Browse Articles
Discover research articles across all indexed journals
Geographic distribution of cancer treatment trials and population-level cancer outcomes in the U.S.
1556 Background: Access to cancer clinical trials varies geographically in the U.S., yet the relationship between trial availability and population-level cancer outcomes is poorly characterized. Understanding this relationship is critical for policymakers and trial sponsors seeking to reduce disparities in cancer outcomes. Methods: We conducted an ecological analysis of U.S.-based cancer treatment trials (1990–2025). Trial availability was measured using population-adjusted cumulative trial density (trials/10,000 population) and classified as none, low, medium, or high. County-level cancer mortality-to-incidence ratios (MIRs) were used to indicate post-diagnosis outcomes (lower MIR is better). Associations with trial density were assessed using logistic regression, with and without adjustment for the structural factors, including area deprivation index (ADI), health system performance, and area type (metro, micro, non-metro/non-core). Analyses were stratified by area type and ADI to stabilize estimates and assess associations across strata. Results: Between 1990 and 2025, 32,332 U.S.-based trials were conducted. The majority of counties (1,680, 52%; 10% of the U.S. population) had no trials. In counties with trials, counts ranged from 1 to 6,840 and density from 0.05 to 1,992 per 10,000 population. Compared to counties without trials, low-, medium-, and high-density counties were more likely to have lower MIR (OR = 1.9, 3.6, and 3.3, respectively; all p < .001). After adjustment for structural factors, associations persisted in medium- and high-density counties (OR = 1.7, 1.9; p < .001), but not in low-density counties (OR = 1.2; p = .30). Associations were strongest in metro counties (Table), with ORs of 1.91, 2.61, and 2.61 in high-ADI areas (all p<0.0001). In low-ADI areas, medium and high density were associated with lower MIR (OR=2.58 and 2.00; p<.001), but low density was not (OR=1.33; p=0.70). Conclusions: Cancer clinical trial availability in the U.S. is highly concentrated and strongly associated with improved population-level cancer outcomes. The co-occurrence of low trial density, structural deprivation, and higher cancer MIRs highlights opportunities to improve cancer outcomes by expanding the geographic reach of trial infrastructure. Associations between trial density and MIR. Areas Trial Density OR* p Adjusted OR* p All counties None Ref Low 1.90 <.001 1.16 .30 Medium 3.64 <.001 1.72 <.001 High 3.34 <.001 1.86 <.001 Metro None Ref Low 2.10 .005 1.37 .30 Medium 3.70 <.001 2.31 .005 High 4.13 <.001 2.83 .001 Micro None Ref Low 1.35 .05 1.10 .60 Medium 2.00 .001 1.33 .20 High 2.35 <.001 1.46 .04 Non-metro/non-core None Ref Low 0.97 .90 0.88 .80 Medium 3.72 .01 3.74 .02 High 2.37 .01 1.86 .10 Low ADI None Ref Low 1.33 .07 1.25 .20 Medium 2.58 <.001 2.38 <.001 High 2.00 <.001 1.91 <.001 Higher ADI None Ref Low 1.91 <.001 1.36 .05 Medium 2.61 <.001 1.59 .02 High 2.61 <.001 1.99 <.001 *Odds of lower MIR.
Efficacy and safety of zanubrutinib in Waldenström macroglobulinemia: A systematic review and meta-analysis of clinical trials.
e19569 Background: Bruton tyrosine kinase (BTK) inhibition has transformed the management of Waldenström macroglobulinemia (WM). Zanubrutinib, a second-generation BTK inhibitor (BTKi), has shown improved selectivity and tolerability in patients with WM, particularly with relapsed/refractory (RR) disease. However, the rarity of WM and limited availability of head-to-head trials restrict the feasibility of comparative meta-analyses of longer-term clinical outcomes. As such, a systematic review and single-arm meta-analysis represents the next best level of evidence to quantitatively assess the clinical performance of zanubrutinib across disease settings while avoiding exclusion of key data from single-arm trials. Methods: We conducted a systematic review and single-arm meta-analysis of prospective clinical trials evaluating zanubrutinib in treatment-naïve (TN) and RR WM. Random-effects models were used to pool response, survival, and adverse event outcomes. Fixed-time survival estimates were extracted directly or reconstructed from Kaplan-Meier curves when not explicitly reported. Results: Five trials comprising six cohorts and 322 WM patients were included, of whom 304 were efficacy evaluable. The pooled ORR was 89.7% (95% CI 82.1–95.6), with a MRR of 76.8% (95% CI 70.2–82.9) and a VGPR rate of 38.0% (95% CI 31.8–44.3). Pooled 12-month PFS and OS were 85.1% and 95.8%, respectively, with corresponding 24-month estimates of 77.8% and 92.1%. Efficacy outcomes were consistent across TN and RR cohorts. Any-grade AEs occurred in 94.8% of patients and grade ≥3 AEs in 64.6%; however, atrial fibrillation (3.5%) and major hemorrhage (3.0%) were infrequent. AE-associated treatment interruption rate was 10.2%. Conclusions: This systematic meta-analysis provides the most comprehensive synthesis of zanubrutinib outcomes in WM to date, demonstrating high response rates, meaningful depth of response, durable survival, and a favorable safety profile. These pooled estimates establish clinically relevant benchmarks that inform treatment selection and contextualize emerging comparative data in WM. Pooled clinical outcomes of zanubrutinib in Waldenström macroglobulinemia stratified by disease subtype. Outcome Pooled Estimate, % [95% CI] TN RR Efficacy Very good partial response rate`Major response rateOverall response rateStable diseaseProgressive disease 31.1 [18.1–45.6]79.0 [61.7–92.5]92.9 [71.2–100]2.2 [0–17.0]3.0 [0–14.4] 44.2 [31.1–57.7]76.6 [68.0–84.4]90.1 [79.4–97.6]2.0 [0–7.0]4.5 [0–15.6] Survival* 12-month PFS12-month OS24-month PFS24-month OS ---- 83.8 [62.9–97.7]93.1 [82.9–99.4]77.8 [51.7–95.8]91.2 [87.8–95.7] CI=Confidence interval, OS=Overall survival, PFS=Progression-free survival, RR=Relapsed/Refractory, TN=Treatment naïve. *Survival outcomes not quantitatively synthesized for TN subgroups due to insufficient data.
AI-based assessment of tumor-infiltrating lymphocytes (TILs) in patients with triple-negative breast cancer (TNBC) receiving neoadjuvant chemotherapy versus chemo-immunotherapy.
e12583 Background: Most patients with early-stage TNBC receive neoadjuvant chemotherapy and immunotherapy, yet biomarkers to guide treatment selection remain limited. Building on prior manual TIL analyses, we evaluated whether AI-derived TIL and immune spatial metrics are associated with response and survival, and whether these associations differ by treatment regimen. Methods: From the prospective DFCI Multicenter TNBC Registry, we identified patients with stage I–III TNBC who received neoadjuvant therapy. Pretreatment H&E–stained slides were analyzed for stromal TILs using both manual assessment and a zero-shot AI pipeline to quantify multiple immune and spatial features, including AI TIL score, immune hotspots, admixed immune infiltration, and immune habitat. Associations with pathologic complete response (pCR) and event-free survival (EFS) were evaluated using modified Poisson regression and Cox proportional hazards models, respectively. Multiple testing was addressed using the Benjamini-Hochberg false discovery rate method (q-values reported). Results: Baseline H&E slides were available from 105 patients who received neoadjuvant therapy (chemotherapy, n=40, (38%); chemo-immunotherapy, n=65, (62%) and included in this analysis. 12 patients (11%) presented with stage I, 44 (42%) stage II, and 49 (47%) stage III disease. 45 (43%) had lymph node involvement, 89 (85%) of patients received the KN522 regimen, and (15%) received a platinum and taxane-containing regimen. Overall, 61 (58%) patients experienced a pCR: 22 (55%) in the chemotherapy-only cohort and 39 (60%) with chemo-immunotherapy. Higher AI-derived TIL metrics were associated with increased likelihood of experiencing pCR, including AI TILs (RR per standard deviation (SD) 1.22, 95% CI 1.12-1.33, p<0.001, q<0.001), admixed immune infiltration (RR per SD 1.14, 95% CI 0.99-1.31, p=0.0625, q=0.120), immune hotspot (RR per SD 1.14, 95% CI 1.00-1.29, p=0.049, q=0.120), and immune habitat (RR per SD 1.15, 95% CI 0.99-1.33, p=0.0638, q=0.120). Associations were observed in the chemotherapy-only and chemo-immunotherapy cohorts combined, without evidence of a statistically significant interaction between AI-TIL metrics and neoadjuvant regimen. With a median follow-up of 2.6 years, the 3-year EFS was 85.3% in the overall cohort. Higher AI-TILs and immune habitat were associated with improved EFS (q<0.10), though event numbers were limited. Conclusions: These findings are consistent with prior manual TIL analyses and demonstrate that AI-TIL metrics are prognostic in early TNBC, though not predictive of benefit from adding pembrolizumab to neoadjuvant chemotherapy. Additional studies with longer follow-up are needed to validate these findings and further explore the potential role of AI-derived TIL metrics in early-stage TNBC.
Evidence supporting <i>RPS20</i> as a colorectal cancer susceptibility gene from a case-control MGPT study.
10617 Background: Ribosomal protein S20 ( RPS20 ) is increasingly included in colorectal cancer (CRC) germline multigene panel testing (MGPT) and clinical management guidelines, despite limited evidence of gene-disease association. Germline loss-of-function (LOF) variants in RPS20 have been observed in a small number of individuals with CRC, without tumor loss-of-heterozygosity, and have typically been classified as variants of uncertain significance (VUS). We conducted a case/control study of individuals heterozygous for RPS20 LOF variants to evaluate the CRC and colorectal polyp risk compared to controls and cases of established CRC susceptibility genes. Methods: Cases and controls underwent MGPT at Invitae (now part of Labcorp) from May 2018-May 2025. Cases included individuals with a single LOF variant in RPS20 who had no identified pathogenic variants (PV) in other hereditary cancer genes and no VUS in established CRC risk genes; controls consisted of individuals with no PV or VUS identified. To reduce ascertainment bias, cases and controls were identified from requisitioned (clinician-ordered, reported) and unrequisitioned (run on MGPT backbone but unreported) data, under WCG protocol 1167406. Ninety controls were randomly sampled and age/sex matched to cases. CRC/polyp diagnoses were determined by clinician-reported ICD10 codes and keywords and compared between cases and controls utilizing Fisher’s exact tests. To compare risk estimates with a well-defined hereditary CRC syndrome, odds of CRC/polyps in cases with RPS20 LOF variants and in cases with mismatch repair gene PVs ( MLH1, MSH2, MSH6, PMS2 ), both compared to controls, were estimated with multivariable logistic regression that included sex, age, requisition status, family history of cancer, and race/ethnicity. Results: Twenty-seven heterozygous RPS20 LOF cases were identified (median age at testing 48 years [IQR 35-37]). CRC prevalence was significantly higher in RPS20 LOF cases (63.0%) compared with controls (11.1%, p= <0.001), while polyp prevalence did not differ (7.4% vs. 10.0%, p=1). In multivariable analysis, RPS20 LOF variants were associated with increased odds of CRC compared with controls (odds ratio [OR] 18.0; 95% confidence interval [CI], 5.2–62.1), comparable to odds observed for MMR gene PGV cases (OR 9.5; 95% CI, 6.6–13.7). Conclusions: Heterozygous RPS20 LOF variants identified through MGPT are associated with substantially increased CRC risk. By incorporating age- and sex-matched controls and unrequisitioned testing data, this study reduces ascertainment bias and supports RPS20 as a CRC susceptibility gene. Larger studies are needed to inform clinical management. Multivariable logistic regression, RPS20 LOF cases compared to controls. Cancer/polyps OR CI p-value CRC 18.0 5.2 - 62.1 <0.001 Colon polyps 0.8 0.2 - 3.0 0.719 CRC/polyps 10.2 3.5 - 29.5 <0.001 Other Cancer 2.5 0.9 - 6.6 0.114
Metabolic health and risk of early-onset cancers in obese young adults: Real-world insights from a large cohort study.
e22670 Background: Early-onset cancers, typically defined as malignancies occurring in adults under the age of 50, are an emerging public health concern with rising incidence worldwide. Obesity is a major modifiable risk factor contributing to this trend and is strongly associated with cancers affecting the digestive system and hormone-sensitive organs. Metabolically healthy obesity (MHO) and metabolically unhealthy obesity (MUO) represent distinct obesity phenotypes, but their differential association with early-onset obesity-associated cancers remains incompletely understood. Additionally, emerging metabolic therapies, including GLP-1 receptor agonists, may hold potential for cancer risk reduction, highlighting the importance of clarifying metabolic risk profiles in young adults. Methods: We conducted a retrospective cohort study using the TriNetX US Collaborative network, including adults aged 18–49 years with a diagnosis of obesity (ICD-10-CM E66.9). MUO was defined as obesity with ≥1 diagnosis of hypertension (I10), type 2 diabetes mellitus (E11), or dyslipidemia (E78). MHO was defined as obesity without any of these conditions. Metabolic comorbidities were assessed up to 5 years prior to the index obesity diagnosis. Propensity score matching (1:1) was performed for age, sex, race/ethnicity, personal history of nicotine dependence, and family history of malignant neoplasm (Z80). The primary outcome was incidence of early-onset obesity-associated cancers, including breast (C50), colon (C18), pancreas (C25), endometrium (C54.1), kidney (C64), liver (C22), esophagus (C15), and gallbladder (C23) cancers, within 5 years following the index date. Results: Before matching, 1,293,901 patients with MUO and 1,500,261 patients with MHO were identified. After matching, each cohort included 966,484 patients. Incidence of early-onset obesity-associated cancers was 0.831% in the MUO group (n = 8,027) compared with 0.727% in the MHO group (n = 7,033), with a risk difference of 0.104% (95% CI: 0.079% to 0.128%; p < 0.0001). The risk ratio was 1.143 (95% CI: 1.107 to 1.181), and the odds ratio was 1.144 (95% CI: 1.108 to 1.182). Conclusions: Among young adults with obesity, those with metabolic dysfunction (MUO) had a modest but statistically significant increased risk of early-onset obesity-associated cancers compared with metabolically healthy obese individuals. These findings emphasize the importance of metabolic health in early cancer prevention strategies and support further investigation into targeted metabolic interventions that may reduce obesity-related cancer risk in younger populations.
Pucotenlimab plus dose-intensified radiotherapy and chemotherapy for locally advanced rectal cancer: Preliminary results from a prospective, single-arm phase II trial (PUCRT).
e15635 Background: The neoadjuvant PD-1 inhibitors plus chemoradiotherapy (nCRT) has improved complete response (CR) rates in patients with locally advanced rectal cancer (LARC). However, the optimal radiotherapy strategy—particularly whether simultaneous tumor-focused dose intensification can further potentiate immunotherapeutic efficacy—remains unclear. This study evaluates pucotenlimab combined with dose-intensified nCRT and explores whether this approach enhances tumor response and clinical outcomes in LARC. Methods: This prospective, single-arm phase II study enrolled patients with pMMR/MSS LARC (cT3–4 and/or N+, M0; tumor ≤10 cm from the anal verge). All participants received neoadjuvant, dose-intensified radiotherapy (57.5 Gy in 25 fractions) targeting the primary tumor and involved nodes, combined with concurrent capecitabine and pucotenlimab (anti-PD-1). This was followed by two cycles of consolidation therapy with pucotenlimab plus capecitabine and oxaliplatin (CAPOX). Based on treatment response and patient preference, patients underwent either radical surgery or a watch-and-wait (W&W) strategy. Subsequently, all patients completed four additional cycles of pucotenlimab plus CAPOX. The primary endpoint was complete response (CR), defined as either pathological CR (pCR) or clinical CR (cCR). Results: From August 2024 to December 2025, 36 patients were enrolled. 63.89% (23/36) cases with tumors were located ≤5 cm from the anal verge, including 38.89% (14/36) with ultra-low tumors (≤3cm). High-risk features (T4a/b, mesorectal fascia involvement, positive lateral lymph nodes, N2 disease, or extramural venous invasion) were present in 88.89% (32/36) of patients. At the data cutoff, 24 patients had completed response evaluation after consolidation therapy, 9 underwent surgery and 15 adopted a W&W strategy. The pCR rate among surgical patients was 55.56% (5/9). All 15 patients in the W&W group achieved cCR, resulting in an overall CR rate of 83.33% (20/24). Organ preservation was achieved in 62.50% (15/24) of all evaluable patientsand in 78.58% (11/14) among those with low rectal tumors. No treatment related death was reported. Conclusions: Pucotenlimab plus dose-intensified chemoradiotherapy achieved a high complete response rate with rapid tumor regression after just two consolidation cycles in patients with pMMR/MSS LARC, including those with high-risk and ultra-low tumors. This immunochemoradiotherapy regimen offers strong early tumor control and significant organ-preservation opportunity, particularly for very low rectal cancers. These promising findings warrant further validation in larger, comparative trials. Clinical trial information: NCT06770270 .
A distinct immune–microbiome axis in early-onset colorectal cancer.
e15736 Background: The incidence of early-onset colorectal cancer (EOCRC; ≤50 years) is rising globally and is increasingly recognised as biologically distinct from late-onset colorectal cancer (LOCRC). Differences in tumour immune contexture and gene-expression programs have been described, and emerging data suggest a role for tumour-associated bacteria and bacterial genotoxins, including colibactin-producing (pks+) E. coli , in EOCRC. However, integrated analyses linking immune phenotype, transcriptomics, and tumour-resident microbiome, including bacterial toxin signatures, remain limited. Methods: Pre-treatment colorectal tumour samples from EOCRC patients (n = 19) were compared with LOCRC controls aged > 65 years (n = 196). RNA sequencing was used to assess differential host gene expression, gene-set enrichment, and consensus molecular subtypes (CMS). Tumour-resident microbiomes were profiled using meta-transcriptomic analysis of unmapped RNA-seq reads, including targeted assessment of bacterial genotoxin signatures. Microbial diversity, taxonomic abundance, and expression of E. coli –associated toxins, including pks⁺ and cytotoxic necrotising factor 1 ( CNF1 ), were compared between age groups. Results: EOCRC tumours showed a distinct transcriptomic profile characterised by enrichment of cell-cycle and proliferative gene sets and depletion of immune-related pathways. When CMS subtypes were grouped by immune phenotype, 94% of EOCRC tumours were immune-excluded (CMS2/3) compared to LOCRC tumours (p = 0.04), with markedly reduced representation of immune-rich CMS1/4 subtypes. Genes associated with chromosomal instability and WNT signalling, including HOXA11-AS , STMN2 , and DACH2 , were overexpressed in EOCRC, while immune-associated genes were enriched in LOCRC. Tumour-resident microbiome analysis revealed significantly reduced microbial species richness in EOCRC tumours (p < 0.01), with distinct enrichment of specific bacterial taxa despite similar overall beta diversity. Expression of E. coli –derived genotoxins was significantly associated with younger age, including colibactin (pks⁺), previously reported in EOCRC, and CNF1 , representing a novel association. Several bacterial taxa enriched in LOCRC tumours have been previously linked to immunogenic host responses. Conclusions: EOCRC is characterised by an immune-excluded tumour microenvironment, proliferative transcriptional programs, and a distinct tumour-resident microbiome, including enrichment of bacterial genotoxin signatures. Integration of known EOCRC-associated colibactin signals with the newly identified association of cnf1 suggests that microbial genotoxicity may contribute to age-related differences in tumour biology. These findings support EOCRC as a biologically distinct disease entity and highlight tumour–microbiome interactions as potential targets for risk stratification and prevention.
Phase 1 study results with PRAME-directed T-cell receptor (TCR) T-cell therapies in synovial sarcoma.
11516 Background: Patients with advanced or recurrent synovial sarcoma (SS) have limited treatment options that can provide sustained disease control beyond 6 mos. SS expresses multiple cancer testis antigens, including PRAME, which is expressed in >50 cancers. Anzutresgene autoleucel (anzu-cel, IMA203) is a PRAME-directed TCR T-cell therapy engineered to recognize an intracellular PRAME-derived peptide presented by HLA and initiate a potent and specific antitumor response. IMA203CD8 is a TCR T-cell therapy that additionally engineers a CD8 co-receptor to allow both CD4+ and CD8+ T cells to detect and destroy PRAME-expressing tumor cells that evade conventional therapeutic approaches. Here we present phase 1 results for the subset of pts with SS from IMA203-101 treated with anzu-cel or IMA203CD8 (hereafter TCR T). Methods: Pts were ≥18 y, HLA-A*02:01+, PRAME+, with R/R solid tumors, measurable disease (RECIST 1.1), ECOG PS 0-1, and no remaining SOC treatment options. Following leukapheresis and TCR T manufacture, pts underwent lymphodepletion (LD) with Cy (500 mg/m 2 × 4 d) and Flu (30 mg/m 2 × 4 d), followed by TCR T infusion +/- low-dose SUBQ IL-2. Results: As of Oct 27, 2025, 21 pts with SS were treated with anzu-cel (n=9) or IMA203CD8 (n=12). The most common TEAEs were LD-related cytopenias (100%). Cytokine release syndrome occurred in 18 pts (86%; G1: 38%, G2: 38%, G3: 10%). One event of G1 immune effector cell–associated neurotoxicity occurred (5%). Twenty pts were evaluable for efficacy and received a low median dose of 1.51 x 10 9 total TCR T cells (range: 0.35-9.36 x 10 9 ). These pts had a median sum of target lesions at baseline of 10.22 cm (range: 1.24-41.1 cm) and received a median of 2.5 prior lines of therapy (range: 1-5). At baseline, 65% had ≥2 previous lines of systemic therapy and 75% received prior radiotherapy. The confirmed objective response rate (cORR), including all pts evaluable for efficacy across escalating doses, was 50% (10/20), including 1 confirmed CR (8%) with IMA203CD8 at a dose of 2.05 x 10 9 total TCR T cells. Median tumor reduction for pts with confirmed objective response was -63.4%. Median duration of response at these escalating dose levels was 12.7 mo (mFU: 23 mo) with 5 responses lasting for ≥12 mo including 2 ongoing PRs at 27 and 23 mo. Median overall survival was 18 mo. Results include 1 pt who received 3 sequential TCR T-cell therapies: the MAGE-A4-targeted afamitresgene-autoleucel (BOR: SD; PFS: 21.4 mos), PRAME-targeted anzu-cel (BOR: cPR; PFS: 24.4 mo), and IMA203CD8 (BOR: cPR; PFS ongoing at 24.4 mo). Conclusions: PRAME-directed TCR T-cell therapies exhibited a safety profile consistent with previous observations, and despite low doses, promising antitumor activity with durable responses that deepened over time in pts with SS. These results warrant further investigation. Clinical trial information: NCT03686124 .
Predictors of tumor lysis syndrome in non-Hodgkin lymphoma hospitalizations.
11178 Background: Tumor lysis syndrome (TLS) is a life-threatening oncologic emergency in non-Hodgkin lymphoma (NHL) caused by rapid tumor cell breakdown and severe metabolic derangements. Although TLS is well recognized, most risk data derive from small cohorts or trials, limiting generalizability. We used a nationally representative inpatient dataset to identify predictors of TLS in hospitalized NHL patients. Methods: We performed a cross-sectional analysis of the National Inpatient Sample (2016–2020). Adult NHL hospitalizations were identified using ICD-10 codes. Survey-weighted chi-square tests and multivariable logistic regression evaluated demographic, clinical, and treatment predictors. Analyses accounted for the complex survey design. Statistical significance was defined as p<0.05. Results: An estimated 350,400 NHL hospitalizations were identified, of which 0.79% had TLS. TLS patients were younger (65.7 vs 69.5 years, p < 0.001) and demonstrated different racial distributions, with lower proportions of White patients (67.8% vs 76.9%) and higher proportions of Black (12.7% vs 9.2%) and Hispanic patients (11.6% vs 8.6%) (p < 0.001). TLS occurred more frequently in hospitals in the West and Northeast (p = 0.036). TLS admissions had significantly longer stays (10.48 vs 5.91 days) and higher charges ($174,972 vs $76,884; p < 0.001). Survey-weighted regression identified several strong predictors of TLS. Bone marrow involvement was one of the strongest independent predictors (OR 13.78 [95% CI 5.45–34.84]). Electrolyte derangements were also powerful predictors: hyperkalemia (OR 8.20 [95% CI 6.82–9.86]), hyperphosphatemia (OR 8.28 [95% CI 6.62–10.36]), and hypocalcemia (OR 7.02 [95% CI 5.67–8.69]). Additional predictors included sepsis (OR 2.29 [95% CI 1.89–2.77]), neutropenia (OR 1.71 [95% CI 1.27–2.31]), HIV infection (OR 1.73 [95% CI 1.05–2.84]), and receipt of inpatient chemotherapy (OR 1.38 [95% CI 1.01–1.87]). Prior chemotherapy (OR 0.44 [95% CI 0.30–0.63]) and radiation therapy (OR 0.61 [95% CI 0.38–0.99]) were protective, suggesting TLS risk is greatest in chemo-naïve or high–tumor burden presentations. Conclusions: TLS in hospitalized NHL patients is associated with distinct demographic patterns and substantial increases in healthcare utilization. Early risk stratification incorporating clinical, laboratory, and treatment factors may improve prevention strategies and outcomes. Predictors of TLS among hospitalized patients with NHL. Predictor Adjusted OR 95% CI p-value Bone marrow involvement 13.78 5.45–34.84 <0.001 Hyperkalemia 8.20 6.82–9.86 <0.001 Hyperphosphatemia 8.28 6.62–10.36 <0.001 Hypocalcemia 7.02 5.67–8.69 <0.001 Sepsis 2.29 1.89–2.77 <0.001 Neutropenia 1.71 1.27–2.31 <0.001 HIV infection 1.73 1.05–2.84 0.031 Inpatient chemotherapy 1.38 1.01–1.87 0.041 Prior chemotherapy 0.44 0.30–0.63 <0.001 Radiation therapy 0.61 0.38–0.99 0.046
ABVD versus BEACOPP for front-line treatment of Hodgkin lymphoma: A systematic review and meta-analysis.
e19039 Background: Hodgkin lymphoma is a curable malignancy, the choice of frontline therapy for advanced-stage disease remains debated. ABVD has been the backbone of treatment for decades because of its reliable efficacy and lower long-term toxicity, while escalated BEACOPP was introduced to improve disease control. Randomized trials show better progression-free survival with BEACOPP but no clear overall survival benefit. BEACOPP carries higher rates of acute toxicity, infertility, and secondary malignancies particularly relevant given patients' young age and long life expectancy. Interpretation of existing literature is limited by differences in study design, patient selection, follow-up duration, and outcome reporting, so many individual trials are underpowered. A systematic review and meta-analysis is needed to define balance between efficacy and toxicity in advanced Hodgkin lymphoma. Methods: A total of 2858 records were retrieved from PubMed, Embase, Scopus, ScienceDirect, and the Cochrane Library from inception January 2026. Eligible randomized controlled trials and comparative observational studies of frontline BEACOPP (standard/escalated) vs ABVD in classical Hodgkin lymphoma reporting overall survival (OS) and progression-free/event-free survival (PFS/EFS), while secondary outcomes included complete or objective response, treatment-related toxicity (including grade 3–4 adverse events), secondary malignancies, and treatment discontinuation due to adverse events, and health-related. Pooled analyses used RevMan random-effects models (p<0.05). Results: Eight studies (n = 5187) assessed the effectiveness of BEACOPP compared to ABVD in patients with classical Hodgkin lymphoma receiving front-line systemic therapy. BEACOPP showed an 18% reduction in the risk of mortality (HR 0.82; 95% CI 0.48–1.41; p = 0.48) and a 38% reduction in the risk of progression (HR 0.62; 95% CI 0.45–0.85; p = 0.003), with moderate heterogeneity (I² = 55%). However, no significant differences were observed in complete response rate (RR 0.99; 95% CI 0.89–1.10; p = 0.82) or overall response rate (RR 1.03; 95% CI 0.95–1.11; p = 0.51) with heterogeneity (I² = 51% and 23%, respectively). BEACOPP was associated with a significantly higher risk of serious adverse events (RR 4.60; 95% CI 2.37–8.93; p = 0.04) and a higher risk of secondary malignancies (RR 1.30; 95% CI 0.47–3.58; p = 0.61), both reported with moderate heterogeneity (I² = 65% and 71%). Additionally, BEACOPP showed a non-significant reduction in the risk of event occurrence (HR 0.36; 95% CI 0.07–1.93; p = 0.23), though this result was limited by high heterogeneity (I² = 97%). Conclusions: BEACOPP improved progression free survival as compared to ABVD but did not significantly improve overall survival and was associated with significantly greater toxicity. This efficacy toxicity trade off must be put into close consideration with regimen choice.
Artificial intelligence comparison of Google MedGemma and ChatGPT-5 in expert-reviewed gastrointestinal and genitourinary oncology cases.
e15509 Background: Large language models (LLMs) are increasingly explored for oncology decision support, yet comparative evidence evaluating medical-specific versus general-purpose LLMs using expert-reviewed clinical cases is limited. We compared a medical-specific LLM (Google MedGemma 27B) with a general-purpose LLM (ChatGPT-5) using gastrointestinal (GI) and genitourinary (GU) oncology vignettes. Methods: This cross-sectional study evaluated 32 de-identified oncology vignettes (26 GI, 6 GU) created by a U.S. board-certified medical oncologist and informed by real-world cases. Vignettes included text-only information on the clinical context, description of pathology, tumor staging, biomarkers, and relevant comorbidities. Both LLMs were prompted to provide National Comprehensive Cancer Network(NCCN) guideline-based treatment recommendations, monitoring strategies, maintenance therapy considerations, and references, along with a self-reported confidence score (0–100). Board-certified U.S. medical oncologists independently reviewed outputs using a structured instrument; two disease specific expert oncologists evaluated each case. Outcomes included correctness and comprehensiveness (1–10 scales), hallucinations (factually fabricated or unsupported statements), NCCN traceability, potential patient harm, and clinician trust. Associations between accuracy and comprehensiveness were assessed using correlation coefficients. Inter-rater reliability was assessed using a two-way random-effects intraclass correlation coefficient (ICC). Results: ChatGPT-5 demonstrated higher mean correctness (8.65 vs 4.68) and comprehensiveness (8.68 vs 5.22) scores compared with MedGemma 27B. Accuracy and comprehensiveness were strongly correlated overall (Pearson r = 0.93, p < 0.001), with significant model-specific correlations (ChatGPT-5 r = 0.86, p < 0.001; MedGemma r = 0.77, p < 0.001). Hallucinations occurred less frequently with ChatGPT-5 (12.3% vs 65%), and NCCN-traceable recommendations were more common (75.4% vs 36.7%). Responses judged as potentially harmful occurred in 18.5% of ChatGPT-5 outputs versus 70% of MedGemma outputs. Clinician trust favored ChatGPT-5 (84.6% vs 15%). Inter-rater reliability improved with rater averaging (ICC(A,k): correctness 0.72; comprehensiveness 0.64). Conclusions: In expert-reviewed GI and GU oncology cases, ChatGPT-5 demonstrated higher clinical accuracy, comprehensiveness, and guideline alignment than a medical-specific LLM; however, a substantial rate of hallucinations and potentially harmful recommendations persisted across models. These findings underscore that, despite performance differences, current LLMs require rigorous clinician oversight and should not be used as standalone decision-support tools in oncology.
Association of cardiotoxicity with chimeric antigen receptor T-cell (CAR-T) therapy: A systematic review.
e19522 Background: Chimeric antigen receptor (CAR) T-cells represent a novel and highly effective option for the treatment of relapsed/refractory hematologic malignancies. However, CAR T-cell therapy also creates the potential for unpredictable and serious side effects, particularly cytokine release syndrome (CRS) and cardiovascular ischemia. Recent evidence has shown that there is an increasing recognition of arrhythmias, heart failure, and left ventricular dysfunction during the post-infusion period. Methods: A systematic review of published literature was performed in accordance with PRISMA guidance. PubMed/MEDLINE, Embase, Scopus, and the Cochrane Library were searched from inception to the most recent available date for studies reporting cardiovascular adverse events following CAR T-cell therapy. Eligible studies included clinical trials, observational cohorts, registry-based studies, and case series in adult or pediatric populations. Data was extracted on patient characteristics, CAR-T product type, CRS severity, cardiovascular event definitions, diagnostic modalities, timing of events, therapeutic interventions, and clinical outcomes. Risk of bias was assessed using study-design appropriate tools. Results: There were 20 studies included in our analysis. Pooled incidence of cardiovascular events observed was approximately 19.68% with arrhythmias (7.70%); heart failure (5.73%); and decreased left ventricular ejection fraction (3.86%) representing the most commonly reported cardiovascular events. Troponin elevation was reported in 23.61% of patients while NT-proBNP elevation occurred in 9.4% of patients studied. Most commonly, cardiovascular toxicity occurs shortly after CAR T-cell infusion and is frequently reported alongside systemic inflammatory toxicity/CRS. Higher CRS grade, elevated inflammatory markers, and baseline cardiovascular comorbidities were consistently associated with increased risk of cardiotoxicity. Management strategies reported included early CRS-directed therapy, hemodynamic support, temporary interruption of cardiotoxic co-medications, and treatment for arrhythmias and ventricular dysfunction. Conclusions: In summary, CAR T-cell therapy-associated cardiotoxicity is a clinically relevant event for approximately 1 in 5 patients; arrhythmias and heart failure represent the most prevalent phenotypes of this form of toxicity. Cardiac biomarker elevation is very common and provides an avenue for accurately detecting early development of toxicity, particularly among patients at risk for the development of cardiotoxicity. Establishing clear definitions for evaluating cardiotoxicity, along with developing surveillance and clinical management strategies within an integrated cardio-oncology clinical pathway, are essential for facilitating early detection and improving prognosis.
Association of post-treatment vaginal structural changes with psychosexual dysfunction in young cervical cancer survivors: Evidence from a low- and middle-income country.
5542 Background: Vaginal stenosis is a known late side effect of pelvic radiotherapy for cervical cancer. In low- and middle-income countries (LMICs), the extent of vaginal changes and their effects on sexual and emotional health are poorly studied due to cultural stigma and limited counseling. This study examined objective vaginal changes and their association with psychosexual distress among young cervical cancer survivors. Methods: This cross-sectional mixed-method study included 300 women aged ≤45 years who completed pelvic radiotherapy, with or without surgery and/or brachytherapy, 6–24 months earlier at Delta Hospital and collaborating centers. Participants completed a bilingual interviewer-administered questionnaire on counseling, intimacy, psychosocial distress, and perceived barriers. A standardized clinical examination was used to assess vaginal changes, including vaginal length, elasticity, adhesions, dryness, and dyspareunia, graded using LENT-SOMA and CTCAE v5.0 criteria. Outcomes were compared between women treated with definitive chemoradiation and those who underwent surgery followed by adjuvant radiotherapy ± brachytherapy. Results: The mean post-treatment vaginal length was 4.1 ± 1.2 cm. Reduced vaginal elasticity and adhesions were found in 68% of participants. Severe vaginal shortening (≤2 cm) or blind-ending occurred more often in women who received surgery followed by adjuvant radiotherapy and brachytherapy (approximately 40%) compared with those treated with definitive chemoradiation (p<0.05). The psychosexual impact was marked: 80% of women reported no return to penetrative intimacy or regular vaginal dilation, and 85% reported loss of sexual interest. The most common barriers were fear of injury or cancer recurrence (72%), pain during intercourse (65%), sociocultural stigma (48%), and partner separation or dysfunction (36%). Although 52% recalled receiving some counseling, only 15% actively discussed sexual health concerns, and fewer than 10% accessed specialized psychosexual care. Conclusions: Objective vaginal shortening and fibrosis are strongly linked to psychosexual distress and avoidance of intimacy among young cervical cancer survivors, especially after multimodality treatment. In low- and middle-income countries, unstructured counseling alone is not sufficient. Survivorship care should include regular vaginal assessment and culturally sensitive, partner-inclusive psychosexual counseling.
Sociodemographic and geographic disparities in early-onset pancreatic cancer mortality in the United States, 1999–2020.
e16459 Background: Early-onset pancreatic cancer (EOPC), defined as pancreatic cancer occurring before age 50,represents an increasingly recognized clinical entity of public health concern. Germ line and somatic mutations are involved in about 30% of EOPC .Comprehensive national data characterizing long-term mortality trends and disparities across race, sex, age, andgeographical location remain limited. We examined demographic and geographic patterns in EOPCmortality in the United States. Methods: We performed a population-based analysis using the CDC WONDER Multiple Cause of Deathdatabase (1999–2020). Deaths with pancreatic cancer as the underlying cause (ICD-10 C25)among individuals aged 25–49 years were included. Mortality rates per 100,000 were calculatedby race, sex, five-year age group, and U.S. state. Temporal trends were assessed using log-linearregression to estimate annual percent change (APC). State-level analyses used aggregated multi-year data to characterize geographic heterogeneity. Results: EOPC mortality showed substantial heterogeneity across demographic and geographicstrata. Black adults consistently had the highest mortality rates, followed by Asian/PacificIslander and White adults. Across all racial groups, men had higher mortality than women. Overtime, modest declines in mortality were observed among Black men (APC –1.09%/year) andBlack women (–1.04%), with smaller declines among White men (–0.64%) and White women(–0.42%). Despite these trends, absolute mortality rates among Black adults remainedpersistently higher.Age-specific analyses revealed variation within early-onset disease. Mortality was highestamong individuals aged 45–49 years (mean 4.20 per 100,000), with a modest decline over time(APC –0.55%). In contrast, mortality among adults aged 30–34 years, while low in absoluteterms, demonstrated a slight increasing trend (APC +0.54%), whereas rates among those aged35–39 years remained relatively stable.Geographic analyses identified marked state-level variation, with the highest mean multi-yearmortality rates observed in states such as West Virginia and Arkansas, with regional clusteringin Appalachian and Deep South areas. Conclusions: Early-onset pancreatic cancer mortality in the United States exhibits persistent disparities byrace, sex, age, and geography. These findings identify populations and regions withdisproportionate burden and may inform targeted awareness efforts and strategies to improvetimely diagnostic evaluation.
Using large language models to assess adherence to ASCO tele-health in oncology standards in post-Covid clinical documentation.
e23238 Background: In 2021, the American Society of Clinical Oncology (ASCO) published telehealth standards and practice recommendations outlining appropriate use, workflow, and required documentation for telehealth encounters in oncology. These standards include explicit documentation elements intended to ensure quality, equity, and continuity of care. In the post-COVID era, telehealth remains widely used; however, adherence to ASCO telehealth documentation standards has not been systematically evaluated. Manual chart review of unstructured electronic health record (EHR) notes is time-intensive and limits scalability. Large language models (LLMs) may enable efficient extraction of telehealth adherence measures from free-text clinical documentation. Methods: We conducted a retrospective EHR study of oncology telehealth encounters occurring between January 1, 2024 and October 1, 2025. Eighty patients with advanced solid or hematologic malignancies were randomly selected, each with at least one synchronous telehealth encounter (audio-only or audio-video). Fourteen documentation and appropriateness domains were derived from ASCO 2021 telehealth standards, including required visit documentation elements and visit-selection guidance. Two physician reviewers independently performed manual chart review with adjudication. A HIPAA-secure LLM was prompted to identify each domain and required generation of supporting source text. Performance was evaluated using sensitivity, specificity, accuracy, and F1 score. Results: Across adherence domains, LLM sensitivity ranged from 0.46-0.94, specificity from 0.34-0.97, and accuracy from 0.56-0.95. Highest performance was observed for documentation of visit modality and visit purpose, while lower performance was noted for patient location, provider location, and visit completion. Audio-video encounters demonstrated higher documentation completeness than audio-only encounters across most domains. The overall hallucination index was low (4.2%). Median review time decreased from 9.1 minutes per encounter with manual chart review to 0.8 minutes using LLM-assisted extraction. Conclusions: LLMs can identify adherence to ASCO telehealth documentation standards from unstructured oncology EHR notes in the post-COVID era. This approach may support scalable telehealth quality monitoring, identify systematic documentation gaps, and inform targeted workflow improvements in oncology practice.
Treatment decision-making after CAR-T and bispecific antibody therapy in relapsed/refractory multiple myeloma: Real-world perspectives from U.S. hematologist/oncologists.
e19525 Background: BCMA-directed CAR T cell therapies and bispecific antibodies (BsAbs) have transformed outcomes for patients with relapsed/refractory multiple myeloma (RRMM). As these therapies move earlier in treatment, an increasing proportion of patients relapse after exposure to one or both T-cell–redirecting modalities. Real-world evidence to guide treatment selection in this setting remains limited, creating uncertainty in clinical practice. Methods: A mixed-methods market research study surveyed 61 hematologist/oncologists from academic (43%, n = 26) and community (57%, n = 35) practices across the U.S. using a structured online questionnaire, supplemented by qualitative interviews conducted via web-based videoconferencing. Fieldwork occurred between October and November 2025. Quantitative measures assessed treatment patterns, sequencing strategies, access barriers, and real-world considerations influencing CAR T and BsAb use in RRMM using Likert-type scales and utilization metrics. Qualitative interviews followed a semi-structured guide exploring physician rationale for treatment selection, sequencing philosophies, and access-related decision-making. Results: Physicians consistently described relapse after CAR T or BsAbs as one of the most challenging aspects of RRMM management. While 56% preferred sequencing CAR T prior to BsAbs, real-world constraints often led to bispecific therapy being used first due to off-the-shelf availability. After progression, 67% reported no clear standard of care, relying on individualized decisions informed by prior response depth and durability, toxicity, performance status, and treatment availability. Re-treatment with an alternate T-cell–redirecting modality was commonly considered, though concerns regarding T-cell exhaustion, diminished efficacy, and cumulative toxicity were frequently cited. Some physicians favored non–T-cell–engaging regimens or clinical trial enrollment, particularly for frail or heavily pretreated patients. There was broad consensus that limited guidance and lack of strong efficacy data for post–T-cell–redirecting therapy represent a critical gap. Conclusions: As CAR T and BsAbs become foundational in RRMM, the lack of guidance for managing relapse after these therapies represents an emerging clinical gap. Physicians report heterogeneous, individualized approaches in the absence of a defined standard of care. Although CAR T remains preferred earlier in sequencing, real-world constraints frequently influence treatment order. These findings highlight the need for prospective studies and real-world evidence to clarify post–T-cell–redirecting therapy sequencing, inform clinical frameworks, support decision-making, and improve outcomes for patients with RRMM.
Multimodal immunotherapy for high-risk acute myeloid leukemia and myelodysplastic syndrome.
6569 Background: CD117 is highly expressed on myeloid leukemia cells (MLC) with limited expression on normal tissues. Our group has developed an anti-CD117 monoclonal antibody and demonstrated its safety in a phase 1 clinical trial for myelodysplastic syndrome (MDS). To potentiate engagement with MLC and immune effector cells, we engineered bispecific antibodies (bsAb) to bind CD117 and CD33 or CD38 and recruit natural killer (NK) cells using an active Fc and CD16 (CM1/3/16/16) and/or NKp46 scFv (CM1/3/16/17). NK cell cytotoxicity against MLC increases several-fold when activated by IL2, IL15 or IL21. We studied the in vitro and in vivo efficacy of bsAbs with fresh NK cells or IL2, IL15 and IL21 (IL21521) activated NK cells (cNK) to deplete MLC. Methods: In vitro cytotoxicity assays were performed on Kasumi-1 (CD117 high) and MOLM-13 (CD117 negative) acute myeloid leukemia (AML) cell lines and a primary AML sample (CD117 low) with bsAb and NK or cNK cells. cNK were exposed to IL21521 for 7 days. To mimic in vivo exposure, cNK were removed from IL21521 for 24, 48 and 72 hours prior to cytotoxicity assays. Using immune deficient NBSGW mice and primary patient sample of high risk (HR)-MDS transformed to AML (MDS/AML), a patient derived xenograft (PDX) model of HR-MDS/AML was created and allowed to progress to >80% CD45+ human leukemia chimerism (HLC) before treatment. Treatment groups included CM1/3/16/16 at 1mg and/or 2e6 NK or cNK given as a single dose per animal. Weekly blood HLC and a terminal 4-week blood, marrow and spleen HLC were performed. Results: In in vitro cytotoxicity assays, among the bsAbs, CM1/3/16/16 at 100nM with NK cells at 1:1 target:effector (TE) ratio produced the highest cytotoxicity against Kasumi-1 (70%) and MOLM13 (50%). cNK cells alone produced >90% cytotoxicity of Kasumi-1 cells at 1:1 TE ratio even at 72 hours after IL21521 washout. In the HR-MDS/AML PDX advanced leukemia model, the group receiving CM1/3/16/16 with cNK demonstrated the highest depletion of MLC, with 70% reduction of HLC seen at 4 weeks post-treatment. Mice in this treatment group had 100% survival at 4 weeks at which time they were euthanized for analysis. Conclusions: Both CM1/3/16/16 and cNK cells independently killed MLC with high efficacy in vitro . A single dose of CM1/3/16/16 combined with cNK cells depleted human MLC in PDX model of HR-MDS/AML with advanced leukemia. Further clinical investigation of this bsAb with allogeneic donor cNK cells in HR-MDS/AML is warranted.
Three decades of breast cancer predisposition genetic testing at a comprehensive cancer center.
e22633 Background: Uptake of germline genetic testing for breast cancer (BC) predisposition has expanded substantially over the past three decades. Temporal trends in pathogenic variant (PV) detection were assessed across Huntsman Cancer Institute’s (HCI) catchment area: Idaho, Montana, Nevada, Utah, and Wyoming. Methods: We conducted a retrospective observational study of individuals identified with BC predisposition PVs at HCI, describing demographics, rurality, and gene-specific PV detection across three time periods: 1994–2005, 2006–2015, and 2016–2025. Results: Among 16,169 patients who underwent germline genetic testing at HCI between 1994–2025, 3,179 (19.7%) were identified with PVs in BC predisposition genes; 3,097 with recorded testing date were included in temporal analyses. There was more than sixfold (612%) increase in identified PV carriers from the earliest to the most recent period (293 in 1994–2005 to 2,088 in 2016–2025). Mean age at testing among PV carriers was approximately 45 years and increased modestly in the most recent period compared with 2006–2015 (41.7 vs 45.9 years; p<0.001) (Table 1). Most PV carriers were White/Caucasian across all periods (84% overall), with minimal change in racial or ethnic diversity over time. BRCA1 and BRCA2 accounted for the largest number of PVs across successive periods ( BRCA1 : 195, 332, 360; BRCA2 : 117, 312, 611). Substantial growth was also observed in detected PVs in non- BRCA genes, including TP53 (15, 77, 213), ATM (0, 14, 320), CHEK2 (0, 11, 501), PALB2 (0, 11, 128), RAD51C (0, 4, 30), RAD51D (0, 1, 24), and STK11 (0, 5, 10). The geographic distribution of PV carriers broadened over time, with the number of counties represented increasing from 38 to 86. While approximately 80% of PV carriers resided in metropolitan areas, the absolute number of carriers from non-metropolitan areas increased 9.5-fold (47 to 448), with no significant change in proportional distribution (p>0.1) (Table 1). Conclusions: Detection of BC predisposition PVs increased markedly at HCI, with growing identification of carriers residing in both metropolitan and non-metropolitan areas, highlighting the need for strengthened infrastructure to support equitable care. PV carriers were identified at a mean age in the mid-40s, with a modest increase over time, suggesting missed opportunities for earlier identification to enable timely screening, risk reduction, and preconception counseling. Age and rurality by temporal changes among pathogenic variant (PV) carriers. Characteristics All PV carriers(n = 3,097) 1994-2005(n=293) 2006-2015(n=716) 2016-2025(n=2,088) Age, mean (SD) 44.7 (18.2) 43.8 (16.4) 41.7 (16.1) 45.9 (18.9) Rurality, n (%) Metropolitan* 2467 (79.7) 246 (84.0) 585 (81.7) 1636 (78.3) Micropolitan* 336 (10.8) 22 (7.5) 71 (9.9) 243 (11.6) Small Town* 100 (3.2) 5 (1.71) 20 (2.79) 75 (3.6) Rural Areas 189 (6.1) 20 (6.8) 39 (5.5) 130 (6.2) *Core, low and high commuting areas.
Derivation and external validation of a prognostic nomogram for HER2-negative early breast cancer in young Chinese females: A population-based study.
e12728 Background: Young breast cancer (YBC) is a distinct subtype, marked by aggressive tumor biology, and worse overall survival. HER2-negative breast cancer, accounting for 70-75% of cases, poses a significant challenge due to the absence of targeted therapies. Despite treatment advances, young women with HER2-negative early-stage breast cancer continue to have poor outcomes, underscoring the need for accurate prognostic tools to guide personalized treatment strategies. Methods: This study analyzed 7,863 patients with HER2-negative early YBC from the SEER database, diagnosed between 2010 and 2015. Independent prognostic factors were identified through univariate and multivariate Cox analyses. These factors were used to develop a nomogram predicting 3- and 5-year overall survival (OS). External validation was performed with data from 138 patients. Results: Tumor grade, estrogen receptor (ER) status, histology, T stage, N stage, and surgery type, were incorporated into the nomogram. The concordance indices (C-index) for the nomogram were 0.739 (95% CI: 0.723-0.755) in the training set, 0.748 (95% CI: 0.723-0.773) in the internal-validation set, and 0.711 (95% CI: 0.573-0.848) in the external validation set. Time-dependent ROC analysis demonstrated strong discrimination. Calibration curves confirmed the high predictive precision, and decision curve analysis (DCA) validated its clinical utility. Risk stratification indicated significantly better OS in high-risk patients compared to low-risk patients (p < 0.05). Conclusions: The prognostic nomogram developed in this study offers a reliable and clinically useful tool for predicting OS in young women with HER2-negative early breast cancer, facilitating more informed treatment decisions and improving patient outcomes. Baseline characteristics of HER2 negative early YBC included from SEER database and Chinese PLA General Hospital. Variable Training Set (N = 5,507) Internal Validation Set (N = 2,356) External Validation Set (N = 138) p-value 1 Age (median, IQR) 36.0 (32.0-38.0) 36.0 (32.0-38.0) 36.0 (33.0-38.0) 0.2 Age Group 18-35 years (49%) 18-35 years (48%) 18-35 years (42%) 0.3 Marital Status Married (61%) Married (61%) Married (95%) <0.001 Grade G1 (9.9%), G2 (35%), G3/G4 (55%) G1 (10%), G2 (34%), G3/G4 (56%) G1 (2.9%), G2 (62%), G3/G4 (35%) <0.001 Histology IDC (92%) IDC (92%) IDC (90%) 0.94 T stage T1 (40%), T2 (46%), T3 (11%) T1 (39%), T2 (47%), T3 (11%) T1 (47%), T2 (48%), T3 (2.9%) 0.052 N stage N0 (52%), N1 (35%) N0 (53%), N1 (35%) N0 (54%), N1 (25%) 0.009 ER Status 2 Positive (71%) Positive (72%) Positive (72%) 0.6 PR Status 2 Positive (63%) Positive (64%) Positive (74%) 0.016 Surgery Type Mastectomy (62%) Mastectomy (61%) Mastectomy (83%) <0.001 1 Kruskal-Wallis rank sum test; Pearson's Chi-squared test. 2 Abbreviations: Estrogen receptor, ER; Progesterone receptor, PR.
Coverage of subcutaneous versus intravenous oncology therapies by US commercial health plans.
1554 Background: Several widely used intravenous (IV) oncology therapies have recently been reformulated for subcutaneous (SC) administration. Compared with IV infusions, SC formulations reduce administration time and chair use, improve patient convenience, and enable more flexible sites of care. At the same time, SC formulations may alter established reimbursement and contracting arrangements, including shifts away from buy-and-bill payment, changes in site-of-care management, and differences in how net costs are realized by payers. We examined how US commercial health plans prioritize SC formulations relative to IV formulations in their drug coverage policies. Methods: We used the Tufts Medical Center Specialty Drug Evidence and Coverage (SPEC) Database to identify US commercial health plans with coverage policies for both SC and IV formulations of oncology therapies. SPEC includes detailed treatment access criteria from 17 large US commercial health plans. Coverage policies were current as of December 2025. The unit of analysis was a plan-drug combination, e.g., UnitedHealthcare’s policies for trastuzumab. Coverage was classified as IV preferred , on par , or SC preferred , depending on if one formulation is required before another. Results: We identified 86 coverage policies across 6 oncology therapies. One payer excluded the SC formulation from coverage in two instances, thereby denying enrollees access to the therapy. Among the remaining 84 policies, 8 (9.5%) were classified as IV preferred and 76 (90.5%) as on par ; no policies classified the SC formulation as preferred (Table 1). Coverage varied across therapies, with the frequency of IV-preferred coverage among payers covering both formulations ranging from 0% to 21%. Among the 76 on-par policies, 55 (72.4%) allowed patients to initiate treatment with either the IV or SC formulation without prior treatment requirements. In contrast, 21 (27.6%) required patients to first try and fail an alternative therapy before coverage for either formulation was granted. These step-therapy requirements were more common for older therapies with available biosimilars, and the required prior therapy was most often a biosimilar of the reference product. Conclusions: Across six oncology therapies, most US commercial plans covered SC and IV formulations on par, with a minority preferring IV and none preferring SC. Preferences varied by product, and access restrictions mainly reflected step-therapy requirements for older therapies with available biosimilars. Overview of coverage for intravenous (IV) and subcutaneous (SC) formulations. Drug # Payers Covering IV & SC Formulations SC Only Preferred Coverage Parity(SC vs IV) IV Only Preferred Drug A 16 0 14 (87.5%) 2 (12.5%) Drug B 15 0 14 (93%) 1 (7%) Drug C 15 0 14 (93%) 1 (7%) Drug D 15 0 15 (100%) 0 (0%) Drug E 14 0 11 (79%) 3 (21%) Drug F 9 0 8 (89%) 1 (11%) Total 84 0 76 (90.5%) 8 (9.5%)