Revisiting the Obesity Paradox in Heart Failure: New Insights?

This editorial refers to ‘Body mass indices and outcome in patients with chronic heart failure‘ by James Futter et al. and ‘Absence of obesity paradox in patients with chronic heart failure and diabetes mellitus: a propensity-matched study‘ by Chris Adamopoulos et al. published in this issue on pages 200–bold and 207–213. In recent years, the public has become very concerned about weight loss, and programmed diets of various types have become blockbuster sellers for many companies. Indeed, in industrialized countries, obesity has reached epidemic proportions. The World Health Organization (WHO) estimates that more than 1 billion adults worldwide are overweight, and of these 300 million are actually clinically obese.1 In view of this alarming development, the aim to be slim and to lose weight has been thoroughly implemented in the public consciousness as a surrogate for physical and mental wellbeing. From the perspective of the clinician, however, if may be advisable to establish a more differentiated view on the association between body composition and morbidity and mortality. Indeed, the principles that have been proven for healthy populations may not be applicable in the setting of chronically diseased patients. Convincing evidence has accumulated that the WHO criteria of normal body mass index (i.e. body mass index 18.5–25.0 kg/m2) may not translate directly to patients with chronic diseases such as chronic heart failure (HF) with regard to optimal survival. Accordingly, the last few years have witnessed research into an entirely contrasting field: patients with chronic illnesses deemed to be incurable and even deadly seem to benefit from mild obesity. Consistent data have been published for chronic HF,2–4 chronic kidney failure,5 chronic obstructive pulmonary disease,6 and cancer.7 All of the studies show that some degree of obesity is associated with improved survival, which has led to the use of the term ‘obesity paradox' to describe this phenomenon. For centuries, the development of cachexia, i.e. involuntary weight loss in the face of a present illness, has been viewed as a sign of poor prognosis,8 and in chronic HF the last few years have seen growing research interest into this area.9,10 Insight into the beneficial role of higher body weight, on the other hand, has only recently begun to sprout. Many have argued that the available data are only retrospective in nature, and that database analyses cannot replace prospective studies. Indeed, there is a paucity of prospective data on prevalence and incidence of both cachexia and obesity in patients with chronic HF, and only relatively small studies are currently available. This issue of the European Journal of Heart Failure contains two articles that provide further insight into the obesity paradox in patients with chronic HF. However, both studies cannot circumvent the fact that a scarcity of prospective clinical studies exists in the field, as both are retrospective in nature. Futter et al.11 studied 2271 patients with chronic HF due to left ventricular systolic dysfunction attending an outpatient clinic. The authors assessed several anthropometric measures such as body mass index, ponderal index (weight divided by height3), body surface area, and, in a subset of patients, the so-called Charles index that also includes waist circumference [defined as weight divided by (waist2 times height)]. Similar to earlier studies, the authors found that measures of body mass were strong predictors of survival. In contrast, waist circumference and the Charles index failed to predict survival among the patients in this study. Further, it is interesting to note that by using bioimpedance data the authors also showed that both fat mass and fat-free mass were predictors of outcome in single-predictor analysis. While fat-free mass may include muscle tissue with obvious functional involvement for physical activity, the role of fat mass for outcome prediction is far more difficult to explain. It has been proposed that larger fat stores could reflect a greater metabolic reserve in a condition of overall increased catabolic turnover. However, after multivariable adjustment, this effect was lost for both compartments, and we are thus once again left in the dark with regard to which body compartment should be supported in our therapeutic endeavours. In addition, the authors left important predictors of survival out of their final multivariable models, specifically left ventricular ejection fraction and N-terminal pro-B-type natriuretic peptide (NT-proBNP). This was because the datasets were deemed to be too incomplete. It is unfortunate, because NT-proBNP was the best predictor of survival in the univariable models as highlighted by the respective χ2 values, and many authors argue that prognostic assessments in HF nowadays should include a natriuretic peptide.12,13 Using a propensity-matched approach with data from the Digitalis Investigation Group (DIG) trial, Adamopoulos et al.14 show that the obesity paradox is only present in non-diabetic patients with HF, but not in patients with manifest diabetes mellitus. Overall, 29% of the 7788 participants in this analysis were diabetic, which is consistent with several other databases from large-scale HF trials. Indeed, the prevalence of diabetes is usually between 20 and 30% in such cohorts15–17 although the numbers may be significantly higher in patients with acute HF.18 In the study by Adamopoulos et al., the presence of obesity in non-diabetics was associated with a 23% reduction in the risk of all-cause mortality (matched hazard ratio 0.77, 95% confidence interval 0.61–0.97, P = 0.025). The authors speculate that the presence of diabetes mellitus is a much stronger predictor of poor outcome than obesity itself, which may offset a protective effect of the latter. It is noteworthy, however, that in their analysis of diabetic patients, overweight was not observed to carry an increased risk of mortality. This is still in contrast to common thinking, as being overweight is usually considered to be a major complication of diabetes in terms of metabolic control, complications, and outcome. Overweight seems not to be a problem in these patients, which may seem to be a paradoxical finding of its own. When analysing these data, it has to be kept in mind that the DIG trial was performed in an era before beta-blockers and aldosterone antagonists became standard treatment of chronic HF. The use of such drugs may have shifted the outcomes of the present study, as is known, for example from analyses of anaemia in HF. In the second Evaluation of Losartan in the Elderly (ELITE II) trial19 and in the Prospective Randomized Amlodipine Survival Evaluation (PRAISE) trial,20 a U-shaped relationship was reported for mortality vs. haemoglobin levels. Indeed, patients in the low and in the high ranges of haemoglobin had increased mortality rates. However, this U-shaped relationship disappeared in patients who were treated with a beta-blocker, and it is possible that such an effect exists in diabetic patients as well. The propensity-score matching used by Adamopoulos et al. compared obese and non-obese patients. Since only grouping into these two categories was applied, this approach did not and cannot account for the commonly observed U-shaped relationship between body composition and mortality. The European Commission recognizes the problems associated with retrospective analyses and the paucity of prospective data for important co-morbidities such as diabetes mellitus, obesity, or cachexia among patients with chronic HF. In October 2009, the Studies Investigating Co-morbidities Aggravating Heart Failure (SICA-HF) was initiated, and patient recruitment has commenced. SICA-HF is a prospective, multicentre, multinational, longitudinal, pathophysiological evaluation study, which is being conducted in 11 centres across six countries.21 One of the main aims of SICA-HF is to characterize the prevalence, incidence, persistence, and phenotype of obesity, cachexia, and type 2 diabetes in patients with chronic HF. The analysis of body composition and its changes over time as well as patterns of exercise capacity at baseline and their changes during up to 4 years of follow-up are some important considerations in this project that is funded under the 7th Framework Programme of the European Commission. In total, SICA-HF aims to recruit more than 1600 patients with chronic HF and two control cohorts including more than 150 healthy control subjects and more than 300 patients with type 2 diabetes mellitus without chronic HF. Until the publication of these results, we will need to rely on the published data from large-scale trials—and there are many that seem to point in the same direction. Concluding from these data, it seems that weight reduction is not good advice for the HF patient. Part of this work was funded by the European Commission under the Seventh Framework Programme 439 (FP7/2007–2013) under grant agreement number 241558 (SICA-HF).