Dm

ORIGINAL ARTICLES

Increased Urinary Excretion of Transferrin in Children with Type 1 Diabetes Me1Iitus P. Martin”, C. Walton”, C. Chapmanb, H.J. Bodansky”, M.H. Stickland” ”University Department of Medicine and bDepartment of Chemical Pathology, The General Infirmary, Leeds, UK

Urinary transferrin excretion was measured by radioimmunoassay in 74 children with Type 1 diabetes mellitus and in 40 normal children, and compared with urinary excretion of albumin, a-1-microglobulin, and N-acetyl-p-D-glucosaminidase. Urinary transferrin excretion was significantly elevated in diabetic (median (range) 186 (18-1671) mg molcreatinine-l) compared with normal (85 (27-668) mg mol-creatinine-l) children (p < 0.001). Seventeen diabetic children had transferrin excretion above the 95th centile for normal children. I n contrast there was no significant increase in urinary albumin excretion in the diabetic children although 8 had urinary albumin excretion which exceeded the 95th centile for normal children (6 of these 8 patients having coexistent urinary hyperexcretion of transferrin). Urinary transferrin excretion correlated significantly with urinary albumin excretion in both normal (rs = 0.62, p < 0.001) and diabetic (r, = 0.61, p < 0.001) children. The indices of proximal renal tubular function (urinary excretion of a-1-microglobulin and N-acetyl-p-D-glucosaminidase) correlated significantly with transferrin excretion in both diabetic (rs= 0.43 and r, = 0.41, p < 0.001) and normal (r, = 0.40, p < 0.02 and r, = 0.53, p < 0.001) children, but not with albumin excretion (r, = 0.20, p > 0.05 and r, = 0.22, p > 0.05). In addition urinary transferrin excretion significantly correlated with urinary glucose concentration (r, = 0.34, p < 0.007) in Type 1 diabetic children. The discrepancy in urinary excretion of transferrin and albumin may reflect impaired proximal renal tubular reabsorption of transferrin and/or altered glomerular basement membrane selectivity for the two proteins. KEY WORDS

T v w 1 diabetes mellitus Children N-acetvl-B-D-nlucosaminidase Transferrin Microalbuminuria a-1-Microglobulin

Introduction In adults with Type 1 diabetes mellitus a raised albumin excretion rate (>30 pg rnin-’ , ‘microalbuminuria‘) predicts progression to persistently Albustix-positive proteinuria, an ominous sign antedating the onset of renal failure by 4 to 5 years.’ Raised urinary excretion of albumin has also been reported in 10-20 % of children with Type 1 diabete~,~ although ,~ the pathogenesis remains unclear. Diabetic nephropathy is characterized by the excessive excretion of protein, generally thought to be related to increased glomerular permeability to p r ~ t e i n .Support ~ for this assumption i s derived from studies identifying elevated glomerular filtration rate early in the course of Type 1 diabetes5and the fact that antihypertensive drugs decrease urinary albumin excretion in both diabetic patientsf6 and others with essential hypertension.’ Furthermore, subtle basement membrane changes in the glomerular capillaries have also been found early in the course of diabetes.8 This interpretation of the early rise in the urinary excretion of albumin being due to increased glornerular permeability takes no account of possible alterations in the tubular component of protein handling; Correspondence to: Mr P. Martin, University Department of Medicine, G Floor, Martin Wing, The General Infirmary, Leeds, LS1 3EX, UK

0742-3071 /90/010035-06$05 .OO

@ 1990 by John Wiley & Sons, Ltd.

-

it i s widely understood that in excess of 95 % of filtered albumin is reabsorbed in the proximal convoluted tubules.9,1 O In the light of recent studies identifying patterns of low molecular weight (tubular) microproteinuria in diabetic children’ 1 , 1 2 we have investigated the relationship between the indices of renal glomerular (albumin, -microglobulin, N-acetyl-Ptransferrin) and tubular (a-1 D-glucosaminidase) handling in a young cohort of children with Type 1 diabetes of short duration. Urinary transferrin excretion was measured as an adjunct to urinary albumin excretion to provide a more detailed profile of glornerular basement membranecharge selectivity for anionic proteins. a-1-Microglobulin and N-acetyI-p-D-glucosaminidase were measured as indices of renal tubular reabsorptive capacity and integrity, respectiveiy .

Methods With local ethical committee approval, 74 children with Type 1 diabetes rnellitus (38 male, 36 female) were studied. All were attending diabetic clinics in Leeds or Huddersfield and no selection criteria were applied other than age less than 16 years (mean r SD 10.4 r 3.5 Accepted 6 September 1989 DIABETIC MEDICINE, 1990; 7: 35-40

35

DrTl

ORIGINAL ARTlCLES years). Duration of diabetes was 3.8 2 2.8 years and glycosylated haemoglobin (HbA1,) was 10.4 5 1.8 % (normal reference range 5.6-7.4 %). A control group of 40 normal children (19 male, 21 female) who were neither attending hospital nor had a family history of diabetes in a first degree relative were studied. Their age was 9.2 ? 3.3 years. The first voided morning urine sample was collected (subject to sterile bacteriological culture) and stored at -20 “C using J g I-’ sodium azide as preservative. Urinary excretion of a protein was determined by calculation of the ratio of that protein to creatinine in the first morning urine specimen. First morning urine samples were preferred to timed overnight collections because in children urinary a1bumin:creatinine ratios in first morning samples d o not change with age.13 In contrast, albumin excretion rates increase with age and correlate with height, weight, and surface area.14 Transferrin (mol. wt. 77 000 Daltons) was measured by double antibody radioimmunoassay. Antiserum to human serum transferrin (Sigma, Poole, UK) was produced in rabbits following subcutaneous multisite injection of 1 mg kg-’. ’251-transferrin label was prepared by the liquid-phase lodogen r n e t h ~ d ’and ~ purified on a Sephadex G-75 (Pharmacia, Uppsala, Sweden) 25 ml column. Specific activity of the tracer was 22 Ci g-’. Rabbit antiserum to human serum transferrin was diluted 25 000-fold in assay buffer (phosphate buffer, 50 mmol I-’, pH 7.5, with 1 g I-’ bovine serum albumin) containing rabbit carrier serum (RCS) optimized against precipitating antibody (2.5 ml I-’). The diluted antiserum (400 pl) was incubated at 25 “C with 50 pl of human serum transferrin standard or unknown and 100 FI transferrin tracer (20 000 counts) solution for 18 h. Sheep antiserum (100 pi of 60 ml 1 - l ) to rabbit IgG (Sigma, Dorset, UK) (raised by subcutaneous multisite injection of 0.3 mg kg-’) in 500 ml I-’ horse serum/phosphate buffer was then added to each sample immediately followed by 500 pI of 100 g I-’ polyethyleneglycol, incubated for 4 0 min at 25 “C, centrifuged and the precipitate counted. Urine specimens were diluted 5fold in assay buffer initially and further diluted where appropriate. Recovery was 98.2 % (range 96.1101.2 %). Cross-reactivity of the antibody with albumin was < 0.1 % at 2 g IF1 albumin. Range of assay was Inter-assay coefficients of variation 70-2200 pg I-’. were: at 100 p,g I-’ 11.8 %, at 290 pg I-’ 5.6 %, at 710 pg I-’ 4.3 %. Albumin was measured by radioimmunoassay (Alb RIA 100, Pharmacia, Uppsala, Sweden); a-1-microglobulin by radial immunodiffusion16 (antiserum and standards supplied by Behringwerke (Marburg, Lahn, FRG); Nacetyl- P- D-glucosaminidase by spectr~fluorimetry;’~ creatinine by the Jaffe method; and glucose by an autoanalyser using a glucose oxidase method. Venous blood samples collected within 24 h of urine samples were assayed for glycosylated haemoglobin (HbA,,) by isoelectric focusing.’8 The inter- and intra-assay 36

coefficients of variation for all methods were less than 15 %. Statistical analyses were performed using the MannWhitney U test, the Spearman Rank correlation test, and multiple linear regression. A two-tailed value was taken for estimation of p. Results are expressed as median (range).

Results There were no significant differences between boys and girls for any of the urinary indices measured, and neither age nor duration of diabetes significantly correlated with urinary excretion of any of the proteins. However, highly significant correlations were observed between the respective urinary protein concentrations and protein:creatinine ratios (r, = 0.58-0.82, p < 0.001 in all cases) and, furthermore, urinary excretion of transferrin, a-1microglobulin and N-acetyl-P-D-glucosaminidase (but not albumin) were significantly elevated in diabetic compared with normal children ( p < 0.001 in all cases). Urinary excretion of transferrin was significantly higher in diabetic than in normal children (186(18-1671) vs 85(27-668) mg mol-creatinine-’; p < 0.001 ), but there was no significant difference in urinary albumin excretion (0.89(0.06-6.85) vs 0.82(0.11-3.67) g mol-creatinine-’; 3 > 0.15). The data from normal children were used to construct centile lines (using a cumulative frequency chart) for urinary excretion of transferrin and albumin. Seventeen diabetic children were above the 95th centile for urinary transferrin excretion (Figure l(a)) and 8 diabetic children were above the 95th centile for urinary albumin excretion (Figure 1 (b)).Of the 8 diabetic children with elevated urinary albumin excretion 6 also had abnormally increased urinary transferrin excretion (Figure 2). Urinary excretion of a-1-microglobulin (Figure 1(c)) and of N-acetyl-P-D-glucosaminidase (Figure 1 (d)) was significantly higher in diabetic than in normal children ( p < 0.001 in both cases). In diabetic children urinary a-1-microglobulin excretion was 0.56 (0.1 1-1.92) g molcreatinine-’ and urinary N-acetyl-P-D-glucosaminidase excretion 177 (1 5-642) U mmol-creatinine-’. In normal children urinary a-1-microglobulin excretion was 0.23 (0.1 1 4 . 9 0 ) g mol-creatinine-’ and urinary N-acetyl-PD-glucosaminidase excretion 36 (15-202) U mmolcreati n i ne- . On univariate analysis the strongest correlation of transferrin was with urinary albumin excretion in both diabetic and normal children (Table 1). A weaker correlation was present between urinary transferrin and both urinary a-1-microglobulin and urinary N-acetyl-PD-glucosaminidase excretion but there was no correlation between albumin excretion and either urinary a-1microglobulin (r, = 0.20, NS) or urinary N-acetyl-p-Dglucosaminidase excretion (r, = 0.22, NS). In children with Type 1 diabetes HbA,, significantly P. MARTIN €7 AL.

DTT7

O R I G I N A L ARTICLES

l-1

c

l0-l

-normal

diabetic

1

0.OlJ

-diabetic

=I

(C)

i

--f---------

=%

2

’

diabetic’

normal

Figure 1. Urinary excretion of transferrin ( (a),p < 0.001), a-1-microglobulin ( (c), p < 0.001), and N-acewl-@-D-glucosaminidase( (d), p < 0.001) was significantly increased in Type 1 diabetic children (closed circles) compared with normal children (open circles) in the absence of significantly altered urinary albumin excretion ( (b),p > 0.1 5 , NS). Bars indicate median values. Broken lines indicate the 95th centiles for normal children

correlated with urinary excretion of a-1-microglobulin (r, = 0.34, p < 0.001), N-acetyl-P-D-glucosaminidase (r, = 0.36, p < 0.001) and albumin (r, = 0.27, p < 0.05) but not with urinary transferrin excretion (r, = 0.01). Urinary glucose concentration was 61.5 (0.1-465.0) mmol I-’ in diabetic children and this significantly correlated with urinary excretion of a-1-microglobulin (r, = 0.52, p < 0.001), N-acetyl-p-D-glucosaminidase (r, = URINARY EXCRETION OF TRANSFERRIN

0.69, p < 0.001) and transferrin (Table 1 ) but not with urinary albumin excretion (r, = 0.21). Due to the multicolinearity of glucose, a-1-microglobu.lin and N-acetyl-PD-glucosaminidase excretion multiple regression analysis revealed urinary transferrin excretion to be significantly associated with urinary excretion of albumin ( p < 0.001) and glucose ( p < 0.01) but not with a-1-microglobulin or N-acetyl-P-D-glucosaminidaseurinary excretion.

37

** 0.1

..

J . I

*

I1111l11

.

I

I I I I

I I

I1111111

100

10

I

I

lllrm

1000

10000

Tronsferrin (rng rnol-creotinine-')

Figure 2. Correlation of urinary transferrin with urinary albumin excretion in Type 1 diabetic children (r, = 0.61, p < 0.001). Broken lines indicate 95th centiles for normal children

Discussion Concern has previously been expressed as to the appropriateness of various methods for calculation of urinary protein excretion rates. The validity of urinary creatinine concentration as a correction factor for first voided morning urine samples has been questioned due to the net renal tubular secretion of creatinine seen in healthy subjects. l 9 In this study, however, the highly significant correlations observed between the respective urinary protein concentrations and pr0tein:creatinine ratios would seem to indicate that the renal tubular handling of creatinine was not a relevant factor. In diabetic children the higher incidence of elevated urinary transferrin excretion (23 %) compared with urinary albumin excretion (11 %) is a surprising finding given that the molecular radius of transferrin (3.8 nm) is greater than that of albumin (3.6 nm).20It would therefore seem unlikely on the basis of this evidence that an increase in glomerular filtration rate could account for these findings.

In the present study serum concentrations of transferrin and albumin were not measured. However, Ellis et a/. have previously demonstrated no relationship between the respective serum and urinary concentrations of transferrin or albumin in children with diabetes.20 Clycosylation-induced changes in the molecular charge of proteins have been reported by several workers. have demonstrated that albumin, when Giggheri et glycosylated, becomes even more anionic and thus more repelled by the glomerular basement membrane polyanion. However transferrin (pl 5.71, which is much less anionic than albumin (pl 4.9), has been shown to be electrophoretically immobile in diabetic patients under conditions where glycosylation of albumin has occurred.22 While these findings may help to explain a change in the relative ratio of urinary excretion of transferrin:albuinin, they do not account for the real increase in urinary transferrin excretion observed in these diabetic children compared with normal control children. Increased urinary transferrin excretion has been reported in diabetic childrenz0but there are no previously published data from children of this age and no reports of elevated urinary transferrin excretion with reliable evidence of accompanying proximal renal tubular dysfunction. Evidence for renal tubular dysfunction in diabetes mellitus has, until recently, been provided by studies of the unreliable urinary marker P-2-microglobulin. However, several studies in Type 1 diabetic adults have shown normal 24-h urinary excretion of this p r ~ t e i n . ~ ~Furthermore, -~~. reported have usually been associated with poor metabolic control, ketoacidosis or exercise-induced hyperexcretion of p-2microglobulin. P-2-Microglobulin is markedly acid-labile and therefore a significant false-negative rate is to be expected whenever it it used as a marker. Alkalinization of the urine does not resolve this problem as a considerable period of time may elapse between the formation of the urine by the kidneys and micturition. The renal tubular markers examined in this paper, a1-microglobul in29 and N-acetyl-~-D-gluco~aminidase,~~ have been found to be more reliable and are stable at low pH. Univariate correlation of urinary transferrin with these indices of proximal tubular function in both diabetic

Table 1. Spearman rank correlations of urinary transferrin excretion in diabetic and nondiabetic children Second variable Albumin a-1 -Microglobulin N-acetyl-P-D-glucosaminidase

Urinary glucose HbA,' Age

Duration of diabetes

Diabetic children rS P 0.61 0.43 0.41 0.34

0.01 0.18 0.1 1