Browse Articles

Landscape of peripheral immune response induced by multimodal thermal therapy combined with chemoimmunotherapy in patients with early-stage HER2-negative breast cancer.

e12654 Background: Multimodal thermal therapy (MTT), consisting of alternated liquid nitrogen cooling and radiofrequency heating, was proven to effectively trigger systemic antitumor immunity. However, their immune response induced by this emerging therapy has not been fully reported in breast cancer. This pilot study (ClinicalTrials.gov: NCT06636591) was conducted to determine the effect of MTT combined with preoperative chemoimmunotherapy in the treatment of early-stage HER2-negative breast cancer. Methods: Twelve HER2-negative breast cancer patients were recruited to preoperatively receive MTT followed by chemoimmunotherapy [6 cycles of Paclitaxel, Carboplatin and Camrelizumab (an anti-PD-1 antibody)]. Single-cell RNA sequencing was performed on peripheral blood mononuclear cells (PBMCs) from all patients at three timepoints [baseline (Day 0); prior to chemoimmunotherapy (Day 7); mid-on-treatment (Cycle 3)]. The primary goal was to explore the changes of PBMCs induced by the combination of MTT and chemoimmunotherapy. Results: MTT and chemoimmunotherapy were well tolerated in combination without any delay in prescheduled surgery. No treatment-related grade III/IV adverse events were observed. Single-cell RNA sequencing analysis of PBMCs revealed early systemic immune priming on Day 7 post-MTT but prior to chemoimmunotherapy. Specifically, increased plasma-cell frequency, elevated KLRK1 expression in NK cells, upregulated HLA molecules in antigen-presenting cells (APCs), and higher CCR2 expression in circulating monocytes were observed in patients achieving favorable axillary/breast responses and total pathological complete response (tpCR). These findings suggest that early peripheral signatures may serve as predictors of clinical benefit from subsequent therapy. By Cycle 3 of chemoimmunotherapy, peripheral plasma cells and APCs increased across the cohort, with expansion being most pronounced in patients with tpCR. This was accompanied by heightened expression of effector molecules and Fcγ receptor–related gene in NK cells and APCs. Furthermore, HLA molecules and CD40 expression on APCs were significantly upregulated than those on Day 7 after MTT, followed by increased CX3CR1 and PDCD1 expression on CD4⁺/CD8⁺ T cells. Additionally, enhanced interactions between monocytes/B cells and effector T/NK cells, coupled with the upregulation of tumor-homing chemokine receptors, indicating reinforced lymphocyte trafficking to tumor sites in patients with favorable response. Conclusions: The study reported global characteristics of the MTT-induced systemic immune response, providing an opportunity for the identification of potential targets to improve the efficacy of combined chemoimmunotherapy. Clinical trial information: NCT06636591 .

Does digital access translate into human capital gains? Assessing information technology use effects on cognitive and non-cognitive development of students in Western Rural China

This study explores how the use of information technology (IT) influences the development of cognitive and non-cognitive abilities among primary school students in rural areas of western China. As IT tools become more integrated into daily life, especially in underdeveloped regions, their influence on education has grown significantly. Relying on Heckman’s human capital development theory, we apply Propensity Score Matching combined with a Difference-in-Differences (PSM-DID) approach to analyze two rounds of panel data collected from third and fourth grade students between 2019 and 2020. The analysis suggests that while IT use does not significantly boost cognitive skills, a slight positive trend is observed, likely reflecting the coexistence of both beneficial and adverse effects. In terms of non-cognitive skills, frequent IT use appears to enhance students’ openness and increase their fondness for school and teachers, yet it is also linked to rising levels of self-blame and anxiety. Additional findings show that the effects of IT use are more pronounced among older children, Han Chinese students, and those from households with lower income or lower parental education. These results contribute to a deeper understanding of IT’s role in shaping human capital in rural settings and offer practical implications for policymakers. In particular, they highlight the need to maximize the positive aspects of IT in promoting equity in education while addressing its potential downsides through more customized support for diverse student populations.

Kevlar-based laser-induced graphene memristors and its relevant applications

Memristors, as key devices for compute-in-memory, are increasingly explored for flexible and wearable applications. The one-step fabrication of Al/laser-induced graphene (LIG)/Al memristor with flexibility and artificial synapse functionality on Kevlar through laser-induced technology has been demonstrated in this paper. The Kevlar-based LIG memristor features multi-level resistance switching, reproducible resistive switching over the measured 270 cycles, and a relatively large ON/OFF ratio of about 70, higher than that of some previously reported polyimide-based LIG memristors. In addition, the Kevlar-based LIG memristor demonstrates synaptic functions, including paired-pulse inhibition, dynamic synaptic response, and long-term potentiation/depression. The device still exhibits distinguishable resistive switching characteristics after 100 days under plastic encapsulation. Furthermore, by integrating the memristor with an LIG pressure sensor, an adjustable-threshold sensory response is realized. These results suggest the Kevlar-based LIG memristor as a multifunctional device that not only integrates resistive switching, synaptic functionality, and threshold sensing but also retains distinguishable switching characteristics under the tested storage conditions, making it attractive for intelligent protective clothing and flexible electronics.

3D printing of novel antimicrobial composite ink with silver nanoparticles and silica aerogel as bifiller incorporated in polymethylmethacrylate as matrix

Correction to “Dual Magnetic Resonances From Site‐Selective Doping of Hexaferrite Yielding Ultrabroadband Microwave Absorption”

Prognostic value of the aggregate index of systemic inflammation in palliative care patients

Abstract Accurate prognostication remains essential for appropriate care planning in palliative care settings. The Aggregate Index of Systemic Inflammation (AISI) represents a comprehensive integration of four inflammatory cell populations, but its comparative prognostic value in palliative care has not been established. This retrospective cohort study analyzed 144 consecutive patients admitted to a palliative care unit between January 2020 and December 2024. We evaluated the prognostic performance of AISI, Systemic Inflammation Response Index (SIRI), Systemic Immune-Inflammation Index (SII), and Neutrophil-to-Lymphocyte Ratio (NLR) for predicting in-hospital mortality, intensive care unit admission, and mechanical ventilation requirement. Receiver operating characteristic curve analysis, multivariable Cox regression, and Kaplan-Meier survival analysis were performed. A total of 144 patients (mean age 75.4 ± 11.8 years; 85 male [59.0%]) were included. In-hospital mortality occurred in 99 patients (68.8%). AISI demonstrated the highest discriminatory capacity for mortality prediction with area under the curve 0.826 (95% CI: 0.754–0.898), significantly superior to NLR (0.734, p  = 0.003) and SII (0.768, p  = 0.042). Optimal AISI cutoff of 1850 yielded sensitivity 79.8% and specificity 77.8%. In multivariable analysis, AISI > 1850 remained the strongest independent mortality predictor (adjusted HR 4.38, 95% CI: 2.41–7.96, p  < 0.001) after adjusting for established prognostic factors. Kaplan–Meier analysis revealed median survival of 9 days versus 35 days for patients with AISI > 1850 versus ≤ 1850 (log-rank p  < 0.001). AISI also independently predicted intensive care unit admission (adjusted OR 3.96, 95% CI: 1.89–8.28, p  < 0.001) and mechanical ventilation requirement (adjusted OR 4.15, 95% CI: 1.98–8.69, p  < 0.001). AISI demonstrates superior prognostic value compared to conventional inflammatory markers in palliative care patients. This readily available, cost-effective biomarker enables robust risk stratification for mortality and critical care interventions. Integration of AISI into comprehensive prognostic assessments could enhance individualized care planning and optimize resource allocation in palliative care settings.

Essential role of MptB in the biosynthesis of phosphatidylinositol mannosides, lipomannan and lipoarabinomannan in mycobacteria

Oncology drug shortages and disruptions to cancer treatment delivery in Mexico: A national oncologist survey.

1599 Background: Oncology drug shortages increasingly disrupt cancer care delivery worldwide, yet quantitative data describing their impact on treatment decisions in middle-income countries are limited. This study evaluates how drug and supply shortages affected cancer treatment delivery in Mexico during 2024–2025. Methods: Between September and December 2025, we conducted a national, cross-sectional, anonymous online survey of practicing oncologists in Mexico. The survey, promoted by the Mexican Society of Oncology (SMEO), consisted of structured multiple-choice questions assessing whether oncology drug or supply shortages required treatment modifications during routine clinical practice within the preceding 12 months. Respondents could select multiple types of treatment adaptations. Descriptive statistics were used to summarize responses. Results: A total of 196 oncologists and hematologists completed the survey. Overall, 191 respondents (97.4%) reported performing at least one therapeutic modification due to drug or supply shortages. The most frequently reported modifications were changes in treatment regimen (79.6%), omission of one or more planned treatment cycles (71.9%), and delayed or non-timely administration (62.8%). Additional adaptations included substitution with alternative drugs within the same pharmacologic class (57.7%), non–clinically indicated dose reductions (37.8%), and use of non–guideline-approved regimens (25.5%). Referral of patients to other institutions because of treatment unavailability wasreported by 12.8% of respondents. Only 2.6% reported no treatment modifications. Conclusions: In this national survey of oncologists and hematologists, drug and supply shortages in Mexico were highly prevalent and associated with frequent, clinically meaningful disruptions to cancer treatment delivery, including regimen changes, treatment omissions, and delays. These findings highlight a substantial challenge for cancer care delivery in a middle-income setting and underscore the need for health-system strategies aimed at preserving treatment continuity. Shortage-related therapeutic modifications reported by oncologists (n = 196). Therapeutic modification due to shortage n (%) Any modification performed 191 (97.4) Change of treatment regimen 156 (79.6) Omission of ≥ 1 planned treatment cycle 141 (71.9) Delayed / non-timely administration 123 (62.8) Drug substitution within same class 113 (57.7) Non-indicated dose reduction 74 (37.8) Non–guideline-approved regimen 50 (25.5) Referral to another institution 25 (12.8) No modifications performed 5 (2.6)

Engineering tumor-infiltrating lymphocytes (TILs) via the T-Editor platform to enhance antitumor activity.

2643 Background: Tumor-infiltrating lymphocyte (TIL) therapy has demonstrated clinical potential in melanoma, cervical cancer, and non-small cell lung cancer following failure of standard-of-care therapies, achieving objective response rates of 25-40%. However, two major challenges remain: the limited understanding of biological mechanisms underlying differential patient responses, and the absence of robust engineering technologies for manipulating fragile TIL populations. These limitations have constrained the application of genetic modification strategies to improve TIL therapeutic efficacy. Methods: By comparing transcriptomic profiles of infused TILs from responders (R) and non-responders (NR) in a Phase I investigator-initiated trial evaluating TIL therapy for solid tumors, we identified more than 10 differentially expressed genes as candidate therapeutic targets. We established the T-Editor platform, a CRISPR-mediated gene editing system optimized for TIL engineering, through systematic optimization of stimulation conditions, electroporation parameters, and CRISPR/Cas9 component dosing. Functional validation was performed by generating knockout (KO) constructs for each candidate gene using T-Editor, assessing their impact on TIL expansion, cytokine secretion, and cytolytic activity. The most promising target was selected for advanced cytosine base editing (CBE) studies, employing sgRNA designs to minimize Cas9-induced double-strand break risks. Base-edited TILs were comprehensively compared with Cas9-KO counterparts in vitro, while in vivo efficacy was evaluated using patient-derived xenograft (PDX) mouse models. Results: The T-Editor platform we established achieved optimal gene editing efficiency with defined stimulation condition, electroporation program and CRISPR/Cas9 doses. Functional screening revealed four critical regulatory genes, with FAM84B emerging as the most significant target where knockout demonstrated an increase in cytolytic activity compared to controls. CBE-mediated C·G→T·A conversion at FAM84B exon achieved high editing efficiency with minimal insertion-deletion events. Base-edited TILs showed comparable expansion kinetics, phenotypic stability, and cytokine production to Cas9-KO counterparts. In vivo studies demonstrated increased tumor growth inhibition in PDX models with FAM84B-edited TILs, which exhibited enhanced memory phenotype, superior effector function and elevated IFN-γ secretion. Conclusions: The T-Editor platform represents a rapid, efficient, and safe CRISPR-based system for TIL engineering, enabling both gene knockout and precise base editing applications. Our discovery of FAM84B as a potential novel inhibitory regulator of TIL function establishes a promising therapeutic target for improving adoptive cell therapy outcomes in solid tumor treatment.

A real-world study of <i>RET</i> fusion in patients with solid tumor.

e20678 Background: RET activates downstream oncogenic signaling pathways such as RAS/MAPK, which regulates essential cellular processes. RET fusion is established actionable target in lung cancer, thyroid cancer, and other solid tumors. Their clinicopathological and molecular features, treatment patterns, and prognostic implications remain less characterized. Methods: This single-center retrospective cohort study included patients with RET fusion-positive solid tumors who were diagnosed and treated at the First Affiliated Hospital of Xi'an Jiaotong University between August 1, 2016, and May 1, 2025. Demographic, pathological, molecular, treatment, and follow-up data were collected from electronic medical records. Survival analysis was performed using the Kaplan–Meier method, with prognostic factors analyzed by Lasso-Cox and multivariable Cox proportional hazards models, and safety was evaluated according to CTCAE version 5.0. Results: The study enrolled 160 patients with RET fusion-positive solid tumors. The median age is 58 years, 56.3% are female, and 73.1% are never smokers. Lung cancer is the predominant primary site, with KIF5B being the most frequent fusion partner and TP53 the most common co-altered gene. The median overall survival (mOS) across tumor types: 48 months for lung cancer (n = 115), not reached for thyroid cancer (n = 21), and 20 months for gastrointestinal tumors (n = 9). In the lung cancer subgroup, baseline liver metastasis (HR 7.2, 95% CI 2.03-25.55, p &lt; 0.05), regional lymph node metastasis (HR 5.7, 95% CI 1.74-18.51, p &lt; 0.05), and adrenal metastasis (HR 9.9, 95% CI 2.85-35.09, p &lt; 0.05) are independent risk factors for OS. Compared to chemotherapy, RET-TKIs exhibited longer PFS in both first-line (median PFS 11 vs. 14 months; HR 1.41, 95% CI 2.17-13.21) and second-line treatment (median PFS 4 vs. 8 months; HR 1.49, 95% CI 1.82-41.04). Among 55 patients treated with RET-TKIs, median PFS was 12 months, with significantly shorter PFS observed in those harboring co-occurring driver gene alterations (4 vs. 14 months, p &lt; 0.05). No statistically significant differences in PFS were detected between selpercatinib and pralsetinib (13 vs. 12 months; p = 0.37), nor between first-line and subsequent-line administration (12 vs. 13 months, p = 0.91). RET-TKIs demonstrated favorable tolerability profiles: selpercatinib was primarily associated with edema, rash, and hypertension, while pralsetinib showed higher incidence of thrombocytopenia and hepatic dysfunction. Conclusions: RET fusion is frequently observed in younger, female, never-smoking patients with lung cancer. RET-TKIs demonstrated a trend toward superior efficacy compared to chemotherapy in the advanced cohort, with manageable safety profiles.

A phase I study of pomalidomide and nivolumab (Pom/Nivo) in patients with virus-associated malignancies with and without HIV.

2663 Background: Up to 15% of cancers worldwide are associated with oncogenic viruses and disproportionately affect people with HIV (PWH). Epstein–Barr virus (EBV) is linked to several malignancies, particularly lymphomas; Kaposi sarcoma herpesvirus (KSHV) to Kaposi sarcoma (KS) and primary effusion lymphoma (PEL); and human papillomavirus (HPV) to anogenital and head and neck cancers. Virus-associated tumors evade immune surveillance through mechanisms such as T-cell exhaustion. In vitro studies show that pomalidomide (Pom), an immunomodulatory drug, enhances T-cell co-stimulation and increases immune surface marker expression in virus-infected cell lines. Pom may synergize with nivolumab (Nivo), an anti-PD-1 monoclonal antibody, to reverse immune exhaustion and enhance antitumor immunity. Methods: This phase I study enrolled participants (pts) with advanced virus-associated malignancies (EBV, KSHV, HPV). PWH were required to receive antiretroviral therapy for ≥4 weeks with an HIV viral load (VL) ≤400 copies/mL. There was no CD4 T-cell requirement; however, major opportunistic infections (with limited exceptions) were not permitted within 6 months of enrollment. Intravenous Nivo (480 mg) was administered every 28 days with Pom given once daily on days 1-21 of a 28-day cycle using a 3+3 dose-escalation design (DL1=3 mg, DL2=4 mg, optional DL-1=2 mg) for up to 24 cycles. The primary objective was safety and tolerability of the combination (using CTCAE v5.0). The secondary objective was to evaluate the anti-tumor activity of Pom/Nivo using disease-specific criteria (RECIST 1.1, Lugano criteria, or modified ACTG criteria for KS). Results: Sixteen pts (75% male, 56% White and 37% Hispanic) were enrolled, including 9 PWH with a baseline median (med) CD4 T-cell of 253 cells/mm³ (range 61-616) and HIV VL 37 copies/mL. After a med of 7 cycles (range 1-24), the most common Grade (G) 1/2 toxicities were anemia, lymphopenia, and maculopapular rash. Five pts had G3 or G4 neutropenia that was managed with growth factor support and Pom dose reductions. The only dose-limiting toxicity was observed in 1 pt with metastatic anal cancer in DL1 who developed G3 dyspnea; the maximum tolerated dose of Pom was 4 mg. Five pts were not evaluable for an objective response (2 died from progressive disease (PD), 2 declined further treatment, 1 was deemed ineligible), among 11 pts who received ≥3 cycles, all 4 pts with KS had a response (2 partial and 2 complete responses), 1 pt with PEL had stable disease (SD), 1 of 4 pts with HPV-malignancies had SD and 1 of 2 with nasopharyngeal carcinoma had SD, and 4 pts had PD. Among PWH, CD4 T-cell count did not change from baseline to the end-of-treatment (p=0.46). Conclusions: Pom/Nivo is a chemotherapy-sparing regimen with acceptable safety in advanced virus-associated cancers and promising activity in KS. Based on these results, an expanded KS cohort will be added to the trial. Clinical trial information: NCT04902443 .

Phase 2 study of adjuvant botensilimab in combination with balstilimab in patients with microsatellite-stable colorectal cancer and persistent circulating tumor DNA following surgery and chemotherapy.

TPS3689 Background: In patients with localized microsatellite stable (MSS) colorectal cancer (CRC), detection of circulating tumor DNA (ctDNA) after completion of definitive therapy indicates the presence of minimal residual disease (MRD) and predicts metastatic recurrence within months. Novel adjuvant therapies to prevent recurrence are very much needed. Botensilimab (BOT) is an Fc-enhanced multifunctional anti-cytotoxic T-lymphocyte-associated protein 4 (CTLA-4) antibody designed to expand therapy to cold/poorly immunogenic solid tumors, such as MSS CRC. BOT plus balstilimab (BAL; PD-1 inhibitor) has previously demonstrated deep, durable responses in treatment-refractory metastatic MSS CRC and no active liver metastases. We hypothesize that treatment of MRD with effective immunotherapy in patents with MSS CRC after definitive therapy will prevent tumor recurrence. Methods: This investigator-initiated, open label, Phase 2 study evaluates the efficacy of BOT/BAL in patients with stage III MSS CRC (cohort 1) or colorectal liver metastases (CRLM) (cohort 2) and MRD following definitive treatment, including surgery and pre-operative chemotherapy. The study is being conducted in 2 parts. Part 1 aims to determine ctDNA clearance, as a surrogate for improved survival. Patients receive BOT/BAL for 6 months followed by BAL alone for 6 months. With a total of 27 patients per cohort, and using an exact single-stage binomial design, we will have 80% power to detect an improvement in ctDNA clearance from 5% to 20% with a type I error rate of 5%. If part 1 is successful, then part 2 of the study may proceed as a randomized, placebo-controlled, double-blind, multicenter study to assess improvement in recurrence-free survival. Translation studies are planned to correlate tumor and immune characteristics with ctDNA clearance and clinical endpoints to further our understanding of immune responsiveness in the setting of MRD in MSS CRC. Enrollment to the study is currently ongoing. Clinical trial information: NCT07227636 .

Validation of an AI-based clinical decision support system for guideline-concordant prostate cancer treatment at a Brazilian cancer center.

1605 Background: Clinical decision support systems (CDSS) based on large language models may improve adherence to evidence-based guidelines in prostate cancer. We developed a Gemini-based CDSS for localized and locally advanced prostate cancer treatment at Instituto do Câncer do Ceará (ICC), Fortaleza, Brazil, and conducted a validation study comparing its recommendations with National Comprehensive Cancer Network (NCCN) guidelines and real-world physician decisions. Methods: We retrospectively included 198 consecutive patients prostate cancer treated at ICC. For each case, the CDSS generated a recommended primary treatment. Independent clinician reviewers assigned reference NCCN-concordant treatment categories. We calculated per-treatment precision, recall, F1-score, and one-vs-rest AUC comparing CDSS and physicians, with particular focus on active surveillance (AS) and radical prostatectomy (RP). Results: Overall NCCN concordance was 87.9% for the CDSS and 82.3% for physicians. Treatment-level analysis showed that the CDSS more often and more accurately identified candidates for active surveillance. For AS, the CDSS achieved precision 1.00, recall 1.00, F1-score 1.00, and AUC 1.00, while physicians had precision 1.00 but substantially lower recall (0.24), F1-score 0.39, and AUC 0.62. In contrast, physicians more frequently recommended RP beyond NCCN indications: for prostatectomy, the CDSS showed precision 1.00, recall 1.00, F1-score 1.00, and AUC 1.00, whereas physicians had lower precision (0.40) with recall 1.00, F1-score 0.58, and AUC 0.90. Conclusions: In this single-center validation study, a Gemini-based CDSS for prostate cancer treatment achieved higher overall NCCN concordance than treating physicians and improved alignment with guidelines for the appropriate use of AS. These findings suggest that integration of large language model–based CDSS into clinical workflows at cancer centers in low- and middle-income settings may reduce overtreatment and support more value-based prostate cancer care. Although we did not include data on social impossibility of AS, the finding leads to reflections on the possibility of greater use of AS, even in a region with limited resources. Performance of a Gemini-based clinical decision support system vs physicians for active surveillance and radical prostatectomy (NCCN-concordant treatment metrics). Treatment modality Metric CDSS (Gemini-based) Physicians Active surveillance (VA) Precision 1.00 1.00 Active surveillance (VA) Recall 1.00 0.24 Active surveillance (VA) F1-score 1.00 0.39 Active surveillance (VA) AUC 1.00 0.62 Radical prostatectomy (P) Precision 1.00 0.40 Radical prostatectomy (P) Recall 1.00 1.00 Radical prostatectomy (P) F1-score 1.00 0.58 Radical prostatectomy (P) AUC 1.00 0.90

Genomic profiling to define molecular drivers of colorectal neuroendocrine carcinoma subtypes.

3666 Background: Neuroendocrine carcinoma (NEC) is a rare and aggressive subtype of colorectal cancer (CRC), often arising in association with poorly differentiated adenocarcinoma. Two histologic subtypes, small cell NEC (SCNEC), and large cell NEC (LCNEC), have been described; however, how these subtypes differ, and whether they share molecular origins with the adenocarcinoma component in mixed adenocarcinoma and neuroendocrine carcinoma (MANEC) tumors, remain unclear. Better characterization of the genomic alterations driving NECs is needed to clarify disease biology and improve outcomes. Methods: We identified 125 patients with NEC of the colon and rectum who received care at Memorial Sloan Kettering Cancer Center. All patients underwent tumor profiling using the institutional targeted sequencing platform MSK-IMPACT. For 11 patients with MANEC, the adenocarcinoma and NEC components were sequenced separately. Results: Among 125 patients, 33 had SCNEC, 50 had LCNEC, and 42 had histologic features intermediate between SCNEC and LCNEC. 62% of patients presented with stage IV disease, and NEC was enriched at metastatic sites in MANEC cases. Clinical outcomes were poor across all groups, with a median survival of 11 months for stage IV patients (9.5m for SCNEC, 11m for LCNEC). SCNEC was more commonly observed in the rectum and genomically characterized by frequent RB1 loss (70%), with a smaller subset harboring PTEN loss (24%). Notably, TP53 mutations were identified in 58% of the cases, lower than the frequency observed in sporadic advanced CRC. In contrast, LCNEC was more often right sided, associated with an adenocarcinoma component and genomically enriched for MYC amplification (28%) and BRAF mutations (38%). We found that MYC amplification frequently occurred on extrachromosomal DNA, as confirmed by both FISH and shallow whole genome sequencing. Based on genomic alterations, 21of the 42 (50%) unclassified cases could be assigned a histologic subtype. A subset of NECs lacked known neuroendocrine associated genomic alterations and are being investigated further. In 11 MANEC cases with separately sequenced components, RB1 loss and MYC amplification were restricted to the neuroendocrine component, supporting a causal role in neuroendocrine differentiation and clonal evolution from adenocarcinoma. Across the cohort, PIK3CA (6%) and SMAD4 (9%) mutations were less frequent than expected and were absent in the NEC component of MANEC cases. DLL3 expression was evaluated in a small subset and was positive in some SCNECs but not in LCNEC. Conclusions: SCNEC and LCNEC represent clinically and genomically distinct subtypes of colorectal NEC. For NECs that cannot be classified based on histology alone, genomic profiling should be used to aid classification. Ongoing studies aim to validate our genomic findings above and to model the biology of NEC to identify more effective therapeutic strategies.

Identifying high-risk cytogenetic subgroups in multiple myeloma using large language artificial intelligence models.

7523 Background: The Revised International Staging System (R-ISS) is the clinical standard for risk stratification in multiple myeloma (MM) and is central to treatment selection and prognostication. However, the cytogenetic abnormalities on which the R-ISS is predicated are typically reported in unstructured formats within bone marrow biopsy reports and clinical notes. Despite the critical importance and relevance of obtaining cytogenetic abnormalities in MM, this information is often missing from large clinical databases, limiting their availability for large scale research and real-world analyses. Here, we created a large language model (LLM)-based algorithm to identify cytogenetic abnormalities from unstructured text and evaluated our model’s adaptability by comparing its performance across definitions of high-risk abnormalities. Methods: We extracted oncology notes and pathology reports within 2 years of diagnosis from the EHR of adult patients with MM registered within the Department of Veterans Affairs (VA) Cancer Registry System between 2000 and 2024. Two physician reviewers iteratively annotated 1600-character snippets from these texts for abnormalities and their type (gain, deletion, translocation) plus normal cytogenetics. From these annotations we identified items with high-risk cytogenetic abnormalities per current guidelines (del(17p) in &gt; 20% plasma cells; TP53 mutation; gain(1q) and/or del(1p32) plus t(4;14), t(14;16), or t(14;20); monoallelic del(1p32) plus gain(1q); biallelic del(1p32)) and previous guidelines (t(4;14), t(14;16), t(14;20), del TP53/17p, del(1p), gain(1q)). We used few-shot prompting with GPT-OSS 120B to extract cytogenetic results and validated our model on a test set of 985 snippets representing 773 patients at 98 VA centers. Results: The test set contained 81 snippets with normal cytogenetics and 464 with abnormalities; 58 were high-risk per current guidelines and 144 per previous guidelines. Our model successfully identified high-risk cytogenetics per current (P: 88.9%, R: 82.8%, F1: 85.7%) and previous guidelines (P: 86.7%, R: 90.3%, F1: 88.4%), as well as type of cytogenetic abnormality (weighted averages; P: 92.3%, R: 92.2%, F1: 92.4%), demonstrating its utility and flexibility for the task as standards change. Applied to a cohort of 7250 unlabeled notes, our model identified high-risk cytogenetics in 501 (6.9%) and any cytogenetic information in 5385 (74.3%). Conclusions: Our model demonstrated the successful capture of cytogenetic information across multiple definitions of high-risk cytogenetic abnormalities, providing a fast, adaptable framework for extraction that traditionally required repeated large-scale human efforts. Leveraging an LLM for this task enables scalable MM risk stratification while reducing clinician burden and supporting rapid adaptation to evolving cytogenetic risk definitions.

Predicting response to ¹⁷⁷ Lu-PSMA-617 in metastatic prostate cancer (mPC) using plasma-derived extracellular vesicles (EVs).

5068 Background: 177 Lu-PSMA-617 (LuPSMA) is a groundbreaking therapy for mPC, but biomarkers are urgently needed to improve patient selection. Plasma-derived EVs are emerging blood-based markers of tumor burden. Here, we analyzed levels of EVs of 123 patients (pts) from 4 cohorts: a pre- and during LuPSMA treatment for a contemporary prospective Mayo Clinic cohort, pre-treatment plasma levels of EVs for a historical Tulane cohort, localized prostate cancer (LPC) after radical prostatectomy (RP), and male healthy donors (HD). Methods: Blood samples were collected from 51 pts pre-cycles (pre-C) 1-3 of LuPSMA at Mayo Clinic from 12/2024-12/2025 and 26 pts pre-C1 at Tulane Cancer Center between 12/2018-1/2025. Small and large EVs were incubated with fluorescently conjugated antibodies (PSMA, STEAP1, CEACAM5, and TROP2) and enumerated using nanoscale flow cytometry. EVs from blood samples of 10 HD and 36 LPC post-RP from 6/2020-8/2024 at Mayo Clinic as controls. AUC-ROC and logistic regression analyses were performed to identify marker signatures associated with PSA50 response. Results: Baseline demographics are shown in the table. Median levels of PSMA, STEAP1, CEACAM5, and TROP2 EV/milliliter (mL) were all significantly higher for pts with mPC (pooled Tulane and Mayo cohorts) compared to LPC post-RP (p&lt;0.001) and HD (p&lt;0.001). In the Mayo cohort, 10 of 51 pts (20%) had undetectable PSA pre-C1. There were no significant differences in median pre-C1 plasma EV levels for these 4 different types of EVs comparing pts with undetectable vs detectable PSA. For 41 Mayo pts with detectable PSA pre-C1, 20 achieved PSA50 response at time of data cutoff (49%), while 7 pts (17%) had persistent PSA rise on LuPSMA. For pts who achieved PSA50, there was a significant decline in both CEACAM5 and TROP2 EVs/mL pre-C1 compared to pre-C3 (19 x 10 6 vs 5.8 x 10 6 , p=0.003 and 16 x 10 6 vs 5.5 x 10 6 , p&lt;0.001, respectively), while no significant changes occurred for the 7 pts with no PSA response. A composite score of pre-C1 CEACAM5 and TROP2 EVs (large- and small-sized EVs) demonstrated the strongest association with PSA50 response (OR=2.96, 95% CI: 1.27-6.44, p=0.038), achieving AUC of 0.73. PSMA and STEAP1 EVs did not show any significant changes from pre-C1 to pre-C3. Conclusions: Plasma-derived EVs are a promising biomarker to identify patients who will benefit from LuPSMA. Longer follow-up and enrolling a larger cohort are underway. Clinical trial information: NCT06200103 . Median values for pre-LuPSMA (Mayo and Tulane), post-RP for LPC, and HD. Mayo mPC (n=51) Tulane mPC (n=26) LPC (n=36) HD (n=10) Age (range) 73 (57-87) 71 (54-83) 66 (56-77) 55 (50-57) Prior lines (range) 1 (1-3) 4 (1-11) - - Gleason score 9 8 7 - PSA (range) 1.0 (0-182) 75 (0.8-503) - - PSMA EV/mL 3.7 x 10 6 14 x 10 6 2.0 x 10 6 1.4 x 10 6 STEAP1 EV/mL 30 x 10 6 45 x 10 6 12 x 10 6 12 x 10 6 CEACAM5 EV/mL 17 x 10 6 5.1 x 10 6 3.7 x 10 6 3.9 x 10 6 TROP2 EV/mL 13 x 10 6 5.2 x 10 6 4.5 x 10 6 3.6 x 10 6

Nivolumab + relatlimab (NIVO + RELA) in advanced melanoma: 5-year update of RELATIVITY-047.

9532 Background: NIVO + RELA improved progression-free survival (PFS) vs NIVO in patients (pts) with advanced melanoma in the randomized, double-blind, phase 3 RELATIVITY-047 trial (NCT03470922), leading to regulatory approval as a fixed-dose combination (FDC). The combination also improved clinically meaningful efficacy outcomes (objective response rate [ORR], overall survival [OS], melanoma-specific survival [MSS]) vs NIVO. Here, we report 5-year trial outcomes that characterize long-term response durability and landmark survival. Methods: 714 pts were randomized 1:1 to receive NIVO 480 mg + RELA 160 mg FDC or NIVO 480 mg intravenously every 4 weeks, as previously described. PFS per RECIST v1.1 (primary endpoint) was assessed by blinded independent central review (BICR); secondary endpoints included ORR per BICR and OS. MSS (censoring nonmelanoma-related deaths) was a post hoc analysis. Results: At data cutoff (25 Sept 2025), minimum potential follow-up was 57.3 months (mo; time from last pt randomization to clinical cutoff date). NIVO + RELA continued to show a clinically meaningful benefit vs NIVO for PFS, OS, MSS, and ORR (table). Improvements in absolute landmark survival rates were observed over 3, 4, and 5 years as follows: PFS, 4%, 7%, and 10%; OS, 7%, 9%, and 10%; MSS, 9%, 10%, and 13%. In addition, NIVO + RELA demonstrated consistent improvement in efficacy outcomes vs NIVO across the majority of key subgroups. Subsequent systemic therapy was received by 40% of pts treated with NIVO + RELA and 41% with NIVO. Secondary efficacy and biomarker analyses will be presented. Grade 3–4 treatment-related adverse events (TRAEs) occurred in 23% of pts treated with NIVO + RELA and 12% with NIVO; TRAEs (any grade) led to treatment discontinuation in 17% and 10% of pts, respectively. No new treatment-related deaths were reported since the 2-year analysis. Conclusions: With a follow-up of 5 years, first-line NIVO + RELA continued to demonstrate durable, clinically meaningful long-term improvements in PFS, OS, ORR, and MSS vs NIVO, with median MSS not reached for the combination. Across landmark timepoints, absolute separations in PFS, OS, and MSS rates were maintained and increased at the 5-year follow-up. Safety remained consistent with prior reports, with no new or unexpected safety signals. Clinical trial information: NCT03470922 . NIVO + RELA (n = 355) NIVO (n = 359) HR/ORR difference (95% CI) Median PFS, mo (95% CI) 10.2 (6.5–15.7) 4.6 (3.5–6.5) 0.77 (0.65–0.92) 60 mo PFS rate, % (95% CI) 29 (24–35) 19 (15–24) — Median OS, mo (95% CI) 54.8 (33.6–81.1) 33.2 (25.2–44.7) 0.78 (0.64–0.95) 60 mo OS rate, % (95% CI) 48 (43–53) 38 (33–43) — Median MSS, mo (95% CI) NE (NE–NE) 47 (34–65) 0.71 (0.57–0.88) 60 mo MSS rate, % (95% CI) 58 (52–63) 45 (40–51) — ORR, % (95% CI) 44 (39–49) 34 (29–39) 10 (3–17) Descriptive analyses; minimum follow-up: 57.3 mo. HR, hazard ratio; NE, not estimable.

T-cell receptor (TCR) repertoire dynamics of response to neoadjuvant chemo-immunotherapy in triple- negative breast cancer (TNBC).

e12660 Background: Neoadjuvant pembrolizumab plus chemotherapy is the current standard for cT2N0-3 TNBC, yet ~40% of patients (pts) do not achieve a pathological complete response (pCR). This increased efficacy of the addition of pembrolizumab comes at the expense of significant immune related and financial toxicities. To better understand drivers of pCR, we analyzed longitudinal TCR dynamics in pCR patients versus those with residual disease (non-pCR). Methods: Blood samples from TNBC pts were collected at baseline (BL), on-treatment (DT), and post-surgery (PS). Peripheral blood mononuclear cells were isolated by density gradient centrifugation. Single-cell RNA sequencing was performed (Parse Biosciences platform). Paired α/β TCR repertoires were analyzed for diversity, overlap, and homeostasis. Longitudinal tracking identified persistent clonotypes across timepoints, with expansion quantified as log₂ fold changes. Transcriptional states of persistent and convergent clones were characterized using GSEA (MSigDB Hallmark and a targeted panel of T-cell signaling pathways [C2/GO:BP]) and differential expression analyses, comparing pCR vs. non-pCR pts and longitudinal clonal dynamics. Results: Among 18 pts, 11 achieved pCR. A total of 21,622 TCR clones were analyzed. pCR pts had 2,682 unique clones at BL, 3,834 at DT, and 5,274 at PS; 78 clones were present at both BL and DT, and 68 were detected across all timepoints. Non pCR pts had 1,864 clones at BL, 2,997 at DT, and 4,538 at PS; 27 clones were present at both BL and DT, and 49 persisted across all timepoints. Most clones were not shared between pCR and non pCR pts. While non pCR CD8+ T cells showed greater numeric expansion (137% vs 76%), their repertoires exhibited a marked loss of large clonotypes (20.2% to 7.2%) compared with pCR pts (24.5% to 17.6%). Persistent pCR clones demonstrated stronger V gene correlations, and their corresponding T cells showed upregulation of NFκB pathways. None of the clonotypes persisting in either pCR or non pCR pts had been previously associated with breast cancer in the literature. Convergent clones were most frequent in CD8+ T cells (6.1%) and exhibited higher degeneracy, particularly at the D2 level, which was increased in pCR patients. These clones showed upregulation of TCR signaling pathways, MMP14 , and AGMAT (markers of tissue homing and proliferation), whereas non pCR pts showed downregulation of NFκB signaling. Conclusions: pCR may be driven by the stability and functional quality of the TCR repertoire rather than by expansion alone. Non pCR was characterized by repertoire instability and loss of inflammatory signaling. In contrast, pCR was associated with the maintenance of robust and persistent clonotypes exhibiting invasive and proliferative phenotypes. Peripheral clonal stability may represent a more informative biomarker of treatment efficacy than simple diversity metrics.

Tirzepatide-induced lipid metabolic remodeling in relation to clonogenicity and migration in colorectal cancer cells: An <i>in vitro</i> and <i>in silico</i> approach.

e15561 Background: Glucagon-like peptide-1 (GLP-1)–based therapeutics have attracted increasing attention beyond metabolic disorders; however, their cellular mechanisms in colorectal cancer (CRC) remain to be explored, particularly the involvement of alternative metabolic pathways. To our knowledge, this study provides the first comparative analysis of tirzepatide-induced lipid metabolic remodeling in matched primary and metastatic CRC cell lines. Methods: The effects of tirzepatide treatment (1–500 nM) were evaluated in primary (SW480) and metastatic (SW620) CRC lines using lipid droplet (LD) staining, wound healing, and colony formation assays. In addition to the in vitro experiments, molecular docking analyses were performed to find the interactions of tirzepatide with targets involved in lipid metabolism. Molecular docking was performed using the HADDOCK 2.4 web server, with the human GLP-1 receptor (GLP-1R) included as a positive control. Publicly available data from the Human Protein Atlas were analyzed to examine the expression of GLP-1/GIP receptors and fatty acid–metabolism–associated genes ACLY and FASN. Results: LD staining indicated a decrease in lipid droplet abundance in both cell lines. Colony formation assay revealed that SW480 cells exhibited enhanced colonogenic capacity in response to tirzepatid treatment, while SW620 cells showed no significant changes. Wound-healing assay indicated that only SW620 cells exhibited enhanced migratory capacity under high-dose tirzepatid treatment at 48h. Molecular docking demonstrated that tirzepatide binds lipid metabolism–related enzymes with favorable affinity, showing strong interactions with FASN (−76.1) and ACLY (−58.2), alongside comparable binding to the GLP-1 receptor (−66.6), predominantly through hydrogen-bond–mediated interactions. The Human Protein Atlas revealed negligible expression of GLP-1 and GIP receptors in both cell lines, suggesting receptor-independent mechanisms, while FASN and ACLY showed marked expression level in both cell lines. Conclusions: Our results showed tirzepatide’s potential in the modulation of lipid metabolism, as also supported by in silico analyses. The divergent clonogenic responses, reduced LD activity, and distinct migratory behaviors observed between SW480 and SW620 cells appear to arise from differential metabolic programming rather than direct GLP-1R/GIPR–dependent signaling, highlighting a potential receptor-independent mechanism underlying tirzepatide’s cellular effects.

Bispecific antibodies in the treatment of relapsed refractory multiple myeloma: A meta analysis.

e19529 Background: Relapsed/refractory multiple myeloma (RRMM) includes relapse after response orprogression within 60 days of the last regimen, and many patients become triple‐class refractory(PI, IMiD, and anti‐CD38 refractory) with historical median overall survival of about 1 year.Bispecific antibodies (bsAbs) targeting B-Cell Maturation Antigen (BCMA) (Teclistamab,Elranatamab), GPRC5D (Talquetamab), or FcRH5 (Cevostamab) have shown substantialactivity even in heavily pretreated, post‐BCMA patients. However, survival comparisons acrossindividual bsAbs are limited by heterogeneous populations, prior BCMA exposure, and evolvingsequencing (e.g., bsAbs as CAR‐T bridging). This meta-analysis therefore aims to provide directand indirect comparative data on the efficacy and safety of bsAbs in RRMM. Methods: We manually searched major medical databases and identified 28 studies for pooledsingle‐arm analyses of bsAbs (5 Cevostamab, 5 Teclistamab, 5 Elranatamab, 10 Talquetamab),plus 7 additional studies for indirect bsAbs versus standard‐of‐care comparisons and forest plotswere created using R software. Results: The pooled objective response rate (ORR) was 0.72 (95% CI 0.68–0.75) forTalquetamab, 0.55 (0.39–0.70) for Elranatamab, 0.60 (0.56–0.65) for Teclistamab, and 0.49(0.24–0.74) for Cevostamab. The pooled complete response rate (≥CR) was 0.33 (0.27–0.40)for Talquetamab, 0.32 (0.19–0.49) for Elranatamab, 0.23 (0.13–0.37) for Teclistamab, and 0.14(0.03–0.43) for Cevostamab. The pooled incidence of cytokine release syndrome (CRS) was0.66 (0.57–0.73) for Talquetamab, 0.52 (0.32–0.71) for Elranatamab, 0.57 (0.50–0.64) forTeclistamab, and 0.69 (0.48–0.84) for Cevostamab. The pooled incidence of ICANS was 0.10(0.07–0.14) for Talquetamab, 0.11 (0.01–0.58) for Elranatamab, 0.10 (0.05–0.17) forTeclistamab, and 0.18 (0.09–0.35) for Cevostamab. Elranatamab was associated with asignificantly higher ORR versus (SOC) (OR 3.84; 1.85–7.96), whereas Talquetamab andTeclistamab showed numerically higher but not statistically significant ORR compared with SOC(OR 5.71; 0.66–49.34) and (OR 4.12; 0.64–26.74) respectively. Conclusions: Talquetamab has the highest pooled ORR and ≥CR among the bispecifics, thoughall four show meaningful activity with similar CRS and ICANS rates. Elranatamab is the onlyagent with a statistically superior ORR versus standard of care; Talquetamab and Teclistamabconfer only numerically higher, non‐significant ORR improvements.