Masayuki Teramoto’s research while affiliated with National Cerebral and Cardiovascular Center and other places

BACKGROUND Coronary heart disease (CHD) is a major cause of morbidity and mortality worldwide. Identifying key risk factors is essential for effective risk assessment and prevention. Machine learning (ML) offers advanced methods for analyzing complex datasets, revealing novel predictors of CHD beyond traditional models. OBJECTIVE This study aims to evaluate the contribution of various risk factors to CHD, focusing on both established and novel markers using machine learning techniques. METHODS The study recruited 7,672 participants aged 30 to 84 years from Suita City, Japan, between 1989 and 1999. Over an average of 15 years, participants were monitored for cardiovascular events. Five ML models—Random Forest (RF), XGBoost, Support Vector Machine (SVM), Logistic Regression (LR), and LightGBM—were used. The optimal model was identified based on accuracy, sensitivity, specificity, and AUC. SHapley Additive exPlanations (SHAP) were then employed to explore the contribution of various risk factors to CHD. RESULTS RF achieved the highest AUC (95% CI) of 0.94 (0.93-0.96), outperforming LR, SVM, XGBoost, and LightGBM. SHAP on the best model identified the top CHD predictors. Intima-media thickness of common carotid artery (IMT_cMax) was identified as the strongest predictor of CHD, highlighting the importance of arterial health. Systolic and diastolic blood pressure, along with lipid profiles (non-HDL cholesterol, HDL cholesterol, and triglycerides), were closely associated with CHD incidence. eGFR underscored the link between renal function and CHD. Novel insights included the impact of lower calcium levels, systemic inflammation (elevated WBC counts), fructosamine levels, and obesity-related factor (body fat percentage). A protective effect in females indicated the need for sex-specific CHD management strategies. CONCLUSIONS ML, particularly the RF model combined with SHAP, effectively identified key risk factors for CHD, including arterial health, blood pressure, lipid profiles, renal function, and novel markers. These findings support a multifactorial approach to CHD risk assessment.

Background: A growing body of evidence has documented unfavorable maternal outcomes attributed to excessive antenatal coffee consumption. Preeclampsia is one of the most common hypertensive disorders of pregnancy with several adverse maternal and neonatal outcomes. However, the association between antenatal coffee consumption and preeclampsia remains debatable. Herein, we performed a systematic review and meta-analysis of available evidence to investigate this association. Methods: After systematically reviewing PubMed and Scopus for eligible studies published until October 2023, we pooled the odds ratios (ORs) and their 95% confidence intervals (CIs) of preeclampsia for women who reported the highest versus the lowest frequencies of antenatal coffee consumption. We used the I² statistic to measure heterogeneity across studies and the funnel plot asymmetry to assess publication bias. Results: This meta-analysis included seven retrospective studies (six case-control studies and one cross-sectional study) investigating 904 women with preeclampsia and 6,257 women without it. Combined, the highest frequencies of antenatal coffee consumption were associated with higher odds of preeclampsia: (pooled OR = 1.39, 95% CI: 1.03, 1.86), with a moderate heterogeneity across studies (I² = 40.34% and p-value for heterogeneity = 0.122) and no publication bias (z = 0.610 and p-value for publication bias = 0.542). However, excluding the cross-sectional study, which contributed to 24.3% of the meta-analysis weight, left the association statistically non-significant: (pooled OR = 1.33, 95% CI: 0.91, 1.95; I² = 44.59%). The association became even weaker after limiting the analysis to studies that excluded women with chronic hypertension: (pooled OR = 1.21, 95% CI: 0.77, 1.89; I² = 41.64%) or after excluding studies with low quality: (pooled OR = 1.24, 95% CI: 0.70, 2.19; I² = 65.79%). Conclusion: The association between antenatal coffee consumption and preeclampsia remains inconclusive. Future prospective cohort studies are needed to better investigate this association.

Background: Serum cholesterols are well-documented markers of cardiovascular diseases; however, their association with cognitive well-being is uncertain. This study investigated the association between serum cholesterol levels and mild-to-moderate cognitive impairment. Methods: Epidemiological evidence on the role of total cholesterol (TC), high-density lipoprotein cholesterol (HDL C), non-HDL-C, and low-density lipoprotein cholesterol (LDL-C) in cognitive impairment was highlighted. Then, data from 6216 Japanese individuals, aged ≥50 years, from the Suita Study were analyzed. Mini-Mental State Examination (MMSE) scores <27 and < 24 were used to define cognitive impairment. Logistic regression was used to calculate the odds ratios (ORs) and 95 % confidence intervals (95 % CIs) for cognitive impairment. Results: Epidemiological studies investigating the association between serum cholesterol and cognitive impairment have shown conflicting findings: elevated risk with certain lipid components in some studies and no association in others. In the Suita Study, HDL-C < 40 mg/dL was associated with cognitive impairment: ORs (95 % CIs) = 1.36 (1.08, 1.72) for MMSE <27 and 1.61 (1.00, 2.60) for MMSE <24. Non-HDL-C ≥ 200 mg/ dL was also associated with cognitive impairment: ORs (95 % CIs) = 1.53 (1.02, 2.31) for MMSE <27 and 1.80 (1.16, 2.79) for MMSE <24. No such associations were detected with TC. Conclusion: While epidemiological evidence remains inconsistent, the Suita Study showed that decreased HDL-C and increased non-HDL-C, but not increased TC, were associated with mild-to-moderate cognitive impairment. Management of serum cholesterol levels should be considered to prevent cognitive impairment.

In this narrative review, we summarized evidence from peer-reviewed articles, published before February 2024, that investigated lifestyle behaviors among children (≤12 years) and adolescents (13-18 years) representing different geographic descents. These behaviors included dietary patterns, screen time, physical activity, smoking, alcohol consumption, oral hygiene, unsafe sex, and sleep duration and quality. We documented the significant impacts of parental, school, and sociodemographic factors on the adoption of numerous lifestyle behaviors in this age category. Several health consequences could be attributed to unhealthy lifestyle behaviors during childhood and adolescence. For example, poor dietary habits can lead to a higher risk of obesity and cardiovascular disease. Physical inactivity contributes to the development of musculoskeletal and psychological disorders. Excessive screen time is associated with visual acuity problems, poor sleep, and psychological and behavioral problems. Tobacco use poses a significant risk for severe respiratory and cardiovascular diseases. Risky sexual behaviors are related to sexually transmitted infections and exposure to violence. Short sleep duration is associated with a lack of physical fitness and poor cognitive function. However, potential interventions, such as school-based health programs, community outreach initiatives, and national health policies and regulations, can improve lifestyle behaviors among children and adolescents.

We leveraged machine learning (ML) techniques, namely logistic regression (LR), random forest (RF), support vector machine (SVM), extreme gradient boosting (XGBoost), and LightGBM to predict coronary heart disease (CHD) and identify the key risk factors involved. Based on the Suita study, 7672 men and women aged 30 to 84 years without cardiovascular disease were recruited from 1989 to 1999, in Suita City, Osaka, Japan. Over an average period of 15 years, participants were diligently monitored until the onset of their initial cardiovascular event or relocation. CHD diagnoses encompassed primary heart attacks, sudden death, or coronary artery disease with bypass surgery or intervention. RF achieved the highest AUC (95% CI) of 0.79 (0.70–0.87), outperforming LR, SVM, XGBoost, and LightGBM. Shapley Additive Explanations (SHAP) on the best model identified the top CHD predictors. Notably, systolic blood pressure, non-HDL-c, glucose levels, age, metabolic syndrome, HDL-c, estimated glomerular filtration rate, hypertension, elbow joint thickness, and diastolic blood pressure were key contributors. Remarkably, elbow joint thickness was identified as a previously unrecognized risk factor associated with CHD. These findings indicated that ML methods accurately predict incident CHD risk. Additionally, ML has identified new incident CHD risk variables.

Background The utility of screening for the degree of common carotid artery (CCA) stenosis as a predictor of cardiovascular disease (CVD) in a general population remains unclear. Methods and Results We studied 4775 Japanese men and women whose CCA was measured using bilateral carotid ultrasonography at baseline (April 1994–August 2001). We calculated the degree of stenosis as a percentage of the stenotic area of the lumen in the cross‐section perpendicular to the long axis. The Cox proportional hazards model was used to calculate multivariable‐adjusted hazard ratios (HRs) with 95% CIs for incident CVD and its subtypes according to the degree of CCA stenosis. During the median 14.2 years of follow‐up, 385 incident CVD events (159 coronary heart disease and 226 stroke) were documented. The degree of CCA stenosis was associated with increased risks of incident CVD, coronary heart disease, and stroke, with multivariable‐adjusted HRs (95% CIs) for <25%, 25%–49%, and ≥50% stenosis with plaque compared with no CCA plaque of 1.37 (1.07–1.76), 1.72 (1.23–2.40), and 2.49 (1.69–3.67), respectively. Adding the CCA stenosis degree to traditional CVD risk factors increased Harrell's C statistics (0.772 [95% CI, 0.751–0.794] to 0.778 [95% CI, 0.758–0.799]; P =0.04) and improved the 10‐year risk prediction ability (integrated discrimination improvement, 0.0129 [95% CI, 0.0078–0.0179]; P <0.001; continuous net reclassification improvement, 0.1598 [95% CI, 0.0297–0.2881]; P =0.01). Conclusions The degree of CCA stenosis may be used as a predictive marker for the development of CVD in the general population.

This study aimed to investigate the association between hypertensive disorders of pregnancy (HDP) and subsequent risk of dementia using a systematic review and meta-analysis of cohort studies. We searched PubMed and Scopus for eligible studies that investigated the association between HDP and dementia risk. Using the random-effects model, pooled hazard ratio (HR) and 95% confidence interval (CI) of dementia risk in women with HDP were calculated. We applied the I2 statistic to measure heterogeneity across studies and the test for funnel plot asymmetry to evaluate publication bias. Six cohort studies were eligible: three from the United States, two from Sweden, and one from Denmark. When combined, HDP was associated with the risk of dementia: pooled HR (95% CI) = 1.31 (1.12, 1.53). The heterogeneity across studies was moderate (I2 = 47.3%, p-heterogeneity = 0.091), but no signs of publication bias were detected. The association of HDP with vascular dementia was stronger than that with Alzheimer's disease: pooled HRs (95% CIs) = 1.66 (1.13, 2.43) and 1.29 (0.97, 1.72), respectively. In conclusion, HDP was associated with a higher risk of dementia and this association was more prominent with vascular dementia.

Background: We previously developed risk models predicting stroke, coronary heart disease (CHD), and cardiovascular disease (CVD) among Japanese people from the Suita Study. Yet, applying these models at the national level was challenging because some of the included risk factors differed from those collected in the Japanese governmental health check-ups, such as Tokutei-Kenshin. We, therefore, conducted this study to develop new risk models for stroke, CHD, and atherosclerotic CVD (ASCVD), based on data from the Suita Study. The new models used traditional cardiovascular risk factors similar to those in the Japanese governmental health check-ups. Methods: We included 7,413 participants, aged 30–84 years, initially free from stroke and CHD. All participants received baseline health examinations, including a questionnaire assessing their lifestyle and medical history, medical examination, and blood and urine analysis. The risk factors of stroke, CHD, and ASCVD (cerebral infarction or CHD) were determined using the multivariable-adjusted Cox regression. The models’ performance was assessed using the C-statistics for discrimination and the Hosmer-Lemeshow for calibration. We also developed three simple scores (zero to 100) that could predict the 10-year incidence of stroke, CHD, and ASCVD. Results: Within 110,428 person-years (median follow-up = 16.6 years), 410 stroke events, 288 CHD events, and 527 ASCVD events were diagnosed. Age, smoking, hypertension, and diabetes were associated with stroke, CHD, and ASCVD risk. Men and those with decreased high-density lipoproteins or increased low-density lipoproteins showed a higher risk of CHD and ASCVD. Urinary proteins were associated with an increased risk of stroke and ASCVD. The C-statistic values of the risk models were >0.750 and the p-values of goodness-of-fit were >0.30. The 10-year incidence of stroke, CVD, and ASCVD events was 3.8%, 3.5%, and 5.7% for scores 45–54, 10.3%, 11.8%, and 19.6% for scores 65–74, and 27.7%, 23.5%, and 60.5% for scores ≥85, respectively. Conclusions: We developed new Suita risk models for stroke, CHD, and ASCVD using variables similar to those in the Japanese governmental health check-ups. We also developed new risk scores to predict incident stroke, CHD, and ASCVD within 10 years.

We aimed to examine the association between hypertension grades and the risk of total and site-specific cancer mortality among Japanese men and women. In the Japan Collaborative Cohort Study, 27,332 participants aged 40–79 years were enrolled and followed up with their mortality until 2009. According to the measured blood pressure (BP) at baseline, we classified the participants into four BP categories based on 2018 European guidelines. The Cox proportional hazard models were used to calculate the multivariable hazard ratios (HRs) with 95% confidence intervals (CIs) of total and site-specific cancer mortality according to the hypertension category. During the 18.5 years of median follow-up, 1,927 cancer deaths were documented. Grade 1 (systolic blood pressure [SBP] 140–159 mmHg or diastolic blood pressure [DBP] 90–99 mmHg) and grade 2–3 hypertension (SBP ≥ 160 mmHg or DBP ≥ 100 mmHg) were associated with an increased risk of total cancer mortality; the multivariable HRs were 1.17 (1.04–1.32) for grade 1, and 1.27 (1.09–1.47) for grade 2–3 hypertension compared to optimal and normal BP (SBP < 130 mmHg and DBP < 85 mmHg). Linear and positive associations were observed between SBP and DBP 10 mmHg increment and the risk of total cancer mortality; HRs were 1.06 (1.03–1.08) for SBP and 1.07 (1.02–1.11) for DBP of 10 mmHg increment. The excess risk was primarily found for esophageal, liver, and pancreatic cancer; the respective multivariable HRs of grade 2–3 hypertension vs optimal and normal BP were 2.57 (1.10–6.04) for esophageal, 1.67 (1.01–2.77) for liver, and 1.95 (1.17–3.23) for pancreatic cancer. Hypertension was associated with the increased risk of total cancer mortality, primarily of esophageal, liver, and pancreatic cancer.