Non-Paradoxical Paradoxes of Heart Failure

The word “paradox” appears over and over again in the AHA scientific statement on comorbidities in heart failure (HF)1, especially in the sections on hypertension and obesity. Indeed, hypertension is a known risk factor which multiplies the chances of developing HF2, and yet once HF develops, higher blood pressure (BP) suddenly marks better prognosis. Paradoxical? Not quite.

Arterial BP is a product of cardiac output and peripheral vascular resistance. Cardiac output remains stable throughout most pre-clinical and symptomatic stages of HF, and BP is mainly determined by peripheral vascular resistance. Hypertension creates increased left ventricular afterload. When this condition persists for years, it may eventually result in HF. To the contrary, in advanced HF, cardiac output decreases. Now systolic BP is not only the afterload, but the measure and reflection of severity of left ventricular dysfunction- a weak ventricle cannot generate high pressure. This is the reason that lower blood pressure is associated with poor prognosis.

Not surprisingly, the relationship between systolic BP and prognosis is different with different degrees of left ventricular dysfunction. While systolic function is preserved, the relationship between systolic BP and mortality is U-shaped, with worse outcomes at both high and low end of BP spectrum. However, when left ventricular ejection fraction decreases below 30%, the association becomes linear, with lower BP linked to higher mortality3.

Another “paradoxical” relationship is between HF and obesity. Superficially, the situation is very similar: in the general population, obesity facilitates HF. And yet, once HF develops, obesity suddenly becomes protective and improves survival. Specifically, obesity increases the risk of de novo HF, with each one-unit increment of body mass index (BMI) increasing this risk by approximately 5% for men and 7% for women4. At the same time, in individuals who already have HF, a five-unit increment of BMI reduces 30-day mortality by 11% and 1-year mortality by 9%5. However, the similarity with hypertension is deceiving. There are no obvious hemodynamic factors that would so dramatically change the role of excessive weight at any stages of HF.

Moreover, weight reduction in HF improves outcomes. One study reported that bariatric surgery resulted in weight reduction and significant increase in left ventricular ejection fraction (+5.1% ±8.3%, P=0.0005). Other authors discovered that, in patients with profound left ventricular dysfunction (left ventricular ejection fraction <25%), bariatric surgery yielded a median BMI reduction of 12.9 kg/m2 (p = 0.017) and postoperative left ventricular ejection fraction improvement, in some cases sufficient to obviate heart transplantation6. And, perhaps most convincingly, another study found that the rate of emergency room visits or hospitalizations for HF dropped from 15.3% in the year preceding bariatric surgery to 9.9% a year after surgery7.

So, obesity is detrimental for new onset HF, and treatment of obesity is beneficial in existing HF, but there is favorable effect of obesity on mortality? This is really counterintuitive and deserves more detailed explanation.

Recently, we collected data on baseline patient characteristics from multiple studies on the obesity paradox and noted a striking pattern across multiple studies, strongly favoring obese patients with HF8. Obese patients were significantly younger than those with normal weight in 17 out of the 19 examined studies, and in the remaining two studies the difference was not significant. There was not a single study where the obese were older. Sometimes, the age difference equaled almost a full decade9. Besides, left ventricular ejection fraction was higher among obese patients. In the same study, mean left ventricular ejection fraction was 42% in obese patients, and 36% in their normal weight counterparts9, and this discrepancy was also consistent across multiple studies.

Systolic blood pressure, which, as we discussed, is a good prognostic sign in HF, was higher in obese individuals in every single study where the difference was significant. Renal function was better in obese patients, with a single exception. A smaller proportion of obese patients had advanced functional class (New York Heart Association III and IV)8.

Moreover, obese patients with HF tended to have less prevalent and less severe valvular regurgitation, lower burden of chronic obstructive pulmonary disease, fewer bundle branch blocks and arrhythmias, better nutritional status and more favorable values of other indicators of wellness including handgrip strength, lymphocyte count, hemoglobin, serum albumin, pre-albumin, and cholesterol.

Being younger and having better nutritional and functional status, cardiac and renal function, and higher blood pressure, obese patients with HF have better survival, and it is very intuitive and non-paradoxical. However, in each study, appropriate statistical methods were applied to adjust for the baseline differences. We recently hypothesized that synergistic effect of all favorable characteristics of the obese with HF can have greater effect on prognosis than each of the factors accounted for individually. In a statistical model, we demonstrated that this unaccounted synergism may result in misleading conclusion that obesity is beneficial when, in fact, it is neutral or detrimental10. Some studies, where the patients with different BMIs were matched into triples on NT-proBNP, age, sex, and New York Heart Association functional class, failed to confirm the obesity paradox11, and sometimes just adding the NT-proBNP to the equation eliminated the significant association between BMI and mortality, even though accounting for other clinical variables had failed to remove the obesity paradox12.

Another potential flaw of the studies demonstrating obesity paradox in HF is that each patient’s weight/BMI is measured only once. If it is normal or overweight, it may represent either stable condition or the result of weight loss from cardiac cachexia. In this case, normal weight actually reflects the loss from previous overweight or obese range. Because this represents an unfavorable change over time, HF patients with normal BMI at a snapshot in time may actually be quite ill; the normal BMI may not stem from a healthy lifestyle but may reflect movement along a downward trajectory in BMI. Stokes and colleagues13 explored this phenomenon--reverse causality--in detail. They compared obese HF patients not to all those with normal weight but only to those who had maintained normal weight over time. With this limitation, the obesity paradox disappeared in their analysis.

Moreover, Zamora and colleagues14 recently demonstrated that mortality was significantly higher in HF patients with unintentional weight loss versus those without such weight loss (27.6% versus 15.3%, P<0.001). Obesity paradox may, therefore, represent a statistical artifact15.

Another interesting interplay occurs between HF and diabetes. This AHA scientific statement1 reviews the role of diabetes as a risk factor for HF and the condition complicating already existing HF16. At the same time, authors recognize that prevalence of diabetes in acute decompensated HF may be greater than in stable HF17. This is a very interesting observation because it connects with the hypothesis that HF may also be a precipitating factor for diabetes. In a large nationwide Danish cohort of patients discharged after myocardial infarction and not having a history of either diabetes mellitus or HF, about 20% developed HF, and 5% of them developed diabetes during the follow-up. Increasing HF severity, measured by loop diuretic requirements, was associated with increased risk of diabetes18. In another cohort of about 60,000 non-diabetic adults, HF was independently associated with an increase in incidence of diabetes by 48%19.

Analyzing the dataset from the Cardiovascular Health Study, we found that, among subjects with normal fasting glucose at baseline, HF significantly increased the odds of developing impaired fasting glucose after 3 or 4 years [odds ratio (OR) 2.18, 95% confidence interval (CI) 1.03-4.61, p = 0.043] or overt DM (OR 4.78, 95% CI 1.84-12.4, p < 0.001)20.

Moreover, recent data from the patient population with left ventricular assist devices indicate that such radical therapy for HF can lead to dramatic changes in insulin requirements and quality of blood glucose control21-23. Potential mechanisms include improved perfusion of pancreas leading to better beta cell function, reduced inflammation and thus lowered insulin resistance, and improved tissue perfusion leading to improved glucose utilization.

Taken together, these data are encouraging. They indicate that better understanding of the interactions between HF and DM can improve management of diabetes in HF patients, with the potential for favorable impact on morbidity and mortality in patients with combination of both conditions.

In summary, paradoxes in HF need to be explored and explained--and used to the benefit of the patients.