1950 DIABETES CARE, VOLUME 22, NUMBER 12, DECEMBER 1999

The incidence of type 1 diabetes hasbeen increasing in the last decades.I n t e re s t i n g l y, the increase has been

re p o rted to be most remarkable in gro u p sof children ,2 years of age (1,2). However,the disease is still rarely manifested in chil-d ren ,2 years of age. As far as we know,

t h e re are no re p o rts on the clinical, autoim-mune, and genetic characteristics of thedisease in such young patients.

T h e re are a lot of data suggesting het-e rogeneity in the etiology and pathogenesisof type 1 diabetes between children andadults. The process of b-cell destruction is

faster in affected children than in aff e c t e dadults (3), and the metabolic deteriorationat the diagnosis of diabetes is severer inyoung children than in older children (4).T h e re are also well-documented diff e re n c e sin autoantibody frequency and levelsamong patients manifesting type 1 diabetesat various ages (5–9). HLA genes associatedwith disease susceptibility are more fre-quent in patients diagnosed during child-hood than in those diagnosed duringadulthood (10–12).

The objective of the present study wastwofold: to evaluate clinical, autoimmune,and genetic characteristics at pre s e n t a t i o nand during the early course of clinical type 1diabetes in subjects presenting with the dis-ease before their second birthday; and toc o m p a re such characteristics with those ofsubjects diagnosed later in childhood.

RESEARCH DESIGN AND M E T H O D S

Study populationAs part of the Finnish nationwide Child-hood Diabetes in Finland (DiMe) study,801 probands ,15 years of age, who werediagnosed as having diabetes during there c ruitment period from 1 September 1986to 30 April 1989, were invited to part i c i-pate in the study. The study protocol hasbeen described in detail previously (13),and it was approved by the ethical com-mittees of all participating hospitals. Thepatients and/or their parents gave form a lconsent. At the time of diagnosis, capillaryor venous blood pH levels, blood glycatedhemoglobin levels, and serum C-peptidelevels were measured. The probands werethen followed in their own outpatient clin-ics (n = 31), and their data were collected at6-month intervals for 2 years.

MethodsAutoantibody tests were perf o rmed in theR e s e a rch Laboratory, Department of Pedi-atrics, University of Oulu, Finland. Islet cellantibodies (ICA) were determined by as t a n d a rd indirect immunoflu o re s c e n c eassay perf o rmed on sections of fro z e nhuman blood group O pancreas (13). Rab-

F rom the Department of Pediatrics (J.K.), Kuopio University Hospital, Kuopio; the Department of Pediatrics( P.K., K.S.), University of Oulu, Oulu; the Department of Pediatrics (R.L.), Keski-Häme Central Hospital,Hämeenlinna; the Turku Immunology Center and the Department of Vi rology (J.I., H.R.), University of Tu r k u ,Turku; the Department of Pediatrics and the University of Ta m p e re Medical School (M.K.), Ta m p e re Uni-versity Hospital, Ta m p e re; and the Hospital for Children and Adolescents (H.K.Å.), University of Helsinki,Helsinki, Finland.

A d d ress correspondence and reprint requests to Jorma Komulainen, MD, Department of Pediatrics, Kuo-pio University Hospital, P.O. Box 1777, FIN-70211 Kuopio, Finland. E-mail: jorm a . k o m u l a i n e n @ k u h . fi.

A b b re v i a t i o n s : DKA, diabetic ketoacidosis; GADA, GAD antibodies; IAA, insulin autoantibodies; IA-2A,IA-2 antibodies; ICA, islet cell antibodies; JDF U, Juvenile Diabetes Foundation units; RU, relative units.

A table elsewhere in this issue shows conventional and Système International (SI) units and conversionfactors for many substances.

Clinical, Autoimmune, and GeneticCharacteristics of Ve ry Young Childre nWith Type 1 Diabetes

O R I G I N A L A R T I C L E

O B J E C T I V E — To study the characteristics of type 1 diabetes in very young childre n .

RESEARCH DESIGN AND METHODS — Clinical outcome, islet cell antibodies(ICA), insulin autoantibodies (IAA), antibodies against GAD (GADA), IA-2 antibodies (IA-2A),and HLA-DQB1–defined genetic risk were analyzed in 35 children diagnosed with type 1 dia-betes before 2 years of age and compared with those in 146 children who were diagnosedbetween 2.0 and 4.9 years of age and with those in 620 children diagnosed between 5.0 and14.9 years of age.

R E S U LT S — The youngest age-group had severer metabolic decompensation at clinicalonset, and their serum C-peptide levels, compared with those of older children, were lower atthe time of diagnosis and during the first 2 years after the diagnosis. The levels of ICA and IAAw e re highest in children ,2 years of age, but there were no diff e rences in GADA levels amongthe three age-groups. The youngest age-group had the lowest IA-2A levels. The HLADQB1*02/*0302 genotype associated with strong genetic susceptibility was more frequent inc h i l d ren diagnosed ,5 years of age, whereas the pro p o rtion of children carrying a genotype,which includes protective alleles, was higher among those diagnosed at $5 years of age.

C O N C L U S I O N S — The clinical presentation of type 1 diabetes at a very young age isassociated with severe metabolic decompensation, poorly pre s e rved residual b-cell function,s t rong humoral autoimmunity against islet cells and insulin, and strong HLA-defined diseasesusceptibility.

Diabetes Care 2 2 :1 9 5 0–1955, 1999

JORMA KOMULAINEN, MD

PETRI KULMALA, MD

KAISA SAVOLA, MB

RAISA LOUNAMAA, MD

JORMA ILONEN, MD

HELENA REIJONEN, MD

MIKAEL KNIP, MD

HANS K. ÅKERBLOM, MD

THE CHILDHOOD DIABETES IN FINLAND

(DIME) STUDY GROUP

E p i d e m i o l o g y / H e a l t h S e r v i c e s / P s y c h o s o c i a l R e s e a r c h

DIABETES CARE, VOLUME 22, NUMBER 12, DECEMBER 1999 1951

Komulainen and Associates

bit anti–human IgG (Behringwerke, Mar-b u rg, Germany) was used to detect ICA.The results were expressed in Juvenile Dia-betes Foundation units ( JDF U) (15). Thedetection limit for ICA was 2.5 JDF U. Thel a b o r a t o ry has participated in the intern a-tional workshops on the standardization ofthe ICA assay with sensitivity, specific i t y,v a l i d i t y, and consistency of 100, 98, 98, and98%, re s p e c t i v e l y, in the pertinent ro u n d .

Insulin autoantibodies (IAA) werem e a s u red by a modification of the liq-uid phase radioimmunoassay originallydescribed by Palmer et al. (16). In thisa s s a y, the specific insulin binding in 105nondiabetic children was 23.3 ± 11.2nU/ml. The cutoff limit for antibody posi-tivity was 57 nU/ml (mean 1 3 SD in 105nondiabetic subjects). The laboratory hasp a rticipated in the international workshopson the standardization of insulin autoanti-body assays and in the pro ficiency testingp rogram for several years. Subsequently,the laboratory has achieved a sensitivity of78%, a specificity of 100%, a validity of92%, and a consistency of 100%.

Antibodies against GAD (GADA) wereq u a n t i fied with a radioligand assay asdescribed by Petersen et al. (17). Theresults were expressed in relative units(RU), which re p resent the specific bindingas a percentage of that obtained with a pos-itive standard serum. The cutoff limit forantibody positivity was set at the 99th per-centile of 372 nondiabetic children andadolescents (i.e., 6.6 RU). The intra-assayc o e fficient of variation was ,5%, and the

interassay coefficient of variation was,10%. The disease sensitivity of the pre s-ent assay was 80%, and the specificity was94% based on the 101 samples included inthe Second International GAD AntibodyWorkshop (18).

IA-2 antibodies (IA-2A) were analyzedwith a radio-binding assay pre v i o u s l ydescribed in detail (9). The results weree x p ressed in RU. The limit for positivity(0.43 RU) was set at the 99th percentile for374 nondiabetic Finnish children and ado-lescents. The interassay coefficient of varia-tion was 12% at an IA-2A level of 0.63 RU,10% at a level of 21.3 RU, and 8% at a levelof 82.6 RU. The disease sensitivity of theassay was 62%, and the specificity was 97%based on 140 samples included in the 1995Multiple Autoantibody Workshop (19).

The HLA-DQB1 typing method hasbeen described in detail earlier (20). Inbrief, 158 bp of the second exon of theDQB1 gene were amplified by the poly-merase chain reaction from each DNA sam-ple using a primer pair with a biotinylated39- p r i m e r. The amplification product wasbound to streptavidin-coated micro t i t r a-tion plates and denatured with NaOH.After washing, bound DNA was analyzedusing two diff e rent hybridization mixture s :one containing a europium-labeled intern a lre p o rter probe for DQB1*0602 and *0603alleles (0602-3) and one containing a ter-bium-labeled consensus probe. The sec-ond hybridization mixture includedterbium-, samarium- and euro p i u m -labeled probes specific for DQB1*02,

*0301, and *0302 alleles, re s p e c t i v e l y. Tom e a s u re probe hybridization, micro t i t e rplates were analyzed by time-resolved flu-o rescence (DELFIA R e s e a rch Fluoro m e t e r ;Wallac Oy, Turku, Finland). Diff e rent emis-sion wavelengths and decay times wereused to distinguish signals of each lan-thanide label.

For the analysis, the subjects weredivided into four groups according totheir DQB1 alleles: the DQB1*02/*0302g e n otype with both risk alleles; theDQB1*0302/x and DQB1*02/y genotypeswith either of the two risk alleles in singleor double dose but without any pro t e c t i v ealleles (*0301, *0602, or *0603); and theDQB1z/z genotype, which re p resents thosegenotypes that include protective alleles.

Random serum C-peptide concentra-tions were analyzed with a radioim-m u n o a s s a y, as described earlier (21), byusing antiserum K6 (Novo Research Insti-tute, Bagsvaerd, Denmark). The intra-assayc o e fficient of variation was 1.8%, and theinterassay coefficient of variation was 10%.We have previously shown that there is as t rong correlation between random seru mC-peptide levels and serum C-peptide con-centrations measured 120 min after a stan-d a rdized meal and a 24-h urinary C-peptides e c retion (6).

S t a n d a rd methods for blood HbA1 a n dH b A1 c analyses were used in the varioushospitals participating in the study. To com-p a re the results, data were expressed as theSD above the mean for nondiabetic subjects.P a rtial remission was defined as a period

Table 1—Clinical data of the study population at time of diagnosis

G roup 1 (,2 years) G roup 2 (2.0–4.9 years) G roup 3 (5.0–14.9 years) P

pH ,7 . 3 0 5 3 . 3 1 4 . 5 2 1 . 8 1 vs. 2 vs. 3: ,0 . 0 0 11 vs. 2: 0.0011 vs. 3: 0.003

I m p a i red consciousness 3 1 . 4 1 1 . 6 1 1 . 1 1 vs. 2 vs. 3: 0.0021 vs. 2: 0.0041 vs. 3: ,0 . 0 0 1

C-peptide (nmol/l) 0.07 (0.05–0.13) 0.13 (0.09–0.20) 0.17 (0.12–0.26) 1 vs. 2 vs. 3: ,0 . 0 0 11 vs. 2: 0.0011 vs. 3: ,0 . 0 0 1

Acute infection 5 4 . 3 4 8 . 6 2 3 . 2 1 vs. 2 vs. 3: ,0 . 0 0 11 vs. 3: ,0 . 0 0 1

Blood glucose (mmol/l) 22.9 (19.2–28.0) 18.8 (14.7–25.1) 20.3 (15.1–25.9) 1 vs. 2 vs. 3: 0.0341 vs. 2: 0.0081 vs. 3: 0.017

G H b 10.7 ± 6.3 10.2 ± 4.4 13.6 ± 6.1 1 vs. 2 vs. 3: ,0 . 0 0 11 vs. 3: 0.027

Data are %, medians (interq u a rtile range), or means ± SD.

1952 DIABETES CARE, VOLUME 22, NUMBER 12, DECEMBER 1999

Diabetes in very young children

in which the daily insulin dose was ,0 . 5IU ? k g– 1 ? d a y– 1 with standardized glycatedhemoglobin (HbA1 or HbA1 c) being ,4 SD.

Diabetic ketoacidosis (DKA) wasd e fined as a capillary or venous blood pH of,7.30. The degree of consciousness anddehydration was assessed by the clinicianexamining the patient at the time of hospitaladmission. Consciousness was estimated tobe either normal or impaired. The re c o m-mended amount of insulin and the weight ofthe proband was re g i s t e red at 3 weeks afterdiagnosis. There a f t e r, the daily insulin dosewas calculated every 6 months.

The data were evaluated statisticallyusing cross-tabulation and x2 s t a t i s t i c s ,one-way analysis of variance, the Kru s k a l -Wallis test, and the Mann-Whitney U t e s t(SPSS for Windows, SPSS, Chicago). Thedistribution of HLA-DQB1 genotypes wasevaluated using the Mantel-Haenszel testfor linear association.

R E S U LT S — A total of 35 (4.4%)patients were diagnosed with type 1 dia-betes before the age of 2 years, 146 (18.2%)patients were diagnosed between 2.0 and4.9 years of age, and 620 (77.4%) patientsw e re diagnosed between 5.0 and 14.9 yearsof age. The majority of the patients wereboys (440 vs. 361; P , 0.005). There wereno significant diff e rences in the pro p o rt i o nof boys among the three age-groups (,2years, 51.4%; 2.0–4.9 years, 55.5%;5.0–14.9 years, 55.0%). Furt h e rm o re ,t h e re were no diff e rences in the fre q u e n c yof family history of type 1 diabetes amongthe age-groups (,2 years, 6.7%; 2.0–4.9years, 15.7%; 5.0–14.9 years, 13.2%).

The clinical findings at the diagnosis oftype 1 diabetes are summarized in Table 1.

The children in the youngest age-group pre-sented more often with DKA, impaired con-sciousness, and signs of an acute infectionthan children in the other two age-gro u p s .They also had lower serum C-peptide con-centrations and higher blood glucose levelsthan the older children. Severe DKA (pH,7.10) was seen in 10.0% of the patients inthe group ,2 years of age, in 2.9% of thepatients in the group 2.0–4.9 years of age,and in 4.7% of the patients in the gro u p5.0–14.9 years of age (NS).

S e rum C-peptide levels were lowest inthe youngest age-group at diagnosis andduring follow-up (Fig. 1). At the 6-monthfollow-up visit, none of the children diag-nosed before the age of 2 years but 27.4%of the children aged 2.0–4.9 years and37.1% of those diagnosed between 5.0 and14.9 years of age (P , 0.001) were in par-tial remission. At the same time, children inthe youngest age-group had higher GHblevels than those in the older age-gro u p s(,2 years, 8.8 SD; 2.0–4.9 years, 4.5 SD;

5.0–14.9 years, 3.8 SD above the mean fornondiabetic subjects; P , 0 . 0 0 1 ) .

ICA were measured in 791 patients atthe diagnosis of type 1 diabetes, while IAAw e re quantified in 789 patients, GADA in769 patients, and IA-2A in 757 patients.Patients in the youngest age-group testedpositive more often for IAA (P , 0 . 0 0 1 )than those in the older age-groups (Fig. 2).T h e re were no diff e rences in ICA, GADA,or IA-2A positivity among the diff e re n ta g e - g ro u p s .

All of the children in the youngest age-g roup (32 of 32, 100%) tested positive for atleast one of the antibodies studied, while130 of 131 (99.2%) and 568 of 583 (97.4%)of the children in the two older age-gro u p s ,re s p e c t i v e l y, were positive for at least oneantibody (NS). The percentages of childre ntesting positive for all four antibodies were53.1 in the youngest age-group, 42.0 in thea g e - g roup between 2.0 and 4.9 years, and26.9 in the oldest age-group (P , 0 . 0 0 1 ) .

S e rum levels of the diff e rent autoanti-bodies are shown in Table 2. Children inthe youngest age-group had higher titers ofICA and IAA than those in the older age-g roups. The levels of GADA were of thesame magnitude in all three age-gro u p s ,w h e reas the IA-2A levels were lowest inc h i l d ren ,2 years of age.

The HLA-DQB1 typing was success-fully carried out in 647 patients. Childre ndiagnosed before the age of 2 years carr i e dthe high-risk genotype DQB1*02/*0302m o re often than the older childre n ,w h e reas the pro p o rtion of those carry i n gp rotective genotypes (DQB1z/z) incre a s e din the oldest age-group (Table 3).

C O N C L U S I O N S — In the pre s e n tseries, 35 of 801 (4.4%) patients diag-nosed with type 1 diabetes before the age

Figure 1—Median serum C-peptide levels in different age-groups at time of diagnosis and during fol -low-up. —, ,2 years; ? ? ?, 2.0–4.9 years; – – –, 5.0–14.9 years; P , 0.001.

F i g u re 2—Autoantibody positivity in three age-groups. h, ,2 years; j, 2.0–4.9 years; j, 5.0–14.9years; *P , 0.001.

DIABETES CARE, VOLUME 22, NUMBER 12, DECEMBER 1999 1953

Komulainen and Associates

of 15 years were ,2 years of age at thetime of clinical onset. The pro p o rtion ofboys in the youngest age-group was51.4%, and the pro p o rtion was somewhath i g h e r, although not signific a n t l y, in theolder age-groups. This is in agre e m e n twith a re p o rt from the Oxford region (22).

We have previously re p o rted that theoverall incidence of DKA in Finland is sim-ilar to that in many other countries (23). Inthe present study, we observed a conspicu-ously severe clinical decompensation at thediagnosis of type 1 diabetes in children ,2years of age. More than one-half of theyoungest children (53.3%) presented withdiabetic ketoacidosis, and close to one-t h i rd (31.4%) presented with impaire dconsciousness. However, similar pro p o r-tions have re p o rted elsewhere. Pinkney etal. (24) found that 19 of 43 (44%) childre ndiagnosed with type 1 diabetes in theO x f o rd region presented with a blood pHlevel of 7.35 or lower. In their series, 12 ofthese children (28%) had severe DKA (pH,7.10) compared with 10% of the chil-d ren ,2 years of age in our surv e y.

The reasons for the poor clinical con-dition in very young children manifestingwith diabetes are probably multifactorial.Their diabetic symptoms might have beend i fficult to recognize, especially as the dis-ease rarely manifests at an early age. Acuteinfections, which in our study were seen in

54.3% of the youngest children at the pre-sentation of type 1 diabetes, may also maskthe diabetic symptoms and signs, thusdelaying the clinical diagnosis of diabetes.On the other hand, the process of b- c e l ld e s t ruction leading to clinical type 1 dia-betes may be faster early in childhood thanlater in childhood. This suggestion is sup-p o rted by the finding that serum C-peptidelevels at diagnosis were lowest in childre ndiagnosed before 2 years of age, where a sglycated hemoglobin was highest in chil-d ren diagnosed $5 years of age, indicatinga longer duration of hyperglycemia in thelatter age-gro u p .

The age at onset of childhood diabeteshas previously been re p o rted to be posi-tively correlated to the postprandial plasmaC-peptide (25). It is also known thathealthy young children have lower seru mC-peptide levels than older ones (26). Wefound that children ,2 years of age losetheir endogenous insulin secre t o ry capacityfaster than older children, even after theclinical manifestation of type 1 diabetes.During the follow-up, the younger chil-d ren had lower random serum C-peptidelevels than the older children, and, actually,their endogenous insulin secretion did notrecover at all after the diagnosis, while asubstantial increase could be seen in thes e rum C-peptide concentrations over thefirst 3 months of overt disease in the oldest

a g e - g roup. Thus, our finding of lowers e rum C-peptide levels in very young chil-d ren versus those in older children pro b a-bly cannot solely be explained by a lowb-cell mass in very young children (27). Inaddition, none of the youngest childre nwas in partial remission at 6 months aftertheir diagnosis, while this phenomenonwas observed in 27.4% of the childre ndiagnosed with type 1 diabetes between2 and 4.9 years and in 37.1% of those inthe oldest age-group. These findings sug-gest that the loss of b-cell mass and func-tion pro g ressed faster in the very youngc h i l d ren than in older children, even afterclinical disease manifestation.

Metabolic control 6 months after diag-nosis, as defined by serum glycated hemo-globin, was poorest in the youngestc h i l d ren. This is partly explained by theird e c reased endogenous insulin secre t i o n .H o w e v e r, the diff e rences in GHb levelsamong the age-groups remained evenwhen only patients without partial re m i s-sion were included (data not shown).A c c o rd i n g l y, we speculate that other fac-tors, such as problems in cooperation, vari-able food intake and physical activity, andp a rental fears of hypoglycemia may havecontributed to poor metabolic control inv e ry young diabetic childre n .

It has previously been shown thatyoung children are more often positive forsome but not all diabetes-related autoanti-bodies at the time of diagnosis (5–9) andthat the age at which antibodies are detectedin the preclinical period is related to thepace of clinical diabetes pro g ression (28).Our data showed that the very young chil-d ren had the highest titers of IAA and ICAlevels. In contrast, the titers of GADA levelsw e re of the same magnitude, and the titersof IA-2 levels in very young children werelower than those of older children. On the

Table 2—Levels of autoantibodies according to age at diagnosis in subjects testing positive for autoantibodies

G roup 1 (,2 years) G roup 2 (2.0–4.9 years) G roup 3 (5.0–14.9 years) P

ICA (JDF U) 160 (113–320) 92 (34–169) 47 (18–141) 1 vs. 2 vs. 3: 0.0011 vs. 2: 0.0091 vs. 3: ,0 . 0 0 1

IAA (nU/ml) 675 (364–1044) 197 (106–494) 125 (79–268) 1 vs. 2 vs. 3: ,0 . 0 0 11 vs. 2: ,0 . 0 0 11 vs. 3: ,0 . 0 0 1

GADA (RU) 36.4 (10.6–64.0) 35.3 (12.8–87.0) 40.8 (16.3–88.5) N SIA-2A (RU) 6.0 (1.5–30.0) 52.1 (4.8–117.9) 71.8 (10.5–134.3) 1 vs. 2 vs. 3: ,0 . 0 0 1

1 vs. 2: 0.011 vs. 3: ,0 . 0 0 1

Data are medians (interq u a rtile range).

Table 3—Distribution of the four HLA-DQB1 genotypes

,2 Years of age 2.0–4.9 Years of age 5.0–14.9 Years of age P

* 0 2 / * 0 3 0 2 9 (31.0) 42 (35.6) 109 (21.8) 0 . 0 1 2* 0 3 0 2 / x 11 (37.9) 39 (33.1) 193 (38.7) N S* 0 2 / y 4 (13.8) 13 (11.0) 65 (13.0) N Sz / z 5 (17.2) 24 (20.3) 133 (26.5) 0 . 0 3 9

Data are n ( % ) .

1954 DIABETES CARE, VOLUME 22, NUMBER 12, DECEMBER 1999

Diabetes in very young children

other hand, children ,2 years of age testedpositive for IAA more often than older chil-d ren, while there were no significant diff e r-ences in the pro p o rtion of children testingpositive for ICA, GADA, and IA-2A amongthe three age-groups. We have pre v i o u s l yre p o rted that children diagnosed at $1 0years of age test positive for GADA moreoften than children ,10 years of age (8).This diff e rence cannot be seen in the pre s-ent study because all children $5 years ofage were grouped together. Contrary toGADA, we have not been able to show anys i g n i ficant association between age at diag-nosis and the frequency of IA-2A in Finnishc h i l d ren and adolescents diagnosed beforethe age of 15 years (9).

M o re than half (53.1%) of the youngestc h i l d ren tested positive for all four antibod-ies, suggesting a strong autoimmune attackagainst b-cells in very young children withnewly-diagnosed type 1 diabetes.

Patients diagnosed with type 1 dia-betes in childhood are even more likelythan adults to carry HLA genes associatedwith disease susceptibility (3,11,12). In ours t u d y, close to one-third of the children inthe two youngest age-groups (31.0 and35.6%, respectively) carried the high-riskgenotype DQB1*02/*0302 compared withonly 21.8% of the children in the oldesta g e - g roup. Corre s p o n d i n g l y, appro x i m a t e l yone-sixth (17.2%) of the children in theyoungest age-group carried pro t e c t i v egenotypes, which were seen in 20.3% ofthe children aged 2.0–4.9 years and 26.5%of the children in the oldest age-gro u p .

It has recently been suggested that theage at diagnosis of type 1 diabetes would begenetically determined (29,30). If so, onecould expect a strong genetic disease sus-ceptibility in patients diagnosed very earlyin life. In the present study, although theDQB1*02/*0302 genotype associated withthe highest risk was more common in chil-d ren #5 years of age, close to one-fif t h(17.2%) of the youngest children still car-ried protective genotypes. These fin d i n g ss u p p o rt the idea that factors other thanH L A - c o n f e rred genetic risk are import a n tin determining the timing of the clinicalmanifestation of type 1 diabetes.

A c k n o w l e d g m e n t s — The DiMe project hasbeen supported by the Juvenile Diabetes Foun-dation (grant 188517), the Association ofFinnish Life Insurance Companies, the SigridJuselius Foundation, the National Institutes ofHealth (DK-37957), the University of Helsinki,

the Nordisk Insulin Foundation, and the Novo-N o rdisk A/S, Denmark.

We are grateful to all participants and theirfamilies and to the respective hospitals and theirs t a ff. We thank L. Toivanen, J. Pitkäniemi, andE. Vi rtala for their skillful assistance in the col-lection and arrangement of the data, and S.Anttila, S. Heikkilä, P. Koramo, and R. Päkkiläfor their valuable technical assistance.

A P P E N D I X

Members of the DiMe Study GroupPrincipal investigators: H.K. Åkerblom andJ. Tuomilehto; Coordinators: R. Lounamaaand L. Toivanen; Data management: E. Vi r-tala and J. Pitkäniemi; Local Investigators: A.Fagerlund, M. Flittner, B. Gustafsson, C.Häggquist, A. Hakulinen, L. Herva, P.Hiltunen, T. Huhtamäki, N.-P. Huttunen, T.Huupponen, M. Hyttinen, T. Joki, R. Jok-isalo, M.-L. Käär, S. Kallio, E.A. Kaprio, U.Kaski, M. Knip, L. Laine, J. Lappalainen, J.Mäenpää, A.-L. Mäkelä, K. Niemi, A. Niira-nen, A. Nuuja, P. Ojajärvi, T. Otonkoski, K.Pihlajamäki, S. Pöntynen, J. Rajantie, J.Sankala, J. Schumacher, M. Sillanpää, M.-R.S t å h l b e rg, C.-H. Stråhlmann, T. Uotila, M.V ä re, P. Varimo, G. Wetterstrand; SpecialInvestigators: A. Aro (National Public HealthInstitute, Helsinki), M. Hiltunen (MedicalSchool, University of Ta m p e re), H. Hurm e(Åbo Akademi, Pedagogic Faculty in Va a s a ) ,H. Hyöty (Medical School, University ofTa m p e re), J. Ilonen (Department of Vi ro l o g y,University of Turku), J. Karjalainen (Depart-ment of Pediatrics, University of Oulu), M.Knip (Department of Pediatrics, Universityof Oulu), P. Leinikki (National Public HealthInstitute, Helsinki), A. Miettinen (Depart-ment of Bacteriology and Immunology, Uni-versity of Helsinki), T. Petäys (Department ofBacteriology and Immunology, University ofHelsinki), L. Räsänen (Department of Nutri-tion, University of Helsinki), H. Reijonen( D e p a rtment of Vi ro l o g y, University ofTurku), A. Reunanen (Social Insurance Insti-tute, Helsinki), T. Saukkonen (Childre n ’sHospital, University of Helsinki), E. Savilahti( C h i l d re n ’s Hospital, University of Helsinki),E. Tu o m i l e h t o - Wolf (National Public HealthInstitute, Helsinki), P. Vähäsalo (Depart m e n tof Pediatrics, University of Oulu), S.M. Vi r-tanen (Department of Nutrition, Universityof Helsinki).

R e f e re n c e s1 . Dahlquist G, Hustonen L: Analysis of a 15

year prospective incidence study of child-

hood diabetes onset: time trends and cli-matological factors. Int J Epidemiol 2 3 :1234–1241, 1994

2 . Tuomilehto J, Podar T, Tu o m i l e h t o - Wolf E,Vi rtala E: Evidence for importance of gen-der and birth cohort for risk of IDDM ino ffspring of IDDM parents. D i a b e t o l o g i a38:975–982, 1995

3 . Karjalainen J, Salmela P, Ilonen J, Surcel H-M, Knip M: A comparison of childhoodand adult type 1 diabetes mellitus. N Engl JM e d 320:881–886, 1989

4 . L e v y - M a rchal C, Papoz L, de Beaufort C,D o u t reix J, Froment V, Voirin J, Czern i-chow P: Clinical and laboratory findings oftype 1 diabetic children at the time of diag-nosis. Diabet Med 9:279–284, 1992

5 . Arslanian SA, Becker DJ, Rabin B, Eber-h a rdt M, Cavender D, Dorman J, Drash AL:C o rrelates of insulin antibodies in newlydiagnosed children with insulin-dependentdiabetes before insulin therapy. D i a b e t e s320:926–930, 1985

6 . Karjalainen J, Knip M, Mustonen A,Åkerblom HK: Insulin autoantibodies at theclinical manifestation of type 1 (insulin-dependent) diabetes: a poor predictor ofclinical course and antibody response toexogenous insulin. D i a b e t o l o g i a 3 1 : 1 2 9 – 1 3 3 ,1 9 8 8

7 . Kolb H, Dannehl K, Grüneklee D, ZielasekJ, Bertrams J, Hübinger A, Gries FA :P rospective analysis of islet cell antibodiesin children with type 1 (insulin-dependent)diabetes. D i a b e t o l o g i a 31:189–194, 1988

8 . Sabbah E, Kulmala P, Veijola R, Vähäsalo P,Karjalainen J, Tu o m i l e h t o - Wolf E, ÅkerblomHK, Knip M, the Childhood Diabetes inFinland Study Group: Glutamic acid decar-boxylase antibodies in relation to otherautoantibodies and genetic risk markers inc h i l d ren with newly diagnosed insulin-dependent diabetes. J Clin Endocrinol Metab81:2455–2459, 1996

9 . Savola K, Bonifacio E, Sabbah E, Kulmala P,Vähäsalo P, Karjalainen J, Tu o m i l e h t o - Wo l fE, Meriläinen J, Åkerblom HK, Knip M, theChildhood Diabetes in Finland StudyG roup: IA-2 antibodies: a sensitive markerof IDDM with clinical onset in childhoodand adolescence. D i a b e t o l o g i a 4 1 : 4 2 4 – 4 2 9 ,1 9 9 8

1 0 . Knip M, Ilonen J, Mustonen A, ÅkerblomHK: Evidence of accelerated b-cell destru c-tion in HLA-Dw3/Dw4 heterozygous chil-d ren with type I (insulin-dependent)diabetes. D i a b e t o l o g i a 29:347–351, 1986

1 1 . Caillat-Zucman S, Garchon HJ, Timsit J,Assan R, Boitard C, Djilali-Saiah I,B o u g n e res P, Bach JF: Age-dependent HLAgenetic heterogeneity of type 1 insulin-dependent diabetes mellitus. J Clin Invest90:2242–2250, 1992

1 2 . Shield J, Wa d s w o rth E, Baum JD: Thegenetic contribution to disease pathogene-sis in childhood diabetes is greatest in the

DIABETES CARE, VOLUME 22, NUMBER 12, DECEMBER 1999 1955

Komulainen and Associates

v e ry young. Diabet Med 12:377–379, 19951 3 . Tuomilehto J, Lounamaa R, Tu o m i l e h t o -

Wolf E, Reunanen A, Vi rtala E, Kaprio EA,Åkerblom HK, the Childhood Diabetes inFinland (DiMe) Study Group: Epidemiol-ogy of childhood diabetes mellitus in Fin-land: background of a nationwide study oftype 1 (insulin-dependent) diabetes melli-tus. D i a b e t o l o g i a 35:70–76, 1992

1 4 . Bottazzo GF, Florin-Christiansen A,Doniach D: Islet cell antibodies in diabetesmellitus with autoimmune polyendocrined e ficiences. L a n c e t ii:1279–1282, 1974

1 5 . L e rnmark Å, Molenaar JL, van Beers WM,Yamaguchi Y, Nagataki S, Ludvigsson J,M a c l a ren NK: The fourth intern a t i o n a ls e rum exhange workshop to standard i z ecytoplasmic islet cell antibodies. D i a b e t o l o -g i a 35:70–76, 1992

1 6 . Palmer JP, Asplin CM, Clemons P, Lyen K,Tatpati O, Raghu PK, Paguette TL: Insulinantibodies in insulin-dependent diabeticsb e f o re insulin treatment. S c i e n c e 2 2 2 : 1 3 3 7 –1339, 1983

1 7 . Petersen JS, Hejnaes KR, Moody A, KarlsenAE, Marshall MO, Hoier-Madsen M, Boel E,Michelsen BK, Dyrberg T: Detection ofGAD65 antibodies in diabetes and otherautoimmune diseases using a simple radio-ligand assay. D i a b e t e s 43:459–467, 1994

1 8 . Schmidli RS, Colman PG, Bonifacio E, par-ticipating laboratories: Disease sensitivityand specificity of 52 assays for glutamic

acid decarboxylase antibodies: the SecondI n t e rnational GADab Workshop. D i a b e t e s44:636–640, 1995

1 9 . Ve rge CF, Stenger D, Bonifacio E, ColmanPG, Pilcher C, Bingley PJ, Eisenbarth GS,p a rticipating laboratories: Combined useof autoantibodies (IA-2 autoantibody,GAD autoantibody, insulin autoantibody,cytopalsmic islet cell antibodies) in type 1diabetes: Combinatorial Islet Autoanti-body Workshop. D i a b e t e s 4 7 : 1 8 5 7 – 1 8 6 6 ,1 9 9 8

2 0 . S j ö roos M, Iitiä A, Ilonen J, Reijonen H,L ö v g ren T: Triple-label hybridization assayfor type 1 diabetes-related HLA alleles.B i o t e c h n i q u e s 18:870–877, 1995

2 1 . Heding LG: Radioimmunological determ i-nation of human C-peptide in serum. D i a -b e t o l o g i a 11:541–548, 1975

2 2 . G a rdner SG, Bingley PJ, Sawtell PA, We e k sS, Gale EAM, the Bart ’s - O x f o rd StudyG roup: Rising incidence of insulin depen-dent diabetes in children aged under5 years in the Oxford region: time tre n danalysis. B M J 315:713–717, 1997

2 3 . Komulainen J, Lounamaa R, Knip M,Kaprio EA, Åkerblom HK, the ChildhoodDiabetes in Finland Study Group: Ketoaci-dosis at the diagnosis of type 1 (insulindependent) diabetes mellitus is related topoor residual beta cell function. A rch DisC h i l d 75:410–415, 1996

2 4 . Pinkney JH, Bingley PJ, Sawtell PA, Dunger

DB, Gale EAM, the Bart ’s - O x f o rd StudyG roup: Presentation and pro g ress of child-hood diabetes mellitus: a prospective pop-ulation-based study. D i a b e t o l o g i a 3 7 : 7 0 – 7 4 ,1 9 9 4

2 5 . Wallensteen M, Dahlquist G, Persson B,Landin-Olssonn M, Lernmark A, SundkvistG, Thalme B: Factors influencing the mag-nitude, duration, and rate of fall of B-cellfunction in type 1 (insulin-dependent) dia-betic children followed for two years fro mtheir clinical diagnosis. D i a b e t o l o g i a 3 1 :664–669, 1988

2 6 . Knip M, Åkerblom HK: Plasma C-peptideand insulin in neonates, infants, and chil-d ren. J Pediatr 99:103–105, 1981

2 7 . Rahier J, Wallon J, Henquin J-C: Cell pop-ulations in the endocrine pancreas ofhuman neonates and infants. D i a b e t o l o g i a20:540–546, 1981

2 8 . Bingley PJ, Bonifacio E, Shattock M, Gill-mor HA, Sawtell PA, Dunger DB, Scott RD,Bottazzo GF, Gale EA: Can islet cell anti-bodies predict IDDM in general popula-tion? Diabetes Care 16:45–50, 1993

2 9 . Wa d s w o rth EJK, Shield JPH, Hunt LP, BaumJD: A case-control study of enviro n m e n t a lfactors associated with diabetes in the under5s. Diabet Med 14:390–396, 1997

3 0 . Fava D, Gardner S, Pyke D, Leslie RDG:Evidence that the age at diagnosis of IDDMis genetically determined. Diabetes Care 2 1 :925–929, 1998