Arterial stiffness nomogram identification by cluster analysis: A new approach of vascular phenotype modeling

Abstract

Arterial stiffness, measured by arterial stiffness index (ASI), can be considered as a major denominator in cardiovascular diseases. Thus, it remains essential to highlight patient phenotyping profiles with high ASI values. A nomogram of arterial stiffness was evaluated by calculation of ASI nomogram. Theoretical ASI can be performed according to age, sex, mean blood pressure, and heart rate, allowing to form an individual ASI nomogram [(measured ASI – theoretical ASI)/theoretical ASI]. An ASI nomogram > 0 defined AS. This study investigates among UK Biobank participants without cardiovascular diseases, the hypothesis that K‐means cluster analysis can be used to identify homogeneous phenotyping subgroups of participants according to ASI levels and then, the phenotype differences observed between these clusters. ASI nomogram was applied on 132 851 participants. K‐means clustering was implemented with 10 clusters (optimal CCC value of 105.246). One cluster showed 100% rate of AS, corresponding to 25 393 participants (41.6% of the AS participants) with ASI nomogram = .26 (.22), ASI = 11.6 (2.3)m/s. A second cluster showed a 100% of non‐AS, corresponding to 27 844 participants (38.8% of the participants with no arterial stiffness) with ASI nomogram = ‐.22 (.13), ASI = 7.1 (1.44)m/s. Threshold values of independent factors for differencing these two clusters were total cholesterol > 5.409 mmol/L ( P < .001), triglycerides > 1.286 mmol/L ( P < .001), smoking pack years > 11.8 pack/years, CRP > .99 ( P < .001), daily alcohol consumption > 1.794 units/days and BMI > 26.641 kg/m ² ( P < .001). Cluster analysis allowed to highlight homogeneous participants profile with or without AS. Determine the markers differencing these clusters participates in the management of cardiovascular preventive strategies.