Citations

Objective Meaningful assessments of how large language models (LLMs) incorporate clinical guidelines require large-scale testing over many queries. Here, we evaluate the prevalence of clinical guideline omissions and hallucinations in a large sample of diagnostic LLM outputs. Methods We used simulated case vignettes and zero-shot prompting to generate diagnostic outputs and rationales from GPT-4.1 and DeepSeek-V3. English case vignettes were created for hypercholesterolaemia and type-2 diabetes mellitus. Each vignette contained identical medical information, while sociodemographic characteristics varied in terms of sex, ethnicity and location. We calculated the prevalence of existing and hallucinated clinical guidelines in LLM outputs across disease, LLM and sociodemographic characteristics. Results We analysed a total of 12 197 LLM outputs, which quantifies three hazard areas: omissions (up to 97% for DeepSeek-V3 and 46% for GPT-4.1), hallucinations (up to 9%) and inconsistencies (guideline citation rate ranging from 0% to 78.39% across sociodemographic vignettes). Omission and hallucination rates were generally similar across vignettes with different sex or ethnicity data, yet were particularly sensitive to patient location. Discussion This study highlights significant variability in clinical guideline prediction across two different diseases, three different sociodemographic variables and two LLMs, even when the LLMs were instructed by identical prompts, establishing clinical guideline prediction in LLM outputs as a stochastic event. Conclusion The stochastic nature of LLMs creates a unique challenge for evidence generation and clinical deployment. Being able to measure and capture this stochasticity within high-quality research designs will be a prerequisite to advancing the responsible deployment of LLMs in healthcare.

Abstract Pancreatic cancer disproportionately affects Black individuals in the United States, but they have limited representation in genetic studies of pancreatic ductal adenocarcinoma (PDAC). To address this gap, we performed admixture mapping and genome-wide association analysis (GWAS) in genetically inferred African ancestry individuals (1,030 cases and 889 controls). Admixture mapping identified three regions with a significantly higher proportion of African ancestry in cases compared to controls (5q33.3, 10p1, 22q12.3). GWAS identified a genome-wide significant association at 5p15.33 ( CLPTM1L , rs383009:T>C, T Allele Frequency=0.51, OR:1.45, P value=1.24×10 -8 ), a locus previously associated with PDAC. Known loci at 5p15.33, 7q32.3, 8q24.21 and 7q25.1 also replicated (P value <0.01). Multi-ancestral fine-mapping identified two potential causal SNPs (rs3830069 and rs2735940) at 5p15.33. Collectively these findings identified novel PDAC risk loci and expanded our understanding of this deadly cancer in underrepresented populations, emphasizing the multifactorial nature of PDAC risk including inherited genetic and non-genetic factors. Statement of Significance To understand how genetic variation contributes to PDAC risk in Black people in North American, we studied individuals of genetically-inferred African ancestry. We identified novel risk loci and differences in the contribution of known loci. This demonstrates that ancestry-informed genetic analyses improve our understanding of PDAC risk and enhances discovery.

OBJECTIVE Overuse of broad-spectrum empiric antimicrobials remains common in children hospitalized with respiratory tract infections (RTI). A negative methicillin-resistant Staphylococcus aureus (MRSA) nasal swab (MNS) polymerase chain reaction (PCR) result has been shown to have a high negative predictive value for MRSA in RTI. In this quality improvement (QI) initiative, we developed a pharmacist-driven approach to reduce the duration of empiric vancomycin for RTI in children. The primary aim was the time to de-escalation of vancomycin, and the secondary aim was the reduction in vancomycin monitoring in patients hospitalized with RTI. METHODS Baseline duration of vancomycin therapy was assessed by conducting a retrospective chart review. After root cause analysis and engaging key stakeholders, an electronic health record OurPractice Advisory (OPA) was implemented, prompting pharmacists to order an MNS PCR during order verification for vancomycin with an RTI indication. RESULTS Preimplementation (01/01/2021 to 08/31/2023) included 89 patients, and 25 postimplementation (09/21/2023 to 05/31/2024). We found a 36% decrease in mean vancomycin duration from 53 to 34 hours after implementation of the pharmacist-driven OPA. Postimplementation, 24% of patients had a vancomycin concentration obtained, compared with 37% preimplementation. No patients were readmitted within 30 days postimplementation, compared with 1 patient preimplementation. CONCLUSIONS This QI initiative provides preliminary data that empiric vancomycin duration in pediatric patients with RTI may be impacted by pharmacist-facing OPA implementation, prompting ordering of MNS PCR at the time of vancomycin verification. This initiative could be replicated at other institutions seeking to reduce empiric vancomycin usage.

Background: Physical inactivity is highly prevalent among individuals with spinal cord injury (SCI), contributing to secondary health conditions and reduced quality of life (QOL). Objectives: To examine the independent associations between adherence to SCI cardiorespiratory fitness guidelines and health and psychosocial outcomes, while controlling for relevant covariates. Method: We conducted a multisite, cross-sectional study of adults with traumatic SCI stratified by self-reported leisure-time physical activity: no activity (28%), some activity (41%), and met guidelines (26%). Multivariable regression models analyzed associations between physical activity (PA) group and outcomes of hospitalization, self-perceived health, resilience (SCI-QOL Resilience), pain, depression (PHQ-8), QOL (SWLS), and social integration (CHART-SF). Results: The study included 361 participants (no activity, 28%; some activity, 41%; met guidelines, 26%). Meeting PA guidelines was associated with better self-perceived health (odds ratio [OR] 2.79, P < .001) and higher resilience (β = 1.78, P = .04) compared to the No Activity group. However, hospitalization risk was unexpectedly higher for both the Some Activity (risk ratio [RR] 1.93, P = .005) and Met Guidelines (RR 1.75, P = .03) groups. After adjusting for covariates, PA was not significantly associated with pain, depression, QOL, or social integration; instead, higher income was consistently associated with better outcomes in these domains. Conclusion: Adherence to SCI exercise guidelines is associated with enhanced self-perceived health and resilience. However, PA alone was not associated with broader psychosocial outcomes in this model, where socioeconomic factors such as income appeared to play a significant explanatory role.

Importance Young adults (YAs; ages 18-39 years) diagnosed with cancer face unique psychosocial challenges, including high distress and poor health-related quality of life (HRQOL). There are few efficacious interventions tailored to their needs. Objective To evaluate the efficacy of Bright IDEAS-YA, a problem-solving skills training intervention tailored for YAs, in reducing distress and improving HRQOL. Design, Setting, and Participants This randomized clinical trial was conducted at 3 academic medical centers and enrolled participants between February 2021 and March 2024. Eligible participants were aged 18 to 39 years, within 4 months of a first cancer diagnosis, and undergoing systemic therapy. Participants completed surveys at baseline and 3, 6, 12, and 24 months. Intervention Bright IDEAS-YA is a 6-session, one-on-one intervention that teaches a structured approach to problem-solving. Enhanced usual care (EUC) included standard psychosocial care plus a list of adolescent and YA resources. Main Outcomes and Measures Primary outcomes were depression and anxiety (Patient-Reported Outcomes Measurement Information System Short Forms) and HRQOL (Functional Assessment of Cancer Therapy-General) at 6 months. Problem-solving ability (Social Problem-Solving Inventory-Revised Short Form) was examined as a mediator. Analyses used linear mixed effects models and mediation analysis. Results Of the 1128 YAs screened, 344 YAs (median [IQR] age, 31.27 [25.78-36.38] years; 251 female [62.5%]) enrolled, with 171 allocated to intervention and 173 allocated to control. Overall, 296 YAs (86.0%) and 280 YAs (81.4%) completed 3- and 6-month surveys, respectively. At 6 months, the intervention group showed significantly greater improvements than EUC in depression (effect size estimate, −3.23 points; 95% CI, −4.93 to −1.53 points; P &amp;lt; .001), anxiety (effect size estimate,−2.43 points; 95% CI, −4.05 to −0.81 points; P = .003), and HRQOL (effect size estimate, 3.40 points; 95% CI, 0.34 to 6.45 points; P = .03). These changes were clinically meaningful. Mediation analyses showed that improvements in depression and anxiety were partially due to increased problem-solving ability, particularly reductions in negative problem orientation. Sex and education were moderators of treatment effect, with the treatment effect on anxiety being greater for males (effect size estimate, −5.29 points; 95% CI, −7.97 to −2.60 points) and the treatment effect on anxiety (effect size estimate, −6.94 points; 95% CI, −10.50 to −3.43 points) and depression (effect size estimate, −7.47 points; 95% CI, −11.20 to −3.74 points; P &amp;lt; .001) being greater for those with lower education. Conclusions and Relevance In this randomized clinical trial of Bright IDEAS-YA, intervention participants had significantly reduced distress and improved HRQOL relative to control. These findings suggest that Bright IDEAS-YA may be offered as supportive care for YAs with cancer to improve psychosocial outcomes. Trial Registration ClinicalTrials.gov Identifier: NCT04585269

The 6th edition of the WHO Classification of Breast Tumours introduces both major and minor changes based on recent advances in our understanding of breast biology, developments in diagnostic modalities, identification of specific molecular targets and new treatment regimens necessitating modifications to pathology reporting and tumour biomarker categorisation. This review summarises the main changes that strive towards a classification of global relevance. In invasive carcinoma, predictive factors increasingly inform modern breast cancer treatment. The 6th edition provides an update on HER2 reporting categories following the DESTINY‐Breast 04 and 06 trials. Terminologies used to classify invasive tumours are clarified, with the term ‘variant’ now reserved for molecular/genetic alterations. Invasive lobular carcinoma (ILC) with extracellular mucin is recognised as a new diagnostic entity with prognostic implications. The diagnosis of mucinous carcinoma (MC) is reserved for mucin secreting carcinomas with grade 1 or 2 morphology and a favourable biomarker profile. The diagnosis of malignant phyllodes tumours requires only four of the original five adverse histological criteria. Classification of neuroendocrine tumours (NETs) is revised, recognising that the unified model, promoted in the 5th edition, is difficult to apply to the breast. New approaches to the classification of adenomyoepithelioma are discussed but the 5th edition system is broadly retained. A new section on ‘Small Diagnostic Samples’ outlines the merits of non‐operative biopsy diagnosis, the B coding system and the importance of multidisciplinary review. Changes to diagnostic practice and the emerging role of artificial intelligence, with advantages and challenges, are discussed in a new section on ‘Digital Pathology’.

We previously reported initial results of the pivotal phase II trial of atezolizumab for patients with alveolar soft part sarcoma (ASPS; ClinicalTrials.gov identifier: NCT03141684 ). Here, we report on three additional years of observation. Fifty-three patients with ASPS received atezolizumab. Median duration of response increased to 37.0 months. Objective response rate (ORR) and median progression-free survival (mPFS) remained essentially as previously reported (35.8% [95% CI, 23.1 to 50.2] and 20.8 months [IQR, 7.6-not reached], respectively). ASPSCR1::TFE3 fusion type was determined for 47/53 patients; ORR and mPFS were higher among the 41 patients expressing type 1 (43.9% [95% CI, 28.5 to 60.2] and 28.3 months [IQR, 9.2-not reached], respectively) than the six patients expressing type 2 (0% [95% CI, 0 to 45.9] and 7.5 months [IQR, 3.9-not reached], respectively, PFS HR, 3.2 [95% CI, 1.01 to 10.2]). Eleven patients chose a per-protocol drug holiday (range, 3.5-26.4 months) after ≥2 years of treatment; two experienced disease progression during the holiday. Nine eligible patients elected to receive bevacizumab plus atezolizumab after progressing on monotherapy; ORR was 0% and mPFS was 18.5 months (IQR, 7.9-21.1) in this small cohort. Long-term results support using atezolizumab to treat ASPS, even for several years; a drug holiday with careful monitoring may be an option for some patients.

Abstract Background: Early detection is the most potent strategy to improve survival in PDA, but its success hinges on an unresolved question: does radiologically invisible disease represent biologically early or indolent disease, or established malignancy beneath the threshold of detection? We leveraged prediagnostic CTs—incidental scans obtained 3-36 months prior to diagnosis without a focal mass—to define the relationship between imaging phenotype and tumor biology. Methods: From a multi-institutional pool of 5,046 biopsy-confirmed PDAs, we assembled a cohort of 346 prediagnostic CTs, which were independently evaluated by three radiologists using a standardized protocol. Based on majority vote, CTs were dichotomized into imaging-occult (IO) (normal pancreas) and imaging-apparent (IA) (abnormal pancreas with indirect signs such as focal atrophy, duct dilatation, etc.) groups. Tumor size and metastases (M) at diagnosis, and overall survival (OS) were compared using multivariable linear (size), logistic (M1), and Cox proportional hazards (OS) models, adjusting for lead time to diagnosis, age, sex, diabetes, CA 19-9, and surgical resection status, to isolate the independent biological relevance of the imaging phenotype. Results: IO phenotype was seen in 150 patients (mean age 66.7±11.2; 37.3% females) while 196 had IA phenotype (mean age 69.9±10.6; 49.5% females). Median lead time was 396 days (range: 90-1088) with the IA group being enriched in 3-12-month lead time (57.7% vs 33.3%; p&lt;0.001). Both groups had comparable diabetes rates (IA 33.2% vs IO 32.7%; p=1.0), median CA 19-9 (IA 193 vs IO 322 U/mL; p=0.41), and surgical resection (IA 38.3% vs IO 38.0%; p = 1.0). While unadjusted mean tumor size at diagnosis was larger in IA group (3.8 cm vs 3.2 cm), there was no difference after adjusting for clinical factors (-7.1% difference; 95% CI: -17.2, 4.2; p=0.21). Metastases were also comparable (M rate: IA 22.4% vs IO 27.3%; adjusted OR 0.97; 95% CI: 0.50, 1.88; p=0.93). Adjusted Cox models demonstrated equivalent OS (median: 18.7 vs 17.6 months; HR 1.06; p=0.66), which persisted even when stratified by surgical status (HR 1.07; 95% CI: 0.83, 1.38; p=0.61). Even in the critical 3-12 months lead time (n=163), OS remained equivalent (HR 0.89; p=0.62), and groups did not differ in either tumor size (-6.4%; 95% CI: -22.4, 12.9; p=0.49) or metastases (OR 0.48; 95% CI 0.14, 1.59; p=0.23). Conclusions: A normal-appearing pancreas on prediagnostic imaging is common yet signifies neither early nor indolent disease—it harbors PDA biologically equivalent to the imaging-apparent pancreatic phenotype. Because the disease at this stage lies beyond human perceptual capacity, AI augmentation represents the only potential path to realizing the survival benefit of early detection. Citation Format: Khurram Khaliq Bhinder, Sovanlal Mukherjee, Armin Zarrintan, Angela Ammirabile, Ahmed Khan Jadoon, Ziding Feng, Suresh T. Chari, Ajit H. Goenka. Dissociation between prediagnostic imaging phenotype and tumor biology defines the early detection challenge in pancreatic ductal adenocarcinoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr LB205.

Abstract Background: Zedoresertib (zedo) and lunresertib (lunre) are oral, highly selective WEE1 and PKMYT1 inhibitors, respectively. Inhibition of these targets leads to premature mitotic entry, and in the presence of high replication stress causes mitotic catastrophe. In preclinical models, the combination of the zedo and lunre is synthetically lethal, supporting this first-in-class trial combination (NCT04855656). Methods: In this Phase 1 trial, patients (pts) with advanced solid tumors with CCNE1 amplification (amp), or FBXW7 or PPP2R1A deleterious mutations (mut) were treated with zedo (daily dosing) plus lunre (3 days on/4 days off), guided by preclinical modelling of the pharmacological contributions of WEE1 and PKMYT1 inhibition in achieving tumor regression. Objectives included safety, tolerability, recommended doses (RD), pharmacokinetics (PK), pharmacodynamics (PDy), antitumor activity and molecular response rate (MRR; ≥50% decline in circulating tumor DNA). Results: As of 18 Nov 2025, 54 pts (55.6% female ; prior lines of therapy 1 - 9, main cancer types: ovarian (OC) (35.2%), colorectal (31.5%), pancreatic (7.4%) and breast (5.6%)) were treated at 5 dose levels (DL) from 150 mg to 260 mg of zedo plus 60 mg or 80 mg of lunre. The most common ≥G3 treatment related adverse events (TRAEs) were anemia (9.3%), neutropenia (7.4%) and Palmar Plantar Erythrodysesthesia (PPE) (7.4%). There were no G5 TRAEs. 4 (9.3%) of 54 pts had a dose-limiting toxicity (G3 pain in extremity and G1 rash in same pt (zedo 150 mg / lunre 60 mg), G3 rash (200 mg / 80 mg), G3 PPE (260 mg / 60 mg), G3 vomiting (260 mg / 60 mg)). PK/PDy analyses showed dose-dependent exposures for both zedo and lunre and target engagement in most pts, with reductions in pCDK1 Thr14 and Tyr15 and yH2AX induction. 24 (55.5%) of 44 RECIST v1.1 evaluable pts showed tumor shrinkage (3 cCR, 5 uPR, 15 SD and 1 PD). Of 16 evaluable pts with advanced OC (12 platinum resistant , 2 unknown and 2 platinum sensitive) with CCNE1 amp (13 pts) or FBXW7 mut (3 pts), 13 (81.3%) showed tumor shrinkage (3 cCR, 3 uPR and 7 SD); clinical benefit rate (pts with CR, PR or SD) was 93.8%; 7 (43.8%) OC pts were on treatment for at least 16 weeks; 4 (25%) OC pts achieved GCIG CA-125 responses. Of 23 pts with a post baseline ctDNA analysis, a MRR of 39% was seen across all DLs. Conclusion: We report the first disclosure of clinical data of the first-in-field synthetic lethal combination of WEE1 and PKMYT1 inhibitors in a molecularly defined pt population. The safety profile of zedo and lunre was expected and manageable. Promising antitumor activity was observed in pts with CCNE1 amp or FBXW7 mut solid tumors, especially pts with OC. Further dose optimization of this combination is ongoing. Citation Format: Timothy A. Yap, Rahul Aggarwal, Elisa Fontana, Martin Hojgaard, Benedito A. Carneiro, Linda Duska, Patricia LoRusso, Siddhartha Yadav, Stephanie Lheureux, Cara Mathews, Ryan H. Moy, Nehal Lakhani, Mia Weiss, Joan Tymon-Rosario, Gerardo Colon-Otero, Ignacio Garrido-Laguna, Elisabeth K. Lee, Esteban Rodrigo Imedio, Jonathan Wessen, Luke Piggott, Vito Dozio, Noemie Luong, Christophe Mas, Alison M. Schram. First data disclosure of the Phase I trial of the first in class combination of WEE1 inhibitor zedoresertib with PKMYT1 inhibitor lunresertib in patients with advanced solid tumors harboring CCNE1, FBXW7, or PPP2R1A genomic alterations [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr CT022.