Browse Articles

Early G-CSF after venetoclax-based induction for acute myeloid leukemia: Multicenter associations with infections, neutrophil recovery, and early mortality.

6518 Background: Venetoclax-based induction has improved remission rates in acute myeloid leukemia (AML) but is associated with prolonged neutropenia and infectious mortality. Optimal G-CSF timing remains uncertain. Methods: Multicenter real-world cohort (2018–2025) using TriNetX data across 21 U.S. institutions. Adults with newly diagnosed AML treated with venetoclax + hypomethylating agent or low-dose cytarabine were included. Early G-CSF was defined as initiation ≤ day 7 of cycle 1; deferred/none = > day 7 or none. The primary endpoint was grade ≥ 3 infection through day 42 (death as a competing risk, Fine–Gray models). Secondary endpoints: febrile neutropenia (FN), time to ANC ≥ 1.0 × 10⁹/L, ICU admission ≤ day 42, 60-day mortality, CR/CRi by day 60, and overall survival (OS). Multivariable models adjusted for age, ECOG, ELN risk, cytogenetics, baseline ANC/WBC, TP53/IDH/NPM1 mutations, antibacterial prophylaxis, and center effects. Results: N = 2,934; median age = 69 y (range 34–87). Early G-CSF = 39 %, deferred/none = 61 %. Early G-CSF lowered grade ≥ 3 infections (sHR 0.61 [95 % CI 0.49–0.76], p < 0.001) and shortened ANC recovery (Δ –3.2 days [–4.6 to –1.9], p < 0.001). FN risk was similar (aOR 0.94 [0.79–1.12], p = 0.49). CR/CRi by day 60 (aOR 1.06 [0.90–1.25], p = 0.47), OS (aHR 0.97 [0.85–1.11], p = 0.65), and 60-day mortality (aOR 0.92 [0.73–1.16], p = 0.49) were comparable. Findings persisted in age ≥ 75 and ELN-adverse subsets and remained robust after inverse-probability weighting and landmark analyses. Conclusions: Early (≤ day 7) G-CSF after venetoclax-based AML induction reduced severe infections and accelerated neutrophil recovery without affecting remission or survival. Results support standardized early G-CSF use to improve supportive-care safety, especially for older or high-risk patients. Early G-CSF timing and clinical outcomes after venetoclax-based AML induction. Outcome Deferred/None Early G-CSF Adjusted effect (95 % CI) P value Grade ≥3 infection ≤ day 42, % 31.4 21.2 sHR 0.61 (0.49–0.76) <0.001 Time to ANC ≥1.0 (days, median) 11.8 8.6 Δ –3.2 (–4.6 to –1.9) <0.001 Febrile neutropenia, % 38.9 37.4 aOR 0.94 (0.79–1.12) 0.49 ICU admission ≤ day 42, % 14.6 11.9 aOR 0.80 (0.65–0.99) 0.04 60-day mortality, % 10.1 9.3 aOR 0.92 (0.73–1.16) 0.49 CR/CRi by day 60, % 63.8 64.9 aOR 1.06 (0.90–1.25) 0.47 Abbreviations: sHR = sub-hazard ratio; aOR = adjusted odds ratio; aHR = adjusted hazard ratio; Δ = difference in medians. All models adjusted for baseline covariates (age, ELN risk, ECOG, cytogenetics, molecular subtype, prophylaxis use) and center effects.

Hypercalcemia as a marker of in-hospital mortality and resource utilization in breast cancer: A national analysis.

e12711 Background: Hypercalcemia is a well-recognized oncologic emergency associated with advanced malignancy and poor outcomes. While its clinical impact has been described broadly across solid tumors, contemporary national data characterizing the burden, temporal trends, and outcomes of hypercalcemia specifically among patients hospitalized with breast cancer remain limited. We evaluated national trends in prevalence, in-hospital mortality, and healthcare utilization associated with hypercalcemia among breast cancer hospitalizations in the United States. Methods: We conducted a retrospective, cross-sectional study using the National Inpatient Sample (NIS) from 2016–2020. Adult hospitalizations with a diagnosis of breast cancer were identified using ICD-10-CM codes. Hypercalcemia was defined by ICD-10 code E83.52. Survey-weighted analyses were performed to estimate national prevalence, temporal trends, in-hospital mortality, length of stay (LOS), and total hospital charges. Multivariable survey-weighted logistic regression was used to assess the association between hypercalcemia and in-hospital mortality, adjusting for age and sex. All analyses accounted for the complex sampling design of the NIS. Results: An estimated 166,127 weighted breast cancer hospitalizations were identified from 2016–2020. The weighted prevalence of hypercalcemia among breast cancer admissions increased from 2.57% in 2016 to 3.49% in 2020. Patients with hypercalcemia experienced significantly higher in-hospital mortality compared with those without hypercalcemia, with consistent differences observed across all study years (2020: 9.4% vs 5.0%). After adjustment, hypercalcemia was independently associated with increased odds of in-hospital mortality (adjusted OR 2.01, 95% CI 1.63–2.47). Hypercalcemia was also associated with longer LOS (2020: 7.37 vs 5.12 days) and higher mean hospital charges ($94,852 vs $70,946). Nationally, hypercalcemia accounted for an estimated 5,585 breast cancer hospitalizations annually. Conclusions: Among hospitalized patients with breast cancer, hypercalcemia is increasingly prevalent and independently associated with higher in-hospital mortality and substantial healthcare utilization. Although hypercalcemia likely reflects advanced disease burden and metastatic involvement, these findings highlight its value as a clinically meaningful marker of acute illness severity, identifying a high-risk inpatient population with disproportionate mortality and resource utilization.

Laminectomy versus laminoplasty in the surgical management of long-segment intradural spinal tumors: Any difference in neurological outcomes?

e14069 Background: Laminectomy (LAMT) and laminoplasty (LAMP) have been widely applied on patients with spinal cord tumors (SCTs). However, the clinical efficacy of LAMP versus LAMT remains controversial. The purpose of this study is to provide an updated assessment of the safety and efficacy of LAMP compared to LAMT in the treatment of SCTs, incorporating clinical evidence up to 2025. Methods: We searched several databases (PubMed, EMBASE, the Cochrane Library) to identify relevant clinical trials or observational studies. The outcome measures included neurological recovery, resection rates, and perioperative safety metrics. Meta-analysis was performed using a random-effects model. Results: Eighteen studies involving 1,212 patients were included. The results showed statistically significant differences favoring LAMP in terms of blood loss (SMD -2.96; 95% CI: -4.18, -1.75; P < 0.0001), hospital stay (SMD -0.97; 95% CI: -1.54, -0.39; P = 0.001), spinal deformity (OR 0.30; 95% CI: 0.12–0.72; P = 0.007), and cerebrospinal fluid leak (OR 0.19; 95% CI: 0.10–0.35; P < 0.0001). No significant differences were observed in surgery duration (SMD -0.30; 95% CI: -0.83 to 0.23; P = 0.27), effective recovery rate (OR 1.67; 95% CI: 0.40, 6.54; P = 0.51), or total resection rate (OR 1.39; 95% CI: 0.83, 2.33; P = 0.21). Conclusions: LAMP is a safer and more effective surgical method for SCTs, offering advantages in perioperative safety and spinal stability. While no significant difference was observed in surgery duration, neurological recovery, or tumor resection, the lower incidence of spinal deformity and CSF leak suggests that LAMP is a favorable surgical option for long-segment cases.

Shifting etiologic burden of hepatocellular carcinoma hospitalizations in the United States, 2016–2022: A National Inpatient Sample analysis.

e16345 Background: The epidemiology of chronic liver disease is evolving with widespread use of direct-acting antivirals (DAAs) for hepatitis C virus (HCV) and rising prevalence of metabolic dysfunction and alcohol use. We evaluated national trends in diagnosis-code–based etiologic attribution among hospitalizations for hepatocellular carcinoma (HCC) in the United States. Methods: We conducted a repeated cross-sectional analysis of the National Inpatient Sample (2016–2022). Adult (≥18 years) hospitalizations with a principal diagnosis of HCC (ICD-10-CM C22.0) were identified. Etiologies were assigned using ICD-10-CM codes in any diagnosis field and categorized as HCV, hepatitis B virus (HBV), alcohol-associated liver disease (ALD), nonalcoholic fatty liver disease/nonalcoholic steatohepatitis (NAFLD/NASH), mixed (≥2 etiologies), or none/other (no listed etiology codes). Trends in etiology shares were assessed with survey-weighted logistic regression (odds ratio [OR] per year). Survey-weighted multivariable models evaluated associations between etiology and in-hospital mortality (adjusted OR [aOR]) and resource utilization (log[LOS+1], log[charges+1]), adjusting for demographics, payer, and hospital characteristics. Results: We identified 17,049 unweighted hospitalizations, representing an estimated 85,245 HCC hospitalizations nationally. Annual volume declined from 12,680 in 2016 (95% CI, 11,789–13,571) to 11,610 in 2022 (95% CI, 10,915–12,305). The proportion attributed to HCV decreased from 34.9% (95% CI, 32.9–37.0) to 19.5% (95% CI, 17.8–21.2; OR per year 0.88, 95% CI 0.86–0.89; p < 0.001). NAFLD/NASH increased from 5.0% to 8.8% (OR per year 1.11, 95% CI 1.08–1.14; p < 0.001) and ALD increased from 7.2% to 9.7% (OR per year 1.04, 95% CI 1.01–1.07; p = 0.003). None/other increased from 37.5% to 48.6%. Compared with HCV-related admissions, mixed etiology had higher in-hospital mortality (aOR 1.34, 95% CI 1.08–1.66) and none/other also had higher mortality (aOR 1.19, 95% CI 1.03–1.39), while NAFLD/NASH showed borderline lower mortality (aOR 0.74, 95% CI 0.54–1.00). NAFLD/NASH and mixed etiologies were associated with higher adjusted charges (ratios 1.21 and 1.11, respectively) and longer LOS (ratios 1.05 and 1.10, respectively) versus HCV. Conclusions: From 2016 to 2022, the inpatient burden of HCC shifted substantially from HCV toward metabolic and alcohol-associated etiologies, alongside a growing proportion of unclassified cases. These trends underscore the need for prevention and care strategies addressing metabolic and alcohol-related risk factors and for improved capture of liver disease etiology in administrative data.

Evaluation of xenograft alignment with patient tumors to inform treatment studies in uterine carcinosarcoma.

e17647 Background: Accurate representation of patient tumors is critical to study the factors affecting treatment response in Uterine Carcinosarcoma (UCS), which is a rare and aggressive uterine malignancy. UCS is categorized as a complex metaplastic carcinoma with sarcomatous component arising from de-differentiation of carcinoma as well as collagen-rich stromal component. Despite the poor prognosis associated with standard chemotherapy, translational efforts to better model UCS architecture and inform therapeutic advancements remain limited. Methods: Primary UCS tumors from the surgical samples of six patients were used to grow patient-derived xenografts (PDXs). Both primary UCS tumor tissue and the matched PDXs underwent dual collagen and hyaluronic acid (HA) staining. High-resolution brightfield images were captured and quantitative spatial parameters, including ECM area fractions, HA : collagen ratios, gray-level co-occurrence matrix (GLCM) texture metrics, and Moran’s I spatial autocorrelation were analyzed. Group comparisons used non-parametric statistics, and principal component analysis (PCA) summarized multivariate ECM signatures. In parallel, patient-derived 3D organoid models are being established from the same frozen UCS tumor specimens. Results: Majority of the patient UCS tumors (n = 5) irrespective of their clinical stage of diagnosis, demonstrated higher collagen area fraction than their matched PDX tumors (0.20 ±0.01 vs 0.05 ±0.04; p<0.05), while HA levels were similar in both. Texture analysis revealed more heterogeneous collagen organization in patient tumors, with higher entropy (8.22 ±0.10 vs 8.07 ±0.02; p<0.05) and lower homogeneity (0.15 ±0.02 vs 0.19 ±0.01; p<0.05). Spatial statistics showed a two-fold increase in collagen Moran’s I in PDX tumors (0.51 ±0.02 vs 0.26 ±0.07; p<0.05), indicating stronger ECM clustering and reduced spatial heterogeneity than the patient tissue. PCA clearly separated patient and PDX samples, driven by HA : collagen imbalance, textural heterogeneity, and spatial metrics. Conclusions: Measurable differences in the stromal profile including reduced collagen content, entropy, and heterogenous organization in PDX tumors, underscore the importance of careful model selection when interpreting treatment-related studies in UCS. Our ongoing efforts of developing the patient-derived 3D organoid models aim to better preserve the UCS features, and will further elucidate factors contributing to the limited clinical benefit observed with standard chemotherapy drugs (carboplatin and paclitaxel) in UCS.

Knowledge of breast cancer warning signs and risk factors and their association with stage at diagnosis among women in Abuja, Nigeria.

e22538 Background: Late-stage presentation of breast cancer remains common in low- and middle-income countries and contributes substantially to poor survival. While early diagnosis depends on access to care, women’s ability to recognize symptoms and understand breast cancer risk factors may influence timely health-seeking and stage at presentation. Methods: We conducted a cross-sectional study between 2024 and 2025 among 94 women with histologically confirmed breast cancer receiving care at two tertiary hospitals in Abuja, Nigeria. Sociodemographic characteristics, knowledge of breast cancer warning signs and risk factors, and stage at diagnosis were obtained using structured questionnaires and medical records. Knowledge scores were categorized as good (≥75%), moderate (50–74%), or poor (< 50%). Associations with stage at diagnosis were assessed using chi-square tests, with statistical significance set at p < 0.05. Results: The mean age of participants was 48.3 ± 11.4 years, with 61.7% aged 31–50 years. Awareness of common warning signs was high, including breast lump (86.2%), axillary lump (85.1%), and breast pain (79.8%). Overall, 44.7% demonstrated good knowledge of warning signs. In contrast, knowledge of breast cancer risk factors was predominantly poor, with 67.0% scoring in the poor category and only 5.3% demonstrating good knowledge. Poor risk-factor knowledge was significantly associated with late-stage diagnosis (p = 0.046), whereas knowledge of warning signs was not (p = 0.521). Educational level was significantly associated with prolonged patient delay prior to diagnosis (p = 0.001). Conclusions: Among women with breast cancer in Abuja, high awareness of symptoms coexists with limited understanding of breast cancer risk factors. Poor risk-factor knowledge, rather than symptom recognition alone, was associated with late-stage diagnosis. Incorporating targeted risk-factor education into routine reproductive and primary health services may help improve early presentation in low-resource settings.

Immune cell profiling by spectral cytometry to predict response to avelumab in advanced bladder cancer patients.

e14529 Background: Bladder urothelial cancer is a common neoplasm with a significant proportion of patients who progress to advanced disease despite optimal local and systemic therapy. Newly immune related therapies have improved patients’ outcome; however, only certain patients achieve durable responses. Hence, there is a need for non-invasive biomarkers for early response prediction. Methods: Peripheral Blood Mononuclear Cells (PBMCs) from advanced bladder cancer patients treated with avelumab (n=11) were obtained prior to immunotherapy administration. One year follow up revealed treatment response, which allowed a cohort division for immune cell signature comparison. A 40-plex spectral cytometry panel was applied, and further computational analysis performed. Results: Unsupervised spectral cytometry analysis revealed a highly heterogeneous circulating immunome landscape, capable of identifying 78 distinct immune clusters. Further analysis discriminated 10 clusters that significantly differed between responders and non-responders, 8 clusters from T, NK & innate lymphoid cells lineage and 2 clusters from Myeloids & antigen presenting cells lineage (Figure 2). Further hierarchical validation confirmed an expansion of effector CD4+ T cells and CD8+ TEMRA cells, an among non-responder patients before therapy. On the contrary, those with subsequent response displayed an overrepresentation of CD2⁻ innate lymphoid cells as well as CD25⁺ classical monocytes, supporting a blood-based immune signature associated with therapeutic benefit. Conclusions: We have demonstrated that spectral cytometry analysis is capable of revealing distinct immune fingerprint among bladder cancer patients in response to immunotherapy. The biomarkers found in this study could therefore be of use for anticipating clinical outcome in these patients, improving disease management.

Comeback from long-course ADT with relugolix (rel) and darolutamide (daro) in hormone-sensitive prostate cancer (PC) (CLEARED).

5083 Background: The oral luteinizing hormone releasing hormone antagonist rel and androgen receptor pathway inhibitor (ARPI) daro are approved for treatment of advanced PC but the combination has not been prospectively investigated, and kinetics of testosterone (T) recovery after 2 years of rel have not been reported. Here we report pharmacokinetics (PK) as well as initial safety and reduction in T and PSA from a phase 2 trial of rel plus daro in high risk localized (HRL), lymph node-positive (N+) or low volume metastatic (LV) PC. Methods: Eligible patients (pts) had PSA > 0.02 ng/ml, T ≥ lower limit of normal (LLN) and were planned for 2 years of ADT+ARPI for HRL, N+ or LV PC. Prior ADT±ARPI was permitted if T recovered to ≥ LLN, but pts with high volume metastatic PC, prior PSA rise with castrate T, or on medications with significant predicted drug-drug interactions with rel or daro were excluded. Pts received rel 360 mg ×1 on day 1 (D1) followed by 120 mg QD and daro 600 mg BID starting on D1, and elected for PK sampling at 0h, 2h, 4h, 8h on D1 (cohort 1) or pre-dose on D1, 2, 8 and 29 (cohort 2). Radiation therapy to local and/or metastatic sites was delivered per standard of care. Pts receive up to 26 28-day treatment cycles with PSA/T monitoring every 3 months (mo) until 18 mo after end-of-treatment (EOT). The primary endpoints are rate of T recovery to > LLN by 18 mo after EOT and safety/adverse events (AEs). Secondary endpoints include PK, rate of treatment discontinuation, and pt-reported outcomes. Results: 33 pts enrolled: 8 with HRL, 11 with N+ and 14 with LV PC. 21 chose cohort 1 and 12 chose cohort 2. Single dose maximum concentrations (C max ) on D1 for rel and daro, and D29 pre-dose trough level (C trough ) for rel were similar to historical monotherapy comparators (comp) with geometric mean ratios (GMR) between 0.80-1.25 (Table), while D8 C trough for daro was slightly outside this range. By cycle 7 (C7) D1, 15 pts experienced Grade ≥3 AEs: 12 lymphopenia (3 possibly related to daro), with the remainder not treatment-related. 0 pts discontinued rel or daro; 3 dose-reduced daro. 33/33 (100%) pts achieved castrate T < 50 ng/dl and 32/33 (97%) achieved profound castrate T < 20 ng/dl (1 with T 24 ng/dl on C7 D1). PSA decline of ≥50% was achieved in 33/33 (100%) and decline ≥90% in 32/33 (97%) by C7 D1. Conclusions: The combination of rel and daro was safe and well-tolerated during the 1 st 6 treatment cycles, with PK parameters similar to historical monotherapy comparators. Appropriate initial T and PSA reduction were observed, and pts will be followed after EOT for the primary endpoint of T recovery (NCT06463457). Clinical trial information: NCT06463457 . PK for rel and daro. Geometric Mean Coefficient of variation (%) GMR rel D1 C max (N=19) [comp] 141.5 ng/ml [125] 145.2% [220%] 1.13 rel D29 C trough (N=11) [comp] 6.34 ng/ml [7.54] 62.6% [91.9%] 0.84 daro D1 C max (N=18) [comp] 2.10 µg/ml [1.77] 29.0% [26.4%] 1.19 daro D8 C trough (N=11) [comp] 2.52 µg/ml [1.82] 24.7% [32.8%] 1.38

Transient prostate-specific antigen flare induced by lutetium Lu 177 vipivotide tetraxetan in patients with metastatic castration resistant prostate cancer.

5078 Background: Prostate-specific antigen (PSA) flare, an initial rise in PSA that subsequently falls, is a well-known phenomenon in prostate cancer patients treated with hormone therapy and chemotherapy. However, its incidence, chronology, and clinical significance for treatment outcome of patients undergoing lutetium Lu 177 vipivotide tetraxetan (Lu-177), a newly FDA-approved radiopharmaceutical agent for metastatic castration-resistant prostate cancer (mCRPC), is uncertain. Our retrospective study, included mCRPC patients treated with Lu-177 between June 2022 and March 2025 at University of Kansas Cancer Center, aims to evaluate the association of PSA flare with clinical outcomes in patients with mCRPC undergoing Lu-177. Methods: PSA flare was defined as any initial increase in PSA level from baseline before start of 3 rd cycle followed by a decrease in PSA level of any degree or a decrease to less than baseline PSA level. The primary endpoint was progression-free survival (PFS). The secondary endpoints were overall survival (OS), PSA50 response, and disease control rate (DCR). The log-rank test was applied to compare PFS and OS between patient group with PSA flare and patient group without PSA flare. Results: The data from 158 patients were analyzed. The median age was 73.5 years old (range: 52-100). The median number of Lu-177 cycles received was 4 (range: 1-6). Forty five patients (28.5%) experienced PSA flare. The median time from treatment initiation to onset of PSA flare was 3 weeks (range: 1-9); whereas the median interval between PSA flare and PSA nadir was 14 weeks (range: 2-72). The median increase in PSA from baseline to flare was 39.1% (range: 3.3-1,133%); whereas the median decrease in PSA from flare to nadir was 70.1% (range: 8.4-98.4%). There was no statistically significant difference in PFS and OS between patient group with PSA flare and patient group without (median PFS: 8.2 vs. 5.6 months, p=0.20; median OS: 14.3 vs. 12.5 months, p=0.29). PSA50 response was observed in total of 76 patients (48.1%). In patient group with PSA flare, 48.9% of the patients achieved PSA50 response. Similarly, 48.7% of the patients achieved PSA50 response in patient group without PSA flare. Majority of patients were also assessed for radiographic response (n=139). Between patient group with PSA flare and group without PSA flare, DCR was achieved in 60% and 49.5% of the patients respectively (RR = 1.21, 95% CI: 0.88 to 1.67, p=0.26). Conclusions: The PSA flare phenomenon is not rare in mCRPC patients treated with Lu-177. Our study suggests the clinical outcome appears consistent regardless of the PSA flare. Therefore, Lu-177 should not be withdrawn early and prematurely in mCRPC patients who have an initial but isolated PSA rise following the therapy. Further studies are warranted to validate our findings.

Development of an endometrial cancer test from a vaginal swab.

5624 Background: Endometrial cancer (EC) is the most common gynecologic cancer and abnormal uterine bleeding (AUB) is the predominant presenting symptom. While <10% of AUB represents underlying EC, endometrial sampling is currently required to rule out EC. To develop an alternative, non-invasive rule-out strategy, we aimed to refine a previously reported EC-associated methylated DNA marker (MDM) panel in vaginal swabs. Methods: Patients ≥18 years old with AUB who met ACOG recommendations for endometrial sampling and those undergoing a hysterectomy for biopsy-proven EC or atypical hyperplasia (AH) were prospectively enrolled in a multicenter cohort study. A vaginal fluid specimen was collected via swab by a provider during a speculum exam. DNA was extracted from samples and bisulfite treated, followed by a target and signal amplification method to detect 8 previously identified MDMs. The cross-validated EC classification model used the mean of standardized MDMs. Clinical classification of benign endometrium (BE), EC, and AH was based on central pathology review. EC diagnoses were based on hysterectomy specimens. Endometrioid and mucinous constituted type 1 EC; serous, carcinosarcoma, clear cell, and mixed histology ECs were collectively analyzed as type 2. Results: Of 693 participants enrolled, 670 were evaluable with valid results: 524 in the AUB cohort and 146 in the suspected EC cohort. Within the AUB cohort, 16 had EC or AH. The final analysis cohort included 508 BE, 156 EC, and 6 AH. Of the evaluable participants 20% identified as Black, 3% Asian or Pacific Islander, 75% White, and 3% Other; 7% reported Hispanic ethnicity. Median age was 66 and 50 years for EC and BE AUB, respectively; 80% of those with EC and 43% with benign AUB reported being postmenopausal. Among participants with EC, 102 (65%), 17 (11%), 30 (19%), 6 (4%) were stage I, II, III, and IV, respectively; one EC was unstaged. Type 2 EC accounted for 42 (27%) of all EC cases. A 4-MDM panel yielded an AUC of 0.97 (0.96-0.99, 95% CI) for EC; models with 5-8 MDMs gave identical performance. All-stage EC sensitivity was 97, 97, 96, 95, and 92% at specificities ranging from 75, 80, 85, 90, and 95%, respectively. At the 85% (82-88%) specificity threshold, sensitivities across EC stage, type, and age were consistently high (Table). Sensitivity for AH was 33% (10-70%). Conclusions: An MDM panel demonstrated high sensitivity for EC across stage and EC type, supporting the potential for a non-invasive vaginal swab-based test to rule out EC. Future studies are necessary to finalize the panel and algorithm, prior to validating the test. Clinical trial information: NCT06294886 . EC Sensitivity (%, 95% CI) at 85% Specificity All Stages (n=156)* 96 (92-98) Stage I (n=102) 94 (88 - 97)    II (n=17) 100 (82 - 100)    III (n=30) 100 (89 - 100)    IV (n=6) 100 (61 - 100) Type 1 (n=114) 95 (89 - 98)    2 (n=42) 100 (92 -100) Age ≤ 66 yrs (n=86) 95 (89 - 98)   > 66 (n=70) 97 (90 - 99) *n=1 EC was un-staged.

Long-term responders to trifluridine/tipiracil plus bevacizumab in previously treated metastatic colorectal cancer: Results from the BeTAS real-world study.

3583 Background: The phase III SUNLIGHT trial established trifluridine/tipiracil plus bevacizumab (FTD/TPI+BEV) as a standard treatment option for metastatic colorectal cancer (mCRC) after failure of at least two prior regimens. However, limited evidence exists regarding patients achieving durable benefit in routine clinical practice. We aimed to characterize the prevalence, clinical features, and outcomes of long-term responders treated with FTD/TPI plus bevacizumab in a large real-world cohort. Methods: BeTAS is a multicenter ambispective real-world study conducted within the Galician Research Group on Gastrointestinal Tumors. Consecutive patients with previously treated mCRC receiving FTD/TPI+BEV after failure or intolerance to ≥2 prior regimens were included. Long-term responders (LTRs) were defined using a pre-specified landmark of progression-free survival (PFS) ≥9 months. Baseline characteristics, treatment response, and survival outcomes were compared between LTRs and non-LTRs. Overall survival (OS) was assessed using a 9-month landmark analysis to minimize immortal time bias. Survival was estimated using the Kaplan–Meier method and compared using the log-rank test. Multivariable Cox regression was performed. Results: Among 576 patients, median PFS and OS were 4.9 months (95% CI 4.3–5.5) and 10.8 months (95% CI 9.8–11.9), respectively. LTRs accounted for 20.3% of the cohort (n=117). Compared with non-LTRs, LTRs more frequently had ≤2 metastatic sites (83.8% vs 67.5%, p =0.001), absence of liver metastases (28.4% vs 16.7%, p =0.004), longer time from metastatic diagnosis ( p =0.046), and Tabernero best prognostic characteristics ( p =0.007). Among evaluable patients, objective response rate was higher in LTRs (22.1% vs 5.5%, p <0.001), as was disease control rate (100% vs 39.0%, p <0.001). Median PFS was 15.3 versus 3.6 months, and landmark OS was 23.5 versus 8.1 months (both p <0.001). Long-term response remained independently associated with improved OS (HR 0.19, 95% CI 0.14–0.26; p <0.001). Conclusions: In real-world practice, approximately one in five previously treated patients with mCRC achieve durable benefit with FTD/TPI+BEV. Identification of clinical factors associated with long-term response may support improved patient selection in later-line treatment.

Large-scale EMR-based machine learning for early risk stratification of colorectal cancer.

e15513 Background: Early detection of colorectal cancer (CRC) substantially improves outcomes; however, age- and symptom-based screening strategies fail to identify many high-risk individuals. Machine learning (ML) applied to longitudinal electronic medical records (EMR) may enable earlier, data-driven risk stratification beyond traditional risk factors. Methods: We analyzed a de-identified EMR event-stream dataset comprising 61,062,985 records from 300,000 patients. A case–control CRC cohort was constructed including 500 incident CRC cases (43,123 records) and 299,500 cancer-free controls with ≥3 years follow-up beyond the last observation. CRC onset was defined as the multidisciplinary tumor board diagnosis date. To reduce diagnostic work-up leakage, events within 6 months prior to onset were excluded.Patient histories were featurized over a primary prediction horizon of −36 to −6 months, summarized across four pre-diagnostic intervals (−6 to −12, −12 to −24, −24 to −36, and > −36 months). Models (random forest, LightGBM, histogram-based gradient boosting) were trained using patient-level splits, class rebalancing (SMOTE), and combined via a soft-voting ensemble.Generalizability was assessed in an independent, non-overlapping validation cohort (~100,000 patients) without restrictions on comorbidities and with CRC prevalence reflecting the general population. Results: In internal testing, the ensemble achieved accuracy 80.0%, sensitivity 72.2%, specificity 80.1%, NPV 95.0%, F1-score 75.9%, and ROC AUC 0.8175. Performance was preserved in the population-representative validation cohort (accuracy 79.0%, sensitivity 70.3%, specificity 79.0%, NPV 99.0%, F1-score 74.4%, ROC AUC 0.8217).Key contributors included age, healthcare utilization patterns across multiple specialties, routine laboratory parameters, and vital signs, rather than cancer-specific markers. Window-specific models for shorter pre-diagnostic intervals are under evaluation. Conclusions: Machine learning applied to large-scale longitudinal EMR data enables early CRC risk stratification up to three years before diagnosis using nonspecific, routinely collected features. Validation in an independent non-overlapping cohort supports feasibility as a phase-1, population-scale triage approach. Limitations include potential confounding by healthcare utilization intensity. Prospective evaluation of clinical yield (including PPV), sensitivity analyses, decision-curve utility, and validation across healthcare systems are planned.

Real-world trends in the evolving use of first-line therapy in metastatic castration-sensitive prostate cancer patients (mCSPC) in the United States.

e23384 Background: Management strategies for frontline therapy of mCSPC is rapidly evolving. Current standards of care include androgen deprivation therapy (ADT) combined with androgen receptor pathway inhibitors (ARPIs), as well as triplet regimens incorporating ADT + docetaxel chemotherapy, and ARPIs and recently with added poly-ADP ribose polymerase inhibitors (PARPi) in selected patients. These approaches have largely superseded ADT monotherapy. However, treatment intensification continues to be reportedly underutilized. This study aims to characterize real-world patterns and changes over time of frontline therapy with ADT doublet, triplet therapy or ADT monotherapy in men with mCSPC and assess survival differences. Methods: A retrospective cohort of men with prostate cancer was derived from the ConcertAI Patient360 Prostate Cancer Dataset. Patient characteristics and demographics were abstracted and treatment using ADT and combination therapies were defined by completing at least 6 months of first-line therapy. Results: Of the 17,736 patients (pts) included in the database from years 1988 to 2024, 5972 patients had stage IV prostate cancer with mCSPC, of whom treatment details were extracted from 2015 to 2024 in 2973 pts. Mean age at time of first treatment was 65.0 years (range 40 – 79). Caucasians made up 2132 pts (72%), Blacks 501 (17%), Asians 47 (2%), Others 293 (10%). The majority of patients were treated in Community centers: 2614 (88%) and Academic centers 359 (12%). Geographically, treatment occurred mainly in the South 1135 (38%), followed by the Midwest 688 (23%), the Northeast 559 (19%) and the West 478 (16%). ADT doublet was the most common first-line therapy utilized in 1615 (54%), with Other (519) and ADT Triplet (506) at 17% each. ADT monotherapy comprised 229 (8%) of pts. ADT combination therapy use has trended upward from 42% to 57% (ADT doublet) and 15% to 28% (ADT triplet) during the study period. Additionally, ADT monotherapy has trended downward from 10% to 5% during the same time. The utilization of primary definitive radiation in mCSPC pts occurred in 526 (18%) while surgery in 278 (9%), although lesser Black pts underwent radiation in 60 (12%) and surgery in 28 (6%), respectively. Conclusions: ADT doublet therapy with ARPI is the most commonly utilized frontline therapy for mCSPC in the US with corresponding decline in ADT monotherapy use over time, suggesting emergence of guideline-concordant care. While triplet therapy use is rising, it remains lower than ADT doublet use. Further investigation is needed to discern trends in survival in a real-world setting.

CIDER: Secure high-throughput extraction of clinical variables from non-English pathology reports with local large language models.

e13676 Background: Clinically relevant information is primarily embedded in unstructured narrative documents, limiting the scalability of clinical research and the development of registries. While large language models (LLMs) enable advanced clinical text mining, adoption is constrained by concerns regarding data sovereignty, multilingual performance, and reproducibility. We developed and validated a secure, on-premise LLM-based pipeline for structured data extraction from non-English pathology reports. Methods: CIDER (ClinIcal Data ExtractoR) is an on-premise, asynchronous pipeline for high-throughput extraction of structured clinical variables from pathology reports. The system uses vLLM-based inference with the open-source Qwen3-VL-32B-Instruct-FP8 model deployed in an air-gapped institutional environment. Validation was performed on 2,073 Hungarian-language histopathology reports. Seven clinical variables were evaluated against expert-curated reference data. Extraction accuracy, technical reproducibility (three independent runs at temperature [T] = 0.1), and robustness across stochastic sampling settings (T = 0–2.0) were assessed. Results: CIDER achieved near-expert concordance with manual abstraction, with accuracy exceeding 98% for sex and year of surgery, > 95% for T stage, and > 92% for N stage. Technical reproducibility was high, with negligible variability across repeated runs. The pipeline substantially improved dataset completeness by recovering clinically valid values omitted during manual abstraction, including 62.8% of missing T-stage annotations and 91.5% of missing tumor size measurements. Performance remained stable across sampling temperatures, with optimal accuracy at low stochasticity (T ≤ 0.2). Conclusions: CIDER demonstrates that locally deployed, open-source LLMs can reliably extract structured clinical data from complex, non-English pathology narratives while preserving full data sovereignty and reproducibility. By enabling scalable transformation of unstructured text into research-ready datasets, CIDER provides a secure and practical solution for clinical registry development and real-world evidence generation.

Bevacizumab rechallenge in platinum-resistant ovarian cancer: Evidence from a propensity score–matched analysis.

5607 Background: Bevacizumab is approved for the treatment of ovarian carcinoma across multiple disease settings, from first-line therapy to late platinum-resistant recurrences (PROC). The greatest relative benefit of bevacizumab has been observed in the platinum-resistant setting. While evidence supports bevacizumab rechallenge in the first platinum-sensitive recurrence, no randomized trial has demonstrated a benefit of bevacizumab rechallenge in PROC. Methods: We conducted a retrospective cohort study including patients with PROC treated with chemotherapy with or without bevacizumab. Patients were matched using propensity score matching based on histology (HGSC vs others), chemotherapy backbone, line of platinum-resistant treatment, age, BRCA pathogenic variant status, and prior bevacizumab exposure. Nearest-neighbor matching with a 1:2 ratio (bevacizumab:non-bevacizumab) and a caliper of 0.2 was applied. Balance between groups was assessed using standardized mean differences. Progression-free survival (PFS) and overall survival (OS) were estimated using Kaplan–Meier methods and compared using log-rank tests. Hazard ratios (HRs) were estimated using Cox proportional hazards models with robust variance accounting for matching. Subgroup analyses were performed according to prior bevacizumab exposure, and treatment–exposure interaction was formally tested. Results: A total of 149 patients with PROC were included. Baseline clinicopathologic characteristics were well balanced between treatment groups. In the overall matched cohort, bevacizumab use was associated with significantly improved PFS compared with chemotherapy alone (median PFS 7.79 vs 4.37 months; HR 0.49, 95% CI 0.36–0.68; p<0.001), as well as improved OS (median OS 23.06 vs 10.45 months; HR 0.47, 95% CI 0.34–0.66; p<0.001). In bevacizumab-naïve patients, bevacizumab remained significantly associated with longer PFS (median 7.79 vs 4.04 months; HR 0.47, 95% CI 0.33–0.68; p<0.001) and OS (median 23.06 vs 10.05 months; HR 0.46, 95% CI 0.32–0.67; p<0.001). Among patients previously exposed to bevacizumab, re-exposure was associated with numerically longer PFS (median 8.49 vs 4.89 months; HR 0.61, 95% CI 0.31–1.20; p=0.15) and OS (median 21.29 vs 10.45 months; HR 0.54, 95% CI 0.26–1.11; p=0.09), although these differences did not reach statistical significance. There was no statistically significant interaction between prior bevacizumab exposure and treatment effect for PFS (p for interaction = 0.60). Conclusions: The addition of bevacizumab to chemotherapy is associated with significant benefit in bevacizumab-naïve patients with PROC. In patients previously exposed to bevacizumab, re-exposure was not associated with a statistically significant improvement in outcomes, underscoring ongoing uncertainty regarding the role of bevacizumab rechallenge in this setting.

Role of nanotechnology in cosmetic development and skin care

Correction to “Rigidity‐Flexibility Regulation and Hard‐Soft Donor Combination: Dual Strategies in Covalent Organic Frameworks Construction for Actinides/Lanthanides Separation”

Near‐Infrared Upconversion Modulation of Intracellular Protons for Autophagy‐Induced Apoptosis

ABSTRACT Protons critically regulate cancer cell behavior, metabolism, and signaling pathways, making intracellular pH modulation a promising therapeutic strategy. Yet, precise spatiotemporal control of proton levels remains a formidable challenge. In this study, we introduce a near‐infrared (NIR)‐controlled nanoscale proton delivery system using upconversion nanoparticles (UCNPs) coated with photoacid (PA) and ferrocene (Fc). Upon 980 nm NIR stimulation, UCNPs emit UV–visible emission (300–500 nm), activating surface‐bound PA to induce transient H + release and acidify the tumor microenvironment in vivo. This acute acidic stress reduces tumor cell glucose uptake by 50% and suppresses mechanistic target of rapamycin (mTOR) signaling, triggering excessive autophagy that functionally drives mitochondrial dysfunction and intrinsic apoptosis—a process we define as proton‐mediated autophagy‐induced apoptosis (PAA). Fc, a biodegradable peroxidase mimic and a non‐fluorescent quencher, is incorporated to enable real‐time visual quantification of proton accumulation via H + ‐triggered biodegradation, restoring the NIR upconversion luminescence (at 800 nm) of UCNPs. Following intravenous administration, the nanoagent achieves a six‐fold reduction in tumor weight and elevates proton levels in glioma, effectively triggering PAA under non‐invasive NIR irradiation. This work establishes a spatiotemporally controlled platform for intratumoral proton dynamics, enabling precision cancer theranostics.

Predicting depressive symptoms among Chinese college students using recurrent neural networks with longitudinal data

Soft tissue and bone/joint sarcoma disaggregation in minorities.

e23100 Background: Hispanic and APPI populations are usually aggregated in large databases when there can be a difference in prognostication and survival among the individual subgroups. This study aimed to evaluate those differences in patients with soft tissue sarcoma and bone/joint sarcoma. Methods: Using the National Cancer Database, we identified patients with stages I-IV soft tissue sarcomas (STS) and bone/joint sarcomas (BJS). Multivariable logistic regression estimated odds of advanced stage (III–IV) at diagnosis (AS). Multivariable Cox regression estimated 3-year overall survival (OS). Models were adjusted for age, comorbidity, insurance, diagnosis year, and income quartile, with Non-Hispanic White (NHW) as reference. Results: Among 22,486 STS patients (NHW 19,806; Hispanic 1,932; AAPI 748) Hispanic patients had higher AS (aOR 1.12, p = 0.040), while AAPI patients had similar AS (aOR 1.00, p = 0.957), compared to NHW. When disaggregated, Pacific Islander patients had higher AS (aOR 1.61, p = 0.048) and worse OS (aHR 1.63, p = 0.002), while South/Central American patients had better OS (aHR 0.72, p = 0.046). Indian/Pakistani (aHR 0.73, p = 0.063) and Korean patients (aHR 0.63, p = 0.075) showed borderline trends towards better OS. Among 11,821 BJS patients (NHW 10,160; Hispanic 1,335; AAPI 326) Hispanic patients had higher AS (aOR 1.23, p = 0.008) when compared to NHW. OS in Hispanic and AAPI patients was similar to NHW. When disaggregated, AS and OS did not significantly differ between Hispanic/AAPI and NHW patients. Conclusions: Odds of advanced stage at diagnosis and 3-year overall survival for STS and BJS were different between aggregated and disaggregated analyses of Hispanic and AAPI groups, highlighting the importance of disaggregation studies to uncover and address hidden subgroup specific disparities. Hispanic and AAPI disaggregated subgroups adjusted odds and hazards ratios compared to non-hispanic whites in soft tissue sarcoma. Subpopulation n aOR (95% CI) p aHR (95% CI) p Non-Hispanic White 19,806 Reference - Reference - Hispanic, Unspecified 1330 1.12 (0.99–1.27) 0.073 0.97 (0.88–1.07) 0.565 Mexican 344 1.14 (0.90–1.44) 0.273 0.93 (0.77–1.12) 0.430 South/ Central America 123 1.31 (0.90–1.92) 0.158 0.72 (0.52–0.99) 0.046 Puerto Rican 68 1.33 (0.76–2.37) 0.320 0.94 (0.62–1.45) 0.794 Dominican Republic 36 1.15 (0.60–2.28) 0.677 0.77 (0.45–1.33) 0.346 Cuban 31 0.86 (0.41–1.77) 0.672 0.69 (0.37–1.28) 0.237 Hawaiian 48 0.69 (0.36–1.31) 0.254 1.24 (0.78–1.96) 0.370 Indian/Pakistani 163 0.78 (0.56–1.10) 0.155 0.73 (0.52–1.02) 0.063 Filipino 138 0.85 (0.59–1.23) 0.391 1.01 (0.75–1.36) 0.945 Chinese 132 1.28 (0.88–1.86) 0.197 1.08 (0.81–1.44) 0.611 Korean 69 0.84 (0.50–1.39) 0.495 0.63 (0.38–1.05) 0.075 Vietnamese 62 0.89 (0.52–1.52) 0.674 0.80 (0.51–1.28) 0.354 Japanese 60 1.00 (0.58–1.75) 0.997 0.77 (0.50–1.20) 0.250