Browse Articles

Real-world treatment patterns and clinical outcomes among US patients with non-muscle invasive bladder cancer (NMIBC) during the BCG shortage.

4612 Background: Bladder cancer is diagnosed in 80,000 US patients (pts) annually; 70% present as non-muscle invasive (NMIBC). Guidelines are risk-based; treatment for high risk (hr) disease includes tumor resection (TURBT) and intravesical therapy, commonly Bacille Calmette Guérin (BCG). Delivery of indicated treatment remains challenged, especially by ongoing (acutely since 2017) BCG shortages. We describe recent real-world (rw) patient characteristics, treatments (BCG, TURBT use), and clinical outcomes for NMIBC pts by risk group. Methods: We analyzed demographic, clinical and treatment data from electronic health records in N-Power Medicine's Real-World Analytical Dataset, largely based in the Midwest. Adults diagnosed with NIMBC between 1/2017 - 6/2022 were included and followed through 2022. Pts were classified as hr (Y/N) by AUA definition. Kaplan-Meier (KM) curves estimated BCG maintenance (MTX) duration, rw MIBC and metastatic free survival (modified PFS), cystectomy free survival (CFS), overall survival (OS), and recurrence free survival (RFS). Logistic and Cox regression was used to identify predictors of MTX start and discontinuation among hr pts including demographics, diagnosis year, smoking status, Charlson comorbidity index (CCI) and physician specialty. Results: Among 3,516 NMIBC pts (median age at diagnosis 72 years), 2,634 (75%) had hrNMIBC; most were non-Hispanic White (88%) and male (75%). Over a median follow-up of 2.7 years, 29% of hrNMIBC pts received BCG induction (IDX); 12% received BCG MTX and median MTX duration was 7.7 months (95% confidence interval [CI]: 6.8-9.2). For hr pts who received IDX, those diagnosed in 2019-2020 were less likely to initiate MTX (2019 vs 2017: OR 0.4; 95% CI [0.2-0.7], 2020 vs. 2017: OR 0.6 CI [0.4-1.0]). Hr pts with higher CCI (>2 vs 0: HR 2.2 [1.2-4.2]) were more likely to discontinue MTX. Most (95%) BCG recipients received no other therapy after BCG. Nearly all pts (>99%) received TURBT. Clinical outcomes (rwmPFS, rwCFS, rwOS) were less favorable for hr pts (Table). Conclusions: We examined treatment during the BCG shortage. BCG IDX and MTX were low among hrNMIBC pts. Clinical outcomes during the shortage highlight the potential to improve outcomes, especially for hr pts, and the need for alternative treatment strategies. KM percent survival (95% CI) of rwmPFS 1 , rwCFS 2 , and rwOS 3 by risk category. Follow up Month rwmPFS rwCFS rwOS Non-Hr Hr Non-Hr Hr Non-Hr Hr 12 97 (96, 98) 90 (89, 92) 97 (95, 98) 91 (90, 92) 90 (78, 100) 72 (69, 75) 24 92 (90, 94) 85 (83, 86) 92 (90, 94) 85 (83, 86) 85 (70, 100) 64 (60, 68) 36 89 (86, 91) 79 (77, 80) 89 (86, 91) 78 (76, 80) 85 (70, 100) 57 (53, 62) 48 87 (84, 90) 75 (73, 77) 87 (84, 90) 74 (72, 76) 75 (56, 100) 53 (48, 58) 60 81 (77, 85) 70 (68, 73) 81 (77, 85) 70 (67, 73) — (—, —) 47 (42, 53) 1 Real-world modified progression free survival. 2 Real-world cystectomy-free survival. 3 Real-world overall survival.

Effects of RNA-based NGS on false-positive fusion calls in colorectal cancer compared with DNA-NGS.

e15575 Background: Gene fusions act as oncogenic drivers across multiple solid tumors and provide opportunities for matched targeted therapies. DNA-based next-generation sequencing (DNA-NGS) is widely used to profile actionable alterations in solid tumors. However, in non-small cell lung cancer and cholangiocarcinoma, RNA-based NGS (RNA-NGS) has been reported to capture additional fusion events relative to DNA-NGS. Although colorectal cancer (CRC) generally shows a lower fusion prevalence, most existing CRC reports are DNA-NGS–based, and whether RNA-NGS can detect a broader and more clinically relevant fusion landscape in CRC remains uncertain. Methods: We included 857 CRC cases tested by DNA-NGS and 1,280 CRC cases tested by RNA-NGS. Fusion prevalence was summarized with 95% confidence intervals (CIs). Between-group differences were assessed using two-sided Fisher’s exact tests, and effect sizes were reported as risk difference (RD), risk ratio (RR), and odds ratio (OR) with 95% CIs. Results: Using DNA-NGS, 13 fusion calls were identified among 857 CRC cases (1.52%, 95% CI 0.89%–2.58%), including FGFR1 (n = 4), FGFR2 (n = 3), ALK (n = 2), and MET/NTRK3/RET/FGFR3 (n = 1 each). RNA-NGS detected 14 fusions among 1,280 CRC cases (1.09%, 95% CI 0.65%–1.83%), with a distinct spectrum: MET (n = 4), RET (n = 3), ROS1 (n = 2), NTRK1 (n = 2), ALK (n = 2), and FGFR2 (n = 1). When comparing overall fusion prevalence (DNA-NGS 13/857 vs RNA-NGS 14/1,280), the difference was not statistically significant (RD = 0.42%, 95% CI −0.54% to 1.57%; RR = 1.39, 95% CI 0.66–2.94; OR = 1.39, 95% CI 0.65–2.98; Fisher’s exact P = 0.432). Importantly, DNA-NGS reported three rearrangements that did not include the kinase domain of the fusion genes (FGFR1, n = 2; FGFR3, n = 1), which may represent DNA-level rearrangements without functional fusion transcripts (i.e., likely false positives). These accounted for 23.1% (3/13; 95% CI 8.2%–50.3%) of DNA-NGS fusion calls. After excluding these three events, the prevalence of canonical/likely functional fusions by DNA-NGS was 10/857 (1.16%, 95% CI 0.64%–2.13%). Compared with RNA-NGS (14/1,280; 1.09%, 95% CI 0.65%–1.83%), canonical fusion prevalence remained similar (RD = 0.07%, 95% CI −0.83% to 1.14%; RR = 1.07, 95% CI 0.48–2.39; OR = 1.07, 95% CI 0.47–2.41; Fisher’s exact P = 1.000). Conclusions: In CRC, RNA-NGS demonstrates a fusion spectrum distinct from DNA-NGS and may reduce false-positive fusion calls arising from non-functional DNA-level rearrangements (e.g., events lacking kinase domains). While overall fusion prevalence was comparable between platforms, RNA-NGS may provide higher specificity for clinically meaningful fusion detection.

Durable efficacy of first-line necitumumab plus pembrolizumab in PD-L1–high advanced NSCLC: Final results of the phase II K-TAIL-202 study.

8573 Background: Pembrolizumab is a standard 1st line treatment for advanced non–small cell lung cancer (NSCLC) with high PD-L1 expression (tumor proportion score [TPS] ≥50%); however, durable benefit is achieved in only a subset of patients. Prior pembrolizumab-based combination strategies (e.g., anti-TIGIT) have inconsistently improved outcomes over pembrolizumab monotherapy in this setting. Epidermal growth factor receptor (EGFR) signaling may promote immune evasion by stabilizing PD-L1 expression and shaping an immunosuppressive tumor microenvironment, providing a rationale for combining the anti-EGFR antibody necitumumab with pembrolizumab. We therefore evaluated the efficacy and safety of necitumumab plus pembrolizumab in treatment-naive patients with PD-L1–high advanced NSCLC. Methods: K-TAIL-202 (jRCT2031200248) was an open-label, multicenter, single-arm phase II study conducted in Japan. Eligible patients were aged ≥20 years with unresectable stage III/IV or recurrent NSCLC, PD-L1 TPS ≥50% (22C3), ECOG PS 0–1, and ≥1 measurable lesion. Patients received necitumumab 800 mg IV on days 1 and 8 plus pembrolizumab 200 mg IV on day 1 every 3 weeks for up to 35 cycles or until disease progression or unacceptable toxicity. The primary endpoint was investigator-assessed objective response rate (ORR) per RECIST v1.1; secondary endpoints included progression-free survival (PFS), overall survival (OS), and safety. The study assumed an expected ORR of 54.8% and a threshold ORR of 39.0% (one-sided α = 0.10; 80% power), requiring enrollment of 50 patients. Results: Between December 2020 and March 2023, 50 patients were enrolled (median age, 72 years). The ORR was 68.0% (95% CI, 53.3–80.5) and the disease control rate was 78.0% (95% CI, 64.0–88.5). Complete response, partial response, stable disease, progressive disease, and not evaluable status were observed in 2.0%, 66.0%, 10.0%, 12.0%, and 10.0% of patients, respectively, meeting the primary endpoint. Median PFS was 16.0 months (95% CI, 9.0–24.0), with a 24-month PFS rate of 35.9%. Median OS was not attained (95% CI, 38.0 months–not estimable), and the 24-month OS rate was 71.3%. The most treatment-emergent adverse events included acneiform dermatitis (66.0%), hypomagnesemia (60.0%), and pneumonitis (12.0%). One grade 5 cardiac arrest occurred in 1 patient (2.0%), for which a causal relationship to study treatment could not be ruled out. Conclusions: At final analysis, first-line necitumumab plus pembrolizumab demonstrated durable antitumor activity and encouraging long-term survival as a chemotherapy-free regimen in patients with PD-L1–high advanced NSCLC. Toxicities were consistent with the known profiles of EGFR-targeted antibodies and PD-1 inhibitor therapy, underscoring the need for careful monitoring. These findings warrant further confirmation in a randomized phase III clinical trial. Clinical trial information: jRCT2031200248.

Light-activated EGFR antibody–drug conjugate: Analysis of selectivity and translational profile for first-in-human evaluation.

3023 Background: Conventional antibody–drug conjugates (ADCs) are limited by systemic toxicity and narrow therapeutic windows. Cetuximab-I21-2 is a first-in-class, EGFR-targeted ADC conjugated to a Heloporfin-derived photosensitizer engineered for activation exclusively at the tumor site using red-light irradiation. This approach enables biologically targeted delivery with externally controlled cytotoxic activation, designed to improve the therapeutic index and support safe translation into first-in-human studies. Methods: In vitro cytotoxicity was evaluated across EGFR-high, intermediate, and low tumor cell lines (A431, HCC827, Huh7, NCI-N87, MIA-PaCa-2) with or without red-light activation (1–5 J/cm²). Cellular ROS generation, mitochondrial membrane potential disruption, and photodynamic injury were assessed by live-cell imaging. In vivo activity was tested in A431 xenograft-bearing B-NDG mice following a single intravenous dose of Cetuximab-I21-2 (35.3 mg/kg) and localized tumor irradiation at 100 or 200 J. Safety evaluation included body-weight trends and clinical observations. Results: Cetuximab-I21-2 showed negligible dark toxicity up to several µg/mL, confirming that the payload remains inactive without illumination. Upon irradiation, the ADC demonstrated potent, energy-dependent cytotoxicity (IC50: 0.05 µg/mL in A431 and 0.14 µg/mL in HCC827 at 5 J/cm²), with reduced effects in EGFR-low models, supporting a target- and activation-dependent mechanism. Illumination triggered rapid singlet-oxygen and ROS production and immediate mitochondrial depolarization. In vivo, Cetuximab-I21-2 achieved irradiation-dependent tumor suppression, reaching 22.2% inhibition at 200 J, and produced histologic photodynamic injury localized to the tumor. Treatment was well tolerated: animals exhibited only transient (<10%) weight loss and no systemic toxicity. Conclusions: Cetuximab-I21-2 demonstrates a strong translational profile with selective activation, robust preclinical antitumor activity, and a favorable safety margin. The spatial control offered by light activation may enable dose-sparing strategies and reduce systemic adverse events typically associated with ADCs. These findings support advancement toward IND-enabling studies, including optimization of clinical illumination parameters, device–drug integration, and early feasibility planning for first-in-human evaluation in EGFR-expressing solid tumors.

H&E–based MSI/MMR testing with artificial intelligence in colorectal cancer: A blinded evaluation across diverse clinicopathological subgroups.

3520 Background: Mismatch repair (MMR) deficiency and microsatellite instability (MSI) testing is integral to colorectal cancer management. Despite guideline recommendations, routine testing remains limited by cost, turnaround time, and pathology resources. PANProfiler Colorectal (PPC) is an artificial intelligence (AI) system that infers MSI/MMR status directly from H&E-stained whole slide images (WSIs), offering a rapid, resource-efficient alternative. This study reports blinded validation of PPC, specifically evaluating performance across clinically relevant subgroups. Methods: A total of 2,636 WSIs from St. James’s University Hospital (Leeds, UK) were retrospectively analyzed, with the dataset representing a broad range of clinicopathological characteristics including age ( < 50 to > 75 years), gender, tumor stage, differentiation, and histological subtypes. PPC returned proficient MMR ( pMMR ), deficient MMR ( dMMR ) or indeterminate (no definitive result), with results evaluated for agreement against immunohistochemistry. Results: PPC generated definitive results for 2,109 WSIs, achieving a test replacement rate (TRR) of 80%. Overall positive percent agreement (PPA) was 97.9% and negative percent agreement (NPA) 93.8% (Table). Performance was consistent across all age groups and genders. PPA and NPA were mostly robust in all cancer stages, with PPA exceeding 99% in Stage II disease, where MMR status primarily informs adjuvant therapy decisions. In Stage III/IV disease, PPA reduced to 82.6%, which may reflect greater histological heterogeneity in advanced tumors. In mucinous and poorly differentiated tumors, PPA remained high (98.7-100%), supporting reliable detection of dMMR samples, while NPA ranged from 75.4-80%, indicating more conservative classification of pMMR tumors that would require lab tests. Conclusions: PPC demonstrated a high overall performance, providing definitive results for the majority of cases. PPA and NPA were consistent in key clinical groups, including younger patients and early stage disease. In more challenging cases like mucinous and poorly differentiated tumors, PPA remained robust despite reductions in NPA, supporting safe guidance of subsequent testing. These findings suggest PPC could streamline MSI/MMR workflows, reducing reliance on conventional testing while maintaining diagnostic safety across diverse clinicopathological profiles. Performance across diverse subgroups. Overall Age <50 Age 50-65 Age 65-75 Age >75 Male Female Stage I Stage II Stage III/IV Well Diff. Moderately Diff. Poorly Diff. Adenocarcinoma Mucinous Definitive (n) 2109 155 159 698 703 1165 944 395 1015 694 245 1655 187 1879 210 PPA (%) 97.9 100 100 92.1 100 95.5 99.4 97.8 99.5 82.6 100 96 100 99.5 98.7 NPA (%) 93.8 91.7 98 91.3 94.1 95.5 91.5 95.1 91.8 95.5 92.5 95.6 80 95.3 75.4 TRR (%) 80 74.5 76.4 81.2 82.2 81.1 78.7 74.5 81 81.8 76.1 80.8 80.3 82.4 63.6

Clinical characteristics and survival outcomes of a large cohort of patients diagnosed with diffuse large B-cell lymphoma: Data from the GOTEL Lymphoma Registry.

e19070 Background: Diffuse large B-cell lymphoma (DLBCL) is the most common subtype of non-Hodgkin lymphoma. It is a heterogeneous and aggressive disease, whose survival has improved significantly since the introduction of immunochemotherapy, achieving cure rates of up to 60–70%. Currently, there is limited real-world evidence in the literature regarding lymphoid neoplasms, and more specifically DLBCL, due to the scarcity of population-based registries. The aim of this study was to describe demographic and clinical characteristics as well as survival outcomes of patients diagnosed with DLBCL treated at Spanish centers. Methods: A retrospective, multicenter study was conducted including patients with DLBCL diagnosed at oncology departments of GOTEL-affiliated centers and registered in the national lymphoid neoplasm registry between 2000 and 2023. Epidemiological and clinical data were collected and analyzed descriptively. Survival probabilities were estimated using the Kaplan–Meier method with 95% confidence intervals. Results: A total of 504 patients were analyzed. Demographic and clinical characteristics are summarized in Table 1. After a median follow-up of 6.5 years, 45.4% of patients were alive and disease-free, 7.7% had active disease, and 41.3% had died. Most patients (94.6%) received first-line immunochemotherapy, with 23.8% developing refractory or relapsed disease. Median relapse-free survival was not reached, and median overall survival (OS) was 10.2 years (95% CI, 8.30–12.46). The 1-, 5-, and 10-year OS rates were 82.9%, 63.3%, and 51.1%, respectively. The leading cause of death was primary disease (54.3%), followed by infectious diseases (13.1%) and secondary malignancies (8.2%). Conclusions: Our study describes the main clinical characteristics and survival outcomes of a large cohort of Spanish patients with DLBCL treated across multiple centers, representing one of the population-based registries with the longest follow-up reported to date. Such studies are essential to characterize real-world populations and to develop specific treatment protocols, as well as follow-up strategies for long-term survivors. Patients with DLBCL (n = 504) Sex Male: 54.0% Female: 46.0% Median age 65 years (17-98) Stage I:16.9% II: 24.8%III: 18.7%IV: 35.5%Unknown: 4.1% ECOG 0: 31.2% 1: 41.1%2: 15.3%3: 6.5%4: 1.2% Unknown: 4.8% IPI 0-1: 29.0% 2-3: 39.7% 4-5: 18.5%Unknown: 12.9% Primary extranodal: 180 (35.7%) Spleen: 1.4%Gastric: 6.5%Liver: 1.4%Intestinal: 3.6%Head and neck: 4.8%Bone: 5.2%Lung: 3.2%Testicular: 2.0%Unknown: 1.0%Others: 5.2% Bulky mass 38.1% LDH increased 51.2% Causes of death (n = 208) Primary tumor: 54.3%Second malignancy: 8.2%Other: Neurological: 5.0% Cardiovascular: 5.0% Infectious: 13.1% Respiratory: 1.5% Myelodysplastic syndrome: 0.5% Unknown: 12.0%

Genomic alterations in metastatic renal cell carcinoma (mRCC): Impact on survival and response to front-line systemic therapy.

e16556 Background: Genomic alterations in RCC can serve as biomarkers for response to therapy. Wild type VHL has shown inferior response to VEGF inhibitors, while patients (pts) with mutated VHL have improved outcomes with immunotherapy (IO)/tyrosine kinase inhibitor (TKI) therapy. We describe the impact of genomic alterations on outcomes and response to front-line therapy. Methods: Pts with mRCC with genomic sequencing between 2015-2025 at Indiana University and treated with front-line systemic therapy were included. Kaplan-Meier method was used to analyze progression free survival (mPFS) and overall survival (OS) using the log rank test to compare groups. Results: 198 pts were included. Median age was 58.6 (range, 24.2-83.7). Tumor histology was 68.7% clear cell, 9.1% papillary, 8.1% unclassified, 4% chromophobe, 2.5% translocation, 1% poorly differentiated, and 6.6% other. 17.2% of pts had sarcomatoid features; 14.1% had rhabdoid. 45.5% pts had metastasis at diagnosis and 54% at relapse. Metastasis sites were lungs 57.1%, regional lymph nodes (LNs) 36.9%, bone 34.3%, distant LNs 27.8%, liver 19.7%, and brain 7.6%. IMDC risk was 22.2% good, 44.4% intermediate, 16.7% poor, and 16.7% unknown. Front-line therapy was IO/IO in 24.7%, IO/TKI in 30.3%, single-agent TKI in 22.7%, single-agent IO in 5.6%, and other in 16.7%. Genetic alterations included VHL in 53.5%, PBRM1 in 32.3%, SETD2 in 25.3%, TP53 in 14.6%, and BAP1 in 13.1%. Median follow-up was 5.0yrs (range 0.03-21.9). Overall mPFS was 0.8yrs (95%CI 0.6-1.0). 5-yr OS was 41.6% (95%CI: 33.3-49.8). mPFS was 1.3yrs (95% CI 0.8-2.1) for those treated with IO/TKI vs. 0.7yrs (95% CI: 0.4-0.9) for IO/IO (p=0.0050), though there was no difference in OS. For pts with a VHL mutation, mPFS was 1.0yrs (95% CI; 0.8-1.4) vs. 0.5yrs (0.3-0.6) for pts without (p=0.01). 5yr OS for those with a VHL mutation was 48.1% (95% CI 35.8-59.4) vs. 35.1% (24.1-46.2) for pts without (p=0.002). For pts with a VHL mutation, mPFS was 0.8yrs (95% CI 0.4-1.8) for pts treated with IO/IO vs. 1.6yrs (95% CI 0.9-4.9) for IO/TKI (p=0.05). There was no difference in OS. Table 1 outlines outcomes by genomic mutation and therapy received. Conclusions: VHL mutations were associated with improved mPFS and OS. For most mutations, IO/TKI resulted in improvements in PFS, though no difference in OS. Median PFS (yrs) 5-yr OS (%) Mutation Present Mutation Absent p-value Mutation Present Mutation Absent p-value Genomic Mutation:TP53VHLPBRM1SETD2BAP1TERT 0.91.01.20.90.50.4 0.80.50.60.70.80.8 0.45 0.01 0.01 0.320.050.10 30.148.147.235.637.721.4 44.235.138.843.842.344.5 0.41 0.002 0.080.850.690.37 Present mutation: 1L therapy IO/IO 1L therapy IO/TKI p-value 1L therapy IO/IO 1L therapy IO/TKI p-value TP53VHLPBRM1SETD2BAP1TERT 0.40.81.20.80.30.3 1.11.61.61.91.70.5 0.03 0.05 0.04 0.03 0.04 0.20 20.041.941.526.143.80 25.953.875.438.1033.3 0.930.160.050.270.800.50

Neoadjuvant and adjuvant (neoadj-adj) enfortumab vedotin (EV) plus pembrolizumab (pembro) in participants with cisplatin-ineligible muscle-invasive bladder cancer (MIBC): An analysis of clinically relevant subgroups in KEYNOTE-905.

4613 Background: Results from the phase 3 KEYNOTE-905/EV-303 study (NCT03924895) showed that neoadj-adj EV + pembro and radical cystectomy with pelvic lymph node dissection (RC + PLND) led to superior event-free survival (EFS; HR 0.40; 95% CI 0.28-0.57), overall survival (OS; HR 0.50; 95% CI 0.33-0.74), and pathological complete response (pCR) rate (57.1% vs 8.6%) vs RC + PLND alone in participants (pts) with MIBC who were ineligible for or declined cisplatin. We report efficacy in subgroups of clinical interest, including advanced age, ECOG PS, cisplatin eligibility status, and baseline clinical tumor stage. Methods: Adults with stage T2-T4aN0M0 or T1-T4aN1M0 MIBC by central review who were cisplatin-ineligible or declined cisplatin were randomly assigned 1:1 to neoadj-adj EV + pembro and RC + PLND vs RC + PLND alone (control). EFS by BICR, OS, and pCR (pT0N0) rate by central review were analyzed in subgroups by age (≥65 to <75 vs ≥75 yr), ECOG PS (0-1 vs 2), cisplatin-ineligibility status (ineligible vs declined), and clinical stage (T2N0 vs T3-T4aN0 vs T1-4aN1). The analyses were exploratory; no formal statistical testing was performed. Results: 170 pts were assigned to EV + pembro and 174 to control. Median follow-up at data cutoff (Jun 6, 2025) was 25.6 mo (range 11.8-53.7). 30 pts in the EV + pembro arm and 32 in the control arm had stage T2N0 MIBC; 133 and 132 pts had stage T3-T4aN0; 7 and 10 had stage T1-4aN1. In pts with T2N0 MIBC, median EFS was not reached (NR) with EV + pembro vs NR with control (HR 0.26; 95% CI 0.08-0.80), and pCR rate was 60.0% vs 18.8%. In pts with T3-T4aN0 MIBC, median EFS was NR with EV + pembro vs 17.2 mo with control (HR 0.43; 95% CI 0.29-0.63); median OS was NR vs 41.7 mo (HR 0.54; 95% CI 0.35-0.82); pCR rate was 57.1% vs 6.8%. In pts with T1-4aN1 MIBC, median EFS was 36.7 vs 7.2 mo (HR 0.35; 95% CI 0.09-1.40); pCR rate was 42.9% vs 0%. OS for T2N0 and T1-4aN1 will be assessed after additional follow-up (currently <10 events). Outcomes by age, ECOG PS, and cisplatin eligibility are shown in the Table. Conclusions: An efficacy benefit was observed with neoadj-adj EV + pembro added to RC + PLND vs RC + PLND alone across clinically relevant subgroups, consistent with the ITT population. Results further support EV + pembro as a potential new standard treatment for pts with cisplatin-ineligible MIBC. Clinical trial information: NCT03924895 . Age ≥65 to <75 yr Age ≥75 yr ECOG 0-1 ECOG 2 Cisplatin-ineligible EV+P C EV+P C EV+P C EV+P C EV+P C N 63 77 78 68 149 148 21 26 142 139 EFS, median, mo NR 13.1 NR 18.8 NR 17.2 NR 9.3 NR 14.2 EFS HR(95% CI) 0.41(0.24-0.70) 0.50(0.30-0.84) 0.44(0.30-0.64) 0.19(0.07-0.52) 0.37(0.26-0.54) OS, median, mo NR 41.7 NR 37.4 NR 42.6 NR 19.2 NR 37.4 OS HR(95% CI) 0.49(0.26-0.94) 0.63(0.36-1.10) 0.59(0.39-0.90) 0.17(0.05-0.59) 0.46(0.30-0.70) pCR rate, % 54.0 9.1 53.8 7.4 53.7 8.8 81.0 7.7 54.9 8.6 C, control; EV+P, EV + pembro.

Clinical outcomes of solid basaloid and classic adenoid cystic carcinoma of the breast.

e12746 Background: Adenoid cystic carcinoma (AdCC) of the breast is a rare ( < 0.1%) form of breast cancer with two variants, solid basaloid (SB-AdCC) and classic (C-AdCC). Due to its rarity, limited outcome data exists. We aimed to describe clinicopathological features and long-term outcomes in patients with SB-AdCC and C-AdCC. Methods: Patients diagnosed with stage I-III AdCC between 2007-2024 were retrospectively identified from an institutional database. Patient characteristics were compared between groups using Fisher’s exact test and Wilcoxon rank sum test. Recurrence free survival (RFS) was estimated using the Kaplan-Meier method and compared between groups using log-rank test. Results: 77 patients were included. The median age was 58 (IQR 48,69). Most patients self-identified as White (n = 64, 86%) and not Hispanic (n = 65, 89%). Most patients (n = 55, 71%) had SB-AdCC histology and 22 patients (29%) had C-AdCC. Median tumor size was 17 mm (IQR 13,26), with no significant size difference between SB-AdCC and C-AdCC (p = 0.5). Most patients (n =44, 57%) presented with clinical stage I disease and 34 (50%) patients had grade 3 disease, with a higher proportion of high-grade disease in patients with solid basaloid subtype (p < 0.001). Eight patients (10%) were clinically node-positive at presentation. Nearly all tumors were ER-/HER2- (88%, n = 68). Most patients underwent breast conserving therapy (70%, n = 52). Forty percent (n = 31) of patients received systemic treatment. Patients with Sb-AdCC were more likely to receive systemic therapy, 53% vs 9% (p<0.001). Among patients who received systemic therapy (n = 31), the majority (n = 19, 61%) received anthracycline–based chemotherapy, 9 (29%) received taxane-based therapy, and 3 (10%) received CMF. Few patients overall (n = 10, 13%) received immunotherapy. All patients who received neoadjuvant systemic therapy (n = 10, 14%) had Sb-AdCC histology, and none achieved pCR. Overall, 67% (n = 47) of patients received radiation, 42 (89%) after lumpectomy and 5 (11%) after mastectomy. Median RFS was 7.2 years; 18 years for C-AdCC and 5.6 years for Sb-AdCC (p = 0.014). There was no significant difference in median RFS by receipt of systemic therapy in patients with Sb-AdCC (p = 0.8). Conclusions: Despite the use of standard systemic therapy, patients with Sb-AdCC have much worse oncological outcomes compared to C-AdCC. Alternative systemic therapy options are urgently needed. Patient age and disease-related factors by AdCC variant. Overall N=771 1 Classic AdCC N=221 1 Solid Basaloid AdCC N=551 1 p-value 2 Age 58 (48,69) 60 (50,70) 56 (48,68) 0.4 Subtype ER-/HER2- 68 (88%) 17 (77%) 51 (93%) 0.11 ER+/HER2- 9 (12%) 5 (23%) 4 (7.3%) Tumor Size (mm) 17 (13, 26) 18 (14, 26) 16 (11, 25) 0.5 Grade 1 6 (8.8%) 5 (26%) 1 (2.0%) <0.001 2 28 (41%) 13 (68%) 15 (31%) 3 34 (50%) 1 (5.3) 33 (67%) 1 Median (IQR); n (%). 2 Wilcoxon rank sum test; Fisher’s exact test.

Translating real-world data into actionable insights: An AI-powered model for prostate cancer survival prediction in a Brazilian cohort.

e23449 Background: Prostate cancer (PCa) outcomes are shaped by a complex interplay of clinical and socioeconomic factors. In resource-limited settings, leveraging routinely collected cancer registry data represent a critical but underutilized source for prognostic modeling. However, data heterogeneity as well as missing key clinical variables often limit their direct clinical applicability; consequently, translating real-word registry data into reliable survival predictions remains a significant clinical gap. We aimed to develop and externally evaluate a machine learning model to predict overall survival in PCa patients using data from a Brazilian Hospital-based Cancer Registry (HbCR), thereby demonstrating a practical application of data science translating into clinical oncology. Methods: We conducted a retrospective cohort study including 10,556 PCa patients recorded in a statewide HbCR from Southeastern Brazil, probabilistically linked to Mortality Information System. The dataset was randimly divided into training (n = 8,418) and testing (n = 2,110) sets. Feature selection was performed using the Boruta algorithm, yielding15 key predictors encompassing demographic, clinical, and treatment-related variables. Multiple machine learning models were trained and compared, with the LightGBM algorithm ultimately selected on discriminative performance. The primary outcome was all-cause mortality; accordingly, the model was trained to predict this outcome. Results: In the independent test set, the final LightGBM model achieved an overall accuracy of 71.3% and a weighted F1-score of 0.72. For mortality prediction, the model yielded a precision of 0.62 and a recall of 0.71 (F1-score = 0.66). For survival, precision was 0.79 and recall was 0.72 (F1-score = 0.75). The most influential predictors of survival identified by the model included age, clinical and pathological TNM stage, initial treatment modality, disease status at the end of treatment, race, and socioeconomic proxies like treatment cost. Model performance remained robust despite the absence of prostate-specific antigen levels and Gleason scores, demonstrating effective learning from heterogeneous registry data. Conclusions: Machine learning applied to routinely collected HbCR data can generate clinically meaningful survival predictions for PCs even in the presence of incomplete clinical information. This population-based LightGBM model offers a scalable, low-cost strategy to identify high-risk patients and can support clinical decision-making and resource allocation in public health systems. By transforming administrative HbCR into actionable survival insights, this approach exemplifies the ‘science and practice of translation' and and highlights the translational potential of real-world data analytics to improve PCa outcomes in LMIC.

Association of early on-treatment cytokine dynamics following induction chemo-immunotherapy with progression-free survival in squamous non–small cell lung cancer.

e20560 Background: Despite advances in immunotherapy-based combinations for advanced non–small cell lung cancer (NSCLC), biomarkers that reflect early on-treatment immune and angiogenic dynamics remain limited. LUNG-IST-127 is an investigator-initiated trial evaluating induction chemo-immunotherapy followed by maintenance pembrolizumab plus cabozantinib, a multitargeted tyrosine kinase inhibitor. We conducted an exploratory correlative analysis to assess whether longitudinal plasma cytokine changes are associated with progression-free survival (PFS). Methods: Peripheral blood samples were collected longitudinally from patients enrolled on LUNG-IST-127 at baseline, post-induction, and during maintenance therapy. Plasma cytokines were quantified using CodePlex Innate bulk cytokine profiling. Associations between cytokine dynamics (baseline, post-induction, and early change) and PFS were assessed using Spearman correlation. Analyses were exploratory and hypothesis-generating. Cytokine concentrations were log10-transformed. Within-patient changes from baseline to post-induction were calculated and assessed using Spearman rank correlations with progression-free survival (PFS). Exploratory comparisons used nonparametric methods, and p-values were nominal and hypothesis-generating due to the pilot sample size. Results: Seven patients with available cytokine data and PFS outcomes were included; six had paired baseline and post-induction samples. Baseline cytokine levels alone were not strongly associated with PFS. In contrast, early on-treatment cytokine changes demonstrated notable associations with clinical outcome. An increase in IL-10 from baseline to post-induction was strongly associated with longer PFS (p = 0.015), while an increase in VEGF over the same interval was associated with shorter PFS (p = 0.021). Higher post-induction VEGF levels were also inversely associated with PFS (p = 0.021). Maintenance-phase longitudinal analysis was underpowered, with only three patients contributing ≥2 maintenance samples. Conclusions: In this exploratory correlative analysis of LUNG-IST-127, early cytokine dynamics following induction chemo-immunotherapy were more informative than baseline measurements. Increases in IL-10 were associated with prolonged PFS, while increases in VEGF identified patients with poorer outcomes. These findings suggest that early angiogenic and immunoregulatory shifts may stratify patients by risk and raise the hypothesis that VEGF-high tumors following induction may represent a biologically relevant subgroup for intensified anti-angiogenic strategies during maintenance. Ongoing accrual and longitudinal sampling will be critical to validate these observations. Clinical trial information: NCT05613413 .

Efficacy outcomes in first-line (1L) non–small cell lung cancer (NSCLC) with maintenance of immune competence: QUILT-2.023 randomized phase 3 study of IL-15R agonist nogapendekin alfa inbakicept (NAI) with checkpoint inhibitor (CPI) ± chemotherapy.

8588 Background: Baseline and treatment-induced lymphopenia are adverse prognostic factors in NSCLC and may limit CPI durability. CPI and chemo-IO can reduce peripheral absolute lymphocyte counts (ALC) over time. NAI expands NK and effector/memory T cells and increases ALC, a biological marker of immune competence. In QUILT-3.055 (≥2L NSCLC after CPI failure), sustained ALC increases were associated with prolonged overall survival (OS). QUILT-2.023 tests whether preserving immune competence with NAI improves response rate, progression-free survival (PFS), and OS when added to 1L standards of care (SOC). To inform the design of the ongoing registrational-intent QUILT-2.023 study, results from 98 enrolled subjects were unblinded for exploratory analysis. Methods: QUILT-2.023 (NCT03520686) is an ongoing randomized multicohort phase 3 trial in untreated advanced/metastatic NSCLC (ECOG 0–1; squamous or nonsquamous). Cohort A (PD-L1 TPS ≥1%) was randomized 1:1 to CPI+NAI vs CPI; ALC over time was analyzed using a mixed model for repeated measures. The primary efficacy endpoint was PFS with hierarchical testing (TPS ≥50% then TPS ≥1%). PFS was analyzed at 8 months to evaluate the temporal relationship of ALC through 27 weeks of CPI alone versus CPI+NAI. Cohorts B (squamous) and C (nonsquamous), with any PD-L1 TPS, were randomized 1:1 to SOC platinum chemo-IO ± NAI; pooled Cohort B+C efficacy (ORR, DCR, PFS, OS) was assessed per RECIST v1.1. Results: Cohort A (N = 63): baseline ALC was comparable between arms; CPI+NAI showed significant, sustained ALC increase vs CPI alone over 27 weeks (P = 0.0065), regardless of PD-L1 TPS. In TPS ≥50% (N = 45), ALC over time significantly improved with CPI+NAI vs CPI alone (P = 0.0420). Correlating with this increased ALC, PFS in the CPI+NAI arm demonstrated median PFS of 7.0 vs 2.2 months (HR 0.40; 95% CI 0.17–0.94; p = 0.0298). Grade ≥3 CPI-related TEAEs occurred in 16% of subjects in both arms. In pooled Cohorts B+C (ITT N = 36), ORR (CR+PR) was 56% with chemo-IO+NAI vs 33% chemo-IO alone; DCR was 83% vs 78%. Median PFS was 18.9 vs 10.6 months (HR 0.48; 95% CI 0.19–1.23; p = 0.1192) and median OS reached significance of 34.7 vs 20.2 months (HR 0.38; 95% CI 0.15–0.98; p = 0.0394). Conclusions: Adding NAI to CPI-based SOC increased ALC and improved PFS in PD-L1 TPS ≥50% 1L NSCLC, with manageable safety. In platinum chemo-IO cohorts, pooled efficacy shows improved response and a significant OS signal, consistent with immune-competence preservation as a modifiable determinant of CPI durability. Enrollment and follow-up are ongoing. These interim findings are consistent with the single-arm trial QUILT-3.055 and support continued accrual in QUILT-2.023 and ResQ201A (NCT06745908), randomized NAI+CPI studies across 1L and ≥2L advanced NSCLC. Clinical trial information: NCT03520686 .

Firmonertinib combined with platinum-based chemotherapy in patients with <i>EGFR</i> -mutant advanced NSCLC following disease progression on first-line third-generation EGFR-TKI.

8610 Background: Patients (pts) who have progressed after first-line 3rd-generation EGFR-TKIs treatment exhibit limited therapeutic options. We report the preliminary antitumor activity and safety results of firmonertinib combined with platinum-based chemotherapy (Cohort C) in pts with EGFR-mutant advanced NSCLC who failed first-line 3rd-generation EGFR-TKIs from a prospective multi-cohort phase 2 trial (NCT06652048). Methods: Cohort C planned to include 20 pts with advanced NSCLC harboring EGFR-sensitizing mutations who progressed after first-line 3rd-generation EGFR-TKIs. Study treatment was firmonertinib (160 mg once daily) with chemotherapy (pemetrexed [500 mg/m 2 ] plus either cisplatin [75 mg/m 2 ] or carboplatin [AUC 5]). The primary endpoint was progression-free survival (PFS) assessed by investigators according to RECIST v1.1. Results: Cohort C completed accrual with 18 pts who received at least one dose of study treatment: median age was 64.5 years (range 43 to 80), ten (55.6%) were female, 16 (88.9%) were ECOG PS 1, nine (50%) were exon 21 L858R, six (33.3%) had CNS metastases at baseline. As of 30-Nov-2025, the median PFS was 8.4 months (95%CI 4.2-NR). The objective response rate (ORR) was 50.0% (95%CI 26.0-74.0). The disease control rate (DCR) was 88.9% (95%CI 65.3-98.6). The median duration of response (DOR) was not reached (95%CI 6.2-NR). The 9-month DOR rate was 75.0% (95%CI 12.8-96.1). The median overall survival (OS) was not reached (95%CI 9.1-NR). The 12-month OS rate was 70.0% (95%CI 16.2-93.3). Treatment-emergent adverse events (TEAEs) of any grade/grade≥3 occurred in 88.9%/44.4% of pts. The most common any-grade TEAEs were anemia (66.7%) and white blood cell count decreased (50%). The most common grade≥3 TEAEs were white blood cell count decreased (16.7%), neutrophil count decreased (11.1%) and anemia (11.1%). No TEAE with fatal outcome was reported. Conclusions: Firmonertinib combined with chemotherapy showed promising antitumor activity with a manageable safety profile in pts with EGFR-mutant advanced NSCLC following disease progression on first-line 3rd-generation EGFR-TKIs. Clinical trial information: NCT06652048 .

Pathologic complete response with neoadjuvant chemoimmunotherapy in clinical stage II–III NSCLC: A large real-world analysis including older adults.

8055 Background: Neoadjuvant chemoimmunotherapy is now the standard for clinical stage II-III non-small cell lung cancer (NSCLC) following CheckMate 816. However, older adults and socioeconomic factors have been underrepresented in clinical trials. This study aims to evaluate the impact of age and basic socioeconomic factors on pathological complete response (pCR) after neoadjuvant chemoimmunotherapy. Methods: Data from the National Cancer Database (NCDB) were obtained for patients diagnosed with clinical stage II-III NSCLC between 2019 and 2023 who underwent neoadjuvant chemotherapy (CT; 2019-2021) or chemoimmunotherapy (CTIO; 2022-2023) with complete T, N, and M staging data, as well as post-pathologic status (ypT, ypN). Those receiving neoadjuvant radiotherapy and with clinical or pathologic metastatic disease were excluded. Groups were stratified based on age (&lt;75 vs ≥ 75 years). Group comparisons were performed using chi-square testing. The primary outcome was pCR, analyzed using univariate and multivariable Cox regression models. A p-value of &lt;0.05 was considered statistically significant. Results: A total of 993 patients received CT alone, while 2,819 received CTIO. Among patients ≥ 75 years, pCR rates were significantly higher in the CTIO group compared with CT alone (25% vs 5%, p&lt;0.001). This pattern was similar in patients &lt;75 years (29% vs 8%, p&lt;0.001). Among CTIO receivers, pCR rates did not differ significantly by age (25% ≥75 vs 29% &lt;75; p=0.051) (Table 1). Patients with ≥10 lymph nodes examined had lower pCR rates compared with those with fewer nodes examined (26% vs 37%; p&lt;0.001). On multivariate analysis, higher nodal stage (N1–N3; OR 1.23, 1.02-1.49, p=0.035) and non-adenocarcinoma histologies (OR 1.26, 1.07-1.49, p=0.006) were independently associated with higher odds of pCR. Notably, age, sex, race, and rural/urban residence were not independently associated with pCR, and no significant differences in race or rural/urban status were observed between groups. Conclusions: To our knowledge, this is the largest real-world analysis of neoadjuvant chemoimmunotherapy outcomes in older adults with clinical stage II-III non–small cell lung cancer. We found comparable pCR rates across age groups, supporting the use of this approach in older individuals. While racial disparities have been described in access to neoadjuvant therapy, our study suggests similar response rates across racial groups who underwent treatment. Higher nodal status was independently associated with pCR, which may be explained by greater tumor immunogenicity. These findings are hypothesis-generating and warrant validation in further prospective studies. Pathologic complete response rate by age status. Factors pCR(n=811) No pCR (n=2008) P value (Chi-square) Age ≥ 75&lt; 75 115 (25%)696 (29%) 345 (75%)1663 (71%) 0.051 pCR=Pathologic Complete Response.

Evolving graph attention networks for dynamic link prediction

Graph neural networks (GNNs), which learn node representations via aggregating their neighbors, have shown superior performance and become the de facto efficient toolkit for analyzing and learning from data with structured properties. However, most existing GNNs are designed for static graphs and assume fixed graph structures and node sets. In many real-world applications, graphs evolve continuously over time—with nodes and edges appearing or disappearing—rendering static models insufficient for capturing these temporal dynamics. In this paper, we propose Evolving Graph Attention Networks (EGAT), a novel framework for dynamic graph representation learning. Specifically, EGAT leverages the anisotropic attention mechanism of Graph Attention Networks (GATs) to capture complex inter-node relationships. Crucially, the multi-head attention weights of the GAT are evolved over time via a recurrent neural network (RNN), enabling the model to adaptively adjust the importance of different neighbors as the graph topology and relational dynamics change. This weight-evolving paradigm couples the anisotropic attention mechanism of GATs with a recurrent subnetwork, enabling the joint modeling of topological evolution and temporal relational dynamics. Extensive experiments on benchmark datasets demonstrate that the proposed model consistently outperforms state-of-the-art baselines.

Homoleptic and Heteroleptic Borylones L1‐B(Ph)‐L2

ABSTRACT Quantum chemical calculation using density functional theory at the BP86‐D3(BJ)/def2‐TZVPP level and ab initio theory at the CCSD(T)/def2‐TZVPP level have been carried out for the homoleptic and heteroleptic borylones L1‐B(Ph)‐L2 with the ligands L1, L2 = PPh 3 , SPh 2 , N 2 , CO, CS, NHC Me , CAAC Me . The computation of the bond dissociation energy suggests that all borylones considered in this work should be stable enough to be observed experimentally. A surprising result is the finding that the homoleptic borylones B(Ph)(N 2 ) 2 and B(Ph)(SPh 2 ) 2 have a higher BDE than the corresponding carbones. It seems possible that bis‐dinitrogen borylones can be synthesized and structurally characterized under appropriate conditions. The boron‐ligand bonds of the heteroleptic borylones L1‐B(Ph)‐L2 influence each other, so that one bond becomes stronger and the other weaker, but to different degrees. The BDEs do not always show the same trend, as the electronic and geometric relaxation of the fragments can strongly influence the energy change due to bond breaking, which affects the thermodynamic stability of the compounds. The EDA‐NOCV results show that Pauli repulsion is often the strongest energy component of the chemical bond, determining the bond strength and bond length.

A narrative review on the role of herbal constituents and their combination with nanotechnology as future hope in the management of diabetic retinopathy

Alkaline Leaching: A Facile Surface Activation Strategy to Improve the Reactivity of Air Electrodes for Solid Oxide Fuel Cells (Adv. Mater. 33/2026)

Dynamic Covalent Peptide‐Drug Conjugates Address the Heterogeneity in Alzheimer's Disease Progression

ABSTRACT A growing understanding of the pathophysiological evolution of Alzheimer's disease (AD) underscores the heterogeneity in its progression as a critical factor undermining the success of various candidate interventions and complicating the establishment of effective pharmacotherapeutic regimens. Here, we introduce the development of a hierarchical‐responsive therapeutic agent self‐assembled from phenylboronate ester‐linked Tjernberg's KLVFF peptide‐curcumin conjugates (CPKNAs), which is designed to dynamically track the spatiotemporal coordinates of biomarkers associated with AD heterogeneity in progression. The dynamic covalent phenylboronate ester bond undergoes varying degrees of dissociation in response to the temporal evolution patterns of amyloid‐β, reactive oxygen species, and glucose, three key indicators for staging AD progression, enabling self‐adaptive regulation of drug distribution and dosing tailored to specific phenotypes. Employing seven cellular models and three types of transgenic mice simulating different AD stages, we demonstrate that CPKNAs effectively minimize the risks of under‐ or overtreatment, achieving consistent therapeutic outcomes in mitigating cellular damage and improving brain dysfunction.

Submanifold sparse convolutional networks for automated 3D segmentation of kidneys and kidney tumours in computed tomography

Abstract Accurate delineation of kidney tumours in Computed Tomography (CT) is essential for downstream quantitative analysis and precision oncology that could enable personalised treatments, but manual segmentation is a specialised task, time-consuming and difficult to scale in routine practice. Automated 3D segmentation remains challenging in medical imaging, where images are large and dense volumes of data, making high-resolution processing with conventional dense convolutional neural networks computationally expensive, and often reliant on downsampling or patch-based inference. To overcome this problem, we propose a two-stage 3D segmentation methodology based on voxel sparsification and submanifold sparse convolutional networks (SSCNs). In Stage 1, a low-resolution sparse network identifies a region of interest (ROI); in Stage 2, a high-resolution sparse network performs refined segmentation within the cropped ROI. This design enables native 3D processing at high resolution while reducing CPU/GPU memory usage and inference time. We evaluate the method on the KiTS23 dataset of renal cancer CT scans using 5-fold cross-validation. Our method achieved Dice similarity coefficients of 95.8% for kidneys + masses, 85.7% for tumours + cysts, and 80.3% for tumours alone, with performance competitive with top KiTS23 approaches. In direct comparisons on the same cross-validation folds, the proposed sparse method achieves tumour + cyst and tumour-only Dice scores comparable to, and slightly higher than, a patch-based nnU-Net baseline, while consistently requiring less VRAM and shorter inference time across the tested hardware. Across the tested GPUs, our sparse model is markedly faster than both nnU-Net and the zero-shot zoom-out/zoom-in foundation model SegVol, which localises kidneys well but underperforms on small heterogeneous lesions. Compared to an equivalent dense implementation of the same architecture, the proposed sparse approach achieves up to a 60% reduction in inference time and up to a 75% reduction in VRAM usage across both CPU and the GPU configurations tested.