Discovery Of Novel Subgroups Of Patients With Essential Hypertension

ObjectivesThe purpose of this study was to divide the essential hypertension cohort into new subgroups by K-means clustering analysis.To diagnose and classify patients with essential hypertension more accurately by comparing the clinical characteristics of diseases in different subgroups after the clustering analysis.MethodsA total of 892 patients with essential hypertension who were hospitalized for evaluation and treatment during January 2016 to December 2017 at the Hypertension Center of Fuwai Hospital were retrospectively included.Unsupervised cluster analysis was performed using the K-means algorithm on the above cohort.The clustering was based on 9 variables related to cardiovascular risk in patients with hypertension,including age at onset,body mass index,systolic blood pressure at first diagnosis,fasting blood glucose,total cholesterol,total triglyceride,high-density lipoprotein cholesterol,serum creatinine and homocysteine.ResultsA total of 892 patients with essential hypertension were divided into 3 subgroups named cluster 1,cluster 2,and cluster 3,among which 303 cases(33.97%)were cluster 1,241 cases(27.02%)were cluster 2,348 cases(39.01%)were cluster 3.ANOVA and rank sum test of multiple independent samples were performed on the variables between the subgroups.The variables among the three subgroups were significantly different(SBP P<0.05,the remaining variables P<0.001).The characteristics of each subgroup are as follows,cluster 1:the age of onset is the highest,the SBP at the first diagnosis is the highest,BMI is at intermediate level,FPG and TC are the highest,HDL-C is the highest,HCY and SCR are the lowest,it is defined as overweight and medium-risk essential hypertension;cluster 2:The age of onset is thelowest,BMI,TC,TG,FPG are the lowest,HDL-C is close to the highest value,it is defined as the low-risk essential hypertension of normal-sized youth;cluster 3:the age of onset is at intermediate level,BMI,SCR,TG and HCY are the highest,TC and FPG are close to the highest value,HDL-C is the lowest,it is defined as obese and high-risk essential hypertension.Conclusions We divided the essential hypertension population into subgroups with significantly different characteristics using K-means clustering analysis.The new subgroup classification may have a certain suggestive effect on the risk of cardiovascular adverse events in patients.The new subgroup classification also has positive significance of exploring individualized treatments and underlying pathogenesis of essential hypertension.