Left ventricular remodelling index: Reference ranges and prognostic associations in 44,000 UK Biobank participants
R Roy, D G Condurache, J Salatzki, L Szabo, T T Le, C W L Chin, S E Petersen, Z Raisi-EstabraghAbstract
Background
Cardiac magnetic resonance (CMR) measures of left ventricular hypertrophy (LVH) predict cardiovascular disease, but the optimal metric to capture clinically relevant LV remodelling remains uncertain. The Remodelling Index (RI), defined as cube-root LV end-diastolic volume divided by maximal wall thickness, is a mechanistically informed marker of LV geometry, yet reference ranges and comparative prognostic data are limited.
Purpose
(1) Characterise healthy demographic variation in RI and derive age-, sex- and ethnicity-specific reference ranges, (2) evaluate RI associations with cardiometabolic risk factors and incident outcomes in a large population cohort, including comparison with established LVH metrics.
Methods
CMR studies were performed/analysed using standardized protocols. In the Healthy Hearts Consortium (HHC; international databank of verified healthy adults), we modelled RI variation by age, sex and ethnicity, and generated reference ranges (2.5th-97.5th centiles using GAMLSS methodology). In UK Biobank, we assessed associations between RI and prevalent cardiometabolic disease (logistic regression), and incident heart failure (HF), myocardial infarction (MI), stroke, and cardiovascular death (Cox regression), adjusting for demographic/clinical factors and established CMR markers (including LV ejection fraction, LV mass and global longitudinal strain).
Results
In HHC (n=8,613; 52% female; median age 61y, 81% White), RI varied significantly with age, sex and ethnicity (Fig 1A), supporting tailored reference ranges (Fig 1B). In UK Biobank (n=44,075, 52% female, median age 65y, median follow-up 4.4y), higher RI was associated with lower odds of prevalent hypertension (OR 0.68 [0.66, 0.70]) and diabetes (OR 0.76 [0.71, 0.81]), but higher odds of prevalent HF (OR 1.48 [1.22,1.80]) and MI (OR 1.39 [1.25, 1.53]) (Fig 2A). For incident outcomes, higher RI was associated with lower stroke risk (HR 0.81 [0.69, 0.95]), but higher HF risk (HR 1.28 [1.11, 1.49]) (Fig 2B). Associations differed by hypertension status: amongst non-hypertensives, higher RI was linked to lower stroke and MI risk, whereas amongst hypertensives, higher RI was associated with higher HF risk (Fig 2C-F). Compared with other LVH pattern metrics, RI demonstrated comparable/greater association than LV mass-to-volume ratio for stroke/MI, which was particularly evident in non-hypertensives. LVM generally showed the strongest associations amongst measures of LVH degree.
Conclusions
In this multi-cohort CMR analysis, the RI emerged as a robust marker of LV remodelling phenotype with significant demographic variation. We provide novel population-specific reference ranges that support its clinical use. The RI captures a spectrum of LV geometric remodelling and offers information complementary to conventional LVH metrics, with potential to improve risk stratification for cardiovascular outcomes, particularly amongst individuals without hypertension.Figure 1Figure 2