Trigger mechanisms of secondary sclerosing cholangitis in critically ill patients

In the case of an incipient circulatory collapse with arterial hypotension, deterioration of the hepatic perfusion with subsequent alterations of the biliary blood supply is to be expected because the perfusion of the liver can be maintained by autoregulation only to a limited extent [15]. All of our 16 patients initially survived an episode of severe haemodynamic instability with a decrease in MAP to <65 mmHg. The later this hypotensive episode occurred during the treatment course, the later cholestasis appeared. Shock states with decreased arterial pressure are common in ICU patients – according to reports in the literature up to 58% of ICU patients are affected [16]. In comparison, the rate of 100% in our series was very high, supporting a possible role of severe hypotension in the initiation of the pathogenesis. When fluid administration fails to restore adequate systemic pressure in such a clinical situation with a critical blood pressure decrease, treatment with vasoactive agents is initiated. Therefore, all 16 of our patients required vasopressors and/or inotropes. However, since the increase in systemic blood pressure does not always correlate with improved hepatosplanchnic perfusion, prolonged liver ischemia can be masked. Commonly used agents, such as epinephrine, norepinephrine, dopamine and dobutamine, all increase systemic blood pressure, but have different effects on hepatosplanchnic blood flow. Dopamine not only increases the mesenteric blood flow, but apparently also has a positive effect on liver perfusion [17]. Only 2 of our 16 patients received dopamine. In contrast, epinephrine and norepinephrine have a dose-dependent negative effect on the perfusion of visceral organs. On account of its vasoconstrictive properties, norepinephrine has the potential to decrease splanchnic blood flow. Epinephrine, in particular, is known for its disadvantageous effect on splanchnic perfusion in septic patients [18,19]. In our series, norepinephrine was the most frequently used vasopressor (13/16 patients), and epinephrine was used only rarely (4/16 patients); one patient received neither drug. The maximum dose of epinephrine or norepinephrine exceeded 0.2 μg · kg⁻¹ · min⁻¹ in the minority of patients (5/16). The majority of patients received catecholamines in the medium or low-dose range. Neither in the epinephrine (E) nor in the norepinephrine (NE) group was the mean continuously administered dose (MCAD) within the high-dose range (>0.2 μg · kg⁻¹ · min⁻¹); in both groups, the MCAD was mostly in the range below 0.1 μg · kg⁻¹ · min⁻¹ (E: n = 2/4, NE: n = 12/13). Overall, our results do not support the earlier assumption that high-dose catecholamines play a decisive role in the pathogenesis of SSC-CIP.

As is known from experimental models, ischemic bile duct damage increases as the size of occluded arteries decreases [20-22]. This indicates that microcirculatory disturbances involving the peribiliary plexus are of particular clinical significance. The question as to why SSC-CIP occurs much more often in surgical patients has yet to be answered. Two-thirds of all published cases of SSC-CIP were patients with trauma, burns or major surgery. Also in our group, 13 of the 16 cases had surgery-requiring primary diseases (Table 1). Polytrauma or extensive surgery is usually accompanied by extensive tissue damage. Clinical studies show that this tissue damage can be an initiator of a hypercoagulable state [23-28]. This might well contribute to occlusion of the peribiliary plexus. Ischemic bile duct damage has frequently been found in a number of conditions known to be associated with hypercoagulopathy, such as the antiphospholipid syndrome, paroxysmal nocturnal haemoglobinuria, Schönlein-Henoch purpura or systemic lupus [29-33]. Other important factors affecting the microcirculation are increased blood viscosity and red cell aggregation. Trauma, burns and major surgery are all conditions that may increase plasma viscosity. Furthermore, extensive muscle destruction, electric power accidents and post-ischemic syndrome are often associated with a marked release of myoglobin (crush syndrome) which can result in an increase in blood viscosity, as shown in animal experiments [34]. However, the infusion of colloidal solutions, such as hydroxyethyl starch (HES), can also increase whole blood viscosity [35,36]. HES is also one of the components of the University of Wisconsin (UW) solution, used in the preservation of donor livers for transplantation. Unfavourable effects of this solution on blood flow through the peribiliary plexus have been reported in clinical studies, and are supported by experimental data [37,38]. In recent years, it became clear that there are risks of RBC transfusion related to RBC storage effects. Stored RBCs adhere to the endothelium or lose their deformability, resulting in obstruction of the microcirculation. Increased mortality rates associated with RBC transfusion have been reported in cardiac surgery patients and after liver transplantation [39-41]. Of our patients, 81% (13/16) received blood transfusions prior to onset of cholestasis. This is a very high proportion compared to the percentage (45%) generally reported in ICU patients [42,43]. Increased blood viscosity caused by colloids or myoglobin release, RBC transfusion and hypercoagulable states are among the factors determining microcirculatory blood flow through the PBP. All 16 patients had at least one of these factors – often several simultaneously – prior to the onset of cholestasis. These observations suggest that apart from severe hypotension, additional microcirculatory disturbances may be involved in the pathogenesis of SSC. Possibly, severe hypotension may have bile duct implications only when it occurs in combination with additional microcirculatory disturbances.

Experimental studies indicate that mechanical ventilation with high PEEP for lung-protective purposes has negative effects on the microcirculation in the gastrointestinal tract [44]. Based on these studies, it was suspected, that, in particular, a high PEEP contributes to biliary ischemia, and that patients with a low oxygenation index are at risk of developing SSC-CIP [2]. Our results demonstrate that the development of SSC-CIP is not conclusively associated with a PEEP >10 cm H₂O or a PaO₂/FiO₂ < 150 mmHg. However, this does not exclude the possibility that mechanical ventilation itself might be of significance, since all cases of SSC-CIP described in the literature as well as our own 16 patients, received mechanical ventilation.

Pro-inflammatory cytokines such as TNF-α, IL-1 or IL-6 down-regulate the function and/or expression of hepatobiliary transporters. Animal models indicate, for example, that the canalicular bilirubin-conjugate-export pump (MRP2) is sensitive to cytokines and that the mRNA of the bile salt export pump (BSEP) is down-regulated under the influence of inflammatory mediators [45-47]. These data have been confirmed in human hepatocytes [48]. However, data regarding the phospholipid export pump (MDR3) are not conclusive, but some suggest a decrease in its hepatic expression in experimental inflammation models [47,49]. At the biliary transporter level, pro-inflammatory cytokines inhibit the activity of the anion exchanger AE2 [50]. In the clinical setting, the development of SIRS reflects cytokine release. Such cases of SIRS developed in all of our patients prior to cholestasis. SIRS is a common finding in patients in the ICU, usually occurring in 68% of the cases [51]. By comparison, the 100% incidence of SIRS in our group was exceptionally high. Therefore, if plausibility, temporality and consistency (Hill criteria) are given, SIRS could play a role in triggering SSC-CIP. Our patients exhibited signs of SIRS within 4.4 ± 2.9 days of ICU admission (range 2 to 11, median 3 days). However, up to this point, a pathogen was isolated in only 10/16 patients. Thus, infections do not adequately explain the development of SIRS in our patients. The release of pro-inflammatory cytokines is not limited to bacterial infections, but is also triggered by trauma, severe burns, cardiac surgery and other major surgery [52]. Moreover, prolonged hypotension or shock can also induce SIRS via generalised tissue ischemia [53]. In 14 of our 16 patients SIRS was preceded by severe hypotension, which could explain the non-infectious SIRS in our series.

All 16 patients in our study had experienced an episode of severe hypotension. Experimental data suggest that ischemia/hypoxia can also down-regulate hepatobiliary transporters [54].

Propofol, which is commonly used for sedation in ICU patients, is thought to influence hepatobiliary transporters via the release of cytokines. While high-dose propofol (5 to 10 mg kg⁻¹ h⁻¹) increases serum levels of TNF-α and IL-10 after haemorrhagic shock, low-dose propofol (1 mg kg⁻¹ h⁻¹) decreases them [55]. Since only one of our patients received high-dose propofol, this mechanism played no key role in our patients.

Clinical studies have shown that prolonged parenteral nutrition can trigger cholestasis. In a mouse model the expression and function of the canalicular phospholipid flippase (Mdr2) significantly decreased under the influence of TPN [56]. However, 25% of our patients did not receive TPN prior to the onset of cholestasis. In the 75% with prior TPN, the average TPN duration was very short (five days). In contrast, TPN-induced cholestasis usually occurred only after two to three weeks of therapy. Likewise, the results of a multicentre study by Grau et al. showed no impairment of liver function within the first six days of TPN [57]. In the light of these findings we do not attribute any major role to TPN in the pathogenesis of SSC-CIP in our patients.

Inhibitory effects of antimicrobial substances, such as rifampicin and fusidate, on the function of hepatobiliary transporters have been reported [58,59]. In a recent study by Yosikado et al. itraconazole was shown to inhibit MDR3-mediated efflux of phospholipids in vitro and led to itraconazole-induced cholestatic liver injury in vivo [60]. Although all of our patients received antibiotic therapy, the spectrum of the substances was so heterogeneous that a common pathogenetic pathway appears improbable.

A number of cases of ketamine-induced cholangiopathies were recently reported [61-63]. A large proportion of our patients (15/16) received ketamine prior to the onset of cholestasis, and it cannot be excluded that this had an unfavourable effect on the course of the illness.