Ventricular remodelling: an equal-opportunity prognosticator.

Abstract

This article refers to ‘Prognostic impacts of dynamic cardiac structural changes in heart failure patients with preserved left ventricular ejection fraction’ by S. Yamanaka et al., published in this issue on pages 2258–2268. Regardless of cause, cardiomyopathies are generally characterized by progressive remodelling of the left ventricle. This process involves myocyte hypertrophy and interstitial fibrosis and, in restrictive myopathies, infiltration of abnormal tissue. The resulting size and shape of the ventricular cavity and walls depend on both the nature of the underlying disease and ambient factors such as baseline heart size and gender.1 Ventricular remodelling has been well described in animal models and patients with ischaemic heart disease2-5 and dilated cardiomyopathy (with a variety of underlying causal factors), characterized by progressive increases in cavity volume, in both diastole and systole, transition to a more globular shape, and reduction in left ventricular ejection fraction (LVEF). The latter may be impacted by reduced contractility and/or abnormal left ventricular (LV) load. However, to a great extent, chronic changes in LVEF are directly derived from the structural dilatation process itself because LV dilatation without major change in heart rate or stroke volume results in an obligate decrease in LVEF.6 Recent studies were mostly performed in patients with reduced LVEF. However, early investigations of LV remodelling did not separate populations according to LVEF, but, rather, examined predictive models across the spectrum of LVEF measurement,2, 3 documenting that adverse changes in both LVEF and LV volumes are strong predictors of subsequent adverse clinical outcomes.2, 3, 7 In this issue of the Journal, Yamanaka et al.8 extend these observations to patients with clinical heart failure (HF) and an LVEF ≥50% [HF with preserved ejection fraction (HFpEF)], showing that LV structure changes variably over time, and that structural abnormalities, both at baseline and at 1 year, are predictors of the primary endpoint of cardiovascular (CV) death or HF hospitalization. They found that, compared with patients without LV hypertrophy (LVH) or LV enlargement (LVE) at baseline, subsequent adverse outcomes were more frequent in patients with LVH without LVE at baseline and were even more frequent in patients with LVE. Using a landmark analysis, the authors show that the worst subsequent outcomes were in patients with LVE at 1 year, regardless of its presence or absence at baseline.8 These important findings, in some ways, extend what has been known in patients with HF with reduced ejection fraction (HFrEF) and suggest a continuity of the prognostic value of LV volumes in patients across a broad range of LVEF measurement (Figure 1). The term ‘HFpEF’ comprises a heterogeneous population with multiple potential diagnoses (or phenotypes).9 For the most part, the commonality is that the myocardium may be characterized by hypertrophy, fibrosis and/or infiltration, but the ventricle has not remodelled with dilatation. Generally, barring LV volume overload (in which actual LV stroke volume exceeds forward stroke volume), a normal LVEF implies that whatever remodelling process is present, it is associated with mild or absent LV dilatation, with myocyte hypertrophy primarily translating into increased wall thickness, driving increased LV mass. Although a variety of underlying diseases may lead to HFpEF, the most prevalent is metabolic CV disease, risk factors for which include diabetes, hypertension, obesity and advanced age, and in which endothelial dysfunction, vascular inflammation, nitric oxide deficiency, myocyte hypertrophy and interstitial fibrosis play pathophysiologic roles.10 Perhaps the next most common diagnosis is cardiac amyloidosis, particularly infiltration with transthyretin amyloid, which is far more commonly recognized now that the disease may be diagnosed non-invasively and that therapy is available.11 Like other registries and clinical trial populations, the population studied in this report8 is likely to include such patients, without whom the findings might have been even stronger. However, the population characteristics at baseline suggest that the preponderance of patients had metabolic CV disease. The pathologically hypertrophied cardiomyocyte handles calcium abnormally and is functionally abnormal during both systole and diastole, and both systolic and diastolic dysfunction are detected in patients with HFpEF.12 In the intact heart, increased wall thickness confers diastolic dysfunction through both slowed myocyte relaxation and reduced LV compliance. It is therefore no surprise that LVH, as defined by these authors,8 even without LVE carries a worse prognosis, as it does in the general population.13 In addition to examining the predictive value of LV mass and volume as categorical variables (LVH +/−, LVE +/−), the authors8 also examined them as continuous variables. In the former analysis, although LVEF progressively decreased as the status of patients grouped by LVH/LVE went from −/− to +/− to +/+, the relationships between structural category and outcome were unchanged within a multivariate analysis adjusting for clinical variables including LVEF (recognizing that LVEF was truncated below 50%). The authors8 incorporated LVEF as one of the continuous variables analysed, along with LV mass and LV diastolic volume, and found that baseline LVEF was not a significant prognosticator. However, decrease in LVEF at 1 year independently predicted adverse outcome, after adjustment for clinical variables. Reduced LVEF as LVE appears is a manifestation of the mathematical link between LVEF and LV volumes. LVEF may be thought of as relative LV stroke volume. If LV remodelling results in increases in LV end-diastolic and end-systolic volume, without change in absolute LV stroke volume, LVEF must decrease. However, the retention of the significance of LVE following adjustment for LVEF, and the greater predictive value of LV volume than LVEF, at least at baseline, is consistent with LV volumes being more precise markers of dilatational remodelling than LVEF, recognizing, nonetheless, that the HFpEF population encompasses a limited range of LVEF. Regardless, this analysis shows that in HFpEF, as in HFrEF, LV dilatation connotes a worse prognosis.8 It should be expected that abnormal remodelling correlates with the frequencies of hospitalization and death because it represents progression of the gross pathology associated with the majority of myocardial disorders. It seems likely that diseases such as ischaemic cardiomyopathy and metabolic CV disease are not confined to any ejection fraction boundary as they may be associated with varying degrees of LV remodelling. If we compare the present findings8 with those of a 1979 investigation2 in patients with coronary artery disease, regardless of baseline LVEF, we see the same pattern of worsening outcomes as baseline LV end-diastolic volume index (LVEDVI) increases. However, the slope of that increase in the prior study2 seems to have been far steeper (greater increase in event rates for a given increase in LVEDVI) than it appears to be today. Granted, the two populations differ, but there is substantial overlap in the ranges of LV volumes within them. The difference is probably linked to the marked improvements in medical therapy over the past 40 years. Clinical trial results in HFpEF have been disappointing. One reason is likely to refer to the heterogeneity in underlying diseases (including the prevalence of transthyretin amyloid). A second is that we are addressing metabolic CV disease too late. That is, many of the therapies investigated may have a greater likelihood of benefit before advanced stages of ventricular remodelling and clinical expression of HF are present. A third reason for slow therapeutic development may be the absence of an effective marker of likely clinical benefit during early stages of investigation. In patients with LVE and reduced LVEF, therapeutically induced improvement in LV volume and LVEF correlates with a therapeutic effect on mortality.7 Therefore, these markers represent excellent indicators, during Phase 2 investigations, of whether the intervention is likely to show outcome benefit in Phase 3. Based on the present findings,8 it will be important to explore such a relationship in patients with preserved LVEF to provide a rationale for proceeding into Phase 3. Since the early 1980s, we have learned much about the importance of ventricular remodelling in understanding both the pathophysiology of HF and the relationship between LV structure and clinical disease progression. Until recently, the majority of that work occurred in patients within a low range of LVEF. The findings in this issue8 extend our knowledge about remodelling, particularly LVE, and outcomes across the ejection fraction boundary into patients with preserved LVEF. An important next step will be to establish a degree of surrogacy for remodelling markers in this population, examining associations between therapeutically mediated structural change and therapeutic impact on clinical outcomes, as has been done in HFrEF.7 The sample size and duration of observation required to detect longitudinal structural change in a diverse HFpEF population (with a limited range of baseline LVEF and LV volume) may challenge the utility of current techniques in reliably establishing treatment effect within small Phase 2 populations. Yet the present report8 should encourage continued investigation. First, we should work to segregate study populations into more homogeneous diagnostic groups to enhance the efficacy signal and reduce measurement noise. Second, we should further perfect imaging techniques that could more effectively detect early, subtle therapeutically derived structural changes. In this way, knowledge gained from the work of Yamanaka et al.8 may yield more than a mere prognostic tool. It may lead to a means for the early identification of therapeutic benefit and for accelerating the development of successful therapies within this population. Conflict of interest: none declared.