Dietary Patterns and Precision Prevention of Heart Failure

Heart Failure (HF), with increasing prevalence over time, now affects more than 6.5 million adults in the US [1]. Although pharmacological options and devices have improved survival of HF patients with reduced ejection fraction (HFrEF), strategies to control HF with preserved ejection fraction (HFpEF) are less well established. Given the difficulty in treating HF, severe debility experienced by HF patients and high cost of HF care [2], prevention of HF is a preferable strategy. Although adopting a healthy diet is a key recommendation in the primary prevention of HF [1], direct evidence from epidemiologic studies that can underpin this recommendation is still limited.

In this issue of JACC, Lara et al present a prospective study that examined major dietary patterns in association with the incidence of HF and its two subtypes, HFpEF and HFrEF, in the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study. The authors derived five major dietary patterns in 16,068 black and white adults aged 45 years or older who were free from coronary heart disease (CHD) at baseline. They reported that a higher adherence to a “plant-based” dietary pattern, featured by high intakes of fruit, vegetables, fish, poultry, and low-fat dairy and low consumption of soda, was associated with a lower risk of HF; the association was independent of established risk factors of CVD and barely changed after further adjustment for potential mediating factors, including measures of body adipose tissue, diabetes, hypertension, atrial fibrillation, and measures of kidney function. In contrast, participants who consumed a dietary pattern high in red and processed meat, sugar-sweetened beverages, high-fat dairy and refined grain, named as “Southern” dietary pattern, showed an increased risk of HF. The remaining three dietary patterns were not associated with the risk of HF. The five dietary patterns were empirically derived through factor analysis that aggregates specific food items on the basis of the degree to which the items are correlated with one another [3]. An attractive feature of the empirically derived dietary patterns is that the derived patterns can be readily translated into eating behaviors in the dietary recommendations because this dietary patterning approach identifies the ways that foods are consumed in combination in actual diets. The use of dietary patterns in epidemiologic studies and randomized trials circumvents the challenges of isolating the effects of specific dietary factors from other highly correlated dietary factors of a diet and complement studies of specific nutrients and foods. We usually have greater confidence when interpreting the associations with dietary patterns as causal than we have for the associations with specific nutrients or foods [4]. In addition, Lara et al’s findings are particularly useful for making recommendations to a general population due to a baseline CHD-free study population and inclusion of black participants with greater susceptibility to HF [5]. Thus, this study possess great potential of informing the population-level strategies for the prevention of HF.

Another highlight of Lara et al’s study is their investigation on the potentially different associations of diets with HFpEF as compared with HFrEF. The authors found that a higher adherence to the Southern pattern was significantly associated with an increased risk of HFrEF, but was not associated with the risk of HFpEF. For the plant-based pattern, an inverse, albeit nonsignificant, association was observed for HFpEF but not for HFrEF. These findings appear to be the first detailed examination of overall diet in the associations with HF subtypes in population-based studies. The differential associations of dietary patterns with the two HF subtypes were mechanistically plausible. In most cases, HFpEF has systemic origins, in which cardiac injury is often a secondary phenomenon [6]. A plant-based dietary pattern with low energy-density and high fiber and antioxidant content [7] may exert its beneficial effects for preventing HFpEF through modifying systematic risk factors, e.g., vascular dysfunction, pulmonary and renal dysfunction, skeletal muscle dysfunction, low grade inflammation, blood pressure and glycemic control. In contrast, CHD and other myopathies account to a great extent for the etiology of HFrEF in which cardiac injury is often primary. Individuals who adopted the Southern dietary pattern had high red meat intake. A recent animal study suggested that trimethylene N-oxide (TMAO), a metabolite produced from metabolism of red meat, may contribute directly to the cardiac dysfunction [8]. Echoing Lara et al’s findings, a preliminary study in the Women’s Health Initiative recently reported a more pronounced positive association of sodium intake with HFpEF than that with HFrEF [9]. We should welcome the nutritional epidemiologic studies targeting on subtypes of HF given the large phenotypic and pathophysiological heterogeneity of the disease. These findings, if confirmed in future studies, will not only contribute to in-depth biological understanding and phenotypic refinement of HF [10], but also inform dietary preventive approaches customized for specific HF phenotypes [11], which perfectly fits into key missions of precision medicine, i.e., understanding large variability between individuals in both the development and the clinical manifestations of the specific disease, as well as variability in individual’s response to dietary, lifestyle and pharmacological interventions [12]. However, we should interpret these findings cautiously due to the limited number of cases in each subtype and consider the findings as only hypothesis-generating. Confirmation should be sought in future sufficiently powered studies.

Notable limitations of this study include one single measurement of diet and covariables. The participants may have changed their diet over the follow-up. Repeated collections of dietary information will help dampen errors embedded in measurements of diet. However, had the diet been measured better, it is likely the observed association between dietary patterns and HF risk would become stronger. The authors mentioned residual confounding as a limitation, which is always a consideration in observational studies. However, as I discussed above, residual confounding is less of a concern for studies on dietary patterns than those on specific foods and nutrients. Relatively short follow-up is another concern. Future investigations with extended follow-ups from this cohort or other prospective cohort studies are imperatively needed, which will provide greater flexibility to address the issue of reverse causation and examine the latency periods between dietary exposures and the onset of HF.

In summary, Lara et al’s study represents an important step forward in establishing a robust evidence base for the dietary prevention of HF. In addition, the authors’ finding on the potentially differential associations of dietary patterns with HFpEF as compared with HFrEF provides an early glimpse of precision dietary prevention of HF, i.e., recommending dietary preventive approaches customized based on different pathophysiological basis of HF subtypes. However, to achieve the precision prevention of HF, much research is needed to accrue robust and reproducible evidence in large epidemiologic studies. In addition, although the traditional classification into HFpEF and HFrEF has proven useful in differentiating distinct pathophysiological mechanisms with clear therapeutic implications, a large proportion of variability in clinical manifestations of HF, especially HFpEF, remains unexplained. Future epidemiologic studies based on a more advanced phenotype classification of the disease would set the stage for the precision prevention of HF [10, 13].