| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Other Original Articles |
Division of Endocrinology and Diabetes, LAC-USC Medical Center (T.A.B.), Keck USC School of Medicine (A.Z., S.J.S.), Los Angeles, California 90033; Division of Medical Genetics, Cedars-Sinai Research Institute and UCLA (L.J.R., J.I.R.), Los Angeles, California 90048; Division of Endocrinology and Metabolism, Children’s Hospital of Los Angeles (G.C.), Los Angeles, California 90027; California Children’s Hospital of Oakland Research Institute (J.P.K., C.W.), Oakland, California 94611; Division of Endocrinology, Metabolism, and Nutrition, University of Washington, VA Medical Center (J.P.), Seattle, Washington 94609; H. Lee Moffitt Cancer Center and Research Institute (J.P.K.), University of South Florida, Tampa, Florida 33612; and Weill Medical College at Cornell University (N.K.M.), New York, New York 10021
Address all correspondence and requests for reprints to: Adina Zeidler, M.D., Division of Endocrinology and Diabetes 1200 North State Street, Los Angeles, California 90033. E-mail:azeidler{at}hsc.usc.edu
Abstract
As part of a genetic study of type 1 diabetes in
Mexican-Americans, 360 first-degree relatives of 108 type 1
diabetic probands were studied. Islet cell antibody (ICA), insulin
autoantibody, glutamic acid decarboxylase (GAD65), and
protein tyrosine phosphatase autoantibodies were measured and human
leucocyte antigen (HLA) class II alleles DRB1 and DQB1
genotyping was performed. ICA was positive in 37% of the probands and
5.8% of the relatives. A subgroup of 26 probands (12 ICA+, 14 ICA-)
was tested for GAD65 and was found positive. 4/14 ICA+
first-degree relatives were GAD65 positive. Four relatives,
positive for two antibodies, subsequently developed type 1 diabetes.
Life-Table
analysis of first-degree relatives with autoantibodies
indicated an 80% disease-free survival at 3.5 yr. HLA-DRB1 was found
to be associated with the presence of ICA in both probands and
relatives, whereas HLA-DPB1 was associated with autoantibody in
relatives of type 1 diabetic probands.
|
STUDIES IN FAMILIES with type 1 diabetes support the concept that this disease results from chronic autoimmune-mediated destruction of the pancreatic ß-cells (1). The clinical manifestation of type 1 diabetes is preceded by a variable prodromal period, during which cellular and humoral immune changes can be detected in the peripheral blood. The progression to clinical disease occurs in those with islet cell autoantibodies (ICA) and abnormal insulin secretion; however, disease progression is faster in those with multiple autoantibodies (2, 3, 4, 5, 6, 7). Individuals at risk of developing type 1 diabetes can also be identified by their genetic markers (8, 9, 10). Of several immunological markers for the development of type 1 diabetes, ICA is currently the most widely used, despite uncertainties about the target antigens involved. Most current models of prediction of type 1 diabetes in first-degree relatives are based on this marker (3), though some researchers believe that measurement of GAD65Ab may soon take precedence, especially if combined with antibodies to tyrosine phosphatase IA-2 termed ICA-512 (11, 12, 13). ICA is detectable in 60–80% of newly diagnosed patients with type 1 diabetes (3) and in 3–5% of their first-degree relatives (5). It has been estimated that 17% of ICA positive human leucocyte antigen (HLA) identical siblings will develop type 1 diabetes, 5% if sharing one HLA haplotype, and only 1% if sharing none (5). When additional markers such as impaired first phase insulin release (2), presence of autoantibody (IAA) and GAD65Ab are detected in the same individual, the estimated risk of developing type 1 diabetes increases significantly (6, 7, 14).
Studies regarding type 1 diabetes in Mexican-Americans and their first-degree relatives are rare (9, 10, 15). Thus, the extent to which the same methods used in Caucasians to predict the risk for type 1 diabetes can be used in Mexican-Americans is not known. The present study was designed to determine the risk of developing type 1 diabetes in first-degree relatives of Mexican-American type 1 diabetic patients, by assessing autoantibodies and HLA class II alleles.
Patients and Methods
Human subjects
Mexican-American families were recruited through our ongoing type 1 diabetes family study at the Los Angeles County Hospital/University of Southern California, Children’s Hospital of Los Angeles, USC Keck School of Medicine. The study was designed to determine the type and prevalence of type 1 diabetes in the Mexican-American population in Southern California (15). Type 1 diabetes was defined according to the National Diabetes Data Group (16). To be considered of Mexican ancestry, the type 1 diabetic probands and control subjects had to have a minimum of 3 out of 4 grandparents born in Mexico (15). Between 1989 and 1992, 360 first-degree relatives from 107 families with a type 1 diabetic proband were tested for autoantibodies and HLA class II DRB1 and DQB1. The number of individuals tested for ICA, IAA, GAD65, and IA-2 antibodies varied, depending on the amount of serum that was available. HLA-DPB1 typing was performed on a subset of the study population. Samples were obtained from 107 families, 108 type 1 diabetic patients, 102 mothers, 61 fathers, 85 brothers, and 112 sisters. Seventy-two healthy individuals (42 males and 30 females) of Mexican ancestry without a known family history of diabetes were recruited from orthopedic surgery clinics and served as controls.
The study was approved by the Institutional Review Board Committee. Written informed consents were obtained from all subjects and/or their parents before their participation in the study. All subjects/parents were requested to complete a detailed questionnaire including demographics, and the medical history of each proband and their first-degree relatives.
ICA, IAA, GAD65, and IA-2 Ab assays
ICA was determined using an indirect immunofluorescence technique using cryosections (4 µm) of fresh, type O human blood group pancreatic tissue as substrate (7). The ICA values are reported in JDF units based on comparison to endpoint titration of the JDF standard serum. A positive test was one with a titer >5 JDF units. The interassay coefficient of variation of the ICA was 9%. The assay was 100% specific and had a sensitivity of 74.4% for new onset diabetic patients of age less than 30 yr. In multiple proficiency testing programs, the laboratory specificity and sensitivity remained close to 100%.
IAA was performed using human recombinant 125I labeled mono-specific A-14 insulin (kindly provided by Eli Lilly & Co., Indianapolis, IN) as the ligand source as previously described (17). Values over 125 µl/ml (±3 SD) were considered to be IAA positive. The 125 µl/ml cutoff value was established from representing the mean ± 3 SD of normal controls (7) of 150 healthy adults and children. IAA was not measured in the type 1 diabetic probands because all patients were treated with insulin. GAD65 was performed using a Depleted ELISA (DELISA system) sensitive nonisotope assay (18). The cutoff for GAD65 was the mean ± 3 SD of normal controls. The intraassay and interassay coefficient of variations were 4.8% and 6.9% respectively. IA-2 assay was performed as previously described (7). Human full-length IA-2 and the internal domain of IA-2ß were expressed from cDNAs in rabbit reticulocytes and used to provide recombinant 35S methionine labeled protein for radio precipitation analyses. The intensity of the reaction was assessed by densitometry and thereby determined to be positive or negative. In the Immunology Diabetes Workshop, laboratory proficiency testing results for GAD65 and IA-2 were 100% accurate, sensitive, and specific. The cutoff levels are 0.02 for GAD65 and 0.04 for IA-2.
HLA class II typing
DNA was prepared either directly from buffy coats isolated from whole blood collected in EDTA-containing tubes or from cell pellets prepared from Epstein-Barr virus transformed lymphoblastoid cell lines (19). HLA class II typing for DRB1, DQB1, and DPB1 was performed using PCR/SSOP methods that have been described previously (20, 21, 22, 23, 24, 25). In brief, DNA is amplified by PCR with primers specific for the second exon of the relevant locus. The resultant PCR product is then denatured and immobilized onto nylon membrane. Membranes are subsequently hybridized with horseradish peroxidase-labeled or, in some cases, biotin-labeled, sequence-specific oligonucleotide probes. Following stringent washing, membranes hybridized to biotin-labeled probes are incubated with streptavidin-horseradish peroxidase. Bound probe is detected by means of a colorimetric detection system using the substrate tetramethylbenzidine. The pattern of probes hybridizing to a given sample indicates the identity of the alleles represented in the PCR product. In some cases, biotinylated PCR primers are used to generate labeled PCR products, which are subsequently hybridized to a set of unlabeled oligonucleotide probes immobilized on a membrane in a method referred to as "reverse dot blot" or "reverse line blot" (23).
Statistical methods
The results are expressed as mean ± SEM.
Comparison between groups for ICA, IAA, GAD65Ab,
and IA-2 were analyzed using the 
square test or the Fisher exact
test. The G-test for heterogeneity was used to compare allele and
haplotype frequencies in autoantibody positive and negative
first-degree relatives and probands. The relative risk (RR) or odds
ratio was calculated from the numbers of antibody positive and negative
subjects with and without the HLA marker. Life Table
analysis was used
to estimate the time of development of type 1 diabetes in the
first-degree relatives. Follow up time for each subject was calculated
from the date when ICA > 5 JDF units was first detected
(26).
Results
The male to female ratio was similar in type 1 diabetes probands and controls, but a predominance of females occurred in the first-degree relatives tested (M:F 1:1.5). The age range for the diabetics was 3–48 yr, mean 16.3 ±8.6 SD and a median of 13 yr. The siblings were similar to the index cases with an age range of 3–38 yr, mean of 16.6 ± 8.6 yr and a median of 13 yr. The parents’ age range was 23–63 yr with a mean of 39.6 ± 8.6 yr and a median of 38 yr.
Islet cell, antiinsulin, glutamic acid decarboxylase, and IA-2 antibodies
Thirty-seven percent of the type 1 diabetes patients, 5% of the
parents, and 7% of the siblings were ICA positive, whereas only 1/72
controls (1.4%) was ICA positive (Table 1). Of 75 parents and 78 siblings tested
for IAA (Table 1), IAA was detected in 4% and 6%, respectively. IAA
was not tested in the probands because all patients were treated with
insulin. Two parents and five siblings were positive for both ICA and
IAA. Twenty-six type 1 diabetic patients (12 ICA+, 14 ICA-) and 14
ICA+ first-degree relatives were tested for GAD65
antibodies. All of the diabetic subjects, both ICA+ and ICA-, were
GAD65+. However, out of the 14 ICA+ relatives,
only 4 (29%) were GAD65+. IA-2 antibodies were
detected only for 1/197 (0.5%) relatives. Members of 23 families (25
siblings and 27 parents) were retested for ICA after 1 yr of follow up;
none of the ICA negative first-degree relatives developed diabetes. In
contrast, 4 of the 21 ICA+ first-degree relatives developed type 1
diabetes over the course of the study (one father [33 yr.], one
mother [58 yr.], and two siblings [7 and 18 yr.]), all from
different families. The first-degree relatives who subsequently
developed diabetes were treated with insulin following diagnosis,
except one mother, who was treated initially with oral hypoglycemic
agents but was changed to insulin within a year. She was positive for
both ICA (320 JDF) and IAA (1167 µU/ml).
Although not reaching statistical significance, the highest observed
prevalence of ICA (50%) in type 1 diabetes patients was noted in the
31- to 40-yr-old group, followed in decreasing order by the 3–10
(44%), 11–20 (43%), and 21–30 (7%) year-old groups. Type 1
diabetic probands with the shortest duration of diabetes had the
highest frequency of ICA (Fig. 1). In
siblings, the 11- to 20-yr-old group had the highest ICA frequency
(10%), whereas the lowest ICA rate (3%) was noted at 21–30 yr. As
expected, parents in the 41- to 50-yr-old group was the least ICA
positive (2%). The cumulative risk for the development of type 1
diabetes in the ICA+ relatives is shown in Fig. 2. An 80% disease-free survival at 3.5
yr in relatives who are positive for ICA.
|
|
The presence of ICA antibodies was associated with HLA-DRB1, both
in probands (P < 0.01) and in nondiabetic first-degree
relatives (P < 0.005) (Table 2). No probands carrying HLA-DRB1*0802 or
1104 were ICA+ (RR 0.1 and 0.2, respectively), whereas DRB1*0403 (RR
10.7) and 0404 (RR 4.6) were associated with an increased risk for ICA+
among probands. As shown in Table 2, DRB1*0701 (RR 0.3), 0802 (RR 0.2),
1104 (RR 0.2), 1302 (RR 0.3), and 1602 (RR 0.2) occurred infrequently
in ICA+ relatives, whereas HLA-DRB1*0301 appeared to be associated with
an increased risk for ICA+ (RR 2.3). DRB1*0403 also had an increased RR
(7.2), but the significance of this finding needs to be clarified given
the small number of 0403 alleles in the sample. No association was
found between HLA-DQB1 and ICA antibodies in either the probands or the
nondiabetic relatives. The presence of IAA was associated with
DPB1*0501 (P < 0.05) in nondiabetic relatives, but not
with DQB1 or DRB1 (data not shown). While the numbers are small (only 4
of 44 relatives with both IAA and DPB1 tested were IAA+), DPB1*0501 was
associated with a markedly increased risk of IAA+ (RR 25.8). No DPB1
association with ICA+ was observed in first-degree relatives, nor was
DRB1 or DQB1 associated with IAA.
|
Discussion
The study was designed to assess the prevalence of autoantibodies in Mexican-American patients with type 1 diabetes and to examine whether the presence of immune markers and specific HLA alleles could be used to predict the risk for type 1 diabetes in their first-degree relatives. The results indicate that Mexican-American patients with type 1 diabetes at various periods after their diagnosis have an overall prevalence of ICA of 37%, which is similar to a previous study in Caucasians (27). The prevalence of ICA and IAA noted in the parents and siblings in this study are not significantly greater than described in Caucasians (7). The prevalence of positive ICA (7%) in siblings of type 1 diabetes patients in this study, compared with our previous study (3%) in 59 siblings of another group of Mexican-American type 1 diabetes probands, was not significantly different, (13/197 vs. 2/55, P = 0.62) (28).
The association between ICA and HLA-DRB1 observed was expected because this locus is known to be associated with type 1 diabetes risk, particularly within type 1 diabetes families (29). The HLA-DRB 0403 subtypes increase the risk of type 1 diabetes and are greatly variable in different populations. Specifically, HLA-DRB1* 0401 and 0404 have been shown to be associated with type 1 diabetes (30, 31, 32, 33, 34, 35). The relative risk ratios seen in the present study suggest that there may not always be a direct correlation between the alleles associated with diabetes risk and those associated with the development of autoantibodies. Further studies involving larger numbers of antibody-positive first-degree relatives will be necessary to determine whether there are some class II alleles that contribute to the risk of antibodies, without necessarily having a similar effect on disease risk, in Mexican-American families.
As expected, we found that the greatest prevalence of positive antibodies in siblings occurred in groups aged less than 20 yr. In the Mexican-American parents, the peak prevalence of ICA was noted in the 21- to 30-yr-old group.
It has been suggested that the presence of ICA can often precede the onset of diabetes by 3–5 yr or longer. Thus, the presence of positive ICA in type 1 diabetic siblings younger than 10 yr gives immunologic support to epidemiologic findings indicating that peak incidence of newly diagnosed type 1 diabetes occurs between 8 and 12 yr (36). The findings of positive ICA and IAA in two parents, who then went on to develop clinical diabetes, suggest that type 1 diabetes in Mexican-American adults may masquerade as type 2 diabetes mellitus (37). The decrease in frequency of positive ICA in type 1 diabetes patients with increasing duration of the disease noted here is similar to reports in Caucasians and supports our previous findings in Mexican-American type 1 diabetic patients (27).
Autoantibodies to GAD65 have been used as a marker for type 1 diabetes. Although the number of patients tested for GAD65 is small, we found that 29% of the ICA positive first-degree relatives were also GAD65Ab positive, which is lower compared with previous reports in Caucasians (7, 38). While the very low prevalence of IA-2 antibodies in our study makes it difficult to draw any conclusions, the fact that none of the 4 relatives who developed diabetes were IA-2 positive suggests that this antibody may not be a very sensitive predictor of risk in this population. All four relatives who developed diabetes tested positive to two autoantibodies (ICA and IAA or ICA and GAD65). Our findings are consistent with a recent report indicating that first-degree relatives of type 1 diabetes of Hispanic origin, who have multiple antibodies, are at significant risk for developing the disease (7).
Previous studies have suggested that prediction of subsequent
development of type 1 diabetes can best be made in subjects with ICA
titer of >40 JDF-U (5, 6) coupled with a decreased
insulin secretion more than 2 SD from the mean (6, 7, 39). However, when two or more autoantibodies are present, the
risk of developing type 1 diabetes is highly increased (6, 7). The Life-Table analysis of relatives of Mexican-American
type 1 diabetics in this study indicates an 80% type 1 diabetes free
survival at 5 yr follow-up. Follow-up studies of larger numbers of
Mexican-American type 1 diabetes families will be necessary to confirm
if the risk to autoantibody positive relatives is similar to that of
Caucasians. Such studies may also help determine if the prevalence of
type 1 diabetes is lower in Mexican-Americans than in Caucasians, as is
currently believed. Our data suggests the possibility that some of the
diagnoses of type 2 diabetes in first-degree relatives of type 1
diabetic patients may, in fact, represent misclassified cases of type 1
diabetes with a later onset.
This is the first study to assess autoimmunity and preclinical type 1 diabetes in families of Mexican-American patients. Better understanding of the natural history of pre-clinical type 1 diabetes in the Mexican-American population with long-term follow up of first-degree relatives will have practical implications for possible prevention of the disease in this ethnic group.
Acknowledgments
We acknowledge Dr. Melchor Alpizar, Jamie Kaplan, Frank Martinez, and Steve Rad for their excellent assistance. Jinyan Lu’s assistance with the statistical analyses is greatly appreciated and we thank William Klitz, Ph.D., for his kindness in making his G-statistic program available to us. We are also thankful to Denise Edwards for her word processing assistance.
Footnotes
This study was supported in part by grants from the Gustavus and Louis Pfeiffer Research Foundation, General Clinical Research Center branch of the National Center for Research Resources (No. M01RR43), and by R01-MD-194-67-12 (to N.M.). We are thankful to Susie Nakao and the GCRC nursing staff for their assistance and commitment in conducting this clinical study.
Abbreviations: GAD65 Glutamic acid decarboxylase; HLA, human leucocyte antigen; IAA, autoantibody; ICA, islet cell antibody; RR, relative risk.
Received March 29, 2000.
Accepted June 27, 2001.
References
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
