Citations

Abstract INTRODUCTION Prospective studies of blood‐based biomarkers reflecting pathogenic processes, other than Alzheimer‐specific pathologies, that contribute to dementia in diverse cohorts are lacking. METHODS We jointly fitted linear mixed‐effect models and proportional hazards models to examine associations of midlife glial fibrillary acidic protein (GFAP), chitinase‐3‐like protein 1 (YKL‐40), soluble cluster of differentiation‐14 (sCD14), neurofilament light chain (NfL), total tau (t‐tau), and ubiquitin C‐terminal hydrolase L1 (UCHL1) levels in serum (collected in 1993 to 1995) with cognitive decline and incident dementia (ascertained over 29 years through 2022) in the community‐based Atherosclerosis Risk in Communities Study. sCD14 and YKL‐40 were measured in 3082 participants and the other four biomarkers in 1766 participants. RESULTS Higher serum GFAP, YKL‐40, sCD14, and NfL were associated with faster cognitive decline and elevated dementia rate (e.g., 1 standard deviation [SD] higher log‐base2 YKL‐40 was associated with −0.11SD faster 25‐year cognitive decline (95% confidence interval [CI]: −0.15,−0.07) and 45% higher dementia hazard (hazard ratio [HR]: 1.45 [95% CI: 1.25, 1.68]). Higher t‐tau and UCHL1 were also associated with faster cognitive decline. DISCUSSION Midlife blood biomarkers reflecting glial and neuronal dysfunction and neuroinflammation are associated with early cognitive impairment.

Abstract Impaired kidney function is a risk factor for cognitive decline and dementia, but most studies use serum creatinine-based equations to estimate kidney function as estimated glomerular filtration rate (eGFR); these equations include age, also a strong driver of cognitive impairment, and produce eGFR estimates with substantial error compared to gold-standard objective measures of GFR (mGFR). The purpose of this study was to determine eGFR and mGFR associations with cognitive function in African American adults, a population disproportionately impacted by both chronic kidney disease and dementia. We conducted a cross-temporal analysis of 368 participants (Age: 66.4 ± 7.4y, 75% Female) in the Genetic Epidemiology Network of Arteriopathy study with eGFR (CKD-EPI 2021 Creatinine Race-Free Equation) and mGFR assessed. eGFR was assessed using the CKD-EPI 2021 Creatinine Race-Free Equation and mGFR (urinary clearance of non-radiolabeled iothalamate). Memory (Rey Auditory Verbal Learning Test; RAVLT) and executive function (Digit Symbol Substitution Test; DSST) were assessed. Linear regression models were used to relate eGFR, mGFR, and serum creatinine to cognitive function with and without adjustment for age and sex. In unadjusted models, eGFR was positively associated with DSST scores (β = 0.19, 95%CI:0.11, 0.27) and RAVLT scores (β = 0.10, 95%CI:0.04, 0.16). mGFR was marginally associated with DSST scores (β = 0.07, 95%CI:-0.00, 0.14), but not RAVLT scores (β = 0.03, 95%CI:-0.01, 0.07). Associations of serum creatinine with cognitive function were mostly null to marginal. Age and sex adjustment greatly attenuated all associations. The observed differential renal function associations with cognition suggest age may largely explain prior reported associations.

Abstract Background Prospective investigations on blood biomarkers for Alzheimer's disease and related‐dementias (ADRDs) and cognitive impairment provide important insights into their clinical utility in early disease diagnosis and risk stratification, but have been primarily limited to AD pathologies (e.g. amyloid‐beta and hyperphosphorylated‐tau aggregation). Studies of biomarkers reflecting other pathogenic processes that contribute to ADRDs, especially in community‐based cohorts, are lacking. We examined associations of blood biomarkers reflecting glial and neuronal dysfunction and neuroinflammation, including soluble cluster of differentiation‐14 (sCD14), chitinase‐3‐like protein‐1 (YKL‐40), glial‐fibrillary acidic protein (GFAP), neurofilament‐light chain (NfL), total‐Tau, and Ubiquitin C‐terminal hydrolase L1 (UCH‐L1) with cognitive decline and incident dementia in ARIC study participants. Methods Blood biomarkers were measured in 1,864 ARIC Visit 3 (1993‐1995) participants using the Meso Scale Discovery assay (YKL‐40), the Quanterix Simoa Neurology 4‐Plex assay (total‐Tau, NfL, GFAP, UCH‐L1), and a commercial ELISA assay (sCD14, R&D Systems); global cognition z‐scores were derived from three cognitive tests (measuring memory, language, and executive function) administered up to six times over 25 years, and incident dementia cases were ascertained via in‐person cognitive assessments (expert‐adjudicated), informant‐based interviews, and hospitalization and death records through 2018. We used shared parameter models to simultaneously fit linear mixed‐effect models and proportional hazard models to estimate 25‐year cognitive decline and dementia rates separately for each biomarker (log‐base2‐transformed), adjusting for covariates (Table). Results Among 1,864 participants (mean±SD age:63±4.5 years, 60% female, 49% self‐reported Black), 27% developed dementia. Higher levels of each biomarker were associated with faster cognitive decline. All but UCH‐L1 were associated with elevated dementia risk (Table). For example, one standard deviation (SD) higher log‐base2 YKL‐40 was associated with ‐0.10 SD steeper 25‐year cognitive decline (95%CI:‐0.15,‐0.04) and 59% higher dementia hazard (hazard ratio (HR):1.59 (95%CI:1.20,2.10); likewise, sCD14 was associated with an additional cognitive decline of ‐0.06 (95%CI:‐0.11,‐0.01) and 38% higher dementia hazard (HR:1.38 (95%CI:1.08,1.76)). Conclusions Blood biomarkers reflecting neuropathological processes, including microglial activation, astrocytic reactivity, and neuronal degeneration, were associated with faster cognitive decline and increased dementia risk, suggesting that they may aid in risk stratification and intervention strategies for ADRDs. Future analyses will examine whether these associations differ by race and APOE‐ε4 s tatus.

Abstract Objectives Heart failure (HF) impacts millions of patients worldwide, yet the variability in treatment responses remains a major challenge for healthcare professionals. The current treatment strategies, largely derived from population based evidence, often fail to consider the unique characteristics of individual patients, resulting in suboptimal outcomes. This study aims to develop computational models that are patient-specific in predicting treatment outcomes, by utilizing a large Electronic Health Records (EHR) database. The goal is to improve drug response predictions by identifying specific HF patient subgroups that are likely to benefit from existing HF medications. Materials and Methods A novel, graph-based model capable of predicting treatment responses, combining Graph Neural Network and Transformer was developed. This method differs from conventional approaches by transforming a patient's EHR data into a graph structure. By defining patient subgroups based on this representation via K-Means Clustering, we were able to enhance the performance of drug response predictions. Results Leveraging EHR data from 11 627 Mayo Clinic HF patients, our model significantly outperformed traditional models in predicting drug response using NT-proBNP as a HF biomarker across five medication categories (best RMSE of 0.0043). Four distinct patient subgroups were identified with differential characteristics and outcomes, demonstrating superior predictive capabilities over existing HF subtypes (best mean RMSE of 0.0032). Discussion These results highlight the power of graph-based modeling of EHR in improving HF treatment strategies. The stratification of patients sheds light on particular patient segments that could benefit more significantly from tailored response predictions. Conclusions Longitudinal EHR data have the potential to enhance personalized prognostic predictions through the application of graph-based AI techniques.

Key Points Using multiple markers may improve GFR estimation especially in settings where creatinine and cystatin C are known to be limited. Panel eGFR is a novel multimarker eGFR equation consisting of age, sex, cystatin C, and nuclear magnetic resonance–measured creatinine, valine, and myo-inositol. eGFR-Cr and eGFR-Cr-CysC may underestimate measured GFR, while panel eGFR was unbiased among younger Black male individuals. Background Using multiple markers may improve accuracy in GFR estimation. We sought to externally validate and compare the performance of a novel multimarker eGFR (panel eGFR) equation among Black and White persons using the Genetic Epidemiology Network of Arteriopathy cohort. Methods We included 224 sex, race/ethnicity, and measured GFR (mGFR) category–matched persons, with GFR measured using urinary clearance of iothalamate. We calculated panel eGFR using serum creatinine, valine, myo-inositol, cystatin C, age, and sex. We compared its reliability with current eGFR equations (2021 CKD Epidemiology Collaboration creatinine [eGFR-Cr] and creatinine with cystatin C [eGFR-Cr-CysC]) using median bias, precision, and accuracy metrics. We evaluated each equation's performance in age, sex, and race subgroups. Results In the overall cohort, 49% were Black individuals, and mean mGFR was 79 ml/min per 1.73 m 2 . Panel eGFR overestimated mGFR (bias: −2.4 ml/min per 1.73 m 2 ; 95% confidence interval [CI], −4.4 to −0.7), eGFR-Cr-CysC underestimated mGFR (bias: 4.8 ml/min per 1.73 m 2 ; 95% CI, 2.1 to 6.7), while eGFR-Cr was unbiased (bias: 2.0 ml/min per 1.73 m 2 ; 95% CI, −1.1 to 4.6). All equations had comparable accuracy. Among Black male individuals younger than 65 years, both eGFR-Cr (bias: 17.0 ml/min per 1.73 m 2 ; 95% CI, 8.6 to 23.5) and eGFR-Cr-CysC (bias: 14.5 ml/min per 1.73 m 2 ; 95% CI, 6.0 to 19.7) underestimated mGFR, whereas panel eGFR was unbiased (bias: 1.7 ml/min per 1.73 m 2 ; 95% CI, −3.4 to 10.0). Metrics of accuracy for all eGFRs were acceptable in all subgroups except for panel eGFR in Black female individuals younger than 65 years (P30: 73.3%). Conclusions Panel eGFR can be used to estimate mGFR and may have utility among Black male individuals younger than 65 years where current CKD Epidemiology Collaboration equations are biased.

Abstract Background Alzheimer’s disease (AD) is a progressive neurological disorder with no specific curative medications. Sophisticated clinical skills are crucial to optimize treatment regimens given the multiple coexisting comorbidities in the patient population. Objective Here, we propose a study to leverage reinforcement learning (RL) to learn the clinicians’ decisions for AD patients based on the longitude data from electronic health records. Methods In this study, we selected 1736 patients from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database. We focused on the two most frequent concomitant diseases—depression, and hypertension, thus creating 5 data cohorts (ie, Whole Data, AD, AD-Hypertension, AD-Depression, and AD-Depression-Hypertension). We modeled the treatment learning into an RL problem by defining states, actions, and rewards. We built a regression model and decision tree to generate multiple states, used six combinations of medications (ie, cholinesterase inhibitors, memantine, memantine-cholinesterase inhibitors, hypertension drugs, supplements, or no drugs) as actions, and Mini-Mental State Exam (MMSE) scores as rewards. Results Given the proper dataset, the RL model can generate an optimal policy (regimen plan) that outperforms the clinician’s treatment regimen. Optimal policies (ie, policy iteration and Q-learning) had lower rewards than the clinician’s policy (mean −3.03 and −2.93 vs. −2.93, respectively) for smaller datasets but had higher rewards for larger datasets (mean −4.68 and −2.82 vs. −4.57, respectively). Conclusions Our results highlight the potential of using RL to generate the optimal treatment based on the patients’ longitude records. Our work can lead the path towards developing RL-based decision support systems that could help manage AD with comorbidities.

Abstract Background and Aims The use of multiple markers may improve the accuracy in glomerular filtration rate (GFR) estimation especially when the interpretation of creatinine and cystatin C is limited. We sought to externally validate a recently developed multi-marker nuclear magnetic resonance (NMR)-based estimated GFR equation (eGFR-NMR) using the biracial Genetic Epidemiology Network of Arteriopathy cohort. Method We included 224 sex, race/ethnicity, and mGFR-category-matched participants. GFR was measured using urinary clearance of iothalamate (mGFR). We calculated eGFR-NMR using serum creatinine, valine, myo-inositol, and cystatin C, age, and sex. We compared the reliability of eGFR-NMR with current eGFR equations (2021 Chronic Kidney Disease Epidemiology Collaboration equations for creatinine [eGFR-Cr] and creatinine with cystatin C [eGFR-Cr-CysC]) using median bias, precision, and accuracy metrics. In particular, we evaluated its performance in age, sex, and race subgroups. Results In the overall cohort, mean age was 63 (±8) years, 54% were females, 49% were Black individuals, and mean mGFR was 78.7 (±24.3) ml/min/1.73 m2. eGFR-NMR overestimated mGFR by 2 mL/min/1.73 m2 (95% CI, 4 to 0.7) while eGFR-Cr-CysC underestimated mGFR by −5 mL/min/1.73 m2 (95% CI, −2 to −7). All equations had acceptable accuracy metrics. When stratified by age, sex, and race, eGFR-NMR performed the best among Black males age <65 years compared to current equations (Figure 1). In this subgroup, eGFR-NMR was unbiased (bias, 2mL/min/1.73 m2 [95% CI, -3 to 10]) compared to substantial biases of eGFR-Cr (bias, 17 mL/min/1.73 m2 [95% CI, 9 to 24]) and eGFR-Cr-CysC (bias, 15 mL/min/1.73 m2 [95% CI, 6 to 20]). In other subgroups, measures of accuracy for eGFR-NMR, eGFR-Cr, eGFR-Cr-CysC were generally similar. Conclusion eGFR-NMR can be used to estimate mGFR and was more accurate than CKD-EPI equations among Black males age <65 years.