Diabetes-Associated Autoantibodies in Relation to Clinical ...

10 downloads 0 Views 161KB Size Report
shown recently in first-degree relatives of patients with type. 1 diabetes ... per 24 h, and a blood sample for glycated hemoglobin (GHb) and serum. C-peptide ...
0021-972X/99/$03.00/0 The Journal of Clinical Endocrinology & Metabolism Copyright © 1999 by The Endocrine Society

Vol. 84, No. 5 Printed in U.S.A.

Diabetes-Associated Autoantibodies in Relation to Clinical Characteristics and Natural Course in Children with Newly Diagnosed Type 1 Diabetes* EMAD SABBAH, KAISA SAVOLA, PETRI KULMALA, RIITTA VEIJOLA, ¨ HA ¨ SALO, JUKKA KARJALAINEN, HANS K. ÅKERBLOM, PAULA VA MIKAEL KNIP, AND THE CHILDHOOD DIABETES IN FINLAND STUDY GROUP† Department of Pediatrics (E.S., K.S., P.K., R.V., P.V., J.K., M.K.), University of Oulu FIN-90220, Oulu; the Children’s Hospital (H.K.Å.), University of Helsinki, FIN-00290 Helsinki; Medical School, University of Tampere and Department of Pediatrics, Tampere University Hospital (M.K.), FIN-33101 Tampere, Finland ABSTRACT We analyzed 747 children, younger than 15 yr of age, with newly diagnosed diabetes, for antibodies to glutamic acid decarboxylase (GADA), the IA-2 protein (IA-2A), insulin (IAA), and islet cells, to evaluate the influence of positivity for GADA, IA-2A, IAA, or multiple ($3) autoantibodies at diagnosis, on the clinical presentation and natural course of the disease over the first 2 yr and to characterize autoantibody-negative patients. At diagnosis, 73.2% of the children tested positive for GADA, 85.7% for IA-2A, 54.2% for IAA, and 72.6% for multiple autoantibodies. Only 17 subjects (2.3%) had no detectable autoantibodies. The patients testing positive for multiple autoantibodies were younger than the remaining children (P , 0.001). A similar age difference was seen when comparing IAA-positive and -negative patients (P , 0.001). There was no significant difference between the GADA-positive and -negative subjects in the degree of metabolic decompensation at diagnosis, whereas those testing positive for IA-2A had reduced serum C-peptide concentrations (P 5 0.003), and those positive for IAA had lower glycated hemoglobin values. The patients with no detectable autoantibodies had higher serum C-peptide levels (P 5 0.007) at diagnosis than did the other

T

YPE 1 DIABETES is thought to be caused by an autoimmune process that destroys the pancreatic b-cells (1, 2). The sequence of this process is still mostly unknown, and it may take years before b-cell destruction has proceeded so far that the disease becomes overt. Young age and definite signs of islet cell-directed autoimmunity are known to be factors that affect the pace of the b-cell destructive process (3). An increasing number of circulating diabetes-associated Received March 16, 1998. Revision received May 19, 1998. Rerevision received January 20, 1999. Accepted February 2, 1999. Address all correspondence and requests for reprints to: Dr. M. Knip, Medical School, University of Tampere, P.O. Box 607, FIN-33101 Tampere, Finland. E-mail: [email protected]. * This research was supported by the Maud Kuistila Foundation, Helsinki, Finland, the Foundation for Diabetes Research in Finland, the Juvenile Diabetes Foundation International (Grants 188517 and 197032), the Novo Nordisk Foundation, the Medical Research Council of the Academy of Finland (Grants 26109, 32757, and 44718), and a grant (to E. S.) from the Centre for International Mobility, Helsinki, Finland. The Childhood Diabetes in Finland project has been supported by grants from the National Institutes of Health (Grant DK-37957), the Sigrid Juse´lius Foundation, the Association of Finnish Life Insurance Companies, the University of Helsinki, and Novo Nordisk A/S, Bagsvaerd, Denmark.

subjects. The children initially positive for IA-2A had decreased serum C-peptide concentrations at 24 months (P 5 0.045), and their daily insulin dose was higher at 18 (P 5 0.005) and 24 months (P , 0.001). The patients who tested positive for multiple autoantibodies at diagnosis had decreased serum C-peptide levels (P , 0.001) and higher insulin doses (P 5 0.005) at 12, 18, and 24 months. A lower proportion of them were also in clinical remission at 12 and 18 months (P 5 0.01). Autoantibody-negative subjects needed less exogenous insulin at 6 and 18 (P 5 0.01) and at 24 months (P , 0.001) than the other subjects, and a higher proportion of them were in clinical remission at 18 months (P , 0.001). We conclude that positivity for multiple diabetes-related autoantibodies is associated with accelerated b-cell destruction and an increased requirement for exogenous insulin over the second year of clinical disease, indicating that multiple autoantibodies reflect an aggressive progression to total b-cell destruction. Patients testing negative for diabetes-associated autoantibodies at diagnosis seem to have a milder degree of b-cell destruction, but their metabolic decompensation is similar to that seen in other affected children, suggesting that they do represent classical type 1 diabetes. (J Clin Endocrinol Metab 84: 1534 –1539, 1999)

autoantibodies have been identified over the last 20 yr, after the initial description of islet cell antibodies (ICA) in the sera of patients with autoimmune polyendocrinopathy (4). So far, insulin is the only fully characterized autoantigen specific to b-cells, whereas other antigens, like glutamic acid decarboxylase (GAD) (5) and IA-2 [the intracellular portion of a protein tyrosine phosphatase (6)], can be detected also in several other tissues (6, 7). Diabetes-associated autoantibodies are, in general, considered as biological indicators of b-cell damage, without † The DiMe Study Group is composed of the following members: principal investigators: H. K. Åkerblom and J. Tuomilehto; coordinators: R. Lounamaa and L. Toivonen; data management: E. Virtala and J. Pitka¨niemi; local investigators: A. Fagerlund, M. Flittner, B. Gustafsson, M. Ha¨ggquist, A. Hakulinen, L. Herva, P. Hiltunen, T. Huhtama¨ki, N.-P. Huttunen, T. Huupponen, T. Joki, R. Jokisalo, M.-L. Ka¨a¨r, S. Kallio, E. A. Kaprio, U. Kaski, M. Knip, L. Laine, J. Lappalainen, J. Ma¨enpa¨a¨, A.-L. Ma¨kela¨, K. Niemi, A. Niiranen, A. Nuuja, P. Ojaja¨rvi, T. Otonkoski, K. Pihlajama¨ki, S. Po¨ntynen, J. Rajantie, J. Sankala, J. Schumacher, M. Sillanpa¨a¨, M.-R. Ståhlberg, C.-H. Stråhlmann, T. Uotila, M. Va¨re, P. Varimo, and G. Wetterstrand; special investigators: A. Aro, M. Hiltunen, H. Hurme, H. Hyo¨ty, J. Ilonen, J. Karjalainen, M. Knip, P. Leinikki, A. Miettinen, T. Peta¨ys, L. Ra¨sa¨nen, H. Reijonen, A. Reunanen, T. Saukkonen, E. Savilahti, E. Tuomilehto-Wolf, P. Va¨ha¨salo, and S. M. Virtanen.

1534

AUTOANTIBODIES IN CHILDREN WITH DIABETES

playing any active role in tissue destruction. Most earlier studies have indicated that initial positivity for ICA in patients with recent-onset type 1 diabetes is associated with decreased endogenous insulin secretion after the diagnosis (8, 9), whereas insulin autoantibodies (IAA) seem to be poor predictors of the clinical course of the disease (10). It has been shown recently in first-degree relatives of patients with type 1 diabetes that positivity for two or more autoantibody specificities is a stronger predictive marker for the development of diabetes than is positivity for a single type of autoantibody (11–13). Accordingly, it can be hypothesized that multiple autoantibodies may reflect a more rapid and aggressive b-cell destruction. In the present study, children with diabetes were evaluated for the presence of circulating antibodies to GAD (GADA), IA-2 (IA-2A), ICA, and IAA at the time of diagnosis and were observed, for the initial 2 yr, to explore whether these diabetes-related autoantibodies are associated with the clinical features at diagnosis and the natural history of the disease thereafter. In addition, we characterized those children who presented with clinical diabetes despite testing negative for all four diabetes-associated autoantibodies analyzed. Materials and Methods Subjects The study population comprised 747 children under the age of 15 yr, representing 93.3% of the 801 probands in whom diabetes was diagnosed during the recruitment period of the Childhood Diabetes in Finland (DiMe) study from the beginning of September 1986 to the end of April 1989. The background and design of the DiMe study have been described in detail previously (14). The mean age of the children was 8.4 yr (range, 0.8 –14.9 yr) at diagnosis, and slightly over half of them were boys (n 5 411; 55.0%). The probands were monitored regularly at their clinical visits, and data were collected in this study at 6-month intervals over a period of 2 yr. The daily insulin dose was recorded at each visit, expressed in international units per kilogram of body weight (IU/kg) per 24 h, and a blood sample for glycated hemoglobin (GHb) and serum C-peptide was taken. The sera of the blood samples obtained at diagnosis and at the follow-up examinations were stored at 220 C until analyzed. The research design was approved by the ethical committees of all the participating hospitals. The degree of consciousness at diagnosis was assessed by the clinician examining the patient at the time of hospital admission. Consciousness was estimated to be either normal or impaired. The degree of dehydration was evaluated from clinical signs and the percent weight loss. A weight loss of 8% was considered to represent severe dehydration in children younger than 12 yr of age, whereas the limit for older subjects was 7%. If there was a discrepancy between the degree of dehydration based on clinical signs and that assessed from weight loss, the former criterion was always used. Diabetic ketoacidosis was defined as a capillary or venous blood pH of less than 7.30, because data on ketonemia was available in a limited number of children. Clinical remission was defined as a period characterized by a daily insulin dose of less than 0.5 IU/kgzday and a GHb value lower than the mean 1 3 sd for nondiabetic subjects.

GADA GADA were measured with a radioligand assay, as described earlier (15, 16). All the samples were analyzed in quadruplicate with and without competition from an excess of unlabeled purified human recombinant GAD65 (1 mg/well) produced in baby hamster kidney cells. The results were expressed in relative units (RU), representing the specific binding as a percentage of that obtained with a positive standard serum. The cutoff limit for antibody positivity was set at the 99th per-

1535

centile in 372 nondiabetic children and adolescents, i.e. 6.6 RU. The interassay coefficient of variation was 18% at a GADA level of 14.6 RU and 12% at GADA levels exceeding 100 RU. The disease sensitivity of the present assay was 79%, and the specificity was 97%, based on the 1995 Multiple Autoantibody Workshop (17).

IA-2 antibodies IA-2 antibodies were analyzed with a radiobinding assay, as described in detail elsewhere (18). The results were expressed in RU, based on a standard curve run on each plate, using an commercial software program (MultiCalc, EG&G Wallac, Inc., Turku, Finland). The limit for IA-2A positivity (0.43 RU) was set at the 99th centile in 374 nondiabetic Finnish children and adolescents. The interassay coefficient of variation was 12% at an IA-2A level of 0.63 RU, 10% at a level of 21.3 RU, and 8% at a level of 82.6 RU. This assay had a disease sensitivity of 62% and a specificity of 97%, based on 140 samples included in the 1995 Multiple Autoantibody Workshop (17).

Insulin autoantibodies Serum levels of IAA were quantified with a microassay modified from that described by Williams et al. (19). Antibody-antigen complexes were precipitated with protein A Sepharose (Pharmacia Biotech, Uppsala, Sweden) after incubation of the serum samples with mono-125ITyrA14)-human insulin (Amersham, Buckinghamshire, UK) for 72 h, in the absence or presence of an excess of unlabeled insulin. The specific binding represented the IAA levels that were expressed in RU, based on a standard curve run on each plate, using the MultiCalc software program (EG&G Wallac). A subject was considered to be positive for IAA when the specific binding exceeded 1.55 RU (99th percentile in 374 nondiabetic Finnish subjects). The performance characteristics of this assay were compared with that run in Bristol (19), based on a sample exchange comprising 100 samples. There was a strong correlation between the 2 assays (r 5 0.96; P , 0.001), and the concordance rate was 94%. The disease sensitivity of our microassay was 35%, and the specificity was 100%, based on 140 samples derived from the 1995 Multiple Autoantibody Workshop (17).

Islet cell antibodies Islet cell antibodies were determined by a standard immunofluorescence method using sections of frozen human group O pancreas (4). End point dilution titers were examined for the positive samples, and the results were expressed in Juvenile Diabetes Foundation (JDF) units relative to an international reference standard (20). The detection limit was 2.5 JDF units. Our laboratory has participated in the international workshops on the standardization of the ICA assay, in which its sensitivity was 100%, specificity 98%, validity 98%, and consistency 98%, in the fourth round.

Endogenous insulin secretion Random serum C-peptide concentrations were analyzed with an RIA, using antiserum K6 as described earlier (21), with a commercial kit (Novo Research Institute, Bagsvaerd, Denmark). The detection limit was 0.02 nmol/L. We have shown previously that there is a close correlation between random postprandial serum C-peptide concentrations, serum C-peptide levels measured 120 min after a standardized breakfast, and 24-h urinary C-peptide excretion (10).

Metabolic control—Glycated hemoglobin (GHb) Standard methods were used for blood hemoglobin and hemoglobin analysis. To make the results comparable, they were expressed as sd scores above the mean for nondiabetic subjects. Blood glucose concentrations and blood gases were determined by routine laboratory methods.

Statistical analysis The data were evaluated statistically by cross-tabulation and with x-square statistics, Student‘s t test (two-tailed), or parametric one-way

1536

JCE & M • 1999 Vol 84 • No 5

SABBAH ET AL.

ANOVA (in the case of normal distribution) and Mann-Whitney U-test or Kruskall Wallis one-way ANOVA (in the case of ordinal data). Logarithmic transformations were performed to normalize skewly distributed continuous variables. Age adjustment was performed with analysis of covariance. Proportions were age-adjusted by direct standardization, by reference to the age distribution of the whole diabetic population (22). The Bonferroni adjustment for multiple comparisons was used when appropriate. All the analyses were performed using the SPSS software package (SPSS, Inc., Chicago, IL).

Results

There were 545 index cases (73.2%) who tested positive for GADA at diagnosis, their GADA levels ranging from 6.6 –208 RU, with a median of 38.7 RU. Six hundred forty children (85.7%) had detectable IA-2A at diagnosis, with levels varying from 0.43–2800 RU. More than half of the probands (405; 54.2%) tested positive for IAA. Their median IAA level was 9.04 RU, with a range from 1.56 –309.8 RU. ICA were detected at diagnosis in 629 probands (84.2%), their levels ranging from 2.5–2473 JDF units, with a median of 53 JDF units. An ICA level of 20 JDF units or more was observed in 491 patients (65.7%). More than one third of the probands (35.6%) tested positive for all 4 autoantibodies, 36.9% for three antibodies, 18.9% for two antibody specificities, and 6.3% for one, whereas 2.3% were negative for all 4 autoantibodies analyzed. Accordingly, multiple autoantibodies (3 or more) were observed in 542 patients (72.6%). The frequency of various antibody combinations is shown in Table 1. The clinical and biochemical characteristics of the index cases who were positive and negative for GADA at diagnosis are presented in Table 2. The patients positive for GADA were older than the negative ones, and the proportion of boys among them was lower. There were no significant differences between the GADA-positive and -negative subjects, in the clinical or metabolic characteristics at diagnosis. No differences could be seen in mean age and sex distribution between the probands testing positive for IA-2A and those testing negative (Table 3). Children with IA-2A had lower serum C-peptide concentrations than the antibody-negative ones. There were no other significant differences in clinical TABLE 1. Frequencies of various autoantibody combinations in 747 children with newly diagnosed diabetes Number

No detectable antibodies One antibody 1 Only GADA 1 Only IA-2A 1 Only IAA 1 Only ICA 1 Two antibodies 1 GADA 1 and IA-2A 1 GADA 1 and IAA 1 GADA 1 and ICA 1 IA-2A 1 and IAA 1 IA-2A 1 and ICA 1 IAA 1 and ICA 1 Three antibodies 1 GADA 1, IA-2A 1, and IAA 1 GADA 1, IA-2A 1, and ICA 1 GADA 1, IAA 1, and ICA 1 IA-2A 1, IAA 1, and ICA 1 All four antibodies 1 Multiple antibodies 1 ($3)

Frequency (%)

17

2.3

16 19 4 8

2.1 2.5 0.5 1.1

25 11 27 10 63 5

3.3 1.5 3.6 1.3 8.4 0.7

16 167 19 74 266 542

2.1 22.4 2.5 9.9 35.6 72.6

TABLE 2. Clinical and biochemical characteristics of type 1 diabetes at diagnosis in cases who were positive or negative for GADA GADA 1 (n 5 547)

8.6 6 3.8 288 (52.7) 70 (12.8) 71 (13.9) (n 5 511) Ketoacidosis 115 (22.6) (n 5 509) pH 7.34 6 0.11 (n 5 510) Blood glucose (mmol/L) 21.4 6 9.3 (n 5 518) Glycated hemoglobin (SDS) 13.1 6 5.8 (n 5 397) Serum C-peptide (nmol/L) 0.19 6 0.14 (n 5 543) Exogenous insulin dose 1.19 6 0.66 over the first 24 h (IU/kg21) (n 5 452) Age (yr) Males (%) Diminished consciousness Severe dehydration

GADA 2 (n 5 200)

P-valuea

7.9 6 3.6 123 (61.5) 19 (9.5) 1 (11.1) (n 5 189) 35 (19.1) (n 5 183) 7.35 6 0.10 (n 5 183) 21.2 6 10.1 (n 5 192) 12.6 6 6.0 (n 5 142) 0.21 6 0.17 (n 5 198) 1.18 6 0.71 (n 5 175)

0.03 0.04 0.24 0.40 0.41 0.16 0.78 0.71 0.37 0.85

Data are n (%) or means 6 SD. a Obtained after age adjustment. TABLE 3. Clinical and biochemical characteristics of type 1 diabetes at diagnosis in cases who tested positive or negative for IA-2A

Age (yr) Males (%) Diminished consciousness Severe dehydration Ketoacidosis pH Blood glucose (mmol/L) Glycated hemoglobin (SDS) Serum C-peptide (nmol/L) Exogenous insulin dose over the first 24 h (IU/kg21) Data are n (%) or means 6

IA-2A 1 (n 5 640)

IA-2A 2 (n 5 107)

8.4 6 3.7 353 (55.2) 78 (12.2) 80 (13.4) (n 5 599) 136 (22.9) (n 5 594) 7.34 6 0.11 (n 5 595) 21.4 6 9.6 (n 5 610) 13.1 6 6.1 (n 5 459) 0.19 6 0.14 (n 5 634) 1.20 6 0.67 (n 5 539)

8.8 6 3.9 58 (54.2) 11 (10.3) 12 (11.9) (n 5 101) 14 (14.3) (n 5 98) 7.36 6 0.11 (n 5 98) 21.3 6 8.9 (n 5 100) 12.1 6 5.9 (n 5 80) 0.25 6 0.18 (n 5 107) 1.09 6 0.65 (n 5 88)

P-value

0.28 0.94 0.69 0.81 0.07 0.14 0.90 0.18 0.003 0.16

SD.

and biochemical signs of metabolic decompensation at diagnosis between IA-2A-positive and -negative patients. The children with detectable IAA were, on average, 2 yr younger at diagnosis than those testing negative for IAA, and they had a substantially lower GHb level than the IAA-negative subjects (Table 4). No other differences were observed when comparing subjects with and without IAA. During the follow-up period, no significant differences were seen between the probands who tested positive and those who were negative for GADA at diagnosis, in subsequent serum C-peptide concentrations, daily insulin doses, or GHb levels over the first 2 yr (data not shown). The proportion of children in clinical remission was, however, lower among GADA-positive subjects at 18 months (2.4% vs. 9.5%; P 5 0.02). Those who initially had detectable IA-2A had lower serum C-peptide levels at 24 months [0.06 6 0.07 (sd) vs. 0.11 6 0.18 nmol/L; P 5 0.045], and they needed more

AUTOANTIBODIES IN CHILDREN WITH DIABETES

exogenous insulin at 18 (0.69 6 0.21 vs. 0.57 6 0.24 IU/ kgzday; P 5 0.005) and 24 months (0.73 6 0.21 vs. 0.64 6 0.21 IU/kgzday; P , 0.001). Children testing positive for IAA at the clinical presentation had lower serum C-peptide levels at 6 months [0.19 6 0.15 vs. 0.28 6 0.19 nmol/L; P 5 0.02], 12 months [0.13 6 0.14 vs. 0.19 6 0.17 nmol/L; P 5 0.005], and 24 months [0.05 6 0.06 vs. 0.09 6 0.12 nmol/L; P 5 0.05], and they needed more exogenous insulin at 24 month (0.75 6 0.22 vs. 0.69 6 0.21 IU/kgzday; P , 0.001). In addition, a lower proportion of them were in clinical remission at 12 months (11.0% vs. 20.1%; P 5 0.01). Both those children with three or more antibodies and those with one or two antibodies were significantly younger at the clinical manifestation than the subjects testing negative for all four antibodies (Table 5). No significant differences TABLE 4. Clinical and biochemical characteristics of type 1 diabetes at diagnosis in cases who were positive or negative for IAA IAA 1 (n 5 405)

7.5 6 4.0 234 (57.8) 52 (12.8) 49 (13.0) (n 5 378) Ketoacidosis 84 (22.8) (n 5 368) pH 7.34 6 0.10 (n 5 369) Blood glucose (mmol/L) 21.2 6 8.5 (n 5 337) Glycated hemoglobin (SDS) 11.8 6 5.5 (n 5 341) Serum C-peptide (nmol/L) 0.19 6 0.15 (n 5 399) Exogenous insulin dose 1.18 6 0.64 over the first 24 h (IU/kg21) (n 5 337) Age (yr) Males (%) Diminished consciousness Severe dehydration

IAA 2 (n 5 342)

9.6 6 3.1 177 (51.8) 37 (10.8) 43 (13.4) (n 5 322) 66 (20.4) (n 5 324) 7.34 6 0.11 (n 5 324) 21.8 6 10.6 (n 5 286) 14.3 6 6.0 (n 5 287) 0.20 6 0.14 (n 5 342) 1.19 6 0.71 (n 5 286)

Data are n (%) or means 6 SD. a Obtained after age adjustment.

P-valuea

,0.001 0.10 0.40 0.88 0.43 0.46

were seen in the clinical characteristics or in the degree of metabolic decompensation at diagnosis among the three groups. The children testing positive for multiple autoantibodies needed higher doses of exogenous insulin at 12, 18, and 24 months (Fig. 1A), and they had lower serum Cpeptide concentrations during the second year after diagnosis than the others (Fig. 1B), whereas there were no differences in GHb levels between the two groups during the observation period (data not shown). A lower proportion of patients positive for multiple autoantibodies were in clinical remission at 12 months (12.4 vs. 22.6%; P 5 0.008) and at 18 months (2.2% vs. 9.6%, P 5 0.004), when compared with the remaining children. Those testing negative for all autoantibodies at diagnosis were older than the other subjects [11.1 6 2.6 vs. 8.4 6 3.8 yr; P , 0.001], but they did not show any significant differences in sex distribution, clinical signs, blood glucose levels, or frequency of ketoacidosis at diagnosis (data not shown). The autoantibody-negative children had higher serum C-peptide concentrations at diagnosis (0.33 6 0.26 vs. 0.19 6 0.14 nmol/L; P 5 0.007) and lower daily insulin doses at 6 (0.27 6 0.16 IU/kg vs. 0.45 6 0.20 IU/kg; P 5 0.01), 18 (0.43 6 0.26 vs. 0.68 6 0.22 IU/kg; P 5 0.01), and 24 months (0.49 6 0.27 vs. 0.73 6 0.21 IU/kg; P , 0.001) than the autoantibodypositive ones, after adjustment for age and multiple comparisons. A higher proportion of those without detectable autoantibodies were in clinical remission at 18 months (44.2% vs. 3.4% in the others; P , 0.001).

0.15

Discussion

0.002 0.19 0.91

1537

Our results indicate that positivity for GADA has no impact on the degree of metabolic decompensation at the clinical presentation of type 1 diabetes, and neither could we find any association between initial GADA positivity and residual b-cell function after diagnosis. These findings are consistent with those reported in a Swedish survey (23). In contrast, Petersen et al. (24) reported a lower glucagon-stim-

TABLE 5. Clinical and biochemical characteristics of type 1 diabetes at diagnosis in cases who were positive for multiple ($3 antibodiesAB), 1 or 2 antibodies, and in those with no detectable antibodies I. $3 AB (n 5 542)

Age (yr) Males Diminished consciousness Severe dehydration Ketoacidosis pH Blood glucose (mmol/L) Glycated hemoglobin (SDS) Serum C-peptide (nmol/L) Exogenous insulin dose during the first 24 h (IU/kg21)

8.2 6 3.8 296 (54.6) 70 (12.9) 70 (13.8) (n 5 507) 120 (23.8) (n 5 504) 7.34 6 0.10 (n 5 396) 21.5 6 9.6 (n 5 396) (542) 12.9 6 3.2 (n 5 385) 0.19 6 0.14 (n 5 396) 1.23 6 0.65 (n 5 396)

II. 1–2 AB (n 5 188)

8.8 6 3.5 104 (55.3) 15 (8.0) 19 (10.7) (n 5 177) 26 (15.1) (n 5 172) 7.35 6 0.11 (n 5 128) 21.2 6 9.0 (n 5 128) (188) 13.1 6 3.4 (n 5 128) 0.21 6 0.15 (n 5 128) 1.09 6 0.61 (n 5 128)

Data are n (%) or means 6 SD. AB, antibodies; F, F-ratio; df, degrees of freedom. a Obtained after age adjustment.

III. No AB (n 5 17)

11.1 6 2.6 11 (64.7) 4 (23.5) 3 (18.8) (n 5 16) 4 (25.0) (n 5 16) 7.29 6 0.16 (n 5 11) 25.0 6 13.1 (n 5 11) (17) 12.6 6 2.6 (n 5 11) 0.33 6 0.26 (n 5 11) 1.20 6 0.95 (n 5 11)

P-valuea

Fdf53 5 6.54; P , 0.001 I vs. III P , 0.001 II vs. III P 5 0.002 x2df52z 5 0.69; P 5 0.71 x2df52z 5 5.48; P 5 0.07 x2df52z 5 1.54; P 5 0.46

x2df52z 5 5.82; P 5 0.06 F 5 0.92; P 5 0.34 F 5 0.21; P 5 0.65 F 5 1.35; P 5 0.25 F 5 2.24; P 5 0.14 F 5 0.00; P 5 0.96

1538

JCE & M • 1999 Vol 84 • No 5

SABBAH ET AL.

FIG. 1. Mean serum C-peptide concentrations (A) and daily insulin doses (B), over the initial 2 yr of type 1 diabetes in patients with multiple autoantibodies (f) and in the other subjects (M). *, P 5 0.005; **, P , 0.001, after adjustment for age and multiple comparisons.

ulated C-peptide response during the first year of clinical disease in young adult patients initially positive for GADA, compared with the antibody negative ones. That observation suggests that positivity for GADA at the diagnosis of type 1 diabetes predicts a more rapid progression to total b-cell destruction. On the other hand, GADA have been reported to remain detectable for a long period of time after the clinical manifestation of diabetes (25, 26). The persistence of GADA in affected subjects without any signs of endogenous insulin secretion (25–28) suggests that the production of these autoantibodies must be stimulated, at least partly, by some other antigen than b-cell-derived GAD, possibly reflecting immune stimulation by small amounts of extrapancreatic GAD or by cross-reactive exogenous antigens. To our knowledge, there are no previous data on the possible relation between IA-2A and metabolic state at diagnosis or the clinical course thereafter in children with type 1 diabetes. In this study, affected children testing positive for IA-2A at diagnosis had lower serum C-peptide concentrations than IA-2A-negative patients, both initially and after a duration of 2 yr. This implies that IA-2A may, to some extent, reflect b-cell destruction. This hypothesis is consistent with previous observations of an association between IA-2A and HLA DR4 (18, 29), which is the HLA DR allele most strongly predisposing to type 1 diabetes.

In the present survey, children positive for multiple autoantibodies had a marked reduction in residual b-cell function after the first 6 months of clinical disease. This phenomenon seems to be of physiological significance, because the children who initially tested positive for multiple antibodies needed more exogenous insulin, over the second year, to achieve the same degree of metabolic control than did those who were positive for only 1 or 2 antibody specificities or who were negative. Age is a well-established factor affecting endogenous insulin secretion in children with recent-onset diabetes, with lower serum C-peptide concentrations recorded in younger children than in older ones (30 –33). The decreased serum C-peptide levels observed in children who tested positive for multiple antibodies remained significantly lower than those seen in subjects negative for multiple antibodies, even after age adjustment, indicating an independent association between residual b-cell function and multiple autoantibodies. One can speculate as to why a small proportion of children with diabetes have no detectable autoantibodies at diagnosis. One reason might be that there is a truly antibody-negative form of diabetes in children and adolescents where the b-cell destruction is mediated by mechanisms other than organspecific autoimmunity. Another explanation could be that the phenomenon is caused by insensitive antibody assays. A third alternative is that the patients have seroconverted to autoantibody negative status in the preclinical period and that they have an exceptionally slow autoimmune b-cell destructive process of low intensity. This hypothesis is supported by our observation that the antibody-negative subjects developed their disease, on average, about 2 yr later than did those with at least one type of diabetes-specific autoantibody. In conclusion, the lack of any significant differences in the degree of metabolic decompensation at diagnosis, between autoantibody-positive and negative patients in this extensive series of children with diabetes, suggests that the intensity of the humoral islet-directed immune response has little influence on the clinical characteristics at diagnosis of type 1 diabetes. The major effect of intensive islet-cell-specific autoimmunity seems to be accelerated target organ destruction, resulting in rapid progression to clinical diabetes (3) and leading to total b-cell destruction after the diagnosis. Acknowledgments We thank Susanna Heikkila¨, Sirpa Anttila, Pa¨ivi Koramo, and Riitta Pa¨kkila¨ for their skillful technical assistance.

References 1. Eisenbarth GS. 1986 Type 1 diabetes: a chronic autoimmune disease. N Engl J Med. 314:1360 –1368. 2. Falorni A, Kockum I, Sanjeevi CB, Lernmark Å. 1995 Pathogenesis of insulindependent diabetes mellitus. Clin Endocrinol Metab. 9:25– 46. 3. Knip M, Va¨ha¨salo P, Karjalainen J, Lounamaa R, Åkerblom HK, the Childhood Diabetes in Finland Study Group. 1994 Natural history of preclinical IDDM in high-risk siblings. Diabetologia. 37:388 –393. 4. Bottazzo GF, Florin-Christensen A, Doniach D. 1974 Islet cell antibodies in diabetes mellitus with autoimmune polyendocrine deficiencies. Lancet. II:1279 –1282. 5. Baekkeskov S, Aanstoot H-J, Christgau S, et al. 1990 Identification of the 64K autoantigen in insulin-dependent diabetes as the GABA-synthesizing enzyme glutamic acid decarboxylase. Nature. 347:151–156. 6. Lan M, Notkins A. 1994 Molecular cloning and identification of a receptor-

AUTOANTIBODIES IN CHILDREN WITH DIABETES

7.

8. 9. 10.

11. 12. 13. 14. 15. 16.

17.

18. 19.

type protein tyrosine phosphatase, IA-2, from human insulinoma. DNA Cell Biol. 13:505–514. Christie MS, Brown TJ, Cassidy D. 1993 Binding of antibodies in sera from type 1 (insulin-dependent) diabetic patients to glutamic acid decarboxylase from rat tissues. Evidence for antigenic and non-antigenic forms of the enzyme. Diabetologia. 35:380 –384. Manna R, Salvatore M, Scuderi F, et al. 1988 Negative correlation between ICA persistence and beta cell restoration after IDDM diagnosis. Diabetes Res. 9:101–103. Peig M, Gomis R, Ercilla G, Casamitjana R, Bottazzo GF, Pujol-Borrell R. 1989 Correlation between residual b-cell function and islet cell antibodies in newly diagnosed type 1 diabetes. Follow-up study. Diabetes. 38:1396 –1401. Karjalainen J, Knip M, Mustonen A, Åkerblom HK. 1988 Insulin autoantibodies at the clinical manifestation of type 1 (insulin-dependent) diabetes—a poor predictor of clinical course and antibody response to exogenous insulin. Diabetologia. 31:129 –133. Bingley PJ, Christie MR, Bonifacio E, et al. 1994 Combined analysis of autoantibodies improves prediction of IDDM in islet cell antibody-positive relatives. Diabetes. 43:1304 –1310. Verge CF, Gianini R, Kawasaki E, et al. 1996 Prediction of type 1 diabetes in first-degree relatives using a combination of insulin, GAD, and ICAbdc/IA-2 autoantibodies. Diabetes. 45:926 –933. Kulmala P, Savola K, Petersen JS, et al. 1998 Prediction of insulin-dependent diabetes mellitus in siblings of children with diabetes. J Clin Invest. 101:327–333. Tuomilehto J, Lounamaa R, Tuomilehto-Wolf E, et al. 1992 Epidemiology of childhood diabetes mellitus in Finland - background of a nationwide study of type 1 (insulin-dependent) diabetes mellitus. Diabetologia. 35:70 –76. Petersen JS, Moody HA, Karlsen AE, et al. 1994 Detection of GAD65 antibodies in diabetes and other autoimmune diseases using a simple radioligand assay. Diabetes. 43:459 – 467. Sabbah E, Kulmala P, Veijola R, et al. 1996 Glutamic acid decarboxylase antibodies in relation to other autoantibodies and genetic risk markers in children with newly diagnosed insulin-dependent diabetes. J Clin Endocrinol Metab. 81:2455–2459. Verge CF, Stenger D, Bonifacio E, et al. 1998 Combined use of autoantibodies (IA-2 antibody, GAD antibody, insilin antibody, cytoplasmic islet cell antibodies) in type 1 diabetes. Combinatorial islet autoantibody workshop. Diabetes. 47:1857–1866. Savola K, Bonifacio E, Sabbah E, et al. 1998 IA-2 antibodies—a sensitive marker of IDDM with clinical onset in childhood and adolescence. Diabetologia. 41:424 – 429. Williams AJK, Bingley PJ, Bonifacio E, Palmer JP, Gale EAM. 1997 A novel microassay for insulin autoantibodies. J Autoimmun. 10:473– 478.

1539

20. Lernmark Å, Molenaar JL, Van Beers WAM, et al. 1991 The fourth international serum exchange workshop to standardize cytoplasmic islet cell antibodies. Diabetologia. 34:534 –535. 21. Heding LG. 1975 Radioimmunological determination of human C-peptide in serum. Diabetologia. 11:541–548. 22. Morris JA, Gardner MJ. 1989 Calculating confidence intervals for relative risks, odds ratios, and standardised ratios and rates. In: Gardner MJ, Altman DG, eds. Statistics with confidence. Confidence intervals and statistical guidelines. 1st ed. London: BMJ Books; 50 – 63. ¨ rtquist E, Falorni A, Scheynius A, Persson B, Lernmark Å. 1997 Age governs 23. O gender-dependent islet cell autoreactivity and predicts the clinical course in childhood IDDM. Acta Paediatr. 86:1166 –1171. 24. Petersen JS, Dyrberg T, Karlsen AE, et al. 1994 Glutamic acid decarboxylase (GAD65) autoantibodies in prediction of b-cell function and remission in recent-onset IDDM after cyclosporin treatment. Diabetes. 43:1291–1296. 25. Christie MJ, Daneman D, Champagne P, Delovitch TL. 1990 Persistence of serum antibodies to 64,000-Mr islet cell protein after onset of type 1 diabetes. Diabetes. 39:653– 656. 26. Kawasaki E, Moriuchi R, Takino H, et al. 1992 Autoantibodies to 64,000-Mr islet cell protein in long-term type 1 (insulin-dependent) diabetic patients. Diabetologia. 35:748 –752. 27. Savola K, Sabbah E, Kulmala P, Va¨ha¨salo P, Ilonen J, Knip M. 1998 Disease associated autoantibodies persist after the diagnosis of IDDM in children. Diabetologia. 41:1293–1297. 28. Zanone MM, Petersen JS, Peakman M, et al. 1994 High prevalence of autoantibodies to glutamic acid decarboxylase in long-standing IDDM is not a marker of symptomatic autonomic neuropathy. Diabetes. 43:1146 –1151. 29. Genovese S, Bonfanti R, Bazzigaluppi E, et al. 1996 Association of IA-2 autoantibodies with HLA DR4 phenotypes on IDDM. Diabetologia. 39:1223–1226. 30. Knip M, Ilonen J, Mustonen A, Åkerblom HK. 1986 Evidence of an accelerated b-cell destruction in HLA-Dw3/Dw4 heterozygous children with type 1 (insulin-dependent) diabetes. Diabetologia. 29:347–351. 31. Sochett EB, Daneman D, Clarson C, Ehrlich RM. 1987 Factors affecting and patterns of residual insulin secretion during the first year of type 1 (insulindependent) diabetes mellitus in children. Diabetologia. 30:453– 459. 32. Schiffrin A, Suissa S, Poussier P, Guttmann R, Weizner G. 1988 Prospective study of predictors of b-cell survival in type 1 diabetes. Diabetes. 37:920 –925. 33. Wallensten M, Dahlquist G, Persson B, et al. 1988 Factors influencing the magnitude, duration, and rate of fall of b-cell function in type 1 (insulindependent) diabetic children followed for two years from their clinical diagnosis. Diabetologia. 31:664 – 669.