Introduction: ¹⁸F‐fluorodeoxyglucose (FDG)‐PET metabolic patterns of brain differ among autoimmune encephalitis with different neuronal surface antigens. In this case report, we compared the topographical relationship of cerebral glucose metabolism and antigen distribution in the patients with anti‐NMDAR and anti‐AMPAR encephalitis. Literature review summarized the common features of brain metabolism of autoimmune encephalitis.Methods: The cerebral glucose metabolism was evaluated by FDG‐PET/CT during acute‐to‐subacute stage of autoimmune encephalitis and after treatment. The stereo and quantitative analysis of cerebral metabolism used standardized z‐score and visualized on three‐dimensional stereotactic surface projection. To map NMDAR and AMPAR in human brain, we adopted genetic atlases from the Allen Institute and protein atlases from Zilles's receptor densities.Results: The three‐dimensional stereotactic surface projection displayed frontal‐dominant hypometabolism in a 66‐year‐old female patient with anti‐AMPAR encephalitis and occipital‐dominant hypometabolism in a 29‐year‐old female patient with anti‐NMDAR encephalitis. Receptor density maps revealed opposite frontal–occipital gradients of AMPAR and NMDAR, which reflect reduced metabolism in the correspondent encephalitis. FDG‐PET hypometabolic areas possibly represent receptor hypofunction with spatial correspondence to receptor distributions of the autoimmune encephalitis. The reversibility of hypometabolism was in line with patients' cognitive improvement. The literature review summarized six features of metabolic anomalies of autoimmune encephalitis: (a) temporal hypermetabolism, (b) frontal hypermetabolism and (c) occipital hypometabolism in anti‐NMDAR encephalitis, (d) hypometabolism in association cortices, (e) sparing of unimodal primary motor cortex, and (e) reversibility in recovery.Conclusions: The distinct cerebral hypometabolic patterns of autoimmune encephalitis were representative for receptor hypofunction and topographical distribution of antigenic receptors. The reversibility of hypometabolism marked the clinical recovery of autoimmune encephalitis and made FDG‐PET of brain a valuable diagnostic tool.

Objectives: Glymphatic system maintains brain fluid circulation via active transportation of astrocytic aquaporin-4 in perivascular space. The diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) is an established method measuring perivascular glymphatic activity, but comprehensive investigations into its influential factors are lacking. Methods: Community-dwelling older adults underwent brain MRI scans, neuropsychiatric, and multi-domain assessments. Blood biomarker tests included glial fibrillary acidic protein (GFAP) for astrocyte injury. Results: In 71 enrolled participants, the DTI-ALPS index was associated with modifiable factors, including lipid profile (high-density lipoprotein, r = 0.396; very-low-density lipoprotein, r = − 0.342), glucose intolerance (diabetes mellitus, standardized mean difference (SMD) = 0.7662; glycated hemoglobin, r = − 0.324), obesity (body mass index, r = − 0.295; waist, r = − 0.455), metabolic syndrome (SMD = − 0.6068), cigarette-smoking (SMD = − 0.6292), and renal clearance (creatinine, r = − 0.387; blood urea nitrogen, r = − 0.303). Unmodifiable associative factors of DTI-ALPS were age (r = − 0.434) and sex (SMD = 1.0769) (all p < 0.05). A correlation of DTI-ALPS and blood GFAP was noticed (r = − 0.201, one-tailed t-test for the assumption that astrocytic injury impaired glymphatic activity, p = 0.046). Their cognitive correlations diverged, domain-specific for DTI-ALPS (Facial Memory Test, r = 0.272, p = 0.022) but global cognition-related for blood GFAP (MoCA, r = − 0.264, p = 0.026; ADAS-cog, r = 0.304, p = 0.010). Conclusion: This correlation analysis revealed multiple modifiable and unmodifiable association factors to the glymphatic image marker. The DTI-ALPS index correlated with various metabolic factors that are known to increase the risk of vascular diseases such as atherosclerosis. Furthermore, the DTI-ALPS index was associated with renal indices, and this connection might be a link of water regulation between the two systems. In addition, the astrocytic biomarker, plasma GFAP, might be a potential marker of the glymphatic system; however, more research is needed to confirm its effectiveness.

Introduction: Appropriate tools and references are essential for evaluating individuals’ cognitive levels. This study validated the Taiwan version of the Alzheimer’s Disease Assessment Scale-Cognitive subscale (ADAS-cog) and provided normative data for the Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), and ADAS-cog in community-dwelling older adults. Methods: MMSE, MoCA, and ADAS-cog were administered to 150 nondemented healthy adults aged 55–85 years during 2018–2020 as part of the Northeastern Taiwan Community Medicine Research Cohort. ADAS-cog was translated from the original English version to traditional Chinese with cultural and language considerations in Taiwan. Cronbach’s alpha (α) tested the reliability of ADAS-cog, and Pearson correlations examined its external validity using MMSE and MoCA as comparisons. Normative data were generated and stratified by age and education, and the one-way analysis of variance compared scores between age and education groups. Another 20 hospital-acquired participants with cognitive impairment joined the 150 healthy participants. Comparisons in the Clinical Dementia Rating (CDR) tiers tested the discriminability of the tests for different cognitive levels. The area under the receiver operating characteristic curve (AUROC) analyzed the power of ADAS-cog in predicting CDR 0.5 from CDR 0. Results: The Taiwan version of ADAS-cog had fair reliability between items (α = 0.727) and good correlations to MMSE (r = −0.673, p < 0.001) and MoCA (r = −0.746, p < 0.001). The normative data of MMSE, MoCA, and ADAS-cog showed ladder changes with age (p = 0.006, 0.001, and 0.437) and education (p < 0.001, <0.001, and <0.001) in the 150 nondemented older adults. Next, in the 170 mixed participants from the communities and the hospital, MMSE, MoCA, and ADAS-cog scores were well differentiable between CDR 0, 0.5, and 1. In addition, ADAS-cog discriminated CDR 0.5 from 0 by an AUROC of 0.827 (p < 0.001). Discussion/Conclusion: The three structured cognitive tests consistently reflect cognitive levels of healthy older adults. The Taiwan version of ADAS-cog is compatible with MMSE and MoCA to distinguish people with mildly impaired from normal cognition. In addition, this study derived MMSE, MoCA, and ADAS-cog norms tailored to demographic factors. The findings highlight the need for stratification of age and education rather than applying a fixed cutoff for defining normal and abnormal cognition.

Background: Timely and accurate outcome prediction plays a vital role in guiding clinical decisions on acute ischemic stroke. Early condition deterioration and severity after the acute stage are determinants for long-term outcomes. Therefore, predicting early outcomes is crucial in acute stroke management. However, interpreting the predictions and transforming them into clinically explainable concepts are as important as the predictions themselves. Objective: This work focused on machine learning model analysis in predicting the early outcomes of ischemic stroke and used model explanation skills in interpreting the results. Methods: Acute ischemic stroke patients registered on the Stroke Registry of the Chang Gung Healthcare System (SRICHS) in 2009 were enrolled for machine learning predictions of the two primary outcomes: modified Rankin Scale (mRS) at hospital discharge and in-hospital deterioration. We compared 4 machine learning models, namely support vector machine (SVM), random forest (RF), light gradient boosting machine (LGBM), and deep neural network (DNN), with the area under the curve (AUC) of the receiver operating characteristic curve. Further, 3 resampling methods, random under sampling (RUS), random over sampling, and the synthetic minority over-sampling technique, dealt with the imbalanced data. The models were explained based on the ranking of feature importance and the SHapley Additive exPlanations (SHAP). Results: RF performed well in both outcomes (discharge mRS: mean AUC 0.829, SD 0.018; in-hospital deterioration: mean AUC 0.710, SD 0.023 on original data and 0.728, SD 0.036 on resampled data with RUS for imbalanced data). In addition, DNN outperformed other models in predicting in-hospital deterioration on data without resampling (mean AUC 0.732, SD 0.064). In general, resampling contributed to the limited improvement of model performance in predicting in-hospital deterioration using imbalanced data. The features obtained from the National Institutes of Health Stroke Scale (NIHSS), white blood cell differential counts, and age were the key features for predicting discharge mRS. In contrast, the NIHSS total score, initial blood pressure, having diabetes mellitus, and features from hemograms were the most important features in predicting in-hospital deterioration. The SHAP summary described the impacts of the feature values on each outcome prediction. Conclusions: Machine learning models are feasible in predicting early stroke outcomes. An enriched feature bank could improve model performance. Initial neurological levels and age determined the activity independence at hospital discharge. In addition, physiological and laboratory surveillance aided in predicting in-hospital deterioration. The use of the SHAP explanatory method successfully transformed machine learning predictions into clinically meaningful results.

Objective: To test whether strokes increase around the time of cancer diagnosis, we comprehensively examined the correlations of cancer and stroke by employing a population-based cohort study design. Methods: One million people insured under the Taiwan's National Health Insurance program in 2005 were randomly sampled to create the study's dataset. According to the presence of cancer and/or stroke, patients were separated into cancer and stroke, cancer-only, and stroke-only groups. Diagnoses of cancer, stroke, and comorbidities were defined according to ICD9-CM codes. Cancer and non-cancer populations were matched by age at cancer diagnosis, gender, and stroke risk factors, and each patient with cancer was matched with two non-cancer controls nested in the same year of cancer diagnosis. The hazards of stroke and cumulative incidences within a year after cancer diagnosis were evaluated using Fine and Gray's subdistributional hazard model. Results: The temporal distribution of first-ever stroke in patients with both cancer and stroke was a sharpened bell shape that peaked between 0.5 years before and after cancer diagnosis. Frequencies of stroke were further adjusted by number of cancer survivors. The monthly event rate of stroke remained nested around the time of cancer diagnosis in all strokes. Brain malignancies, lung cancer, gastric cancer, prostate cancer, and leukemia patients obtained higher ratio of stroke, while breast cancer and thyroid cancer patients had low percentage of combining stroke. When compared to non-cancer matched control, the hazard of stroke within one year after cancer diagnosis was increased by cancer at a subdistributional hazard ratio of 1.72 (95% confident interval 1.48 to 2.01; p < 0.0001). Conclusions: Cancer increased the risk of stroke and stroke events were nested around the time of cancer diagnosis, occurring 0.5 years prior to cancer on average regardless of stroke type.

Background: Angiotensin receptor blockers (ARBs) may have protective effects against dementia occurrence in patients with hypertension (HTN). However, whether telmisartan, an ARB with peroxisome proliferator-activated receptor γ (PPAR-γ)–modulating effects, has additional benefits compared to other ARBs remains unclear.Methods and findings: Between 1997 and 2013, 2,166,944 type 2 diabetes mellitus (T2DM) patients were identified from the National Health Insurance Research Database of Taiwan. Patients with HTN using ARBs were included in the study. Patients with a history of stroke, traumatic brain injury, or dementia were excluded. Finally, 65,511 eligible patients were divided into 2 groups: the telmisartan group and the non-telmisartan ARB group. Propensity score matching (1:4) was used to balance the distribution of baseline characteristics and medications. The primary outcome was the diagnosis of dementia. The secondary outcomes included the diagnosis of Alzheimer disease and occurrence of symptomatic ischemic stroke (IS), any IS, and all-cause mortality. The risks between groups were compared using a Cox proportional hazard model. Statistical significance was set atp< 0.05. There were 2,280 and 9,120 patients in the telmisartan and non-telmisartan ARB groups, respectively. Patients in the telmisartan group had a lower risk of dementia diagnosis (telmisartan versus non-telmisartan ARBs: 2.19% versus 3.20%; HR, 0.72; 95% CI, 0.53 to 0.97;p= 0.030). They also had lower risk of dementia diagnosis with IS as a competing risk (subdistribution HR, 0.70; 95% CI, 0.51 to 0.95;p= 0.022) and with all-cause mortality as a competing risk (subdistribution HR, 0.71; 95% CI, 0.53 to 0.97;p= 0.029). In addition, the telmisartan users had a lower risk of any IS (6.84% versus 8.57%; HR, 0.79; 95% CI, 0.67 to 0.94;p= 0.008) during long-term follow-up. Study limitations included potential residual confounding by indication, interpretation of causal effects in an observational study, and bias caused by using diagnostic and medication codes to represent real clinical data.Conclusions: The current study suggests that telmisartan use in hypertensive T2DM patients may be associated with a lower risk of dementia and any IS events in an East-Asian population.