Timing of onset of autoimmunity is a prerequisite for unmasking triggers and pathogenesis of type 1 diabetes. We followed 4,590 consecutive newborns with 8 or 3% HLA-DQB1 conferred risk for type 1 diabetes at 3-, 6-, or 12-month intervals up to 5.5 years of age. Islet cell autoantibodies (ICAs) and, in the 137 children with ICAs, insulin autoantibodies (IAAs), GAD65 autoantibodies (GADAs), and IA-2 protein autoantibodies (IA-2As) were measured. Children with high genetic risk developed ICAs more often than those with moderate risk (log-rank P = 0.0015); 85 and 91% remained ICA negative by 5 years of age, respectively. The time of appearance of biochemical autoantibodies was then compared with the appearance of ICAs. IAAs and GADAs emerged usually before ICAs (means −1.8 and −1.5 months, respectively) and IA-2As after ICAs (mean 2.0 months). Ninety-five percent of all IAAs, GADAs, and IA-2As seroconversions occurred in a cluster (−12 to 8 months) around the ICA seroconversion. We conclude that diabetes-associated autoantibodies emerged in children with predisposing HLA-DQB1 alleles after 3 months of age at a constant tempo, determined by the genetic risk level, usually in the order of IAA, GADA, ICA, and IA-2A. Seroconversion to multiple autoantibody positivity usually occurred tightly clustered in time.

In pre-type 1 diabetes, accurate timing of the appearance of diabetes-associated autoantibodies may reveal environmental triggers of autoimmunity and improve the accuracy of the assessment of risk for clinical disease. The disease-predictive value of autoantibodies has been studied mainly in first-degree relatives of patients with type 1 diabetes (19). However, they represent poorly the true risk population, because ∼90% of patients with new-onset type 1 diabetes have no close affected relative (10). Only fragmentary data exist on the time of appearance of diabetes-associated autoantibodies (6,8,11), and the time interval from the emergence of autoantibodies to the onset of the disease is poorly defined. We now report when islet-cell antibodies (ICA) emerge and how they behave in children selected from the general population because of increased genetic risk for type 1 diabetes. We also describe when autoantibodies formed against three biochemically characterized autoantigens, insulin (IAA), GAD65 (GADA), and the IA-2 protein (IA-2A), appear in these children and how they behave during the first 5 years of life.

Design and subjects.

The Type 1 Diabetes Prediction and Prevention Project in Finland is based on the screening of HLA-DQB1-associated genetic susceptibility to type 1 diabetes in infants born at the university hospitals of Turku, Oulu, and Tampere (12). This report comprises data on 4,590 consecutively born children with increased genetic diabetes susceptibility and whose parents gave informed consent for prospective follow-up of the child. A total of 982 children were at high genetic risk (8% risk during childhood), and 3,608 were at moderate genetic risk (2.5% risk) (13).

The children in Oulu and Tampere were studied at the ages of 3, 6, 12, 18, and 24 months and annually thereafter, whereas the children in Turku were studied at 3-month intervals for the first 2 years of life and then twice a year. After ICA seroconversion, children were followed at 3-month intervals in all centers. Because genetic screening and recruitment of the at-risk children to the follow-up continue, young children and short follow-up are overrepresented (median 1.5 years, range 0.12–5.5). ICA alone was analyzed at every visit, but after seroconversion, all previous and subsequent samples were analyzed for insulin autoantibodies (IAAs), GAD65 autoantibodies (GADAs), and IA-2 protein autoantibodies (IA-2As). In a separate randomized, placebo-controlled, double-blind trial, the efficacy of nasal insulin in prevention of type 1 diabetes was evaluated in children who tested positive for ICAs on two consecutive occasions.

The Ethics Committees of all participating universities and hospitals approved the study. The parents of all study children gave separate informed consent for the genetic testing and immunological follow-up.

Genetic screening.

Genetic susceptibility to type 1 diabetes was analyzed from cord-blood spots dried on filter paper as described (14,15). Gene sequences were amplified by PCR and hybridized with HLA-DQB1 allele-specific probes labeled with europium, terbium, or samarium lanthanide chelates. Time-resolved fluorescence of the labels was measured to detect different polymerase chain reaction products of the DQB1 alleles *02, *0301, *0302, *0602, and *0603 in all children and DQA1 alleles *0201 and *05 in boys who were positive for DQB1*02 in the Turku cohort. Children with DQB1*02/*0302 were categorized to the group with high genetic risk, whereas children with DQB1*0302/x (x ≠ *02, *0301, or *0602) and boys with DQB1*02/y-DQA1*05/z (y ≠ *0301, *0302, *0602, or *0603; z ≠ *0201) genotypes were categorized to the group with moderate genetic risk.

Autoantibody assays.

ICAs were analyzed using a standard indirect immunofluorescence assay on a section of frozen human pancreas from a blood group O donor (16,17). End-point dilution titers were determined for the ICA-positive sera, and the results were expressed in Juvenile Diabetes Foundation units (JDFU). The detection limit of the assay is 2.5 JDFU. Our laboratory had a sensitivity of 100% and a specificity of 98% in the fourth round of the International Workshop on the Standardization of the ICA Assay (18).

GADAs were measured using a radioligand assay as described (19). The results are expressed in relative units (RU) based on a standard curve run on each plate, and the cutoff limit for positivity is 5.35 RU, representing the 99th percentile in a group of 373 healthy children. The disease sensitivity of the assay was 69% and the specificity was 100%, based on the 140 samples included in the 1995 Multiple Autoantibody Workshop (20).

IA-2As were determined using a modification (21) of a radioligand method (22). The cutoff limit for IA-2A positivity is 0.43 RU, representing the 99th percentile in a group of 374 healthy children and adolescents. The disease sensitivity was 62% and specificity was 97% in 140 samples included in the 1995 Multiple Autoantibody Workshop (20).

IAAs were measured by a radioligand assay in a microplate format using a modification of the method described by Williams et al. (23). The cutoff limit for positivity is 1.55 RU, representing the 99th percentile in a group of 371 nondiabetic Finnish subjects. The disease sensitivity of the assay was 35% and the disease specificity was 100% in 140 samples included in the 1995 Multiple Autoantibody Workshop (20). The assay was compared with the microassay run in Bristol (23) by analyzing blindly 100 samples in both laboratories. The two assays correlated well (r = 0.96, P < 0.001) and showed 94% concordance.

When antibodies to biochemically characterized autoantigens were measured, all samples with titers between the 95th and 99.5th percentiles were reanalyzed to confirm antibody status. In cases of discrepancy, the samples were tested for a third time. Possible assay drift over time was monitored by analyzing blindly three standards (low, medium, and high antibody titers) once a month in each assay.

Statistical methods.

Kaplan-Meier method was used to construct a life table for the likelihood of developing ICAs. The follow-up time was calculated from birth to the first ICA-positive sample or to the last available sample if the child remained ICA-negative. Log-rank test was used for comparison of the survival distributions and hazard ratio to give an estimate of the relative event rates. Distributions of the ICA titers were skewed, and nonparametric Mann-Whitney U and Kruskal-Wallis tests were used in comparison of the groups. Differences between the groups were evaluated using two-sided t tests or χ2 tests depending on whether the variables were continuous or categorical. Only serum samples that were drawn before the onset of diabetes or starting preventive therapy were included. Samples that contained maternal antibodies were excluded from the study. The SPSS software package (Version 9.0.1 for Windows; Chicago, IL) was used for statistical analyses.

Genetic risk and development of autoantibodies.

With a maximum follow-up time of 5.5 years (median 1.5), 137 children (3.0%) had seroconverted to ICA positivity. A total of 47 of them (34%) had high genetic risk, and 90 (66%) had moderate genetic risk. The first seroconversions occurred between the ages of 3 and 6 months, and the proportion of ICA-positive children increased steadily thereafter.

The high-risk children seroconverted to ICA positivity at 1.8 times higher rate (95% CI 1.3–2.8) than the moderate-risk children (log-rank P = 0.0015; Fig. 1) and also to multiple (≥2) autoantibody positivity more often than the moderate-risk children (P = 0.01; Fig. 2). However, the age at ICA seroconversion did not differ between the children with high or moderate genetic risk (P = 0.9). The ICA titers in the first positive samples were also closely similar in the two groups (median 10 vs. 8 JDFU; range 3–436 and 5–110, respectively; P = 0.068), but the maximum ICA titers during the follow-up were slightly higher in the group with high genetic risk than in the group with moderate genetic risk (median 15 vs. 8 JDFU, range 5–436 and 5–1,742, respectively; P = 0.047).

Children with ICA only.

When a child seroconverted to ICA positivity, IAAs, GADAs, and IA-2As were also measured in that child’s all previous, current, and future samples, and the sampling interval was reduced to 3 months if it had been longer before. Half of the ICA-positive children (n = 76) remained negative for other antibodies during the median follow-up of 9.2 months (range 0–30) after ICA seroconversion. Twelve ICA-positive children (9%) reverted to ICA negativity after one to five consecutive ICA-positive samples taken at 3-month intervals. None of these children had any other autoantibodies, and their maximum ICA titers were low (median 9 JDFU, range 5–18). Maximum ICA titers were similarly low (median 8 JDFU, range 5–28) in those children (n = 64) who so far have remained positive for ICA (1–11 consecutive samples taken at 3- to 6-month intervals).

Children with multiple autoantibodies.

Maximum ICA titers during the follow-up were significantly higher in the ICA-positive children who had additional autoantibodies (n = 61) than in those who had ICAs only (n = 76) (median 55 vs. 8 JDFU; range 8–1,742 and 5–28, respectively; P < 0.001). When the number of autoantibodies increased, the maximum ICA titers during the follow-up increased correspondingly (P < 0.001; Fig. 3). Similarly, the first measured positive ICA titer was already higher in the group of children who had or later developed multiple autoantibodies than in the group who remained positive for ICAs only during the follow-up (median 15 vs. 8 JDFU, P < 0.001), but the median ages at ICA seroconversion were almost the same in both groups (1.5 vs. 1.9 years, P = 0.1).

Among the 137 children who during the follow-up seroconverted to ICA positivity, 56 (41%) also had IAAs, 46 (34%) had GADAs, and 36 (26%) had IA-2As at least at some time point during the follow-up. Only 5 children (8%) with multiple autoantibodies were constantly negative for IAAs, 15 (25%) for GADAs, and 25 (41%) for IA-2As. The species of autoantibodies that emerged during the follow-up in the children with high or moderate genetic risk differed slightly (Fig. 4). We then studied the order of appearance of antibodies by comparing the time of appearance of biochemical antibodies with that of ICAs. When other species of autoantibodies emerged, they usually emerged in a cluster with 95% appearing between −12 and 8 months of the ICA seroconversion. There was no consistent order, but IAAs and GADAs usually appeared earlier than ICAs (mean difference −1.8 months, 95% CI −3.1 to −0.6, P = 0.005; and mean difference −1.5 months, 95% CI −3.0 to −0.09, P = 0.038, respectively), and IA-2As usually later than ICAs (mean difference 2.0 months, 95% CI 0.9–3.2, P = 0.001; Fig. 5). However, in occasional children any one of the four autoantibodies could precede the others.

This study shows for the first time that diabetes-associated autoantibodies emerged in children who were selected from the general population by virtue of increased genetic risk for type 1 diabetes at a constant pace after 3 months of age and that the pace depends on the genetic risk of the children as defined by the HLA-DQ gene alleles. Furthermore, IAAs and GADAs usually appeared earlier than ICAs and IA-2As but any of the four autoantibodies occasionally appeared first in children who were younger than 5 years. Finally, if multiple autoantibodies emerged, strongly increasing the risk of clinical diabetes, they usually emerged within a short time period, suggesting that the tentative trigger(s) of these autoimmune markers may well be identical or linked mechanistically with each other.

The first children with HLA-conferred genetic predisposition to type 1 diabetes seroconverted to ICA positivity very early in life. When the age of the oldest children in the follow-up was 5.5 years and the median follow-up time 1.5 years, 137 children had tested positive for ICAs at least once. Our survival analysis showed that the proportion of children who seroconverted to ICA positivity increased steadily, at least for the first 5 years of life. Children with a high genetic risk, as defined in this study, have an estimated risk of developing type 1 diabetes during childhood that is approximately three times higher than that in children with a moderate genetic risk (13). It is interesting that the proportion of children with ICAs increased almost twice as fast in the high-risk children than in the moderate-risk children. Whether this difference remains constant in the follow-up can be answered only after several more years have passed. Because this difference in the proportion of seroconverted children was smaller than the expected difference in the occurrence of clinical diabetes, we propose that the ICA-positive children with high genetic risk have a higher risk for progressing to clinical disease than the ICA-positive children with moderate genetic risk. This hypothesis is consistent with our data showing that the children with high genetic risk seroconverted to multiple autoantibody positivity more often than the children with moderate genetic risk. Children with multiple autoantibody species are more likely to progress to clinical diabetes than those who are positive for only a single autoantibody species (4,5,7,2427). In contrast to our expectations, genetic risk failed to influence the age at which seroconversion occurred, and its effect on the ICA titers in these two groups of children was minimal. Consequently, the HLA-DQ genes mainly increase risk for seroconversion to autoantibody positivity and for progression via multiple autoantibody positivity to overt type 1 diabetes in young children selected from the general population.

The disease risk increases in children if ICA seroconversion occurs, and the risk increases further if the ICA titers and the number of other autoantibodies increase (19). It is interesting that the number of autoantibody species found in a child correlates also more strongly with the maximum ICA titer than with the first ICA titer measured (data not shown), suggesting that the predictive value of randomly measured autoantibody titers in children may vary substantially, depending on the behavior of that and the other autoantibodies both before and after the sampling in that particular child. Low ICA titers were found in both children with ICA only and children with multiple antibodies, suggesting that low ICA titers alone are of limited value in diabetes prediction. If no biochemical antibodies are found when ICA titer is low, then one or several consecutive samples are needed to improve the risk estimation, as multiple antibodies usually appeared in a cluster. As ICA titers above 28 JDFU (Fig. 3) were seen only in children with multiple antibodies, such titers alone confer high risk without the need to measure biochemical antibodies in that and consecutive samples.

Transient seroconversion to ICA positivity occurs in young children relatively infrequently, as only 9% of the ICA-positive children have so far reverted back to ICA negativity during the follow-up. All of these children were negative for the other three autoantibodies studied, and their ICA titers never exceeded 18 JDFU. However, it is important to recognize that 16 of the 61 children who were positive for multiple autoantibodies had ICA titers that so far have not exceeded 18 JDFU.

As we studied the children with increased genetic risk primarily for ICAs only, we do not know how many and which children of those who remained ICA-negative developed other autoantibodies. Figure 4 thus shows only that there was barely any difference in the species of biochemical antibodies found in ICA-positive children with high or moderate genetic risk. The proportions of children with biochemical antibodies only may be different in ICA-negative children than what we have shown for the children with ICAs. However, studies in Finland and elsewhere show that properly standardized ICA is the most sensitive single autoantibody predicting type 1 diabetes in young first-degree relatives of patients with type 1 diabetes (35,7). In our study, most of the children who developed multiple types of autoantibodies seroconverted to positivity for additional autoantibodies within a short time window. In the majority of these children, the different autoantibodies appeared in consecutive samples taken at 3-month intervals rather than simultaneously. Any one of the four diabetes-associated autoantibodies was at least occasionally found as the first or last emerging autoantibody, and no constant order of appearance of the autoantibodies was recognized. However, IAAs emerged an average of 1.8 months earlier than ICAs, which appeared at the median age of 18 months in children with multiple autoantibodies.

ICA measurement is hampered by problems such as difficulties in obtaining well-suited cadaver pancreases and standardization of the assay. Combined analyses of IAAs and GADAs is a sensitive alternative approach for prediction (5,7,2427). If in our study we had screened for IAAs and GADAs, then we would have missed only three children with multiple autoantibodies and detected antibodies in 31 children (50% of the children with multiple antibodies) in the median 5.8 months earlier than when we used primarily only ICA. Measurement of IAA + GADA would have delayed the detection of multiple autoantibodies in four children by 3, 3, 16, and 5 months (the true delay in this fourth child was probably 3.5 years, as the value of the second antibody was just above the limit of positivity only in one sample, 5 months after the appearance of ICA, but then it disappeared; the child then developed multiple antibodies 3.5 years after ICAs). For identifying children with multiple autoantibodies as early as possible, measurement of all four autoantibodies obviously would be the most sensitive approach, but the cost-efficacy of such an approach clearly would be lower than that of our current strategy. If concurrent measurement of all four antibodies is not feasible, then the most sensible antibody or combination of antibodies should be preferred, as the autoantibodies usually appeared in a narrow time window. However, in practice, the time interval between the samples drawn creates the greatest delay in the recognition of the autoantibody positivity, not the choice of antibody or antibodies measured. The specificity in diabetes prediction then can be improved by measuring the other autoantibodies from the sample(s) of interest. We have repeatedly analyzed our data during the study years to improve our prediction, also taking into account the adverse effects caused by finding “false positive” children, who probably will never progress to overt diabetes. The unfounded psychological burden in such families caused by such false alarms may be substantial. Thus far, we find ICAs well-suited for the primary immunological marker measured in studies like ours, but the predictive values of the autoantibodies and their combinations may change when children become older.

Nearly half of the ICA-positive children developed multiple autoantibodies, and the other half so far has had ICAs only. The former group has clearly higher risk for progression to type 1 diabetes. ICA alone was occasionally the first autoantibody to emerge in such young children, but if other antibodies appeared, they appeared within a short time window in the majority of cases. Accordingly, a second sample taken a few months later provided essential predictive information in children who had had ICAs only in the first positive sample. However, in occasional children with ICAs only, the titer increased or other autoantibodies appeared after a long silent period, thereby probably increasing the risk substantially. We obviously need longer follow-up times to accurately estimate the risk for diabetes in such children.

In conclusion, the proportion of ICA-positive children increased steadily, at least during the first 5 years of life, among children who were selected from the general population based on increased HLA-conferred genetic risk for type 1 diabetes. As the first children progress to clinical type 1 diabetes during the first few years of life, only repeated autoantibody measurements will identify these children early enough for possible preventive interventions. Screening of the at-risk children for ICAs only is a relatively efficient screening strategy, as in young children multiple autoantibodies in most cases appear during a short time window, if they are to appear. Although the different autoantibodies may appear in any order, IAAs and GADAs usually emerged slightly earlier than ICAs, and IA-2As as the last antibody type.

FIG. 1.

Kaplan-Meier survival curve of children who remained ICA-negative during the 5-year postnatal follow-up. Children with moderate or high genetic risk for type 1 diabetes were compared. At ages 0, 1, 2, 3, 4, and 5, the number of children at moderate genetic risk was 3,608, 2,598, 1,489, 792, 380, and 57, respectively. At ages 0, 1, 2, 3, 4, and 5, the number of children at high genetic risk was 982, 712, 427, 250, 128, and 22, respectively.

FIG. 1.

Kaplan-Meier survival curve of children who remained ICA-negative during the 5-year postnatal follow-up. Children with moderate or high genetic risk for type 1 diabetes were compared. At ages 0, 1, 2, 3, 4, and 5, the number of children at moderate genetic risk was 3,608, 2,598, 1,489, 792, 380, and 57, respectively. At ages 0, 1, 2, 3, 4, and 5, the number of children at high genetic risk was 982, 712, 427, 250, 128, and 22, respectively.

FIG. 2.

The proportion of ICA-positive children who had moderate or high genetic risk for type 1 diabetes and who were positive for ICA only or also positive for one, two, or three additional autoantibodies (abs).

FIG. 2.

The proportion of ICA-positive children who had moderate or high genetic risk for type 1 diabetes and who were positive for ICA only or also positive for one, two, or three additional autoantibodies (abs).

FIG. 3.

Scatter-plot diagram of maximum ICA titers according to the number of autoantibodies. Note that other diabetes-associated autoantibodies were measured only in children who had been ICA positive at least once.

FIG. 3.

Scatter-plot diagram of maximum ICA titers according to the number of autoantibodies. Note that other diabetes-associated autoantibodies were measured only in children who had been ICA positive at least once.

FIG. 4.

The combinations of IAA, GADA, and IA-2A positivity in ICA-positive children with moderate or high genetic risk for type 1 diabetes. Absolute numbers (%) are shown.

FIG. 4.

The combinations of IAA, GADA, and IA-2A positivity in ICA-positive children with moderate or high genetic risk for type 1 diabetes. Absolute numbers (%) are shown.

FIG. 5.

The time of ICA seroconversion was standardized for all 137 ICA-seroconverted children to be time 0 on the x-axis. The seroconversion times of IAAs (red circles), GADAs (green), and IA-2As (blue) were compared with ICA seroconversion time above the x-axis. The mean differences (95% CI) in the seroconversion times are presented. The current follow-up times after ICA seroconversion of the children who constantly have remained negative for IAAs, GADAs, or IA-2As in all samples are shown underneath the x-axis.

FIG. 5.

The time of ICA seroconversion was standardized for all 137 ICA-seroconverted children to be time 0 on the x-axis. The seroconversion times of IAAs (red circles), GADAs (green), and IA-2As (blue) were compared with ICA seroconversion time above the x-axis. The mean differences (95% CI) in the seroconversion times are presented. The current follow-up times after ICA seroconversion of the children who constantly have remained negative for IAAs, GADAs, or IA-2As in all samples are shown underneath the x-axis.

This study was supported by Juvenile Diabetes Foundation International (Grants 4-1998-274, 197032, and 4-1999-731); the Sigrid Jusélius Foundation; the Academy of Finland; the Päivikki and Sakari Sohlberg Foundation; the Novo Nordisk Foundation; the Jalmari and Rauha Ahokas Foundation; the Foundation for Pediatric Research, Finland; the Foundation for Diabetes Research, Finland; the Research Foundation of ORION Corporation; the Signe and Ane Gyllenberg Foundation; and Special Grants for University Hospitals of Turku, Oulu, and Tampere.

We thank all families participating in the Diabetes Prediction and Prevention Project; Terttu Laurén, Mia Karlsson, and Ritva Suominen for genetic screening in practice; Susanna Heikkilä, Tuovi Mehtälä, Riitta Päkkilä, and Päivi Salmijärvi for measuring autoantibodies; Paula Asunta, Helena Haapanen, Reija Hakala, Anu-Maaria Hämäläinen, Teemu Kalliokoski, Susanna Lunkka, Ulla Markkanen, Elina Mäntymäki, Birgitta Nurmi, Hilkka Pohjola, Sirpa Pohjola, Kaija Rasimus, Kaisu Riikonen, Riikka Sihvo, Aino Stenius, Aila Suutari, Anna Toivonen, Maija Törmä, and Mia äikäs for dedicating time to the best of the study children and families; and the personnel at the Departments of Obstetrics and Gynecology of the Turku, Oulu, and Tampere University Hospitals for collaboration.

1.
Riley WJ, Maclaren NK, Krischer J, Spillar RP, Silverstein JH, Schatz DA, Schwartz S, Malone J, Shah S, Vadheim C, Rotter JI: A prospective study of the development of diabetes in relatives of patients with insulin-dependent diabetes.
N Engl J Med
323
:
1167
–1172,
1990
2.
Bonifacio E, Bingley PJ, Shattock M, Dean BM, Dunger D, Gale EA, Bottazzo GF: Quantification of islet-cell antibodies and prediction of insulin-dependent diabetes.
Lancet
335
:
147
–149,
1990
3.
Bingley PJ, Christie MR, Bonifacio E, Bonfanti R, Shattock M, Fonte MT, Bottazzo GF, Gale EA: Combined analysis of autoantibodies improves prediction of IDDM in islet cell antibody-positive relatives.
Diabetes
43
:
1304
–1310,
1994
4.
Kulmala P, Savola K, Petersen JS, Vähäsalo P, Karjalainen J, Löppönen T, Dyrberg T, Åkerblom HK, Knip M: Prediction of insulin-dependent diabetes mellitus in siblings of children with diabetes: a population-based study: the Childhood Diabetes in Finland Study Group.
J Clin Invest
101
:
327
–336,
1998
5.
Yamamoto AM, Deschamps I, Garchon HJ, Roussely H, Moreau N, Beaurain G, Robert JJ, Bach JF: Young age and HLA markers enhance the risk of progression to type 1 diabetes in antibody-positive siblings of diabetic children.
J Autoimmun
11
:
643
–650,
1998
6.
Schenker M, Hummel M, Ferber K, Walter M, Keller E, Albert ED, Janka HU, Kastendiek C, Sorger M, Louwen F, Ziegler AG: Early expression and high prevalence of islet autoantibodies for DR3/4 heterozygous and DR4/4 homozygous offspring of parents with type I diabetes: the German BABYDIAB study.
Diabetologia
42
:
671
–677,
1999
7.
Maclaren N, Lan M, Coutant R, Schatz D, Silverstein J, Muir A, Clare-Salzer M, She JX, Malone J, Crockett S, Schwartz S, Quattrin T, DeSilva M, Vander Vegt FP, Notkins A, Krischer J: Only multiple autoantibodies to islet cells (ICA), insulin, GAD65, IA-2 and IA-2β predict immune-mediated (type 1) diabetes in relatives.
J Autoimmun
12
:
279
–287,
1999
8.
Colman PG, Steele C, Couper JJ, Beresford SJ, Powell T, Kewming K, Pollard A, Gellert S, Tait B, Honeyman M, Harrison LC: Islet autoimmunity in infants with a type I diabetic relative is common but is frequently restricted to one autoantibody.
Diabetologia
43
:
203
–209,
2000
9.
Becker DJ, LaPorte RE, Libman I, Pietropaolo M, Dosch HM: Prevention of type 1 diabetes. Is now the time?
J Clin Endocrinol Metab
85
:
498
–506,
2000
10.
Veijola R, Reijonen H, Vähäsalo P, Sabbah E, Kulmala P, Ilonen J, Åkerblom HK, Knip M: HLA-DQB1-defined genetic susceptibility, beta cell autoimmunity, and metabolic characteristics in familial and nonfamilial insulin-dependent diabetes mellitus: the Childhood Diabetes in Finland (DiMe) Study Group.
J Clin Invest
98
:
2489
–2495,
1996
11.
Rewers M, Norris JM, Eisenbarth GS, Erlich HA, Beaty B, Klingensmith G, Hoffman M, Yu L, Bugawan TL, Blair A, Hamman RF, Groshek M, McDuffie RS Jr: Beta-cell autoantibodies in infants and toddlers without IDDM relatives: Diabetes Autoimmunity Study in the Young (DAISY).
J Autoimmun
9
:
405
–410,
1996
12.
Kupila A, Muona P, Simell T, Arvilommi P, Savolainen H, Hämäläinen A, Korhonen S, Kimpimäki T, Sjöroos M, Ilonen J, Knip M, Simell O: Feasibility of genetic and immunological prediction of type I diabetes in a population-based birth cohort.
Diabetologia
44
:
290
–297,
2001
13.
Nejentsev S, Sjöroos M, Soukka T, Knip M, Simell O, Lövgren T, Ilonen J: Population based genetic screening for type 1 diabetes mellitus risk in Finland: selective genotyping of the markers in the HLA-DQB1, HLA-DQA1 and HLA-DRB1 loci.
Diabet Med
16
:
985
–992,
1999
14.
Sjöroos M, Iitiä A, Ilonen J, Reijonen H, Lövgren T: Triple-label hybridization assay for type-1 diabetes-related HLA alleles.
Biotechniques
18
:
870
–877,
1995
15.
Ilonen J, Reijonen H, Herva E, Sjöroos M, Iitiä A, Lövgren T, Veijola R, Knip M, Åkerblom HK: Rapid HLA-DQB1 genotyping for four alleles in the assessment of risk for IDDM in the Finnish population: the Childhood Diabetes in Finland (DiMe) Study Group.
Diabetes Care
19
:
795
–800,
1996
16.
Bottazzo GF, Florin-Christensen A, Doniach D: Islet-cell antibodies in diabetes mellitus with autoimmune polyendocrine deficiencies.
Lancet
2
:
1279
–1283,
1974
17.
Karjalainen JK: Islet cell antibodies as predictive markers for IDDM in children with high background incidence of disease.
Diabetes
39
:
1144
–1150,
1990
18.
Lernmark Å, Molenaar JL, van Beers WA, Yamaguchi Y, Nagataki S, Ludvigsson J, Maclaren NK: The Fourth International Serum Exchange Workshop to standardize cytoplasmic islet cell antibodies: the Immunology and Diabetes Workshops and Participating Laboratories.
Diabetologia
34
:
534
–535,
1991
19.
Savola K, Sabbah E, Kulmala P, Vähäsalo P, Ilonen J, Knip M: Autoantibodies associated with type I diabetes mellitus persist after diagnosis in children.
Diabetologia
41
:
1293
–1297,
1998
20.
Verge CF, Stenger D, Bonifacio E, Colman PG, Pilcher C, Bingley PJ, Eisenbarth GS: Combined use of autoantibodies (IA-2 autoantibody, GAD autoantibody, insulin autoantibody, cytoplasmic islet cell antibodies) in type 1 diabetes: Combinatorial Islet Autoantibody Workshop.
Diabetes
47
:
1857
–1866,
1998
21.
Savola K, Bonifacio E, Sabbah E, Kulmala P, Vähäsalo P, Karjalainen J, Tuomilehto-Wolf E, Meriläinen J, Åkerblom HK, Knip M: IA-2 antibodies—a sensitive marker of IDDM with clinical onset in childhood and adolescence. Childhood Diabetes in Finland Study Group.
Diabetologia
41
:
424
–429,
1998
22.
Bonifacio E, Lampasona V, Genovese S, Ferrari M, Bosi E: Identification of protein tyrosine phosphatase-like IA2 (islet cell antigen 512) as the insulin-dependent diabetes-related 37/40K autoantigen and a target of islet-cell antibodies.
J Immunol
155
:
5419
–5426,
1995
23.
Williams AJ, Bingley PJ, Bonifacio E, Palmer JP, Gale EA: A novel micro-assay for insulin autoantibodies.
J Autoimmun
10
:
473
–478,
1997
24.
Verge CF, Gianani R, Kawasaki E, Yu L, Pietropaolo M, Jackson RA, Chase HP, Eisenbarth GS: Prediction of type I diabetes in first-degree relatives using a combination of insulin, GAD, and ICA512bdc/IA-2 autoantibodies.
Diabetes
45
:
926
–933,
1996
25.
Hagopian WA, Sanjeevi CB, Kockum I, Landin-Olsson M, Karlsen AE, Sundkvist G, Dahlquist G, Palmer J, Lernmark Å: Glutamate decarboxylase-, insulin-, and islet cell-antibodies and HLA typing to detect diabetes in a general population-based study of Swedish children.
J Clin Invest
95
:
1505
–1511,
1995
26.
Bingley PJ, Bonifacio E, Williams AJ, Genovese S, Bottazzo GF, Gale EA: Prediction of IDDM in the general population: strategies based on combinations of autoantibody markers.
Diabetes
46
:
1701
–1710,
1997
27.
Strebelow M, Schlosser M, Ziegler B, Rjasanowski I, Ziegler M: Karlsburg type I diabetes risk study of a general population: frequencies and interactions of the four major type I diabetes-associated autoantibodies studied in 9419 schoolchildren.
Diabetologia
42
:
661
–670,
1999

Address correspondence and reprint requests to Antti Kupila, MD, Department of Pediatrics, University of Turku, Box PL 52, FIN-20521 Turku, Finland. E-mail: antti.kupila@tyks.fi.

Received for publication 16 March 2001 and accepted in revised form 19 November 2001.

GADA, GAD65 autoantibody; IA-2A, IA-2 protein autoantibody; IAA, insulin autoantibody; ICA, islet cell autoantibody; JDFU, Juvenile Diabetes Foundation units; RU, relative units.