Abstract
Background
Population-based studies provide a great opportunity to assess the causes of apparent resistant hypertension in real-world setting. Anti-hypertensive drug (AHD) treatment information is typically collected through interviews supported by drug barcode scanning. However, the reliability of such information is to be proven. To this aim, we implemented an unsupervised cluster analysis in Cooperative Health Research in South Tyrol (CHRIS) study, a population-based study from an Alpine environment, comparing AHD treatment-based grouping against RAAS biomarker-based clustering, eventually testing cluster association with blood pressure (BP) levels.
Methods
Among all 2105 CHRIS participants with documented AHD treatment, we identified five main groups: untreated; ACEi, plain or in combination; ARB, plain or in combination; beta-blocker, plain or combined with diuretics, and other AHDs. We simultaneously quantified aldosterone and equilibrium AngI and AngII with RAAS Triple-A liquid chromatography high throughput approach combined with tandem mass spectrometry analysis. We ran K-means unsupervised cluster analysis and assessed agreement against known treatments, extending the analysis to include clinically-relevant information. The extent of additional information provided by the clusters on BP was quantified using linear models, accounting for relevant covariates and polygenic background.
Results
K-means-based silhouette analysis identified an optimal number of five clusters, with an ACEi-specific cluster (kappa=0.55; sensitivity=0.48; specificity=0.99) and ARBs preferentially split in two different clusters, one with kappa=0.18 (sensitivity=0.31, specificity=0.85) and one with kappa=0.27 (sensitivity=0.37, specificity=0.88). The unsupervised clusters were associated with systolic BP beyond age, sex, RAAS biomarkers, ACEi/ARB metabolites, clinical covariates and BP polygenic scores.
Conclusions
RAAS biomarkers-based unsupervised clustering appears to capture partially distinct profiles with respect to reported AHD treatment and BP levels.