Cystatin C is a more sensitive marker than creatinine for the estimation of GFR in type 2 diabetic patients

Kidney International, Vol. 61 (2002), pp. 1453–1461

CLINICAL NEPHROLOGY – EPIDEMIOLOGY – CLINICAL TRIALS

Cystatin C is a more sensitive marker than creatinine for the estimation of GFR in type 2 diabetic patients MICHELE MUSSAP, MICHELE DALLA VESTRA, PAOLA FIORETTO, ALOIS SALLER, MARIACRISTINA VARAGNOLO, ROMANO NOSADINI, and MARIO PLEBANI Department of Laboratory Medicine, and Department of Medical and Surgical Sciences, University of Padova, Padova; Chair of Endocrinology and Metabolic Diseases, University of Sassari, Sassari; and Center of Biomedical Research, Veneto Region, Castelfranco Veneto, Italy

Cystatin C is a more sensitive marker than creatinine for the estimation of GFR in type 2 diabetic patients. Background. Glomerular filtration rate (GFR) is the best overall index of renal function in health and disease. Inulin and 51Cr-EDTA plasma clearances are considered the gold standard methods for estimating GFR. Unfortunately, these methods require specialized technical personnel over a period of several hours and high costs. In clinical practice, serum creatinine is the most widely used index for the noninvasive assessment of GFR. Despite its specificity, serum creatinine demonstrates an inadequate sensitivity, particularly in the early stages of renal impairment. Recently, cystatin C, a low molecular mass plasma protein freely filtered through the glomerulus and almost completely reabsorbed and catabolized by tubular cells, has been proposed as a new and very sensitive serum marker of changes in GFR. This study was designed to test whether serum cystatin C can replace serum creatinine for the early assessment of nephropathy in patients with type 2 diabetes. Methods. The study was performed on 52 Caucasian type 2 diabetic patients. Patients with an abnormal albumin excretion rate (AER) were carefully examined to rule out non-diabetic renal diseases by ultrasonography, urine bacteriology, microscopic urine analysis, and kidney biopsy. Serum creatinine, serum cystatin C, AER, serum lipids, and glycosylated hemoglobin (HbA1c) were measured. GFR was estimated by the plasma clearance of 51Cr-EDTA. In addition the Cockcroft and Gault formula (Cockcroft and Gault estimated GFR) was calculated. Results. Cystatin C serum concentration progressively increased as GFR decreased. The overall relationship between the reciprocal cystatin C and GFR was significantly stronger (r ⫽ 0.84) than those between serum creatinine and GFR (r ⫽ 0.65) and between Cockcroft and Gault estimated GFR and GFR (r ⫽ 0.70). As GFR decreased from 120 to 20 mL/min/ 1.73 m2, cystatin C increased more significantly that serum

Key words: cystatin C noninvasive measurement of GFR, glomerular filtration rate, type 2 diabetes, diabetic nephropathy, serum creatinine, end-stage renal disease. Received for publication March 29, 2001 and in revised form October 18, 2001 Accepted for publication October 31, 2001

 2002 by the International Society of Nephrology

creatinine, giving a stronger signal in comparison to that of creatinine over the range of the measured GFR. The maximum diagnostic accuracy of serum cystatin C (90%) was significantly better than those of serum creatinine (77%) and Cockcroft and Gault estimated GFR (85%) in discriminating between type 2 diabetic patients with normal GFR (⬎80 mL/min per 1.73 m2) and those with reduced GFR (⬍80 mL/min/1.73 m2). In particular, the cystatin C cut-off limit of 0.93 mg/L corresponded to a false-positive rate of 7.7% and to a false-negative rate of 1.9%; the serum creatinine cut-off limit of 87.5 ␮mol/L corresponded to a false-positive rate of 5.8% and to a falsenegative rate of 17.0%. Conclusions. Cystatin C may be considered as an alternative and more accurate serum marker than serum creatinine or the Cockcroft and Gault estimated GFR in discriminating type 2 diabetic patients with reduced GFR from those with normal GFR.

Diabetic nephropathy is the single most frequent cause of end-stage renal disease (ESRD) in the United States and in Europe [1–3], and is predominantly due to type 2 diabetes mellitus. An increasing number of type 2 diabetic patients live long enough for nephropathy and ESRD to develop, since the treatment of diabetes, hypertension and coronary heart disease has improved. Therefore, prevention of diabetic renal disease, or at least the postponement or slowing down of the disease process, has emerged as a key issue [4, 5]. Microalbuminuria is the first detectable functional abnormality [6]; a proportion of microalbuminuric patients then progress to overt nephropathy, characterized by the presence of proteinuria [7]. Although measurements in albumin excretion rate (AER) are very useful in clinical practice, there is a 40% day-to-day variability in AER [8]. In addition, the efficacy of the therapeutic treatment for diabetic nephropathy is commonly evaluated by assessing its positive effects on glomerular filtration rate (GFR), while those on AER alone are not sufficient to prove a clinical benefit. Indeed, the rate of the glomerular filtration is widely believed to be the best overall index of renal

1453

1454

Mussap et al Cystatin C in type-2 diabetes mellitus

function in health and disease [9–11]. While inulin and 51 Cr ethylenediaminetetraacetic acid (51Cr-EDTA) clearances are traditionally considered the gold standard for measuring GFR, they require specialized technical personnel over a period of several hours [12]. In addition, a number of practical considerations (cumbersome methods of determination, radioactivity, high costs) have limited the use of these techniques in clinical practice. Serum creatinine and creatinine clearance are the most widely used indices for the routine noninvasive estimation of GFR. Serum creatinine is considered relatively specific, but not very sensitive since its levels significantly increase when more than 50% of the GFR is reduced [13]. The serum creatinine concentration may be significantly influenced by several extra-renal factors (muscle mass, changes in tubular secretion, dietary intake, analytical interferences, etc.): especially in elderly female patients with reduced muscle mass, measurement of serum creatinine may grossly underestimate the reduction in the GFR. Finally, numerous drugs and endogenous substances also interfere with the measurement of creatinine, leading to falsely high or low creatinine values [14]. Consequently, there is a need to provide an alternative method to creatinine that is analytically more reliable and as or more clinically reliable. Because the kidney plays a major role in the metabolism of low molecular mass (low-Mr) plasma proteins, it was postulated that their serum levels might reflect changes in GFR. Most low-Mr plasma proteins are freely filtered by the glomerulus and then almost completely reabsorbed and catabolized by proximal tubular cells [15, 16]. Among these proteins, cystatin C has been recently suggested as a serum marker of changes in GFR. Human cystatin C is a basic low-Mr plasma protein of 13,359 Daltons [17] with 120 amino acid residues. It belongs to the cystatin superfamily [18] and is synthesized and secreted by all nucleated human cells [19]. Cystatin C does not cross the placental barrier [20] and it does not seem to be significantly influenced by gender and age beyond the first year of life [21]. Inflammation does not influence serum cystatin C variations, and thus it cannot be considered an acute phase protein [22]. In healthy adults, cystatin C reference intervals, assessed by nephelometry, were found to range from 0.51 to 0.98 mg/L, being independent from gender and age [23]. Because of its lowMr and its positive charge at physiological pH, cystatin C easily crosses the glomerular filter; after filtration, the proximal tubular cells reabsorb and catabolize virtually all of the filtered cystatin C rate [24]. It was demonstrated that the renal clearance of cystatin C is closely related to GFR, measured as 51Cr-EDTA clearance [25]. Indeed, the serum concentration of cystatin C increases considerably in patients with renal failure, as does that of other low-Mr plasma proteins [26]. Thus, cystatin C has been proposed as a new sensitive endogenous serum marker

for the early assessment of changes in GFR [27, 28]. One of the most significant advantages of cystatin C in comparison with traditional markers of renal impairment is that very small reductions in GFR cause significant increases in cystatin C serum levels [29]. This study was designed to compare traditional markers of changes in GFR with cystatin C in patients with type 2 diabetes, and to determine whether diagnostic accuracy of serum cystatin C is better than that of serum creatinine and Cockcroft and Gault estimated GFR for the early assessment of diabetic nephropathy. METHODS Patients The study was performed on 52 Caucasian type 2 diabetic patients (33 males, 19 females, aged between 48 and 73 years) during day-hospital admission in the Department of Medical and Surgery Sciences at Padua. Type 2 diabetes was diagnosed in all patients after age 40 years. The therapeutic treatment consisted of either diet alone or in association with oral hypoglycemic agents and/or insulin. No patient received insulin over the first two years after diagnosis. Hypertensive patients were treated with angiotensin-converting enzyme (ACE) inhibitors and/or with calcium channel blockers. When necessary, diuretics were added. Patients with abnormal AER (ⱖ20 ␮g/min) were carefully examined to rule out non-diabetic renal diseases by performing ultrasonography, urine bacteriology, microscopic urine analysis, and kidney biopsy. Renal function was assessed by measuring urinary AER, serum creatinine, serum cystatin C, and by determining the 51Cr-EDTA plasma clearance. We also calculated the Cockcroft-Gault formula. Furthermore, levels of hemoglobin A1c (HbA1c), glucose, and plasma lipids [total cholesterol, low-density lipoprotein (LDL) cholesterol, triglycerides] were measured to assess risk factors for nephropathy [30]. All 52 patients accepted to participate in the study, giving informed consent. The study was approved by the local ethical committee. Methods Blood pressure was measured by mercury sphygmomanometers with the patients standing; the average of four determinations was recorded, using extra cuff size if upper-arm circumference exceeded 32 cm. Body mass index (BMI, kg/m2) was calculated by dividing weight, expressed in kg, for height, expressed in meters. Percutaneous renal biopsies were performed under ultrasound guidance; tissue was examined under dissecting microscope to ensure adequate number of glomeruli and processed for light, electron, and immunofluorescent microscopy [31, 32]. GFR was estimated by plasma clearance of 51Cr-EDTA [33]. After a single bolus injection of 1

Mussap et al Cystatin C in type-2 diabetes mellitus

mg Ci/kg of 51Cr-EDTA within 30 seconds, 12 venous blood samples were drawn at 5, 15, 30, 60, 90, 120, 150, 180, 210, 240, 270, and 300 minutes. Finally, the 51Cr radioactivity was measured in duplicate 1-mL aliquots of plasma in a gamma counter with energy windows set to 240 to 400 keV (Cobra-5002 CPM; Camberra Packard, Milan, Italy). All biochemical examinations were done centrally and the assays were unchanged during the study period. Urinary albumin measurements were performed on three consecutive 24-hour urine collections, after urine cultures were found to be sterile. Analytical method is based on a commercial available nephelometric immunoassay [34] on the fully-automated BN II nephelometer (Dade Behring, Milan, Italy). The detection limit of the assay is 8.0 mg/L. Serum cystatin C was measured by a nephelometric immunoassay based on rabbit monospecific anti-human cystatin C antiserum, covalently coated with 80-nm diameter chloromethylstirene particles [35] (Dade Behring Diagnostics). Serum creatinine was measured by the Vitros娃 enzymatic assay [36] (Vitros娃 950; Ortho Diagnostic Division, Milan, Italy). We estimated GFR also by applying the Cockcroft and Gault formula [37]: Cockcroft and Gault estimated GFR (mL/min) ⫽ (140 ⫺ age) · weight (kg)/␬ · serum creatinine (mg/dL), where age is expressed in years, and ␬ is a constant (0.72 in males, 0.85 in females). HbA1c was assessed by HPLC (BioRad Laboratories, Milan, Italy). Serum glucose was measured by the glucose-oxidase enzymatic method on a fully-automated multichannel analyzer (Hitachi 200, Roche, Milan, Italy). Total cholesterol and triglycerides were measured enzymatically on the same fully-automated multichannel analyzer (Hitachi 200). LDL cholesterol (LDL-c) was determined by using a homogeneous method for directly measuring serum LDL-c on a fullyautomated analyzer (Hitachi 912, Roche). Statistical analysis The statistical analysis was made by using the MS Excel ‘98 software (Microsoft娃 Co., Seattle, WA, USA), the Astute娃 statistical package (Diagnostic Development Unit, University of Leeds, Leeds, UK), and the StatView SE⫹Graphics娃 4.01 statistical software (Abacus Concepts Inc., Berkeley, CA, USA) on a Macintosh PowerBook G3 computer (Apple Computer Inc., Cupertino, CA, USA). Results were presented as median (50th percentile), range of measured levels, and percentiles [38]. To check the Gaussian distribution, data were preliminarily evaluated by applying the Kolgomorov-Smirnov test, taking P ⬍ 0.001 as a significant result. Cystatin C and creatinine were found normally distributed; for other variables, we calculated their logarithm before statistical treatment. All data were finally evaluated by using standard parametric tests. In particular, differences in concentrations of the variables between groups were investigated by applying the one-way analysis of variance

1455

(ANOVA) followed by the Scheffe´ multiple comparison test and by the unpaired Student t test. P ⬍ 0.05 was considered statistically significant. Reciprocal values of serum cystatin C and serum creatinine were calculated for the correlation with the 51Cr-EDTA clearance. This procedure resulted in a linearization of the curvilinear relationship between renal filtration and the serum marker concentration. Correlations between variables were investigated by using the simple linear regression analysis and by calculating the coefficient of regression (r) [39]. To assess the diagnostic value of the individual markers, receiver operating characteristic (ROC) were obtained and calculated. Non-parametric ROC curves were generated by plotting the sensitivity versus 1-specificity, giving the ideal test a sensitivity equal to 1 and a specificity equal to 1 (corresponding to 1-specificity equal to zero). The area under the curve (AUC) represents the most commonly used measure of the diagnostic efficiency of a test. Traditionally, this AUC is always ⬎0.5, values ranging from 1 (ideal perfect separation of the test values between groups of patients and controls) to 0.5 (no apparent distribution difference between the two groups). According to Hanley and McNeil [40], we calculated areas under the curves, 95% confidence intervals (CI) [41], and differences between ROC curves. A value ⬎1.96 of the critical ratio z, defined as z score, was considered significant, as indicated elsewhere [42]. RESULTS On the basis of the 51Cr-EDTA plasma clearance values, type 2 diabetic patients were divided in two groups: 28 patients with reduced GFR (ⱕ80 mL/min/1.73 m2) and 24 with normal GFR (⬎80 mL/min/1.73 m2). Their main characteristics are summarized in Table 1. In the group of patients with reduced GFR, 5 patients were normoalbuminuric (AER ⬍20 ␮g/min), 5 microalbuminuric (AER 20 to 200 ␮g/min), and 18 macroalbuminuric (AER ⬎200 ␮g/min). Among patients with normal GFR, 7 were normoalbuminuric, 11 microalbuminuric, and 6 macroalbuminuric. Body mass index (BMI) and AER significantly differed between the two groups (P ⫽ 0.01). A significantly inverse correlation between GFR and AER was found in the overall cohort (r ⫽ ⫺0.50). This inverse correlation was stronger (r ⫽ ⫺0.60) in patients with reduced GFR than in patients with normal GFR (r ⫽ ⫺0.36). In the group with normal GFR, serum creatinine values significantly differed (P ⫽ 0.0001) between males and females (91.7 ⫾ 14 ␮mol/L vs. 72.4 ⫾ 10 ␮mol/L, respectively), while no difference (P ⫽ 0.21) was observed for cystatin C values (males, 0.76 ⫾ 0.09 mg/L; females, 0.72 ⫾ 0.13 mg/L). Likewise, in the group with reduced GFR we found that serum creatinine values in males (128 ⫾ 43 ␮mol/L) significantly differed (P ⫽ 0.008) from those in females (90 ⫾ 13 ␮mol/L), while

1456

Mussap et al Cystatin C in type-2 diabetes mellitus Table 1. Baseline characteristics of patients with type 2 diabetes GFR ⬍80 mL/min/1.73 m2 (N ⫽ 28)

Males/females N Age years (range) Duration of the disease years (range) Body mass index kg/m2 Mean blood pressure mm Hg Fasting glucose mmol/L HbA1c % Urinary albumin lg/min Total cholesterol mmol/L LDL cholesterol mmol/L Triglycerides mmol/L

61 16.5 25.5 107 8.5 7.8 386 5.97 4.11 1.50

⬎80 mL/min/1.73 m2 (N ⫽ 24)

19/9 (48–72) (4.0–33.0) (19.0–33.0)a (90–140) (5.5–19.8) (4.9–11.3) (3–4524)a (2.69–7.50) (1.16–5.71) (0.81–4.27)

58 10.5 30.5 107 11.0 8.0 68 5.49 3.43 1.75

All patients (N ⫽ 52)

14/10 (43–73) (1.0–25.0) (20.0–36.0) (90–133) (6.1–17.1) (6.0–11.7) (6–1410) (3.21–7.58) (1.32–6.70) (0.97–5.67)

59 14.5 28.5 107 10.1 7.9 120 5.61 3.84 1.66

33/19 (43–73) (1.0–33.0) (19.0–36.0) (90–140) (5.5–19.8) (4.9–11.7) (3–4524) (2.69–7.58) (1.16–6.70) (0.81–5.67)

Results are expressed as median values (50th percentile) and ranges, except those referred to the number of males and females. a Statistical comparison between groups: P ⫽ 0.01, unpaired Student t test after the logarithmic conversion of data

Table 2.

51

51

Cr-EDTA clearance, serum creatinine, Cockcroft and Gault estimated GFR, and serum cystatin C in type 2 diabetic patients

Cr-EDTA clearance mL/min/1.73 m2 Serum creatinine lmol/L Cockcroft and Gault estimated GFR mL/min Serum cystatin C mg/L

GFR ⱕ80 mL/min per 1.73 m2 (N ⫽ 28)

GFR ⬎80 mL/min per 1.73 m2 (N ⫽ 24)

All patients (GFR: 20–120 mL/min per 1.73 m2) (N ⫽ 52)

59 (16–80) 105.6a (70.7–221.0) 62b (24–113) 1.12c (0.78–2.66)

97 (81–137) 79.6a (61.9–123.8) 91b (58–150) 0.73c (0.54–1.03)

77 (16–137) 92.8 (61.9–221.0) 76 (24–150) 0.89 (0.54–2.66)

Results are expressed as median values (50th percentile) and ranges. Statistical comparison between the group of patients with normal GFR and that with reduced GFR: aP ⬍ 0.001; bP ⬍ 0.0001; cP ⬍ 0.00001

serum cystatin C values in the two groups were not significantly different (males, 1.36 ⫾ 0.55 mg/L; females, 1.09 ⫾ 0.09 mg/L; P ⫽ 0.07). In patients 61 to 70 years old, serum creatinine was significantly higher (P ⫽ 0.02) than in patients 51 to 60 years old with normal GFR (mean values 92 and 78 ␮mol/L, respectively), while serum cystatin C did not differ between patients of these two decades (0.74 and 0.76 mg/L, respectively). Table 2 compares the results obtained in the two patient groups and in the overall cohort. The overall relationship between cystatin C (expressed as reciprocal value of the serum concentration) and GFR was significantly stronger (r ⫽ 0.84, P ⫽ 0.0001, 95% CI ⫽ 1.05 to 1.15) than those between serum creatinine (expressed as reciprocal value of the serum concentration) and GFR (r ⫽ 0.65, P ⫽ 0.0001, 95% CI ⫽ 0.010 to 0.011). Figure 1 makes evident that the correlation between cystatin C and GFR as well as that between serum creatinine and GFR were higher in patients with reduced renal function (r ⫽ 0.75, 95% CI ⫽ 0.80 to 0.93, and r ⫽ 0.56, 95% CI ⫽ 0.009 to 0.01, respectively) in comparison to those with normal or nearnormal renal function (r ⫽ 0.40, 95% CI ⫽ 1.29 to 1.45, and r ⫽ 0.25, 95% CI ⫽ 0.011 to 0.013, respectively). The overall correlation between Cockcroft and Gault

estimated GFR and the 51Cr-EDTA plasma clearance was: Cockcroft-Gault estimated GFR ⫽ 28.7 ⫹ 0.611 51 Cr-EDTA plasma clearance, r ⫽ 0.70, 95% CI ⫽ 70.7 to 80.7. This correlation was found to be substantially stable in that no difference in the results was observed between patients with reduced GFR and those with normal GFR (r ⫽ 0.46, P ⫽ 0.01 and r ⫽ 0.46, P ⫽ 0.02, respectively). Furthermore, when the relationship between serum cystatin C and serum creatinine was investigated, the correlation was stronger in patients with reduced renal function (r ⫽ 0.80, P ⫽ 0.0001, 95% CI ⫽ 0.007 to 0.012) than that in patients with normal renal function (r ⫽ 0.59, P ⫽ 0.002, 95% CI ⫽ 0.70 to 0.78). By comparing the two serum markers, we found that as GFR decreased from 120 to 20 mL/min/1.73 m2, cystatin C increased more significantly that serum creatinine, giving a stronger signal in comparison to that of creatinine over the range of the measured GFR (Fig. 2). In other words, by calculating the multiple time of the upper reference range for both serum markers, cystatin C mean values measured for each GFR grouping were always higher than those of creatinine. Non-parametric ROC plots for serum cystatin C and serum creatinine demonstrated that area under curve (AUC) of cystatin C (0.954)

Mussap et al Cystatin C in type-2 diabetes mellitus

1457

Fig. 1. Correlation between 51Cr-EDTA clearance and the reciprocal serum cystatin C concentration (A) or between the reciprocal serum creatinine concentration (B) in 52 type 2 diabetic patients.

was significantly greater than that of serum creatinine (0.812), and than that of Cockcroft and Gault estimated GFR (0.873), as reported in Figure 3. This was confirmed by the statistical comparison between areas, which provided significant z-score values: 3.37 (95% CI ⫺0.27 to ⫺0.013) by comparing cystatin C versus serum creatinine and 1.98 (95% CI ⫺0.03 to 0.19) by comparing cystatin C versus Cockcroft and Gault estimated GFR. The maximum diagnostic accuracy of serum cystatin C (90%) was significantly better than those of serum creatinine (77%) and Cockcroft and Gault estimated GFR (85%) in discriminating between the type 2 diabetic patients with a normal GFR (⬎80 mL/min/1.73 m2) and those with a reduced GFR (ⱕ80 mL/min/1.73 m2). In particular, the cystatin C cut-off limit of 0.93 mg/L corresponded to a false-positive rate of 7.7% (81% specificity) and to a false-negative rate of 1.9% (97% sensitivity); the creati-

nine cut-off limit of 87.5 ␮mol/L corresponded to a falsepositive rate of 5.8% (89% specificity) and to a falsenegative rate of 17.0% (62% sensitivity). A sensitivity of 100% for serum creatinine and for the Cockcroft and Gault estimated GFR would require cut-off limits (133 and 129 mL/min, respectively) that would yield very low specificities (29 and 12%, respectively), while a sensitivity of 100% for cystatin C corresponds to a false-positive rate of 19%. A specificity of 100% for serum creatinine and the Cockcroft and Gault estimated GFR would require a cut-off limit (71 and 58 mL/min, respectively) yielding low sensitivities (12 and 36%, respectively), while a specificity of 100% for cystatin C corresponds to a false-negative rate of 15%. Table 3 summarizes the ROC analysis results and reports the diagnostic efficiencies of the investigated markers in different theoretical conditions.

1458

Mussap et al Cystatin C in type-2 diabetes mellitus

Fig. 3. Nonparametric receiver operating characteristic (ROC) plots to assess the diagnostic efficiency of serum cystatin C (䊉), serum creatinine (䉭), and Cockcroft and Gault estimated GFR ( ) by distinguishing between normal and reduced GFR in 52 patients with type 2 diabetes mellitus. Fig. 2. Relationship between glomerular filtration rate (GFR), grouped into six ranges of variation, and proportional changes in cystatin C (䊉) and creatinine (䊊), both expressed as multiple times of the upper reference range. Vertical lines show the standard deviation for the distribution of results in each GFR grouping.

DISCUSSION Although cystatin C has been proposed as a reliable serum marker of GFR [21, 23, 24, 27–29], the results in patients with diabetes mellitus are controversial. Harmoinen et al reported that serum cystatin C is a better marker of GFR than creatinine and creatinine clearance in patients with type 2 diabetes [43], and our results confirm these findings. By contrast, Oddoze et al found that serum creatinine was better than serum cystatin C for the estimation of GFR in microalbuminuric and proteinuric diabetic patients [44]. To explain these discordant results, some differences among studies should be kept in mind. First, Oddoze et al used only three blood samples to estimate 51Cr EDTA clearance, the last one at 240 minutes, and therefore their protocol may be less accurate in assessing GFR than that used in our investigation (12 blood samples over 300 min). Second, Oddoze et al selected a heterogeneous group of type 1 and type 2 diabetic patients, each of them with diabetic nephropathy (presence of microalbuminuria or proteinuria), while we studied 12 normoalbuminuric patients. The decline in renal function in type 2 diabetic patients leads to a reduction in GFR and in a proportional increase of AER, as demonstrated by the significant in-

verse correlation between these two indices. However, assessment of AER cannot replace the GFR estimation, because they may represent different aspects of renal damage. Analysis of data on the basis of gender shows that the mean concentration was 5.5% higher in males than in females for cystatin C (not statistically significant). For serum creatinine, this difference was 28% (statistically significant). Because only 4 patients were aged under 50 years and 3 over 70 years, the relationship between the patients’ age and variables was investigated only by comparing two decades (51–60 vs. 61–70 years). Despite this limitation, results show that in patients with normal GFR the serum creatinine concentration is significantly related to age, being 18% higher in patients aged 61 to 70 than in those aged 51 to 60 years. By contrast, serum cystatin C concentration was 2.7% higher in those between 61 and 70 years than between 51 and 60 years. Thus, we can argue that in our patients cystatin C is independent from gender and age, while creatinine is significantly higher in males, as previously described [29], perhaps as a consequence of the different muscle mass. This makes serum cystatin C more specific than serum creatinine in evaluating renal function. When data from all 52 patients were included in the calculations, both the serum creatinine and the cystatin C concentrations were significantly related to GFR. In Figure 1, the much tighter distribution of values around the regression line for cystatin C confirms the better correlation between cystatin C and GFR with respect to that between

1459

Mussap et al Cystatin C in type-2 diabetes mellitus Table 3. ROC analysis for cystatin C and creatinine as biochemical markers of glomerular filtration in type 2 diabetes mellitus

Creatinine lmol/L Cockcroft and Gault estimated GFR mL/min Cystatin C mg/L Creatinine lmol/L Cockcroft and Gault estimated GFR mL/min Cystatin C mg/L Creatinine lmol/L Cockcroft and Gault estimated GFR mL/min Cystatin C mg/L a

Cut-off limit

Sensitivity

Specificity

Positive predictive value

Negative predictive value

Diagnostic efficiency

87.5

62

89

83

73

77a

75 0.93 133

82 97 100

87 81 29

88 88 54

81 94 100

85a 90a 61

129 1.05 71

100 100 12

12 61 100

57 69 100

100 100 57

60 79 60

58 0.78

36 67

100 100

100 100

57 78

65 85

Maximum diagnostic efficiency

serum creatinine and GFR. Consequently, cystatin C measurement has been shown to be a reliable reflector of GFR. Cystatin C and creatinine are probably two independent markers of GFR, as described elsewhere [45]. In fact, they closely correlated in patients with a reduced GFR because of being strongly influenced by renal glomerular impairment, whereas the correlation was lower in patients with a normal GFR, as a consequence of different pathophysiological factors that may affect the serum concentration in that cohort. As shown in Figure 2, serum cystatin C levels rise earlier and more rapidly than those of serum creatinine as GFR decreases, reflecting the greater sensitivity of cystatin C as a predictor of GFR in type 2 diabetic patients. This greater sensitivity positively influences the diagnostic accuracy of cystatin C, as it is significantly higher than that of serum creatinine and Cockcroft and Gault estimated GFR, as shown in Figure 3. By using the nonparametric ROC plot, we describe the alterations in diagnostic sensitivity and specificity which occur when the cut-off limit is gradually increased. Since values of the z-score (3.37 and 1.98, respectively) exceed the limit of 1.96, the AUCs are significantly different, meaning that the diagnostic accuracy of the serum concentration of cystatin C is superior to those of serum creatinine and Cockcroft and Gault estimated GFR for the assessment of changes in GFR in type 2 diabetes. The optimum cut-off for cystatin C—defined as the serum value at which the sum of sensitivity and specificity is greatest—was 0.93 mg/L, that of serum creatinine was 87.5 ␮mol/L, and that of Cockcroft and Gault estimated GFR was 75 mL/min. In a previous study of 47 type 2 diabetic patients [43], the greatest diagnostic efficiency for cystatin C was found to be 98%, requiring a cut-off level of 1.32 mg/L, that of serum creatinine was 85%, requiring a cut-off of 91 ␮mol/L, and that of creatinine clearance was 77% with a cut-off of 71 mL/min per 1.73 m2. The most important difference between the two studies concerns the value of the greatest diagnostic efficiency of cystatin C and its related cut-

off level. This difference depends by the choice of the analytical method for measuring serum cystatin C concentration. We measured cystatin C by a nephelometric immunoassay, and as described elsewhere [35], this method significantly differs from the turbidimetric immunoassay used previously. Moreover, we calculated creatinine clearance by using the Cockcroft and Gault formula, while in the previous study creatinine clearance was traditionally estimated by measuring serum and urinary creatinine. In our study, both when the object is to exclude with certainty type 2 diabetic patients with normal or near-normal GFR (specificity) and when it is important to identify those with GFR impairment (sensitivity), serum cystatin C is a better diagnostic tool than serum creatinine and Cockcroft and Gault estimated GFR. By analyzing our results, we argue that the poor sensitivity of creatinine may be due to analytical and/or pathophysiological factors; in other words, this means that GFR can change before serum creatinine becomes abnormal, as found in previous studies [45]. On the other hand, Cockcroft and Gault estimated GFR offers a better diagnostic efficiency than the traditional creatinine clearance test; by using a cut-off value of 71 mL/min per 1.73 m2, the previous study found a diagnostic efficiency that was significantly lower (77%) in comparison to that found in this study (85%) with a cut-off limit of 75 mL/min. In conclusion, estimation of GFR using an appropriate method is a reliable measure of the kidney function and impairment in type 2 diabetic nephropathy. Together with the availability of new, sensitive, non-invasive serum markers, the accurate estimation of GFR appears to be important because nearly 40% of ESRD patients receiving renal replacement therapy have diabetes mellitus. The inadequacy of the traditional markers in detecting early changes in GFR and particularly in monitoring the course of advanced diabetic nephropathy calls for alternative non-invasive methods in clinical nephrology. Cystatin C seems to be an alternative and more accurate

1460

Mussap et al Cystatin C in type-2 diabetes mellitus

serum marker than serum creatinine in discriminating type 2 diabetic patients with a reduced GFR from those with a normal or near-to-normal GFR. The more prominent rise in serum cystatin C values, as found in this study, allows a more rapid diagnosis of decline in GFR, with an earlier therapeutic treatment. Because our results show that cystatin C is superior for renal glomerular function assessment in a well-defined patient group, its measurement may be recommended in the routine management of diabetic patients. Finally, the cystatin C nephelometric assay offers better accuracy and reproducibility than the colorimetric methods for creatinine and, more importantly, is not significantly influenced by interfering substances, as has been previously reported [36]. Further prospective studies are required to evaluate the response of cystatin C concentration to the different renal replacement modalities in ESRD patients. Reprint requests to Mario Plebani, M.D., Azienda Ospedaliera di Padova, Servizio di Medicina di Laboratorio, Via N. Giustiniani, 2, 35128 Padova, Italy. E-mail: [email protected]

REFERENCES 1. Ritz E, Rychlı´k I, Locatelli F, Halimi S: End-stage renal failure in type 2 diabetes: A medical catastrophe of worldwide dimensions. Am J Kidney Dis 34:795–808, 1999 2. Ritz E, Orth SR: Nephropathy in patients with type 2 diabetes mellitus. N Engl J Med 341:1127–1133, 1999 3. Ismail N, Becker B, Strzelczyk P, Ritz E: Renal disease and hypertension in non-insulin-dependent diabetes mellitus. Kidney Int 55:1–28, 1999 4. Mogensen CE, Keane WF, Bennett PH, et al: Prevention of diabetic renal disease with special reference to microalbuminuria. Lancet 346:1080–1084, 1995 5. Jacobs C, Selwood NH: Renal replacement therapy for end-stage renal failure in France: Current status and evolutive trends over the last decade. Am J Kidney Dis 25:188–195, 1995 6. Mogensen CE: Microalbuminuria predicts clinical proteinuria and early mortality in maturity-onset diabetes. N Engl J Med 310:356– 360, 1984 7. Caramori ML, Fioretto P, Mauer M: The need for early predictors of diabetic nephropathy risk: Is albumin excretion rate sufficient? Diabetes 49:1399–1408, 2000 8. Mathiesen ER, Oxenbøll B, Johansen K, et al: Incipient nephropathy in type 1 (insulin dependent) diabetes. Diabetologica 26:406– 410, 1984 9. Lavender S, Hilton PJ, Jones NF: The measurement of glomerular filtration rate in renal disease. Lancet i:1216–1219, 1969 10. Renkin EM, Robinson RR: Glomerular filtration. N Engl J Med 290:785–789, 1974 11. Sawicki PT, Berger M: Measuring progression of diabetic nephropathy. Eur J Clin Invest 24:651–655, 1994 12. Duarte CG: Renal Function Tests. Boston, Little, Brown and Company, 1980, p 320 13. Perrone RD, Madias NE, Levey AS: Serum creatinine as an index of renal function: New insights into old concepts. Clin Chem 38:1933–1953, 1992 14. Young DS: Effects of Drugs on Clinical Laboratory Tests (2nd ed), Washington DC, AACC Press, 1990, p 1465 15. Bernier GM, Cohen RJ, Conrad ME: Microglobulinaemia in renal failure. Nature 218:598–599, 1968 16. Jung K, Schulze BD, Sydow K, et al: Diagnostic value of lowmolecular mass proteins in serum for the detection of reduced glomerular filtration rate. J Clin Chem Clin Biochem 25:499–503, 1987

˚ X-ray crystal structure 17. Bode W, Engh R, Musli D, et al: The 2.0 A of chicken egg white cystatin and its possible mode of interaction with cysteine proteases. EMBO J 7:2593–2599, 1988 18. Saitoh E, Isemura S, Sanada K, Ohnishi K: The human cystatin gene family: Cloning of three members and evolutionary relationship between cystatins and Bowman-Birk type proteinase inhibitors. Biomed Biochim Acta 50:599–605, 1991 19. Olafsson I: The human cystatin C gene promoter: functional analysis and identification of heterogeneous mRNA. Scand J Clin Lab Invest 55:597–607, 1995 20. Cataldi L, Mussap M, Bertelli L, et al: Cystatin C in healthy women at term pregnancy and in their infant newborns: Relationship between maternal and neonatal serum levels and reference values. Am J Perinatol 16:287–295, 1999 21. Bo¨kenkamp A, Domanetzki M, Zinck R, et al: Cystatin C – A new marker of glomerular filtration rate in children independent of age and height. Pediatrics 101:875–881, 1998 22. Grubb A, Lo¨fberg H: Human ␥-trace. Structure, function, and clinical use of concentration measurements. Scand J Clin Lab Invest 45(Suppl 177):7–13, 1985 23. Finney H, Newman DJ, Price CP: Adult reference ranges for serum cystatin C, creatinine and predicted creatinine clearance. Ann Clin Biochem 37:49–59, 2000 24. Simonsen O, Grubb A, Thysell H: The blood serum concentration of cystatin C (␥-trace) as a measure of the glomerular filtration rate. Scand J Clin Lab Invest 45:97–101, 1985 25. Tenstad O, Roald AB, Grubb A, Aukland K: Renal handing of radiolabelled human cystatin C in the rat. Scand J Clin Lab Invest 56:409–414, 1996 26. Maack T, Johnson V, Kau ST, et al: Renal filtration, transport, and metabolism of low-molecular-weight proteins: A review. Kidney Int 16:251–270, 1979 27. Denium J, Derkx FHM: Cystatin for estimation of glomerular filtration rate? Lancet 356:1624–1625, 2000 28. Coll E, Botey A, Alvarez L, et al: Serum cystatin C as a new marker for noninvasive estimation of glomerular filtration rate and as a marker for early impairment. Am J Kidney Dis 36:29–34, 2000 29. Newman DJ, Thakkar H, Edwards RG, et al: Serum cystatin C measured by automated immunoassay: A more sensitive marker of changes in GFR than serum creatinine. Kidney Int 47:312–318, 1995 30. Ravid M, Brosh D, Ravid-Safran D, et al: Main risk factors for nephropathy in type 2 diabetes mellitus are plasma cholesterol levels, mean blood pressure, and hyperglycemia. Arch Intern Med 158:998–1004, 1998 31. Fioretto P, Mauer M, Brocco E, et al: Patterns of renal injury in NIDDM patients with microalbuminuria. Diabetologia 39:1569– 1578, 1996 32. Fioretto P, Steffes MW, Mauer M: Glomerular structure in non proteinuric IDDM patients with various levels of albuminuria. Diabetes 43:1358–1364, 1994 33. Sambataro M, Thomaseth K, Pacini G, et al: Plasma clearance rate of 51Cr-EDTA provides a precise and convenient technique for measurement of glomerular filtration rate in diabetic humans. J Am Soc Nephrol 7:118–127, 1996 34. Solerte SB, Severgnini S, Locatelli M, et al: Nephelometry in the clinical assessment of glomerular proteinuria and tubular function in diabetic nephropathy. Clin Nephrol 48:151–158, 1997 35. Mussap M, Ruzzante N, Varagnolo M, Plebani M: Quantitative automated particle-enhanced immunonephelometric assay for the routinary measurement of human cystatin C. Clin Chem Lab Med 36:859–865, 1998 36. Smith CH, Landt M, Steelman M, Ladenson JH: The Kodak Ektachem 400 analyzer evaluated for automated enzymic determination of plasma creatinine. Clin Chem 29:1422–1425, 1983 37. Cockcroft DW, Gault MH: Prediction of creatinine clearance from serum creatinine. Nephron 16:31–41, 1975 38. Strike PW: Statistical Methods in Laboratory Medicine. Oxford, Butterworth-Heinemann, 1991, pp 108–112 39. Godfrey K: Simple linear regression in medical research, in Medical Uses of Statistics (2nd ed), edited by Bailar JC III, Mosteller F, Boston, NEJM Books, 1992, pp 201–232 40. Hanley JA, McNeil BJ: The meaning and use of the area under

Mussap et al Cystatin C in type-2 diabetes mellitus a receiver operating characteristic (ROC) curve. Radiology 143:29– 36, 1982 41. Gardner MJ, Gardner SB, Winter PD: Confidence Interval Analysis Microcomputer Program. London, British Medical Journal Publishing Group, 1989 42. Hanley JA, McNeil BJ: A method of comparing the area under receiver operating characteristic curves derived from the same cases. Radiology 148:839–943, 1983

1461

43. Harmoinen APT, Kouri TT, Wirta OR, et al: Evaluation of plasma cystatin C as a marker for glomerular filtration rate in patients with type 2 diabetes. Clin Nephrol 52:363–370, 1999 44. Oddoze C, Morange S, Portugal H, et al: Cystatin C is not more sensitive than creatinine for detecting early renal impairment in patients with diabetes. Am J Kidney Dis 38:310–316, 2001 45. Plebani M, Dall’Amico R, Mussap M, et al: Is serum cystatin C a sensitive marker of glomerular filtration rate (GFR)? A preliminary study on renal transplant patients. Ren Fail 20:303–309, 1998