The increasing global prevalence of type 2 diabetes mellitus (T2DM) and chronic kidney disease (CKD) has prompted research efforts to tackle the growing epidemic of diabetic kidney disease (DKD; also known as diabetic nephropathy). The limited success of much of this research might in part be due to the fact that not all patients diagnosed with DKD have renal dysfunction as a consequence of their diabetes mellitus. Patients who present with CKD and diabetes mellitus (type 1 or type 2) can have true DKD (wherein CKD is a direct consequence of their diabetes status), nondiabetic kidney disease (NDKD) coincident with diabetes mellitus, or a combination of both DKD and NDKD. Preclinical studies using models that more accurately mimic these three entities might improve the ability of animal models to predict clinical trial outcomes. Moreover, improved insights into the pathomechanisms that are shared by these entities — including sodium–glucose cotransporter 2 (SGLT2) and renin–angiotensin system-driven glomerular hyperfiltration and tubular hyper-reabsorption — as well as those that are unique to individual entities might lead to the identification of new treatment targets. Acknowledging that the clinical entity of CKD plus diabetes mellitus encompasses NDKD as well as DKD could help solve some of the urgent unmet medical needs of patients affected by these conditions.
Cardiovascular mortality and progression to end-stage renal disease are the two major unmet medical needs in patients with chronic kidney disease (CKD) plus diabetes mellitus.
In patients with diabetes mellitus, primary prevention of kidney disease, regardless of aetiology (diabetic kidney disease (DKD; also known as diabetic nephropathy)) or nondiabetic kidney disease (NDKD)), is essential and includes appropriate control of glucose, blood pressure, and body weight and avoidance of nephrotoxic drugs.
The clinical entity CKD plus diabetes encompasses DKD and NDKD or a combination of the two; diagnosis of these entities by kidney biopsy is important for disease management and research.
Biopsy studies show that NDKD is common in patients with type 2 diabetes mellitus (T2DM); as most patients with T2DM entering clinical diabetes trials do not undergo kidney biopsy, the pathophysiology underlying their kidney disease remains uncertain.
A disconnect exists between animal models used in preclinical studies of DKD and clinical studies with regard to differences in age, obesity status, renal function at onset, and use of co-medications; this disconnect might contribute to the poor predictability of animal studies for clinical trial outcomes.
Findings from clinical trials suggest that hyperfiltration driven by the sodium–glucose cotransporter 2 and the renin–angiotensin system is a common upstream mechanism that drives kidney disease in both DKD and NDKD in the context of diabetes mellitus.
The global prevalence of diabetes mellitus (DM) has increased enormously over the past few decades, mostly driven by an increase in the prevalence of type 2 diabetes mellitus (T2DM) secondary to obesity and the metabolic syndrome1. Tight glucose control can prevent the development of microvascular complications in patients with T2DM2, but achieving glucose targets in the physiological range (4–6 mmol/l when fasting and <7.8 mmol/l within 2 hours postprandially) is often not only difficult but also no longer recommended for patients with T2DM, as it can negatively affect mortality3,4. Therefore, the global prevalence of microvascular and macrovascular complications associated with DM is increasing dramatically1,5. Microvascular changes within the kidney often lead to chronic kidney disease (CKD), an entity referred to as diabetic kidney disease (DKD) or diabetic nephropathy6. This disease is characterized by a distinct histopathological pattern of glomerular basement membrane (GBM) thickening, mesangial matrix expansion, nodular glomerulosclerosis, and arteriolar hyalinosis (Fig. 1). This histopathological pattern is frequently observed in young and lean patients with type 1 DM (T1DM), but biopsy samples from patients with T2DM often indicate the presence of other pathogenic factors, such as primary glomerulopathies, ageing-related nephropathy, or previous episodes of acute kidney injury (AKI). Thus, the term DKD lacks precision in describing the most prevalent form of kidney disease in patients with DM.
The above observations are in line with the understanding that increased longevity, changes in diet and lifestyle, and an increased incidence in episodes of AKI (ischaemic and toxic) have increased the global prevalence of CKD independent of DM7,8,9,10 and imply that the high prevalence of CKD in adults with DM is in part attributable to the presence of nondiabetic kidney disease (NDKD), including glomerulonephritis, minimal change disease, primary and secondary forms of focal segmental glomerulosclerosis, and paraprotein-related kidney injury (Fig. 1), which may not be suspected on the basis of clinical signs or urine abnormalities11,12. Further complicating matters, the diabetic environment can alter the course of NDKD. Thus, coincident CKD in a patient with DM can be true DKD (as a result of the DM), NDKD, or a combination of NDKD and DKD: these three entities can be reliably distinguished only by kidney biopsy11,13. In clinical practice, however, diagnostic kidney biopsy is rarely performed in patients with DM11,13. Moreover, in the absence of a diagnostic biopsy, registries often assign a diagnosis of DKD to patients with DM plus CKD — an action that inflates the true prevalence of DKD7,9. Consequently, DKD has been identified as an unmet medical need of high priority, leading to the allocation of enormous resources for research and drug development programmes14,15,16. However, the likelihood that some patients with suspected DKD in fact have NDKD or NDKD plus DKD (Fig. 2a) has not been appropriately acknowledged in preclinical or clinical studies despite the need for these disease entities to be distinguished in order to improve understanding of disease pathomechanisms and develop appropriate preclinical models for drug development.
In this Review, we outline the unmet medical needs of patients with T2DM plus CKD and highlight recent advances in the therapeutic arena. We discuss why NDKD needs to be considered a separate entity from DKD and describe the shared and distinct pathomechanisms by which the diabetic environment and NDKD contribute to nephropathy in patients with DM. Finally, we discuss how improved understanding of underlying disease processes has the potential to improve the translational yield of preclinical studies in this field.
CKD plus diabetes: unmet medical needs
To improve the quality of life and outcomes of patients with CKD plus DM, translational research must focus on the unmet medical needs of these patients. In the context of CKD plus DM, the most prominent unmet medical needs are mortality, in particular cardiovascular mortality, and progression to end-stage renal disease (ESRD)17.
The leading cause of mortality in patients with CKD plus DM is cardiovascular death, caused by events such as stroke, sudden cardiac death, myocardial infarction, and other fatal complications of diabetic cardiomyopathy17. Numerous studies have documented the association of accelerated vascular, valvular, and cardiac ageing — characterized by soft tissue calcifications, vascular wall degeneration, and myocardial fibrosis — with mortality in patients with DM plus CKD. In fact, the majority of patients with CKD plus DM die from a cardiovascular event before reaching ESRD17. Thus, addressing the high risk of cardiovascular disease (CVD) in these patients should be an area of priority18. Until very recently, metformin was the only agent known to reduce all-cause mortality in patients with T2DM, with the 1998 UK Prospective Diabetes Study (UKPDS) demonstrating that metformin could reduce the risk of mortality of overweight patients with T2DM by 36% compared with the mortality risk of conventional therapy19. Of note, findings from the RENAAL trial showed that inhibitors of the renin–angiotensin system (RAS) do not reduce all-cause mortality despite their ability to slow progression of microvascular complications20. In the past few years, however, randomized controlled trials (RCTs) have demonstrated beneficial effects of two new classes of antidiabetic agents — sodium–glucose cotransporter 2 (SGLT2) inhibitors and glucagon-like peptide 1 (GLP1) agonists — on cardiovascular mortality in patients with T2DM. In the EMPA-REG OUTCOME trial and CANVAS Program, the SGLT2 inhibitors empagliflozin and canagliflozin reduced cardiovascular mortality by 38%21 and 14%, respectively22. In the LEADER trial, the GLP1 analogue liraglutide reduced cardiovascular mortality by 22%23. Despite some improvements, cardiovascular morbidity and mortality remain a key concern for patients with T2DM, in particular for those with CKD regardless of the underlying cause24. Thus, the identification of approaches to delay CKD progression in patients with DM will not only delay progression to ESRD but also will reduce the risk of CVD18.
As mentioned above, CKD progression is closely linked with cardiovascular risk; hence, targeting CKD progression is essential for reducing cardiovascular mortality as well as for preventing the development of ESRD in these patients18. Despite some progress in reducing cardiovascular complications in patients with DM (for example, through intensive glycaemic control, lifestyle modifications, and the use of RAS inhibitors), the rate of CKD progression to ESRD in patients with DM has remained unchanged over the past 2 decades, with around 25 of 10,000 adults with DM developing ESRD per year25. The numerous complications, comorbidities, and use of polypharmacy and renal replacement therapy associated with advanced CKD and ESRD impose an enormous economic burden on health-care systems. Thus, approaches to slow the progression of CKD in patients with DM have the potential to vastly reduce the morbidity and cost associated with renal dysfunction. For the past 25 years, RAS inhibitors were the only approach to attenuate CKD progression in patients with DM. Although RAS inhibitors elicit renoprotective effects in animal models of DKD and in patients with T1DM and T2DM20,26,27, additional therapeutic options to attenuate CKD progression in DM have until now been limited25.
Similar to their beneficial effects on cardiovascular mortality, findings from the recent EMPA-REG OUTCOME and CANVAS Program studies have now shown that the SGLT2 inhibitors empagliflozin and canagliflozin have renoprotective effects, preventing the decline in glomerular filtration rate (GFR) by 44% and 40%, respectively, in patients with T2DM22,28. Along these lines, the LEADER trial demonstrated that liraglutide administration was associated with a 43% lower risk of developing persistent macroalbuminuria and other parameters indicative of CKD progression in patients with T2DM29. The beneficial effects of SGLT2 inhibitors and liraglutide on cardiovascular and renal outcomes indicate that these agents target pathways that are common to both outcomes in patients with T2DM30. Importantly, these three studies are the first in over a decade to successfully demonstrate renoprotective effects in the context of DM20,31,32.
The failure of earlier RCTs to demonstrate beneficial effects of various therapeutics on renal outcomes might be due to a number of reasons, including the fact that patients with DKD represent a heterogeneous population in whom renal disease is driven by the diabetic environment (DKD), by factors unrelated to DM (NDKD), or by a combination of both33,34,35. Improved understanding of the mechanisms underlying these entities is likely to aid in the identification of further treatment targets and lead to further successes in the clinic.
Mechanisms of DKD
As mentioned above, improved understanding of the shared and distinct mechanisms driving DKD and NDKD in patients with DM is likely to improve outcomes for affected patients. Below, we dissect the mechanisms that drive CKD progression in these patients with the aim of providing a conceptual framework from which to identify suitable drug targets.
DKD is a major microvascular complication of DM and is responsible for up to 50% of all cases of ESRD in Western populations7,9. As mentioned above, DKD is an important risk factor for cardiovascular mortality in patients, particularly when accompanied by hypertension. The earliest detectable clinical manifestation of DKD is microalbuminuria, and in the absence of early intervention, approximately 50% of patients with established microalbuminuria will progress to macroalbuminuria, which is associated with a tenfold higher risk of progression to ESRD than that of patients with normoalbuminuria36,37. Of note, risk factors for the development of microalbuminuria in patients with T2DM include classic cardiovascular risk factors, such as a high urinary albumin:creatinine ratio, older age, high haemoglobin A1C (HbA1c), elevated blood glucose, and hypertension, but not haemodynamic factors such as estimated GFR (eGFR)38.
The incidence of microalbuminuria in patients with T1DM is around 30% and largely depends on medication compliance and blood glucose control39. The incidence of microalbuminuria in patients with T2DM is also around 30% but is usually associated with hypertension40,41. In both cases, albuminuria does not occur in the absence of hyperglycaemia and glucose control is the main determinant of progression to overt DKD2,42. The importance of hyperglycaemia in the development of DKD is illustrated by the fact that improving HbA1c (for example, from 9.3% to 8.7%) can prevent progression to ESRD in patients with T1DM and proteinuria43. Moreover, pancreas transplantation in patients with T1DM can completely reverse diabetic glomerulosclerosis in native kidneys following 10 years of normoglycaemia44.
Beyond glycaemic control, risk factors for DKD include morbid obesity, low birthweight, and genetic susceptibility factors, which may explain why some but not all patients with DM develop DKD45,46. Numerous pathomechanisms, described in detail below, contribute to the development of DKD; however, as mentioned earlier, not all patients with CKD and DM necessarily have DKD. To dissect differences in the pathological mechanisms of DKD and NDKD, it is important to first understand the early versus late effects of hyperglycaemia on the kidney.
Immediate effects of hyperglycaemia
Hyperglycaemia leads to an immediate increase in the amount of glucose filtered through the glomerular filtration barrier, inducing hyper-reabsorption of glucose in the proximal tubule47 (Fig. 2b). Hyper-reabsorption of glucose involves induction of the expression of glucose transporters and massive increases in energy-consuming transport processes in the proximal tubular cells47; this drastically increases oxygen demand in the renal cortex and the outer medulla, which induces relative ischaemia and increases the expression of cellular stress markers, such as neutrophil gelatinase-associated lipocalin (NGAL) and kidney injury molecule 1 (KIM1; also known as HAVCR1)47,48. The increased workload of the proximal tubule leads to proximal tubule hypertrophy and elongation, contributing to kidney hypertrophy42,47 (Fig. 3). As SGLT2 cotransports sodium, the recovery of sodium in the proximal tubule is massively increased, reducing the sodium chloride concentration in the distal tubule and at the macula densa47. The low sodium chloride concentration at the macula densa deactivates the tubuloglomerular feedback mechanism and induces dilation of the renal afferent arteriole30,47. Concomitant secretion of renin promotes vasoconstriction of the efferent arteriole. The consequences of these haemodynamic effects are a persistent increase in single-nephron GFR (SNGFR), glomerular hyperfiltration, and glomerular hypertension. Glomerular pressure declines following the development of glomerular hypertrophy, but glomerular hyperfiltration persists30,47 (Fig. 3). The involvement of SGLT2 and renin in these processes might explain why SGLT2 and RAS inhibition exert profound effects on renal haemodynamics in patients with DKD within a fairly short period of time22,28 (Figs. 2,3).
Late effects of hyperglycaemia
In contrast to the beneficial effects of targeting the upstream haemodynamic pathways discussed above, targeting the downstream and delayed effects of hyperglycaemia, such as inflammation and endothelial dysfunction, might have different effect sizes, particularly in heterogeneous groups of patients with DKD versus NDKD plus DM. Nevertheless, an understanding of these downstream effects is required to aid the development of treatment strategies for patients with late-stage disease, for example. Some of the main mechanisms that drive the late effects of hyperglycaemia on the kidney are discussed below.
Systemic endothelial cell dysfunction — characterized by increased vascular permeability, inflammation, and dysregulation of vascular tone — is a pathogenic mechanism involved in several microvascular and macrovascular complications of DM and is linked to cardiovascular mortality49,50. In the kidney, dysfunction of the glomerular endothelium, resulting in microvascular permeability, presents as microalbuminuria. Hence, microalbuminuria is an indicator of systemic endothelial dysfunction, in line with its recognized role as a risk factor for cardiovascular mortality49,50. The glomerular endothelium forms part of the glomerular filtration barrier and is a highly specialized, fenestrated layer that is fully covered by a negatively charged glycocalyx49, which is connected to the endothelium by several proteoglycans and glycoproteins that form a network with soluble molecules, such as versican and perlecan49. The removal of certain glycocalyx components, including heparan sulfate, chondroitin sulfate, dermatan sulfate, keratan sulfate, or hyaluronic acid51, increases permeability of the endothelium, highlighting the importance of endothelium–glycocalyx interactions in barrier function52. Hyperglycaemia damages the endothelial glycocalyx and promotes microalbuminuria53 probably through the generation of reactive oxygen species (ROS) that interfere with the production of heparan sulfate54. Hyperglycaemia-induced downregulation of thrombomodulin and activated protein C further contributes to endothelial dysfunction in models of DM55,56.
Glomerular basement membrane changes and mesangial cell expansion
Thickening of the GBM is an early histopathological finding in DKD and is caused by the abnormal turnover and modification of extracellular matrix produced by endothelial cells and podocytes57 (Figs. 3,4). Altered GBM remnants contribute to the expanding mesangial matrix, but hyperglycaemia also stimulates mesangial cells to proliferate and produce matrix58 through activation of transforming growth factor-β (TGFβ) and vascular endothelial growth factor (VEGF), which directly induce the transcriptional activation of matrix collagens59. The turnover of collagens in the hyperglycaemic environment is also affected by alterations in the expression and function of matrix metalloproteinases60. GBM thickening can be interpreted as an early sign of endothelial and podocyte activation, but a functional role for this structural change has not been documented. Similarly, the association between mesangial cell expansion and DKD is well documented, but a functional role for this process has not been demonstrated. Importantly, both lesions can be reversed by combined kidney and pancreas transplantation61.
DM can also induce a podocytopathy, characterized by cellular hypertrophy, foot process effacement, and podocyte loss62 (Fig. 5). The reduction in podocyte number in patients with T2DM correlates with albuminuria and GFR decline62. In rodent models of T1DM and T2DM, we detected an initial loss of podocytes of up to 10% followed by stabilization of podocyte numbers, confirming our in vitro observation that podocytes adapt to high glucose levels63,64. High glucose levels change the composition of the intracellular scaffold complex of the podocyte. The mechanisms underlying podocyte loss include detachment due to loss of α3β1 integrin expression or to direct glucose or TGFβ-induced apoptosis64,65,66. Immunofluorescence studies using Wilms tumour protein (WT1) and synaptopodin as podocyte markers showed that SGLT2 inhibition ameliorated podocyte loss and damage in diabetic db/db mice67. Interestingly, however, the reduced expression of nephrin, characteristic of murine models of DKD, was not influenced by SGLT2 inhibition, indicating that the beneficial effects of empagliflozin are not due to direct effects on the cells but are most likely secondary to reduced hyperfiltration and shear stress47,67. Nephrin is an important podocyte protein as it is not only the cornerstone of an important intracellular signalling hub but also directly involved in podocyte survival pathways and actin dynamics68. Its expression is reduced in murine models of DKD69, in part owing to glucose-induced endocytosis of nephrin in a process orchestrated by protein kinase C α-type (PKCα) and β-arrestin70,71. PKCα might have dual functions in this context as deletion of PKCα ameliorates the TGFβ-induced pro-apoptotic signalling in podocytes71. Available evidence suggests that high glucose levels change the post-translational modifications of nephrin and other slit diaphragm components by downregulating the adaptor protein CD2-associated protein (CD2AP) and upregulating the adaptor protein CIN85 (also known as CD2BP3), inducing a ubiquitylation response that leads to an increase in nephrin endocytosis72. In addition to affecting the structural integrity of the podocyte cytoskeleton, the reduction of nephrin expression by DM has direct consequences for insulin signalling because the cytoplasmic domain of nephrin enables insulin recognition and intracellular signalling in podocytes73,74. Mice with podocyte-specific deletion of the insulin receptor develop features of DKD75, highlighting the relevance of insulin signalling to podocyte function. Podocyte loss is compensated for by podocyte hypertrophy, but such compensation is possible only if the cell number remains above a certain threshold76. Beyond induced cell stress and death, podocyte loss under diabetic conditions can occur through mechanical force-induced detachment from the GBM. Glomerular hyperfiltration mediated by SGLT2 and the RAS induces an increase in the hydraulic pressure gradient across the filtration barrier47. Concurrently, the elastic counterforces of the GBM are decreased owing to the molecular changes in the GBM described above. Possible cellular counterforces to filtration flow provided by the podocyte might be impaired with reduced podocyte numbers and an expanded GBM surface45. Progressive podocyte loss (for example, as a consequence of mitotic catastrophe77 as a result of the molecular and mechanical events described above77) manifests clinically as macroproteinuria and promotes diabetic glomerulosclerosis, with subsequent, irreversible nephron loss (Figs. 3,4). Of note, podocyte loss is a feature of many glomerular diseases and is not specific to DKD.
Processes associated with sterile inflammation initiate fairly early following the onset of DKD but contribute to the progressive deterioration of renal function. In particular, the mechanisms and consequences of monocyte and macrophage infiltration within the kidney are areas of great interest78,79. In a 2017 phase II RCT, a Spiegelmer that inhibits CC-chemokine ligand 2 (CCL2; also known as MCP1) reduced proteinuria in patients with T2DM and CKD80. Similar results were obtained with an inhibitor of the respective chemokine receptor, CC-chemokine receptor 2 (CCR2)81, highlighting the importance of chemokine signalling in DKD pathogenesis. In rodents, the extent of inflammatory cell accumulation in the diabetic kidney correlates with the decline in renal function; inhibition of inflammatory cell recruitment into the kidney protects rodents from experimental DKD82,83. Several researchers have proposed that inflammasome-mediated release of IL-1 contributes to the development of DKD in rodents and patients84,85,86. The inflammasome is an intracellular multiprotein complex that senses various signals, including mitochondrial ROS, advanced glycation end products (AGEs), high mobility group protein B1 (HMGB1), and other danger-associated molecular patterns (DAMPs). These stimuli typically activate the NACHT, LRR and PYD domains-containing protein 3 (NLRP3) inflammasome87,88, which then recruits and assembles with apoptosis-associated speck-like protein containing a CARD (ASC) and caspase 1, resulting in caspase 1 activation, maturation of IL-1β and IL-18, and promotion of immune cell recruitment87. Initially thought to be restricted to immune cells, the inflammasome is now understood to function in non-immune cells, including podocytes84,89. Genetic or pharmacological inhibition of inflammasome activity improves markers of DKD in rodent models of T1DM and T2DM90. Of note, a recombinant IL-1β receptor antagonist not only prevented but also reversed markers of DKD in diabetic db/db mice84. Finally, the IL-1β antagonist canakinumab showed beneficial effects on CVD risk in a population at high cardiovascular risk, of whom around 40% had DM91.
Studies on the long-term changes caused by hyperglycaemia have shown that high blood glucose can induce complications such as retinopathy and macrovascular diseases long after normoglycaemia has been re-established2,4. This phenomenon is termed ‘metabolic memory’ (REF.92) and is thought to involve the consolidation of metabolic cues and transcriptional changes through epigenetic programming93,94. Other factors, such as diet, also affect the epigenome95,96, and pathways that integrate signals for cellular nutrient uptake and intracellular metabolism, such as those regulated by mechanistic target of rapamycin (mTOR), might mediate the establishment of metabolic memory97,98. Interestingly, the mTOR pathway is an important driver of podocyte hypertrophy, de-differentiation, and loss in progressive DKD99,100,101, prompting the suggestion that permanent mTOR activation by hyperglycaemia through transcriptional or epigenetic mechanisms leads to profound metabolic reprogramming of kidney epithelial cells. By contrast, pyruvate kinase M2 activity seems to protect against metabolic flux, podocyte demise, and progression of DKD in a process that can be specifically targeted for therapy102. Thus, various signalling pathways are activated by metabolic cues in the context of hyperglycaemia, and cell-specific targeting of glycolytic or mitochondrial enzymes might represent a promising therapeutic intervention for DKD.
Diabetes-related modifiers of DKD
DM is often associated with other disease states that can trigger or drive CKD independently of hyperglycaemia and hyperinsulinaemia. The presence of these entities in patients with DM can exacerbate vascular complications such as DKD. For example, obesity, which often accompanies T2DM, is associated with an increased filtration load and levels of circulating adipokines that can increase SNGFR and induce nephron hypertrophy103. These effects are particularly evident in patients with morbid obesity or comorbidities such as T2DM. Indeed, obese adolescents and young adults with T2DM develop a considerably higher number of diabetes complications, including DKD, than those with T1DM5. Pregnancy in patients with uncontrolled DM can also induce renal dysfunction owing to the combined effects of hyperglycaemia and pregnancy-induced hyperfiltration104. Patients with T2DM often have arterial hypertension. Indeed, many diagnoses of DKD might actually relate to the combined effects of high blood pressure on top of the diabetes-induced loss of renal autoregulation, causing damage to the glomerular filtration barrier. Finally, kidney injuries related to urinary tract infections or drug toxicity are more common among patients with DM than among the healthy population105 owing to the immunosuppressive effects of hyperglycaemia and the high medication burden of this population.
NDKD in DM
In addition to the aforementioned causes and modifiers of CKD related to the diabetic milieu, diabetes-independent factors can cause CKD, even in patients with DM. Genetic kidney disorders, immunoglobulin A (IgA) nephropathy, infection-related glomerulopathies, secondary focal segmental glomerulosclerosis and minimal change disease, cholesterol embolism, and all types of AKI can cause precedent, concomitant, or subsequent kidney injury in patients with DM10,12,106 (Fig. 1).
CKD and DM are major non-communicable diseases of global importance. They can manifest in patients causally or co-incidentally; importantly, not all cases of CKD in patients with DM are DKD. The overall prevalence of CKD stage G1–G5 of any cause is 7–12% in different regions, with the highest prevalences reported for Latin America, Europe, the Middle East, and East Asia10. By contrast, the global prevalence of DM is tightly linked to that of obesity and varies from <6% to >14% (corresponding to obesity rates of <10% to >30%8), with the highest prevalences in North America, the Middle East, Indonesia, and New Zealand1. Of note, most of the increase in DM prevalence over the past few decades is due to an increase in the prevalence of T2DM; the prevalence of T1DM has increased but not by much107. According to the US Renal Data System, 14.8% of the US population in the years 2011–2014 had CKD G1–G5, of whom ~40% had DM9. However, as mentioned earlier, CKD can occur co-incidentally to DM, and a diagnostic renal biopsy is required to determine the aetiology of CKD in patients with DM.
Reports from centres that endorse kidney biopsy in patients with T2DM indicate that the prevalence of NDKD ranges from 33% to 72.5% among patients with T2DM (Table 1). Glomerulonephritis and focal segmental glomerulosclerosis are the most common forms of NDKD in patients with T2DM in Asia and the United States, respectively12,108. These studies likely overestimate the true prevalence of NDKD in patients with DM, as indications for biopsy are biased towards atypical cases (for example, those with nephrotic syndrome, abnormalities of the urinary sediment, or other indicators of NDKD)11. On the other hand, risk factors for CKD that are independent of hyperglycaemia, such as low birthweight, nephron loss due to an AKI episode, or ageing nephropathy, may not be evident even on kidney biopsy; thus, despite its benefits, kidney biopsy is not a guaranteed approach for determining disease aetiology.
As mentioned above, renal biopsy is the only way to determine whether CKD in a patient with DM is in fact a direct consequence of the diabetic environment. Although we encourage the increased adoption of diagnostic renal biopsies for patients with suspected DKD, we realize this approach might not always be possible. In these situations, consideration of clinical factors, such as the timing of CKD onset, can aid diagnosis.
CKD before the onset of diabetes
The most obvious indication that CKD in a patient with DM is NDKD is the presence of kidney dysfunction before the onset of DM. This scenario is most likely for patients with, for example, early-onset genetic forms of CKD or CKD following an episode of AKI. In these patients, CKD can manifest early in life but can be exacerbated by the development of comorbidities such as obesity and/or T2DM later in life (Fig. 3). Of note, the presence of autoimmune and renal disease can increase susceptibility to T2DM, consequently exacerbating pre-existing CKD. For example, long-term steroid use can induce T2DM in patients with autoimmune diseases such as lupus nephritis and renal vasculitis and in kidney transplant recipients109.
CKD present at the time of diabetes diagnosis
T2DM and CKD are often diagnosed together, but the time at which they developed is unknown. As DKD typically develops many years after the onset of DM110, a diagnosis of CKD G3–G5 at the time of DM diagnosis most likely indicates the presence of NDKD (Fig. 6).
Undiagnosed CKD at the time of diabetes diagnosis
Current approaches to the diagnosis of CKD, including estimates of GFR, urine analysis, and imaging studies37, can be unspecific, and concern exists that certain forms of CKD remain unrecognized because of their lack of specific symptoms, apparently normal serum creatinine levels and, hence, apparently normal eGFR, and their absence of urine abnormalities. Of note, serum creatinine-based estimates of GFR are inaccurate for the diagnosis of CKD stage G1 or G2 (REF.111). CKD due to polycystic kidney disease, kidney hypoplasia, previous AKI episodes, or ageing nephropathy is not usually associated with urinary abnormalities on presentation, and the presence of a normal serum creatinine level can mask a clinically relevant stage of CKD112. The sudden increase in SNGFR and filtration pressure induced by the onset of DM can unmask such ‘silent’ CKD. Therefore, such forms of precedent CKD are diagnosed only years after the onset of DM — a series of events that increase the likelihood of misdiagnosing NDKD as DKD (Fig. 6).
Historically, nephrology counselling and kidney biopsy were considered to be dispensable for patients presenting with DM, especially for patients with poorly controlled T1DM given the seemingly obvious diagnosis of DKD3 — a concept that persists in current guidelines3. Although protocol biopsies performed in homogeneous cohorts such as 30–50-year-old Pima Indians with T2DM demonstrate a rather consistent DKD phenotype113,114, the presence of CKD in patients with T2DM despite the availability of improved options for glucose control and the drastic increase in the prevalence of T2DM in older age groups implies that NDKD may be more common than anticipated, especially in more heterogeneous populations. Nevertheless, kidney biopsy is still infrequently performed in patients with DM, especially when proliferative diabetic retinopathy is present, indicating the presence of diabetic microvascular changes. However, as noted earlier, reports from biopsy studies indicate that the prevalence of NDKD among patients with DM is 33–72.5% and have provided insights into the most common types of NDKD.
The presence of CKD G2–G5 in DKD and NDKD implies the loss of nephrons, the result of which is increased filtration load to the remnant nephrons, which manifests as increased SNGFR115,116 (Figs. 3,4). The associated supraphysiological filtration pressure triggers a compensatory increase in filtration surface via nephron hypertrophy, which lowers filtration pressure and accommodates the higher SNGFR45. The changes in renal haemodynamics, SNGFR, and compensatory growth that occur in the remnant kidney after kidney donation are a good example of what happens to remnant nephrons in states of low nephron endowment or following injury or ageing-related nephron loss117. The reserve provided by compensatory hypertrophy and hyperfiltration of the remnant nephrons maintains total GFR for a period of time, but continued nephron loss leads to GFR decline, which eventually manifests as CKD45,116 (Figs. 3,4). At the point at which CKD is clinically evident, only 20–40% of the normal nephron number might be present. This finding implies that even an apparently normal GFR and normal kidney size can mask considerable nephron loss and compensatory increases in SNGFR115.
As described earlier, the subsequent onset of DM in this scenario of decreased renal reserve immediately activates SGLT2 and RAS-driven glomerular hyperfiltration47,116. In the case of precedent CKD, this additional hyperfiltration cannot be accommodated as easily as in otherwise healthy kidneys42. The result is intraglomerular hypertension and accelerated progression to macroproteinuria, eventually leading to podocyte loss, glomerulosclerosis, and rapid CKD progression30,116.
In addition to facilitating diagnosis, kidney biopsy can also help predict CKD progression in patients with DM. A multicentre, prospective observational cohort study that followed 2,484 patients with DM and either diabetic nephropathy or NDKD in Northeast Japan found that 281 patients (11.3%) developed ESRD within a mean of 4.44 years follow-up. For patients with CKD stage G3 at baseline, the hazard ratios for development of ESRD were 7.1 (95% CI 2.46–20.49) and 0.89 (95% CI 0.19–4.24) for DKD and NDKD, respectively118, highlighting the deleterious effect of DKD on CKD progression. A study from Korea supported this finding119. Thus, kidney-biopsy-based stratification of DKD and NDKD might enable the identification of patients at low or high risk of CKD progression. Of note, regional differences exist in the prevalence of different types of NDKD. For example, chronic IgA nephropathy, which is usually a slowly progressive disease, is much more prevalent in East Asia than in other parts of the world and might have contributed to the slower rate of CKD progression among NDKD patients in the Japanese study. Therefore, biopsy studies with long-term follow-up are needed from other countries. Also of note is the fact that kidney biopsy might not always clarify whether a patient has a fast-progressing NDKD that builds on precedent DKD or whether DKD accelerates the progression of a precedent but otherwise rather inactive NDKD (Fig. 6).
Hierarchy of pathomechanisms
As described above, available evidence suggests that hyperglycaemia is the central upstream driver of DM-related CKD irrespective of whether DM is the sole cause (in the case of DKD) or a contributing factor (in the case of patients with NDKD; Fig. 6a). Thus, tight glucose control will likely improve renal outcomes for patients with DM, regardless of disease aetiology2. Findings from the EMPA-REG OUTCOME and CANVAS-R trials show that dual SGLT2 and RAS blockade substantially slow CKD progression and development of ESRD in patients with CKD plus DM, despite the fact that the study participants likely comprised patients with DKD and those with NDKD22,28. Given that SGLT2 and RAS inhibitors reduce glomerular hyperfiltration and hyper-reabsorption of glucose and sodium in the proximal tubule, it seems that SGLT2 transport and the RAS have fundamental upstream roles in promoting CKD progression in any patient with DM30,47 (Fig. 6a). Podocyte loss, the development of secondary glomerulosclerosis, renal ischaemia, and fibrosis are downstream pathomechanisms that are also shared by all entities. Whether AGEs, pyruvate kinase M2 activity, or other DM-specific signalling abnormalities qualify as therapeutic targets in patient populations with an unknown contribution of DKD and NDKD remains unclear and might depend on the temporal association of DM and CKD as well as the age of disease onset (Fig. 6b).
Implications for translational research
As mentioned earlier, findings from studies in animal models of DKD have so far poorly predicted outcomes of RCTs, probably because of the existence of wide gaps in the design of experimental and clinical trials14,120. Of note, the majority of clinical trials initiated as part of the drug approval process have been performed in obese patients aged >60 years with T2DM plus CKD G2–G4 and probably include a considerable percentage of individuals with NDKD as a single cause or as a major contributor to CKD. By contrast, animal studies of DKD are usually performed in juvenile lean mice that are housed in artificially clean conditions and do not have precedent kidney injury — a scenario that does not match the disease complexity of the target patient population. Key differences between animal models and human disease were highlighted in a cross-species comparison of glomerular transcriptional networks in patients with T2DM and three mouse models, which identified pathways unique to each of the human–mouse networks as well as shared pathways121.
Histomorphological studies, which are often extensively performed in experimental studies, are typically not performed in clinical trials, whereas the essential end points of RCTs, such as GFR decline or cardiovascular outcomes, are rarely addressed in preclinical studies14,120. These differences between preclinical and clinical studies represent a major disconnect122,123 that needs to be addressed in order to select the most promising drug candidates and improve the predictability of human trial outcomes122,123 (Box 1). Identification of transcriptional networks that are shared across species should be particularly useful in choosing suitable animal models for preclinical studies of DKD121.
Implications for clinical management
On the basis of the available evidence showing that patients with CKD and DM often have NDKD, we recommend that all patients with DM and signs of CKD (as defined by parameters of renal excretory dysfunction, urine, or imaging abnormalities that persist for more than 3 months) undergo diagnostic assessment as is recommended for patients without DM37. Such an approach involves careful assessment of medical history for evidence of familial or genetic forms of CKD, preterm birth or low birthweight, previous AKI episodes, toxin exposures, or systemic disease states (other than DM) that potentially affect the kidney, including chronic infections (for example, viral hepatitis and HIV), autoimmune disease (for example, systemic lupus erythematosus and systemic vasculitis), haematological malignancies (for example, lymphoma, myeloma, and monoclonal gammopathy), or solid tumours. This evaluation can include extensive diagnostic work-up, including genetic testing and/or diagnostic kidney biopsy11,37.
Despite the nonspecificity of certain histomorphological features, kidney biopsy can usually discern DKD from NDKD or document their coincidence on the basis of the presence or absence of characteristic histomorphological patterns (Box 1; Table 1). In experienced hands, the risks associated with renal biopsy are typically minimal124. Bleeding complications are a concern mainly in patients with low platelet counts or advanced CKD124. As mentioned above, kidney biopsy may also aid prognostication119.
Targeting cardiovascular mortality
Reducing overall mortality is the main priority in managing patients with CKD plus DM and requires rigorous control of cardiovascular risk factors such as hyperglycaemia, smoking, obesity, dyslipidaemia, and hypertension. Management of hypertension is preferably achieved with RAS inhibitors3. Among the antidiabetic drugs, metformin, SGLT2 inhibitors, and the GLP1 inhibitor liraglutide have demonstrated survival benefits, as described above18,21,22,23. Although clinical trials in this domain often refer to patients having DKD, this diagnosis is almost never confirmed by kidney biopsy. However, as all patients with CKD plus DM (that is, those with DKD, NDKD, or a combination of both) are at increased risk of cardiovascular disease, all patients should benefit from the rigorous control of cardiovascular risk factors. Further research should focus on the mechanisms underlying vascular dysfunction in DM, including those underlying endothelial dysfunction, vascular wall calcification and atheroma formation, glycocalyx and endothelial barrier dysfunction, and microvascular occlusions (including retinal and cerebral ischaemia, dementia, glomerulosclerosis, heart failure, and cardiac fibrosis). Defining the specific molecular mechanisms driving these complications is a prerequisite for developing innovative cures.
Targeting CKD progression
A missed diagnosis of NDKD represents a missed opportunity to identify specific approaches to slow the progression of kidney diseases such as glomerulonephritis and those caused by paraproteinaemia or amyloidosis, which have pathogenic processes that are independent of increased SNGFR122. Diagnosis of any NDKD requires initiation of management approaches (for example, immunosuppressive therapy for entities such as lupus nephritis or renal vasculitis). For patients with concurrent DM, use of SGLT2 and RAS inhibition to reduce glomerular hyperfiltration (and hence total GFR) is essential to limit GFR decline30. SGLT2 inhibitors are currently approved only for patients with a GFR >45 ml/min/1.73 m2, although data concerning their efficacy and safety in more advanced stages of CKD are eagerly awaited. GLP1 activation with liraglutide also elicits renoprotective effects but does not seem to have immediate effects on total GFR29. Whether a combination of SGLT2 and RAS inhibition and a GLP1 analogue will have additive effects on CKD progression is currently unknown. Glucose and body weight control will also help to reduce hyperfiltration in patients with NDKD plus DM45,116,125. Improved understanding of the less specific, downstream mechanisms of CKD progression under hyperglycaemic conditions, including epigenetic programming and signalling defects, mechanisms of podocyte detachment and loss, oxidative stress and inflammation-related effects, and the cellular and molecular mechanisms of renal scarring, will aid in the identification of further targets for both DKD and NDKD.
Cardiovascular death and CKD progression are the two main unmet medical needs for patients with CKD plus DM regardless of whether the underlying renal dysfunction is the direct result of the diabetic environment (that is, DKD) or represents an NDKD superimposed on DM. The metabolic and haemodynamic consequences of DM on the kidney require glycaemic control and dual SGLT2 and RAS inhibition, but the diagnosis of NDKD might identify additional therapeutic approaches. The important contribution of NDKD to the global epidemic of CKD in patients with obesity-related T2DM is rarely acknowledged or mirrored in the design of preclinical studies. A greater number of protocol biopsy studies are needed to better define the true prevalence of NDKD in DM and to guide treatment, given the possibility that a diagnosis of NDKD may offer additional treatment options. Refining strategies in basic research, drug development, and preclinical drug testing to better align preclinical models with the complex processes underlying CKD in DM — as well as the use of biopsy to aid in the enrolment of more homogeneous patient populations in clinical trials — might help to address the unmet medical needs of these patients, increase the predictability of preclinical studies for clinical trial outcomes, and improve the overall effectiveness of translational research in this important domain of kidney research.
H.-J.A. is supported by the Deutsche Forschungsgemeinschaft (DFG; AN372/24-1) and the European Union (EU)’s Research and Innovation programmes (under grant agreements Horizon 2020 and NEPHSTROM No. 634086). T.B.H. is supported by the DFG (CRC1140, CRC 992), by the Federal Ministry of Education and Research (BMBF; 01GM1518C; Germany), by the European Research Council grant 616891, and by the Horizon 2020 Innovative Medicines Initiative 2 consortium BEAt-DKD (Biomarker Enterprise to Attack DKD). M.S. was supported by BMBF grant 01GM1518A and a Fritz Thyssen Grant (10.16.2.026MN). The views expressed here are the responsibility of the authors only. The EU Commission takes no responsibility for any use made of the information set out.
- Diabetic kidney disease
(DKD). Also known as diabetic nephropathy. Often defined as a clinical syndrome of albuminuria in patients with diabetes mellitus but more accurately defined as a distinct histopathological pattern of kidney injury, characterized by arteriolar hyalinosis and nodular glomerulosclerosis, induced by hyperglycaemia.
- Sodium–glucose cotransporter 2
(SGLT2). A transporter expressed in the S2 segment of the (convoluted) proximal tubule that reabsorbs freely filtered glucose and cotransports sodium.
- GLP1 analogue
Glucagon-like peptide 1 analogues are incretin mimetics that reduce meal-related hyperglycaemia and have a low risk of causing hypoglycaemia. Liraglutide reduced cardiovascular and renal end points in patients with type 2 diabetes mellitus.
- Kidney hypertrophy
A diffuse nephron hypertrophy that increases the size of the kidney.
- Tubuloglomerular feedback
An autoregulatory mechanism of glomerular perfusion that maintains a submaximal single-nephron glomerular filtration rate (SNGFR) and responds to changes in intravascular fluid volume. Hyperglycaemia deactivates tubuloglomerular feedback and leads to a persistent increase in SNGFR.
- Glomerular hyperfiltration
The consequence of an increase in single-nephron glomerular filtration rate due to impaired autoregulation of glomerular haemodynamics or a reduction in the nephron number:body mass ratio.
- Glomerular hypertension
The increased pressure gradient across the glomerular filtration barrier that occurs, for example, when glomerular hyperfiltration is not associated with a respective increase in filtration surface.
- Sterile inflammation
Non-infectious causes of inflammation such as those that occur in autoinflammatory or autoimmune disorders upon trauma or toxic tissue injury.
L-Ribonucleic acid aptamers that mirror structures of natural RNA oligonucleotides, protecting them from enzymatic degradation. Spiegelmers can bind and neutralize small proteins.
- Danger-associated molecular patterns
(DAMPs). Natural cellular components, often released by dying cells, that induce inflammation via specific pattern recognition receptors of the innate immune system.
Cytokines secreted by the adipose tissue, including leptin, adiponectin, and apelin.
- Nephron hypertrophy
The increased dimensions of nephrons, usually as a consequence of glomerular hypertension and hyperfiltration.
- Total GFR
In poorly controlled diabetes, the total glomerular filtration rate (GFR) is close to maximal GFR.
- Nephron number
A critical parameter for kidney function. The number of nephrons multiplied by single-nephron glomerular filtration rate (GFR) equals the total GFR.
- Kimmelstiel–Wilson lesions
A lesion characterized by nodular glomerulosclerosis that is commonly found in biopsy samples from patients with diabetes mellitus.