Comparison of HRD scores in primary vs. recurrent disease and untreated vs. previously treated disease.

Clinical characteristics and prior treatments of study participants.

Exploratory analysis of the three components of the HRD score.

PURPOSE:Copy number-high (CN-H)/p53-abnormal endometrial cancers are high-grade uterine malignancies characterized by TP53 mutations, copy-number alterations, mismatch repair proficiency (MMRp), and absence of POLE mutations. A subset may be homologous recombination deficient (HRD), potentially conferring sensitivity to poly (ADP-ribose) polymerase (PARP) inhibitors. We aimed to test the combination of PARP and immune checkpoint inhibitors in this subgroup, leveraging possible synergy from immune priming. PATIENTS AND METHODS:We conducted a single-arm, open-label, phase II trial evaluating the efficacy and safety of olaparib (300 mg orally twice daily) plus pembrolizumab (200 mg intravenously every 3 weeks) in patients with persistent or recurrent CN-H/p53-abnormal endometrial cancer. Eligible patients had p53-abnormal, MMRp, and POLE-negative disease and up to 3 prior lines of therapy. The primary endpoint was the best overall response rate (ORR) at 24 weeks. RESULTS:Of the 25 patients evaluable for efficacy, 2 patients achieved a complete response, and 6 achieved a partial response, resulting in an ORR of 32% [90% one-sided confidence interval (CI), 19.6%-100%]. The median duration of response was 11.2 months (80% two-sided CI, 6.4-11.9). The median progression-free survival was 3.9 months (80% two-sided CI, 2.1-5.8), and the median overall survival was 16.5 months (80% two-sided CI, 9.6-23.6). No new safety signals were identified. Genomic analyses suggested that responders had a numerically higher frequency of HRD tumors than nonresponders (50% vs. 17%). CONCLUSIONS:The combination of olaparib plus pembrolizumab has promising activity with durable responses in patients with persistent or recurrent CN-H/p53-abnormal endometrial cancer. Molecular biomarkers may be helpful for patient selection in future studies of this combination.

5615 Background: Carcinosarcoma is a rare and aggressive histology of gynecologic carcinomas that has limited treatment options and is often excluded from larger therapeutic clinical trials. In KN775, treatment with multikinase inhibitor lenvatinib plus PD-1 inhibitor pembrolizumab led to significantly longer progression-free survival (PFS) and overall survival (OS) than chemotherapy among patients with previously treated advanced non-carcinosarcoma endometrial carcinomas. We report results from a phase 2 trial of lenvatinib + pembrolizumab in pts with endometrial or ovarian carcinosarcoma. Methods: In this open label, single institution, investigator initiated, phase 2 study, pts had confirmed recurrent/persistent endometrial or ovarian carcinosarcoma with progression after at least 1 prior platinum chemotherapy but no more than 3 prior lines, measurable disease, and ECOG performance status ≤1. Pts received oral lenvatinib 20 mg/day plus pembrolizumab 200 mg intravenously every 3 weeks. The primary endpoints were objective response rate (ORR) per RECIST v1.1 and PFS rate at 27 weeks (PFR27). Secondary endpoints included median duration of response (mDOR), clinical-benefit rate (CBR: complete response (CR) + partial response (PR) + stable disease (SD) at 27 weeks), median overall survival (mOS), and safety. Next generation sequencing, MSK-IMPACT, was performed on tumors from all 27 patients for exploratory analysis. Results: As of data cutoff on 12/15/2025, 27 pts enrolled; median age was 63 years (range: 51−79); and at least 26.4% pts were non-white. Primary tumor site was 63% (17/27) endometrial and 37% (10/27) ovarian; tumors were 19% (5/27) dMMR and 81% (22/27) pMMR. Confirmed ORR was 18.5% (5/27; 0 CRs, 3 PRs in endometrial, 2 PRs in ovarian), meeting the interim analysis primary endpoint. Of the 5 confirmed responses, 4 occurred in pts with pMMR tumors, and 1 in a pt with dMMR tumor. There are 2 additional unconfirmed PRs. Responses in pts with ovarian tumors included 1 platinum-sensitive and 1 platinum-resistant. PFR27 was 35.7% (90% CI: 24 - 100), mDOR was 26.5 months (80% CI: 3.8 - NE), CBR was 37% (80% CI: 24.4 - 51.2), and mOS was 15.3 months (95% CI: 7.1 – 24.4). Two pts remain on treatment and were censored for time to event analyses. All pts had at least one treatment-emergent adverse event (TEAE). The most common TEAEs were fatigue (74%), hypertension (56%), diarrhea (52%), anorexia (48%), and weight loss (48%). Grade 3 and 4 TEAEs occurred in 44% and 7.4% patients, respectively. There were no grade 5 TEAEs. 5 Pts (19%) discontinued study treatment due to TEAEs. Further genomic analyses are underway. Conclusions: Encouraging efficacy and durable responses were observed in endometrial and ovarian carcinosarcoma pts treated with lenvatinib + pembrolizumab, regardless of MMR status or primary site. The study met its predefined interim analysis endpoint. No new safety signals were identified. Clinical trial information: NCT02501096 .

Representativeness of study participants.

Background Accurate CT-based segmentation of primary esophageal tumors is critical for disease management. While AI models show promise for automated segmentation, the underrepresentation of certain subgroups may limit their robustness and generalizability. However, it remains unclear which factors contribute to reduced model effectiveness. Methods In this retrospective study, segmentation models using nnU-Net were trained on baseline CT images. We systematically excluded tumors based on histological subtype and anatomical location to assess the impact on segmentation performance, measured by Dice similarity coefficient (DSC) and tumor detection rate. Additionally, we explored the effect of including intravenous contrast-enhanced scans in the training set. Results 275 patients were included: 71 from the Erasmus University Medical Center for training and internal testing, and 204 from the Memorial Sloan Kettering Cancer Center for external testing. Excluding squamous cell carcinoma tumors from the training set significantly reduced average DSC for these lesions on the external test set (p < 0.0001; Cohen’s d=0.79), while results for adenocarcinoma remained stable (p > 0.1; Cohen’s d=-0.03). Similarly, excluding mid and upper esophageal tumors decreased segmentation performance (p < 0.01; Cohen’s d=0.90 and 0.72 for the mid and upper ones). Finally, incorporating contrast-enhanced CT improved performance with a large effect size compared to a model trained only on non-contrast scans, with a mean DSC increase of 39% on contrast-enhanced images. Conclusion Underrepresentation of histological subtypes, tumor locations, and imaging protocols in training data introduces representation bias that can considerably reduce segmentation performance. Expanding the training set to include these factors helps mitigate this bias and improves the robustness and generalizability of AI models.

HER2 heterogeneity and RTK alterations are more frequent among patients with shorter PFS. HER2 testing and tissue/plasma ctDNA NGS upon progression demonstrate that 8 of 16 (50%) of patients had HER2 negative tumors post-treatment, and 2 ERBB2 amplified tumors pre-treatment were non-amplified post-treatment. Additionally, frequent alterations in PI3K, cell cycle proteins, Ras-Raf, EGFR, FGFR1/2, and MET were found upon progression, and potentially associated with treatment resistance.

PIK3CD S367L demonstrate increased phosphorylation of AKT. Upon progression, 1 patient developed a PIK3CD S367L mutation (location indicated in gene map), a variant of undetermined significance in the carboxy terminal C2 domain. Above, western blot demonstrating pAKT expression in 3T3 cells stably expressing p110δ WT or mutations at p110d E1021K, L806M, S367L, R894Q, G971E, or R338W relative to non-transformed parental 3T3 cells, which do not express p110δ. Below, lollipop plot of PIK3CD.

Metastatic Sites Involved Prior to HER2-Directed Therapy.

BACKGROUND:To identify baseline rectal MRI characteristics that may serve as predictive factors for recurrence in patients with rectal adenocarcinoma after surgical resection. METHODS:This retrospective, single-center study included 269 consecutive patients (median age, 55 years [interquartile range, 47-65]; 144 men and 125 women) diagnosed with rectal cancer from January 2015-December 2017 who underwent baseline rectal MRI followed by surgical resection. MRI characteristics were collected from rectal MRI synoptic reports. Recurrence-free survival was defined as the time between surgical resection and recurrence (local recurrence and/or metastatic disease) or death. Statistical analysis included Cox proportional hazards to determine associations between baseline rectal MRI/clinical characteristics and recurrence. RESULTS:The median recurrence-free survival in the study sample was 6.4 years. Baseline rectal MRI characteristics associated with recurrence at univariable analysis were: age > 55 years (P = 0.044), low rectal tumor location (P = 0.04), craniocaudal length ≥ 5.0 cm (P = 0.007), anal canal involvement (P = 0.011), presence of suspicious total mesorectal excision (TME) lymph nodes > 0.5 cm (P = 0.03), mesorectal fascia involvement (P = 0.04), T3 stage (P = 0.024), T4 stage (P = 0.008), and M1 stage (P = 0.024). At multivariable analysis, only age > 55 years (P = 0.012) and the presence of suspicious TME lymph nodes > 0.5 cm (P = 0.049) remained associated with recurrence. CONCLUSION:Advanced age and the presence of suspicious TME adenopathy > 0.5 cm on baseline rectal MRI are associated with higher risk of recurrent disease in patients with resected rectal cancer.

Representativeness of Study Participants.

Metastatic Sites Involved Prior to HER2-Directed Therapy.

Multivariable Cox Proportional Hazard for PFS in HER2-targeted retrospective cohort (n=217).

Early change in tumor-matched maxVAF in patients who received induction trastuzumab and pembrolizumab stratified by 6 month PFS (n=16 evaluable).

Neoadjuvant therapy (NAT) for patients with rectal cancer is intended to reduce the risk of local recurrence and distant metastasis while preserving quality of life. Choosing the optimal approach after NAT is complex, requiring a personalized plan that considers the unique characteristics of each patient and their tumor, as well as the resources and capabilities of the treating institution. Advances in treatment, guided by insights from randomized clinical trials and increasing acceptance of organ preservation strategies, also known as nonoperative management or the watch-and-wait approach, emphasize the importance of precise treatment response assessment and multidisciplinary communication. A comprehensive evaluation using digital rectal examination, endoscopy, and MRI with a dedicated rectal cancer protocol ensures accurate clinical and locoregional response assessment. This integrative approach enables clinicians to make decisions regarding organ preservation, surgery, treatment de-escalation, or even additional NAT, with MRI having a critical role in surgical planning when resection is needed. The authors provide a comprehensive guide for interpreting postneoadjuvant rectal cancer MRI and applying the Society of Abdominal Radiology Colorectal and Anal Cancer Disease-focused Panel (SAR DFP) synoptic MRI restaging report template. The aim is to improve the quality, consistency, and clarity of MRI interpretations across different readers and institutions. Each section corresponds to the SAR DFP synoptic MRI restaging report template, addressing common areas of confusion and providing essential background material to ensure that clinically relevant information is clearly communicated to the treatment team, supporting effective decision making and enhancing patient outcomes. ©RSNA, 2025 Supplemental material is available for this article.

Response to pembrolizumab and trastuzumab induction according to pre-treatment 89Zr-trastuzumab PET avidity.

Survival and response in patients receiving trastuzumab/chemotherapy + PD-1 inhibition on- or off-protocol (n=66). a, Progression-free survival and objective response rate by cohort. b, Overall survival by cohort.

Serial plasma ctDNA in an individual patient demonstrating pre-treatment genomic heterogeneity and selection for resistant clones. Plasma ctDNA at diagnosis revealed ERBB2 amplification in a patient with HER2+ EG cancer, as well as numerous pre-existing resistance mechanisms including EGFR amplification, PIK3CA amplification, and an FGFR3-TACC3 fusion that was not identified on tumor sequencing (Figure 2b).