and inconsistent reporting limit accurate identification of HER2-low and HER2-ultralow mBC, underscoring the need for standardized workflows, clearer reporting, and improved multidisciplinary coordination between pathologists and oncologists.
e13086 Background: Adverse inpatient outcomes with COVID-19 have been well characterized for aggregations of all cancer types. However, the unique pathophysiology and treatment of breast cancer warrant specific analysis. Therefore, a critical gap exists with regard to breast cancer hospitalizations specifically to better inform clinical decision making and resource planning. We aimed to characterize inpatient mortality, morbidity, and resource utilization among breast cancer patients with concurrent COVID-19 infections to address this gap. Methods: We identified hospitalization cases from the National Inpatient Sample (NIS), Healthcare Cost and Utilization Project (HCUP), Agency for Healthcare Research and Quality from 2020-2022 that met the inclusion criteria of female adults ≥ 18 years old with hospitalizations containing a primary or secondary ICD-10 code for diagnosis of breast cancer. Non-normally distributed results were compared using Mann-Whitney U test, and categorical variables were compared using Pearson’s chi-square test. Multivariate logistic regression models compared inpatient mortality and COVID-19 and survey-weighted negative binomial regression estimated the association between COVID-19 and inpatient length of stay. All analyses were performed using Stata version 19 (StataCorp LP, College Station, TX). Results: We identified 97,288 hospitalizations meeting inclusion criteria of which 4,933 (5.1%) cases were diagnosed with COVID-19. COVID-19 was independently associated with statistically significantly higher odds of inpatient mortality (11.1% vs 4.9%), septic shock (5.3% vs 2.9%), acute respiratory failure (38.6% vs 10.9%), mechanical ventilation (11.7% vs 5.1%), and longer length of hospital stay (3.0-10.0 days vs 2.0-6.0) (p < 0.001). Increased mortality was additionally associated with age 80 years and older (AOR 2.915, 95% CI 2.209–3.845, p < 0.001), non-Hispanic Black patients (AOR 1.214, 95% CI 1.117-1.320, p < 0.001), metastatic cancer (AOR 3.230, 95% CI 3.032-3.440, p < 0.001), chronic renal failure (AOR 1.309, 95% CI 1.170-1.464, p < 0.001), congestive heart failure (AOR 1.106, 95% CI 1.001-1.223, p = 0.048), and chronic pulmonary disease (AOR 1.169, 95% CI 1.068-1.280, p = 0.001). Conclusions: Breast cancer patients hospitalized with COVID-19 experienced statistically significant higher inpatient mortality, greater morbidity, and longer hospital stays compared with those without COVID-19 even after adjustment for comorbidities and hospital factors, highlighting the increased vulnerability of this population during infectious disease surges. Targeted clinical monitoring, timely prevention strategies, and early intervention may help mitigate these risks, underscoring the need for continued attention to breast cancer–specific outcomes during and beyond public health emergencies.
e23248 Background: Sinonasal malignancies are rare, heterogeneous tumors characterized by nonspecific early symptoms and frequent diagnostic delay, leading to advanced disease at presentation. While outpatient management and multimodal therapy have been extensively reviewed, national inpatient presentation patterns and rescue outcomes remain poorly defined. Methods: A survey-weighted analysis of the National Inpatient Sample from 2016–2023 was performed, identifying adult hospitalizations with sinonasal malignancies (C30–C31) in any diagnosis position and excluding admissions with palliative care coding. A diagnostic delay phenotype was defined by nonelective admission accompanied by acute organ failure at presentation, including sepsis, shock, respiratory failure, or acute kidney injury. Outcomes included in-hospital mortality, ICU escalation, failure-to-rescue (death following a major complication), length of stay, hospitalization cost, and discharge disposition. Outcomes were contextualized through comparisons with other rare cancers and common malignancies. Results: Sinonasal malignancies accounted for 0.66% (95% CI, 0.64%–0.69%) of cancer hospitalizations and were predominantly nonelective at 57.16% (95% CI, 55.63%–58.69%). Acute organ failure at presentation occurred in 21.53% (95% CI, 20.34%–22.72%) of sinonasal admissions, with sepsis in 7.70% (95% CI, 6.93%–8.46%) and respiratory failure in 8.77% (95% CI, 7.98%–9.56%). Overall in-hospital mortality was 1.64% (95% CI, 1.33%–2.02%). Among admissions with major complications, failure-to-rescue mortality was 4.23% (95% CI, 3.39%–5.27%). ICU escalation occurred in 4.93% (95% CI, 4.33%–5.52%), with post-ICU mortality of 27.74% (95% CI, 26.25%–29.27%). Compared with common cancers, sinonasal malignancies demonstrated higher rates of nonelective admission and acute organ failure but similar overall mortality, indicating preserved rescue despite high-acuity presentation. Conclusions: Sinonasal malignancies exhibit a distinct inpatient diagnostic delay phenotype marked by emergency presentation and acute organ failure at admission. Despite high physiologic acuity, inpatient mortality and failure-to-rescue are not disproportionately worse, suggesting that delayed recognition primarily shifts how patients present rather than whether they survive hospitalization. These findings complement existing clinical literature by identifying inpatient presentation and rescue as critical, understudied dimensions of sinonasal cancer care.
6016 Background: Standard chemoradiation (CRT) for unfavorable risk p16+ head and neck squamous cell carcinoma (HNSCC) is associated with high rates of toxicity and treatment failure. We hypothesized that concurrent and adjuvant pembrolizumab (pembro) could achieve superior PFS with acceptable toxicity compared to CRT in this population. Methods: KEYCHAIN was an open-label randomized phase II trial conducted at 6 US academic centers. Eligible patients had newly diagnosed unresected p16+ HNSCC with unfavorable risk (AJCC 8 th edition stage II-III oropharynx or stage III-IVB non-oropharynx HNSCC). Patients were randomized 1:1 to RT (70 Gy in 35 fractions) plus pembro (200 mg, 2 weeks prior to RT then every 3 weeks starting day 1 of RT up to 20 total cycles) or RT plus concurrent cisplatin (cis) (100 mg/m 2 every 3 weeks). Age, ECOG status, and oropharynx site were stratification factors. The primary endpoint was PFS, defined as time from randomization to first progression or death from any cause. Primary analysis was in the modified intent-to-treat population, including all patients having ≥ 1 dose of study medication and ≥ 1 efficacy evaluation after baseline. The study design had 80% power with 1-sided α = 0.15 (log-rank test) and planned sample size of 50 analyzable subjects per arm. The study is registered with clinicaltrials.gov (NCT03383094). Results: Between Feb 2019 and Aug 2025, 108 patients were randomized (53 pembro, 55 cis), with 102 analyzable (50 pembro, 52 cis). Median follow-up was 26.5 months. 79% had minimum follow-up of 2 years. At 2 years, there were 7 PFS events and 1 death in the pembro arm, and 14 PFS events and 7 deaths in the cis arm. Two-year PFS was 84% (95% CI: 74-96%) vs. 70% (95% CI: 58-85%) in the pembro vs cis arm (HR 0.57, 95% CI: 0.25-1.31; 1-sided p = 0.09) (see Table). Two-year OS was 98% (95% CI: 94-100%) vs. 85% (95% CI: 76-96%) for pembro vs. cis (HR 0.33, 95% CI: 0.09-1.28; 1-sided p = 0.048). HRs adjusted for T3-4 and N2-3 category were 0.68 (PFS) and 0.31 (OS). As of the latest follow-up, there were 24 total PFS events and 12 deaths. Grade ≥ 3 adverse events definitely or probably related to treatment were 36% vs. 46% in the pembro vs. cis arm. The median number of cycles received in the pembro arm was 18. In the cis arm, 94% completed ≥ 2 cycles. Results by CPS status will be reported when available. Conclusions: Pembro plus RT met the pre-specified significance criterion for improvement in PFS compared to cisplatin plus RT in this phase II trial. A phase III trial of RT with pembrolizumab for unfavorable risk p16+ HNSCC is warranted. Clinical trial information: NCT03383094 . Pembro Cisplatin N 50 52 Mean age 63.5 62.5 Male 87% 89% White 85% 89% Age > 65 42% 35% ECOG 0 83% 70% Oropharynx 94% 96% T1 8% 10% T2 22% 12% T3 38% 38% T4 32% 40% N0 6% 8% N1 42% 29% N2 46% 54% N3 6% 10% 2-yr PFS 84% 70% 2-yr OS 98% 85% Completed RT w/in 56 days 96% 92% Grade ≥ 3 mucositis 18% 17% Grade ≥ 2 mucositis 72% 63% Grade ≥ 3 dysphagia 14% 6% Grade ≥ 2 dysphagia 48% 35%
7053 Background: Second primary malignancies (SPMs) are the leading cause of long-term treatment-related mortality in Hodgkin Lymphoma (HL). Although substantial improvements have been made to treatments over recent decades, few studies have demonstrated reduced SPM risk, particularly following modern treatments, and how risks should inform screening. Methods: We assembled a national cohort of 7,428 women treated for HL aged <36 across England & Wales 1954-2010, with 99% complete follow-up until 2018. We analyzed SPM incidence from national cancer registry linkage. Treatment data were collated from >250 treatment centers. Standardized incidence ratios (SIRs) and Absolute Excess Risks (AERs) for SPMs were calculated, and multivariable analyses (Hazard Ratios, HR) were undertaken to assess the impact of changes in treatment and assess changing incidence trends over time. Results: Twelve hundred women (16%) developed 1,467 SPMs with mean follow-up of 22 years (range 0-62 years). Overall SIR to develop any SPM was 3.3 (95% CI 3.1-3.5), with breast cancer contributing the greatest excess risk (AER 30.8 95% CI 27.7-34.1). Radiotherapy use halved from 1954-1980 to 2000-2010 (98% to 47%) and mean dose dropped from 48Gy to 33Gy. Conversely chemotherapy use doubled from 55% to 97%, with anthracyclines used in 94% and classic alkylators in 31% of treatments in the most recent period compared with 6% and 48% respectively pre-1980. Radiotherapy conferred the greatest treatment-specific risk factor for SPMs, (SIR 3.5 95% 3.3-3.7), with a strong dose-response relationship trend (HR 1.01/Gy p<0.001), and highest relative risks seen in those treated with radiotherapy aged <15years (SIR 7.4 (95% CI 5.7-9.1). There was a 25% reduction in SPM risk from the earliest treatment period (<1990) to most recent (2000-2010) and 34% reduction in solid SPM risk (Table). The decrease in risk for SPMs became smaller and non-significant after adjusting for reduced radiotherapy use and dose (HR 0.89, p trend 0.11). There was no attenuation after adjustment for chemotherapy. Despite declining risks, risks remained significantly elevated beyond 40 years after treatment. Conclusions: Modern HL treatments are associated with a substantial reduction in SPM risks, which appears to be largely attributable to decreased radiotherapy use and dose. However large persistent excess risks, particularly for breast and lung cancers, underscore the need for targeted screening in high-risk survivors treated in more recent eras. Risk of SPMs by treatment era. SPMs & treatments Treated <1990HR (ref) Treated 1990-1999HR (95% CI) Treated 2000-2010HR (95% CI) p trend All SPMs Any treatment 1.0 0.77 (0.65-0.90 0.75 (0.57-0.99) 0.001 Adjusted for radiotherapy & dose 1.0 0.83 (0.69-1.00) 0.90 (0.63-1.28) 0.11 Solid SPMs Any treatment 1.0 0.71 (0.60-0.85) 0.66 (0.48-0.90) <0.001 Adjusted for radiotherapy & dose 1.0 0.79 (0.65-0.96) 0.81 (0.55-1.20) 0.03
7094 Background: Waldenström Macroglobulinemia (WM) remains incurable with current available therapeutic options. Treatment options for relapsed/refractory pts are limited and often continuous, underscoring the need for new agents with novel mechanisms of action and pt-centered, fixed duration therapy. Iopofosine I 131 is a novel radiopharmaceutical compound composed of a lipid raft-targeting phospholipid ether covalently bound to 131 I, a beta emitting radioisotope. Here, we report preliminary sub-group results of the CLOVER-WaM study [NCT02952508] in WM pts treated with a Bruton’s Tyrosine Kinase inhibitor (BTKi) immediately prior to iopofosine. Methods: Eligible WM pts had received >2 prior therapies and had symptomatic disease requiring therapy according to consensus recommendations for WM. Treatment consisted of 2 cycles of iopofosine I 131 administered at 15 mCi/m 2 on days 1 and 15 of each 57-day cycle. The primary efficacy endpoint was major response rate (MRR; partial response or better, as determined by the modified IWWM-VI). Secondary and exploratory endpoints included overall response rate (ORR), treatment-free survival (TFS), progression free survival (PFS), duration of response (DOR), and clinical benefit rate (CBR; stable disease or better). Herein, efficacy endpoints were assessed in the modified intent-to-treat (mITT) population (all pts who received ≥60 mCi total administered dose of iopofosine I 131), and whose treatment regimen prior to iopofosine included a BTKi. Safety was evaluated using CTCAE v4.03 in pts who received ≥1 dose of iopofosine. Data cut off for this analysis was Jan. 16, 2026. Results: As previously reported, 65 pts were dosed and 55 are included in the mITT population. Of those, 22 (40%) had received a BTKi containing regimen immediately prior to iopofosine; and 11 who previously received at least 2 regimens utilizing BTKis and 12 who were refractory to rituximab. In this subgroup, median age was 70.5 (range 55-88), 17 (77%) were male, and 5 (23%) were female. The median number of prior therapies was 3 (range 2-7), 15/22 (68%) were MYD88 mutated and 18/22 (82%) were TP53 wildtype. The MRR was 15/22 (68.2%), ORR was 19/22 (86.4%) and CBR was 100%, median PFS was 12 months (range 1.6 to 28.0) and median TFS was 19.1 months (range 2.5-34.4). Grade ≥ 3, related treatment emergent adverse events (TEAE) occurring in > 10% of pts included anemia (41%), leukopenia (41%), neutropenia (58%), and thrombocytopenia (86%). Conclusions: Iopofosine I 131 demonstrated significant efficacy; 100% CBR, 86.4% ORR and 68.2% MRR with a median PFS of 12 months (range: 1.6 – 28.0 months) in a pt population with no current standard of care therapy. The treatment was well-tolerated with a manageable toxicity profile with cytopenias as the only Grade 3 or greater TEAE. Based on the results of this pivotal trial, iopofosine I 131 could be a promising novel treatment for WM pts after BTKi therapy. Clinical trial information: NCT02952508 .
e23432 Background: Published time-to-event results are most often presented as Kaplan-Meier (KM) plots, but the underlying individual participant data (IPD) are rarely accessible, limiting secondary analyses and evidence synthesis. Methods for reconstructing IPD from KM curves exist but remain technically demanding, non-reproducible, or slow to scale. We developed a provenance-preserving computational platform that enables accurate and reproducible IPD reconstruction directly from published survival figures. Methods: We developed and implemented an open-source, modular, semi-supervised architecture linking each study, figure, and curve to its calibration parameters and metadata. Study PDFs from published randomized control trials (RCTs) are uploaded and metadata including DOI and PMID are extracted automatically. Plots containing KM curves are identified and cropped by the user to establish boundaries based on Cartesian coordinates. KM curves are then masked to isolate each individual treatment curve, at which point the digitized KM plots are geometrically normalised, parsed for number-at-risk tables, and reconstructed using the validated Wei-Royston algorithm, which produces the patient-level data that would have been in the original arm in the RCT. This process was done systematically across an expert-curated selection of clinical trials relevant in urological oncology. Accuracy was assessed by comparing hazard ratios (HRs) and median survival values against those reported in 30 published clinical trials with 73 total curves spanning oncology and non-oncology indications. Comparative evaluation was performed against the fully automated KM-GPT pipeline published elsewhere for three additional studies. Results: Our approach reproduced published HRs with near-perfect concordance (slope = 1·0059, R² = 0·994). The mean difference between reconstructed and reported HRs was -0·0058 (SD 0·0176). Reconstructed survival curves visually overlapped their published counterparts, and absolute error distributions were normally distributed, indicating random rather than systematic variance. Compared with KM-GPT, our approach achieved equivalent or superior accuracy while maintaining full provenance control such that users can run simulations and counterfactual analyses to analyze different scenarios, compare arms from different trials, or conduct population-level analysis to measure benefit at the population level. Conclusions: Our approach demonstrates that reliable, low-latency reconstruction of time-to-event data can be achieved at scale with transparent, auditable provenance. This platform provides a foundation for clinician-guided, reproducible evidence synthesis from the published biomedical literature.
e21518 Background: Malignant melanoma and cerebrovascular diseases (CVDs) pose major U.S. public health burdens, with melanoma mortality of 2.0–2.7 per 100,000 and > 160,000 annual cerebrovascular deaths. Linked through hypercoagulability, vascular disruption, and frequent brain metastases in advanced melanoma, their combined mortality patterns remain poorly defined despite therapeutic advances. This study examines national trends from 1999 to 2023 to identify vulnerable populations that require targeted interventions. Methods: This population-based study analyzed U.S. mortality data from the Centers for Disease Control and Prevention Wide-Ranging Online Data for Epidemiologic Research (CDC WONDER). Age-adjusted mortality rates (AAMRs) were calculated for malignant melanoma (ICD-10 C43) associated with cerebrovascular diseases (ICD-10 I60–I69) among decedents aged ≥45 years from 1999 to 2023. Trends were stratified by age, sex, race/ethnicity, urbanization, and Census region using Joinpoint regression. Analyses were conducted between December 2025 and January 2026. Results: Nationally, melanoma-associated cerebrovascular mortality increased by 1.4% annually from 1999 to 2023 (P < .05). Rates rose more steeply among individuals aged ≥65 years (AAPC, 1.59%; post-2012 APC, 3.46%; P < .001) compared with those aged 45–64 years. Non-Hispanic White individuals experienced the largest increase (AAPC, 2.8%; post-2018 APC, 6.5%; P < .001). Mortality rates were higher among males than females, though trends were parallel. Nonmetropolitan areas (AAPC, 2.05%) and the South and Midwest showed faster increases than metropolitan, Northeastern, or Western regions. Conclusions: Despite overall declines in isolated melanoma and cerebrovascular mortality, combined melanoma-associated cerebrovascular deaths increased steadily over two decades. Older adults, non-Hispanic White individuals, men, rural residents, and populations in the South and Midwest bear a disproportionate burden. These findings highlight the need for integrated vascular-oncologic care, enhanced surveillance, and equitable access to advanced therapies. Age-adjusted mortality trends for melanoma-related cerebrovascular deaths in U.S. adults ≥45 years (1999–2023). S. No. Characteristic AAMR (1999-2023) AAPC (95%CI) P Value JP, No. 1. Overall 0.30-0.40 1.37 (0.92–1.81) <0.001 1 2. Female 0.21-0.26 1.13 (0.22–2.05) 0.023 1 3. Male 0.50-0.60 1.38 (0.88–1.88) 0.000 1 4. Hispanic 0.31-0.45 1.50 (1.01–1.99) 0.000 1 5. Non-Hispanic - White 0.37-0.54 1.59 (1.12–2.06) 0.000 1 6. 45-64 years 0.10-0.14 1.07 (0.04–2.10) 0.053 0 7. ≥65 years 0.66-0.87 1.52 (1.05–1.99) 0.000 1 8. Census Region- Midwest 0.21-0.42 1.97 (1.12–2.82) 0.000 1 9. Census Region- West 0.38-0.48 1.45 (0.72–2.19) 0.001 1
e17520 Background: Cervical cancer remains a public health concern in Europe despite effective prevention strategies, including HPV vaccination and organized screening. Although mortality has declined in many countries, progress has been uneven, with persistent regional disparities. We assessed long-term regional trends in cervical cancer incidence, mortality, and disability-adjusted life years (DALYs) across Europe. Methods: We conducted a GBD estimates-based analysis of cervical cancer incidence, mortality, and DALYs among females in Europe from 1990 to 2023 using publicly available data. Estimates were stratified by Western, Central, and Eastern Europe and expressed as age-standardized rates per 100,000 population. Temporal trends were evaluated using Joinpoint regression to estimate annual percent change (APC). Average annual percent change (AAPC) and 95% confidence intervals (CIs) were calculated as weighted averages across the study period. Results: From 1990 to 2023, cervical cancer incidence declined across Europe, with marked regional variation. Incidence decreased most in Western Europe (AAPC: −1.42%; 95% CI: −1.59 to −1.24) and Central Europe (AAPC: −1.27%; 95% CI: −1.35 to −1.19), driven by sustained declines after 2007 in Central Europe (APC: −2.41%; 95% CI: −2.48 to −2.35) and during 2001–2023 in Western Europe (APC: −1.89%; 95% CI: −2.03 to −1.75), following modest increases or stabilization during the 1990s. Eastern Europe showed an early rise (1990–1993; APC: 1.48%; 95% CI: −0.21 to 3.21) followed by prolonged declines after 2003, resulting in a small and non-significant overall change (AAPC: −0.09%; 95% CI: −0.42 to 0.24). In contrast, mortality and DALY rates declined across all subregions, with the largest reductions observed in Western Europe (mortality AAPC: −1.63%; 95% CI: −1.71 to −1.56; DALYs AAPC: −1.74%; 95% CI: −1.81 to −1.66), followed by Eastern Europe (mortality AAPC: −0.55%; 95% CI: −0.82 to −0.28; DALYs AAPC: −0.88%; 95% CI: −1.15 to −0.61) and Central Europe (mortality AAPC: −0.39%; 95% CI: −0.49 to −0.29; DALYs AAPC: −0.70%; 95% CI: −0.79 to −0.60). Trends were non-linear, with early increases in Eastern Europe (1990–1993; mortality APC: 1.65%; 95% CI: 1.05 to 2.25; DALYs APC: 4.60%; 95% CI: 4.07 to 5.14), followed by the steepest declines after 2011 and more gradual reductions in Central and Western Europe. Females experienced greater reductions in mortality, incidence, and DALYs, while trends for both sexes followed similar regional patterns. Conclusions: Despite substantial declines in cervical cancer mortality and DALY burden across Europe, incidence trends remain uneven, with the highest burden persisting in Central and Eastern Europe, underscoring the need for equitable strengthening of HPV vaccination and screening to accelerate cervical cancer elimination.
11075 Background: Effective patient education is critical in oncology but hindered by the complexity of treatment guidelines and the time constraints of clinicians. Traditional static materials often fail to address the personalized needs of diverse demographics, particularly the elderly. We developed an Intelligent Medical Science Popularization Platform integrating Multimodal Large Language Models (MLLM) with Retrieval-Augmented Generation (RAG) to automate the production of reliable, guideline-based oncology educational content. Methods: The platform utilizes a three-tier architecture deployed at a tertiary cancer center. The core engine combines a self-developed MLLM with RAG, anchoring content generation to authoritative sources (e.g., CSCO/NCCN guidelines and peer-reviewed journals) to ensure clinical accuracy. The system features: (1) Physician Portal: Automates the conversion of clinical protocols into patient-friendly text, layouts, and animations; (2) Multimodal Synthesis: Generates virtual avatar videos explaining diagnoses and treatments; (3) Patient Interface: Delivers personalized, accessible content via mobile apps, supporting voice interaction for elderly adherence. Results: Implementation data demonstrated significant efficiency and engagement gains. The MLLM+RAG model achieved > 98% accuracy in content generation as verified by expert oncologists. The production time for educational videos was reduced from 72 hours to ~10 minutes (> 99% reduction), enabling rapid updates aligned with new trial data. Animation production costs decreased by 10-fold. Daily content output exceeded 15 units. Notably, in a cohort of elderly cancer patients, the platform increased user retention rates by 40% and extended average session duration by 50 seconds, indicating improved engagement with health information. Conclusions: This policy-compliant AI platform validates the utility of RAG-based MLLMs in oncology. By shifting from passive information supply to active, personalized intelligent services, it significantly reduces the workload for oncologists while bridging the digital divide for vulnerable patient populations. This model offers a scalable solution for enhancing health literacy and treatment adherence in cancer care.
e24161 Background: Chemotherapy-induced thrombocytopenia (CIT) is a frequent complication, resulting in increased bleeding risk, chemotherapy dose reductions, delays or discontinuation. Thrombopoietin receptor agonists (TPO-RA), such as romiplostim (romi), represent a promising therapy for CIT by stimulating platelet production. Methods: Despite National Comprehensive Cancer Network guidance supporting romi or other TPO-RA use in CIT, real-world adoption remains inconsistent across all malignancies. We conducted a retrospective, single-center analysis of romi use for CIT from January 2020 to December 2024. Patients were included if they received romi within 6 months following chemotherapy. This study evaluated achievement of platelet count (×10 9 /L) ≥100 and > 50, time to platelet recovery, dosing patterns, thromboembolic and bleeding events, and chemotherapy dose delay/reductions. Results: Of 121 patients included (median age 58; 69% female), 56% had solid tumors and 44% had hematologic malignancies. The most common chemotherapy regimens prior to the development of thrombocytopenia included carboplatin-based regimens, azacitidine/venetoclax, and FOLFOX-based regimens. Prior to romi initiation, median platelet count was 26. About 12% of patients had a platelet count ≥ 100 prior to romi initiation. After romi initiation, 74% of patients achieved platelets ≥ 100 with a median time of 16 days (patients with a platelet count > 100 were not included in this data point). The median romi dose was 5 mcg/kg during treatment. Dose delays or reductions were avoided in 70% of patients. During romi therapy, bleeding and thrombotic events occurred in 10% and 0.8% respectively. Conclusions: In our study, romi was effective at achieving a platelet goal of > 100 in the majority of our patients. Limitations include a single center study, heterogenous patient population and a lack of comparison to patients who did not receive romi despite similar degrees of thrombocytopenia. Further studies are needed to determine if romi affects overall survival, and to develop an effective dosing strategy for ROMI in CIT as a cost analysis. Outcomes Solid Hematologic Achieved platelets > 100 x 10 9 (n, %) 43, 63.2% 33, 62.3% Achieved platelets > 50 x 10 9 (n, %) 26, 38.2% 36, 68% Not requiring chemotherapy delay or dose reduction (n, %) 48, 70.6% 37, 69.8% Median romi dose during treatment (mcg/kg) 3.5 6.4 Side Effect/Safety Profile Bleeding events recorded during romi (n, %) 6, 8.8% 6, 11.3% Thrombotic events recorded during romi (n, %) 1, 1.5% 0, 0%
9521 Background: 2-Hydroxybutyric acid (α-HB) is an established biomarker in metabolic disease of insulin resistance and oxidative stress but its role in immuno-oncology remains unclear. This study conducted an exploratory analysis to evaluate whether α-HB could be used as a biomarker in advanced melanoma patients treated with ipilimumab-nivolumab, and its association with treatment response and toxicity. Methods: In this single-center retrospective study, 36 patients with unresectable/metastatic melanoma received ipilimumab-nivolumab. α-HB levels were measured by targeted mass spectrometry and measured relative metabolomic signal intensities. Descriptive statistics assessed α-HB across subgroups defined by age, sex and BRAF status. Associations between α-HB and grade ≥3 immune-related toxicity, treatment response, and survival at last contact were tested using univariable logistic regression with α-HB entered as a continuous predictor (log2-transformed). Prespecified sensitivity analyses dichotomized α-HB at the median (1.13) and 75th percentile (1.55). For mortality, we additionally derived an exploratory ROC-based Youden threshold to define a “high α-HB” group and report diagnostic performance (specificity, PPV, NPV) and odds ratios. Results: Median age was 53.5; 63.9% were male; 80.6% had cutaneous melanoma; 33.3% were BRAF-mutant; 80.6% received first-line therapy. Baseline α-HB was lower in patients who developed grade ≥3 toxicity (mean 1.01 vs 1.54; p=0.013) and in responders (1.09 vs 1.60; p=0.021), while it was higher among those who died (1.49 vs 1.06; p=0.045). In logistic models using log2-transformed α-HB, each doubling in α-HB was associated with reduced odds of grade ≥3 toxicity (OR 0.165, 95%CI 0.036–0.744; p=0.019) and response (OR 0.268; 95% CI 0.075–0.961; p=0.043; AUC = 0.79; HL p=0.803, where higher p indicates better fit), consistent with lower α-HB marking both toxicity and response. Higher α-HB trended toward increased death risk (OR 3.3 per doubling; p=0.066). Using a Youden-derived cutoff (α-HB ≥1.58), high α-HB defined a high-risk subgroup (OR 10.0; 95% CI 1.09–91.4; p=0.041) with high specificity for death (94%); PPV 89% and NPV 56%. Conclusions: Elevated α-HB was associated with reduced response, lower incidence of immune-related toxicity, and higher rates of death, suggesting a metabolically suppressed immune phenotype. Further validation is needed to determine whether metabolic interventions could improve immunotherapy outcomes in selected patients.
11102 Background: Cancer stage is often used as a proxy for inpatient risk, yet physiologic reserve may better determine outcomes. It was hypothesized that among gastrointestinal (GI) cancer hospitalizations without coded metastasis, malnutrition/cachexia identifies a vulnerable subgroup with complication burden, ICU escalation, and post-complication mortality approaching that of metastatic disease. Methods: A survey-weighted analysis of the National Inpatient Sample (NIS), 2016–2023, was conducted. Adult hospitalizations with any GI malignancy (ICD-10-CM C15–C20, C22–C25) were included; encounters with palliative care coding (Z51.5) were excluded. Metastatic disease was defined by C77–C79. Vulnerability was defined as malnutrition/cachexia (E43, E44, E46, R64). Major complications included sepsis, bleeding/transfusion, venous thromboembolism, acute kidney injury (AKI), bowel injury/peritonitis, and respiratory failure. ICU escalation proxies included mechanical ventilation and/or shock. Outcomes were compared across three groups: non-metastatic/no vulnerability, non-metastatic/vulnerable, and metastatic. Failure-to-rescue was defined among non-metastatic surgical hospitalizations as death following ≥1 major complication. Results: The cohort included 748,734 weighted hospitalizations; 39.6% had coded metastasis. Among non-metastatic admissions, vulnerability prevalence increased from 15.6% in 2016 to 21.9% in 2023. In-hospital mortality in the non-metastatic/vulnerable group (4.14% [95% CI, 4.00–4.29]) approximated metastatic mortality (3.96%) and exceeded non-metastatic/no vulnerability mortality (1.86% [95% CI, 1.81–1.91]). Compared with non-metastatic/no vulnerability, vulnerable admissions demonstrated higher ICU escalation (7.75% [95% CI, 7.56–7.94] vs 3.54%), AKI (27.30% [95% CI, 26.98–27.62] vs 17.25%), and any major complication (59.31% [95% CI, 58.95–59.67] vs 42.93%). Resource utilization was higher in the vulnerable group (LOS 9.33 vs 5.49 days; cost $33,156 vs $22,879). Among non-metastatic admissions with complications, mortality was higher with vulnerability (6.46% [95% CI, 6.24–6.69] vs 3.88%). In the non-metastatic surgical subcohort, failure-to-rescue mortality was 5.90% (95% CI, 5.59–6.22) with vulnerability versus 2.65%. Conclusions: Among U.S. GI cancer hospitalizations without coded metastasis, malnutrition/cachexia identifies a high-risk inpatient phenotype with complication burden, ICU escalation, and post-complication mortality closely resembling metastatic disease. These findings suggest physiologic vulnerability, rather than stage alone, drives inpatient risk and failure-to-rescue, supporting targeted inpatient risk stratification and earlier supportive interventions.
5575 Background: Patients with platinum-resistant ovarian cancer (PROC) have few therapeutic options. More than 60% of high-grade serous ovarian cancers overexpress (≥ 2+ by immunohistochemistry) Trop2, a key regulator of growth pathways and marker of aggressive disease. Sacituzumab govitecan (SG) is an antibody-drug conjugate consisting of a Trop-2–specific antibody conjugated with SN-38, an active metabolite of irinotecan. Methods: NCT06028932 is a single-institution open-label phase II study. Patients received SG at 10 mg/kg intravenously days 1 and 8 of a 21-day cycle until prohibitive toxicity or progression. The primary endpoint was objective response rate (ORR) by RECIST 1.1. Secondary endpoints included progression-free (PFS) and overall (OS) survival, as well as safety. Results: Twenty patients were enrolled. Median age was 67 years (range: 45-84). Most participants (85%, n=17) were White. The median number of prior lines prior to enrollment was 3 (range: 1-8). ECOG performance status was 0 (n=16) or 1 (n=4). Most patients had a poorly differentiated tumor (90%, n=18) with advanced disease (stage III/IV) at initial diagnosis (85%, n=17). Serous histology predominated (75%, n=15), followed by clear cell (15%, n=3), endometrioid (5%, n=1), and carcinosarcoma (5%, n=1). Across a median follow-up of 9.9 months, there were 15 progressions and 4 deaths. ORR was 35% (7/20); there were no complete responses, and stable disease was achieved in 40% (8/20). Median PFS was 8.0 months (95% CI: 3.8-14.8); median OS was not yet reached. No new safety signals were observed. The most common grade 3-4 treatment-emergent adverse events were neutropenia (n=15), hypokalemia (n=3), and anemia (n=2). The most frequent grade 1-2 events included diarrhea (n=59), fatigue (n=27), abdominal pain (n=20), nausea (n=20), alopecia (n=16), anorexia (n=14), and vomiting (n=13). Conclusions: SG exhibits encouraging efficacy for PROC with manageable toxicities. Future studies are warranted. Clinical trial information: NCT06028932 .
e15624 Background: Colon cancer remains a leading cause of cancer-related mortality. While immune checkpoint inhibitors (ICIs) like pembrolizumab have revolutionized treatment for advanced microsatellite instability-high (MSI-H/dMMR) colon cancer, methods to further improve survival is needed. Recent data have confirmed the safety of mRNA COVID vaccines in patients on ICIs, with emerging studies in melanoma and NSCLC even suggesting a potential synergistic effect whereby peritherapeutic mRNA COVID vaccines may enhance ICI efficacy and lead to improved survival. However, the clinical impact of this association in colon cancer remains unexplored. This study sought to evaluate whether mRNA COVID vaccination improves survival in patients with stage IV colon cancer receiving pembrolizumab. Methods: A retrospective cohort study was conducted using the TriNetX database. We identified stage IV colon cancer patients treated with pembrolizumab between 12/11/2020 and 01/26/2023. Group 1 (Vaccine) received an mRNA COVID vaccine within ±12 months of their first pembrolizumab infusion; Group 2 (Control) did not. Patients with skin or respiratory malignancies synchronous with colon malignancy were excluded to account for pembrolizumab treatment not targeting colon cancer. Propensity score matching (1:1) was performed based on age, sex, race, and comorbidities (diabetes, hypertension, obesity, heart disease, tobacco use, and liver disease). Primary outcome was median overall survival at 3 years (end date: 01/26/2026). Results: Following 1:1 propensity score matching, 374 patients were included in the final analysis (Vaccine group, n = 188; Control group, n = 186). Kaplan-Meier analysis revealed no statistically significant difference in survival between patients the vaccine group and the control group. Median survival for the vaccine cohort was 879 days, while median survival for the control cohort was not reached. At the end of the 3-year period, survival probability was 43.3% in the vaccine group and 50.4% in the monotherapy group, with a log-rank p-value of 0.308, suggesting that COVID-19 vaccination did not significantly impact survival outcomes in this cohort. Cox proportional analysis revealed no statistically significant difference in survival (HR 1.163; 95% CI 0.870-1.553). Conclusions: In this retrospective cohort study, mRNA COVID vaccination within a year of the first pembrolizumab infusion did not affect survival in stage IV colon cancer. This suggests that previous positive findings regarding vaccination during pembrolizumab therapy in other cancers might not extend to colon cancer. Limitations of our study include our assumption that all stage IV colon cancer patients treated with pembrolizumab were MSI-H due to inherent limitations preventing stratification by MSI-H status on the TriNetX platform.
INTRODUCTION:Performance status assessments may overlook important health vulnerabilities in adults with cancer. While geriatric assessment (GA) systematically evaluates multiple domains including physical function, cognition, psychological health, comorbidity, and nutrition, it has been predominantly studied in older adults. We evaluated the prevalence and patterns of GA-identified deficits among adults of all ages initiating systemic therapy for lymphoma. MATERIALS AND METHODS:Adults with lymphoma enrolled in a prospective registry at an academic cancer center from 2018 to 2025 completed a modified Cancer and Aging Research Group (CARG) GA at treatment initiation. The assessment included validated measures across eight domains: mobility (Timed Up and Go), falls history, instrumental activities of daily living (IADLs), medication burden, comorbidities, weight loss, cognition (Blessed Orientation-Memory-Concentration test), and psychological health (Mental Health Index-13). Karnofsky Performance Status (KPS) was documented. Prevalence of deficits was compared between adults <65 and ≥ 65 years. RESULTS:Among 97 participants (mean age 57 years, range 22-84; 39% ≥65 years), 94% had normal KPS (≥80). Despite preserved performance status, 56% had multiple (≥2) GA-identified deficits, and only 21% had zero deficits. The most common deficit was unintentional weight loss (43% overall). Among younger adults (<65 years), substantial proportions reported weight loss (34%), falls (27%), IADL impairment (27%), and anxiety or depression (45%). Most deficits did not differ significantly between age groups, though weight loss, multimorbidity (≥4 comorbidities), and severe polypharmacy (≥10 medications) were more prevalent in older adults. DISCUSSION:GA reveals high prevalence of health-related deficits across multiple domains in adults with lymphoma regardless of age, despite preserved performance status. These findings support a shift toward needs-based rather than age-based approaches to identify individuals who may benefit from supportive care interventions.
