Histological classification of the glomerular lesions was according to the World Health Organization standard (17). The following definitions of the histological entities were applied.

Degree of severity of the diabetic lesions was estimated (from 1 to 3) by amount of matrix in mesangium of non-sclerosed glomeruli.

The kidney biopsies revealed diffuse and/or nodular DG with no other lesions in 27 patients (1 woman), whereas the remaining 8 patients had minimal lesion (n = 4), mesangioproliferative glomerulonephritis (n = 2), mesangioproliferative glomerulonephritis superimposed on glomerulosclerosis (n = 1), and sequelae after glomerulonephritis (n = 1), as described in great detail previously (14).

The patients were considered to have type 2 diabetes if they were treated with diet alone or in combination with oral hypoglycemic agents, or if they were treated with insulin and had an onset of diabetes after the age of 40 years and a body weight in excess of the ideal body weight at the time of diagnosis (20). All insulin-treated patients were lean at the time of diagnosis and had a glucagon test performed, and type 2 diabetes was diagnosed if a stimulated C-peptide value was >0.60 pmol/ml (20).

The patients were scheduled to visit the outpatient clinic every 3–6 months. At each visit, albuminuria, arterial blood pressure, HbA_1c , and body weight were measured, and the antidiabetic treatment and antihypertensive therapy were adjusted. Total serum cholesterol and serum HDL cholesterol were measured approximately every second year and GFR was determined at least once a year from the onset of persistent albuminuria.

GFR was measured after a single intravenous injection of 3.7 MBq ⁵¹Cr-labeled EDTA by determination of the radioactivity in venous blood samples taken from the other arm 180, 200, 220, and 240 min after the injection (21,22). Additional blood samples were obtained after 270 and 300 min when the GFR dropped below 20 ml · min^–1 · 1.73 m^–2. The small underestimation (10%) of ⁵¹Cr-EDTA clearance versus clearance of inulin was corrected by multiplying EDTA clearance by 1.10 (21).

All plasma clearance studies were carried out between 0900 and 1330. Patients had their usual breakfast and morning medication before the investigation. They drank 200 ml tap water per hour during the clearance study.

Urinary albumin excretion was measured by radioimmunoassay in all 24-h urine samples until 1992 (23). After 1992, an enzyme-linked immunosorbent assay method was used (24). The correlation between the two methods was r = 0.99% (24). Serum creatinine concentration was assayed by a kinetic Jaffe method (25).

Arterial blood pressure was measured after 10 min of rest at each visit to the clinic. The measurements were taken with an appropriately sized cuff. Arterial hypertension was diagnosed according to the World Health Organization criteria of systolic blood pressure >160 mmHg and/or diastolic blood pressure >95 mmHg or if antihypertensive treatment was being prescribed.

Retinopathy was assessed by direct ophthalmoscopy at baseline and after 1989 by fundus photography after pupillary dilation and graded as nil, simplex, or proliferative retinopathy. BMI was calculated as body weight per height squared (kg/m²). Serum total cholesterol, serum HDL cholesterol, and serum albumin were measured using conventional laboratory techniques.

The patients diabetic diet was unaltered during the study. But they were put on a protein-restricted diet (i.e., 0.8 g · kg^–1 · 24 h^–1) when GFR was <20 ml · min^–1 · 1.73 m^–2.

The cause of death in patients with serum creatinine >500 µmol/l when dying was classified as ESRD.

All patients gave informed consent and the study was performed in accordance with the Helsinki Declaration.

All comparisons of normally distributed parameters were done with a Student's t test using paired design for intragroup and unpaired design for intergroup comparisons. Values for GFR, serum creatinine, albuminuria, and follow-up time are expressed as median and range. Changes in GFR, serum creatinine, and albuminuria during follow-up were examined using the Wilcoxon's test. Dichotomous characters were compared with Fisher's exact test. A P value <0.05 was considered significant (two-tailed).

Linear regression analysis was used to assess the decline in renal function, the rate of decline being calculated from all GFR determinations. Values for albuminuria were logarithmically transformed before inclusion in the regression analysis because of their positive skewed distribution.

Univariate linear regression and multivariate backward stepwise linear regression analysis were used to assess the association between the putative progression promoters (GFR at entry, baseline albuminuria, and mean values during follow-up of systolic blood pressure, diastolic blood pressure, mean arterial blood pressure [MABP], total serum cholesterol, serum HDL cholesterol, and HbA_1c) and the rate of decline in GFR. All variables significant at P < 0.05 in the univariate analyses were included in the multiple regression analyses.

All calculations were made using SPSS for Windows (SPSS, Chicago).

Diabetic retinopathy was not present in the NDG group, whereas 65% of the patients in the DG group had retinopathy at entry. At the end of follow-up, nearly the same number of patients in the two groups had retinopathy, but proliferative retinopathy was only present in the DG group (Table 1).

During the follow-up time of 7.7 years (1.0–14.2) [median (range)], GFR was measured 11 times (3–20), albuminuria 27 times (5–87), arterial blood pressure 36 times (4–87), HbA_1c 13 times (1–35), and serum total cholesterol 12 times (2–24). Data on the course of GFR, serum creatinine, albuminuria, and arterial blood pressure are presented in Table 2. The difference in rate of decline in GFR between the two groups was significant (P < 0.05). The lowest value of baseline GFR in the NDG group was 89 ml · min^–1 · 1.73 m^–2; the rate of decline in GFR in patients with diabetic glomerulosclerosis with GFR below and above 89 ml · min^–1 · 1.73 m^–2 was 5.8 and 4.5 ml · min^–1 · year^–1, respectively.

Albuminuria increased in the DG group, whereas it decreased in the NDG group (P = 0.05 between groups).

MABP decreased from 118 ± 3 to 104 ± 3 mmHg (P < 0.05) (DG group), whereas it remained unchanged in the NDG group (106 ± 3 vs. 105 ± 3 mmHg). The percentage of patients with arterial hypertension receiving antihypertensive treatment was about the same in the two groups, but the number of DG patients receiving two drugs or more had increased from 23 to 73% (P < 0.01) (Table 1).

Of the patients, 35% (nine patients) in the DG group received treatment with an ACE inhibitor for a median time of 70 months (range 1–90) with a median dose of 50 mg (12.5–100) captopril per day. In the NDG group, two patients (25%) were treated with an ACE inhibitor; one with a dose equal to 100 mg captopril for 28 months and the other with 75 mg captopril for 119 months.

The course of kidney function and arterial blood pressure in the four type 2 diabetic patients with minimal renal lesion is shown in Table 3. Two of the four patients had nephrotic range albuminuria at baseline, and two had a reduction in albuminuria into the microalbuminuric range at the end of the study, whereas one patient had a slight increase in albuminuria during the follow-up period. Total cholesterol was increased in all four patients and ranged from 6.2 to 7.7 mmol/l, and serum albumin was slightly reduced in all four patients and ranged from 526 to 571 µmol/l (lower normal value, 575 µmol/l).

In the DG group, univariate analysis revealed that the rate of decline in GFR correlated significantly with GFR at entry (r = –0.45, P < 0.05) and baseline albuminuria (r = 0.55, P < 0.01). The same analysis based on mean values during the follow-up period revealed that systolic blood pressure (r = 0.62, P < 0.001), MABP (r = 0.52, P < 0.001), and albuminuria (r = 0.55, P < 0.005) as progression promoters for the rate of decline in GFR. Patients with nephrotic range albuminuria (>3 g/day) during follow-up had a faster rate of decline in GFR of 8.3 ml · min^–1 · year^–1 (5.6–21.6) as compared with 3.3 ml · min^–1 · year^–1 (0.3–18.6) in the non-nephrotic patients (P < 0.01). No significant correlation was found between rate of decline in GFR and mean values during follow-up of HbA_1c, diastolic blood pressure, serum cholesterol, and serum HDL cholesterol.

A multiple regression analysis was performed in the DG groups with baseline GFR, mean values during follow-up of albuminuria, and systolic blood pressure as independent variables, and the rate of decline in GFR as a dependent variable. The mean systolic blood pressure was the only variable associated with the rate of decline in kidney function (r = 0.62, P < 0.001).

Causes of death are shown in Table 4. The overall mortality was 58% in the DG group versus 50% in the NDG group during the follow-up period (NS). Of the patients, 50% (95 30–70) in the DG group versus 12.5% (0.3–53) in the NDG group had a doubling or more of serum creatinine during follow-up (P = 0.10). Mean time to doubling of baseline creatinine was 99 months (80–119) in the DG group; at this time, only one patient in the NDG group had a doubling of baseline serum creatinine. In the DG group, one patient received a renal transplantation and one patient received dialysis. In addition, 27% (12–48) of the patients in the DG group developed or died from ESRD, defined as serum creatinine >500 µmol/l, whereas none (0–37) of the NDG patients reached that stage.

Fioretto et al. (personal communication) have measured GFR repeatedly (at least three determinations per patient with ⁵¹Cr-EDTA plasma clearance) during 4 years (range 1.5–6) in 64 type 2 diabetic patients with persistent microalbuminuria. The causes of microalbuminuria were evaluated by a percutaneous kidney biopsy. The study revealed that only patients with typical diabetic glomerulopathy (12 of 64 [19%]) had a significant decrease in GFR as compared with patients with mainly tubulointerstitial and arteriolar changes (26 of 64 [40.5%]) and patients with normal or near-normal renal structure (26 of 64 [40.5%]) (GFR [ml/min/year]: –6.4 ± 6.7, –0.7 ± 7.8, and 2.0 ± 7.1 [P < 0.05], respectively).

It should be stressed that the design of our study is not case-controlled, but a prospective observational study of a cohort of consecutive albuminuric type 2 diabetic patients who had a kidney biopsy performed. All patients in the original cohort were followed up and accounted for in the present analysis, except one male patient with glaucoma whom we had to exclude because of long-term treatment (years) with acelazolamide. This drug induces major reduction in GFR (27,28).

The sole indication for kidney biopsy in our consecutive type 2 diabetic cohort was persistent albuminuria. All biopsies were reviewed by two experienced nephropathologists who were unaware of the patients clinical features. Minimal lesions were described if glomeruli were normal or a slight global and diffuse mesangial hypercellularity (a maximum of four cell nuclei) associated with negative immunofluorescence and absence of glomerular deposits by electron microscopy was apparent. Furthermore, fusion of the foot processes of the visceral epithelial cells was demonstrated in two of three patients. The term "minimal lesions" applied in our study indicates normal or near-normal renal structure, as described below by Fioretto et al. (29). Recently, three studies also applying an unbiased indication for kidney biopsy have been reported in micro- and macroalbuminuric type 2 diabetic patients (15,16,29). Fioretto et al. (29) have described by light microscopy a heterogeneity in renal structure in 34 microalbuminuric type 2 diabetic patients; in fact, 29% have normal or near-normal renal structure. Cordonnier et al. (16) investigated 26 type 2 diabetic patients with proteinuria ranging from 70 to 4,210 mg/24 h and reported nonspecific vascular and glomerular change in 15% of the patients. Schwartz et al. (15) investigated 34 hypertensive proteinuric type 2 diabetic patients with serum creatinine between 133 and 265 µmol/l and found 6% of the patients had nondiabetic glomerulopathy. However, the chance in that study of enrolling patients with normal or near-normal renal structure is clearly reduced because of the above-mentioned criteria of progressive kidney disease. Finally, it should be recalled that a much higher prevalence of nondiabetic kidney diseases frequently has been reported in other biopsy series, but this is mainly due to selective referral of patients with nondiabetic renal disease, as reviewed by Olsen and Mogensen (30).

The normal or near-normal renal structure in proteinuric patients may reflect one of the following causes: minimal-change nephropathy, focal segmental glomerulosclerosis (undetected), silent DG (electron microscopic glomerulopathy), and finally, a new entity with normal renal structure but increased glomerular permeability to macromolecules (size/charge defects).

The relevance of the present study to the scientific community is to provide information for conduct of subsequent treatment trails in albuminuric type 2 diabetic patients with an accelerated course of GFR due to diabetic glomerulosclerosis. By applying strict clinical (14) or histological criteria for diabetic nephropathy, it is possible to enroll albuminuric type 2 diabetic patients with a high risk of chronic progressive renal disease. This approach will increase the power and reduce the number of patients enrolled in future treatment trials.

Exact knowledge of the underlying cause of albuminuria may also play an important role in offering correct treatment to patients, as demonstrated in several studies in type 2 diabetic patients suffering from NDG (31–33). Even though two of our four patients with minimal change nephropathy had clinical evidence of minimal change disease (i.e., nephrotic range albuminuria, hypoalbuminuria, and hypercholesterolemia), steroid treatment was not initiated because they had a small rate of decline in GFR (Table 3).

Our study revealed a strong correlation between arterial blood pressure and the rate of decline in GFR, which is in agreement with previous investigators. Furthermore, antihypertensive treatment reduces the rate of decline in GFR and thereby postpones ESRD (8,11,34–36). It should be mentioned that even though this is an observational study, the level of arterial blood pressure and percentage of patients receiving antihypertensive treatment are comparable in the two groups in the present study. Even though the number of patients in the DG group receiving two drugs or more increased significantly and MABP decreased during follow-up, these patients experience a faster decline in GFR compared with patients in the NDG group, who had nearly the same blood pressure level during follow-up.

Earlier studies dealing with type 2 diabetes have found an increasing prevalence of arterial hypertension with increasing albuminuria (37–39). The patients with DG had a significant increase in albuminuria, whereas the patients with NDG had a significant decrease in albuminuria. Albuminuria is assumed to be an independent risk factor for the progression of renal diseases, and a reduction in albuminuria is important to preserve kidney function (40–43). We found a significant correlation between the rate of decline in GFR and albuminuria at baseline and during follow-up, which supports the concept that albuminuria is a risk factor for progression in DG. Furthermore, an accelerated loss in kidney function was demonstrated in patients with nephrotic range albuminuria.

The impact of hyperglycemia on the progression in diabetic nephropathy is debated. In agreement with our findings, several other studies have failed to demonstrate a significant correlation between glycemic control and progression of GFR in albuminuric type 2 diabetic patients (44,45). In contrast, several studies in type 1 diabetic patients have shown that hyperglycemia is a progression promoter of diabetic nephropathy (11,46,47).

In conclusion, our study demonstrated a more rapid decline in GFR and a progressive rise in albuminuria in type 2 diabetic patients with DG compared with type 2 diabetic patients with NDG.