Urinary Albumin Excretion Is Associated with Renal Functional Abnormalities in a Nondiabetic Population

Microalbuminuria (MA; 30 to 300 mg/24 h) is an important early sign of incipient diabetic nephropathy in insulin-dependent diabetes mellitus (IDDM) (1,2) and, albeit less specific, in non—insulin-dependent diabetes mellitus (NIDDM) (3,4). Hyperfiltration has been observed in both IDDM (5,6,7,8) and NIDDM (9,10) and is thought to occur before the renal function loss (11). Therefore, in diabetic patients, the development of nephropathy typically follows a biphasic pattern, characterized by MA and hyperfiltration in the initial phase, followed by progression to overt proteinuria and renal function decline once overt nephropathy ensues.

Whether this biphasic pattern is typical for diabetic nephropathy or can also be observed in nondiabetic renal disease is still unknown. From animal experimental data, it has been suggested that a period of hyperfiltration might precede progressive renal function loss in nondiabetic renal disease (12,13). Furthermore, from other animal experiments, there is evidence that hyperfiltration induces proteinuria (14), suggesting that the concomitant increased intraglomerular pressure in hyperfiltration might be responsible for both albuminuria/proteinuria and the later renal function loss (15,16).

It is interesting that MA also occurs frequently in nondiabetic subjects and is associated with several risk factors, such as hypertension (17), hyperlipidemia (18,19), obesity (20), and older age (21). Little is known about the relation between albuminuria and renal function in nondiabetic subjects, especially whether nondiabetic subjects also exert early renal functional changes such as hyperfiltration, because they usually come to medical attention only after the development of renal damage. We therefore hypothesized that an increased urinary albumin excretion (UAE) is associated with early and late renal functional abnormalities, such as an elevated filtration and a diminished filtration in a nondiabetic population. To test this hypothesis, we studied renal function in relation to UAE in a large cross-sectional study of nondiabetic subjects in the city of Groningen, the Netherlands.

Materials and Methods

Results

The characteristics of our study group according to the four different UAE groups are shown in Table 1. Age, BMI, minimal waist circumference, systolic and diastolic BP, and glucose showed a significant linear trend with each subsequent UAE level. Similarly, the frequencies of male gender, obesity, a positive family history, and the use of antihypertensive treatment also showed a linear trend with each subsequent UAE level. These differences remained statistically significant after adjusting for age and gender. There was no significant difference in smoking among the four different albuminuria groups. Concerning the renal characteristics, serum creatinine also followed a linear trend with each subsequent UAE level (P < 0.001). In contrast, the mean values for the 24-h Ccr followed a parabolic trend with each subsequent UAE level (P < 0.001). In other words, Ccr in the HNA group was elevated as compared with the control group, whereas it showed a decreasing trend in the MA and macroalbuminuria groups as compared with the HNA group.

To investigate whether this observed parabolic pattern could still be found after adjustment for the two most important confounders, age and gender, we analyzed Ccr in a general linear model. This additional analysis supported the finding of this parabolic pattern (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Creatinine clearance (Ccr) according to the four different albuminuria groups. The error bars represent the 95% confidence intervals. The numbers in the bars represent the age-and gender-adjusted mean of the Ccr (ml/min per 1.73m²).

Because the mean Ccr of a certain group is only the sum of subjects with a normal Ccr and subjects with an abnormal Ccr, this does offer optimal information about the relationship between albuminuria and an elevated or a diminished filtration as compared with normal subjects. Therefore, we also analyzed our data for subjects with an elevated or a diminished filtration. In a linear regression model, including age and gender, mean Ccr was calculated with the 2× SD borders. This is shown in Figure 2 for. Mean Ccr was 103 ± 19 ml/min per 1.73m² in males 30 yr of age and 85 ± 19 ml/min per 1.73m² in males 70 yr of age. In females 30 yr of age and 70 yr of age, the Ccr was 101 ± 18 ml/min per 1.73m² and 76 ± 18 ml/min per 1.73m², respectively. The 2× SD borders show the subjects with an elevated Ccr (n = 263) and with a diminished Ccr (n = 181), taking men and women together.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Ccr (ml/min per 1.73m²) as a function of age, with the two lines representing the upper and lower 2× SD borders of the mean according to the regression analysis. (A) Males. (B) Females. The subjects above the upper line represent subjects with an elevated filtration, and the subjects below the lower line represent subjects with a diminished filtration.

To investigate further the relation between an elevated or diminished Ccr and albuminuria, we analyzed our data in a logistic regression analysis (Table 2). We first analyzed our data adjusting only for age and gender. After also adjusting for all other possible confounders, such as BMI, plasma glucose, a positive family history for diabetes, BP, antihypertensive medication, and smoking, the results remained the same. An elevated filtration was associated with albuminuria, BMI, and plasma glucose, whereas it was inversely associated with the use of antihypertensive medication. Furthermore, a diminished filtration was associated with albuminuria, BMI, male gender, and antihypertensive medication, whereas it was inversely associated with plasma glucose. HNA and MA were associated with an increased RR for an elevated filtration (RR 1.8 [95% CI, 1.30 to 2.51], P < 0.001; and RR 1.7 [95% CI, 1.17 to 2.45], P < 0.01, respectively). Macroalbuminuria was also associated with an increased RR for an elevated filtration (RR 1.8 [95% CI, 0.64 to 5.23]), but this was not significant probably because of the small sample size (n = 82). Conversely, macroalbuminuria was associated with an increased RR for a diminished filtration (RR 4.3 [95% CI, 1.97 to 9.36], P < 0.001), whereas HNA was associated with a decreased RR for a diminished filtration (RR 0.3 [95% CI, 0.13 to 0.59], P < 0.001). In other words, HNA and MA are characterized by an elevated filtration, whereas macroalbuminuria is characterized by a diminished filtration. Moreover, a diminished filtration is not observed in association with HNA.

Discussion

This is the first, albeit cross-sectional, study that might shed some light on renal functional abnormalities in nondiabetic subjects in an early phase. Because diabetic subjects are usually followed for their diabetic disease, the course of their renal function is usually well known. Nondiabetic subjects come to medical attention only after the development of renal damage, which explains why little is known about early renal functional abnormalities in these subjects.

This study shows that the GFR in nondiabetic subjects might have a parabolic pattern that is associated with the different levels of UAE. In other words, the MA group has more subjects with an elevated filtration, whereas in the macroalbuminuria group more subjects with a diminished filtration can be found. It is interesting that the association between an elevated filtration and albuminuria was even more pronounced in subjects with a UAE thus far considered to be normal: 15 to 30 mg/24 h.

One could argue that those subjects with an elevated filtration or a diminished filtration have such a filtration only by coincidence, because of miscollection or because of the variability of the measurement. To test the validity of our data, we also analyzed our data after excluding subjects by the Bland-Altman method. The Bland-Altman method investigates the agreement between two measurements, which should measure the same. When there was more than a 2× SD difference of the mean between the two measurements, either Cockcroft versus Ccr, to correct for miscollections, or first versus second 24 h Ccr, to correct for variability, we excluded these subjects from the validity analysis. In other words, we analyzed our data only with subjects in whom the estimated Cockcroft clearancé was similar to the calculated 24-h Ccr and in whom the first versus the second calculated 24-h Ccr was similar. Because there was no difference in outcome, we did the final analysis without excluding any subjects on the basis of these grounds to avoid selection bias. We conducted a second validation by analyzing our data with Ccr not corrected for BSA. The uncorrected Ccr gave, if any, a somewhat higher RR. Because it is known that body mass influences renal function and because subjects with higher rates of albuminuria were also more obese, we thought it more appropriate to use the data of Ccr corrected for BSA.

We observed an increase in both Ccr and serum creatinine in the HNA group as compared with the control group (Table 1). This discrepancy could be due to a difference in body composition between the high-normal albuminuria group and the control group. Males were more prevalent in the HNA group, consistent with a greater muscle mass, which might explain the observed discrepancy.

The observed elevated or diminished filtration in the different albuminuria groups might be explained by other factors, such as age, gender, hypertension or antihypertensive treatment, hyperglycemia, and obesity. Indeed, hypertension (27,28), hyperglycemia (29), and obesity (30) have been described to be related to hyperfiltration, and it has been shown that antihypertensive treatment may also influence GFR (31). Because these factors were more prevalent in the groups with higher UAE, this may have introduced bias in the presented relation between albuminuria and Ccr. However, even after adjusting for systolic and diastolic BP, antihypertensive treatment, plasma glucose, and BMI, the association between albuminuria and an elevated or a diminished filtration persisted. It is therefore suggested that HNA and MA are independently associated with an elevated filtration, whereas macroalbuminuria is independently associated with a diminished filtration. The finding that HNA is associated with a lower risk for a diminished filtration emphasizes the finding that this group is mainly characterized by an elevated filtration.

The parabolic pattern in Ccr as shown in Figure 1 seems similar to the widely known parabolic or biphasic pattern in renal function as observed in diabetes. Although the renal hemodynamic abnormalities in diabetes are likely to be specific for the diabetic condition (32), it has been suggested that determinants of progressive renal function loss are similar for patients with established nondiabetic and diabetic renal disease (33).

One of the underlying mechanisms responsible for the observed early renal functional abnormalities in our nondiabetic subjects may be the consequence of insulin resistance. As mentioned before, subjects with an elevated UAE had higher BP, higher BMI, higher minimal waist circumference, and higher plasma glucose levels. These characteristics have been related to the insulin resistance syndrome (34). The insulin resistance syndrome may not only underlie NIDDM but can also be found in nondiabetic subjects (34). Within this context, MA has been argued to be part of the insulin resistance syndrome in diabetic (35,36,37,38) and nondiabetic subjects (19,39). The underlying mechanism for hyperfiltration or an impaired filtration might be related to hyperinsulinemia, associated with insulin resistance. In both animal (40) and human experimental research (41,42), hyperinsulinemia has been reported to influence GFR and has even been related to hyperfiltration. It has also been argued that insulin may also be responsible for an altered permeability of the glomerular membrane for albumin (43).

Because this study population after the selection procedure was enriched for albuminuria, one could argue that the population was also enriched for prediabetes. After correction in a multivariate model for characteristics of prediabetes, i.e., plasma glucose, BMI, systolic and diastolic BP, and a diabetic family history, the results remained the same. Therefore, there seems to be a robust effect of albuminuria on renal function, independent of prediabetes or insulin resistance. It is, however, difficult to unravel the responsibility of each unique part on renal function in a cross-sectional study. Nevertheless, these were nondiabetic subjects with early renal functional abnormalities in relation to albuminuria, which has not been shown before.

Apart from the observed elevated filtration in the HNA and microalbuminuric subjects, we observed an independent association with a diminished filtration in macroalbuminuric subjects. Taking together this parabolic pattern in Ccr, it is tempting to speculate that the group with “hyperfiltration” will eventually progress to renal functional impairment. However, we conducted only a cross-sectional study and cannot draw that conclusion without a longitudinal follow-up. When a longitudinal follow-up can confirm that subjects with an elevated filtration will indeed progress to renal functional impairment with a concomitant increase in UAE, an elevated UAE might be an important and early sign for progressive renal function loss in a nondiabetic population.

In conclusion, we found an association between UAE and renal function, i.e., an elevated Ccr in the stages of slightly elevated albuminuria and a lower Ccr in groups with higher UAE. Moreover, in these nondiabetic subjects, increased UAE is associated with a metabolic profile resembling the insulin resistance syndrome. Screening for albuminuria may identify nondiabetic subjects who are at risk for developing renal functional loss.