Nontargeted Metabolomics Revealed Novel Association Between Serum Metabolites and Incident Acute Coronary Syndrome: A Mendelian Randomization Study.

Background This study was performed to identify metabolites associated with incident acute coronary syndrome (ACS) and explore causality of the associations. Methods and Results We performed nontargeted metabolomics in a nested case-control study in the Dongfeng-Tongji cohort, including 500 incident ACS cases and 500 age- and sex-matched controls. Three metabolites, including a novel one (aspartylphenylalanine), and 1,5-anhydro-d-glucitol (1,5-AG) and tetracosanoic acid, were identified as associated with ACS risk, among which aspartylphenylalanine is a degradation product of the gut-brain peptide cholecystokinin-8 rather than angiotensin by the angiotensin-converting enzyme (odds ratio [OR] per SD increase [95% CI], 1.29 [1.13-1.48]; false discovery rate-adjusted P=0.025), 1,5-AG is a marker of short-term glycemic excursions (OR per SD increase [95% CI], 0.75 [0.64-to 0.87]; false discovery rate-adjusted P=0.025), and tetracosanoic acid is a very-long-chain saturated fatty acid (OR per SD increase [95% CI], 1.26 [1.10-1.45]; false discovery rate-adjusted P=0.091). Similar associations of 1,5-AG (OR per SD increase [95% CI], 0.77 [0.61-0.97]) and tetracosanoic acid (OR per SD increase [95% CI], 1.32 [1.06-1.67]) with coronary artery disease risk were observed in a subsample from an independent cohort (152 and 96 incident cases, respectively). Associations of aspartylphenylalanine and tetracosanoic acid were independent of traditional cardiovascular risk factors (P-trend=0.015 and 0.034, respectively). Furthermore, the association of aspartylphenylalanine was mediated by 13.92% from hypertension and 27.39% from dyslipidemia (P<0.05), supported by its causal links with hypertension (P<0.05) and hypertriglyceridemia (P=0.077) in Mendelian randomization analysis. The association of 1,5-AG with ACS risk was 37.99% mediated from fasting glucose, and genetically predicted 1,5-AG level was negatively associated with ACS risk (OR per SD increase [95% CI], 0.57 [0.33-0.96], P=0.036), yet the association was nonsignificant when further adjusting for fasting glucose. Conclusions These findings highlighted novel angiotensin-independent involvement of the angiotensin-converting enzyme in ACS cause, and the importance of glycemic excursions and very-long-chain saturated fatty acid metabolism.

Use of Antihypertensive Medications and Survival Rates for Breast, Colorectal, Lung, or Stomach Cancer.

Using time-dependent Cox regression models, we examined associations of common antihypertensive medications with overall cancer survival (OS) and disease-specific survival (DSS), with comprehensive adjustment for potential confounding factors. Participants were from the Shanghai Women's Health Study (1996-2000) and Shanghai Men's Health Study (2002-2006) in Shanghai, China. Included were 2,891 incident breast, colorectal, lung, and stomach cancer cases. Medication use was extracted from electronic medical records. With a median 3.4-year follow-up after diagnosis (interquartile range, 1.0-6.3), we found better outcomes among users of angiotensin II receptor blockers with colorectal cancer (OS: adjusted hazard ratio (HR) = 0.62, 95% confidence interval (CI): 0.44, 0.86; DSS: adjusted HR = 0.61, 95% CI: 0.43, 0.87) and stomach cancer (OS: adjusted HR = 0.62, 95% CI: 0.41, 0.94; DSS: adjusted HR = 0.63, 95% CI: 0.41, 0.98) and among users of ß-adrenergic receptor blockers with colorectal cancer (OS: adjusted HR = 0.50, 95% CI: 0.35, 0.72; DSS: adjusted HR = 0.50, 95% CI: 0.34, 0.73). Better survival was also found for calcium channel blockers (DSS: adjusted HR = 0.67, 95% CI: 0.47, 0.97) and diuretics (OS: adjusted HR = 0.66, 95% CI: 0.45, 0.96; DSS: adjusted HR = 0.57, 95% CI: 0.38, 0.85) with stomach cancer. Our findings suggest angiotensin II receptor blockers, ß-adrenergic receptor blockers, and calcium channel blockers might be associated with improved survival outcomes of gastrointestinal cancers.

Diabetes Medication Use in Association with Survival among Patients of Breast, Colorectal, Lung, or Gastric Cancer.

PURPOSE: Studies suggest that regular use of metformin may decrease cancer mortality. We investigated the association between diabetes medication use and cancer survival. MATERIALS AND METHODS: The current study includes 633 breast, 890 colorectal, 824 lung, and 543 gastric cancer cases identified from participants of two population-based cohort studies in Shanghai. Information on diabetes medication use was obtained by linking to electronic medical records. The associations between diabetes medication use (metformin, sulfonylureas, and insulin) and overall and cancer-specific survival were evaluated using time-dependent Cox proportional hazards models. RESULTS: After adjustment for clinical characteristics and treatment factors, use of metformin was associated with better overall survival among colorectal cancer patients (hazards ratio [HR], 0.55; 95% confidence interval [CI], 0.34 to 0.88) and for all four types of cancer combined (HR, 0.75; 95% CI, 0.57 to 0.98). Ever use of insulin was associated with worse survival for all cancer types combined (HR, 1.89; 95% CI, 1.57 to 2.29) and for the four cancer types individually. Similar associations were seen for diabetic patients. Sulfonylureas use was associated with worse overall survival for breast or gastric cancer (HR, 2.87; 95% CI, 1.22 to 6.80 and HR, 2.05; 95% CI, 1.09 to 3.84, respectively) among diabetic patients. Similar association patterns were observed between diabetes medication use and cancer-specific survival. CONCLUSION: Metformin was associated with improved survival among colorectal cancer cases, while insulin use was associated with worse survival among patients of four major cancers. Further investigation on the topic is needed given the potential translational impact of these findings.

A machine learning-based framework to identify type 2 diabetes through electronic health records.

OBJECTIVE: To discover diverse genotype-phenotype associations affiliated with Type 2 Diabetes Mellitus (T2DM) via genome-wide association study (GWAS) and phenome-wide association study (PheWAS), more cases (T2DM subjects) and controls (subjects without T2DM) are required to be identified (e.g., via Electronic Health Records (EHR)). However, existing expert based identification algorithms often suffer in a low recall rate and could miss a large number of valuable samples under conservative filtering standards. The goal of this work is to develop a semi-automated framework based on machine learning as a pilot study to liberalize filtering criteria to improve recall rate with a keeping of low false positive rate. MATERIALS AND METHODS: We propose a data informed framework for identifying subjects with and without T2DM from EHR via feature engineering and machine learning. We evaluate and contrast the identification performance of widely-used machine learning models within our framework, including k-Nearest-Neighbors, Naïve Bayes, Decision Tree, Random Forest, Support Vector Machine and Logistic Regression. Our framework was conducted on 300 patient samples (161 cases, 60 controls and 79 unconfirmed subjects), randomly selected from 23,281 diabetes related cohort retrieved from a regional distributed EHR repository ranging from 2012 to 2014. RESULTS: We apply top-performing machine learning algorithms on the engineered features. We benchmark and contrast the accuracy, precision, AUC, sensitivity and specificity of classification models against the state-of-the-art expert algorithm for identification of T2DM subjects. Our results indicate that the framework achieved high identification performances (∼0.98 in average AUC), which are much higher than the state-of-the-art algorithm (0.71 in AUC). DISCUSSION: Expert algorithm-based identification of T2DM subjects from EHR is often hampered by the high missing rates due to their conservative selection criteria. Our framework leverages machine learning and feature engineering to loosen such selection criteria to achieve a high identification rate of cases and controls. CONCLUSIONS: Our proposed framework demonstrates a more accurate and efficient approach for identifying subjects with and without T2DM from EHR.