Assessing Cardiac Remodeling in Aortic Regurgitation Using Indexed Extracellular Volume

It is well established from pathological studies that pressure overload in aortic stenosis (AS) and volume overload in aortic regurgitation (AR) are associated with left-ventricular (LV) remodeling, characterized by increased interstitial fibrosis and myocyte hypertrophy (1). Pathological studies have demonstrated that AR is associated with an increase in left-ventricular mass (LVM), increased myofibril diameter (hypertrophy), increased percent interstitial fibrosis (% IF), and increase in fibrous content (FC = % IF • LVM), with a normal volume fraction of myofibrils, compared with heart transplantation donor hearts (1). These microstructural changes may precede the development of symptoms and the deterioration of LV systolic function. Cardiovascular magnetic resonance (CMR) is the ideal tool to study ventricular remodeling in valvular heart disease (VHD), as it enables the assessment of reactive fibrosis using T1 mapping and extracellular volume (ECV) and replacement fibrosis using late gadolinium enhancement (LGE). The presence of myocardial fibrosis in AS has been associated with less functional recovery and significantly worse outcomes following AVR (2), raising the question of whether assessment of myocardial fibrosis by CMR could provide additional risk stratification in patients with VHD.

There is growing interest in using CMR to provide novel insights into myocardial hypertrophy and fibrosis in VHD. Native and postcontrast CMR T1 mapping can be used to determine the ECV, which has been shown to be a surrogate marker for processes such as diffuse myocardial fibrosis (3). ECV is the volume fraction of extracellular space on a voxel-by-voxel basis and, as such, is inherently normalized. To quantify the total fibrosis burden of the heart, indexed total extracellular volume (iECV), and indexed total cellular volume (iCV) are defined by the following formulas: iECV = ECV • LVMi/1.05, and iCV = (1–ECV) • LVMi/1.05 when 1.05 g/mL is the specific gravity of myocardium and LVMi is the LVM indexed to the body surface area (4).

In a previous CMR study of the progression of disease in AS that included 61 patients, both LVMi and iECV increased over time, but ECV did not increase, suggesting a balanced increase in both the cellular and extracellular compartments (5). Of note, in a large multicenter prognostic study of 440 patients with AS followed for a median of 3.8 years, ECV was an independent predictor of all-cause mortality in multivariate analyses that included LVMi, whereas iECV was not associated with all-cause mortality or cardiac mortality in univariate analyses (6). Two studies looking at the change in LV remodeling in AS following AVR both demonstrated a significant reduction in LVM, iECV, and iCV, with an apparent paradoxical increase in ECV following AVR (4,5). This increase in ECV reflected a more rapid regression of cellular hypertrophy compared with the regression of fibrosis following AVR, resulting in a relative increase in the extracellular space on a voxel-wise basis (4). This finding is consistent with previous pathological studies that demonstrate an increase in interstitial fibrosis at intermediate follow-up of patients following AVR for AS (1).

ECV has also been studied as a marker of diffuse fibrosis in mitral regurgitation (MR). An ECV >30% has been associated with adverse remodeling in asymptomatic patients with chronic moderate or severe MR without a class I indication for surgery (7). Also, increased ECV has been shown to portend an adverse outcome, independent of the severity of MR, among those patients with a regurgitant fraction (RF) >40% (8).

To date there had not been a large CMR study looking at myocardial fibrosis and LV remodeling in AR. The study by Senapati et al in this issue of iJACC (9) addresses this shortcoming by using CMR metrics of ECV, iECV, iCV, and LVMi to study 177 patients with isolated chronic AR followed for a median of 2.5 years. The authors demonstrate that iECV and iCV significantly increased as a function of AR severity, whereas ECV and LGE did not. This suggests a balanced increase in the size of the cellular and extracellular compartments as LV remodeling advances. Using survival analysis, the authors demonstrated that iECV was independently associated with the composite endpoint, with the worst prognosis in the group of patients with a RF >30% and an iECV >24 ml/m². By comparison, ECV% was not associated with the composite endpoint.

We want to commend the authors for their work on publishing the largest study to date investigating tissue remodeling in patients with isolated AR using CMR. One challenge is understanding the relationships among %ECV, iECV, iCV, and LVMi and how these parameters relate to the underlying pathophysiology of LV remodeling in chronic AR. Consider the following scenarios, each of which could result in an increase in iECV but with differing interpretation on a regional basis. 1) ECV increases without an increase in LVMi; in this situation, there would be clear expansion of the extracellular space relative to the intracellular space on a voxel-wise basis. 2) LVMi increases without an increase in ECV; in this situation, there is a change in total fibrotic volume and total cellular volume without a change in relative extracellular expansion on a voxel-wise basis. 3) ECV and LVMi both increase; in this scenario, there is both an absolute and relative increase in the extracellular space and an increase in the total cellular volume. How iECV will increase depends on this relationship between total and relative interstitial expansion.

As iECV is the product of LVMi and ECV, it describes the statistical interaction between LVMI and ECV. From a modeling perspective, the added use of this term can be assessed by considering the 2 nested models: Model 1: ECV + LVMi; Model 2 (interaction model): ECV + LVMi + LVMi • ECV. This analysis would enable one to determine whether the relationship deviates from the expected multiplicative relationship in the Cox model (10). The authors provide a hint at this question in a supplemental analysis by considering nested models that included LVMi or LVMi and iECV. They demonstrate an improvement in the model fit in terms of the log-likelihood chi-square statistic (P = 0.04) of using iECV over LVMi; however, the 2 models produced identical C-statistics, demonstrating similar model performance. Why ECV was not predictive of outcomes in AR–as has been shown for AS–remains to be explained further.

With the increased use of CMR in VHD, and potential implications for treatment, it is important to consider the relationship among iECV, iCV, ECV, and LVMi. Whether or not iECV and iCV have additional information to the primary parameters of ECV and LVMi, and whether or not they provide a useful single summary parameter, requires further consideration and further studies. Furthermore, it will be interesting to see how these parameters change following AVR. The current paper provides important data for furthering our understanding of the role of diffuse fibrosis in AR.