Research Articles HLA-DRB1-DQA1-DQB1 Genotype

Home , Serotype

VIRAL IMMUNOLOGY Volume 25, Number 3, 2012 Research Articles ª Mary Ann Liebert, Inc. Pp. 187–192 DOI: 10.1089/vim.2012.0001

HLA-DRB1-DQA1-DQB1 Genotype and Frequency of Enterovirus in Longitudinal Monthly Fecal Samples from Healthy Infants

Elisabet Witsø,1 Ondrej Cinek,2 German Tapia,1 Trond Rasmussen,1 Lars C. Stene,1 and Kjersti S. Rønningen 3

Abstract

Enterovirus infections may be involved in the etiology of type 1 diabetes (T1D), which is strongly associated with certain human leukocyte antigen (HLA) class II haplotypes. Our aim was to assess whether HLA genotypes conferring varying degrees of risk for T1D were associated with enterovirus gut infections. From the general Norwegian population, 190 healthy infants at high-risk for T1D (DR4-DQ8/DR3-DQ2), and 383 infants without this genotype were identiﬁed. Non-DR4-DQ8/DR3-DQ2 genotypes were further categorized as conferring either an increased-to-moderate risk (DR4-DQ8 or DR3-DQ2), were protective (DQB1*06:02), or were neutral (all other genotypes). A total of 4626 monthly fecal samples taken between age 3 and 12 mo were tested for enterovirus RNA using real-time PCR. Enterovirus prevalence was 11.5% among high-risk children, and 12.2% in other children (adjusted odds ratio: 1.23, p = 0.12). The prevalence was 11.3% in those with increased-to-moderate risk, 13.0% in the protective group, and 12.6% in the neutral group (likelihood ratio test, 3 d.f.: p = 0.37). In conclusion, there was no statistically signiﬁcant association between HLA genotype and the occurrence of human enterovirus gut infections.

Introduction damage and diabetes after viral infection (10–12). Genes within the HLA region are candidate genes for modifying the nteroviruses are small, single-stranded, positive- likelihood or course of viral infection and disease. For in- Esense RNA viruses from the family Picornaviridae (1). stance, HLA has been associated with dengue, hepatitis C, The Enterovirus genus comprises over a hundred serotypes. HIV, and Hantaan virus infection (13). It has been speculated The primary replication of enteroviruses takes place in the that the observed association between enterovirus infection lymphoid tissues of the gut (2), and the infection route is and T1D may be explained by HLA genotype (14). With a few primarily fecal-oral. Infections are very common in infancy exceptions, studies investigating whether T1D-associated and childhood, and are spread by contacts among siblings HLA genotypes may modify susceptibility to enterovirus in- and preschool children (3). fections have been limited in size, have investigated only pa- Type 1 diabetes (T1D) is a polygenic multifactorial disease, tients (not controls), were limited to enterovirus serology, and and susceptibility is mostly conferred by particular haplo- the results have been inconsistent (15–24). types of human leukocyte antigen (HLA)-DRB1, -DQA1 and The objective of the current study was to test whether the -DQB1 (4), with contributions from certain DRB1*04 sub- frequency of enterovirus in serial fecal samples from healthy types (5,6). Despite the fact that enteroviral infections are infants differs according to HLA genotypes previously es- usually asymptomatic and self-limiting (7), they are consid- tablished to confer varying degrees of risk for T1D. ered possible environmental triggers or accelerators of islet autoimmunity (8,9). Nevertheless, the potential pathogenic Materials and Methods mechanism is not known, as is whether a speciﬁc enterovirus Study population serotype or sequence motif is responsible for any possible diabetogenic effect. Data from murine models of diabetes This study involved children participating in the Norwe- autoimmunity suggest that host genetics inﬂuence beta-cell gian prospective cohort study entitled ‘‘Environmental

1Norwegian Institute of Public Health, Oslo, Norway. 2Second Faculty of Medicine, Charles University Prague, Prague, The Czech Republic. 3Oslo University Hospital, Rikshospitalet, Department of Paediatrics, Oslo, Norway.

187 188 WITSØ ET AL.

Triggers of Type 1 Diabetes: The MIDIA study’’ (2001–2007) Detection of enterovirus RNA in fecal samples (25). MIDIA tested 46,939 newborns from the general pop- Fecal samples were collected, processed, and tested for ulation, which identified the 2.1% of Caucasians carrying enterovirus RNA using internally controlled quantitative the heterozygous HLA-DRB1*04:01-DQA1*03-DQB1*03:02/ reverse transcriptase PCR as previously described (7). DRB1*03-DQA1*05-DQB1*02 genotype (often referred to as DR4-DQ8/DR3-DQ2), which confers the highest risk for HLA typing and grouping T1D. This genotype is associated with a 5–10% risk of a T1D diagnosis by the age of 15 y in populations of European HLA genotyping for identification of high-risk T1D alleles origin. From May 2004 to June 2006 the study also recruited has been described previously (25). Briefly, DQA1 was typed a ‘‘control cohort’’ from the general population with HLA for five broad specificities (DQA1*01, *02, *03, *04, and *05), genotypes other than DR4-DQ8/DR3-DQ2, on average ap- while only alleles associated with significant T1D risk or proximately twice as many children with these other geno- protection were included for DQB1 typing (DQB1*02, *03:01, types (the ratio varied slightly during the recruitment *03:02, *03:03, *03:04, *04:02, *06:02, and *06:03). HLA- period). Children with and without the high-risk genotype DRB1*04 was typed for eight subtypes. Sequences and pro- were followed in the same manner with monthly fecal tocols for the detection of allele-specific PCR products are samples for viral analysis. The present analysis is based on available from the authors on request. 4626 fecal samples collected at age 3–12 mo from 190 children The children were categorized into four broad HLA risk with the HLA-DR4-DQ8/DR3-DQ2 genotype, and 383 chil- genotype groups as determined by the odds for developing dren without this genotype (Fig. 1). Written informed con- T1D found in previous studies (4,26–29). The high-risk ge- sent was provided by participating families. The study was notype was DR4-DQ8/DR3-DQ2. Increased-to-moderate approved by The Regional Committee for Medical Research risk genotypes were DQ8/DQ8, DQ8/X, DQ2/DQ2, and Ethics and the Norwegian Data Inspectorate. DQ2/X (X sDQB1*06:02). The protective genotype was at

FIG. 1. Flow chart illustrating the details of the study’s children and their sampling procedure (N, number of children’ n, number of fecal samples; MIDIA, Environmental Triggers of Type 1 Diabetes: The MIDIA study). HLA DRB1-DQA1-DQB1 AND ENTEROVIRUS INFECTION 189 least one DQB1*06:02 allele. All other genotypes were clas- Table 1. Characteristics of Children and Fecal sified as neutral or low risk. Furthermore, 11 children who Samples from Children with and without a carried DR4-DQ8/DR3-DQ2 with DR4 subtypes other than the HLA-DR4-DQ8/DR3-DQ2 Genotype DRB1*04:01 were excluded from the analyses for the sake of Children carrying Children without consistency. the DR4-DQ8 the DR4-DQ8 genotype genotype Statistical analysis DR3-DQ2a DR3-DQ2a Association analyses were performed using logistic re- genotype: genotype: % (n total % (n total gression models, with enterovirus infection (yes or no) as the samples), samples), response, HLA genotype (group) and other variables as Variable n = 1510 n = 3116 covariates, and a random intercept to account for potential intra-individual correlation of infections, using STATA ver- Age (months) of last fecal 12.2 (3.1–12.9) 12.1 (3.2–12.9) sion 11 (StataCorp, College Station, TX). Information on sample [median (range)] breastfeeding and other variables were collected using Female sex 52.4 (791) 53.9 (1679) structured questionnaires completed by the parents when Season sample taken the infants were 3, 6, 9, and 12 mo of age. We adjusted for January–March 23.1 (349) 28.1 (876) age, sex, calendar year, season of sample collection, the April–June 23.6 (357) 21.2 (660) July–September 27.4 (413) 21.2 (660) number of full siblings, and total breastfeedings reported October–December 26.9 (391) 29.5 (920) every month, as previously described (3). In accordance with Breastfeeding this previous analysis, we included an interaction term in the No 32.7 (492) 30.0 (935) model to account for the fact that the association between Yes 67.3 (1016) 70.0 (2181) breastfeeding and enterovirus differed by age. Number of full siblings Finally, we conducted sensitivity analyses, for which the 0 45.6 (688) 44.9 (1398) main analyses were repeated after restricting to (1) infections ‡ 1 54.4 (822) 55.1 (1718) with high quantities of enterovirus RNA (defined as higher aDRB1*04:01-DQA1*03-DQB1*03:02/DRB1*03-DQA1*05-DQB1*02. than the median quantity, 10,000 copies per lL, among positive samples), and (2) infectious episodes (counting only the first enterovirus RNA-positive sample in a series of two derline significant (OR = 1.32, 95% CI 0.97,1.80), but the or more consecutively positive samples), and (3) prolonged global test for association was not significant (LR test, 3 d.f.: infection episodes (at least two consecutive monthly samples p = 0.26; Supplementary Table S2; see online supplementary positive for enterovirus RNA). material at http://www.liebertonline.com).

Results Discussion The summary characteristics of subjects and their sam- To our knowledge, the current study involving testing of ples are shown in Table 1. There were few differences more than 4500 fecal samples from 573 children is the largest between the groups with and without the high-risk HLA- of its kind. Although there were some suggestive trends, we DR4-DQ8/DR3-DQ2 genotype for T1D, except a slight var- did not find any statistically significant association between iation in the season of the samples, which may involve HLA genotypes conferring varying degrees of risk for T1D varying ratios of children in the two groups recruited over and the frequency of gut infections with enterovirus in time. The prevalence of enterovirus infections increased healthy infants. Previously, we suggested a slightly lower linearly from ages 3 to 12 mo, and varied by season, similarly enterovirus frequency in those with the DR4-DQ8/DR3-DQ2 for children with and without the DR4-DQ8/DR3-DQ2 ge- genotype than in children without this genotype (3), al- notype (Fig. 2). though HLA genotype was not the major predictor of in- The observed prevalence of enterovirus was slightly lower terest, and therefore appropriate measures such as restriction among those with the DR4-DQ8/DR3-DQ2 genotype (11.5%) to comparable dates of birth (and hence dates of enrollment) than for the other genotypes, with the highest observed and age of follow-up in the two HLA groups of children prevalence among the T1D-protective genotype (13.0%), but were not taken. After having tested additional samples for the difference was not statistically significant (Table 2). The enterovirus, restricting to comparable dates of birth and age frequency of enterovirus RNA-positive fecal samples was of follow-up for the groups, as well as further genotyping of similar in children carrying at least one DR4-DQ8 haplotype non-DR4-DQ8/DR3-DQ2 children in the current study, it and those carrying at least one DR3-DQ2 haplotype (Table seems that our previous finding may have been at least 2). (Associations with enterovirus for the other variables partially biased (3). used in the adjusted analysis in Table 2 are shown in Sup- It has been speculated that the possible relation between plementary Table S1; see online supplementary material at enterovirus and T1D may be confounded by associations or http://www.liebertonline.com.) The above pattern was es- interactions with HLA genotype (14). In addition to general sentially consistent in sensitivity analyses restricted to either importance for enterovirus epidemiology and immunity, this high-viral-load infections, prolonged infections, or when would have a great impact on the interpretation of studies of counting infection episodes rather than positive samples enterovirus and T1D (8,9). Most previous studies in the field (Supplementary Table S2; see online supplementary material have been relatively small and limited to serologic testing at http://www.liebertonline.com). For high-viral-load infec- of enterovirus in patients with T1D (16–18). One relatively tions, the suggestive pattern was slightly stronger and bor- large study found that T1D patients who had the DR3-DQ2 190 WITSØ ET AL.

FIG. 2. Frequency (%) of enterovirus-positive fecal samples in children with and without the DR4-DQ8/DR3-DQ2 genotype, according to age ( a), and month of sample collection ( b). haplotype were less likely to have enterovirus RNA in their of patients with type 1 diabetes, children who according to blood and/or fecal samples compared to those with other serological HLA typing carried HLA-DR3 and/or HLA-DR4 DQB1 genotypes, but the association was not consistent over had higher levels of anti-enterovirus antibodies than those with ethnic groups (15). It should be noted that an observed as- HLA-DR2 (19). Unpublished results from DAISY in Colorado sociation between host genotype and enterovirus among (31) showed that children with high- or moderate-risk HLA patients with T1D could be explained either by a similar genotypes had signiﬁcantly higher IgM levels for some en- association in the general population, or an interaction be- terovirus serotypes (coxsackie B1, echo 11, and coxsackie A9), tween enterovirus and HLA in the general population such compared to other or low-risk HLA genotypes (M. Rewers, that the association between enterovirus and T1D differs by personal communication). This may consequently result in a HLA genotype (30). It is therefore necessary to assess the lower frequency of infection and/or virus quantity for children potential association between HLA genotype and enterovi- with HLA T1D-associated haplotypes. rus infection among healthy subjects (representative of the In contrast to our ﬁndings, in a recent Finnish article in- general population), not only in patients with T1D. cluding children positive for T1D-associated autoantibodies We are not aware of any previous large scale study in which (who eventually developed T1D), and matched autoantibody- the association between HLA genotype and frequency of en- negative controls, more samples were enterovirus RNA- terovirus RNA has been tested in healthy children. We are positive with the high-risk HLA-DR4-DQ8/DR3-DQ2 genotype aware of two studies of the association between HLA genotype than in children who carried moderate-risk genotypes with and frequency of anti-enterovirus antibodies, one of which is the DR4-DQ8 haplotype (23). However, studies of other co- unpublished (see below). In the Finnish DiMe study of siblings horts of healthy children with HLA genetic susceptibility

Table 2. Enterovirus Infections and HLA Genotypes with Varying Degree of Established Risk for Type 1 diabetes (n = 4626)

Adjusteda Enterovirus RNA-positive Unadjusted HLA genotypeb samples % (n samples/n total) OR (95% CI) OR (95% CI) p Value

High-risk: DR4-DQ8/DR3-DQ2 11.5 (173/1510) 1.0 (ref.) 1.0 (ref.) 0.37c Non-high-risk, grouped as: 12.2 (381/3116) 1.05 (0.80,1.37) 1.23 (0.94,1.61) 0.12d Increased-to-moderate risk 11.3 (124/1101) 0.95 (0.68,1.35) 1.15 (0.82,1.61) Neutral risk 12.6 (124/988) 1.10 (0.78,1.56) 1.35 (0.96,1.90) Protective 13.0 (133/1027) 1.10 (0.78,1.56) 1.22 (0.87,1.71) DR4-DQ8 11.3 (64/565) 1.0 (ref.) 1.0 (ref.) DR3-DQ2 11.2 (60/536) 0.96 (0.56,1.65) 0.91 (0.53,1.57) 0.74e

aOR was mutually adjusted for the covariates listed in Supplementary Table 1 previously found to predict enterovirus frequency (3), and calendar year. bHigh-risk: DR4-DQ8/DR3-DQ2 = DRB1*04:01-DQA1*03-DQB1*03:02/DRB1*03-DQA1*05-DQB1*02; Non high-risk = other DR-DQ genotypes than the high-risk genotype; Increased-to-moderate risk = DR4-DQ8 homo- or heterozygotes and DR3-DQ2 homo- or heterozygotes; Protective = carrying at least one DQB1*06:02 allele; Neutral = all other genotypes. cLikelihood-ratio (LR) test (3 d.f.): testing the model assumption of the HLA categorical variable (4 categories). dLR test (1 d.f.): testing the model assumption of the HLA high-risk/non-high-risk variable (2 categories). Even after testing for a trend over HLA genotypes assuming an increasing risk for infection with decreasing risk for type 1 diabetes, there was no significant association between HLA and enterovirus prevalence (LR test, 1 d.f.: p = 0.15). eLR test (1 d.f.). OR, odds ratio; CI, confidence interval. HLA DRB1-DQA1-DQB1 AND ENTEROVIRUS INFECTION 191 toward T1D screened for the development of islet auto- This study and the MIDIA project were funded by the antibodies have not found any consistent association be- Norwegian Organization for Health and Rehabilitation tween HLA genotypes conferring varying degrees of risk for (2008/0182), the Ministry of Health of the Czech Republic T1D and enterovirus infections as defined by serology or the (IGA MZ 11465-5), the Research Council of Norway (grants presence of enterovirus RNA in serum or fecal samples (20– 135893/330, 155300/320, 156477/730, and 166515/V50), the 22,24). It should be noted that many of these studies have Norwegian Diabetes Association, and Sigurd K. Thoresen’s been small or of moderate size compared to ours. Legacy. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of Limitations the manuscript.

Among the limitations of the current study are the lack of Author Disclosure Statement control of past immunity acquired from maternal antibodies and infections occurring before 3 mo of age, and a lack of No competing financial interests exist. characterization of the enterovirus serotype present in the positive samples, which has also not been done in other References studies. However, to be able to study enterovirus serotype 1. Pallansch MA, and Roos RP: Enteroviruses: polioviruses, separately with sufficient statistical power will require much coxsackieviruses, echoviruses, and newer enteroviruses. In: larger sample sizes due to the large number of serotypes. Fields Virology , 5th ed. (Knipe DM, Howley PM, Griffin DE, Similarly, we did not attempt to test for all possible alleles Lamb RA, Martin MA, Roizman B, and Straus SE, eds.). Lip- within the HLA complex, as most of them are not relevant to pincott Williams & Wilkins, Philadelphia, 2007, pp. 839–894. diabetes risk, but we focused our study on predefined 2. Bodian D: Emerging concept of poliomyelitis infection. Sci- groups according to the degree of predisposition for T1D. As ence 1955;122:105–108. enteroviruses are shed in feces for weeks after an infection, 3. Witsø E, Cinek O, Aldrin M, et al. : Predictors of sub-clinical we likely to detected the majority of enterovirus infections, enterovirus infections in infants: a prospective cohort study. but we cannot exclude the possibility that a different effect Int J Epidemiol 2010;39:459–468. may be seen for virus infections in the blood and those in 4. Undlien DE, Friede T, Rammensee HG, et al. : HLA-encoded other organs (i.e., different HLA genotypes may have a dif- genetic predisposition in IDDM: DR4 subtypes may be as- ferent immune response [weaker or stronger]), and therefore sociated with different degrees of protection. Diabetes in some individuals infections may be more invasive and 1997;46:143–149. symptoms more severe. 5. Van der Auwera B, Van WC, et al. : DRB1*0403 protects against IDDM in Caucasians with the high-risk heterozy- Potential explanations for our results gous DQA1*0301-DQB1*0302/DQA1*0501-DQB1*0201 genotype. Belgian Diabetes Registry. Diabetes 1995;44:527–530. Another factor that may be relevant to our results is that T- 6. Thorsby E and Rønningen KS: Particular HLA-DQ mole- cell-mediated immune responses probed by viral antigens cules play a dominant role in determining susceptibility or presented by HLA class II molecules may be weaker in very resistance to type 1 (insulin-dependent) diabetes mellitus. young children than in older children, who have developed Diabetologia 1993;36:371–377. stronger adaptive immune responses (32). While the associ- 7. Cinek O, Witsø E, Jeansson S, et al. Longitudinal observation ations of HLA with autoimmune disease have been strong of enterovirus and adenovirus in stool samples from Nor- and well acknowledged (33), there have been few studies wegian infants with the highest genetic risk of type 1 dia- finding significant evidence for an association between HLA betes. J Clin Virol 2006;35:33–40. polymorphisms and infectious diseases, possibly due to a 8. Tauriainen S, Oikarinen S, Oikarinen M, and Hyo¨ty H: En- teroviruses in the pathogenesis of type 1 diabetes. Semin lack of statistical power (34). Although the present study Immunopathol 2011;33:45–55. included the largest dataset examined to date of HLA asso- 9. Tracy S, Drescher KM, Jackson JD, Kim K, and Kono K: ciations with enterovirus RNA, we cannot exclude a weak Enteroviruses, type 1 diabetes and hygiene: a complex re- effect of HLA genotype. lationship. Rev Med Virol 2010;20:106–116. In conclusion, our study did not show any significant as- 10. Ross ME, Onodera T, Brown KS, and Notkins AL: Virus- sociation in healthy infants between HLA genotypes known induced diabetes mellitus. IV. Genetic and environmental to confer varying degrees of risk for T1D, and the frequency factors influencing the development of diabetes after infec- of gut infections with enterovirus. tion with the M variant of encephalomyocarditis virus. Diabetes 1976;25:190–197. Acknowledgments 11. Kruppenbacher JP, Mertens T, Muntefering H, and Eggers HJ: Encephalomyocarditis virus and diabetes mellitus: We thank the public health care nurses, in particular Turid studies on virus mutants in susceptible and non-susceptible Wetlesen and Lene Gustavsen, for their effort in the re- mice. J Gen Virol 1985;66(Pt 4):727–732. cruitment to the MIDIA study and follow-up of participating 12. Yoon JW, and Notkins AL: Virus-induced diabetes in mice. families. The staff at the Biobank, Norwegian Institute of Metabolism 1983;32:37–40. Public Health, helped with sample registration and initial 13. Blackwell JM, Jamieson SE, and Burgner D: HLA and in- handling. We also thank Prof. Marian Rewers, Barbara Davis fectious diseases. Clin Microbiol Rev 2009;22:370–385. Center for Childhood Diabetes, University of Colorado 14. Graves PM, Norris JM, Pallansch MA, Gerling IC, and Re- School of Medicine, for access to unpublished results from wers M: The role of enteroviral infections in the develop- DAISY. Last, but not least, we express our deepest gratitude ment of IDDM: limitations of current approaches. Diabetes to all the families for participating in this study. 1997;46:161–168. 192 WITSØ ET AL.

15. Craig ME, Howard NJ, Silink M, and Rawlinson WD: Re- age: design and early results from the MIDIA study. J Au- duced frequency of HLA DRB1*03-DQB1*02 in children toimmun 2007;29:44–51. with type 1 diabetes associated with enterovirus RNA. J 26. Pociot F and McDermott MF: Genetics of type 1 diabetes Infect Dis 2003;187:1562–1570. mellitus. Genes Immun 2002;3:235–249. 16. D’Alessio DJ: A case-control study of group B Coxsack- 27. Bjørnvold M, Undlien DE, Joner G, et al. : Joint effects of ievirus immunoglobulin M antibody prevalence and HLA- HLA, INS, PTPN22 and CTLA4 genes on the risk of type 1 DR antigens in newly diagnosed cases of insulin-dependent diabetes. Diabetologia 2008;51:589–596. diabetes mellitus. Am J Epidemiol 1992;135:1331–1338. 28. Rønningen KS, Spurkland A, Iwe T, Vartdal F, and Thorsby 17. Eberhardt MS, Wagener DK, Orchard TJ, et al. : HLA E: Distribution of HLA-DRB1, -DQA1 and -DQB1 alleles and heterogeneity of insulin-dependent diabetes mellitus at di- DQA1-DQB1 genotypes among Norwegian patients with agnosis. The Pittsburgh IDDM study. Diabetes 1985;34: insulin-dependent diabetes mellitus. Tissue Antigens 1991; 1247–1252. 37:105–111. 18. Schernthaner G, Banatvala JE, Scherbaum W, et al. : Cox- 29. Rønningen KS: Genetics in the prediction of insulin- sackie-B-virus-speciﬁc IgM responses, complement-ﬁxing dependent diabetes mellitus: from theory to practice. Ann islet-cell antibodies, HLA DR antigens, and C-peptide se- Med 1997;29:387–392. cretion in insulin-dependent diabetes mellitus. Lancet 1985; 30. Khoury MJ and Flanders WD: Nontraditional epidemiologic 2:630–632. approaches in the analysis of gene-environment interaction: 19. Sadeharju K, Knip M, Hiltunen M, A ˚ kerblom HK, and case-control studies with no controls! Am J Epidemiol Hyo¨ty H: The HLA-DR phenotype modulates the humoral 1996;144:207–213. immune response to enterovirus antigens. Diabetologia 31. Graves PM, Rotbart HA, Nix WA, et al. : Prospective study of 2003;46:1100–1105. enteroviral infections and development of beta-cell autoim- 20. Salminen K, Sadeharju K, Lo¨nnrot M, et al. : Enterovirus in- munity. Diabetes autoimmunity study in the young (DAI- fections are associated with the induction of beta-cell auto- SY). Diabetes Res Clin Pract 2003;59:51–61. immunity in a prospective birth cohort study. J Med Virol 32. Marttila J, Hyoty H, Nanto-Salonen K, Simell O, and Ilonen 2003;69:91–98. J: Epitopes recognized by CBV4 responding T cells: effect of 21. Stene LC, Oikarinen S, Hyo¨ty H, et al. : Enterovirus infection type 1 diabetes and associated HLA-DR-DQ haplotypes. and progression from islet autoimmunity to type 1 diabetes: Virology 2004;319:27–35. the Diabetes and Autoimmunity Study in the Young (DAI- 33. Steck AK, and Rewers MJ: Genetics of type 1 diabetes. Clin SY). Diabetes 2010;59:3174–3180. Chem 2011;57:176–185. 22. Fuchtenbusch M, Irnstetter A, Jager G, and Ziegler AG: No 34. Singh R, Kaul R, Kaul A, and Khan K: A comparative review evidence for an association of coxsackie virus infections of HLA associations with hepatitis B and C viral infections during pregnancy and early childhood with development of across global populations. World J Gastroenterol 2007;13: islet autoantibodies in offspring of mothers or fathers with 1770–1787. type 1 diabetes. J Autoimmun 2001;17:333–340. 23. Oikarinen S, Martiskainen M, Tauriainen S, et al. : En- terovirus RNA in blood is linked to the development of type Address correspondence to: 1 diabetes. Diabetes 2011;60:276–279. Dr. Elisabet Witsø 24. Sadeharju K, Hamalainen AM, Knip M, et al. : Enterovirus Norwegian Institute of Public Health infections as a risk factor for type I diabetes: virus analyses Postboks 4404 Nydalen in a dietary intervention trial. Clin Exp Immunol 2003; 0403 Oslo, Norway 132:271–277. 25. Stene LC, Witsø E, Torjesen PA, et al. : Islet autoantibody E-mail: [email protected] development during follow-up of high-risk children from the general Norwegian population from three months of Received January 3, 2012; accepted March 1, 2012.