APMIS 100: 709-712. 1992

The prevalence of selective IgA deficiency in type 1 diabetes mellitus

ROLAND S. LIBLAU, SOPHIE CAILLAT-ZUCMAN, ANNE-MARIE FISCHER, JEAN-FRANCOIS BACH and CHRISTIAN BOITARD

Department of Clinical Immunology, TNSERM U 25 and CNRS UA 122, and Department of Haematology, Necker Hospital, Paris, France

Liblau, R. S., Caillat-Zucman, S., Fischer, A-M., Bach, J-F. & Boitard, C. The prevalence of selective IgA deficiency in type 1 diabetes mellitus. APMIS 100: 709-712, 1992.

A significant increase in the prevalence of selective IgA deficiency has been observed in patients with autoimmune disorders such as systemic lupus erythematosus and rheumatoid arthritis. Insulin- dependent diabetes mellitus (IDDM) is an autoimmune disease and susceptibility to both IDDM and IgA deficiency is associated with HLA DQBl alleles encoding non-Asp amino acids at position 57. In order to assess whether the prevalence of selective IgA deficiency is increased in IDDM, we have screened a homogeneous series of adult patients with IDDM for selective IgA deficiency. One patient (1:261) was found to have a selective IgA deficiency. The prevalence of selective IgA deficiency among adult French blood donors is 1:1400. Thus, although IDDM and selective IgA deficiency are both associated with the presence of non-Asp amino acids at position 57 of the HLA DQP chain, the frequency of this immunodeficiency in adult IDDM patients is not significantly increased.

Key words: Selective IgA deficiency; autoimmune disease; insulin-dependent diabetes mellitus; HLA- DQ molecule.

Roland S. Liblau, Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA 94305, USA.

Selective IgA deficiency has been defined on the basis of four criteria: 1) serum IgA level below 0.05 g/l, 2) no deficiency of serum IgG and IgM, 3) normal humoral antibody production, 4) nor- mal or near normal cellular immunity (Burks & Steele 1986). It is the commonest primary immu- nodeficiency with a reported prevalence of ]:I400 in French blood donors (Ropars et al. 1982). Selective IgA deficiency is often asympto- matic, but has been observed with a higher fre- quency in patients with autoimmune diseases such as systemic lupus erythematosus (SLE) and rheumatoid arthritis (Cassidy et al. 1968, Burks & Steele 1986). This immunodeficiency is associated with HLA susceptibility genes (Cun- ningham- Rundles 1990) and a recent report

Received January 23, 1992. Accepted March 5, 1992.

underlines its association with homozygosity for HLA DQBl alleles encoding non-Asp residues at position 57 (Olerup et al. 1990).

Insulin-dependent diabetes mellitus (IDDM) results from an autoimmune destruction of pan- creatic Beta cells and susceptibility to IDDM is also associated with HLA DQBl alleles encod- ing non-Asp residues at position 57 (Todd 1990). We have therefore assessed the prevalence of IgA deficiency in a homogeneous population of adult patients with recent-onset IDDM.

PATIENTS AND METHODS

Two hundred and sixty-one consecutive adult diabetic patients, symptomatic for less than six months and receiving insulin therapy for less than two months, were included in this study. All the patients had ke-

709

LIBLAU rt a1.

toacidosis at the time of diagnosis. The ages ranged from 15 to 48 years (mean 26.4). Patients with haemo- chromatosis, chronic pancreatitis or other types of secondary diabetes were excluded. One hundred nor- mal blood donors (ages ranging from 18 to 60) served as control subjects.

Quantitative determination of serum IgA, IgG and IgM was performed by laser nephelometry. Islet cell antibodies (ICA) were determined in 227 diabetic patients using indirect immunofluorescence on group 0 human frozen pancreas sections as previously de- scribed (Boitard et al. 1987). HLA class I1 oligotyping of 11 5 diabetic patients and 100 control subjects was performed as described (Mach & Tiercy 1991). Brief- ly, the second exons of DRB1, DQAl and DQBl genes were independently amplified according to the polymerase chain reaction procedure; dot-blot hy- bridization using sequence specific oligonucleotide probes allowed identification of 21 DRB1, 8 DQAl and 13 DQBl alleles.

Statistical analyses: The size of the diabetic group permitted detection with high probability of a 20-fold increase in the prevalence of IgA deficiency among diabetic patients compared to blood donors. A 20- fold increase was selected because such an increase has been observed for another autoimmune disease, namely SLE (Cassidy et al. 1968, Yewdall et al. 1983, Rijle et al. 1988). According to the Poisson approxi- mation to the binomial distribution, a sample size of 261 patients is sufficient to give a one-tailed a=O.O5 level test with power above 85%.

RESULTS

As shown in Table 1, IgA levels did not differ between the control group and the diabetic pa- tients. None of the control subjects had IgA deficiency. Among the 261 diabetic patients, 63% had significant ICA levels (142 out of 227 tested) and one was found to have a previously

TABLE 1 . Serum IgA levels in 261 type I diabetic vatients and 100 control subiects

unknown selective IgA deficiency. Thus, the prevalence of this immunodeficiency in adult type 1 diabetic patients was 0.38% (95% confi- dence interval, 0.01% to 2.1%). This prevalence is not significantly different from that of French blood donors (p = 0.17, one-tailed Fisher's exact test; odds ratio, 5.4; the exact 95% confidence interval of the odds ratio is 0.13 to 3 1.2). Among the diabetic patients typed for HLA class 11, the frequency of homozygotes for DQP non-Asp residues at position 57 was significantly in- creased and the frequency of homozygotes for Asp residues at position 57 was significantly decreased as compared to the control group (Table 2). The serum IgA levels from diabetic patients homozygous for DQP non-Asp residues at position 57 (2.19k0.92 g/l; n=64) did not differ significantly from those of non-Asp /Asp heterozygous patients (1.91 k 0.72 g/l; n = 43).

The IgA-deficient diabetic patient was a 17- year-old Caucasian boy with no significant medical history before developing diabetes. His serum IgG level was 19.7 g/l (reference values, 6.4-14.8 g/l); IgM, 0.78 g/l (reference values, 0.5-2.9 g/l); IgA, less than 0.04 g/ l (reference values, 0.7-3.2 g/l); and IgE, less than 10 UI/ ml. IgA was not detectable by serum immuno- electrophoresis and the anti-kappa and anti- lambda light chain reactivity displayed no ab- normality. Serum IgG subclass levels were within normal ranges. In the saliva, IgA was less than 0.04 g/l, IgM was 0.08 g/l and IgG 0.62 g/l. Anti-IgA antibodies were not found. ICA were detected at a dilution of 1:8. Antinuclear antibodies were present at a dilution of 150 with a speckled fluorescence. The HLA typing of the patient was HLA A 26 -; B 14-39; DR

and he was thus DQP non-Asp/Asp hetero- zygous at position 57. The mother, father and

1-8; DQAl* 0101*-0401; DQBI* 0501*-0402

IgA levelsa Diabetic patients Normal subiects

<0.05 g/l 1 0.05-0.55 & / I 1 0.56b-3.84 g/l" 249 3.85-5.0 gil 6 >5.0 ell 4

0 0

96 4 0

"IgA G e l s (mean+SD) in control group: 2.20 gllS0.82; in diabetic patients: 2.19 g/1_+0.89 (p= 0.91, unpaired two-tailed t test). Mean value of control group -2 SD. Mean value of control group + 2 SD.

710

TABLE 2. Amino acid residues at position 57 of the HLA-DQ/3 molecule in 115 insulin-dependent diabetic

Datients and 100 control subiects Non-Asp/ Non-Asp/ AspIAsp Non-Asv ASD

Diabetic patients 64d 43 Normal controls 26 47 27 a Chi-square= 18.1, p < 0.0001 when compared to

the control group. Chi-square= 14.3, p=0.0002 when compared to the control group.

IgA DEFICIENCY IN TYPE 1 DIABETES MELLITUS

TABLE 3. Prevalence of' selective IgA deficiency in patients with autoimmune diseases Disease Authors Number of patients Number of patients Criterion for selective

(age) with selective IgA IgA deficiency deficiency

Systemic Cassidy et al." 87 4 (4.6%) Undetectable

erythematosus Yewdull et al.' 138 2 (1.45%) <0.05 g/ l R f l e et idd 72 2 (2.5%)) <0.05 g/l

Juvenile Bluestone et ul.' 200 4 (2%) Not specified

lupus Gershwin et al. I14 3 (2.6%) CO.10 g / l

rheumatoid Cussidy et ul.' 324 14 (4.3%) <0.01 g/l arthritis Barkley et a1.P 582 12 (2.1%) co.10 g/l

Pelkonen et al. 350 10 (2.9%) <0.02 g / l Rheumatoid Cassidy et aLd 61 I (1.6%) Undetectable arthritis Myasthenia Behan et al.' 51 2 (3.9%) Undetectable gravis Libluu et al.' 333 1 (0.3%) <0.05 g/ l Insulin-dependent Smith et al. 421 (> 16 years) 0 (O.O%,) <0.10 g/l diabetes 366 (I 16 years) 9 (2.5%) CO.10 g/l mellitus Hoddinott et al.' 129 (8 to 51 years) 2 (1.6%) <0.05 g/ l

Present study 261 ( > 16 years) 1 (0.4%) <0.05 g/l Cassidy et al. N. Engl. J. Med. 280: 275, 1968. Gershwin et al. J. Pediatr. 89: 377-381, 1976. Yewdull et

al. J. Clin. Lab. Immunol. 10: 13-18, 1983. Bluestone et a f . J . Pediatr. 77: 98-102, 1970. ' Cassidy et al. Arthritis Rheum. 20: 181-183, 1977. Barkley et al. J. Rheumatol. 6: 219-224, 1979. Pelkonen et al. Scand. J. Rheumatol. 12: 273-279, 1983. ' Behan et al. Lancet I : 593-594, 1976. Libluu et al. Neurology 42: 516-518, 1992. Smith et al. Diabetes 27: 1092-1097, 1978. ' Hoddinott et ul. Diabetologia 23: 326-329. 1982.

R$e e f al. Ann. Med. Int. 139: 134137, 1988.

brother of the proband were in good health and had normal serum Ig levels.

DISCUSSION

The prevalence of IgA deficiency is increased in patients with several autoimmune disorders (Table 3); it is about 20 times greater among patients with SLE than among normal subjects (Cussidy et al. 1968, Yewdull et al. 1983, Rifle et ul. 1988). There is strong evidence suggesting that IDDM is mediated through an autoim- mune process, and in the population that we studied, 63% of the patients tested had signifi- cant ICA. The patients were selected for very recent-onset IDDM and, therefore, the autoim- mune process was presumably still active. De- spite the relatively high incidence of IDDM, only a few cases associated with IgA deficiency have been reported (Smith rt al. 1978, Hoddinott et al. 1982). In a population of 261 adult diabetic patients, we detected only one case of such an association. Our data are in agreement with those of Smith et ul. (1978). These authors did

not find an increased prevalence of IgA de- ficiency in a population of adult IDDM patients (0 of 421 patients). This is surprising since HLA DQBl alleles encoding non-Asp residues at po- sition 57 are associated with susceptibility to both IDDM and selective IgA deficiency, giving a possible basis for an increased co-occurrence of the two diseases (Olerup et al. 1990). Never- theless, in the diabetic population we studied, the high proportion of DQBl non-Asp 57 homozygotes was clearly present. An increased frequency of IgA deficiency was found in dia- betic children (Smith et al. 1978), and this may reflect a higher frequency of non-Asp 57 HLA DQBl alleles in juvenile-onset as compared to adult-onset IDDM. Our study does not support the hypothesis of an increased prevalence of IgA deficiency among adult IDDM patients. This might be due to: 1) the small increase in IgA deficiency prevalence to be detected, which may thus require a much larger population to be screened, 2) the fact that patients affected by both disorders may not survive until adulthood, as previously suggested (Smith et al. 1978), or 3) the involvement of different environmental

71 I

LIBLAU et ul.

and genetic factors (outside or within the HLA complex) in these two conditions, since both IDDM and IgA deficiency are polyfactorial in nature (Cunningham-Rundles 1990, Todd 1990).

We greatly thank D. Shapiro for help with statistical analysis and Dr C. Lock for critical reading of the manuscript.

REFERENCES

1.

2.

3.

4.

5 .

Boitard, C., Feutren, G., Castano, L.. Debray- Sachs, M., Assan, R., Hors, J. & Bah, J-l?: Effect of cyclosporine A treatment on the production of antibody in insulin-dependent (type I) diabetic patients. J. Clin Invest. 80: 1607-1612, 1987. Burks, W A. & Steele, R. W.: Selective IgA de- ficiency. Ann. Allergy 57: 3-8, 1986. Caddy , J. T , Burt, A , , Petty, R. & Sullivan, D. : Selective IgA deficiency in connective-tissue diseases. N. Engl. J. Med. 280: 275, 1968. Cunningham-Rundles, C. : Genetic aspects of im- munoglobulin A deficiency. Adv. Human Genet. 11: 235-266, 1990. Hoddinott, S., Dornan, J. & Farid, N. R.: Im- munoglobulin levels, immunodeficiency and HLA in type 1 (insulin-dependent) diabetes melli- tus. Diabetologia 23: 326329, 1982.

6. Mach, B. & Tiercy, D. M.: Genotypic typing of HLA class 11: from the bench to the bedside. Human Immunol. 30: 278-281, 1991.

7. Olerup, O., Smith, C. I. E. & Hammarstrom, L.: Different amino acids at position 57 of the HLA- DQP chain associated with susceptibility and re- sistance to IgA deficiency. Nature 347: 289-290, 1990.

8. Rijle, G., Bielefeld, P, Chalopin, J. M., Besancen- ot, J. l?, Guiguet, M., Mousson, C., Tanter, Y , Poirier, D. & Cortet, P: Selective IgA deficiency and systemic lupus erythematosus. Ann. Med. Int. (Paris) 139: 134-137, 1988.

9. Ropars, C., Muller, A., Paint, N., Beige, D. & Avenard, G.: Large scale detection of IgA de- ficient blood donors. J. Immunol. Methods 54:

10. Smith, W I., Rabin, B. S., Huellmantel, A., Van Thiel, D. H. & Drash, A.: Immunopathology o f juvenile-onset diabetes mellitus. 1. IgA deficiency and juvenile diabetes. Diabetes 27: 1092-1097, 1978.

11. Todd, J. A . ; Genetic control of autoimmunity in type 1 diabetes. Immunol. Today 11: 122-129, 1990.

12. Yewdall, J!, Cameron, J. S., Nathan, W., Nield, G., Ogg, C. S. & Williams, D. G.: Systemic lupus erythematosus and IgA deficiency. J. Clin. Lab. Immunol. 10: 13-18, 1983.

183-189, 1982.

712