Citations

Purpose This study aimed to develop and validate a tool to predict the primary tumor site in patients with metastatic cancer of unknown primary using demographic and clinical variables, including imaging-derived sites of metastasis, without pathology or laboratory data. Methods In this population-based study, data were extracted from the SEER database. Patients diagnosed between 2010 and 2020 with a known primary cancer site, metastatic staging, and metastasis in the bone, liver, brain, and lungs were included. Patients were randomly split into training (70%) and/or testing (30%) sets. A total of 27 primary tumor sites were included. A gradient boosting model was trained to predict the primary tumor site, with random forest and classification tree models evaluated as comparators. Random guess and frequency-weighted guessing models were used for baseline comparison. Results A total of 689,715 patients were included from 9,208,295 queried. The most common primary cancer sites were non–small cell lung cancer (NSCLC) (n = 211,338), intestines (n = 64,187), and pancreas (n = 60,310). The baseline models achieved Top-5 accuracies of 0.179 for a random guess and 0.476 for a frequency-weighted guess. The gradient boosting model achieved Top-1, Top-3, and Top-5 accuracies of 0.40 ± 0.001, 0.680 ± 0.001, and 0.790 ± 0.001, respectively, across 100 Monte-Carlo replicates. Top-5 accuracy was highest for primary NSCLC (0.990 ± 0.001) and intestinal cancer (0.900 ± 0.003). Our model was most influenced by sex (gain = 0.27) and metastasis to bone (gain = 0.20). A web app was constructed: primarycancerpredictor.netlify.app. Conclusion This study developed and evaluated a gradient boosting machine learning model and a user-friendly web tool to aid in the diagnosis of cancer of unknown primary.

Abstract Patients with rare genetic disorders often wait years for a correct diagnosis, highlighting the urgent need for early and efficient triage for those at risk. Electronic health records (EHRs) are rich in useful information, yet the extent to which this data is sufficient to generate prediagnostic signals is uncertain. We trained machine learning models on billing codes and phenotypes extracted from clinical notes and applied these across eight conditions in a longitudinal cohort of roughly 3 million patient records from the Mayo Clinic, seeking early identification of conditions. The number of patients identified early varied by condition, ranging from 10% to 89% at 99% specificity. Median lead times exceeded one year prior to the first diagnostic billing code for most conditions, including Fabry disease (1,408 days), hereditary angioedema (2,755 days), neurofibromatosis type 1 (1,067 days), and hereditary hemorrhagic telangiectasia (2,320 days). We show that ML models trained on EHR data can prioritize patients for clinician review and confirmatory genetic testing and that the integration of diverse phenotypic data sources provides superior predictive value.

Background: Autosomal dominant polycystic kidney disease (ADPKD) is the most common inherited disorder leading to kidney failure. Tolvaptan, the only FDA-approved therapy, modestly slows disease progression but is limited by significant adverse effects, highlighting the need for more effective and better-tolerated treatments. Ciliary polycystins are central to ADPKD pathogenesis. We previously demonstrated that a cilia base-localized pool of CDK6 restricts ciliary polycystin abundance by suppressing axonemal polyglutamylation. However, whether CDK6 inhibition can restore polycystin function and suppress ADPKD progression in vivo remains unknown. Methods: Abemaciclib was tested in in vitro 3D cyst assays and ex vivo metanephric organ cultures. In vivo efficacy of early- and late-stage intervention were assessed in F1(129/B6) Pkd1 R3277C/R3277C mice receiving daily abemaciclib. Tolvaptan served as a therapeutic control. Cyst burden, fibrosis in the kidneys, kidney function, and primary cilia of renal epithelial cells were analyzed at study endpoints. Results: CDK6 was prominently upregulated in cyst-lining cells in human and murine PKD kidneys. At doses substantially lower than those used in oncology, abemaciclib robustly suppressed cyst growth in vitro and ex vivo . In vivo , long-term early treatment arrested cyst progression more effectively than tolvaptan without detectable toxicity. Late intervention largely halted cyst growth in females and reversed cyst progression in males. Mechanistically, abemaciclib restored axonemal polyglutamylation and increased ciliary polycystin levels. Conclusions: Abemaciclib has greater therapeutic efficacy and tolerability than tolvaptan in this preclinical model, largely driven by CDK6 inhibition–mediated restoration of axonemal polyglutamylation and increased ciliary polycystin abundance.

Background. Rare diseases collectively affect an estimated 300 million people worldwide, yet patients wait an average of 4-8 years for a correct diagnosis after visiting multiple physicians, accumulating unexplained findings, and suffering preventable disease progression before a unifying diagnosis is reached. Electronic health records (EHRs) capture this pre-diagnostic trajectory in rich detail, but conventional methods typically collapse years of longitudinal phenotypic signals into a single static snapshot; our approach instead models this temporal structure directly. Methods. We conducted a retrospective cohort study using longitudinal EHR data from approximately 3 million patients at Mayo Clinic Platform_Accelerate. Cases were patients with confirmed diagnoses of 10 rare diseases; controls (~5,200 per disease) had no record of any rare disease. HPO phenotype terms were extracted from clinical notes and laboratory results using a medspaCy NLP pipeline; ICD codes were grouped at three-character level. We developed and trained two temporal models, GRU-Attn and Conformer, that encode patient histories as quarterly time-binned sequences spanning up to 30 years, and compared these against three static baselines (CatBoost, XGBoost, logistic regression). Models were evaluated systematically at prediction horizons h = 1-10 years before diagnosis. Results. In the full held-out cohort benchmark, temporal models outperformed static baselines across the large majority of diseases and prediction horizons, with the advantage widening substantially as the horizon lengthened. At 1-2 years prior to diagnosis, mean AUROC was 0.928 (GRU-Attn) versus 0.871 (CatBoost); at 6-10 years prior to diagnosis, 0.726 versus 0.693, with static models degrading to near-chance for some diseases. Combining HPO and ICD features yielded the strongest performance, with a mean gain of +0.063 AUROC over the best single source at 1-2 years prior to diagnosis. Separately, in a manually curated cohort of 143 confirmed-undiagnosed cases with extended EHR histories, GRU-Attn identified 138 cases before first disease mention and all 143 cases before formal diagnosis at the prespecified 99th-percentile specificity threshold, with median lead times of 1.1-8.3 years and 1.9-21.7 years, respectively. Conclusions. By learning from the temporal trajectory of phenotypic signals rather than their static aggregate, temporal EHR models improved retrospective early detection performance across several rare diseases, most powerfully for conditions where pre-diagnostic findings accrue gradually and heterogeneously over years. These results show that longitudinal EHR trajectories carry rare-disease signal beyond static phenotypic burden and support prospective evaluation of temporal EHR modeling for rare-disease risk stratification.

Abstract The apolipoprotein E ε4 ( APOE4 ) allele is the strongest genetic risk factor for late-onset Alzheimer’s disease (AD), yet the mechanistic underpinnings by which it alters human microglial function remain poorly understood. Here, we utilize CRISPR/Cas9-engineered isogenic human induced pluripotent stem cell-derived microglia-like cells (iMGs) expressing APOE3 or APOE4 to systematically interrogate genotype-dependent effects on microglial proteomes and their secreted extracellular vesicles (EVs) under basal and fibrillar amyloid-β (fAβ)-challenged conditions. Quantitative proteomics revealed that APOE4 iMGs exhibit downregulated RNA processing, impaired phagocytosis, and exacerbated pro-inflammatory signaling, particularly upon fAβ exposure. Functionally, APOE4 iMGs demonstrated increased NF-κB activation, enhanced neuronal apoptosis and tau pathology in 2D and 3D co-culture systems. Interestingly, APOE4 iMGs released greater amount of EVs enriched in proteasomal components over APOE3 iMG. Integrated proteomic analyses further highlighted APOE4 -specific selective packaging of EV cargo linked to neuron-glia communication, inflammatory signaling, and proteostasis dysregulation, suggesting a potent, non-cell-autonomous mechanism driving neuronal damage. Consistently, these APOE4 iMGEVs were preferentially internalized by neurons, leading to disrupted neuronal proteostasis, impaired neuronal firing, and increased tau phosphorylation, particularly under pathological stress. Notably, pharmacologic inhibition of NF-κB in APOE4 iMGs reduced EV cargo-loading of proteasomal molecules and mitigated iMGEV-driven neuronal dysfunction and tau pathology. These findings establish a mechanistic link between APOE4 , dysregulated microglial homeostasis, and microglial EV-mediated propagation of proteostatic dysfunction and neurodegeneration, advancing our understanding of how APOE4 contributes to AD pathogenesis.

Health care organizations are rapidly adopting artificial intelligence (AI)-enabled tools, yet nurse leaders often lack a clear framework to evaluate, govern, and safely scale these technologies. This article presents a practical, system-level AI governance model grounded in risk mitigation, interdisciplinary partnership, and nursing leadership decision rights commonly enacted through nursing professional governance structures. We outline core components of an effective governance structure, including clinical usefulness, safety and reliability, fairness and inclusiveness, transparency and explainability, privacy and security, and organizational accountability. Real-world examples illustrate how nurse executives can guide responsible AI adoption to reduce cognitive load, improve documentation efficiency, and enhance patient outcomes. The framework equips nursing leaders with actionable strategies and practical workflow checklist to lead AI transformation in complex health care environments.

The intestinal epithelium is exposed to diverse combinations of microbiota-derived compounds; however, the mechanisms by which the host integrates these signals remain poorly defined. Studying two highly abundant microbial metabolites, we identified the purine metabolite hypoxanthine as an effector metabolite that directly drives signaling and the short-chain fatty acid butyrate as a regulatory metabolite that conditions host responsiveness. Specifically, hypoxanthine activates the adenosine A1 receptor–TRPC4 axis in enterochromaffin (EC) cells, triggering calcium influx and serotonin release resulting in accelerated gastrointestinal transit locally and increased platelet activation systemically. In contrast, butyrate epigenetically upregulates specific G protein–coupled receptors and ion channels to enhance response to hypoxanthine and the neurotransmitters norepinephrine and dopamine. These findings define a cooperative signaling framework and highlight the role of EC cells as a distinct epithelial signaling hub that senses and integrates microbial metabolite signals to drive physiological responses. Our findings provide a mechanistic foundation for therapeutic strategies that leverage combinatorial microbial signaling.

Background. The VPS35 p.D620N variant causes autosomal dominant Parkinson's disease (PD) and has been shown to activate the LRRK2 kinase pathway, resulting in increased Rab substrate phosphorylation in peripheral immune cells and elevated urinary bis(monoacylglycero)phosphate (BMP) levels. Recently, a VPS35 variant of unknown significance (c.959C>T; p.A320V) was described in two late-onset sporadic PD patients. Methods. We ascertained a family from the Canary Islands in which six siblings were chronically exposed to high doses of pesticides. Three siblings developed levodopa-responsive, akinetic-rigid PD, while the other three remained unaffected. Whole-exome sequencing was performed in the three affected siblings. The frequency of the resulting candidate variant was assessed in 23,327 PD patients and 9,235 controls from four independent cohorts. Members of this pedigree and unrelated controls were assessed for LRRK2 kinase activity in monocytes and neutrophils and BMP levels in urine. Results. The three affected siblings were all heterozygous for p.A320V, whereas the three unaffected siblings did not carry the variant. In the combined PD case-control cohort, p.A320V was identified in six patients and one control. However, unlike p.D620N, heterozygous carrier status for p.A320V was not associated with increased LRRK2 kinase activity or elevated urine BMP levels. Conclusions. While VPS35 p.A320V co-segregated with PD in this family, it did not exhibit the characteristic LRRK2-associated biomarker signature observed in VPS35 p.D620N carriers. It is possible that p.A320V exerts a subtle effect on VPS35 function that was not captured by the assays performed and that chronic pesticide exposure contributed to disease penetrance in this pedigree.