Citations

Background: Post-treatment prognosis and monitoring are critical for determining the timing of salvage treatment in glioblastoma patients but has been challenging due to difficulties differentiating progression from treatment effects in conventional images. This exploratory study aimed to establish the correlation of radiomics image features from time series of amino acid tracer 18F-DOPA PET images, with outcomes, using machine learning and dimension reduction analysis. Methods: 18F-DOPA PET images were collected for a patient cohort with wild-type IDH and unmethylated MGMT who underwent dose-escalated radiation therapy. Quantitative features were derived from the high uptake region (T/N > 2.0) in pre- and post-radiation therapy follow-up 18F-DOPA PET images. A customized workflow was utilized for pre-selecting predictive features, followed by manifold learning. Machine learning algorithms were employed to establish associations between imaging features and remaining survival (RS), defined as the time between a follow-up scan and date of death. Results: The ML models exhibited 81–83% ROC_AUC in predicting RS evaluated on an independent test dataset. A RS map is proposed for monitoring tumor alterations through serial 18F-DOPA PET scans, demonstrating superior sensitivity and better correlation with survival compared to the RANO criteria. Conclusions: Our study demonstrates that ML models utilizing FU 18F-DOPA PET images have the potential to effectively predict future survival outcomes in patients with glioblastoma treated with dose-escalated radiation therapy. The capability to assess changes in tumor over time through imaging can potentially assist in patient stratification and the selection of salvage treatments, while also aiding in distinguishing treatment effects from genuine tumor progression.

Abstract Background Few studies have assessed the burden of chronic medical conditions in adolescent and young adult (AYA) cancer survivors. This study estimated the risk of these conditions among AYA survivors compared to a matched cohort without cancer. Methods This retrospective cohort study included 2‐year survivors ( n = 14,917) of 11 common AYA (15–39 years) cancers diagnosed at the integrated health care organizations of Kaiser Permanente (KP) Southern and Northern California during 2006–2020. A comparison cohort ( n = 149,164) without cancer (matched 10:1 by age, sex, calendar year, and KP site) was included. Cumulative incidence (CMI) accounting for death as a competing risk was calculated. Poisson regression estimated the incidence rate ratio (IRR) of each condition in cancer survivors versus the matched cohort, adjusting for age, sex, and race/ethnicity. Results The 5‐year CMI was highest for thyroid (17.4%), respiratory (6.6%), cardiovascular (5.0%), and liver (4.8%) diseases. At 10 years, the CMI of any condition was 39% in survivors versus 26% in the matched cohort. Survivors had a 2‐fold increased risk of being diagnosed with any medical condition (IRR, 2.0; 95% confidence interval [CI], 1.9–2.0) as well as two or more conditions (IRR, 2.3; 95% CI, 2.2–2.5). Risk was highest among survivors of hematologic cancers and those diagnosed with distant stage disease. Elevated risks were observed within all sociodemographic groups of this insured population. Conclusion AYAs with cancer had a higher risk of chronic medical conditions compared to those without cancer. Long‐term surveillance, risk mitigation through lifestyle modifications and effective disease management are crucial to reduce premature mortality.

Importance The definition of cure in stage II to III colon cancer (CC) remains unclear due to limitations in conventional end points, which include deaths and second primary tumors as events. These can complicate communication with patients regarding long-term outcomes. Objective To distinguish relapses from competing health-related events to classify long-term outcomes years after surgery and explore when the incidence of true relapses of the initial CC approaches 0% to define cure in this disease. Design, Setting, and Participants This pooled analysis of individual patient–level data from 15 phase 3 randomized clinical trials assessed time to CC-related relapse using Kaplan-Meier and Aalen-Johansen methods, with death and second primary tumors treated as competing risks. Cox regression models evaluated prognostic associations, stratified by sex, stage, and tumor. Patients with stage II to III CC who underwent adjuvant chemotherapy were included. All patients had undergone radical surgery for CC and received adjuvant chemotherapy with a median follow-up of at least 6 years. The Adjuvant Colon Cancer Endpoints (ACCENT) and the International Duration Evaluation of Adjuvant Chemotherapy (IDEA) databases included adjuvant studies conducted between 1996 and 2015. Data were analyzed from February 2022 to June 2025. Exposures Adjuvant chemotherapy regimens varied across trials, including fluoropyrimidines alone or in combination with oxaliplatin or biologic agents. Main Outcomes and Measures The primary outcome was time to CC-related recurrence. The predefined threshold for cure was a recurrence risk below 0.5%. Results Of 35 213 included patients, 19 346 (54.9%) were male, and the mean (SD) age was 60.2 (10.8) years. The incidence rate of recurrence peaked at 6.4% (1993 of 31 373) between month 6 and month 12 and decreased continuously until year 10 of follow-up never exceeding 0.5%. Recurrence rate appeared to increase again after year 10 and peaked at 2.0% during year 12.5 to year 13, a pattern observed exclusively in the MOSAIC trial. Competing-event analysis revealed that death and second primary tumors inflated the apparent recurrence rate, especially for older patients. The overall cumulative incidence of relapse with death as competing risk was lower among female patients (hazard ratio, 0.58; 95% CI, 0.45-0.76; P < .001). Conclusions and Relevance In this pooled analysis of phase 3 randomized clinical trials, a recurrence rate less than 0.5% occurred after 6 years from surgery, supporting a practical definition of cure. Recognizing this milestone may improve patient communication, guide follow-up duration, and reduce unnecessary long-term surveillance.

Abstract With the rapid advancements in large language model technology and the emergence of bioinformatics‐specific language models (BioLMs), there is a growing need for a comprehensive analysis of the current landscape, computational characteristics, and diverse applications. This survey aims to address this need by providing a thorough review of BioLMs, focusing on their evolution, classification, and distinguishing features, alongside a detailed examination of training methodologies, datasets, and evaluation frameworks. We explore the wide‐ranging applications of BioLMs in critical areas such as disease diagnosis, drug discovery, and vaccine development, highlighting their impact and transformative potential in bioinformatics. We identify key challenges and limitations inherent in BioLMs, including data privacy and security concerns, interpretability issues, biases in training data and model outputs, and domain adaptation complexities. Finally, we highlight emerging trends and future directions, offering valuable insights to guide researchers and clinicians toward advancing BioLMs for increasingly sophisticated biological and clinical applications.

Abstract Introduction Pancreatic ductal adenocarcinoma (PDAC) remains one of the deadliest cancers. While total neoadjuvant therapy (TNT) has improved outcomes, recurrence rates remain high. Major pathologic response after TNT is prognostic of better survival, but is rarely achieved. Current pathologic systems primarily consider residual tumor percentage, nodal status, and perineural invasion but overlook spatial organization within the tumor microenvironment (TME). We aimed to evaluate spatial features of the TME to identify biologically and clinically relevant patterns of TME conformation and configuration after TNT and surgical resection of PDAC. Methods We retrospectively analyzed patients with PDAC treated with TNT followed by surgery at Mayo Clinic (2010-2022) with partial or no response per College of American Pathologists (CAP) criteria. Hematoxylin and eosin-stained whole slide images (WSIs) were processed using PathExplore™ PDAC (v1.0) to quantify spatial TME features. These were correlated with disease-free survival (DFS). Results A total of 203 patients were included: 47.3% female, median age 66.5 years. All received TNT; 73.4% had neoadjuvant chemotherapy followed by chemoradiation. Surgical procedures included pancreaticoduodenectomy (62.1%), distal pancreatectomy (22.2%), and total pancreatectomy (15.8%). CAP-defined response was partial in 82.1% and absent in 17.9%. As expected, pathologic T stage (p=0.001), N stage (p=0.002), lymphovascular invasion (p=0.048), and R2 resection (p=0.007) were significantly associated with DFS. However, CAP response category (p=0.053), residual cancer area (p=0.47), and cancer-to-stroma ratio (p=0.25) were not. We analyzed 297 WSIs from 203 patients. Of 31 spatial features evaluated, 5 were associated with DFS (p<0.20). Recursive partitioning using ln(mean cancer shape index [CSI]) <1.775 and ln(SD of stroma shape index [SSI]) >0.651 identified a high-risk group (n=63) with median DFS of 7.23 months (95% CI: 5.19–9.63), compared to 11.57 months in the low-risk group (95% CI: 9.24–17.92) (HR=1.82; 95% CI: 1.31–2.52; p=0.0003). This remained significant after adjusting for T and N stage (HR=1.75; 95% CI: 1.17–2.63; p=0.007). An alternate prognostic index derived via backwards elimination, calculated using 0.2701(mean stromal area) + 0.3905(edge density), also stratified patients. High-risk patients (n=51) had a significantly shorter (p<0.001) DFS (6.2 months) than those in the low-risk category (20.0 months). This significant difference holds in a multivariable analysis controlling for pathologic stage, N stage, and resection margins (HR=1.94, 95% CI: 1.21-3.10, p=0.006). Discussion Current assessment of residual PDAC after TNT is limited by subjective grading and lack of spatial context. Using Tumor Landscape Analysis, we identified high-risk spatial features of the TME that independently predict recurrence. These findings support a novel spatially-informed framework for risk stratification in resected PDAC with residual disease. Citation Format: Miranda Lin, Luis Cisneros, Merih D. Toruner, Andrea Maraone, Paul Dizona, Qian Shi, Martin E. Fernandez-Zapico, Ryan M. Carr. Spatial features of the tumor microenvironment predict recurrence risk in resected pancreatic ductal adenocarcinoma after total neoadjuvant therapy [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: Advances in Pancreatic Cancer Research—Emerging Science Driving Transformative Solutions; Boston, MA; 2025 Sep 28-Oct 1; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2025;85(18_Suppl_3):Abstract nr A051.

Abstract Background: Spatial Reasoning AI (SRAI) systems are emerging as a transformative class of clinical artificial intelligence, enabling real-time interpretation of complex anatomy in imaging, surgery, and chronic disease management. Unlike traditional AI approaches constrained to tabular or 2D inputs, SRAI models incorporate geometric priors, multimodal fusion, and 3D anatomical reasoning to support high-impact tasks such as volumetric segmentation, lesion localization, intraoperative guidance, and prognostic modeling. Objective: This scoping review aimed to (1) map the emerging landscape of SRAI in healthcare, (2) assess its clinical performance across diagnostic, prognostic, and workflow applications, and (3) identify opportunities and barriers for safe translation into practice. Methods: Following PRISMA 2020 guidelines, we systematically searched PubMed, Springer Nature, and IEEE Xplore (including CVPR, ICCV, ISBI, and EMBC) for original studies implementing spatially aware AI models with clinical validation or workflow relevance. Data extraction captured model architecture, input modality, anatomical scope, application domain, and performance outcomes. Results: We identified eligible studies spanning oncology, neurology, cardiology, surgery, and infectious diseases. SRAI systems consistently outperformed conventional methods in segmentation, lesion classification, and risk stratification. A paradigm shift is underway from task-specific CNNs to foundation-level models, including MedSAM and MedCLIP for zero-shot segmentation and retrieval, and NeRF-based pipelines for anatomically faithful 3D reconstruction. Randomized evidence shows that SRAI can improve workflow efficiency without compromising diagnostic accuracy. Conclusion: SRAI models demonstrate strong potential to advance diagnostic precision, prognostic accuracy, and workflow integration across multiple clinical domains. Their safe adoption will require standardized evaluation metrics, prospective multi-institutional validation, and interpretable outputs. As foundation models evolve to better encode spatial priors, SRAI is poised to become a cornerstone of workflow-integrated precision medicine.

Abstract Background Completing adjuvant chemotherapy and reducing toxicities are critical for maximizing survival after a colon cancer diagnosis. Sex, as a biological variable, may affect colon cancer chemotherapy completion, toxicities, and survival differently. Methods From a National Cancer Institute–sponsored trial conducted among patients with stage III colon cancer (Cancer and Leukemia Group B (Alliance)/Southwest Oncology Group 80702), we included 2201 patients receiving the standard adjuvant chemotherapy regimen of fluorouracil, leucovorin, and oxaliplatin. We calculated relative dose intensity to indicate chemotherapy completion and considered reduced relative dose intensity (values <85%) as a clinically significant deviation from standard fluorouracil, leucovorin, and oxaliplatin. Using National Cancer Institute’s Common Terminology Criteria for Adverse Events, we defined severe adverse events (grade ≥3) as the occurrence of any following event, including neutrophil count decrease, nausea, platelet count decrease, hypertension, peripheral neuropathy, diarrhea, fatigue, gastritis, creatinine level increase, gastric ulcer, myocardial ischemia, and cerebral ischemia. The primary survival outcome was disease-free survival (time from enrollment to colon cancer recurrence or death from any cause), and secondary survival outcomes were recurrence-free and overall survival. Results Compared with men, women were at statistically significantly higher risks of experiencing reduced relative dose intensity (adjusted odds ratio = 1.59, 95% CI = 1.29 to 1.96; P < .001) and severe advance events (adjusted odds ratio = 1.72, 95% CI = 1.41 to 2.11; P < .001). Yet, women had statistically significantly better disease-free survival (adjusted hazard ratio = 0.72, 95% CI = 0.59 to 0.87; P < .001) as well as better recurrence-free and overall survival. Conclusions Our findings suggested that women with colon cancer are more likely to have worse chemotherapy completion rates and more severe adverse events, but they have better survival. Sex as a biological variable warrants further consideration in chemotherapy administration and survivorship management after colon cancer diagnosis.

Abstract Background: Spatial Reasoning AI (SRAI) systems are reshaping clinical practice by enabling real-time interpretation of complex anatomical structures in medical imaging, surgery, and chronic disease management (Lin et al., 2023, Qiao et al., 2021; Song et al., 2024) . Unlike traditional AI approaches that rely on tabular or 2D data, SRAI models incorporate geometric priors, multimodal fusion, and 3D anatomical reasoning to support high-impact tasks such as volumetric segmentation, disease localization, and intraoperative guidance (Milletari et al., 2016; Hatamizadeh et al., 2022; Zhang et al., 2022) . Objective: This scoping review aimed to (1) characterize the emerging landscape of spatial reasoning models in healthcare, (2) evaluate their clinical performance across diagnostic and prognostic applications, and (3) identify architectural trends that underpin successful translation into practice. Methods: Following PRISMA guidelines, we systematically screened PubMed to identify studies that implemented spatially aware AI models with real-time or clinically validated outcomes (Page et al., 2021) . Data extraction captured model architecture, modality, anatomical scope, clinical application, and performance metrics. Results: We identified 124 eligible studies across the clinical domains of oncology, neurology, cardiology, orthopedic surgery, and infectious diseases (Nie et al., 2019; Hao et al., 2020; Tasken et al., 2023) . SRAI systems consistently outperformed conventional methods in segmentation, lesion classification, and risk stratification. (Lin et al., 2023; Qiao et al., 2021; Mahootiha et al., 2024) . A paradigm shift from task-specific CNNs to foundation models that reason across spatial structures has emerged, beginning with adaptations of generalist models such as MedSAM and MedCLIP for zero-shot segmentation and cross-modal retrieval, and NeRF-based reconstruction for anatomically faithful 3D modeling (Ma et al., 2024; Mildenhall et al., 2020; Radford et al., 2021). Beyond technical performance, randomized evidence shows that SRAI systems can improve workflow efficiency without compromising outcomes (Lang et al., 2023) . Conclusion: SRAI models hold promise for enhancing diagnostic accuracy, prognostic precision, and clinical efficiency across diverse applications. Their successful integration will depend on standardized evaluation metrics, robust validation, and interpretable outputs. As foundation models evolve to better encode spatial priors, SRAI is poised to become a cornerstone of personalized and procedural medicine.

AbstractBackgroundInfants born via Cesarean section (C‐section) often have a distinct gut microbiome and higher risks of atopic and immune‐related conditions than vaginally delivered infants. We evaluated whether a microbiome‐based program could shift gut microbiome composition and improve microbiome‐associated health outcomes in C‐section born infants.MethodsThis open‐label, randomized, controlled trial included full‐term C‐section‐born infants aged 0–3 months, randomized to an intervention (n = 25) or control arm (n = 29). Over 6 months, the intervention arm received two microbiome reports, personalized recommendations based on their microbiome, educational materials, and coaching calls focused on microbiome health. Parents reported health conditions via surveys. Primary outcome: Difference between study arms in relative abundance of key gut microbiome taxa and functional genes. Other outcomes: Changes in a C‐section index—a taxonomy‐based metric comparing C‐section‐associated taxa to vaginally‐associated taxa—and prevalence of atopic conditions.ResultsCompared to controls, the intervention arm had higher Bifidobacterium (p = .025, q = .121) and higher abundance of genes associated with human milk oligosaccharide degradation (e.g., α‐L‐fucosidase, p = .019, q = .046) at timepoint 2. In the intervention arm, the C‐section index decreased to a level similar to vaginally born infants (p = .807, q = .807). At the end of the intervention, atopic dermatitis prevalence was lower in the intervention arm than in controls (odds ratio, 0.17 [95% CI, 0.023–0.723], p = .031).ConclusionA personalized microbiome‐based program can modulate the gut microbiome of C‐section‐born infants and may reduce the risk of atopic conditions (ClinicalTrials.gov: NCT06424691).