Targeting inflammation for the treatment of Diabetic Kidney Disease: a five-compartment mechanistic model

During stress or inflammation, such as those triggered by hypoxia, ATP or other damage-associated molecular patterns (DAMPs) in the DKD microenvironment, can result in kidney cells expressing chemokines and adhesion molecules. Together this attracts circulating leukocytes into the damaged renal tissue and activates resident cells [23, 42].

Glomerular endothelial cells, mesangial cells, podocytes and tubular epithelial cells express multiple cytokines, including the interleukin (IL) family members (e.g. IL-1, IL-6 and IL-19) tumor necrosis factor-α (TNF-α); multiple chemokines, including C–C motif ligand 2 (CCL2); and multiple adhesion molecules, including vascular cell adhesion molecule 1 (VCAM-1), intercellular adhesion molecule 1 (ICAM-1) and selectins (Table 1) [27]. Activated monocytes/macrophages recruited into the kidney amplify the response by producing pro-inflammatory cytokines (e.g. IL-1β, TNF-α, IL-6 and IL-18) [22]. Activated immune cells also produce multiple molecules that can cause further renal injury, including metalloproteases, reactive oxygen species, advanced glycation end-products and complement proteins [22, 43, 44].

Recruitment of monocytes and macrophages into the kidney is a key step in the pathophysiology of DKD [22]. Macrophage accumulation in the kidney correlates strongly with serum creatinine levels, interstitial myofibroblast accumulation and interstitial fibrosis scores [12, 14, 15]. Resident macrophages and dendritic cells in the tubulointerstitium also contribute to disease progression by recruiting and activating lymphocytes [45]. Non-classical renal ‘patrolling’ monocytes may further orchestrate immune cell responses at the glomerular vascular interface, including recruitment and activation of neutrophils [46, 47]. Interestingly, cell-to-cell communication of resident immune cells, such as macrophages, with renal cells has also been shown to regulate transendothelial transport of immune complexes, as well as other immunoregulatory pathways [23, 48].

Macrophages may progress from the M1-like pro-inflammatory phase to the M2-like tissue repair stage, however, both forms coexist during chronic inflammation in DKD and represent a spectrum of macrophage phenotypes, leading to fibrosis [21]. Pro-fibrotic mediators released by macrophages induce extracellular matrix deposition, leading to fibrosis and impaired renal function [10, 49].

In the early stages of DKD, glomerular hyperfiltration may cause kidney injury and contribute to disease progression by increasing physical stress and oxygen demand to drive reabsorption [29, 50]. Current standard of care, inhibition of the renin angiotensin aldosterone system and sodium-glucose co-transporter-2 (SGLT2), is reducing glomerular injury and can provide significant clinical benefit for patients [29, 50, 51]. However, isolated glomerular hyperfiltration does not predict development of advanced DKD, consistent with a pathophysiological role for chronic inflammation [51].

Chronic inflammation adversely affects all three components of the glomerular filter: the fenestrated endothelium, the basement membrane and the epithelium, comprising podocytes and slit diaphragms (Fig. 3) [52]. TNF-α secreted by resident and infiltrating macrophages is cytotoxic to glomerular mesangial and epithelial cells, and impairs glomerular hemodynamics and filtration [27]. ICAM-1 and E-selectin expression in the glomerular endothelium is induced by pro-inflammatory cytokines and promotes leukocyte recruitment [21, 27].

Inflammation-mediated alterations impair the filtration function of the glomerulus, which relies on size and charge separation. Filtration rate depends on the pressure across the filter and its coefficient of filtration, which is determined by its structural composition and surface area (Fig. 4). The filter’s size and charge selectivity are lost and its permeability increases as glomerular endothelial cells lining the capillaries lose their characteristic fenestrations and as their overlying glycocalyx layer is degraded [53]. Thickening of the glomerular basement membrane due to extracellular matrix deposition correlates with progression of albuminuria and declining GFR, and is one of the earliest histological signs of DKD [8, 52]. The membrane may double in thickness in people with diabetes, from a normal thickness of around 300–350 nm [8]. Podocyte stress, injury and eventual loss are also key factors in disease progression. Podocytes create slit diaphragms for filtration with their foot processes, secrete basement membrane components, communicate with fenestrated endothelial cells, and endocytose proteins that pass through the barrier [54]. Injured podocytes retract their foot processes, disrupting the structure of the filter, and this ‘effacement’ leads to development of proteinuria [55].

Excess protein, advanced glycation end-products, growth factors and complement proteins in glomerular filtrate harm proximal tubular epithelial cells and activate pro-inflammatory responses [45]. Inflammatory and pro-fibrotic cytokines trigger de-differentiation of tubular epithelial cells, leading to loss of resorptive and secretory activities and acquisition of mesenchymal cell-like features, including production of extracellular matrix [56]. Disruption of normal uptake of proteins from the glomerular filtrate by proximal tubular epithelial cells contributes to development of proteinuria [57]. The relative contribution of tubular pathology might increase if SGLT2 inhibitor use slows progression of glomerular pathology, potentially increasing the importance of this compartment. Tubular epithelial cells produce pro-inflammatory and pro-fibrotic cytokines and chemokines including CCL2, IL-8, TGF-β and CCL5 (also known as RANTES) [29]. The resulting inflammation leads to tissue damage, with resulting hypoxia, apoptosis, tubular atrophy, disconnection of the tubule from the glomerulus, and eventual renal failure [58].

Mesangial expansion with eventual fibrosis is a histopathological hallmark of DKD. Excessive accumulation of extracellular matrix leads to fibrosis and impaired function throughout the chronically inflamed diabetic kidney [59]. Fibrosis prevents mesangial cells from expanding and contracting to control capillary blood pressure and the surface area of the glomerular filter, with resulting impairment in filtration [60]. Tubulointerstitial fibrosis expands the space between the tubular basement membrane and the peritubular capillaries, leading to reduced blood flow, hypoxia, and further epithelial damage and inflammation, and eventual tubular atrophy [59].

Pro-fibrotic mediators released by activated macrophages and injured epithelial cells upregulate extracellular matrix production in various cell types [22, 61]. These include mesangial cells, which normally secrete extracellular matrix to provide structural support for the glomerular capillary tuft [60], and perivascular fibroblasts, which normally provide structural support to the kidney microvasculature [62]. Potential biomarkers of renal fibrosis include TGF-β and matrix metalloproteinase 2 (MMP-2) [63]. TGF-β produced by pro-fibrotic macrophages upregulates expression of extracellular matrix proteins in kidney endothelial and epithelial cells [59, 61]. Oxidative and mechanical stress also directly enhance extracellular matrix production and contribute to fibrosis [61].

Ischaemia is a key trigger of inflammation in DKD, especially in the highly metabolically active tubular epithelium. Impaired perfusion leads to hypoxic injury, triggering inflammation, fibrosis, tubular atrophy and progression of DKD [64]. Type 2 diabetes is characterized by a more ischaemic kidney phenotype than the oxidative and proliferative pattern seen with type 1 diabetes [65, 66]. Oxygen demand in the tubules leads to sustained activation of the intrarenal renin–angiotensin–aldosterone system, causing glomerular capillary hypertension that further damages glomerular endothelial and mesangial cells and podocytes [67]. Excessive oxygen levels in glomerular tissues lead to oxidative stress and formation of reactive oxygen species and advanced glycation end-products, with resulting inflammation and fibrosis [64]. Angiotensin II also directly activates pro-inflammatory and pro-fibrotic signaling pathways, contributing to endothelial cell injury and loss [68].

High glucose levels and advanced glycation end-products impair nitric oxide production by kidney endothelial cells, leading to impaired vasodilation in patients with DKD. Nitric oxide deficiency exacerbates oxidative stress, leading to further dysfunction and injury not only of epithelial cells but also adjacent cells, including podocytes [69]. Under oxidative stress, dimeric endothelial nitric oxide synthase decouples and produces superoxide instead of nitric oxide, which exacerbates both oxidative stress and nitric oxide deficiency [70]. Microvascular rarefaction in DKD eventually results from damage and apoptotic loss of endothelial cells together with impaired function of endothelial progenitor cells due to the impact of oxidative stress, advanced glycation end-products, and an inflammatory, pro-fibrotic milieu [69].

Most rodent models of DKD involve induction of diabetes-like phenotypes by streptozotocin treatment, spontaneous mutations, or genetic manipulation in laboratory mice (Mus musculus). Current mouse models have been successfully used with clinical standard of care in DKD and may address some inflammatory compartments, but study of ‘immune cell recruitment’ and ‘perfusion’ have remained challenging (Table 2) [71].

Streptozotocin is a cytotoxic glucose analogue that ablates pancreatic islet β cells, with severity of diabetes-like features depending on the mouse strain. In C57BL/6 J or Balb/c mice, streptozotocin induces only mild or moderate disease, but severity can be increased using genetic modification, crossing with other strains or a high-fat diet. Combing streptozotocin with hyperlipidaemia by knocking out Apoe (encoding apolipoprotein E) accelerates and worsens renal injury in C57BL/6 mice [8, 72]. Mutations in Akita and OVE26 mice cause pancreatic β cells toxicity, and the severity of the diabetes-like disease also varies depending on mouse strain. In db/db mice, a genetic defect in the leptin receptor leads to obesity, diabetes, and some signs of DKD. Surgical removal of one kidney (uninephrectomy) accelerates progression of kidney pathology in db/db mice [72]. Genetic knockout of Nos3 (encoding eNOS) or overactivation of the renin–angiotensin system in TTRhRen mice also accelerate loss of kidney function in mouse models (Table 2) [8, 72].

Zebrafish (Danio rerio) have been used to study ‘filtration’ and ‘immune cell recruitment’ in DKD. Zebrafish offer low cost, high throughput, and an advanced transgenic toolbox for molecular genetic manipulation, at the cost of potential low translatability to human disease. The kidney spontaneously regenerates in fish, providing challenges to test the effectiveness of potential therapeutics. Zebrafish have been used to study genetic variants that lead to podocyte damage and DKD [73, 74].

Models in domestic pigs (Sus scrofa) recapitulate features of DKD in all five compartments. A high-fat high-fructose diet induces renal hypertension, endothelial dysfunction and inflammation, and streptozotocin plus high-fat diet induces renal injury and proteinuria. Genetic manipulation to insert the Akita mutation into pigs leads to diabetes, but changes in kidney function have not yet been described [72].

Non-human primates (Macaca mulatta) are the gold standard for animal models of human kidney disease, but ethical considerations, high cost and the difficulty of genetic manipulation limit respective investigations. Streptozotocin administration leads to histopathological changes, proteinuria and impaired GFR, with the fastest disease progression observed using uncontrolled blood glucose levels and a high-fat, high-salt diet [72]. Still, aged dysmetabolic non-human primate models may offer the most suitable disease model, particularly as aging may be an important factor in DKD [75].

Organoids are three-dimensional tissue structures derived by in vitro differentiation of induced pluripotent or other stem cells [76]. Blood vessel organoids with capillary networks, develop thickening of the basement membrane after exposure to high glucose levels and inflammatory cytokines, potentially providing a model for investigation of microvascular aspects of DKD [77]. The latest kidney organoids comprise connected nephrons and collecting ducts, and research is ongoing into nephropathy and fibrosis for DKD target validation [76].

No anti-inflammatory drugs for the treatment of DKD have progressed beyond phase 2 clinical trials, except maybe for finerenone (Table 3). Finerenone, a non-steroidal selective mineralocorticoid receptor antagonist that induces natriuresis with reduced hyperkalaemia compared with steroidal antagonists (e.g. spironolactone), may retain some potentially beneficial anti-inflammatory and anti-fibrotic effects [38]. In two large phase 3 studies in patients with DKD (FIDELIO- and FIGARO-DKD), finerenone was significantly more effective than standard of care including ACEi and ARBs in slowing the decline in estimated glomerular filtration rate (eGFR) and improving cardiovascular outcomes, with non-significant reductions in end-stage kidney disease and all-cause mortality [38, 97].

A key challenge for the development of new anti-inflammatory medicines is the limited understanding of relevant surrogate endpoints for early clinical development before phase 3 [98] Most published trials have used the traditional markers of UACR and eGFR both as baseline patient selection criteria and as efficacy measures (Table 3). So far, the most reliable surrogate kidney endpoints seem to be a 30% improvement in albuminuria within 6 months, time to 30% decline in eGFR from baseline, and mean reduction in the slope of eGFR decline greater than 0.5–1.0 mL/min/1.73 m²/year over at least 2 years [99‐102]. A sample size of approximately 100 patients per arm provides 80% power to detect a 30% reduction in UACR at 6 months, making it an appealing surrogate endpoint for a phase 2 efficacy study [103]. However, short-term UACR parameters have a legacy from studies of anti-hypertensive medications which may not be appropriate for studies of anti-inflammatory drugs. Rates of progression in UACR vary substantially from patient to patient due to differences in underlying pathophysiology, as well as race, comorbidities and sex [104, 105]. The resulting inflammation-independent diversity presents a challenge to adequate statistical powering of clinical studies, even assuming that the investigational drug can improve short-term UACR in ‘all comers’ with DKD. Also, albuminuria-based endpoints do not differentiate between glomerular and tubular loss of protein (the ‘filtration’ and ‘resorption’ compartments of our conceptual model). In the future, the over-all prevalence of albuminuria in patients may, furthermore, decline substantially with increasing take-up of standard care involving anti-hypertensive medications and SGLT2 inhibitors [106‐108].

Lack of efficacy on renal functional outcomes like UACR or eGFR led to failure of several drug classes at phase 2, including those targeting pathways implicated in the pathophysiology of DKD (e.g. IL-1β antibodies and Janus kinase inhibitors). Unpredictable adverse drug reactions led to failure for bardoxolone (activator of the Nrf2 pathway and an inhibitor of the NF-κB pathway) and the vascular adhesion protein 1 inhibitor ASP8232, despite promising phase 2 kidney function outcomes (Table 3) [109, 110]. The generally disappointing results of trials of anti-inflammatory therapies to date hence indicate a need for improved, compartment-focused selection criteria and outcome measures.

Ziltivekimab, a fully human monoclonal antibody directed against the IL-6 ligand, was evaluated in a randomised, double-blind, phase 2 trial involving individuals (n = 264) with elevated high-sensitivity CRP and chronic kidney disease [111]. The primary study outcome was percentage change from baseline in high-sensitivity CRP after 12 weeks of treatment with ziltivekimab (7.5 mg, 15 mg and 30 mg) compared with placebo. Biomarker and safety data were collected over 24 weeks of treatment. After 12 weeks, median high-sensitivity CRP levels were reduced by 77% for the 7·5 mg group, 88% for the 15 mg group, and 92% for the 30 mg group compared with 4% for the placebo group. Dose-dependent reductions in fibrinogen, serum amyloid A, haptoglobin, secretory phospholipase A2, and lipoprotein(a) were observed. Ziltivekimab was well tolerated. Based on these data showing markedly reduced biomarkers of inflammation and thrombosis relevant to atherosclerosis a further trial is planned to investigate the effect of ziltivekimab in patients with chronic kidney disease, increased high-sensitivity CRP, and established cardiovascular disease.

Maximizing the potential benefits of new treatments involves identifying the compartments most affected by immunopathology in individual patients with DKD. The five-compartment model may serve as a guide for development of tools and therapies that will enable physicians to provide the right treatment to the right patients consistently and accurately, ideally without the need for kidney biopsy. Approaches that will allow patient classification include genomic and transcriptomic studies and identification of novel fluid-phase and imaging biomarkers. However, robust interventional trials are still needed to fully validate these exploratory endpoints.

Circulating biomarkers may allow patients to be identified based on molecular features of inflammation, and stratified based on predominant immunopathology [112]. For example, plasma levels of TNFR-1, TNFR-2 and kidney injury molecule 1 (KIM-1) are associated with decline in eGFR, even after adjustment for baseline albuminuria and eGFR, in multiple cohorts of patients with type 2 diabetes [5, 35, 87, 89, 90, 113]. A proteomics study recently identified a ‘kidney risk inflammatory signature’ comprising a cluster of 17 circulating inflammatory biomarkers that strongly associate with development of end-stage renal disease in multiple ethnicities [34]. Although the cluster included pro-inflammatory mediators already implicated in DKD (e.g. CCL2), it also included chemokines and ILs that were not known to be associated with the disease [34]. This suggests that relevant immunopathological pathways remain to be elucidated.

Inflammatory responses stimulated by toll-like receptors (TLRs), notably TLR4, appear to play a decisive role in the progression of DKD [114, 115] while complement dysregulation, may also contribute to progression [116]. Furthermore, the renal NF-κB pathway, implicated in the development and progression of experimental DKD, may also become an important therapeutic target [117].

Genomic and transcriptomic studies also offer routes to discovery of novel markers. Academic-industry systems biology consortiums aim to share molecular target identification efforts and expertise to accelerate novel drug development (e.g. the Renal Pre-Competitive Consortium [RPC²]) [118]. In a recent RNA-seq study, micro-dissected glomerular and tubulointerstitial kidney biopsy tissue fractions were analyzed from patients with DKD and matched living kidney donors. The results confirmed inflammatory responses, complement activation and extracellular matrix deposition as key pathophysiological processes [119]. A whole-exome sequencing study, in 3315 patients with chronic kidney disease and 9536 controls, used in vivo and in vitro approaches to validate the most strongly associated genes as potential novel diagnostic or therapeutic targets [120]. The analysis identified 93 genes with a strong CKD correlation which after ranking based on literature data supporting a link to CKD relevant biology resulted in 31 genes that were further evaluated in vitro and in vivo. Ultimately a single gene was identified as a CKD target that entered the pipeline for drug discovery.

Magnetic resonance imaging (MRI) can detect changes that precede albuminuria and GFR decline in patients with DKD [121]. Ischaemic regions of the kidney can be identified using blood oxygenation level-dependent MRI, and these signals are predictive of chronic kidney disease progression [122, 123]. Renal fibrosis identified using diffusion-weighted MRI may detect DKD progression earlier than eGFR [124]. Early impairment in renal perfusion in patients with diabetes can be identified using arterial spin labelling MRI, and these signals correlate with reduced GFR [125].

Combining these rapid advances in histology, genetics, ‘omics’, and imaging may unlock the potential of anti-inflammatory therapies in DKD. Eventually, patient stratification by specific and relevant pathophysiology, integrated with pre-clinical models of these disease processes, may allow intervention with novel targeted therapies in the right patients at the right time.

Nevertheless, limitations of the five-compartment model include that not all patients with histopathological indicators of DKD will ultimately develop the condition [126].