Venous Congestion and Renal Function in Heart Failure
For quite some time, venous congestion has been suspected as a cause of renal dysfunction (RD). Notably, in experiments dating back to the 1860s, it was shown that partial occlusion of the renal vein led to an immediate decline in renal blood flow, glomerular filtration rate (GFR), and sodium excretion, with resolution of the abnormalities after relief of the congestion. The concept of decreased perfusion pressure (arterial pressure – venous pressure) causing organ dysfunction is an intuitive explanation for these observations. Perhaps as a result, the recent publication of several studies demonstrating an association between venous congestion and RD appeared to confirm that the observations from animal models translate to human heart failure (HF). However, upon close inspection of these data, we see that many of the reported associations were actually not consistent with findings from animal models. Notably, the association between baseline RD and venous congestion was heterogeneous across studies, and either improvement, no change, or even worsening of renal function (WRF) was observed with relief of the congestion.

One possible explanation for these seemingly contradictory results could be that some studies were wrong. However, it is more likely that the interaction between biology, treatment, and epidemiology resulted in a situation substantially more complex than that found in an animal with a partially occluded renal vein. That being said, even in the experimental setting, the physiology appears to be more complicated than a simple direct hydraulic effect of decreased perfusion pressure causing worsened organ function. One of the primary mechanisms actually appears to involve an increase in renal vascular resistance, including an increase in afferent arteriolar resistance. Notably, this reflex appears to involve the sympathetic nervous system, since intrarenal infusion of phentolamine or renal denervation largely abolishes the effect. Additionally, a direct effect of increased renal interstitial pressure causing tubular collapse and opposition of flow into Bowman's space has been proposed. However depending on the hydration status of the animal, an experimental increase in renal venous pressure produces differential effects on renal interstitial pressure and renal function, highlighting the secondary importance of direct hydraulic effects. Notably, in dogs without volume overload, an increase in renal venous pressure up to 40 mmHg actually caused an increase in sodium excretion and no change in GFR or renal blood flow.

However, it must be stressed that the venous congestion induced in these experimental set-ups is completely different from that seen clinically in HF. In these experiments, the most common approach was to constrict the renal veins selectively. In human HF, the entire venous system is congested. This is important due to the pronatriuretic/renal-preserving effects of overall venous congestion that may counterbalance the direct local effects of renal venous congestion (i.e. increased natriuretic peptide release and decrease in renal sympathetic nerve activity with cardiac distention). Furthermore, the techniques by which venous congestion is investigated in human research vary by study, and each has its own limitations, such as limited reproducibility and large interobserver variability. For example, the most commonly employed surrogate for venous congestion has been central venous pressure (CVP). However, we often forget that due to the remarkable compliance of the venous system, pressure and volume have very poor correlation. A recent meta-analysis of studies comparing CVP with directly measured blood volume found a pooled correlation coefficient of 0.16, indicating that only ~2.5% of the variability in CVP across a population can be explained by differences in blood volume. The reason for this lack of correlation stems from the fact that venous pressure is the combined representation of both volume and venous tone, a relationship evolved for the critical purpose of maintenance of cardiac preload over the physiological range of blood volume encountered in life. This is relevant to the study of venous congestion-induced RD for two reasons. (i) In animal models, volume status appears to dictate the effect of venous pressure on renal function. As such, using a metric of congestion that does not correlate with volume may be problematic. (ii) The primary mediator of increased venous tone is neurohormonal activation. Therefore, any association between CVP and renal function reflects not only congestion, but also the neurohormonal status of the patient and their response to neurohormonal blocking therapies.

The situation is further complicated by the remarkably complex epidemiology of changes in renal function in patients with acute HF. It is clear that renal function represents the intersection between the patient's specific clinical HF profile, treatment, disease severity, and co-morbid conditions. The net amalgamation of these characteristics, which is what produces the observed renal function, obviously varies dramatically across populations. Furthermore, venous congestion is not the only parameter influencing GFR in these patients. Depending on the specific HF phenotype, the therapeutic approach, and the patients' response to this approach, changes in renal function that result from attempts at decongestion can vary considerably. Additionally, the mechanism underlying WRF can vary depending on when the WRF is detected. Very early WRF can occur before effective treatment begins, primarily reflecting the decompensation that caused the admission. This could be followed by a period of improvement in renal function with successful decongestion; however, with continued diuresis the creatinine could worsen again, a pattern which may actually be common. Even if every patient in a hypothetical study had exactly this pattern of change in creatinine, depending on when the renal function was investigated, very different conclusions regarding the relationship between venous congestion and changes in renal function would result. Finally, human observational studies on venous congestion will always be threatened by potential confounding by disease severity. Since patients are not randomized to their degree of congestion, congestion will always identify sicker patients. Thus any associations we find may be cause and effect from the congestion or the natural history of the disease/treatment of these sicker patients.

In the present issue of the journal, Aronson and co-workers present yet another piece of the complicated puzzle linking congestion, diuresis, and changes in renal function. Using the Vasodilation in Management of Acute CHF (VMAC) trial population, the authors studied 475 patients with decompensated HF, of which 238 had protocolized right heart catheterization data available over the first 24 h. The authors found that baseline CVP had a weak association with baseline renal function (r = –0.17), but there was no association between baseline or change in CVP with the volume removed over the first 24 h, nor was there an association with the incidence of WRF up to 14 days. The authors did, however, find a strong association between an increased volume of diuresis in the first 24 h of the hospitalization and a lower incidence of WRF. Although these findings would seem to be in conflict with several studies in the literature, for the reasons outlined above these results should be regarded not as contradictory but rather complementary to the existing literature, providing an additional vantage point of this very complicated topic. However, also for these same reasons, interpretation of the current findings is challenging. For example, the observation that less WRF was found in patients with the greatest degrees of diuresis could mean that aggressive treatment of congestion improves renal outcomes. An equally probable explanation is that patients who are non-responders to initial treatment are sicker and thus predisposed to developing WRF as treatment is escalated, a situation which may be called diuretic resistance. This scenario is supported by the fact that the plasma refill rate (the rate at which fluid removed from the vascular space can be replaced from extravascular stores) in severely congested patients has been reported to be in excess of 750 mL/h early in the treatment of decompensated HF. Even with the removal of ~7.5 L of fluid (over the first 24 h), blood volume has been reported to remain constant in severely congested patients. In VMAC, the median rate of fluid removal over the first 24 h averaged <50 mL/h. As such, in all likelihood, the intravascular volume probably changed very little during this period, making it difficult to conclude that relief of congestion drove the lower rate of WRF.

Despite research on the topic for >150 years, a clear picture of pathophysiology and clinical/treatment implications of venous congestion-induced RD in HF has yet to emerge. However, there are several points that the study of Aronson et al. and other studies have highlighted. (i) Experimentally induced selective renal venous congestion usually results in significant WRF, in several animal species. (ii) Data are accumulating that this physiology may also be important in human HF. (iii) CVP is probably not the ideal metric for the study of this phenomenon or evaluation of patients with congestion-induced RD. (iv) WRF is an epidemiologically complex phenomenon with associations that vary dramatically depending on how it is defined, when it is measured, and the clinical setting in which it is being evaluated. That being said, the fact remains that volume overload is incredibly common in HF, and RD influences a great deal of our therapeutic options for these patients. As such, continued efforts by the research community to better understand and treat this complicated physiology will probably yield significant benefit for the care of our patients.