Humoral β-cell autoimmunity is rare in patients with the congenital rubella syndrome

Abstract

The congenital rubella syndrome (CRS) is associated with increased risk for diabetes and thyroid disease. However, the mechanisms by which the rubella virus may cause these diseases are poorly characterized. Previous studies were carried out before modern immunological methods were available. The present study aimed at evaluating whether autoimmune mechanisms are involved in the pathogenesis by analysing antibodies to biochemically characterized autoantigens. The incidence of clinical diabetes, thyroid disease, coeliac disease and related antibodies (islet cell antibodies, ICA; insulin autoantibodies, IAA; antibodies to the tyrosine phosphatase related IA-2 molecule, IA-2 A and glutamic acid decarboxylase, GADA; thyroid peroxidase, TPO; tissue transglutaminase, TTGA; and gliadin, AGA) and HLA risk genotypes were analysed in 37 subjects affected by or exposed to rubella during fetal life (mean age 22·5 years). One patient had diabetes and four patients had clinical hypothyroidism at the time of the examination. ICA, IAA, GADA or IA-2 A were not detected in any of the patients, while five patients tested positive for TPO antibodies. Coeliac disease or TTGA were not observed. Eight patients carried the HLA-DR3–associated HLA-DQB1*02-DQA1*05 haplotype. These results provide no evidence of an increased frequency of markers for humoral β-cell autoimmunity in patients with CRS suggesting that diabetes in CRS may be caused by other than autoimmune mechanisms.

INTRODUCTION

Rubella virus infection during pregnancy is known to spread to the fetus in the majority of seronegative mothers. If the infection occurs during the first trimester of pregnancy, there is a high risk of serious organ damage in the fetus [1]. A variety of clinical abnormalities is seen in the congenital rubella syndrome (CRS), including sensorineural deafness, mental retardation, retinopathy and heart defects. Increased incidences of diabetes mellitus and thyroid disease have also been reported in these patients. In addition, other endocrine diseases such as Addison's disease and growth hormone deficiency have been implicated [1–3].

The increased frequency of diabetes in patients with CRS has been known since the early 1970s. Clinical diabetes has been reported in 0–20%[4–13], and in one report even in 40%[14] of CRS patients after follow-up for 7–50 years. Less than 10% of the CRS patients have been described as insulin-dependent in earlier reports. Islet cell surface antibodies (ICSA) have been reported in 20–25% of CRS patients [9,15] whereas no islet cell antibodies (ICA), which are more closely related to immune-mediated diabetes, have been detected [5,8]. Insulin autoantibodies (IAA) have been analysed with ELISA in one study of non-diabetic CRS patients, in which the prevalence was reported to be as high as 31%[9]. The pathogenesis of CRS-induced diabetes has remained open and a series of mechanisms have been proposed. Immunological cross-reactivity has been documented between glutamic acid decarboxylase and the rubella virus [16]. The rubella virus can also infect human islet cells in vitro [17,18] and cause hyperglycaemia and hypoinsulinaemia in animal models [4,19].

HLA alleles play a role in the pathogenesis of autoimmune disorders and the risk for diabetes in CRS patient has been linked to the same risk alleles as in type I diabetes in general; the HLA-DR3 allele with increased risk and HLA-DR2 with decreased risk for diabetes [6]. Other HLA alleles may also modulate the risk [20,21]. Hypothyroidism, hyperthyroidism and thyroiditis have also been reported in CRS patients and the incidence of thyroid disease has been about 5%[3]. A high prevalence of thyroid microsomal (Tm) or thyroglobulin (TG) antibodies (19–34%) has also been observed among patients with CRS, suggesting that immune-mediated mechanisms may be important [5,7–9].

The present study aims at elucidating the mechanisms by which CRS may lead to diabetes, and other immune-mediated diseases. We addressed particularly the issue of whether patients with CRS have an increased frequency of markers for β-cell autoimmunity which would support the hypothesis that CRS-associated diabetes is immune-mediated similarly to classical type IA diabetes.

MATERIALS AND METHODS

Study population

Finnish patients with CRS were identified from hospital case records according to the following criteria: typical clinical signs of congenital rubella syndrome together with clinical or serologically confirmed rubella infection of the mother or serologically confirmed intrauterine infection. In addition, three subjects had intrauterine exposure to virus (confirmed maternal infection during pregnancy), but no clinical signs of CRS. To increase further the number of participants, a questionnaire was sent to senior physicians at central institutions for disabled people to identify patients with CRS. Altogether, 25 Finnish subjects (mean age 28·5 ± 6·4 years) were willing to take part in the study, 15 of whom came to an out-patient visit where a blood sample was taken. Three patients were interviewed by telephone and the blood sample was drawn in the local health care centre. Seven patients were institutionalized permanently because of severe symptoms of CRS. Polish subjects were traced after a rubella epidemic in 1985–86 in southern Poland where the mothers had serologically confirmed infection during pregnancy. The Polish subjects have been described previously in detail [22]. A blood sample was collected from 12 patients from Poland (mean age 9·9 ± 0·3 years) (Table 1). The study protocol was approved by the local ethical committees and the study was conducted according to the Declaration of Helsinki.

Table 1.

Clinical and immunological characteristics of the study subjects

					Clinical status				HLA
	Gender	Basis of diagnosis	Time of infection (trimester)	Age at sampling	CRS symptoms	Diabetes	Coeliac disease	Hypo- thyroidism	DQB1	DQA1	ICA	IAA	GADA	IA-2 A	TPO	AGA	tTG
Finland
1	M	Clinical and serological	II	25	Impaired hearing, retinopathy, strabismus	0	0	0	02, 0603	0201	0	0	0	0	0	0	0
2	F	Clinical and serological	I	19	Deafness	0	0	0	X		0	0	0	0	0	0	0
3	F	Clinical and serological	I	26	Deafness, retinopathy, growth retardation	0	0	1	0302, 0602		0	0	0	0	0	0	0
4	F	Clinical and serological	II	19	Impaired hearing, diplegia spastica	0	0	0	0301, 0302		0	0	0	0	0	0	0
5	F	Clinical	II	32	Impaired hearing, retinopathy, open Ductus arteriosus, growth retardation	0	0	0	X		0	0	0	0	0	0	0
6	M	Serological	ND	14	–	0	0	0	02	05	0	0	0	0	0	0	0
7	M	Serological	III	19	–	0	0	0	02, 603	05	0	0	0	0	0	0	0
8	F	Clinical and serological	III	30	Impaired hearing, retinopathy, palatoscisis	0	0	0	ND		0	0	0	0	0	IgA 0·31	0
9	F	Serological	II	20	–	0	0	0	02		0	0	0	0	0	0	0
10	F	Clinical and serological	I	30	Deafness	0	0	0	0602		0	0	0	0	0	0	0
11	F	Clinical and serological	I	24	Deafness, retinopathy, VSD, physical and mental retardation	0	0	1	02	0201	0	0	0	0	203	0	0
12	F	Clinical	ND	28	Impaired hearing, cataract, LPI, physical retardation	0	0	0	0604		0	0	0	0	0	0	0
13	F	Clinical and serological	I	34	Deafness, retinopathy, mental retardation	0	0	1	X		0	0	0	0	0	0	0
14	M	Clinical and serological	II	33	physical reImpaired hearing,tardation	0	0	0	02	05	0	0	0	0	0	0	0
15	M	Clinical and serological	I	36	Deafness, mental retardation, epilepsy	0	0	0	0301		0	0	0	0	0	0	0
16	F	Clinical and serological	II	34	Diplegia spastica, retinopathy, epilepsy	0	0	1	02, 0501	05	0	0	0	0	184	0	0
17	F	Clinical and serological	I	39	Impaired hearing, retinopathy, epilepsy	0	0	1	0302, 0501		0	0	0	0	63	0	0
18	M	clinical and serological	I	30	Deafness, physical retardation	0	0	0	0603, 0501		0	0	0	0	1939	IgG>100 IgA 0·45	0
19	M	Clinical and serological	I	24	Deafness, cataract, retinopathy, palatoscisis, physical and mental retardation, pulmonal stenosis	1	0	0	02	05	0	0	0	0	0	IgA 0·38	0
20	M	Clinical and serological	I	29	Impaired hearing, retinopathy, micro- phtalmia, cataract, physical and mental retardation	0	0	0	0602/03/04†		0	0	0	0	0	IgG 10·9	0
21	M	Clinical and serological	I	33	Deafness, retinopathy, open ductus arteriosus, microcephalia, palatoscisis, physical and mental retardation	0	0	0	02, 0301	03/201	0	0	0	0	0	0	0
22	F	Clinical and serological	I	34	Impaired hearing, cataract, microphtalmia, VSD, physical and mental retardation	0	0	0	0302		0	0	0	0	0	0	0
23	M	Clinical	ND	34	Impaired hearing, retinopathy, pulmonal stenosis, mental retardation	0	0	0	0603		0	0	0	0	0	0	0
24	M	Clinical	I	33	Deafness, micro- phtalmia, cataract, mental retardation	0	0	0	0301		0	0	0	0	0	0	0
25	M	Clinical and serological	ND	33	Deafness, physical and mental retardation, cataract, VSD	0	0	0	0602		0	0	0	0	0	0	0
Poland
1	F	Clinical and serological	I	10	Deafness, VSD,epilepsy	0	ND	0	0602		0	0	0	0	0	0	0
2	M	Clinical and serological	I	10	Ophtalmic defect, excitability	0	ND	0	0301		0	0	0	0	0	0	0
3	M	Serological	I	10	–	0	ND	0	X		0	0	0	ND	0	0	ND
4	M	Serological	I	10	–	0	ND	0	02, 0301	05	0	0	0	0	0	0	0
5	F	Serological	I	10	–	0	ND	0	0301, 0602		0	0	0	ND	0	0	ND
6	M	Serological	II	10	–	0	ND	0	0302, 0602		0	0	0	ND	563	0	ND
7	M	Serological	II	9	–	0	ND	0	02	05	0	0	0	0	0	0	0
8	M	Clinical and serological	II	10	Deafness	0	ND	0	0301, 0602		0	0	0	0	0	IgG 25·2	0
9	M	Serological	II	10	–	0	ND	0	0302		0	0	0	0	0	0	0
10	M	Serological	III	10	–	0	ND	0	0301		0	0	0	ND	0	0	ND
11	F	Serological	III	10	–	0	ND	0	X		0	0	0	0	0	0	0
12	M	Serological	III	10	–	0	ND	0	0301		0	0	0	0	0	0	0

ND, not determined; VSD, ventricular septal defect; LPI, lysinuric protein intolerance;

^†genotype may be either 0602/0603, 0603 or 0603/0604

Detection limits for positivity: TPO ab: 60–100 kU/l low positive, >100 kU/l positive. AGA-IgA: children 0·30–0·50 EU/ml low positive, >0·50 EU/ml positive; adults: 0·20–0·59 EU/ml low positive, >0·50 EU/ml positive. AGA-IgG: children 15–25 EU/ml low positive, >25 EU/ml positive; adults 10–20 EU/ml low positive, >20 EU/ml positive; tTG ab: >8 positive. Male: M; female: F.

Autoantibody analyses

ICA were analysed by a standard indirect immunofluorescence method, and IAA, GADA and IA-2 A by specific radiobinding assays as described previously [23]. The detection limit was 2·5 JDF-units for ICA. The cut-off limit for IAA positivity was 1·56 relative units (RU), for GADA positivity 5·36 RU and for IA-2 A positivity 0·43 RU, representing the 99th percentile in the Finnish non-diabetic background population [23].

Antibodies to the thyroiditis related thyroid peroxidase (TPO) were measured using a commercial fluoroenzyme-immunoassay kit (Pharmacia Diagnostics, Freiburg, Germany). Based on our own reference material comprising 288 subjects, a cut-off value of 100 IU/ml was used with a grey zone ranging from 60 to 100 IU/ml [24]. IgA and IgG-class antigliadin (AGA) antibodies were measured according to a standard enzyme immunoassay with crude gliadin (Sigma G3375, St Louis, MO, USA) as antigen [25]. According to our own reference material the lower limit for positivity in the IgA class antibody assay was 0·30 EU/ml in children and 0·20 EU/ml in adults. In the IgG class antibody assays the limits were 15 EU/ml in children and 10 EU/ml in adults, respectively. IgA-class tissue transglutaminase antibodies (TTGA) were measured with a commercial immunoassay (Celikey™ Pharmacia Diagnostics, Freiburg, Germany) according to the manufacturer's instructions. The cut-off value used was 8 IU/ml (grey zone 5–8 IU/ml).

Genotyping

The presence of HLA class II haplotypes associated with the risk for or protection against type I diabetes were analysed as described earlier [26,27]. Typing for the relevant HLA-DQB1 alleles (DQB1*02, DQB1*0301, DQB1*0302, DQB1*0602, DQB1*0603) was first performed and DQB1*02 positive samples further studied for the presence of DQA1*0201 (neutral or protective DR7 haplotype) or DQB1*05 (risk-associated DR3 haplotype) alleles.

RESULTS

A total of 25 patients with different symptoms of CRS and 12 subjects whose mother had had serologically confirmed rubella during the pregnancy were included in the study (Table 1). The mother had had a rubella infection during the first trimester of the pregnancy in 49% of the cases, during the second trimester in 27% and during the third trimester in 14% of the cases. In four cases (11%) the time of maternal infection was not documented in detail. The patients suffered from a variety of CRS symptoms: among the Finnish CRS patients 95% (21/22) had impaired hearing or deafness, 55% (12/22) retinopathy, 45% (10/22) mental retardation, 32% (7/22) had heart defects and 27% (6/22) had cataracts. Three Finnish subjects had no clinical signs of CRS (Table 1). Nine of the 12 Polish subjects were asymptomatic, while the other three had signs of CRS (Table 1). Two of them were deaf; one of these two also had a ventricular septal defect and epilepsy and the third suffered from ophthalmic defects, anaemia and excitability. The rubella infection in the milder cases of CRS had occurred during the second and third trimester, whereas all known infections in institutionalized subjects had occurred during the first trimester of pregnancy.

No diabetes-associated autoantibodies (ICA, IAA, GADA or IA-2 A) were detected in any of the study subjects (Table 1). Only one subject was affected by clinical diabetes at the time of the examination (a 24-year-old male whose diabetes had been diagnosed at the age of 14). He had no history of ketoacidosis, and the diagnosis was based on repeatedly increased blood glucose concentrations. His current insulin dose was 0·61 IU/kg, fasting plasma glucose 9·1 mmol/l and postprandial glucose 13·0 mmol/l. His serum C-peptide concentrations were 0·25 nmol/l before breakfast and 0·26 nmol/l after breakfast. He was negative for all autoantibodies but carried the HLA-DR3 –associated HLA DQB1*02-DQA1*05 haplotype.

TPO antibodies were detected in 13·5% (5/37) of the subjects (Table 1). The diagnosis of hypothyroidism had been made in three of the TPO-positive patients, while the remaining two had no clinical signs of thyroid disease. In addition, two patients had previously diagnosed hypothyroidism without detectable TPO antibodies at the time of the present examination. According to the hospital charts autoimmune thyroiditis had been diagnosed in one of these two subjects.

All subjects tested negative for tTG antibodies. One individual had high levels of IgG AGA, two had marginally elevated IgG AGA and three had marginally increased IgA AGA (Table 1). None of the subjects had a clinical diagnosis of celiac disease.

The HLA-DR3-associated HLA-DQB1*02-DQA1*05 haplotype was seen in seven of 36 patients with CRS (21% of the Finnish subjects and 17% of the Polish subjects). The DQB1*0302 allele indicating the presence of the other common susceptibility haplotype for type IA diabetes was observed in six subjects (17% in both Finns and Poles). None of the patients was heterozygous for both risk haplotypes for type I diabetes (HLA-DQA1*05-DQB1*02 and *0302) (Table 1).

DISCUSSION

In the present study no signs of β-cell autoimmunity were detected in the 37 subjects with CRS or exposure to rubella virus during fetal life. This is in contrast with some earlier reports, suggesting that approximately 20% of CRS patients test positive for diabetes-associated autoantibodies. However, the autoantibody assays used in the previous studies were not optimal and may thus have provided false positive findings. The predictive value of islet cell surface antibodies (ICSA) and IAA measured by ELISA, which were reported to be frequent in patients with CRS in two previous studies, was later shown to be poor [28–30]. In two other studies, analysing islet cell antibodies (ICA), all CRS patients were observed to be ICA-negative [5,8] (Table 2). In classical autoimmune diabetes, the predictive value of autoantibodies is high if multiple autoantibodies are detected, whereas positivity for a single antibody specificity is associated with only a slightly increased positive predictive value [31]. In addition, the predictive value is affected by the HLA-conferred susceptibility to type I diabetes [32].

Table 2.

Major reports related to diabetes in patients with CRS

Country of subjects	Reference	No. of CRS patients	Mean age or age range of patients	Clinical diabetes or impaired glucose tolerance*	Clinical diabetes with insulin treatment†	ICA	ICSA	IAA	Tg or Tm
Australia	Menser et al. 1967 [13]	50	25 years	2%	0%
	Forrest et al. follow-up 1971 [12]	44		20%	0%
	Menser et al. follow-up 1974 [14]	45		40%	0%
	Menser et al. follow-up 1978 [4]	45		18%	2%
	McIntosh et al. follow-up 1992 [10]	40	50 years	23%	2%
	Menser et al. 1974 [14]	87	34 years	16%	5%
	Menser et al. 1978 [4]	318		0%	3%
USA	Rubinstein et al. 1982 [6]	173	14 years	3%	9%
	Ginsberg-Fellner et al. 1985 [15]	242	17 years	6%	6%		20%		34%
	McEvoy et al. 1988 [9]	187‡	17 years	0%	0%		25%	31%	26%
	Clarke et al. 1984 [8], Shaver et al. 1985 [20]	201	17 years	21%	not reported	0%			23%
	Rabinowe et al. 1985 [5]	16	18 years	0%	6%	0%			19%
UK	Gumpel 1972 [37], Smithells et al. 1978 [11]	83	3–19 years	0%	0%
	Smithells et al. 1978 [11]	482	0–7 years	0%	0·5%
	Fine et al. 1985 [38]	605§	up to 40 years	0·3%	not reported

Tm = thyroid microsomal antibodies;Tg = thyroglobulin antibodies; ICA = islet cell antibodies; ICSA = islet cell cytotoxic or surface antibodiesi; IAA = insulin autoantibodies.

^*Proportion of all CRS patients, insulin-dependent diabetes not included;

^†proportion of all CRS patients;

^‡non-diabetic CRS population selected for study;

^§subjects exposed to rubella during pregnancy.

In our study, only one patient with CRS had insulin-dependent diabetes. He tested negative for all four diabetes-associated autoantibodies. About 60% of patients with type I diabetes have at least one autoantibody of ICA, GADA and IA–IIA detectable 10 years after the clinical diagnosis of the disease [33]. This patient had still some C-peptide secretion, although no clear C-peptide response to a meal was observed. His daily insulin requirement was also relatively low confirming that he had not completely lost his endogenous insulin secretion. About 15% of patients with type I diabetes display residual β-cell function after a disease duration of more than 10 years [34]. Accordingly, it is difficult to draw conclusions about the nature of the clinical disease on the basis of this single case of diabetes.

A higher prevalence of TPO antibodies was detected among CRS patients than in laboratory reference material (13·5%versus 6·9%), although the difference was not statistically significant [24]. The TPO-antibodies were not associated with the HLA DR3-DQB1*02-A1*05 haplotype predisposing to autoimmunity. The frequency of this autoimmunity-associated haplotype being present in 21% of the Finnish patients with CRS does not differ significantly from the prevalence observed in affected family-based artificial controls (91/622 = 15%) in a family study of type I diabetes [35]. The DR4-associated DQB1*0302 allele was detected in 17% of the subjects, which is also close to the frequency of 19% observed in the controls of the above-referenced family study. Overt coeliac disease and/or transglutaminase antibodies were not detected in any of the subjects. Five patients had marginally elevated gliadin antibodies, most probably reflecting non-specific reactivity.

Patients with CRS have been reported to have a transient immunological defect which is overcome after the first few months of life [36]. The presence of TPO antibodies in more than 10% of the subjects with CRS suggests that these patients do have a working humoral immune system.

In conclusion, our observations support the idea that diabetes may not be immune-mediated in patients with CRS. This is in accordance with the fact that a considerable proportion of previously reported CRS patients with diabetes do not require insulin injections and have only impaired glucose tolerance and thus may not represent typical type IA diabetes. The age distribution and the diagnosis criteria of diabetes have, however, differed between studies impeding the interpretation of the results.