2.1 Microalbuminuria

Persistent microalbuminuria was defined as a urinary albumin excretion rate between 30 and 300 mg/24 h in at least two out of three consecutive samples, as recommended in a consensus report (8, 9). The upper boundary of microalbuminuria corresponds to dipstick (Albustix) positive urines (albumin levels are ~100 to 200 mg/l), and thus microalbuminuria can not be detected by conventional stix. Type 1 diabetic patients with an albumin excretion rate in the microalbuminuric range (from 30 to 300 mg/24 h) have long lasting preserved kidney function (10). Time of onset of microalbuminuria was defined as the first recorded positive urine sample in at least two out of three consecutive samples.

2.2 Diabetic nephropathy

Persistent albuminuria was defined as a urinary albumin excretion rate above 300 mg/24 h in at least two out of three consecutive samples (8). Diabetic nephropathy was diagnosed clinically if the following criteria were fulfilled: persistent albuminuria, duration of diabetes of more than 10 years, presence of diabetic retinopathy, and absence of clinical or laboratory evidence of other kidney or renal tract disease (7, 8, 11). If these criteria were not fulfilled, a diagnostic kidney biopsy was performed, which was required to show only diabetic glomerulosclerosis. Time of onset of diabetic nephropathy was defined as the first recorded positive urine sample in at least two out of three consecutive samples.

2.3 Nephrotic range albuminuria

Patients with nephrotic range albuminuria have a very poor prognosis (12, 13). Nephrotic range proteinuria has previously been defined as persisting proteinuria above 3500 mg/24 h (14). A correction factor between proteinuria and albuminuria of 0.70 was applied (unpublished data), and thus in our studies, nephrotic range albuminuria was defined as persisting albuminuria above 2500 mg/24 h in at least two out of three consecutive 24 hours urine collections.

2.4 Progression of diabetic nephropathy

The decisive hard endpoint of chronic progressive kidney diseases is ESRD, where renal replacement therapy becomes essential for survival. To evaluate disease progression and therapeutic interventions in clinical practise in the predialysis state, other endpoints are required. One possibility is to study the progression of renal morphological lesions, as renal structural lesions can be detected even in the early stages of diabetic nephropathy, where GFR is still within the normal range (15). However, to evaluate structural progression of kidney disease, the primary endpoint need to be clearly defined, and requires serial biopsies over time. Due to several practical and ethical issues, this has only been done in few studies (16, 17). Even though surrogate markers of decline in GFR - such as doubling of serum creatinine - is cheap and convenient, the sensitivity is low, and the endpoint is only relevant when a sufficient number of events can be expected (18). The Food and Drug Administration in the USA have approved the rate of decline in GFR as an endpoint in kidney diseases, if GFR is measured for a period of two to three years (19). In patients with short duration of follow-up, the variability of the decline in GFR is large, with loss of precision of the outcome. Furthermore, short term changes after initiation or intensified antihypertensive treatment may dilute the potential beneficial effect of intervention. Therefore, the following requirements should be fulfilled in order to make a valid determination of the rate of decline in GFR: the applied GFR method should have a good accuracy and precision, repeated measurements of GFR (approximately every 6-12 months) should be performed, and the observation period should be extended to at least 2 years (20).

Intervention that reduces the progression of diabetic nephropathy has always been associated with a reduction in proteinuria as reviewed by Parving (21). In addition, the initial reduction in albuminuria during antihypertensive treatment predict the long-term decline in kidney function in both diabetic and non-diabetic nephropathies (22-24). Therefore, in studies of shorter duration, changes in albuminuria can be used as a surrogate endpoint for the rate of decline in GFR. In our studies of progression of diabetic nephropathy, we have measured GFR with 51 Cr-EDTA clearance (see section 3) repeatedly at least yearly for a minimum of three years in each patient.

2.5 Remission

2.5.1 Regression to normoalbuminuria

After microalbuminuria has developed, some patients may revert to normoalbuminuria for a shorter or longer period of time. Especially after the introduction of early secondary prevention strategies in the treatment of diabetic nephropathy, regression of microalbuminuria occur. We defined regression of microalbuminuria to normoalbuminuria as a urinary albumin excretion rate below 30 mg/24 h in two out of three consecutive 24 hours urine collections, in addition to a decrease in urinary albumin excretion rate of at least 30% from the microalbuminuria level, sustained for at least one year. Spontaneous regression to normoalbuminuria was classified as regression observed in patients not receiving blood pressure lowering agents before regression was obtained.

2.5.2 Remission and regression of diabetic nephropathy

In our studies of remission and regression of overt diabetic nephropathy, remission of diabetic nephropathy was defined as a decrease in albuminuria to the microalbuminuric range (below 300 mg/24 h (equivalent to 200 μ g/min)) (8) in at least two out of three consecutive 24 hour urine collections, and a decrease of at least 30% from pre-remission levels, sustained for at least one year during follow-up. Regression of diabetic nephropathy was defined as a rate of decline in GFR equal to or less than 1 ml/min/year during the entire observation period. This decline in GFR is equivalent to the decline in GFR due to the natural ageing process in non-diabetic patients without kidney disease (25). Spontaneous remission and regression was defined as remission or regression observed in patients not receiving blood pressure lowering agents before or during follow-up.

2.5.3 Remission of nephrotic range albuminuria

Remission of nephrotic range proteinuria has previously been classified as a reduction from nephrotic range proteinuria (proteinuria above 3500 mg/24 h) to below 1000 mg/24 h (14). Keeping the above-mentioned correction factor between proteinuria and albuminuria of 0.70 in mind, we used the following definition of remission of nephrotic range albuminuria: a reduction in albuminuria from nephrotic range albuminuria (albuminuria above 2500 mg/24 h) to below 600 mg/24 h, sustained for at least one year during the follow-up period.

3.1 Glomerular filtration rate

Glomerular filtration rate was measured in the supine position after a single intravenous injection of 3.7 MBq 51 Cr-EDTA by determination of the radioactivity in venous blood samples taken 180, 200, 220 and 240 minutes after the injection (26). Glomerular filtration rate measured with 51 Cr-EDTA plasma clearance underestimates GFR as compared with the gold standard - Inulin clearance - by 10% (27), and our results were multiplied by a factor 1.10. Extra renal loss was corrected for by subtracting 3.7 ml/min. Finally, results were standardised for 1.73m 2 body surface, using the patient's surface area at the start of the study throughout. The mean variability in GFR of each patient from day to day was 4%. Linear regression analysis, least square method, was used to determine the rate of decline in GFR for each patient.

3.2 Statistical methods

Statistical analysis was performed by standard statistical methods using a commercially available program, SPSS version 8.0 to 11.0 (SPSS Inc., Chicago, Illinois, USA). More sophisticated analysis were performed by a biostatistician, and includes evaluation of non-linear effects of progression promoters on the decline in GFR (28), and Cox regression analysis with delayed entry used in the study of remission of nephrotic range albuminuria (29). These calculations were performed by the freely available software R (http://www.ci. tuwien.ac.at/R). A p-value < 0.05 (two sided) was considered statistically significant in all analyses.

7.1 Genetic susceptibility

The fact that some diabetic patients are vulnerable, while others seems to be protected against the impact of hyperglycaemia lead to the search for non glycaemic factors modulating the risk of the renal complications in diabetes. The striking racial variation (95) in the prevalence of diabetic kidney disease supports a genetic heredity of vulnerability to the harmful effects of hyperglycaemia. Furthermore, several studies indicate a familial clustering of diabetic nephropathy as well as cardiovascular disease (96-101). The occurrence of familial clustering strongly supports the hypothesis of genetic factors being involved in the development of diabetic kidney disease. However, mutual environmental risk factors in the affected patients could in fact explain the observed familial clustering. So far, none of the established risk factors for the development of diabetic kidney disease exhibits large enough effects, and the concordance rates are most likely explained by shared genetic traits. At present time, diabetic nephropathy is considered to be a complex genetic trait with a combination of alleles at several genes in addition to environmental factors providing the increased risk of the disease. Many studies of different putative candidate genes have been carried out, including diabetes susceptibility genes, genes involved in glucose metabolism and in regulation of blood pressure, genes regulating cardiovascular risk factors, and glomerular structural as well as growth factor genes as recently reviewed by Tarnow (102). Owing to small sample sizes, differences in designs and populations studied, results have been contradictory and subsequently difficult to interpret. The ACE I/D polymorphism have functional impact of the encoded product, as level of circulating ACE is related to genotype with patients with the DD genotype having twice as high plasma ACE as those with the II genotype (103), and may therefore have pathophysiological significance due to the actions of angiotensin II. Some have reported the ACE/ID polymorphism to be related to the development of diabetic nephropathy (104, 105), while others have failed to find an association (106). In a meta analysis, a modest effect of the D-allele of the ACE/ID polymorphism as a risk factor for the development of diabetic nephropathy in Caucasians has been demonstrated (107). In our study of the inception cohort of 286 type 1 diabetic patients followed from onset of diabetes, we did not find an association between the ACE/ID polymorphism and the development of micro-macroalbuminuria (unpublished data), but the study is likely to be underpowered for this type of analysis. In order to evaluate complex genetic traits like diabetic nephropathy, strict criteria of phenotyping, and a sufficient large number (>1000) of individuals are needed. Many studies of varying size have evaluated the impact of other genes on the risk of development of diabetic nephropathy with inconsistent results. To enhance the cost benefit of screening, and early treatment of patients at high risk of developing diabetic nephropathy, early identification of these patients is of utmost importance. In spite of the challenges met, evaluation of the genetics of diabetic nephropathy remains to be a key factor in that process. Furthermore, if new candidates genes can be identified, functional abnormalities related to the genes might be detected, which may give novel insight in the pathogenesis of the development of diabetic microangiopathy.

7.2 Metabolic changes

Hyperglycaemia

In vitro, hyperglycaemia induces basement membrane thickening, increases the synthesis of extracellular matrix, and vasoconstriction (108-110). Furthermore, hyperglycaemia is associated with disturbed cell cycle and accelerated death of endothelial cells (111). On the molecular level, four mechanisms have been implicated in this glucose mediated vascular damage: increased polyol pathway flux, increased advanced glycation end product (AGE) formation, activation of protein kinase C (PKC), and increased hexosamine pathway flux (112). The contribution of the polyol pathway may be very much species specific, and negative clinical trials with pharmacological inhibition of aldose reductase have questioned the relevance of this mechanism in humans (113). Glucose form early, non enzymatically, reversible glycosylation products with proteins (Schiff bases), which in turn rearrange into Amadori products. Through complex chemical rearrangements, irreversible AGE's are formed, which are capable of forming covalent bonds with other proteins (114). Recent studies have demonstrated Amadori products to modulate mesangial cell growth and collagen gene expression in vitro (115), and to be correlated with markers of endothelial dysfunction, and independently associated to diabetic nephropathy in humans (116). Advanced glycosylation end products can damage target cells in at least three different ways. 1) Intracellular proteins can be modified by AGE formation; thereby the function of these proteins can be altered. 2) Extracellular matrix components can be altered by AGE precursors, and subsequently interact abnormally with other matrix proteins, induce cross link formation, and interact with receptors for matrix proteins on cells. 3) Plasma proteins can be changed by AGE formation, bind to AGE receptors on endothelial and mesangial cells and macrophages, inducing production of reactive oxygen species (112). In animal models, the importance of AGE formation on the pathogenesis of microvascular complications have been demonstrated, as treatment with an AGE inhibitor (Aminoguanidine) is associated with reduced renal AGE accumulation, retardation in the development of albuminuria, and attenuation of mesangial expansion (117). A large randomized trial in type 1 diabetic patients with this compound, were discontinued prematurely due to toxicity. Meanwhile, other AGE inhibitors (118) as well as the role of specific AGE receptors (119) are under investigation.

Protein kinase C isoforms may be activated directly by hyperglycaemia (120) or indirectly through AGE receptors, and increased activity of the polyol pathway (112). Various actions of cytokines, e.g. vascular endothelial growth factor (VEGF) (121) and transforming growth factor- β (TGF- β ) (122) are shown to be PKC dependent. Animal models have demonstrated inhibitors of PKC- β to attenuate glomerular hyperfiltration, reduce albuminuria, and decrease expression of TGF- β (122, 123). Pharmacological inhibition of PKC- β in clinical trials in diabetic retinopathy are ongoing (124), although recent results have been disappointing (Abstract: Milton et al, Diabetes, Vol. 52, suppl. 1, page A127, 2003). Lastly, activation of the hexosamine pathway by hyperglycaemia may result in altered gene expression and protein function contributing to the pathogenesis of diabetic complications (112).

A common link between the four basic mechanisms of microvascular damage induced by hyperglycaemia has been suggested to be overproduction of superoxides by the mitochodria (125, 126). This implies that disruption of the superoxide overproduction potentially could normalize all four mechanisms at one time. In clinical practice, this is however hardly possible by conventional antioxidants, as attempted in the Heart Outcomes Prevention Evaluation trial (HOPE) and the Heart Protection Study (HPS), where low dose vitamin supplementation did not succeed in reducing neither risk for cardiovascular nor renal disease in diabetic patients (127, 128).

In man, the relation between glycaemic control on one side, and risk of development of microangiopathy on the other was reported by Pirart (129). A close correlation between level of hyperglycaemia and development of microalbuminuria in type 1 diabetic patients has been firmly established in subsequent observational studies (83, 90, 92, 130). In a meta-analysis on 16 smaller randomized trials of intensive therapy aiming at tight blood glucose control, Wang et al reported that intensive treatment was associated with reduced risk of nephropathy and retinopathy progression (131). A finding confirmed and extended by the DCCT trial comprising 1441 type 1 diabetic patients, where the beneficial effect of intervention aiming at strict glycaemic control as a means of preventing or delaying development of diabetic microvascular complications was clearly demonstrated (68). Intensive blood glucose control during 6.5 years reduced the occurrence of microalbuminuria (>40 mg/24 h) by 39% as compared to conventional therapy (68). The time, effort, and cost required in the DCCT were however considerable in order to achieve good glycaemic control, and the incidence of severe hypoglycaemia was three times higher in the intensive treated group (68). Whether the benefits from the DCCT persisted after the trial ended, was evaluated in the EDIC study (69, 132). The reduction in risk of microvascular disease in intensive group persisted eight years after end of trial; however the difference in glycaemic control narrowed during follow-up (during DCCT 7.2 vs. 9.1%, during EDIC 8.0 vs. 8.2%). The latter emphasize that substantial effort is needed to obtain and maintain excellent glycaemic control.

In our prospective study of an inception cohort, we demonstrate poor glycaemic control at onset of diabetes to be an important predictor for later development of microalbuminuria (89). This observation strongly support the conclusion from the EDIC study (69, 132), and implies that vigorous efforts to normalize glycaemia should be implemented from diagnosis of diabetes, and maintained as long as possible.

Dyslipidaemia

Abnormalities in the lipid metabolism may be of importance in the pathogenesis of glomerulosclerosis. Experimental induced hypercholesterolaemia in rabbits induces glomerular lesions (133), and pharmacologic reduction of serum lipids in Zucker rats did not only reduce serum cholesterol and fasting triglyceride levels, but was also associated with concomitant reduction in urine albumin excretion as well as mesangial matrix expansion and cellularity (134). Dyslipidaemia is a common feature even early in the course of diabetic nephropathy (135, 136). Histologic data in humans support the concept of a relation between dyslipidaemia and diabetic kidney disease, as glomerular lipid deposits have been demonstrated in kidney biopsies from patients with diabetic nephropathy (137). Type 2 diabetic patients with the metabolic syndrome including dyslipidaemia, have a higher prevalence microalbuminuria or macroalbuminuria as compared to type 2 diabetic patients without the metabolic syndrome (23 vs. 7%, p = 0.003) (138). Even in patients with type 1 diabetes, there is a strong, positive, and independent correlation between urinary albumin excretion rate and lipoprotein profile in the DCCT cohort (139), as well as in the European Diabetes Prospective Complications Study (EURODIAB) (140). In a short term observational study, low-density lipoprotein was found to be a significant independent predictor of development of microalbuminuria (130). Further longitudinal studies with repeated concomitant measurements of lipoprotein profile and urinary albumin excretion rate before and after development of microalbuminuria are needed to establish whether dyslipoproteinaemia is a predictor of development - or a consequence - of diabetic kidney disease. The association between dyslipidaemia and development of microalbuminuria has not yet been fully clarified.

7.3 Haemodynamic abnormalities

Glomerular filtration rate is determined by four factors; glomerular plasma flow, systemic oncotic pressure, glomerular transcapillary hydraulic pressure, and the glomerular capillary ultrafiltration coefficient (141). In animal models, renal vasodilatation causes an increase in single nephron GFR due to increase in glomerular plasma flow and glomerular transcapillary hydraulic pressure. In progressive kidney disease, the initial hit is suggested to be intraglomerular hypertension, soon followed by altered permselective properties of the glomerulus, increased protein filtration, protein accumulation in the mesangium, proliferation of mesangial cells, and matrix accumulation leading to glomerulosclerosis (34). In accordance, the two kidney Goldblatt hypertensive model indicates that systemic and local, intrarenal haemodynamic changes can induce the characteristic structural changes in diabetic animals (142). In humans, glomerular hyperfiltration occurs early in the course of diabetes, thus renal haemodynamic alterations associated with hyperglycaemia could initiate the process of glomerulosclerosis (34). Furthermore, the cardiac output and peripheral vasodilatation is increased during hyperglycaemia in diabetic patients1 (43, 144).The landmark studies performed by Brenner and co-workers did not only provide evidence of the importance of haemodynamic factors in the development of diabetic glomerulosclerosis. Moreover, the important role of reducing intraglomerular capillary pressure was emphasized using low protein diet and inhibitors of the renin angiotensin system (RAS) (145-147). The role of the RAS in the development and progression of diabetic nephropathy has been extensively evaluated during the last decades. However, plasma components of the RAS are low or normal in diabetes, except for the precursor of renin: prorenin. In normotensive type 1 diabetic patients, increase in prorenin has been reported even before development of microalbuminuria, and level of serum prorenin was found to be a predictor of development of microalbuminuria and diabetic retinopathy (148). In overt nephropathy, prorenin as well as renin have been shown to be elevated in plasma (149). All components of the RAS, including enzymes and receptors are present within the kidney, and most likely the RAS is activated locally within the kidney. The end product of the RAS, angiotensin II, has haemodynamic effects which include contraction of the efferent arteriole in the glomeruli. In addition, non haemodynamic effects of angiotensin II with importance for the development and progression of diabetic kidney disease, such as cell growth, proliferation and apoptosis through a range of pathways have been demonstrated (150, 151). Angiotensin II has also been demonstrated to induce extracellular matrix accumulation through stimulation of TGF- β (152). The haemodynamic and non haemodynamic effects can be hard to separate from each other, and a link between the deleterious effects of hyperglycaemia, haemodynamic changes with intraglomerular hypertension, and cytokines has now been established. The elastic and compliant property of the glomeruli allows pressure to regulate volume, and implies cyclic changes in the glomerular volume. These cyclic alterations may be aggravated by impaired autoregulation, which are often observed after longstanding diabetes (153), leading to downstream transmission of systemic pressure leaving end organs unprotected from the higher systemic blood pressure. Glomeruli will undergo expansive cycles of distension-contraction under conditions of systemic hypertension and impaired autoregulation. Cyclic stretch in tissue culture induces accumulation of extracellular matrix through stimulation of TGF- β , and extracellular matrix accumulation is markedly enhanced in presence of hyperglycaemia (154). Therefore, restriction of glomerular distension may be of utmost importance in preventing development and progression of glomerulosclerosis, however no effective treatment can improve glomerular pressure autoregulation (155). In contrast, it has recently been demonstrated that dihydropyridine calcium channel blockers may impair or even abolishes autoregulation in type 2 diabetic patients (156). In order to prevent and retard development of micro- and macroalbuminuria, reduction of systemic blood pressure and intraglomerular pressure is necessary, and agents reducing intraglomerular pressure may be advantageous in the prevention of diabetic nephropathy.

In humans, arterial hypertension is observed frequently in patients with diabetic nephropathy (157). Whether elevated blood pressure has a causative role and initiates development of nephropathy, or the development of nephropathy causes blood pressure to rise is not fully clarified. In cross sectional studies, some have demonstrated blood pressure to be a risk marker for development of microalbuminuria (92), while others have failed to find an association (83, 90). Our study suggests that systemic blood pressure elevations even at the onset of diabetes and within the normal range, play an important role in the development of persistent microalbuminuria (89). This is in accordance with the concept of the critical role of glomerular hypertension in the initiation of diabetic kidney disease (34), and further supported by the studies that demonstrate a familial predisposition to hypertension increases the risk of development of diabetic nephropathy (158, 159). Recently, nocturnal hypertension has emerged as a new risk factor for development of microalbuminuria (160). Suggesting that either changes in systemic blood pressure are very small, which makes their detection difficult unless carefully validated and precise methods are used, or arterial blood pressure has been measured at the wrong time (160), i.e. at daytime and not during the night.

In normotensive microalbuminuric type 1 diabetic patients a beneficial effect of treatment with an ACE inhibitor was originally demonstrated by Marre (161), and confirmed by others as reviewed by Mogensen (9). A meta-analysis of secondary prevention studies of diabetic nephropathy using ACE inhibitors conclude that in normotensive type 1 diabetic patients, ACE inhibitors significantly reduces progression to macroalbuminuria, and increases chances of regression to normoalbuminuria (45). The reduction in systemic blood pressure could not entirely explain the beneficial findings. Recently, two very important issues have been reported. Firstly, prevention of diabetic nephropathy through ACE inhibition has been demonstrated to be associated with long lasting preservation of kidney function (10), and secondly implementation of the knowledge acquired from the treatment trials of ACE inhibition in microalbuminuric patients is feasible in everyday life in a tertiary referral center (46). However, primary prevention using ACE inhibitors in normotensive normoalbuminuric patients have not yet been convincingly shown in trials. In a randomised placebo controlled trial in normotensive type 1 diabetic patients with normoalbuminuria or microalbuminuria, the ACE inhibitor Lisinopril was compared to placebo. At inclusion, more than 80% (n=440) of the patients had normoalbuminuria, and the study failed to show a statistical significant treatment difference (12.7% 95% CI: -2.9 to 26.0, p=0.1) (162). In a similar population of 89 type 1 diabetic patients, it was demonstrated that urinary albumin excretion can be significantly reduced by treatment with an ACE inhibitor with lowering of systemic blood pressure (163). However, no information of the impact of ACE inhibition on primary prevention of diabetic nephropathy, i.e. development of persistent microalbuminuria, was reported (163).

7.4 Growth factors and cytokines

Growth hormone and Insulin-like growth factor-1

The growth hormone (GH) hypothesis suggest that GH is one causal factor in the development of diabetic microangiopathy (164). The hypothesis was based on the findings that hypophysectomy improved diabetic retinopathy, that 24 hour levels of GH was found to be elevated in diabetic patients, and returned to normal levels after blood glucose normalization (164). Pituitary GH induce insulin-like growth factor-1 (IGF-1) in various organs. In diabetic animals, the rapid increase in kidney size and function after induction of diabetes is preceded by an increase in renal tissue IGF-I concentration, and strict glycaemic control through insulin treatment abolishes both the increase in kidney IGF-I and renal hypertrophy (165). In dwarf rats with isolated GH deficiency, a smaller degree of renal and glomerular enlargement was observed, and an attenuated rise in albuminuria was observed as compared to diabetic rats with normal functioning pituitary gland (166). The role of GH and IGF-1 in the early changes of diabetic glomerulopathy is further supported by animal studies, where treatment with a long acting somatostatin analogue for 6 months was associated with significant reductions of increase in kidney weight, kidney IGF-1 concentrations and urinary albumin excretion rate compared with untreated diabetic rats (167). Treatment with a GH antagonist in diabetic mice showed comparable results (168). A cross sectional study of 155 children and adolescents with type 1 diabetes demonstrated a relationship between kidney weight and IGF-1 (169). In addition, the 14 patients with microalbuminuria had significantly higher levels of urinary IGF-1 and GH as compared to the normoalbuminuric patients (169). In diabetic patients with pre proliferative retinopathy, a long-term multicenter trial on the effect of the somastatin analogue Sandostatin on progression to proliferative retinopathy is at present time ongoing. Long-term prospective studies to investigate the role of GH in humans on development and progression of diabetic nephropathy are lacking, as well as treatment trials to establish the effect of intervening in the GH/IGF-1 system on development and progression of diabetic kidney disease.

Cytokines

The cytokines constitute a group of small potent regulatory peptides with multiple and often overlapping biological activities. They have been implicated in numerous biological processes in the kidney including cell hypertrophy, proliferation, regulation of matrix synthesis and degradation, inflammatory responses, and regulation of vascular tone (170). Cytokine production is usually transient and tight regulated, and the half life of cytokines in the circulation ranges from hours to a few days. High affinity receptors are expressed on target cells in many organs, and the various cytokines interacts in a complex and not fully clarified way. The investigation of the role of cytokines in development and progression of diabetic glomerulopathy is of interest since these peptides potentially could serve as early markers of development of disease, be targets for intervention, and markers used for monitoring efficacy of other treatment modalities. Transforming growth factor β (TGF- β ) is a prosclerotic cytokine family, the major isoform TGF- β 1 is considered to be the major mediator of collagen formation in the kidney (171). In addition, TGF- β 1 inhibits the production of matrix metalloproteases, stimulates the synthesis of metalloprotease inhibitors, thereby having an effect on matrix degradation pathways (171). The TGF- β expression increases as a response to a wide range of stimuli including high glucose concentration, AGE's, oxidative stress and superoxide overproduction, cyclic stretch, and high levels of intrarenal angiotensin II as reviewed by Ziyadeh (172). These factors are all relevant for the development and progression of diabetic nephropathy. TGF- β has been found to be over-expressed in the kidneys of animals as well as humans with diabetic nephropathy (173), and treatment with TGF- β antibodies reduces extracellular matrix production in diabetic mice (174). TGF- β therefore seems to be involved in the progressive matrix accumulation and fibrosis observed in diabetic glomerulopathy, and moreover to be an important link between the haemodynamic and metabolic pathways in the initiation of diabetic nephropathy. The synthesis of another prosclerotic cytokine, connective tissue growth factor (CTGF) is stimulated by TGF- β , hyperglycaemia and cyclic mechanical stretch (175). In experimental diabetes, CTGF is increased in glomeruli of diabetic animals, and CTGF has been reported to be elevated early in the course of diabetic glomerulopathy (175). A correlation between level of urinary albumin excretion rate and CTGF has been shown in humans (176). However, the role of CTGF in the development and progression of diabetic nephropathy has not yet been studied to a large extent in humans, and prospective studies are lacking.

Vascular endothelial growth factor (VEGF) is a potent cytokine family that induces angiogenesis and markedly increases endothelial permeability (121). In diabetic microangiopathy, the hallmark of endothelial dysfunction is an increase in permeability, and VEGF has been found to be elevated in ocular fluid in diabetic proliferative retinopathy, with the level of VEGF declining after successful laser photocoagulation (177). In experimental diabetic kidney disease, VEGF expression is increased, predominantly in the glomerular podocytes (178). In 110 type 2 diabetic patients, plasma VEGF was higher among patients with micro- and macroalbuminuria as compared with the normoalbuminuric patients, with a tendency towards increasing levels of VEGF with increasing levels of albuminuria (179). In a large case-control study of type 1 diabetic patients with and without diabetic nephropathy, we found that plasma VEGF was significantly elevated early in the course of diabetic nephropathy in male subjects (180). Since a large inter-individual variability in plasma levels of VEGF was observed in our study, it is likely that VEGF and other cytokines might act predominantly as paracrine hormones in diabetic renal disease. In accordance, elevated VEGF levels in the vitreous fluid in eyes of patients with proliferative diabetic retinopathy has been reported to be produced locally by the ischemic retina, whereas circulating VEGF was not elevated (181). Furthermore, platelet derived growth factor have been implicated in the development of glomerulopathy and tumor necrosis factor- α have recently been suggested to play a role for the development of microalbuminuria (182). In the latter case, only kidney tissue and urine samples showed positive results, emphasizing the need for sampling of material close to the organs of interest. In order to establish new early markers for development of diabetic nephropathy in high risk patients, long-term studies with repeated measurement of urine samples during the course of the disease are needed. Only hereby, evaluation of the complex interplay of the various cytokines and the development of diabetic nephropathy can be performed.

7.5 Other factors associated with
development of diabetic nephropathy

Glomerular hyperfiltration (35, 38), low birth weight (183), short stature (184), presence of retinopathy (64, 83), gender and onset of diabetes at young age (185) have been demonstrated to be associated with development of diabetic nephropathy. A thoroughly discussion of these factors is beyond the scope of this thesis.

8.1 Risk factors for progression

8.1.1 Genetics

Along with the extensive investigation of genetic risk factors for development of diabetic kidney disease, an understanding of a genetic influence on progression of established diabetic and non diabetic nephropathy has evolved. In a prospective follow-up study of 35 type 1 diabetic patients with diabetic nephropathy, a significant steeper decline in GFR was found in patients with the DD genotype of the ACE I/D polymorphism during antihypertensive treatment with an ACE inhibitor (193). Despite similar arterial blood pressure and albuminuria during follow-up, the patients with the DD genotype had a rate of decline in GFR of 5.7 vs. 2.6 ml/min/year in patients with the II/ID genotypes. In a three years study where patients and investigators were blinded to ACE genotypes, the effect of blocking of the angiotensin II receptor with losartan was investigated on decline in kidney function in a similar population (194). In contrast to the study using ACE inhibitors, treatment with losartan had similar beneficial renoprotective effects on progression of diabetic nephropathy in patients with ACE II and DD genotypes (194). Recently, we have demonstrated a deleterious impact of increasing number of D alleles on time to doubling of serum creatinine or ESRD during ACE inhibitor treatment, with a risk ratio of 1.81 (95% CI: 1.09 to 3.03) per additional D allele (195). Furthermore, a model of several so-called "bad alleles" within the RAS yielded further evidence of gene-gene interaction on renal endpoints. Although many non genetic factors may influence the effects of medications, there is now growing evidence for genetic factors, as well as the frequency of side effects, being responsible for individual responses to therapy. The concept of pharmacogenetics, defined as the study of the role of genetics in drug response is by no means a new idea (196). Inter-individual differences in response to medication are dependent on variability in pharmacokinetics, i.e. absorption of the drug, distribution within different compartments, metabolism and excretion, and differences in pharmacodynamics. It is estimated that genetics can account for 20 to 95 percent of variability in drug deposition and effects (197). As reviewed by Evans, there are now examples of inter-individual differences in drug responses due to sequence variants in genes encoding drug-metabolizing enzymes, drug transporters, or drug targets (198). In most cases however, interplay between several gene products are involved in the response to a certain drug, and the determinants of this complex drug response are polygenic (198). Genetic factors of importance for the progression of diabetic nephropathy could possibly differentiate between responders and non responders. One of the main objectives of pharmacogenetics is to establish a more individualized drug therapy, beyond the "one size fits all" approach (199). Unfortunately, DNA for genotyping was only available in a subset of the investigated patients in our studies. With increasing knowledge, genotyping of study participants may be routinely carried out as an extension of the usual baseline information in clinical studies to adjust for these as covariates, to reduce selection bias and increase the effect of randomized treatment. In the future, relevant testing for genetic variations may be used to individualize drug therapy and thereby select the best drug at the optimal dose for each individual patient.

The influence of genetic susceptibility on progression of diabetic nephropathy seems at present to be very complex, and is complicated by gene-gene and gene-environment interactions. To assess progression of nephropathy, new large scale prospective studies are needed. Information from the new techniques with high throughput genotyping may help to shed light on this important issue. Identification of specific genotypes, or more likely a combination of these, that predispose to a rapid decline in GFR may help to target early aggressive intervention of modifiable risk factors in susceptible patients.

8.1.2 Blood pressure

Increased intraglomerular capillary hydraulic pressure plays a major role in the development of diabetic nephropathy. Does the increased intraglomerular pressure only initiate glomerulopathy as a necessary initial hit, or continuously enhance the deterioration of kidney function during the disease? The importance of intraglomerular pressure on progression of glomerulopathy was emphasized when prevention of progression was achieved by reducing the systemic and intraglomerular pressure in diabetic animals (200). Long-term treatment with an ACE inhibitor kept intraglomerular pressure normal and prevented a rise in albuminuria as opposed to conventional antihypertensive treatment, where progression in urinary albumin excretion rate was only postponed.

In one of the studies of the natural history of diabetic nephropathy, arterial blood pressure at the end of follow-up was correlated to the preceding decline in GFR before antihypertensive treatment was commenced (49). In an observational study in type 1 diabetic patients with diabetic nephropathy on antihypertensive treatment, Rossing et al found arterial blood pressure to be correlated to the decline in GFR (192), and a retrospective analysis of 58 type 1 diabetic patients with nephropathy reported an independent role of arterial blood pressure on decline in kidney function (201). We evaluated progression promoters in 301 consecutive type 1 diabetic patients with diabetic nephropathy, using annual measurement of plasma clearance of 51 Cr-EDTA. In our long-term prospective observational study with 7 years of follow-up, we confirm and extend the observations that arterial blood pressure is a progression promoter independent of albuminuria, glycaemic control expressed as HbA 1c , and serum cholesterol (28). The relationship between arterial blood pressure and diabetic nephropathy seems to be rather complex. Elevated blood pressure play a major role in the initiation and acceleration of progression of kidney disease, and nephropathy is closely associated with the increase in blood pressure observed in these patients. Impaired autoregulation in long-standing diabetes may aggravate the intraglomerular hypertension and thereby promote progression, as systemic blood pressure can be transmitted without restraint downstream to the glomerulus (202).

8.1.3 Albuminuria

As previously discussed, the initial hit in progressive kidney disease is suggested to be intraglomerular hypertension leading to altered permselective properties of the glomerulus (34). The mechanisms responsible for the abnormal urinary excretion of proteins in glomerular diseases are first of all a loss of charge and size selectivity of the glomerular capillary wall, leading to the transglomerular passage of albumin and other high molecular weight proteins, that usually do not cross the glomerular barrier (47, 215). Due to the increased workload and/or the toxic effect derived from the increased load of the abnormally amount of filtered proteins in the tubular lumen, the reabsorption will subsequently be impaired leading to augmented albuminuria (215). The quantity as well as the molecular weight and radius of the proteins that reach the tubular lumen increase progressively with increasing severity of the disruption of the glomerular capillary wall. Whether proteinuria is a simple marker of the extent of the glomerular damage or may by itself promote progression of glomerulosclerosis has not been fully clarified. However, animal studies have found that proteinuria may precede the structural glomerular epithelial cell changes (216). Other experimental studies have provided evidence for pathways where increased protein traffic across the glomerular capillary wall may promote glomerulosclerosis (217). Recently, it has been suggested that proteins are degraded during renal passage and excreted in protein fragments even in healthy persons (218). A contributing factor to increased levels of intact albumin in urine observed in kidney diseases may therefore be promoted by metabolic events that affect the degradation of proteins during the passage through the kidney, a cellular defect distal to the glomerular basement membrane. These observations needs to be confirmed, and thorough evaluation of the clinical relevance are required. At present, it is not known whether some proteins or protein fragments are more harmful than others, but future analyses of urine samples of proteins and their degradation products - so called proteomic analyses - may shed light on this matter.

Watkins originally demonstrated that diabetic patients with the highest level of albuminuria have the poorest prognosis (12).

In non-diabetic kidney disease, the MDRD study evaluated target blood pressure for renoprotection in relation to the level of albuminuria (219). It was demonstrated that the severity of proteinuria determined the optimal blood pressure target for renoprotection. In patients with 1 to 3 grams of proteinuria per day a target of mean arterial blood pressure of 98 mm Hg was suggested, whereas additional benefit in patients with >3 grams proteinuria per day of lowering mean arterial blood pressure to 92 mm Hg was found (219). Differentiated targets for antihypertensive treatment based on the level of albuminuria is now reflected in some (220), but not all (221-223) of the latest guidelines on treatment of hypertension. The renoprotective effect of lowering of albuminuria was confirmed in the Ramipril Efficacy In Nephropathy trial in non diabetic kidney disease (224). In diabetic nephropathy, intervention that has ameliorated the progression of diabetic nephropathy has always been associated with a reduction in proteinuria in type 1 diabetes (74, 186, 187, 189, 225, 226). We have confirmed and extended these observations in our study (28). Dividing the 301 patients in quintiles based on independent progression promoters revealed a linear relationship between rate of decline in GFR and arterial blood pressure, HbA 1c , and serum cholesterol ( Figure 2 ). However, the correlation between rate of decline in GFR and albuminuria seemed to be curve linear. A threshold for albuminuria on the rate of decline in GFR was found at 436 μ g/min, p<0.01 (approximately 600 mg/24 h), whereas none of the other progression promoters showed a significant threshold (28).

As we used mean of all values of albuminuria during follow-up, it is however not possible to evaluate whether patients with low level of albuminuria per se have a slow deterioration in kidney function in our study population or that a reduction in albuminuria in fact will lead to an improved prognosis. However, it has previously been demonstrated that the initial reduction in albuminuria is a useful clinical guideline predicting a beneficial effect of antihypertensive treatment on the long-term decline in kidney function in diabetic and non-diabetic nephropathies (22-24).

Our findings in type 1 diabetic patients are in accordance with the concept of deleterious effects of increased protein trafficking through the mesangium. However, the beneficial outcome in patients with lower levels of albuminuria might well be explained by improvement in the filtration barrier or lowering of the intracapillary hydraulic pressure. The effect of a reduction in albuminuria during follow-up has been analyzed thoroughly in three of our subsequent studies and will be discussed in section 9.

8.1.4 Metabolic alterations

8.1.5 Protein intake

In our study of progression of diabetic nephropathy, the patients had a diabetic diet containing 45-55% carbohydrates, 30-35% fat, and 15-20% protein, and no sodium or protein restrictions were applied during the study (28). The protein intake was not assessed in our study, but other observational studies in type 1 diabetic patients with diabetic nephropathy have not found dietary protein intake to be associated with the rate of decline in GFR (232, 244). Protein restriction in diabetic rats leads to normalisation of intraglomerular hydraulic pressure, prevents development of albuminuria, and glomerular damage normally observed in these animals1 (46, 245). The beneficial effects of a low protein diet found in diabetic animal models, have not been reproduced convincingly in humans with various kidney diseases. During 2.2 years of follow-up in the MDRD study of predominantly non diabetic kidney diseases including 585 patients with baseline GFR between 25 and 55 ml/min/year/1.73 m 2 , the mean rate of decline in GFR was 1.2 ml/min/year lower in the low protein diet group as compared to the normal protein diet group (p=0.30) (246). In a meta-analysis by Pedrini et al, several smaller studies including type 1 diabetic patients were summarized (247). This meta-analysis demonstrated that a low protein diet significantly slowed the increase in urinary albumin level or the decline in GFR or creatinine clearance (247). However, due to dubious designs, methodology, and insufficient adjustment of other progression promoters in the five studies including 108 type 1 diabetic patients included in the meta-analysis, the results should be interpreted with caution. In a four year prospective controlled trial of 82 type 1 diabetic patients with progressive diabetic nephropathy assigned to either low or normal protein diet, the rate of decline in GFR was reduced significantly from pre study progression rates in both the low and the normal protein diet (248). No difference in the rate of decline in GFR between groups during the study (3.8 vs. 3.9 ml/min/year, NS) was found, and albuminuria remained stable in both groups which might be ascribed to the modest difference in protein intake obtained during follow-up: 0.89 (95% CI: 0.83 to 0.95) in the low protein diet group vs. 1.02 (95% CI 0.95 to 1.10) g/kg/24 h in the normal protein diet group, p = 0.005 (248). A relative risk of the combined endpoint of ESRD or death was found to be 0.23 (95% CI: 0.07 to 0.72), p = 0.01 in patients in the low protein diet group as compared to the normal protein diet group (248). Even though the mechanisms for this favourable outcome of protein restriction is yet unknown, similar findings have been reported in non diabetic nephropathies (249). In conclusion, low protein diet is associated with improved survival free of ESRD in patients with diabetic nephropathy.

8.1.6 Smoking

Several prospective observational studies have found an association between smoking tobacco and progression from normoalbuminuria to microalbuminuria and overt nephropathy (64, 66, 81-83). The mechanisms for the nephrotoxic influence of smoking on the kidneys are not well understood. The smoking induced renal damage have been suggested to be caused by sympathetic activation influencing systemic arterial blood pressure and renal haemodymics, endothelial dysfunction due to reduced nitric oxide availability, and diminished vasodilatation. Activation of the RAS through induction of increased renin production stimulated by activation of the sympathetic nervous system and release of catecholamines have also been implied as one of the pathogenic mechanisms (250). Smoking of tobacco is a well established risk factor for cardiovascular, respiratory and cancer diseases, and the evaluation of smoking as a risk factor for progression of overt diabetic kidney disease seem somewhat academic, and without consequences in the counselling and the care of the diabetic patients in everyday life. However, when evaluating potential risk factors for progression of kidney disease in order to be able to explain the huge variability in the rate of decline in GFR among patients, evaluation of smoking as a progression promoter is highly relevant. Some prospective studies have suggested cigarette smoking to be a risk factor for progression of diabetic nephropathy. In a one year study of 93 type 1 diabetic patients, the impact of smoking on progression of kidney disease was assessed using creatinine clearance, and smoking was suggested to worsen kidney function (251). However, since arterial blood pressure (143/84 vs. 137/85 mm Hg) as well as the level of albuminuria (2547 vs. 1592 mg/24 h) was higher in smokers as compared with non smokers, a causal role can not be established. Others have found that type 1 diabetic patients with diabetic nephropathy smoking more than 10 cigarettes per day (n=8 patients), have a higher rate of decline in creatinine clearance than non smokers, 14.9 vs. 10.3 ml/min/year (252). Again, arterial blood pressure was significantly higher in smokers compared to non smokers, 159/89 vs. 141/82 mm Hg, and levels of albuminuria was not presented (252). In a study by Mühlhauser et al, only 26 patients (4% of the study population) had macroproteinuria or elevated serum creatinine at baseline and no separate analyses on progression of diabetic kidney disease in these patients was performed (66). The studies performed so far are characterized by inadequate methods for measuring rate of decline in GFR, short duration of follow-up in small number of patients, and insufficient adjustment for other well established progression promoters (66, 251, 252). Therefore, we analyzed data from our large long-term prospective observational study of 301 type 1 diabetic patients with diabetic nephropathy (253). In our study, we followed the patients for at least three years with annually GFR measurements with a valid method (253). Patients were classified as smokers, ex smokers and non smokers. Smokers being defined as patients smoking more than 1 cigarette per day during a part or the whole observation period. No difference in the rate of decline in GFR was demonstrated between non smokers (n=94): mean (95% CI) 4.5 (3.7 to 5.4), ex smokers (n=31): 3.1 (1.6 to 4.7), and smokers (n=176): 3.9 (3.3 to 4.5) ml/min/year respectively (NS) (253). During follow-up, the smokers had significantly lower blood pressure as compared to the non and ex smokers, while no difference in the level of albuminuria, HbA 1c , or serum cholesterol was apparent. Adjustment for the difference in arterial blood pressure among groups did not alter the results significantly, with rate of decline in GFR: 4.1 ml/min/year in non smokers, 3.1 ml/min/year in ex smokers, and 4.1 ml/min/year in the smoking group respectively, NS ( Figure 3 ). The definition of smoking status varies between studies. Some use a cut-off at ten cigarettes per day (252), other have used the definition of pack years ([mean number of cigarettes smoked per day/20] × number of smoking years) (66, 251). Due to assessment of smoking habits using questionnaire with the possibility of underreporting, we used the most conservative definition of smoking: smoking more than 1 cigarette per day. However, we analyzed the impact of heavy smoking versus light smoking by dividing our patients according to the median number of cigarettes smoked per day (median number of cigarettes smoked was 20). We did not find a difference in rate of decline in GFR between heavy and light smokers (253). In conclusion, data from our long-term prospective observational cohort study do not support that deterioration in kidney function in type 1 diabetic patients suffering from diabetic nephropathy is associated with tobacco smoking. The relative mortality risk for smoking diabetic patients are substantially higher than for non smoking patients (254, 255). Even in the absence of convincing data supporting the concept that smoking accelerates renal disease progression in overt diabetic kidney disease in type 1 diabetes, cigarette smoking remains a hazard to health. The results from our study can by no means be taken in to account for a general advice to diabetic patients with overt nephropathy who currently smoke to carry on with this risky habit. Conversely, our data can indeed be used to urge for the continued search for new yet unknown progression promoters, as we at present are not able to explain all the variation in renal disease progression from the current knowledge.

In summary, several potential modifiable risk factors for decline in kidney function have been identified. To illustrate the impact of a combination of risk factors for loss of renal function, a two hit model with mean arterial blood pressure (an established modifiable progression promoter) combined with albuminuria, HbA 1c , and serum cholesterol is shown in Figure 4 .

Patients with diabetic nephropathy encounter a continuing loss of renal function and have in addition an increasing risk of cardiovascular disease. The main risk factors for micro- as well as macrovascular disease are closely interrelated in the diabetic patients. For some treatment modalities the evidence for a renoprotective effect is lacking, whereas the effect on macrovascular disease is beyond doubt. Clearly, to improve the prognosis in the diabetic patients with nephropathy, the treatment and care of the patients must deal with all aspects of the disease.

8.2 New predictors of progression
of diabetic nephropathy

The rate of loss in renal function is extremely variable once overt diabetic nephropathy has evolved. Information of the impact of progression promoters during the course of the disease is essential in every day treatment of the patients to postpone, or even prevent, the progression to ESRD. In a study of 18 type 1 diabetic patients with diabetic nephropathy followed for 10 years and treated with ACE inhibitors, two thirds (66%) of the variation of the decline in kidney function was explained by albuminuria and HbA 1c during follow-up (232). In the Captopril Collaborative study with a median follow-up time of three years including 409 type 1 diabetic patients with diabetic nephropathy, baseline progression promoters were evaluated (256). Onset of diabetes later in life, parental diagnosis of type 1 diabetes, abnormal electrocardiogram, elevated mean arterial blood pressure, high blood glucose, increased serum creatinine, enhanced proteinuria and presence of oedema were found to predict doubling of serum creatinine independently (256). In our study of 301 type 1 diabetic patients with kidney disease, we were only able to explain 29% of the variation of the rate of decline in GFR by the impact of arterial hypertension, albuminuria, glycaemic control, and dyslipidaemia (28). Inclusion of genetic risk markers from the RAS as risk factors for decline in renal function in the statistical analysis in 169 type 1 diabetic patients with diabetic nephropathy, treated with ACE inhibitors and followed for 6 years, did not improve the degree of explanation of the variation considerably (195). In the search for new potentially modifiable risk factors for progression of diabetic nephropathy, we have analyzed the impact of plasma VEGF, plasma total homocysteine (tHcy), and plasminogen activator inhibitor-1 (PAI-1), measured at a defined baseline (180, 257). Of the 199 type 1 diabetic patients enrolled in the Steno case-control study, 157 were followed for at least 3 years with yearly GFR measurements from baseline examination in 1993 until 2001. The three years of follow-up was required to be able to make a valid determination of the rate of decline in GFR in the individual patient. Vascular endothelial growth factor has been suggested to play a role in development and progression of diabetic microangiopathy, and it was therefore an apparent candidate as a risk factor for nephropathy progression. Vascular endothelial growth factor was found to be elevated in men with nephropathy, but quite disappointing, no correlation was found between VEGF and the rate of decline in GFR (180). Hyperhomocysteinaemia is an independent risk factor for atherosclerosis, and evidence is emerging that homocysteine induces endothelial dysfunction by promoting oxidative damage (258). Since there are pathophysiological similarities between glomerulosclerosis and atherosclerosis, risk factors for these processes may be identical. Plasma tHcy has been positively correlated to the level of albuminuria in both type 1 and type 2 diabetes (259-262), although an independent association has been questioned (263). Homocysteine becomes elevated when kidney function is impaired, however it remains unclear whether the elevated tHcy is due to impaired metabolism or to reduced excretion in these patients (258). In non diabetic rats, elevated tHcy has been implicated in the pathogenesis of glomerular damage independent of arterial blood pressure (264). In our study of progression of diabetic nephropathy, a borderline significant correlation between tHcy and the rate of decline in GFR was demonstrated (257). A significant association between tHcy and decline in GFR was found when analyzing tHcy divided in tertiles, suggesting a non-linear relationship between tHcy and the outcome (257). However, tHcy was not found to be an independent predictor of the rate of decline in GFR (257). In accordance with our findings, studies in non diabetic nephropathies have failed to establish an association between the level of tHcy and decline in kidney function (265, 266).

Plasminogen activator inhibitor-1 has a critical inhibitory role in the regulation of intravascular fibrinolysis in addition to involvement in tissue repair and remodelling, and is directly induced by angiotensin II (267). A shift in the delicate balance of proteases and their inhibitors occurs during matrix expansion in the kidneys, where PAI-1 is suggested to be one of several modulators. In an animal study by Fogo et al, mRNA overexpression of PAI-1 was found to be localized to injured mesangial and endothelial areas, levels of PAI-1 were increased in sclerotic glomeruli, and the upregulation of PAI-1 was inhibited by treatment with drugs blocking the RAS (268). In type 1 diabetes, we have previously demonstrated that plasma PAI-1 is elevated in men with diabetic nephropathy as compared with normoalbuminuric diabetic men (269). However, no association between PAI-1 and decline in kidney function was established, and neither tHcy nor PAI-1 were however demonstrated to be independent predictors of decline in GFR (257). Thus emphasizing the interrelationship between many risk factors often confounds the relative significance of individual risk factors, and that the "classical" well established risk factors are relatively powerful as such, and always must be taken into account when evaluating new risk factors for nephropathy progression. The lack of evidence in diabetic patients for plasma tHcy or PAI-1 levels as predictors of progression of renal disease can have several explanations. Firstly, it is not clear whether circulating markers, in our study tHcy and PAI-1 levels, in fact does reflect the local expression within the kidney. Secondly, in our study the majority of patients receiving antihypertensive treatment were treated with ACE inhibitors, which may inflect on the level of PAI-1. Furthermore, it is possible that PAI-1 expression play a major role in the initiation of diabetic glomerulosclerosis, whereas the effect on progression of overt diabetic kidney disease could be more modest. Lastly, it must be recognized that results obtained from in vitro studies or animal models can not always be reproduced in man.

Meanwhile, the search for new powerful predictors of progression of diabetic kidney disease continues. Anaemia is frequently observed when renal function deteriorates, and have recently been suggested to be involved in the progressive decline in GFR in type 2 diabetic patients (270). Furthermore, the role of inflammation (271), asymmetrical dimethylarginine which is an endogenous inhibitor of nitric oxide synthase (272), elevated uric acid (273), and increasing levels of aldosterone during RAS blockade (Abstract: Schjoedt et al, Journal of the American Society of Nephrology, Vol. 14, page 7A, 2003) has been brought in to focus.

9.1 Regression of microalbuminuria
to normoalbuminuria

In order to prevent overt diabetic nephropathy, investigators have searched for early predictors for development of the disease. During the early 1980s, urinary albumin excretion rate below the range detectable by standard laboratory methods was introduced as an early and robust predictor of subsequent risk for the development of overt diabetic nephropathy. In a consensus report, the term microalbuminuria was defined as urinary albumin excretion rate ranging from 30 to 300 mg/24 h, equal to 20 to 200 μ g/min (8). A suitable predictor for development of diabetic nephropathy should be identifiable early enough in the disease process for timely and effective intervention to be applied. Furthermore, those subjects at high risk for development of renal complications should be identified. In type 1 diabetic patients, the level of microalbuminuria, and the risk of development of overt nephropathy, can be modified by strict glycaemic control and the use of ACE inhibitors (45, 91).

In the initial observational studies, development of overt nephropathy in microalbuminuric type 1 diabetic patients was found to occur in approximately 80% of the patients during 6 to 14 years of follow-up (41-43). However, the validity of microalbuminuria as a predictor for development of macroalbuminuria has recently been questioned, and regression from microalbuminuria to normoalbuminuria suggested to be a frequent phenomenon (274-276).

In a review, Caramori (274) has estimated the rate of progression from microalbuminuria to overt nephropathy to be 30% during the following 5 to 10 years. However, among several inclusion criteria for studies to be included in this review, a follow-up time from baseline examination of more than five years in the studies were required. Thus, many randomized trials lasting from 2 to 4 years were excluded, introducing selection bias (274). Except for one trial (277), the rate of progression to macroalbuminuria in the placebo groups in the excluded trials were considerably higher, ranging from 5.5 to 13%/year (70-72, 162, 278, 279). Despite the use of ACE inhibitors as secondary prevention strategy in the majority of the microalbuminuric patients in our inception cohort, nearly half of the patients progressed further from microalbuminuria to overt nephropathy during 7.5 years (89), a rate approaching that of the short term trials.

Spontaneous regression of microalbuminuria, i.e. without secondary intervention aiming at reducing the level of urinary albumin excretion rate, has recently been reported to occur frequently in type 1 diabetic patients (275, 276). A cumulative incidence of regression of microalbuminuria, defined as reduction of 50% or more in the urinary albumin excretion from one two-year period to the next, was reported to be 58% during 6 years of follow-up (276). Since the use of ACE inhibitors was not associated with regression to normoalbuminuria, the data suggests spontaneous normalization. This is based on a study of 386 type 1 diabetic patients with microalbuminuria followed for a median of 4.4 years with a drop out rate of 43% (276). Urinary albumin excretion rate was measured in random spot urine samples and recalculated to mg of albumin excreted per minute. In our inception cohort of type 1 diabetic patients followed from onset of diabetes, 79 patients developed microalbuminuria (89). The median follow-up time after onset of microalbuminuria was 7.5 years, the drop out rate was 14%, and data until last visit in the clinic were ascertained. To reduce the confounding influence of variability in urinary albumin excretion rate, we defined regression to normoalbuminuria from microalbuminuria as an urinary albumin excretion rate below 30 mg/24 h in two out of three consecutive 24 hour urine collections, and a decrease of at least 30% from the microalbuminuria level, sustained for at least one year (89.) Regression to normoalbuminuria occurred in 35% of the microalbuminuric patients, with more than half of these subsequently relapsing to persistent microalbuminuria. The 6 year cumulative incidence of regression was 31% (95% CI: 20-42) (89). Regression occurred in 28 patients of the 79 patients, and only 13% developed spontaneous permanent regression to normoalbuminuria, i.e. without antihypertensive treatment (89). None of the patients developing macroalbuminuria spontaneously regressed to microalbuminuria or normoalbuminuria. The difference in the proportion of patients who regress in the two studies may partly be explained by differences in design, as the study by Perkins et al (276) evaluated a prevalence cohort, introducing the possibility of inclusion, exclusion and survival bias. In conclusion, microalbuminuria is at present the best documented non invasive predictor of development of overt diabetic nephropathy in type 1 diabetes. The presence of microalbuminuria in most cases does imply an unalleviated progressive kidney affection.

9.2 Remission and regression
of overt diabetic nephropathy

Implicit in the term ESRD is the understanding of a sequence of injuries on the kidney, starting with an initial insult, and ending with total loss of kidney function. The historical studies suggest that progression is inevitable, but major changes in the treatment of the patients have been implemented since then (49-51). Are the processes involved in the deterioration of kidney function in diabetic kidney disease strictly progressive or can the course of the disease be reversed? In spontaneously hypertensive 5/6 nephrectomized rats, tight blood pressure control decreased the apoptosis rate, mainly in tubular cells (280). In other animal studies using high doses of ACE inhibitors, remodelling of vascular sclerosis, tubulointerstitial fibrosis, and existing glomerulosclerosis is possible in chronic kidney disases (210, 211). Thus regression of biopsy proven glomerulosclerosis can be achieved in animal models of chronic nephropathies. Could the same be the case in humans? In patients with non diabetic nephropathies, long-term ACE inhibition induced preservation and even improvement in GFR (281). When normalizing the diabetic milieu, reversal of morphological lesions of glomerulopathy can be achieved, as demonstrated by performing transplantation of kidneys with diabetic structural lesions in to non diabetic recipients with subsequent resolution of mesangial expansion (282). Furthermore, approximating normalization of blood glucose levels over a long period of time by pancreatic transplantation could reverse renal morphological lesions in a small number of patients with urinary albumin excretion ranging from normal to clearly abnormal levels (233). The latter two studies in humans must be regarded as casuistic and as a "proof of concept", as these treatment modalities are not widely applicable in everyday life. Systemic blood pressure as well as local intraglomerular hydraulic pressures are key factors in the vicious circle of progressive renal damage and progressive injury. However, many other components have been implicated in the initiation and progression of diabetic nephropathy, for instance proteinuria, glycaemic control, reactive oxygen species and cell-specific injury, growth factors and cytokines. To achieve regression of prevailing sclerosis, the dynamic control of cell proliferation, apoptosis and regeneration must be altered, and matrix degradation exceeds matrix synthesis. Theoretically, endothelial cells must regenerate, and mesangial cells and new capillary loops within the glomeruli must regrow. Finally, the shift in balance of proteases and their inhibitors occurring during matrix expansion must be reverted to induce reabsorption of excess matrix deposits. The data from experimental studies have shown that this is feasible in animal models, and casuistic reports in humans are available. We have evaluated the concept of remission and regression in our long-term prospective study of 301 type 1 diabetic patients suffering from diabetic nephropathy (283). Since no general accepted definitions for remission and regression of diabetic nephropathy yet had been established, we applied very conservative and strict criteria. Remission of diabetic nephropathy was defined as going towards normalization of the disease process: decrease in albuminuria to the microalbuminuric range. Regression was defined as normalization in the rate of decline in GFR: equal to or less than 1 ml/min/year during the entire observation period, equivalent to the natural ageing process in non-diabetic patients without kidney disease (25). When applying these definitions in our study, 31% obtained remission, and 22% achieved regression (283). In patients obtaining remission, the rate of decline in GFR was significantly lower (2.2 ml/min/year) as compared with the no remission group (4.8 ml/min/year), p<0.001. The association between prevalence of remission and arterial blood pressure is shown on Figure 5 , and the association between prevalence of regression and arterial blood pressure is shown on Figure 6 . Besides the expected lower rate of decline in GFR, mean arterial blood pressure, albuminuria, HbA 1c , and serum cholesterol during the observation period were lower in patients achieving regression of diabetic nephropathy. Arterial blood pressure, albuminuria, and HbA 1c were independently associated with regression: adjusted odds ratio [95% CI] for a 10 mm Hg decline in blood pressure: 2.14 (1.33-3.44), a tenfold lowering of albuminuria: 2.79 (1.35-5.69), and a reduction of 1% in HbA 1c : 2.00 (1.46-2.73). In the study population were 30 so-called genuine normotensive patients, i.e. normotensive patients not prescribed any antihypertensive medication. These patients had a significantly lower rate of decline in GFR as compared to the hypertensive patients, 1.9 ± 0.5 versus 4.3 ± 0.3 ml/min/year respectively, p<0.01. We did not find any association between the use of ACE inhibitors and remission or regression of diabetic nephropathy. In clinical trials, an additional non haemodynamic renoprotective effect of ACE inhibitors has been demonstrated (189, 190). However, in observational studies such as ours, confounding by factors associated with both treatment and outcome can not be excluded (284). In addition, the dosing of ACE inhibitors was titrated on the basis of arterial blood pressure, since no optimal dosing for renoprotection of ACE inhibitors has yet been ascertained. In conclusion, aggressive antihypertensive treatment in type 1 diabetic patients can induce remission and regression in a sizeable fraction of patients with diabetic nephropathy. Several risk factors for progression of diabetic nephropathy also seem important for regression of diabetic nephropathy in type 1 diabetic patients. Thus, we have demonstrated that the functional changes can be reversed; however, kidney biopsies were unfortunately not available from our study population. In modern chemotherapeutic strategies aiming at eradication of malignancy, multiple agents are used to bring the patients "in remission". A similar comprehensive renoprotective approach in the treatment of diabetic nephropathy may be necessary to prevent patients from ever progressing to ESRD.

9.3 Remission of nephrotic range albuminuria

In 1972, Watkins et al (12) identified the subset of diabetic patients with proteinuria above 3000 mg/24 h to carry a very poor prognosis, as all these patients died between 2 to 6 years of follow-up (12). Albuminuria can therefore be used as an indicator of severity of disease, as patients with the highest level of albuminuria - nephrotic range albuminuria - are unfortunate with the poorest prognosis (5, 12, 13). We have confirmed this finding, as patients with the highest level of albuminuria had the steepest decline in kidney function in our study of progression of diabetic nephropathy (28). The beneficial effect of reduction in albuminuria during antihypertensive treatment on decline in kidney function in diabetic and non diabetic chronic kidney diseases (22-24), implies that reduction in albuminuria may change the prognosis, even in patients with nephrotic range albuminuria. The feasibility of strikingly reductions in proteinuria in diabetic patients with nephrotic range proteinuria was originally demonstrated by Hebert (14). Remission, defined as reduction of proteinuria from above 3500 mg/24 h to below 1000 mg/24 h, was demonstrated in 8 out of 108 patients with nephrotic range proteinuria participating in the Collaborative Study Group Multicenter Controlled Trial of Captopril Therapy in Patients with Type 1 Diabetes Mellitus and Nephropathy (14). A follow-up study of these patients demonstrated long-term remission of nephrotic range proteinuria with a remission rate of 16.7% in patients assigned to captopril treatment, and 1.5% in the patients treated with conventional antihypertensive treatment (285). To evaluate the cumulative incidence of nephrotic range albuminuria in a large population of type 1 diabetic patients, we extended our cohort of 301 type 1 diabetic patients with diabetic nephropathy to include 321 consecutive patients (286). The frequency of remission, and the impact of remission on rate of decline in GFR, was evaluated in the subset of patients with nephrotic range albuminuria. As ~40% of patients with diabetic nephropathy progressed to nephrotic range albuminuria (>2500 mg/24 h) during follow-up, the progressive nature of diabetic kidney disease was confirmed in our study (286). By applying comparable criteria (285) of remission of nephrotic range albuminuria - reduction of albuminuria from above 2500 mg/24 h to below 600 mg/24 h, we demonstrated a remission rate of 22% of nephrotic range albuminuria (286). Remission did not occur spontaneously, but was induced by antihypertensive treatment. Apart from this treatment modality, no other major target organ saving procedure was introduced during the follow-up period. The frequency of remission must, however, be regarded as conservative, since only 33% of the patients in the no remission group achieved a blood pressure below 140/90 mm Hg. As opposed to the initial findings (285), we did not find ACE inhibitors to be superior in inducing remission, although the majority of the patients obtaining remission were treated with ACE inhibitors in our study. As our study was not designed to evaluate effects of different drug regimens, no firm conclusion on this topic can be drawn from our results. Patients obtaining remission of nephrotic range albuminuria had a rate of decline in GFR during follow-up of ~50% of the patients not obtaining remission (3.8 vs. 7.5 ml/min/year, p<0.001). Furthermore, during the entire follow-up period, patients in the remission group had significantly lower arterial blood pressure and serum cholesterol ( Figure 7 ). In those patients where the rate of decline in GFR could be assessed before and during remission of nephrotic range albuminuria, a significantly lower decline in GFR was found during remission. With the exception of gender, no differences in known risk factors for progression of kidney disease at onset of nephrotic range albuminuria was apparent (286). The various reasons for some patients responding better to treatment than others is of vast interest to improve the treatment and prognosis for all patients. Numerous factors may be implicated, e.g. genetic factors, physiological as well as psychological patient related factors. Patient compliance with drug therapy has been recognized as one of the major reasons why antihypertensive therapy fails in the United States and elsewhere (287), and subsequently non compliers may be misclassified as non responders. The probability of poor compliance increases with increasing number of daily doses as well as number of different drugs. To control blood pressure in patients with diabetic nephropathy, antihypertensive treatment with several different antihypertensive agents is however often necessary. Poor adherence to antihypertensive treatment can be observed in patients with uncontrolled hypertension as well as in well controlled blood pressure (288), which implies that even at the same blood pressure level, non adherence to e.g. ACE inhibitor therapy will render the patients without the non haemodynamic beneficial effects of these compounds (189, 190). In conclusion, aggressive antihypertensive treatment can induce long lasting remission in a sizeable fraction of type 1 diabetic patients with nephrotic range albuminuria. The patients obtaining remission have a slow mean decline in GFR and improved cardiovascular risk profile.

9.4 Remission of nephrotic range albuminuria
and improved survival

Before the general use of antihypertensive treatment in patients with diabetic nephropathy, the disease was associated with an extremely high mortality, the average survival time from onset of proteinuria being only 5 to 7 years (2, 5). In studies where antihypertensive treatment was used randomly and infrequent, the mortality rate was 50 to 77% after 10 years with proteinuria (2, 3, 289). As compared to the background population, the relative mortality was ~40 times higher in type 1 diabetic patients with proteinuria, as opposed to only ~2 times higher in patients with normoalbuminuria (290). Historically, the high relative mortality in these patient was mainly due to ESRD, which was the cause of death in 66% of the patients (2). The remarkable reduction in the average rate of decline in GFR (11), has additionally been associated with improvement in median survival time, and accompanied by a change in the pattern in causes of death. Comparing two groups of type 1 diabetic patients with proteinuria without and with the use of antihypertensive treatment have showed that after introduction of this treatment modality, the 8 year mortality declined from 52% to 13% (291). In a prospective observational study including 263 patients with diabetic nephropathy, Rossing et al found a median survival time after onset of nephropathy of 14 years (6). The main causes of death were ESRD (35%), and cardiovascular disease (33%). The reduction in decline in kidney function is not surprising central in postponing time to ESRD, but a change in pattern of causes of death have also been observed. Antihypertensive treatment in these studies was primarily conventional in the sense of the use of agents other than ACE inhibitors. The Collaborative Study Group of Angiotensin Converting Enzyme inhibition with Captopril in Diabetic Nephropathy demonstrated a risk reduction in the captopril treated patients of the combined endpoint of death, dialysis or transplantation of 61% (95% CI: 26 to 80%) in patients with serum creatinine levels >133 μ mol/l versus placebo. In patients with serum creatinine <133 μ mol/l, the risk reduction was 46% (95% CI: -22 to 76%) (190). Albuminuria itself is associated with increased cardiovascular mortality (58), and the relative risk for cardiovascular death has recently been reported to be 2.97 in patients with proteinuria as compared to normoalbuminuric patients (6). Furthermore, a 10 fold in increase in urinary albumin excretion rate increased the risk for all cause mortality with 45% (6). Others have confirmed that higher levels of albuminuria is associated with increased mortality, as mortality ratios have been reported to be 1.5 (95% CI: 1.1 to 2.0) in type 1 diabetic patients with "light" albuminuria, and 2.9 (95% CI: 2.2 to 3.9) in patients with "heavy" proteinuria as compared to patients with no albuminuria (292). Thus, confirming a higher risk of mortality with increasing levels of albuminuria.

Even though the Food and Drug Administration in the USA have approved the rate of decline in GFR as an endpoint in kidney diseases (19), deterioration in kidney function may be regarded as a surrogate endpoint for the ultimate endpoint: ESRD in terms of dialysis or kidney transplantation. We have demonstrated that remission of nephrotic range albuminuria not only is feasible, it is also associated with a diminished rate of decline in GFR in these patients. In addition, remission coincided with an improved cardiovascular risk profile (286). Prevention or at least postponing the progression to ESRD would be an expected and positive consequence of the reduced rate of decline in GFR. However, evidence for the assumption that remission of nephrotic range albuminuria in type 1 diabetic patients will transfer in to a better long-term prognosis and survival has not been provided yet. To assess the outcome of remission of nephrotic on the hard endpoints death, dialysis or transplantation, we analyzed data from our long-term prospective observational study of type 1 diabetic patients with nephrotic range albuminuria due to diabetes (29). The patients were followed until the 31st of December 2003 or to death, dialysis or transplantation. During the median follow-up of 12.4 years, 25% of the patients in the remission group progressed to ESRD or died, whereas 74% of the patients in the no remission group progressed to the composite endpoint. Cardiovascular disease was the major cause of death in the no remission group (64% of deaths). Cox proportional hazard regression analysis with delayed entry revealed that obtaining remission was associated with a decrease in risk of reaching the composite endpoint (relative risk 0.28 (95% CI:0.13 to 0.59)) (29). Analyzing death separately resulted in a relative risk of this endpoint of 0.37 (95% CI: 0.14 to 0.94) in the remission group (29). The effect of remission of nephrotic range albuminuria was strong, with a reduction of both ESRD and mortality by ~60%.

In type 2 diabetes, the Reduction of Endpoints in NIDDM with the Angiotensin II Antagonist Losartan study (RENAAL), the baseline levels as well as the initial reduction in albuminuria has been demonstrated to be a powerful predictor of ESRD and cardiovascular morbidity. Every 50% reduction in proteinuria within the first six months halved the risk for cardiovascular endpoints and heart failure during follow-up (293). No other studies have been published concerning the impact of reduction in albuminuria on all cause mortality and cardiovascular events in diabetic patients.

In summary, remission of nephrotic range albuminuria is not only associated with a slower progression in diabetic nephropathy. In addition a substantially improved survival in these patients was demonstrated (29). Even though uremia is still prevalent, cardiovascular disease has become increasingly more frequent as cause of death. Early detection and treatment of cardiovascular disease in the diabetic patients have therefore become increasingly urgent as the kidney prognosis has improved during the last decades.