Effect of the molecular adsorbent recirculating system and Prometheus devices on systemic haemodynamics and vasoactive agents in patients with acute-on-chronic alcoholic liver failure

The natural course of chronic liver disease is often complicated by acute episodes of potentially reversible decompensation, triggered by a precipitating event such as infection or upper gastrointestinal bleeding, and is frequently referred to as acute-on-chronic liver failure (AoCLF) [1‐3]. This complication is associated with a further aggravation of the systemic haemodynamic dysfunction associated with portal hypertension, also called the hyperdynamic state. The hyperdynamic state is characterized by a reduced systemic vascular resistance and mean arterial pressure (MAP), as well as an increased cardiac index, heart rate and total plasma volume.

The pathophysiology is still incompletely understood but an overload of endogenous vasodilatory substances, including nitric oxide (NO), are presumed to play a major role [4‐6]. The combination of vasodilatation and expanded intravascular volume is associated with abnormal distribution of the plasma volume with hypervolaemia in the splanchnic region (splanchnic hyperaemia) and other noncentral vascular territories, while relative hypovolaemia prevails in the central thoracic compartment. Sensor mechanisms interpret this relative central hypovolaemia as a reduced effective circulating volume, which results in an activation of endogenous vasoconstrictor and water-retaining and sodium-retaining systems such as the renin–angiotensin–aldosterone system and the sympathic nervous system, and in the nonosmotic release of arginine vasopressin. Persistent activation of these systems appears to worsen the increased intrahepatic resistance, thus playing a central role in the aggravation of portal hypertension, and promotes vasoconstriction with ischaemia in essential organs, which may lead to the development of the hepatorenal syndrome, portopulmonary hypertension and, finally, multiorgan failure [6‐10]. Several studies have emphasized the relation between the degree of arterial hypotension in cirrhosis and the severity of portal hypertension, hepatic dysfunction, signs of decompensation and survival [8, 11, 12].

AoCLF is at present managed by treating the precipitating event and offering supportive therapy for end-organ dysfunction in the hope that liver function recovers. Unfortunately, 30–50% of these patients will die within 30 days [13].

Extracorporeal liver detoxification devices such as the molecular adsorbent recirculating system (MARS) [14] and fractionated plasma separation, adsorption, and dialysis (FPSA) – the Prometheus system [15] – have recently been proposed as novel treatment options for severe liver failure and its complications (Figure 1). For the MARS device, besides detoxification, several studies have ascribed a beneficial effect on portal and/or systemic haemodynamics [3, 16‐21]. In contrast, the impact of the Prometheus device on these variables remains currently unknown. We therefore aimed to compare, in a controlled manner, potential changes on systemic haemodynamics induced by the MARS or Prometheus system in a group of patients with alcoholic AoCLF and concomitant hyperdynamic circulation, and to evaluate associated changes in vasoactive substances.

After obtaining approval of the Medical Ethics Committee of the University Hospital Gasthuisberg, signed informed consent was obtained from all patients or through their next of kin.

The current concept of the pathogenesis of circulatory dysfunction in cirrhosis is based on the peripheral vasodilation hypothesis [4, 6‐8, 29]. This hypothesis proposes that splanchnic arterial vasodilation is the initiating factor in the systemic haemodynamic dysfunction. At the early stages of disease, there is a homeostatic increase in the cardiac output as a result of the decrease in cardiac afterload and the stimulation of the sympathetic nervous system. With progression of the disease and intensified splanchnic vasodilation and subsequent systemic vasodilation, the cardiac compensation is no longer sufficient to balance the decreasing afterload. The arterial pressure decreases, which leads to a baroreceptor-mediated stimulation of sympathetic nervous activity and of the renin–angiotensin–aldosterone system, and to an increased nonosmotic release of vasopressin in an attempt to preserve circulatory homeostasis. Activation of these systems leads to renal salt and water retention, and to ascites. Additionally, the activation is also responsible for a further aggravation of the active intrahepatic vascular resistance and the development of multiorgan failure in cirrhosis [6, 8, 30].

The exact underlying pathophysiological mechanism to the hyperdynamic state remains unknown but probably represents a multifactorial phenomenon and may involve impaired neurogenic responses, accumulation of vasodilators and diminished responsiveness to a variety of vasoconstrictors [31]. Ample evidence suggests a central role for excessive production of NO in both the splanchnic and systemic vascular territories [31‐36].

In the present study we observed an improvement of the hyperdynamic state during MARS therapy. It is unlikely that this haemodynamic improvement in the MARS group was induced by changes in fluid balance. Indeed, there were no differences between the treatment groups with regard to central filling, serum albumin and creatinine levels. In both dialysis groups, pump speeds and fluid exchange rates were similar. Furthermore, the favourable effect on the MAP in patients treated with MARS disappeared within four days after cessation of treatment, which further emphasizes a causal relationship to MARS. The improvement in the SVRI and MAP therefore strongly suggests temporary changes and/or elimination in endogenous vasoactive substances.

We observed a reduction in NOx levels that correlated negatively with the improvement in the SVRI. Additionally, this was associated with a drop in endogenous vasopressor systems, which are considered to act as counteracting factors for excessive splanchnic and systemic vasodilation, most probably induced by NO. The question of whether NO is directly removed from the circulation by the MARS device is difficult to explore as the metabolic fate of NO is still poorly understood. NO is a labile species with a half-life of only a few seconds [31]. NO therefore acts locally and is degraded or converted rapidly into intermediate metabolites. Putative intermediate metabolites include an array of low molecular weight and high molecular weight thiols, nitrosoglutathione, nitrosohaemoglobin and nitrosoalbumin, some of which might be present in sufficient quantities to exert biological effects [37‐41]. Whether S-nitroso adducts of serum albumin act as a 'haemodynamically active' circulating depot of NO in vivo, as suggested by Stamler and colleagues [38] and Rafikova and colleagues [39], and result in the removal of NO in this way by MARS, remains subject to practical and theoretical criticism [38‐41]. Furthermore, the value of measuring NOx as a parameter of NO synthase activity can also be discussed, but at present it remains by far the most 'simple' and direct index of NO generation [37].

A second explanation of how MARS might affect the systemic characteristics of portal hypertension is that it might directly target the active, modifiable component of the increased intrahepatic vascular resistance by a decrease in the intrahepatic action of vasoconstrictors [16, 42]. In the present study we found a reduction in vasopressor hormones during MARS treatment, suggesting either decreased production or elimination, or a combination of both. Combined elimination and decreased production seems most probable. We could demonstrate the presence of pressor hormones, such as norepinephrine and aldosterone, in the closed loop albumin dialysate, whereas renin was undetectable in dialysate samples despite decreased serum levels after MARS treatment. Because of its molecular size, renin is not removable by MARS, indicating that MARS exerts control on the degree of counteractivation of endogenous vasoconstrictor systems. This latter observation suggests that the reduction in vasopressors is a consequence of the improved haemodynamic situation, therefore favouring the hypothesis of primarily eliminating a systemic vasodilating substance, such as NO.

A third possibility involves removal of inflammatory mediators due to the inflammation-related precipitant of the AoCLF in all of these patients, more specifically alcoholic steatohepatitis. Tumour necrosis factor alpha, a highly expressed proinflammatory cytokine, particularly in alcoholic steatohepatitis, is currently considered the most important vasoactive inflammatory mediator in this context [43]. More specifically, it has been shown that antagonism of tumour necrosis factor alpha with anti-tumour necrosis factor alpha antibody attenuates portal hypertension and the associated hyperdynamic state, both experimentally [44] and in humans [45].

Sen and colleagues [46] studied the effect of MARS on tumour necrosis factor alpha in patients with alcoholic AoCLF and found no changes in plasma levels after seven days of MARS treatment, despite a documented removal of tumour necrosis factor alpha and its receptor TNF-R1, suggesting that the concurrent cytokine production due to the disease process itself balanced any removal. In the current study, inflammatory markers, such as leukocytosis and C-reactive protein, remained unchanged in all groups, which supports the findings of Sen and colleagues [16, 46]. Together these data re-emphasize the hypothesis that MARS removes a putative circulating vasoactive circulating factor from the peripheral blood independent of changes in vasoactive cytokines.

Our results further suggest that MARS is effective in temporarily improving haemodynamics, while the Prometheus system does not. In contrast, the Prometheus device showed a higher ability of clearing albumin-bound toxins [24, 47] than MARS. A possible explanation for this finding relates to conceptual differences (Figure 1). The Prometheus device is based on FPSA, which involves selective filtration of the native albumin ('fractionated plasma separation') through a specific albumin-permeable polysulfon filter (with a cutoff value of 250 kDa) into a secondary circuit. In this secondary circuit, the albumin fraction is then directly purified by adsorption on a neutral resin adsorber and an anion exchanger, and thereafter returned to the plasma. In addition to the FPSA step, high-flux haemodialysis is performed. The filtration step presumably is the predominant factor in the lack of haemodynamic effects [48].

Besides a decreased potential to eliminate the investigated endogenous vasoactive substances, as documented by the lower calculated reduction ratios for these hormones compared with those of MARS, the FPSA step of the Prometheus system might also have been responsible for minimal changes in colloid osmotic pressure during the procedure. These minimal changes could have caused subclinical volume shifts that prohibited the deactivation of the homeostatic baroreceptor response [15]. Indeed, in our study the concurrent production of endogenous vasoactive substances itself balanced any removal since no differences in the levels of these agents were observed at the end of treatment compared with the beginning of treatment. The increased heart rate, as a compensatory mechanism, might also be an indirect index of subclinical volume shifts.

Several factors make patients with AoCLF extremely sensitive to volume changes, such as the decreased effective circulating volume with persistent counteracting vasopressor response, the increased cardiac output with impaired end-diastolic filling and the autonomic vascular dysfunction [49].

The present study demonstrates that MARS improves haemodynamic disturbances in patients with acute on chronic alcoholic liver failure with associated hyperdynamic circulation. These beneficial haemodynamic effects can be explained by the removal of circulating endogenous vasoactive substances and are in agreement with the current hypothesis of the pathogenesis of portal hypertension.