Renal haemodynamic, microcirculatory, metabolic and histopathological responses to peritonitis-induced septic shock in pigs

Despite the fact that acute kidney injury (AKI) in critically ill patients is predominantly caused by sepsis and septic shock [1], the pathophysiology of septic AKI is still poorly understood [2]. Although renal vasoconstriction and consequent renal ischaemia and acute tubular necrosis (ATN) occupy a central role in AKI development in hypodynamic states, during sepsis the role and character of haemodynamic alterations within the kidney still remain controversial [2‐4].

It must be stressed that the majority of studies reporting a reduction in renal blood flow were derived from fairly heterogenous, short-term and mostly hypodynamic models characterised by a reduced cardiac output, which therefore only have a limited resemblance with human pathophysiology [2, 3]. By contrast, utilising clinically more relevant models of hyperdynamic sepsis in sheep, Langenberg and colleagues have recently challenged the conventional presumption of renal vasoconstriction as a prerequisite for the development of AKI during hyperdynamic bacteraemia in a sheep model [5‐8]. As suggested earlier [9], the authors provided 'proof of concept' that septic AKI may represent a unique form of hyperaemic AKI [2]. However, further research is needed to establish whether this concept is valid in other clinically relevant models of sepsis-induced AKI. [8, 10].

In addition, the vast amount of experimental studies report on haemodynamic changes only, without providing their relationship to microcirculatory, metabolic and histopathological responses. Prompted by these facts, we dynamically assessed the pattern of renal haemodynamics in a long-term porcine model of progressive hyperdynamic sepsis induced by faecal peritonitis. Furthermore, a potential link between renal haemodynamics and renal cortex microcirculatory, metabolic and histological changes was simultaneously analysed.

Animal handling was in accordance with the European Directive for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes (86/609/EU). The experiments were approved by the Committee for Experiments on Animals of the Charles University Medical School, Plzen, Czech Republic.

There were no statistically significant differences in any measured variables between the sham-operated and peritonitis-induced pigs at baseline.

In this clinically relevant model of hyperdynamic septic shock AKI developed without apparent renal vasoconstriction, renal oxygen consumption did not change and renal histology revealed only subtle changes despite significant kidney cortex microvascular and metabolic stress. Renal venous congestion might contribute to the pathogenesis of septic AKI.

The renal haemodynamic, microvascular and metabolic responses to and morphological sequelae of sepsis remain inconsistent because of marked heterogeneity attributable to the use of different species, models of sepsis, experimental settings and supportive treatment. Hence, the next fundamental step to understand the pathophysiology of septic AKI is the use of animal models designed to meet the criteria of human sepsis/septic shock [13, 14]. However, the majority of studies have been derived from very heterogenous, short-term and mostly hypodynamic models characterised by a reduced CO. By contrast, the sepsis model used in our study replicates many of the features of adequately resuscitated human septic shock (i.e. hyperdynamic circulation, inflammatory response accompanying with nitrosative, oxidative and metabolic stress). The substantial instrumentalisation used offers a broad insight into organ haemodynamic and metabolic pathways, thereby making it an appealing sepsis model in studies of AKI. Moreover, this is underpinned by the fact that the pig kidney is more similar to the human kidney than that of dog, rat or mice because of similar renal anatomy, architecture and lymphatic pattern, urinary concentrating ability, tolerance to ischaemia and medullary thickness [15, 16].

To our knowledge, our study is the first to tackle the issue of directly measured renal venous pressure allowing both the determination of RVR and true renal perfusion pressure in a large animal sepsis model. In keeping with recent reports [5, 6, 8], our study provides further evidence against the widely held concept that sepsis increases RVR [1]. Importantly, it is apparent from the present study that, at least in this model, renal venous congestion leading to decreased renal perfusion pressure might play an important role in mediating fall in glomerular filtration despite clinically acceptable MAP and CO.

It is amazing that there are no animal studies that provide information on the behaviour of renal venous pressure in their sepsis models. The combined impact of both venous congestion due to elevated right atrial pressure and sepsis-induced capillary leak promoting the development of tissue oedema and abdominal hypertension could explain the elevated renal venous pressure and associated reduction in the filtration gradient. Of note, our data suggests that the assumption that renal perfusion pressure is essentially equal to MAP might not be valid under conditions of severe capillary leak and aggressive fluid resuscitation.

The well-preserved renal blood flow does not guarantee adequate perfusion to microvascular beds. Indeed, peritonitis-induced sepsis caused significant reduction in cortical microvascular perfusion in our model, supporting the emerging evidence that renal microvascular dysfunction may be a culprit of septic AKI [17]. Within the limitations arising from renal cortex laser Doppler flowmetry measurements, we can unambiguously determine neither the most affected part of the nephron nor the fate of intrarenal distribution of blood flow in deeper cortex layers and medulla in our model. Nevertheless, recent long-term rodent models of sepsis-induced acute renal failure demonstrated marked decline in cortical peritubular capillary perfusion [18‐20] that was associated with tubular redox stress and preceded the development of renal failure [19]. In addition, the distribution of blood flow from the cortex towards medulla has been suggested by several studies [21‐23], although contradictory results have also been reported [24].

Only a few studies with conflicting data have been performed investigating simultaneous renal haemodynamics and oxygen [23, 25‐28]. In our study, the apparent kidney metabolic stress as evidenced by gradually worsened renal venous L/P ratio (a marker of redox state) and acid base status occurred despite unchanged renal oxygen consumption. The design of the present study does not allow conclusions to be drawn about what processes are responsible for altered kidney energy metabolism. Nevertheless, taking into account an increased renal oxygen extraction at the end of the experiment, our results could indicate that these metabolic alterations may be attributable to the deterioration in microcirculatory perfusion and related tissue hypoxia. Of note, highly heterogenous renal tissue perfusion and oxygen consumption within the kidney make any extrapolation of the total oxygen uptake measurement potentially erroneous and regions or energy requiring pathways (e.g. tubular sodium reabsorption) suffering from hypoxia might have been overlooked [29]. In support of this notion, in endotoxaemic rats, Johannes and colleagues recently provided the evidence for the presence of microvascular hypoxic areas, even though renal oxygen consumption was not significantly reduced and no hypoxia was detected in the average microcirculatory pO2 measurements [30]. Finally, Porta and colleagues recently showed that kidney mitochondrial function was preserved in a prolonged porcine endotoxaemia with well-maintained renal blood flow [28], making the disturbed cellular energy machinery independent of tissue oxygen availability a less plausible explanation for the renal metabolic stress.

Analogous to the renal hypoperfusion paradigm in septic AKI, ATN is generally regarded as the most frequent mechanism of renal failure in critically ill patients [1]. However, there is no published study in septic patients, that would provide conclusive histopathological evidence for the presence of ATN in sepsis-induced AKI and very few experimental studies simultaneously evaluate physiological features of AKI and underlying histopathological changes allowing data to be put into the relevant complex picture. In our model, only subtle histological changes without any signs of ATN occurred despite marked microvascular and metabolic changes. Admittedly, a longer duration of the experiment could have been required for the development of more severe kidney dysfunction and corresponding histological changes. On the other hand, our results are consistent with a recently published systematic review showing only mild, non-specific changes in the majority of clinical and experimental studies [31].

Our study has several limitations. Despite a gradual severity of the septic process, the pigs developed relatively mild AKI, probably as a result of early and aggressive haemodynamic management. Nevertheless, even small changes in serum creatinine (delta 30 μmol/l in our study) achieved within 22 hours suggest significant renal injury, confirmed by some histological evidence. One could also argue that fluid resuscitation with a large dose of hydroxyethylstarch could contribute to the renal dysfunction in this model. However, the available data still remains inconclusive and the safety profile of a new colloid generation as used in our study (6% hydroxyethylstarch 130/0.4) needs to be further clarified. Due to logistical limitations, other important variables of renal function, such as creatinine clearance or tubular reabsorption functions, were not measured. In addition, we did not directly measure intra-abdominal pressure and the absence of techniques to precisely assessing intraglomerular and peritubular microvasculature and tissue oxygenation/energetics do not allow any robust conclusions to be drawn. Finally, the long-term effects of sepsis over several days might give different results.

Within the boundaries of the limitations, the results of our study support the recent evidence arguing against the concept of renal vasoconstriction and ATN as physiological and morphological substrates of early septic AKI and show that renal venous congestion might be a hidden and clinically unrecognised factor in the development of kidney dysfunction.

Work was performed at the animal research laboratory of the 1^st Medical Department at Charles University Medical School.