Evidence of Impaired Glomerular Charge Selectivity in Nondiabetic Subjects With Microalbuminuria
Relevance to Cardiovascular Disease
Abstract Microalbuminuria is associated with excess cardiovascular morbidity and mortality in diabetic and nondiabetic subjects. Loss of glomerular charge selectivity may explain the development of microalbuminuria in diabetic subjects. The primary population in this cross-sectional study was 124 subjects aged 40 to 75 years without glucose intolerance and with a previous (3 years before the present study) urinary albumin excretion rate (UAE) in the normal (<20 μg/min) or microalbuminuric (>20 μg/min) range. The secondary population consisted of 39 offspring aged 15 to 40 years. The main outcome measures included UAE, urinary IgG/IgG4 selectivity index (SI), and the presence of ischemic heart disease as determined by questionnaire or ECG. Among the primary population, a significant inverse correlation was found between SI and UAE (r=−.40, P<.001). Reduction in SI could be demonstrated in subjects with UAE >10 μg/min. In multiple regression analysis reduction in SI was found with increasing age, independent of UAE. In 20 subjects with clinical cardiovascular disease a reduction in SI was found (1.9±0.6 versus 2.6±1.3, P=.001) without concomitant elevation in UAE (P=.99). Offspring from parents with microalbuminuria had an SI comparable to offspring from parents with normal UAE (2.7±0.7 versus 3.3±1.6, P=.17). In nondiabetic subjects the development of microalbuminuria is associated with reduced SI, suggesting impairment of glomerular charge selectivity. SI may offer a more sensitive monitoring of abnormalities in glomerular permselectivity than does measurement of UAE, but the ability of SI to predict development of cardiovascular disease needs to be evaluated prospectively.
- Received August 16, 1995.
- Accepted November 20, 1995.
Elevated UAE (microalbuminuria) is associated with the presence of CVD in diabetic as well as nondiabetic subjects.1 2 3 Thus in the Islington Diabetes Study, microalbuminuria was associated with a sixfold increase in prevalence of ischemic heart disease independent from other risk factors.1 In diabetes, microalbuminuria predicts progressive glomerular disease, but the course of albuminuria in nondiabetic subjects remains less clear, as does its underlying mechanism.
The glomerular filtration barrier markedly restricts the passage of albumin due to its size- and charge-selective properties. In insulin-dependent diabetes mellitus patients with elevated UAE, the earliest renal functional abnormality seems to be loss of glomerular charge selectivity.4 5 Whether a similar mechanism governs the development of microalbuminuria in nondiabetic subjects is not known.
Measurement of glomerular charge selectivity should ideally be conducted by using inert tubular probes, eg, differently charged dextrans.6 A less impeccable method, which is more feasible for population studies, is the assessment of the ratio between the urinary excretion of differently charged but similarly sized endogenous proteins, ie, the SI (IgG/IgG4) (IgG: Stokes’ radius 55 Å, pI 7.3 [range, 6 to 10]; IgG4: Stokes’ radius 55 Å, pI 5.8).5 Measurement of the SI may provide evidence of impaired glomerular charge selectivity, which possibly underlies the development of elevated UAE. In addition, as UAE is subject to substantial day-to-day variation, other parameters reflecting glomerular permselectivity may prove more efficient in defining the pathophysiological state in order to resolve the association between glomerular and general vascular dysfunction.
We hypothesized that elevated UAE in nondiabetic subjects who are free from other evidence of renal disease is associated with impairment of glomerular charge selectivity. We further aimed to see whether an association between impaired glomerular charge selectivity and CVD could be demonstrated. As part of an ongoing prospective study evaluating the association between albuminuria and risk of development of CVD, we have measured SI in 124 nondiabetic subjects and in 45 offspring of previously studied nondiabetic subjects with or without elevated UAE. SI was compared with UAE and with evidence of the presence of CHD.
One hundred twenty-four Caucasians aged 40 to 75 years were selected from a cohort of 959 subjects participating in the Goodinge Health Centre Study in North London.7 8 This study (hereafter referred to as “the screening phase”) was conducted 3 years (SD=3.5 months) prior to the present study. Forty-two subjects were selected from the 108 who were microalbuminuric during the screening phase (UAE >20 μg/min in either a 2-hour daytime sample obtained during an oral glucose tolerance test or during a 10-hour overnight sample). Of these 108 subjects, 1 individual with impaired glucose tolerance was excluded, and a random selection of 86 of the remaining subjects (80.5%) was traced. Of these, 3 had died, 35 did not respond to two invitations, 1 had had a stroke, and 5 were on antihypertensives or anticoagulants which could not be stopped, leaving 42 for the present study (55% of those traced). The remaining 82 subjects studied were selected9 by matching for age, gender, BMI, and CHD status from the subjects who were normoalbuminuric at the screening phase (UAE <20 μg/min), including 4 subjects with only one urine sample available at screening. Further characterization of the study population is presented in Table 1⇓.
A secondary study population comprising 39 offspring was identified as follows. During the time of the screening study, offspring (aged 15 to 40 years) of participants found to be microalbuminuric were invited to attend for similar tests. For each subject investigated, 2 to 4 control subjects, representing offspring of normoalbuminuric parents, were selected. At screening 33 Caucasian offspring of microalbuminuric parents and 85 of normoalbuminuric parents were studied, all of whom were reinvited for study at recall. In all, 16 (48%) offspring of microalbuminuric parents and 23 (27%) of normoalbuminuric parents were investigated in this study population.
All subjects included had a normal glucose tolerance at screening as assessed by a 75-g oral glucose tolerance test. Three subjects were on nitrates, 11 on salicylic acid, 2 on lipid-lowering (bezafibrate) drugs, 12 on hormone replacement therapy, and 11 on oral contraceptives. None received antihypertensive medication. Twenty-one subjects were on other drugs, while 103 received no regular medication.
Studies were performed with the approval of the ethics committee of the Islington Health Authority, and all participating subjects recalled gave informed oral consent.
Ischemic Heart Disease, UAE, and SI
At screening a 2-hour urine collection and a timed overnight collection were requested from all participants for calculation of 2-hour and overnight UAE. A history of angina was taken by using the Rose and Blackburn questionnaire, and a 12-lead ECG was classified according to the Minnesota code. CHD was defined as a positive history of angina or myocardial infarction or ECG criteria 1.1 to 1.3, 4.1 to 4.4, 5.1 to 5.3, or 7.1. No ECG was available on six patients.
Recall examinations were conducted at the Department of Medicine, University College of London Medical School, at the Whittington Hospital. All subjects completed a daytime (8 am to 10 pm) and nighttime (10 pm to 8 am) urine collection. The samples were tested for blood, protein, leukocyte, and nitrite levels by using an Ames Multistix (Ames-Miles plc) and, if positive, a midstream specimen was sent for culture. For subjects with a positive culture, another 24-hour urine collection was obtained after treatment of urinary infection.
IgG/IgG4 Assay and Sample Conditions
Urinary and plasma total IgG and IgG4 concentrations were measured by using enzyme-linked immunosorbent assays.10 11 Urine was preserved,10 and samples were shipped on dry ice and assayed within 2 months. Prior to the analysis recall phase, daytime and overnight urine samples were mixed according to the individual volumes to obtain a representative diurnal sample. Venous plasma samples were obtained immediately after the overnight urine sampling. The SI was calculated as (urine/plasma total IgG)/(urine/plasma IgG4). SI could not be calculated in 7 subjects (3 primary study subjects and 4 offspring) because urinary IgG4 concentration was below the assay detection limit of 2 μg/L.
Urinary albumin concentration was measured by using an in-house enzyme-linked immunosorbent assay (similar to that described by Chesham et al12 ) that was validated against a commercial radioimmunoassay (Pharmacia, LKB). The day-to-day variability (coefficient of variation) of UAE for the 124 subjects as calculated from the paired urine samples provided at recall was 28%.
Subjects were weighed and measured with an attached height measure (Seca), and BMI was calculated as body weight in kilograms divided by height in meters squared. Waist and hip circumferences were measured in triplicate by using a steel tape at the level of the umbilicus and greater trochanter, respectively, for the calculation of the waist-hip ratio. BP was measured in duplicate with a random zero sphygmomanometer. A fasting plasma sample was taken for lipid level and glucose estimation by using enzymatic methods.8 9 Any subject with fasting plasma glucose >5 mmol/L had a full glucose tolerance test performed.
Two-tailed parametric tests were used with transformation of log-normal distributed variables. Data are expressed as mean±SD except for UAE which, as a log-normal distributed variable, is expressed as geometric mean×/÷ tolerance factor.
SI and UAE in the Primary Study Population
In the primary study population, SI distribution appeared to be skewed to the right (Fig 1a⇓), but in formal testing it did not significantly differ from the normal distribution (2.5±1.2; Kolmogorov-Smirnov goodness of fit test P>.3 for men and women separately tested). The 24-hour UAE was log-normally distributed (8.5×/÷2.4 μg/min) (Fig 1b⇓).
SI in women was significantly lower than in men (2.2±0.9 versus 2.8±1.3, P=.002), whereas UAE did not differ significantly (women, 7.6×/÷2.0 versus men, 9.4×/÷2.7; P=.17).
SI and UAE were inversely correlated (Fig 2⇓) (r=−.40, P<.001, n=121). The correlation was maintained when only subjects with UAE <20 μg/min were considered (r=−.27, P=.005, n=105) but was of borderline significance in subjects with UAE <10 μg/min (r=−.19, P=.09, n=79). In accordance with the relationship shown in Fig 2⇓, the SI in subjects with elevated UAE at recall (>20 μg/min on 24-hour collection) was significantly reduced compared with normoalbuminuric subjects (1.5±0.5 [n=16] versus 2.7±1.2 [n=105], P<.001).
A significant inverse correlation was found between age and SI (r=−.21, P=.016, n=121), whereas UAE displayed only a borderline significant positive correlation with age (r=.17, P=.062, n=124). The partial correlation between SI and UAE when controlling for age remained significant (r=−.38, P<.001, n=118).
Patients were selected for this study based on whether they were microalbuminuric or normoalbuminuric at screening (see “Methods”). A weak correlation between overnight UAE at screening and current UAE (r=.26, P=.004, n=119) was found, and a considerable number of participants initially classified as microalbuminuric reverted to normoalbuminuria (30 of 42) and vice versa (12 of 78). The UAE at recall of participants who were reclassified from previous microalbuminuria to current normoalbuminuria did not differ significantly from those with persistent normoalbuminuria (6.1×/÷1.7 versus 6.0×/÷1.6, P=.9), whereas SI was marginally reduced (2.4±0.9 versus 2.9±1.3, P=.064) (Table 2⇓). Those who converted from normoalbuminuria to microalbuminuria had SIs that were similar to the persistently microalbuminuric subjects (Table 2⇓). In a subgroup of the present population (n=17), the day-night coefficient of variation of UAE (median, 14.7%; range, 0% to 141%) was found to exceed that of day-night SI variation (median, 3.1%; range, 0% to 18%), suggesting that SI may serve as a more stable parameter.
SI and CHD
Twenty subjects had clinical CVD (history of angina, myocardial infarction, or positive ECG criteria). They were slightly older than subjects without significant CVD and had higher BPs. SI was significantly reduced in patients with CHD (P=.001), whereas UAE was similar in subjects with and without CHD (P=.99) (Table 3⇓ and Fig 3⇓). The relationship between CHD and SI remained of borderline significance after adjusting for age and gender (P=.076).
No significant associations between SI and recorded risk factors for development of CHD (serum triglyceride, total cholesterol, or HDL cholesterol levels, BP, and BMI) were observed (data not shown).
SI in Offspring
Offspring of parents with microalbuminuria during the study screening phase had marginally low SI (2.7±0.7) compared with offspring from parents with normal UAE (3.3±1.6; P=.17 when corrected for variance inhomogeneity). UAE was not elevated in offspring of microalbuminuric parents (6.2×/÷1.5 versus 8.4×/÷2.1, P=.13).
No difference in age or BP was found between these two groups of offspring (data not shown).
This is the first study to evaluate glomerular charge selectivity in relation to UAE, CHD, and inheritance in nondiabetic subjects. We found elevation of UAE within the microalbuminuric range to be associated with a reduction in SI. In accordance with our hypothesis, elevated UAE in nondiabetic subjects without other evidence of renal disease may thus be explained by impairment of glomerular charge selectivity. In addition, subjects with evidence of CVD had a reduced SI without measurable elevation of UAE, although the reduction was of only borderline significance after adjustment for differences in age and gender. The reduced SI among women was an unexpected finding.
The SI, which reflects net renal charge selectivity, has been used in studies of the early phases of the development of diabetic nephropathy. SI may not solely reflect specific changes in glomerular handling of differently charged molecules, but it may under certain conditions reflect subtle changes in glomerular size selectivity.13 In addition, as tubular reabsorption is to some extent influenced by charge14 (anionic proteins having reduced reabsorption rates), even nonglomerular pathophysiology may influence our measurements. Since specific measurements of glomerular size selectivity or tubular function were not undertaken, we may not be able to distinguish these factors from changes specifically related to glomerular charge selectivity. Ascribing the finding of reduced SI to a glomerular event allows for the interpretation of a loss of function, whereas ascribing it to some tubular dysfunction would imply increased capability in the sense that increased tubular charge selectivity would have to be invoked to explain a relative increase in urinary excretion of the anionic IgG4. We think it more likely that the present changes in SI reflect glomerular malfunction, principally involving renal charge selectivity.
We defined microalbuminuria as UAE >20 μg/min, taking the definition from studies related to the development of diabetic nephropathy, which has also been highly predictive of CVD in a nondiabetic population.1 In the nondiabetic population, however, as well as recently in non–insulin-dependent diabetes mellitus patients, others have demonstrated significant associations between CVD and lower UAE rates (>5 to 10 μg/min).15 16 Our present finding of reduced SI in subjects with UAE >10 μg/min might suggest that choosing a lower cutoff level than 20 μg/min to define microalbuminuria might increase sensitivity for predicting CVD (eventually at the expense of specificity). We consequently reexamined our data by analyzing SI in relationship to changes from normoalbuminuria to microalbuminuria using a cutoff level of 10 μg/min (data not shown); the results were essentially identical to those presented.
The mechanism underlying the impairment of glomerular charge selectivity remains unknown. In the case of diabetic nephropathy, loss of glomerular charge selectivity has been suggested as arising from loss of glomerular basement membrane heparan sulfate. In relation to our present finding of an apparent association between low SI and the presence of CHD, we note with interest that a generalized loss of heparan sulfate at the vascular endothelial surface and extracellular matrix has been suggested as a key pathobiochemical event in explaining the development of premature atherosclerosis in patients with diabetic nephropathy. Another characteristic of patients with diabetic nephropathy, the general increase in the transcapillary albumin escape rate, may also be found in otherwise healthy nondiabetic subjects with elevated UAE.17 However, as the etiology of the diabetic glomerular disease includes hyperglycemic damage, and so may be very different from that of a nondiabetic subject with minor elevation in UAE, it would seem hazardous to postulate similar fundamental biochemical changes.
The lower SI in subjects with CHD deserves further attention. First, it indicates ongoing abnormal glomerular selectivity, which is in keeping with the supposed generalized vascular leakage in this group of patients. Second, it is displayed in the absence of any differences in UAE, suggesting that SI may be more sensitive to changes in glomerular function in relation to development of general vascular disease. Whether low SI could somehow be a more sensitive marker of ongoing CVD than that of elevated UAE remains to be established.
In a large cohort of young normoalbuminuric short-term insulin-dependent diabetes mellitus patients, low SI was found to be associated with male gender,18 which was thought to be in good agreement with the known excess male risk for developing diabetic glomerular disease. How the present finding of reduced SI in women should be interpreted is not clear, although we note that women in the present study have a relatively high prevalence of CVD.
No significant reduction in SI among offspring of microalbuminuric parents was found, nor did UAE differ between the offspring groups. A limited number of offspring and parent pairs (19 offspring from 13 parents) participated in the recall study; the majority (n=11) of these parents belonged to the group initially classified as microalbuminuric but presently normoalbuminuric.
In conclusion, microalbuminuria in nondiabetic subjects is associated with loss of glomerular charge selectivity. The present results suggest that low SI could be a more sensitive marker of prevalent CVD than elevation of UAE. Nevertheless, the predictive value of low SI in relation to the development of CVD needs to be evaluated in a prospective study.
Selected Abbreviations and Acronyms
|BMI||=||body mass index|
|CHD||=||coronary heart disease|
|SI||=||urinary IgG/IgG4 selectivity index|
|UAE||=||urinary albumin excretion rate|
We would like to thank the Wellcome Trust for support for the screening phase and the British Heart Foundation for funding the recall phase of this study. We are also grateful to Diabetes and Related Diseases Research, the Mrs Joan Oliver Bequest, and the Mrs Sue Hammerson Trust for additional funding and support for the study.
Damsgaard EM, Frøland A, Jørgensen OD, Mogensen CE. Microalbuminuria as predictor of increased mortality in elderly people. BMJ. 1990;300:297-300.
Guasch A, Deen WM, Myers BD. Charge selectivity of the glomerular filtration barrier in healthy and nephrotic humans. J Clin Invest. 1993;92:2274-2282.
Gould MM, Mohamed-Ali V, Goubet SA, Yudkin JS, Haunes AP. Microalbuminuria: associations with height and sex in non-diabetic subjects. BMJ. 1993;306:240-242.
Foyle WJ, Carstensen E, Fern·ndez MC, Yudkin JS. Longitudinal study of associations of microalbuminuria with the insulin resistance syndrome and sodium-lithium countertransport in non-diabetic subjects. Arterioscler Thromb Vasc Biol. 1995;15:1330-1337.
Kofoed-Enevoldsen A, Jensen K, Beck TC. Measuring urinary IgG and IgG4 excretion. Clin Chem. 1991;37:1136-1137.
Chesham J, Anderton SW, Kingdon CFM. Rapid competitive enzymoimmunoassay for albumin in urine. Clin Chem. 1986;32:669-671.
Christensen EI, Rennke HG, Carone FA. Renal tubular uptake of protein: effect of molecular charge. Am J Physiol. 1983;244:F436-F441.
Kofoed-Enevoldsen A, Jensen K, Deckert T. Glomerular charge selectivity in normoalbuminuric type 1 diabetic patients. Diabetologia. 1992;35(suppl 1):A144. Abstract.