Introduction
Sepsis is a severe and frequent clinical condition in the intensive care unit (ICU) with an associated mortality rate varying between 35 and 50% in septic shock [
1‐
3]. The host response to infection leads to organ failure in patients with sepsis [
1]. One of the severely affected organs is the kidney, with sepsis being the leading cause of acute kidney injury (AKI) in critically ill patients [
2‐
4]. Moreover, patients with sepsis-AKI are twice as likely to die as patients with sepsis without AKI [
5,
6].
For a long time, acute tubular necrosis (ATN) due to hypoxia together with the severe hyperinflammatory response were thought to be the main drivers of renal failure in patients with sepsis-AKI [
4]. Renal tubuli have a marginal oxygen supply combined with high oxygen consumption, yet a reduction in renal blood flow below critical limits is not uniformly reported in clinical sepsis-AKI studies [
5]. Additionally, hemodynamic impairment has not been found to be the main driver of renal failure in patients with sepsis [
6‐
8]. During the last three decades numerous clinical trials targeting supposedly pathophysiological mediators within the devastating cascade of inflammatory mediators in sepsis have failed to improve patient outcome. As a result, mortality in sepsis remains unacceptably high [
9]. This failure undermines the validity of the hypothetical causality of these mediators in sepsis and underscores the limited understanding of the pathogenesis of sepsis and resulting organ dysfunction [
10]. Recently, an elegant sheep model of gram-negative sepsis with intensive care treatment, combined with sequential biopsies, also suggested that hypoxia and inflammation cannot fully explain sepsis-AKI [
11]. Histological data addressing morphological damage such as ATN in patients with sepsis-AKI are scarce [
12]. In the past, most data were acquired from post-mortem studies conducted hours or even days after death. As a result, tissue autolysis and post-mortem processes may have hampered pathophysiological interpretation [
13]. However, recently 2 important studies reported histopathological findings on post-mortem biopsies taken immediately after death [
14,
15]. Collectively, the available human tissue data also do not support the hypothesis that hypoxia and inflammation are the major underlying causes of sepsis-AKI, since only limited inflammation, coagulation and cell death have been identified [
12,
14,
15]. A better understanding of the underlying pathophysiological mechanisms of AKI in patients with sepsis is indispensable for the development of a therapy that will improve outcome [
16,
17]. Therefore, renewed orientation on human tissue with a broader scope might reveal additional pathophysiological mechanisms [
13,
18,
19].
The aim of this study was therefore to investigate and expand the histopathological profiles of the glomerular and tubulointerstitial compartments in human sepsis-AKI, including inflammation, coagulation, cell cycle, and repair using immunohistochemical analysis. Compartmentalized detection and quantification of these features may allow us to set the first steps towards therapeutic interventions that more specifically target cell subsets and organ niches in sepsis-AKI.
Discussion
The failure of clinical trials that aim to improvwe the outcome of patients with sepsis-AKI stresses the need for a more detailed understanding of the pathophysiological processes underlying the development of human sepsis-AKI. The aim of this study was therefore to investigate the missing histopathological information on the glomerular and tubulointerstitial compartments in sepsis-AKI, such as renal inflammation, fibrin deposition, cell proliferation, and repair.
The results from our study corroborate findings from recent animal and human studies that show that sepsis-AKI cannot be explained solely by morphological changes. The landmark study by Takasu and colleagues showed that tubular injury in sepsis was common but focal [
14]. Moreover, in a sheep model in which sepsis was induced by continuous intravenous infusion of live bacteria, the anatomical structure of the kidney was intact [
30]. We found renal tubular damage was unequally distributed and limited in sepsis-AKI. These findings agree with the Takasu study and support the premise that tubular damage cannot fully explain the renal function impairment often found in patients with sepsis [
14]. Moreover, our findings also support those found by Lerolle and colleagues in which they demonstrated that AKI in 19 patients with sepsis was associated with intense infiltration of glomeruli, interstitial capillaries, and occasionally tubular lumens, by predominantly monocytic leucocytes [
15].
Interestingly, the glomerular architecture was hardly affected in sepsis-AKI. While structural defects were not apparent, the molecular mechanisms regulating endothelial permeability, one of the determining factors of glomerular filtration, may explain the decreased glomerular filtration rate in these patients. Our previous studies identified altered mRNA levels of the endothelial molecules regulating vascular permeability in this cohort of patients [
22,
23], and in animal models of lipopolysaccharide (LPS)-induced AKI [
23].
Although apoptosis is considered one of the main mechanisms of tubular damage in AKI, only 21 of the 102 experimental studies on sepsis-AKI focused on apoptosis [
31]. In these studies, 158/170 animals had tubular epithelial cell apoptosis. In previous studies of patients with sepsis there were only slight increases in the amount of apoptotic tubulointerstitial cells [
12,
13], or the absence thereof [
32], suggesting a limited role for apoptosis in sepsis. Lerolle and colleagues observed apoptosis of tubular cells and occasionally glomerular cells in patients with sepsis [
15]. In contrast, in sepsis-AKI we identified a varying (up to high) extent of apoptosis in the tubulointerstitium but not in the glomeruli. Our findings, and the discrepancy between animal and human studies, suggest that the role of apoptosis in the tubulointerstitial compartment needs to be revisited by expanding these types of studies in human AKI before rightful conclusions on its role in AKI can be reached.
Local microvascular thrombosis, a possible consequence of low flow and/or microvascular endothelial activation [
33], might be another mechanism propagating tissue damage in sepsis-AKI. We found low amounts of fibrin deposition in the glomeruli, similar to Lerolle et al., who observed glomerular fibrin depositions in only 1/19 patients with sepsis [
15]. In contrast, while we found abundant fibrin deposition in the peritubular capillaries, these authors report partial or complete thrombi in the afferent arterioles in four patients, but no significant fibrin deposition in the peritubular capillary system [
13]. Similarly, Takasu et al. did not find fibrin deposition in peritubular capillaries in post-mortem kidneys from patients with sepsis [
14].
Among the cellular players held responsible for tissue damage in sepsis-AKI, neutrophils are the first to infiltrate the tissues. Neutrophil blockade or depletion in experimental animal models have varying results in preventing AKI [
34‐
36]. Here, we found a notable but limited number of neutrophils in the glomeruli and tubulointerstitium, including in the peritubular capillaries. However, no renal cell injury could be attributed to this neutrophil influx. Lerolle et al. reported that neutrophil infiltration was limited to the peritubular capillaries [
15]. Neutrophil function is ambiguous as these cells are also required for tissue regeneration [
34,
37‐
39]. Which neutrophil function prevails in human sepsis-AKI is an important future research focus.
Lymphocytes play an important role in the cytokine storm characterizing sepsis. In murine endotoxemia, T lymphocytes were shown to be modulators of kidney function and responsible for renal neutrophil recruitment [
40]. Moreover, T-lymphocytes are important in the development of AKI in experimental ischemia reperfusion injury (IRI) [
41]. T lymphocytes were not found to be important in sepsis-AKI in our study, as these cells hardly invaded the kidney. The role of B lymphocytes in sepsis-AKI has not been studied extensively. The limited experimental evidence available suggests that B lymphocytes may slow down or limit the repair process of the kidney after an ischemic insult. Moreover, B-lymphocyte deficiency was shown to be protective against ischemia reperfusion injury [
42,
43]. We hardly found any B lymphocyte infiltration in the kidneys of patients with sepsis-AKI. Together these findings suggest that inflammatory damage mediated by B lymphocytes or T lymphocytes seems to play a minor role in human sepsis-AKI. Nevertheless, functional contributions of these cells cannot be excluded in this observational study.
The dual role of macrophages in damage and repair in sepsis-AKI is elusive. Renal mononuclear phagocytic cells consist of different subtypes, of which type I macrophages are involved in inflammation and tissue destruction, while type II macrophages attenuate the inflammatory response and are active during tissue repair [
26,
27,
29]. We observed an accumulation of macrophages of both subtypes around the glomeruli and in the glomerular capillaries, but not in the tubulointerstitium of patients with sepsis-AKI. The immediate vicinity of both macrophage subtypes may suggest inflammation and repair being simultaneously active in time and space. Type II macrophages in sepsis-AKI glomeruli may originate from the systemic circulation, but also from activated, proliferating local cells [
44]. We also found occasional macrophages expressing markers of both macrophage subtypes. A similar observation was previously described in the kidneys of rats with glycol-induced rhabdomyolysis [
45]. These cells were proposed to transition from type I to type II and were shown to increase in numbers during recovery [
45].
We also investigated cell proliferation as a marker of tissue repair. We found increased proliferation in glomeruli and tubulointerstitium of sepsis kidneys, concurring with the findings of Takasu et al. [
14]. Although the identity of proliferating cells was not explored, their presence suggests increased regeneration of renal tissue as a mechanism of repair. Interestingly, the amount of α-SMA-expressing myofibroblasts, which initially contribute to repair through matrix deposition, was reduced in sepsis-AKI glomeruli and unchanged in the tubulointerstitium. Sandbo et al., identified inhibition of α-SMA expression at the protein and mRNA level in LPS-stimulated vascular smooth muscle cells of the human aorta, human coronary artery, and rat aorta [
46]. They found that decreased levels of α-SMA correlate with diminished cellular contractile function. Renal scar tissue formation and fibrosis was not abundant in sepsis-AKI although obviously, long-term effects were not studied. The role of repair mechanisms in sepsis-AKI is understudied but preliminary data showing signs of overactivity and underactivity justify further research.
The strength of this current study lies in the early retrieval of kidney biopsy tissues directly post mortem, minimizing the influence of autolysis on subsequent analyses. Indeed, our current and previous studies and studies by others [
22,
23,
47,
48] show the feasibility of advanced immunohistochemical, mRNA, proteomic, and metabolomic analyses of these biopsies. When integrated, the resulting datasets will allow better and extensive patient characterization and will hopefully pave the way towards precision and even personalized medicine in sepsis-AKI [
19].
Despite this future promise, our study has limitations, mainly the heterogeneity of patients with sepsis-AKI and therefore the number of included patients. To correlate our heterogeneous findings with the heterogeneous clinical characteristics of our patients and draw meaningful conclusions is not possible. Large data and tissue banks will be required for multivariate analyses. Most patients with sepsis are relatively old and vary widely in clinical presentation. Comorbidities and causative microorganisms influence renal pathophysiology. Additionally, some patients may have had unknown previous renal disease. We must emphasize that the time between the start of sepsis and death varied considerably between patients. Additionally, the biopsies were taken after a prolonged, variable process of illness and dying, both of which might have interfered with the observed AKI-related changes. We only studied biopsy tissues from deceased patients, which confines our results to the sickest patient group. Control patients have comorbidity as well, and the surgical procedure to procure the kidney with clamping of the renal artery might have already induced histopathological changes [
49]. The choice of control group can be debated. Furthermore, it is of note that all biopsies were reviewed by one experienced renal pathologist which improves consistency but interobserver variability is well-known among pathologists. Moreover, the histopathologic scoring followed the routine pathology procedure in our hospital and therefore could not be blinded, which may have introduced some observer bias. We also do not know whether our pathophysiological findings in septic kidneys are the effect of renal processes or processes elsewhere in the body. Our observations described here might be an epiphenomenon, which makes the question “What causes renal function loss in sepsis?” even more enigmatic.