Mechanistic Insights into Loop Diuretic Responsiveness in Heart Failure

In the setting of acute decompensated heart failure, decongestion has become a “holy grail,” because patients who are effectively decongested by treatment do better both from a neurohormonal standpoint and clinically. “Congestion” is variably defined, but a recent position statement from the European Society of Cardiology presents a schema that relies on both clinical and imaging criteria (1). The central importance of decongestion is indicated by its effects on treatment complications; although worsening kidney function (typically called worsening renal function in the cardiology literature and defined as a creatinine increase ≥0.3 mg/dl during hospitalization) is often considered a poor prognostic factor and used as a “harm marker” in large clinical trials, worsening kidney function associates with better, not worse, outcomes in patients who have been decongested successfully (2). This observation, which has been confirmed, suggests that diuretic treatment should be maintained in the setting of worsening kidney function as long as decongestion is taking place, at least for patients with heart failure with reduced ejection fraction. For nephrologists long accustomed to assuming that worsening kidney function is always a bad thing, these data have required cognitive adjustment.

Although some patients with heart failure present with congestion as a result of fluid redistribution, most have recent expansion of the extracellular fluid volume. It is for this reason that loop diuretics remain first lines of treatment for patients with acute decompensated heart failure, being used in approximately 90% of such patients (3). The congruence of the facts that decongestion is vital for therapeutic success and that diuretics are the main approach to achieve decongestion implies that it is important to understand mechanisms of diuretic responsiveness to design safe and effective approaches for symptomatic heart failure.

In the simplest construct, to be effective, a loop diuretic must (1) be ingested and absorbed or administered parenterally, (2) transit to the kidney via the circulation, (3) undergo secretion by the proximal tubule, (4) be delivered to the luminal surface of thick ascending limb cells, and (5) bind to and inhibit the Na-K-2Cl cotransporter. Additionally, natriuretic effects may be modified by processes upstream and downstream from the active site. Because many patients with acute decompensated heart failure are or become resistant to loop diuretics, knowing which of these steps plays the dominant role should inform effective treatment approaches.

The paper by Charokopos et al. (4) in this issue of CJASN, explores the possible roles of hypoalbuminemia and proteinuria in diuretic resistance in heart failure. To be delivered to their site of action, loop diuretics, which are “small molecules,” must be retained within the bloodstream; their tendency to bind to albumin is a key component of this. In an earlier study, the effect of diuretic binding on response to loop diuretics was examined in analbuminemic rats: rats with a spontaneous mutation that prevents albumin synthesis (5). The furosemide volume of distribution was 545 ml/kg in analbuminemic rats, whereas it was only 60 ml/kg in normal rats. This difference was associated with a marked rightward shift in the furosemide natriuretic dose-response relationship, suggesting that the increased volume of distribution caused the diuretic resistance. This observation was translated to a small (n=20) series of patients who were hypoalbuminemic; when patients received furosemide mixed with albumin, the urine volume was twice what it was when furosemide was administered alone. Of note, the mean serum albumin concentration in this series was 2.0 g/dl.

Subsequent studies have attempted to reproduce and extend these findings, but the results have not been as clear. Chalasani et al. (6) examined the response to furosemide with or without albumin in 13 patients with cirrhotic ascites using a randomized crossover design. The authors detected no effect of concomitant albumin infusion on the response to furosemide in this population. The mean serum albumin in this group, however, was 3.0 g/dl; because the albumin infusion did not alter urinary furosemide excretion in this study, this raises the possibility that the negative results may have reflected the modest degree of hypoalbuminemia. Several meta-analyses have summarized the variable literature, but they have generally concluded that its limited quality precludes firm conclusions.

Charokopos et al. (4) have taken a creative approach to examine the role that hypoalbuminemia may play in the setting of heart failure. They studied consecutive patients who were treated and evaluated in an innovative outpatient transitional care center. Although the group was not selected for acute decompensation or resistance, a second cohort of inpatients admitted with acute decompensated heart failure was used as validation. Unlike most other studies, rather than selecting for patients with hypoalbuminemia, they categorized patients into those whose albumin was above or below the group median and then, compared responses. The results suggest that, in these populations, neither hypoalbuminemia nor proteinuria had a substantial effect on diuretic responsiveness; although there was an association of hypoalbuminemia with less diuretic efficiency, this correlation was eliminated when the confounding effects of inflammation on serum albumin concentration were addressed by adjusting for IL-6 levels.

The results are convincing given the patient population studied, but some limitations deserve emphasis. Although the serum albumin in the outpatient cohort ranged from 2.4 to 4.9 g/dl, hypoalbuminemia was generally mild (mean in the “low” group was 3.4 g/dl). This is similar to the range of values in the well designed study by Chalasani et al. (6) of patients with cirrhotic ascites (range, 2.1–4.3 g/dl), a study that also suggested an absence of benefit. In comparison, the mean serum albumin concentration in the highly cited paper on albumin-furosemide mixing was 2.0 g/dl, with a range of 0.2–3.5 (5). Thus, because the data from analbuminemic rats seem both biologically plausible and scientifically convincing, it remains possible that patients with severe hypoalbuminemia (<2.0 g/dl, for example) may exhibit an increase in diuretic volume of distribution, low tubule delivery, and diuretic resistance. Taken together, therefore, it seems that patients with hypoalbuminemia >2.0 g/dl are unlikely to benefit from albumin infusions whether mixed or administered separately. However, for patients with extremely low serum albumin (<2.0 g/dl), common in some forms of nephrotic syndrome and in critically ill patients, theoretical considerations and limited clinical data support a possibility of benefit.

A second focus of the study by Charokopos et al. (4) was to test whether proteinuria was associated with diuretic resistance. The results again support the null hypothesis, and here again, interpretive caution is indicated. The mean urine albumin concentrations in the two groups were 23 and 90 mg/g creatinine, with the majority of subjects falling in a range indicative of “moderately increased albuminuria” (also known as microalbuminuria). Thus, although these results are relevant for the population studied, they may not be representative for patients with higher or nephrotic levels of proteinuria. This conclusion may have therapeutic implications, because one recent novel hypothesis for diuretic resistance involves the excretion of proteolytic enzymes. Proteolytic cleavage activates epithelial sodium channels (ENaCs). It has been suggested that filtered proteases in the setting of nephrotic syndrome activate ENaC and contribute to sodium retention (7). A recent study concluded that filtered proteases appear in a rat model of heart failure; the same group also detected similar proteases in urine from humans with heart failure (8). Although the recent Aldosterone Targeted Neurohormonal Combined with Natriuresis Therapy in Heart Failure trial did not detect a benefit of diuretic (higher) doses of spironolactone for patients with acute decompensated heart failure, protease ENaC activation might be less dependent on aldosterone action (9). In this case, amiloride, which directly blocks ENaC, might prove to be effective; this hypothesis should be tested.

Unfortunately, the largely negative results of this trial leave us seeking other answers to the conundrum of diuretic resistance in heart failure. Although some resistance is certainly mediated by severe depletion of the “effective arterial blood volume” (a poorly characterized concept), another contributor may be the treatment itself. Kidney tubule cells turn over and proliferate on the basis of local and systemic signals; this process can lead to tubule remodeling in heart failure, much like the better-known cardiac remodeling that occurs after heart damage. We are beginning to understand factors that stimulate kidney tubule remodeling, often in segment-specific ways. Hypokalemia is sensed, for example, by proximal tubules, distal tubules, and collecting ducts, leading very rapidly to cell proliferation; conversely, preventing cell potassium sensing leads to tubule shortening (10). Because these effects are associated with increases or decreases in diuretic responsiveness, they suggest potential new preventive and therapeutic approaches.

In the end, symptomatic heart failure with reduced ejection fraction is typically a homeostatic kidney response to cardiac dysfunction leading to congestion. A goal of treatment is to “trick” the kidneys into excreting salt when they perceive that they should not. We typically choose a therapeutic approach that has been likened to squeezing a balloon, relying mostly on loop diuretics. In the paper by Charokopos et al. (4), for example, 93% of outpatients were receiving loop diuretics before entry, whereas only 12% received a thiazide-type diuretic, and 25% were receiving aldosterone antagonists; additionally, aldosterone antagonists are most commonly administered to patients with heart failure at “nondiuretic” doses. A more balanced approach to diuretic treatment, involving multiple tubule segments targeted earlier in the disease process, would be predicted to minimize the limitations inherent in squeezing the therapeutic “balloon” and perhaps, would better preserve the dynamic range of kidney tubule function. Such an approach should be evaluated in clinical trials, because heart failure is increasingly common and persistently morbid.

Disclosures

None.