Introduction

Diabetic nephropathy has become the leading cause of end-stage renal disease in the United States, Europe and Japan. Furthermore, the number of affected patients in India and China and other parts of the developing world is growing rapidly, creating a worldwide burden both for patients and health care systems. The syndrome is characterised by persistent albuminuria (>300 mg/24 h or 200 μg/min), early elevation of arterial blood pressure, a relentless decline in renal function, and increased cardiovascular morbidity and mortality. Previous studies found a cumulative risk of 25–40% after a diabetes duration of at least 25 years in type 1 and type 2 diabetes [13]. More recent epidemiological data suggest that it is possible to improve the outcome and reduce the number of patients developing overt diabetic nephropathy and other diabetic microvascular complications with improved diabetes care, although these positive results may so far be limited to dedicated centres [4]. In order to reduce the number of patients who develop diabetic nephropathy, it is important to understand the mechanisms underlying progression. This should make it possible to identify additional initiators and factors promoting progression, potentially leading to targeted therapy. It will also enable patients at high risk of initiation or of accelerated progression of diabetic nephropathy to be identified, enabling clinical care to be focused upon them. This review will focus upon identification of patients at risk for development and progression of diabetic nephropathy, consideration of modifiable progression promoters, and possible ways of improving treatment.

Risk factors for development of diabetic nephropathy

Progression from normoalbuminuria to microalbuminuria defines the initiation of diabetic nephropathy, and the transition from microalbuminuria to overt diabetic nephropathy resulting in deterioration of renal function and end-stage renal disease constitutes its progression. The presence of microalbuminuria is associated with increased cardiovascular morbidity and mortality, and regular screening is recommended in guidelines for diabetes care [5]. In an attempt to find risk factors and markers for progression from normoalbuminuria to microalbuminuria, we followed a cohort of 537 normoalbuminuric type 1 diabetic patients for 10 years. Identified in a cross-sectional study in 1984 [6], these patients had a median duration of diabetes of 20 years at baseline, and 25% developed microalbuminuria during follow-up. In a more recent study [7], we followed a separate cohort of 277 type 1 diabetic patients from onset of diabetes (i.e. an inception cohort) for 18 years, during which time 29% developed microalbuminuria.

Genetic risk factors

The observation of familial clustering of diabetic nephropathy [810], and of racial variation in the development of diabetic renal complications [11], suggest that genetic factors are involved in the development of diabetic nephropathy. This is currently considered a complex genetic trait, requiring a combination of alleles of several genes in addition to environmental factors. Several candidate genes have been studied, including diabetes susceptibility genes and genes involved in glucose metabolism and regulation of blood pressure, in addition to the growth factor genes reviewed by Tarnow [12]. Owing to small sample sizes and differences in design and the populations studied, the results have been conflicting. Genes involved in the renin-angiotensin system (RAS) have attracted particular interest, and the insertion/deletion polymorphism in the gene for ACE, related to circulating levels of ACE, was shown in a meta-analysis to have a modest effect on the risk for development of diabetic nephropathy in patients of European extraction [13]. In accordance with the concept of a complex genetic trait, it has been suggested that much larger studies (>1,000 patients) will be needed before analysis of combinations of genetic traits will become possible.

Constitutional factors

It has been suggested that intrauterine growth retardation will lead to a reduction in nephron number, and hence to an inborn deficiency in renal functional reserve capacity [14] (Fig. 1). In our follow-up study of the inception cohort we demonstrated that slightly elevated urinary albumin excretion at onset of diabetes already predicted the development of microalbuminuria [7]. It is possible that this slight increase in urinary albumin excretion is a marker of abnormal glomerular haemodynamic and permselective conditions predisposing to diabetic glomerulopathy as a result of a reduction in renal functional reserve capacity. Birthweight is directly associated with adult height, and could therefore also explain why short stature (low birthweight) in this and previous studies was related to initiation of microalbuminuria [7] and the presence of diabetic nephropathy [15], as well as to development of microalbuminuria in non-diabetic men [16]. In type 1 diabetic women with intrauterine growth retardation (birthweight below the 10th percentile) we demonstrated an increased risk for development of diabetic nephropathy [17]. This supports the hypothesis that factors operating in utero or early life and genetic predisposition are both associated with initiation of diabetic nephropathy, a suggestion that has also been made for other renal diseases [18].

Fig. 1
figure 1

The relationship between low birthweight, reduced functional renal reserve capacity (nephron number) and increased susceptibility to renal disease [115]

Systemic hypertension

Our study of the inception cohort suggests that at onset of diabetes a small increase in systemic blood pressure, within the normal range, is already important for the initiation of diabetic nephropathy [7]. This is in accordance with several other studies, but not all were able find such an association [6, 1922], possibly due to differences in study design and study population or to differences in methods of blood pressure measurement. Since the changes in blood pressure are small and may be more pronounced at night, 24-h ambulatory blood pressure measurements may be more precise [24]. A study of 75 type 1 diabetic patients followed for a mean of 5 years with repeated measurements of ambulatory blood pressure further suggested that a change in diurnal variation in blood pressure with loss of nocturnal dipping could be observed prior to the development of microalbuminuria, although non-dipping at baseline was not predictive for progression [24].

Oral contraceptives–RAS activation

The importance of the activated RAS in the development of diabetic renal disease is reflected by the predictive effect of elevated prorenin levels [24, 25]. Since oral contraceptives also activate the RAS, we evaluated the impact of their use on the RAS in diabetic subjects, and found an elevated activity in those on oral contraceptives as compared with non-users. This was assessed by the renal haemodynamic effects of acute RAS blockade with the ACE inhibitor captopril in subjects on a high-sodium diet [26]. The clinical impact was evaluated in a longitudinal follow-up study in which 18% (six of 33) of oral contraceptive users compared with 2% (two of 81) of non-users of oral contraceptives developed overt diabetic nephropathy (p=0.008 after adjustment for known risk factors) (Fig. 2) [26]. Large prospective studies are required to investigate this relationship further, but a population-based study also reported an association between use of oral contraceptives or hormonal replacement therapy and the presence of microalbuminuria [27].

Fig. 2
figure 2

Oral contraceptive (OC) use and the development of macroalbuminuria in an inception cohort of 114 type 1 diabetic women (81 OC users) followed for a mean period of 20 years. *p=0.003, and in a Cox model adjusted for known risk factors relative risk=8.9 (95% CI 1.79–44.36, p=0.008). Copyright 2005 American Diabetes Association. From Diabetes Care (2005) 28:1988–1994. Reprinted with permission from The American Diabetes Association [26]

Hyperglycaemia

The relationship between hyperglycaemia and the development of microvascular complications was described by Pirart [28] and has since been demonstrated in many observational studies [6, 7, 1921, 29, 30]. Furthermore, intensive therapy aimed at improving glycaemic control reduced the risk for development of diabetic nephropathy and retinopathy (primary prevention) in the smaller randomised studies combined in a meta-analysis by Wang [31], and these findings were confirmed and extended in the Diabetes Control and Complications Trial [32]. It has been suggested that glucose induces vascular damage through four pathways: (1) increased polyol pathway flux; (2) increased advanced glycation end-product formation; (3) activation of protein kinase C; and (4) increased hexosamine pathway flux [33]. Overproduction of superoxides by the mitochondria is considered a common link between these pathways [34, 35]. This suggests new treatment options for the future, but blocking these pathways has not so far been feasible or effective in man.

Smoking

We observed an increased risk for the development of microalbuminuria in smokers in our 10-year follow-up study [29]. Several other observational studies have found the same [30, 36], although others failed to demonstrate this relationship [21, 37]. An observational study of 943 type 1 diabetic patients followed for 4 years found that smoking modified the effect of hyperglycaemia, magnifying the influence of poor glycaemic control [30]. Smoking is thus an obvious modifiable risk factor, although intervention studies have not been performed to confirm this. The mechanisms are poorly understood, but an abrupt rise in systolic blood pressure and heart rate has been observed during the course of heavy smoking, although there was no effect on microalbuminuria or GFR [38]. Smoking has also been proposed as a modifiable progression promoter [39]; however, when we followed 301 type 1 diabetic patients with overt diabetic nephropathy for at least 3 years with a precise measure of GFR we found no impact of smoking on the rate of decline in GFR [40].

Risk assessment for development of microalbuminuria

Over 10 years of follow-up, 25% of initially normoalbuminuric patients developed microalbuminuria. This enabled us to create a simple risk assessment scheme (Fig. 3), demonstrating that the risk of progression from normoalbuminuria to microalbuminuria and macroalbuminuria within 10 years was 70% if the patient had a combination of four risk factors, namely, retinopathy, urinary AER>10 mg/24 h, HbA1c>8.6% and smoking, as against only 10% risk of progression if none of these risk factors were present [6]. Similar data emerged for type 2 diabetic patients (P. Rossing, unpublished data). This type of risk assessment is convenient to use in the clinical situation and can be used to identify high-risk individuals for intervention trials.

Fig. 3
figure 3

Individual risk of progression to microalbuminuria during 10-year follow-up in type 1 diabetic patients with normoalbuminuria. Based on the presence or absence of HbA1c >8.6%, urinary AER >10 mg/24 h, and the number of additional risk markers (smoking and/or diabetic retinopathy). Green 0 additional risk factors, yellow 1, red two. Copyright 2002 American Diabetes Association. From Diabetes Care (2002) 25:859–864. Reprinted with permission from The American Diabetes Association [6]

Does microalbuminuria predict progression to overt nephropathy?

The presence of microalbuminuria was originally found to be predictive of overt diabetic nephropathy in 80% of type 1 diabetic patients untreated with antihypertensive agents, after 6–14 years of follow-up [4143]. More recent studies found overt nephropathy in only 30–45% of microalbuminuric patients after 10 years [7, 44]. This decrease may be due to earlier and more aggressive treatment of blood pressure, and ACE inhibitors in particular have been shown to reduce progression to overt nephropathy in normotensive type 1 diabetic patients with microalbuminuria [45]. It has further been suggested that as many as 58% of patients with microalbuminuria may revert to normoalbuminuria [46], bringing the value of microalbuminuria as a predictor of overt nephropathy into question [47]. Other studies with longer follow-up periods found spontaneous regression to normoalbuminuria in only 13–15% of patients; additional treatment-induced regression was seen in a similar number [7, 44], suggesting that microalbuminuria remains an important risk factor.

Microalbuminuria in diabetes of long duration

In the initial studies, in which up to 80% of patients progressed to overt nephropathy, the duration of diabetes was short. In an observational study of longer term type 1 diabetes (duration >15 years) the rate of progression over 10 years of follow-up was only 28% [48], and a small study from the UK found that 32% of patients with microalbuminuria and a type 1 diabetes duration of >30 years progressed to overt nephropathy [49]; these studies excluded patients with a short duration of diabetes. In a prospective observational study of 181 microalbuminuric type 1 diabetic patients, we found that the rate of progression to overt nephropathy was 45% in those with a diabetes duration of <15 years, as compared with a rate of 26% in patients with a duration ≥15 years [44]. In microalbuminuric type 1 diabetes patients with >40 years of diabetes, the incidence of overt nephropathy was 4% per year.

Raised arterial blood pressure was not associated with risk of progression from microalbuminuria to overt nephropathy in observational studies of type 1 diabetic patients [44, 5052] although more sensitive methods, such as 24-h blood pressure measurements, should possibly have been used [53]. Changes in blood pressure were only apparent close to the development of overt nephropathy or in relation to metabolic control [54]. Despite this, treatment with antihypertensive medication, ACE inhibitors in particular, has been the most successful means of avoiding progression from microalbuminuria to overt nephropathy (secondary prevention). In a meta-analysis of all randomised trials of at least 1 year’s duration in microalbuminuric type 1 diabetic patients, ACE inhibitors reduced progression to overt nephropathy by 62%, and increased regression to normoalbuminuria three-fold compared with placebo [45]. Similar findings came from microalbuminuric type 2 patients treated with an angiotensin II receptor blocker [55]. The effect of ACE inhibition on urinary albumin excretion in microalbuminuric type 1 diabetic patients is long lasting, and GFR was preserved in an 8-year randomised study [56]. What is more, this treatment strategy can be implemented effectively in everyday clinical practice [57].

The importance of glycaemic control for progression to nephropathy has been demonstrated in several studies [44, 48, 50, 51, 58], but has been hard to demonstrate in randomised controlled trials with intensive metabolic control [52, 59]. This might be due to inadequate sample sizes, insufficient separation of mean values for ‘intensive’ and ‘conventional’ glycaemic control, or insufficient follow-up time. In type 1 diabetic patients, pancreas transplantation resulting in normoglycaemia can reverse glomerulopathy in patients with normoalbuminuria (n=3) or microalbuminuria (n=4), but more than 5 years of normoglycaemia are required for this to occur [60].

Improved prediction with cardiovascular risk markers

Additional tests will be needed to improve upon the accuracy of existing risk estimates for development of diabetic nephropathy. Variables that have been investigated include hyperlipidaemia [50], features of the insulin resistance syndrome [22, 50, 61, 62], and a family history of cardiovascular disease [63, 64]. Cardiovascular risk assessment involves much the same risk factors and markers as renal risk assessment, and there has recently been much interest in biochemical markers or factors associated with putative pathophysiological mechanisms for cardiovascular disease, such as endothelial dysfunction, dysregulation of coagulation and fibrinolysis, complement activation and inflammation [65]. Inflammatory activity is increased in type 1 diabetes [66], and a recent cross-sectional study found an independent association between markers of inflammation (C-reactive protein, IL-6 and TNF-α in a combined index) and the presence of microvascular complications [67]. It has been suggested that microalbuminuria is only associated with progression to overt nephropathy in the presence of endothelial dysfunction [68]. Inflammation and complement activation via the mannose-binding lectin pathway may also play a role in the pathogenesis of diabetic microvascular complications, and a prospective study found that increasing levels of mannose-binding lectin early in the course of type 1 diabetes were significantly and independently associated with the later development of persistent microalbuminuria or macroalbuminuria [69]. Cytokines such as TGF-β, a pro-sclerotic cytokine considered to be the major mediator of collagen formation in the kidney, and connective tissue growth factor (another pro-sclerotic cytokine) are other possible markers, but large and prospective studies are lacking. Vascular endothelial growth factor, a cytokine that induces angiogenesis and increases vascular permeability, was elevated in plasma early in the course of diabetic nephropathy, as compared with normoalbuminuric controls, but only in men [70]. The predictive value of these biomarkers and their response to intervention is not yet clear, one problem being that plasma or urinary concentrations of these markers may not reflect local concentrations in the kidney or other organs. These potential markers are, however, of great interest as possible new targets for drug action.

Structural markers

It has been suggested that structural parameters can be used as outcome predictors, and 5-year follow-up of normoalbuminuric type 1 diabetic patients showed that glomerular basement membrane width was increased at baseline in patients developing microalbuminuria over the study period [71]. It has been argued that morphological changes are sometimes present before functional changes such as microalbuminuria, thereby potentially allowing earlier identification of high-risk individuals [47], but there are few prospective data. It has also been suggested that molecular biology techniques could be applied to renal biopsies. For example, tissue levels of RNA encoding molecules known to be involved in the pathogenesis of renal disorders, determined in routine clinical kidney biopsies, predicted loss of renal function in more advanced stages of nephropathy [72]. The requirement for renal biopsy limits the usefulness of such approaches.

Prediction of progression and treatment efficacy in overt diabetic nephropathy

When a patient develops overt diabetic nephropathy with persistent macroalbuminuria and elevated arterial blood pressure, kidney function starts to decline. In the microalbuminuric stage, hyperfiltration is often present, and the GFR is usually high or normal at the onset of nephropathy. Before the introduction of antihypertensive therapy, the rate of decline in GFR was 10–20 ml/min per year [7376]. Today, antihypertensive medication, particularly blockade of the RAS, has reduced the mean rate of decline in GFR to 2.0–10 ml/min per year in randomised controlled trials [7780]. There is, however, still large inter-individual variation in the rate of decline, ranging from stable kidney function to a rapid decline leading to renal replacement therapy or death within a few years. Factors accounting for this large variation in the loss of kidney function have been termed progression promoters, and some possible progression promoters [81, 82] are depicted in Fig. 4.

Fig. 4
figure 4

Putative promoters for progression of diabetic nephropathy

A recent review of putative genetic progression promoters [83] found that the focus has, for the most part, been on genes involved in the RAS, although these genes, even in combination, make only a small contribution to the loss of kidney function [84]. They may, however, contribute to the choice of treatment. In type 1 diabetic patients with overt nephropathy, the rate of decline in renal function during treatment with an ACE inhibitor was accelerated in patients homozygous for the deletion allele D of the insertion/deletion polymorphism as compared with patients homozygous for the insertion allele I or the heterozygous patients [85], whereas the decline in renal function was similar in these patients when treated with an angiotensin II receptor blocker [86]. This implies that the optimal means of blocking the RAS depends upon the genotype. It is also possible that genetic factors related to pharmacodynamic aspects of medication could influence the therapeutic effect [87].

Hypertension and proteinuria are established and modifiable progression promoters [76, 88], and intervention targeting these factors has been the most successful renoprotective treatment in diabetic nephropathy. The primary endpoint in clinical trials of diabetic nephropathy has been the rate of decline in renal function, development of end-stage renal disease or death, whereas a reduction in urinary albumin excretion has been considered a surrogate endpoint, a predictor of a beneficial outcome, and a reduction in albuminuria shortly after onset of antihypertensive therapy is the best predictor of long-term preservation of renal function [89, 90]. Instead of relying solely upon the urinary AER, we may in future be able to study treatment-induced changes in the pattern of the large number of polypeptides excreted in the urine using proteomics with high-throughput mass spectrometry of urine. In a study of type 2 diabetic patients with diabetic nephropathy treated with angiotensin II receptor blockade in a cross-over study, it was possible to identify treatment-induced changes in several excreted polypeptides, inducing partial normalisation of the initial polypeptide pattern that typifies overt diabetic nephropathy [91].

Optimising the intervention in the renin–angiotensin–aldosterone system

Despite the success of ACE inhibitors in treating diabetic nephropathy, not all patients obtain satisfactory control of blood pressure, albuminuria and decline in renal function. Studies have therefore tried to optimise the blockade of the renin-angiotensin-aldosterone system by: (1) increasing the doses of medication beyond the doses used to treat hypertension; (2) combining ACE inhibition and angiotensin II receptor blockade (dual blockade); or (3) inhibition of aldosterone.

The doses used have been based on studies of hypertension, and dose-response curves have not been constructed for ACE inhibitors in diabetic nephropathy. The optimal dosing for renoprotection might differ from the doses used in hypertension, although the optimal doses for each were similar in a short-term study of type 1 diabetic patients with nephropathy treated with the angiotensin II receptor blocker losartan [92]. In contrast, a study of the angiotensin II receptor blocker candesartan revealed a different optimal dose for blood pressure response (8 mg/day) and renal effects (16 mg/day) in type 2 diabetic patients with nephropathy [93]. Another study showed that, in microalbuminuric type 2 diabetic patients, ultra-high doses of the angiotensin II receptor blocker irbesartan (900 mg/day) offered an additional antiproteinuric effect of 15% compared with the usual high dose of 300 mg/day (p=0.02) [94].

As ACE inhibitors and angiotensin II receptor blockers act on different sites in the RAS, combination of the two agents (dual blockade) has been tested. ACE inhibitors have the advantage of increasing bradykinin, but may be bypassed by non-ACE conversion of angiotensin I to angiotensin II [95]. Angiotensin II receptor blockers cannot be bypassed, but incomplete blockade of the angiotensin II receptor type 1 or effects on other angiotensin II receptors are of importance in animal models [96]. The first study to use dual blockade in diabetes was the Candesartan and Lisinopril Microalbuminuria (CALM) study. This included 199 hypertensive microalbuminuric type 2 diabetic patients and found that the combination of lisinopril and candesartan was more effective than either agent alone in reducing blood pressure [97]. In a study of type 1 diabetic patients with overt nephropathy and albuminuria >1,000 mg/24 h despite antihypertensive therapy including ACE inhibition, addition of the angiotensin II receptor blocker irbesartan reduced albuminuria by 37% [98]. Similar results were found in type 2 diabetic patients with overt nephropathy [99]. It has been argued that increasing the dose of a single agent might be equally effective [100], but in a study of type 1 diabetic patients with overt nephropathy, addition of the angiotensin II receptor blocker irbesartan to the maximum recommended dose of ACE inhibition produced a 25% further reduction in albuminuria and a further significant reduction in blood pressure [101]. Since all these studies have been of rather short duration (up to 1 year), longer term trials are needed to establish the role of dual blockade of the RAS in diabetic nephropathy. As regards patients with non-diabetic nephropathy, the Combination Therapy of Angiotensin-II Receptor Blocker and Angiotensin-Converting Enzyme Inhibitor in Non-diabetic Renal Disease (COOPERATE) study, which included 263 patients followed for a median of 3 years, found that only 11% of patients on dual blockade (trandolapril and losartan) progressed to the clinical endpoint (doubling of serum creatinine or end-stage renal disease), in contrast to 23% receiving either agent as monotherapy (p=0.02) [102].

There has recently been an additional focus on aldosterone as a mediator of renal and cardiovascular disease via effects on fibrosis, necrosis and inflammation [103]. Clinical trials with blockade of the renin–angiotensin–aldosterone system with ACE inhibitors or angiotensin II receptor blockers demonstrated that aldosterone levels increase in some patients after an initial reduction (the aldosterone escape phenomenon) [104, 105]. It was further demonstrated that aldosterone levels that rose during treatment were associated with a significantly faster rate of decline in GFR over long-term follow-up [105]. Blocking the effect of aldosterone with spironolactone (25 mg/day) in 13 type 2 diabetic patients with aldosterone escape during treatment with an ACE inhibitor induced a significant reduction in albuminuria [104]. This has been confirmed in controlled studies of type 1 and type 2 diabetic patients with diabetic nephropathy [106, 107].

Prognosis

Diabetic patients with microalbuminuria have an increased morbidity and mortality [108111]. This may be due to progression to overt diabetic nephropathy, or related to an increased risk of the microalbuminuric stage per se. Since urinary albumin excretion has been measured only at baseline in most follow-up studies, this issue could not initially be resolved. We were, however, able to demonstrate in a 10-year follow-up study of microalbuminuric type 1 diabetic patients, which included regular measurements of urinary AER, that increased mortality was related to progression to overt nephropathy, and not to the presence of microalbuminuria per se [108]. This implies that the prognosis of microalbuminuric type 1 diabetic patients can be improved provided progression to overt nephropathy can be avoided.

Early studies describing the prognosis of overt diabetic nephropathy observed a median survival of 5–7 years after the onset of persistent proteinuria [1, 112]. End-stage renal failure was the primary cause of death in 66% of patients. When deaths attributed only to end-stage renal disease were considered, the median survival time was 10 years [2]. All this was before patients were offered antihypertensive therapy. Long-term antihypertensive therapy was evaluated prospectively in 45 type 1 diabetic patients who developed overt diabetic nephropathy between 1974 and 1978. Ten years after onset of diabetic nephropathy the cumulative death rate was 18%, and the median survival was more than 16 years [113]. We went on to examine whether antihypertensive therapy also improved survival in an unselected cohort of 263 patients with diabetic nephropathy followed for up to 20 years, and observed a median survival of 13.9 years; only 35% of patients died because of end-stage renal failure (serum creatinine>500 μmol/l) [108]. Fortunately, survival continues to improve, and we recently showed a median survival rate of 21 years after onset of diabetic nephropathy [114] (Fig. 5).

Fig. 5
figure 5

Cumulative death rate from onset of diabetic nephropathy in type 1 diabetic patients during the natural history of diabetic nephropathy (red line, n=45, Knowles [112]; yellow line, n=360 Andersen et al. [1]) compared with patients who had effective antihypertensive treatment (orange line, n=45, Parving et al. [113]; black line, n=263, Rossing et al. [108]; green line, n=199, Astrup et al. [114]). From [114] with permission from Blackwell Publishing

In conclusion, a number of potentially modifiable risk factors, including poor glycaemic control, raised blood pressure, increased urinary albumin excretion and smoking, predict the initiation and/or progression of diabetic nephropathy. Microalbuminuria carries a high risk for development of overt diabetic nephropathy. New biomarkers may in future permit more precise risk assessment, but at present it is essential to screen for known risk factors. Interventions aimed at strict glycaemic control, to avoid initiation of diabetic nephropathy, and blockade of the RAS, to avoid its progression, are of demonstrated value in clinical trials and are feasible in clinical practice.