*Departments of Pediatric Gastroenterology
†Immunology, University Medical Center Utrecht, Utrecht, The Netherlands
‡Departments of Pathology
§Pediatric Gastroenterology, VU University Medical Center, Amsterdam, The Netherlands.
Address correspondence and reprint requests to Amani Mubarak, Wilhelmina Children's Hospital, Department of Pediatric Gastroenterology, KE 01.144.3, PO Box 85090, 3508 AB, Utrecht, the Netherlands (e-mail: A.Mubarak@umcutrecht.nl).
Received 11 May, 2012
Accepted 14 October, 2012
The authors report no conflicts on interest.
Celiac disease (CD) has a strong genetic component mainly related to human leukocyte antigen (HLA) class II genes (1). In fact, earlier studies indicate that CD can only occur in the context of 2 specific HLA molecules: HLA-DQ2.5 and HLA-DQ8 (1–5). HLA-DQ is an αβ-heterodimer, of which the α- and β-chains are encoded by the HLA-DQA1 and HLA-DQB1 genes, respectively. HLA-DQ2.5 is either expressed in cis (encoded by HLA-DR3-DQA1*05:01, -DQB1*02:01) or trans configuration, encoded by HLA-DR11-DQA1*05:05, -DQB1*03:01 (HLA-DQ7.5), and HLA-DR7-DQA1*02:01, -DQB1*02:02 (HLA-DQ2.2). In the latter case, the α-chain from the HLA-DQ7.5 (DQA1*05:05) and the β-chain from the HLA-DQ2.2 (DQB1*02:02) combine together to form HLA-DQA1*05:05, -DQB1*02:02, which is molecularly highly similar to HLA-DQA1*05:01, -DQB1*02:01 and therefore also called HLA-DQ2.5. HLA-DQ8 is encoded by HLA-DR4-DQA1*03:01, -DQB1*03:02.
The extremely high percentage of patients with CD carrying either HLA-DQ2.5 or HLA-DQ8 has led to the common practice of excluding CD in patients without these HLA types (1–5); however, other reports describe patients with CD who lack both HLA-DQ2.5 and -DQ8, making the validity of this strong negative predictive value questionable (6–8). Therefore, we set out to study the distribution of HLA subtypes in pediatric patients with CD in 2 medical centers in the Netherlands.
The study consisted of 2 parts. A retrospective study was carried out in the VU University Medical Center, Amsterdam, the Netherlands, and included all biopsy-proven (Marsh III) pediatric patients with CD (n = 70; 50 girls, 20 boys) in whom HLA typing was performed between 2003 and 2011. In this cohort HLA typing had been performed without any earlier selection. To avoid skewing of the study population, all patients were diagnosed independent from the HLA typing. The average age at diagnosis in this group was 5.7 years, and the average age at the time that HLA typing was performed was 7.6 years.
A prospective study was performed in the Wilhelmina Children's Hospital, Utrecht, the Netherlands. In this part of the study, HLA typing was performed in all consecutive biopsy-proven (Marsh III) pediatric patients with CD (n = 85; 60 girls, 25 boys) in whom the diagnosis was made between December 2009 and June 2011. The average age at diagnosis was 6.2 years, whereas HLA typing was performed at an average of 6.5 years. In both the centers, the study was carried out according to the guidelines of the local medical ethical board.
Genomic DNA was isolated from ethylenediaminetetraacetic acid-anticoagulated blood. In the VU University Medical Center, polymerase chain reaction (PCR)–amplified exon 2 amplicons were generated for low- to medium-resolution HLA-DQA1 and -DQB1 genotyping in a combined, single-stranded conformation polymorphism–heteroduplex assay by a semiautomated electrophoresis and gel-staining method on the PhastSystem (Amersham Pharmacia Biotech, Uppsala, Sweden) (9).
In the samples from the Wilhelmina Children's Hospital, typing the HLA-DQA1 and -DQB1 alleles was performed using the sequence-specific oligonucleotide Primed PCR (PCR-SSO) technique using the Luminex-based OneLambda LABType SSO Class II DQA1/DQB1 typing kit, following the recommendations of the manufacturer (One Lambda Inc, Canoga Park, CA). Descriptive statistics (SPSS for Windows, version 15.0; SPSS Inc, Chicago, IL) were used to calculate the frequencies of the most common HLA types.
As expected, the most prevalent HLA-type turned out to be HLA-DQ2.5. Remarkably, apart from HLA-DQ8, HLA-DQ2.2 (HLA-DQA1*02:01, -DQB1*02:02) was also found in a substantial number of patients.
In the retrospective cohort, the majority of the patients carried HLA-DQ2.5 (n = 63, 90%), of whom 20 (28.6%) patients also had HLA-DQ2.2 and 6 (8.6%) HLA-DQ8, leaving 37 (52.9%) patients who solely had HLA-DQ2.5 (Table 1). Of these patients, 6 (8.6%) were homozygous for HLA-DQ2.5. No patient was seen with only HLA-DQ8 because the 2 (2.9%) patients with this HLA-type also carried HLA-DQ2.2. Finally, a total of 5 (7.1%) patients lacked the typical HLA-DQ2.5 and -DQ8 genotypes. Interestingly, all of them carried HLA-DQ2.2, including 3 (4.3%) patients with homozygosity for the encoding genes. The 2 (2.9%) patients with heterozygous HLA-DQ2.2 also carried either HLA-DQ6.4 (HLA-DQA1*01:02, -DQB1*06:04) or -DQ2.3 (HLA-DQA1*03:02, -DQB1*02:02).
In the prospective cohort, the distribution of HLA types among the patients with CD was virtually identical to the distribution in the retrospective cohort (Table 1). Most patients carried the typical HLA-DQ2.5 genes (n = 76, 89.4%). A total of 23 (27.1%) also had HLA-DQ2.2 and 5 (5.9%) HLA-DQ8. Of the 48 (56.5%) patients who carried only the HLA-DQ2.5 genotype, 15 (17.6%) patients were homozygous. HLA-DQ8 was present in 10 (11.8%) patients, of whom 6 (7.1%) patients also had either HLA-DQ2.5 (n = 5, 5.9%) or HLA-DQ2.2 (n = 1, 1.2%). A total of 4 (4.7%) patients only had HLA-DQ8, of whom 2 (2.4%) patients were homozygous. In addition, 4 (4.7%) patients were negative for both HLA-DQ2.5 and -DQ8. All these patients had HLA-DQ2.2 in the heterozygous form (Table 1). The other HLA-DQ subtypes in these 4 patients were HLA-DQ5.1 (HLA-DQA1*01:01, -DQB1*05:01) (twice), HLA-DQ5.3 (HLA-DQA1*01:04, -DQB1*05:03) and HLA-DQ9.3 (HLA-DQA1*03:02, -DQB1*03:03).
In the combined cohorts the most common HLA type amongst patients with CD was HLA-DQ2.5 (n = 139, 89.7%), although a significant number of these patients also carried HLA-DQ2.2 (n = 43, 27.7%) and less frequently HLA-DQ8 (n = 11, 7.1%) (Table 1). Twenty-one patients were homozygous for HLA-DQ2.5. HLA-DQ2.2 was the second most common HLA-type being present in more than one-third of the patients (n = 55, 35.5%), in most cases, however, combined with either HLA-DQ2.5 or HLA-DQ8. In 5.8% (n = 9) of the patients no HLA-DQ2.5 or HLA-DQ8 was found; however, in all those patients HLA-DQ2.2 was present, of whom 3 were homozygous (Table 1). These 9 patients were all symptomatic, had positive CD serology, and showed a good clinical and serological response to gluten elimination. One of these patients was of Jewish origin; all others were of Dutch descent.
Finally, HLA-DQ8 was the least frequent HLA genotype (n = 18, 11.6%) and most commonly present in combination with either HLA-DQ2.5 (n = 11, 7.1%) or HLA-DQ2.2 (n = 3, 1.9%). In only 4 (2.6%) patients HLA-DQ8 was present without HLA-DQ2.5 or HLA-DQ2.2 (2 homozygous and 2 heterozygous patients; Table 1).
A major proportion of the genetic predisposition to CD is derived from the HLA complex. Indeed, up to 90% of the CD patients carry the HLA-DQ2.5 heterodimer, historically called HLA-DQ2, while most of the remaining patients with CD are reported to express HLA-DQ8 (1–5). The finding that CD is virtually restricted to these HLA-heterodimers has led to the practice of considering HLA-DQ2.5 and HLA-DQ8 negative patients as being not at risk for CD.
However, in the current study, a different distribution of HLA types was seen. Although almost 90% of the patients indeed carried the HLA-DQ2.5 genotype, of the 16 (10.3%) patients lacking this HLA type only 7 (4.5%) had HLA-DQ8, while 9 (5.8%) patients were negative for both classical HLA types and would therefore be missed if relying on current practise (Table 1). Interestingly, those 9 symptomatic patients, who all had positive CD serology and quickly responded to the diet, possessed the HLA-DQ2.2 genotype. Of these 9 patients, 3 patients were homozygous and one patient also carried HLA-DQ2.3 (so homozygous DQB1*02), which is associated with a 5-fold increased risk of CD (10). The remaining 5 patients carried various other HLA types, so homozygosity for HLA-DQB1*02 cannot explain the development of CD in all patients lacking the typical HLA types.
A similar prevalence of HLA-DQ2.2 in CD patients has been described before in a European study and more recently in a retrospective American study (7,8). Both demonstrated a ∼4% prevalence of patients with CD who had HLA-DQ2.2, but no HLA-DQ2.5 or DQ8. In addition, a study performed in consecutive Spanish and Finnish patients with CD showed that HLA-DQ2.2 was present in 3.2% of the Spanish patients, but in none of the Finnish patients (6). Finally, Zubillaga et al also found a significantly increased prevalence of HLA-DQ2.2 in patients with CD (11).
Many other studies have stated that the development of CD is almost restricted to individuals with either HLA-DQ2.5 or HLA-DQ8; however in some studies HLA typing was limited to these 2 types, potentially missing patients with HLA-DQ2.2 (1–5,12–16). In addition, some typing methods do not distinguish between HLA-DQ2.5 and HLA-DQ2.2 (17,18). The fact that HLA-DQ2.2 is only rarely mentioned in previous studies might also be explained by sampling errors, especially as the HLA-DQ2.2 prevalence in the general population may vary between different countries and most studies were performed in small cohorts, which increases the risk of sampling errors. Given this possible selection bias and/or incomplete typing in previous studies, and with 4 reports, including the current one, now describing a 3.2% to 5.8% prevalence for HLA-DQ2.2 in CD, we propose that HLA-DQ2.2 should also be considered as a CD-related genotype. This will have important clinical implications, because HLA typing is an essential part of the new ESPGHAN guidelines for the diagnosis of CD (10).
In summary, the current study demonstrates that HLA-DQ2.2 is at least as frequent in patients with CD as HLA-DQ8. Consequently, to avoid missing patients with CD, the HLA-DQ2.2 genotype should be considered as one of the HLA types related to CD.
1. Sollid L, Markussen M. Evidence of a primary association of celiac disease to a particular HLA-DQ α/β heterodimer. J Exp Med 1989; 169:345–350.
2. Djilali-Saiah I, Caillat-Zucman S, Schmitz J, et al. Polymorphism of antigen processing (TAP,LMP) and HLA class II genes in celiac disease. Hum Immunol 1994; 40:8–16.
3. Spurkland A, Sollid LM, Polanco I, et al. HLA-DR and -DQ genotypes of celiac disease patients serologically typed to be non-DR3 or non-DR5/7. Hum Immunol 1992; 35:188–192.
4. Balas A, Vicario JL, Zambrano A, et al. Absolute linkage of celiac disease and dermatitis herpetiformis to HLA-DQ. Tissue Antigens 1997; 50:52–56.
5. Ploski R, Ascher H, Sollid LM. HLA genotypes and the increased incidence of coeliac disease in Sweden. Scand J Gastroenterol 1996; 31:1092–1097.
6. Polvi A, Arranz E, Fernandez-Arquero M, et al. HLA-DQ2-negative celiac disease in Finland and Spain. Hum Immunol 1998; 59:169–175.
7. Karell K, Louka AS, Moodie SJ, et al. HLA types in celiac disease patients not carrying the DQA1*05-DQB1*02 (DQ2) heterodimer: results from the European Genetics Cluster on Celiac Disease. Hum Immunol 2003; 64:469–477.
8. Harmon GS, Lebeck LK, Weidner N. Gluten-dependent enteropathy and atypical human leukocyte antigen alleles. Hum Pathol 2011; 42:1112–1116.
9. Csizmadia CG, Mearin ML, Oren A, et al. Accuracy and cost-effectiveness of a new strategy to screen for celiac disease in children with Down syndrome. J Pediatr 2000; 137:756–761.
10. Husby S, Koletzko S, Korponay-Szabó IR, et al. ESPGHAN Working Group on Coeliac Disease Diagnosis; ESPGHAN Gastroenterology CommitteeEuropean Society for Pediatric Gastroenterology, Hepatology, and Nutrition guidelines for the diagnosis of coeliac disease. J Pediatr Gastroenterol Nutr 2012; 54:136–160.
11. Zubillaga P, Vidales MC, Zubillaga I, et al. HLA-DQA1 and HLA-DQB1 genetic markers and clinical presentation in celiac disease. J Pediatr Gastroenterol Nutr 2002; 34:548–554.
12. Fasano A, Berti I, Gerarduzzi T, et al. Prevalence of celiac disease in at-risk and not-at-risk groups in the United States: a large multicenter study. Archives of Internal Medicine 2003; 163:286–292.
13. Hadithi M, von Blomberg BM, Crusius JB, et al. Accuracy of serologic tests and HLA-DQ typing for diagnosing celiac disease. Ann Intern Med 2007; 147:294–302.
14. Megiorni F, Mora B, Bonamico M, et al. HLA-DQ and risk gradient for celiac disease. Hum Immunol 2009; 70:55–59.
15. Alarida K, Harown J, Di Pierro MR, et al. HLA-DQ2 and -DQ8 genotypes in celiac and healthy Libyan children. Dig Liver Dis 2010; 42:425–427.
16. Neuhausen SL, Weizman Z, Camp NJ, et al. HLA-DQA1-DQB1 genotypes in Bedouin families with celiac disease. Hum Immunol 2002; 63:502–507.
17. Liu J, Juo SH, Holopainen P, et al. Genomewide linkage analysis of celiac disease in Finnish families. Am J Hum Genet 2002; 70:51–59.
18. Mustalahti K, Sulkanen S, Holopainen P, et al. Coeliac disease among healthy members of multiple case coeliac disease families. Scand J Gastroenterol 2002; 37:161–165.