HLA-DRB, -DQA, and -DQB polymorphism in celiac disease and enteropathy-associated T-cell lymphoma

ELSEVIER

HLA-DRB, -DQA, and -DQB Polymorphism in Celiac Disease and Enteropathy-Associated T-Cell Lymphoma Common Features and Additional Risk Factors for Malignancy W. Martin Howell, Shu Tong Leung, David B. Jones, Imad Nakshabendi, Margaret A. Hall, Jerry S. Lanchbury, Paul J. Ciclitira, and Dennis H. Wright ABSTRACT: CD is a gluten-sensitive enteropathy, strongly associated with expression of the DQA1*0501, DQB 1’0201 genotype. CD patients have an increased risk of malignancy, particularly EATCL. However, it is controversial as to whether adults with EATCL represent a subgroup of patients with CD or should be regarded as a distinct entity. To investigate the genetic relationship between CD and EATCL, HLA class II DRBl, DQAl, and DQB 1 typing of peripheral blood, frozen or paraffinembedded biopsy tissue obtained from Caucasian patients with CD (n = 9 1) or EATCL (n = 47) was performed by PCR-SSOP typing. Genotype frequencies were compared with those observed in 15 1 unrelated control individuals. A total of 83 (91%) of 9 1 CD patients were’ of DQA1*0501, DQB1*0201 genotype (p, < 10e6, RR = 522.2), compared with 40 (93%) of 43 EATCL patients (p, < 10e6, RR = 44.2) with amplifiable DNA versus 35 (23%) of 151 controls. DRB1*03 frequencies were also elevated in both patient groups (79 of 91 in CD [87%;p, C 10-6, RR = 24.51 and 38 of 40 in EATCL E95%;& < 10-6, RR = 70.71) compared with controls (32 of 15 1, 21%). These results confirm previous studies of HLA associations in CD and also suggest that EATCL arises in individuals with the DQA 1*050 1, DQB 1*020 1 CD-predisposing genotype. However, the frequency of

ABBREVIATIONS CD celiac disease DMSO dimethyl sulfoxide EATCL enteropathy-associated T-cell lymphoma EDTA ethylene diaminetetra-acetic acid &sodium salt

From the Mohiar lmmunoiogy Group, Tenovus Laboratory ( W. M. H.) and Univwsity Patboiogy (S. T. L., D. B. J., D. H. W. ), Soutbampton General Hospital, Southampton; the Gastroextmlogy Unit (I.N.), Gkzsgow Royal Infirmary, Glasgow; the Molecular Immunogenetics Laboratory (M.A.H., J.S.L.), Division of M ed’mne; and the Rayne Institute (P.J.C.), United Medical and Dental Schools, University of London, London, United King&. Human Immunology 43, 29-37 (1995) 0 American Society for Histocomptibility

and Immunogenetics,

1995

DRB 1*03,04 heterotygotes was significantly increased in the EATCL group (16 of 40, 40%) compared with both control individuals (3 of 151, 2%; p, < 10m6, RR = 32.9) and uncomplicated CD patients (6 of 9 1, 7 % ; p, = 0.04, RR = 9.4). Conversely, DRB1*03,03 homozygotes and DRB lx03 ,07 heterozygotes were significantly increased in frequency in the CD but not the EATCL series, compared with controls (for DRB1*03,03, p, = 0.006, RR = 8.5; forDRB1*03,07,p, = 0.06, RR = 7.9), reflecting a significant decrease in DRB1*07 frequency in the EATCL patient group (8%) as compared with the CD series (39%; p, = 0.017, RR = 0.13). In addition, none of the 38 DQA1*0501, DQB1*0201 EATCL patients were homozygous for DQB 1*0201, compared with 44 of 83 (53%) DQA1*0501, DQB1*0201 CD patients (p, < 10m6, RR < 0.02) and 12 of 32 (38%) DQA1’0501, DQB1*0201 controls (p, < 0.02, RR < 0.05). These results suggest that while EATCL arises in individuals with the DQA1’0501, DQB 1*020 1 CD-predisposing genotype, additional HLA-DR/DQ-associated alleles acting independently or in association with known CD-associated genotypes may represent a risk factor for EATCL. Human Immunology 43, 29-37 (1995)

MHC PCR SSOP

major histocompatibility complex polymerase chain reaction sequence-specific oligonocleotide probe

Address reprint requests to Dr. W. M. Howell, Molecular Immunology

Group, Tenouus Labwatory, Southampton General Hospital, Tremona Road, Southampton SO16 6YD, UK. Received(E) June 3, 1994; accepted Nouember 2, 1994.

019%8859/95/$9.50 SSDI 0198-8859(94)00130-I

30

W. M. Howell et al.

INTRODUCTION Celiac disease (CD) is an apparently immune-mediated disease of the small intestinal mucosa characterized by flattening of the small intestinal villi and nonspecific malabsorption, elicited by the ingestion of gluten, a protein found in several cereals including wheat, rye, barley, and oats. It has been known for some time that susceptibility to CD is associated with particular polymorphisms of the major histocompatibility complex (MHC) genes 11, 21 and the extensive studies performed in the 1 lth International Histocompatibility Workshop have confirmed that susceptibility to CD is primarily associated with expression of a particular HLA-DQ heterodimer (HLA-DQ[a 1*050 1, f3 lx020 l]) 131. The DQA lx050 1, DQB 1*020 1 alleles encoding this heterodimer may be located in cis or trans configuration. In northern European populations, these alleles are found in strong linkage disequilibrium with DRB lx030 1, explaining the well-known association between CD and HLA-DR3 in these populations 141. In southern European populations, HLA-DR3 is less frequent and a secondary association between DRB 1” 1101,0701 and CD is seen, reflecting the presence of the DQA1’0501, DQB l”O201 alleles in trans configuration in DRB1*1101,0701 individuals {5, 67. While the precise genetic mechanism involved in disease susceptibility is not clear, it has been shown that activated T cells derived from CD small bowel lamina propria biopsies can recognize a-gliadin (gluten)-derived peptides when presented by cells bearing the DQA 1*050 1, DQB 1*020 1 heterodimer, encoded in cis or trans, demonstrating that HLADQ molecules can effectively present antigenic peptides to T cells in the target organ {7]. Patients with CD are at a greater risk of developing malignancy than the general population. These are principally malignant lymphomas, including gastrointestinal lymphomas ES, 97. The latter are of high grade, have a T-cell genotype, and have been designated enteropathyassociated T-cell lymphoma (EATCL) [ 101. The similarity of the adjacent gut in EATCL to that in CD 1111 and the fact that there is often a preceding history of CD in these patients [12] suggests that the two conditions are related and that EATCL arises as a complication of CD. However, EATCL has certain features which suggest that at least some cases arise independently without preceding CD. The majority of EATCL cases develop in patients with adult-onset CD of short duration, and in one third of patients the evidence of enteropathy is only obtained at the time of diagnosis of the lymphoma 1131. In addition, it has been reported that patients with EATCL do not display a humoral immune response to wheat protein, rarely respond to a gluten-free diet, and are usually male, whereas patients with uncomplicated CD usually

have raised levels of a-gliadin antibody, always respond to a gluten-free diet, and are usually female { 141. Also, previous studies from our laboratory have shown that T-cell clonality may be present in the enteropathic bowel prior to the histologic appearance of overt EATCL [151. These results plus those of Alfsen et al. 1161suggest that the development of overt EATCL may be preceded by a previously unrecognized low-grade lymphomatous state and that EATCL and uncomplicated CD should be regarded as separate entities. In order to further investigate the genetic relationship between CD and EATCL and to distinguish between these conflicting models of the origins of malignant enteropathy, we studied HLA class II DRB 1, DQAl, and DQB 1 polymorphism in patients with uncomplicated CD and EATCL. All HLA typing was performed by nonradioactive polymerase chain reaction (PCR)-based methods using peripheral blood, frozen or paraffinembedded biopsy tissue for analysis, as previous studies from our laboratory have demonstrated that sufficient high-molecular-weight DNA may be obtained from paraffin biopsy sections for PCR amplification and sequence-specific oligonucleotide probe (SSOP) typing f 171. HLA class II genotypic frequencies so obtained were compared with those previously observed in a panel of healthy, unrelated Caucasoid individuals {l&l.

PATIENTS,

MATERIALS,

AND

METHODS

CD Patients Frozen biopsy material was obtained from a series of 17 CD patients (14 female and 3 male; age range, 15-63 years; mean age, 46.5 years). The CD patients were subgrouped into early onset (6 patients: 5 female and 1 male; disease onset age range, 4-24 years; mean onset age, 12.2 years,) and adult onset (9 patients: 7 female and 2 male; disease onset age, 34-59 years; mean onset age, 43.9 years). Two of the 17 CD patients could not be subgrouped in this way because the age of disease onset was not known. In addition, 74 unrelated UK Caucasoid CD patients HLA-DRB, -DQA, and -DQB typed in the 1 lth International Histocompatibility Workshop [3] were also included in the study. EATCL

Patients

Paraffin-embedded biopsy specimens from a series of 47 histologically confirmed EATCL patients were obtained from our own archival material (16 cases), with additional patient biopsy specimens kindly supplied by the Departments of Pathology at Glasgow Royal Infirmary (5 cases), University College and Middlesex Hospitals London (13 cases), University of Bristol (7 cases), and

31

HLA-DR and -DQ in CD and EATCL patients

the University College Hospital, Galway, Ireland (6 cases). In each case, the diagnosis of CD was confirmed by histologic examination with a documented histologic response to gluten withdrawal. The diagnosis of EATCL was based on histologic examination confirmed by immunostaining and demonstration of T-cell clonality using molecular techniques. All patients were of Caucasoid ethnic origin. Controls A total of 15 1 unrelated British Caucasoid bone marrow and cadaveric renal donors from the south of England previously DRB, DQA, and DQB typed in the laboratory [lS} were included as controls for the HLA class II CD/EATCL association studies. DNA

Extraction

Frozen Biopsies. A small piece of biopsy tissue was crushed and ground to a powder in liquid nitrogen. Following a wash in 0.1% NP40, the ground biopsy material was incubated for 3 hours at 55°C in lysis mix (200 mM Tris-HCl pH 8, 20 mM ethylene diaminetetra-acetic acid disodium salt [EDTA), 200 mM NaCl, 78 mM DTT, 0.04% SDS) containing 40 pg/ml proteinase K. Extracted DNA was purified by standard phenol-chloroform procedures. Paraffin biopsies. Two 20+m paraffin sections were dewaxed in xylene and (1: 1) xylene-ethanol washes and vortexed for 20 seconds. This procedure was repeated. The resulting cellular material was pelleted by spinning at 3000 g for 5 minutes, washed in absolute ethanol and pelleted twice, and dried at 50°C for 10-20 minutes. The dried pellet was then incubated at 55°C overnight in 100-300 ~1 of lysis mix, depending on the size of the pellet (100 mM Tris-HCl pH 8, 4 mM EDTA, 0.45% NP40, 0.45% Tween 20) containing 400 p.g/ml proteinase K (added fresh). After the incubation, the mixture was boiled for 5 minutes followed by phenol-chloroform extraction (if required). DNA concentration and purity was determined by UV spectrophotometry. PCR amplijcation and SSOP typing. HLA-DRB, -DQA, and -DQB genomic typing was achieved by PCR amplification of the second (variable) exon of the class II gene in question using PCR primers derived from the 11th International Histocompatibility Workshop {19}. DNA, 1 pg, was used per 100 PJ DRBl, DQBl, and DQAl PCR reaction mix, along with 1 unit of Taq polymerase (Boehringer Mannheim), 5% dimethyl sulfoxide (DMSO), and 50 pm01 of each primer. The reaction mix was denatured at 94°C for 30 seconds, annealed at 52’C for 30 seconds for 20 cycles, followed by denaturing at 94’C for 30 seconds, annealing at 50°C for 1 minute 30

seconds, and elongating at 72’C for 1 minute, for a further 20 cycles. The reaction mix was initially denatured for 4 minutes at 94°C before entering the first 20 cycles and was finally elongated for 7 minutes at 72°C after the 40 cycles of amplification. Following dot blot preparation, SSOP probing was performed using panels of probes selected from the 1 lth Workshop panel: DRB, 18 probes; DQA, 10 probes; DQB, 16 probes. All probes were 3’ end-labeled with digoxigenin (dig- 11-ddUTP). Dot blot membrane Nevinny-Stickel et al (201 blocking, prehybridization, hybridization, and stringency washing were performed according to the protocol of Spenser et al. 1111. Detection of bound probe was achieved using the chemiluminescent substrate Lumigen-PPD (Boehringer Mannheim) according to the manufacturers’ protocol and visualized following 10-40 minute exposure to x-ray film. PCRSSOP typing of paraffin biopsy DNA was performed in triplicate, as yields of intact DNA were generally low, resulting in low yields of PCR product; thus, weak signal strengths for some probes were often observed. Statistical analysis. HLA-DRB 1, -DQAl, and -DQB 1 genotype frequencies were scored in patients and control groups. Frequencies were compared among the CD, EATCL, and control groups and significance evaluated using the chi-squared test or the Fisher’s exact test if any value in a 2 X 2 table was less than 5. All probability values were corrected for the number of comparisons made (30 DRB 1, DQA 1, DQB 1 markers in three groups [CD, EATCL patients, and controls); i.e., 90 comparisons) according to the protocol of Svejgaard and Ryder 12 11. However, when comparing the frequencies of heterozygous individuals in the total patient and control series (i.e., results in Table 4), probability values were conservatively corrected for the number of combinations of alleles at each locus detected in the three series (i.e., 172 combinations in three groups = 5 16 comparisons). These analyses were performed using the University of Southampton Faculty of Medicine MEDSTAT package.

RESULTS Due to the degraded nature of the DNA extracted from paraffin-embedded biopsy specimens, in seven biopsies DRB 1 genotypes could not be assigned, while DQAl/ DQBl genotypes couId not be assigned in four cases. DRB 1, DQA 1, and DQB 1 genotype frequencies for all remaining individuals are shown in Tables l-3, respectively. Examples of DRB 1, DQAl, and DQB 1 dot blot results obtained from DNA derived from two paraffin biopsy specimens and probed with DRBl (four probes), DQB 1 (five probes), and DQA 1 (five probes) are shown in Fig. 1.

32

TABLE

W. M. Howell et al.

1

HLA-DRBl

genotype frequencies in CD and EATCL patients and in controls

DREIl*

CD patients (?I = 91) n (%)

EATCL patients (n = 40) ?I (%)

Controls (n = 151) n (%)

PCI

01 15 16 03 04 11 12 13 14 07 08 09 10

3 (3.3) 18 (19.8) 0 (0.0) 79 (86.8) 14 (15.4) 8 (8.8) 0 (0.0) 5 (5.5) l(l.1) 35 (38.5) 0 (0.0) 1 (1.1) 0 (0.0)

l(2.5) 9 (22.5) 0 (0.0) 38 (95.0) 18 (45.0) 0 (0.0) 0 (0.0) l(2.5) 0 (0.0) 3 (7.5) l(2.5) 0 (0.0) 0 (0.0)

34 (22.5) 54 (35.8) 3 (2.0) 32 (21.2) 49 (32.5) 23 (15.2) 4 (2.7) 28 (18.5) 5 (3.3) 38 (25.2) 3 (2.0) 2 (1.3) 1 (0.7)

0.0013 NS NS <1o-6 NS NS NS NS NS NS NS NS NS

PCII NS NS NS
PC111 NS NS NS NS NS NS NS NS NS 0.017 NS NS NS

In Tables l-5: p, I = p, CD vs controls; p, II = p, EATCL vs controls; p, III = p, CD vs EATCL; and NS, not significant.

DRBl

genotypic frequencies. There was a significant increase in DRB 1*03 frequency in both the CD (86.8%;~~ < lo-$ RR = 24.5) and EATCL (95%; p, < 10e6, RR = 70.7) patient groups compared with the control series (2 1.2%). There was also a decrease in DRB l*O 1 frequency in both CD (3.3%) and EATCL (2.5%) patients compared with controls (19.8%), and DRBl” 15 was also decreased in both groups (19.8% in CD and 22.5% in EATCL compared with 35.8% in controls). Of these comparisons, however, only the decrease in DRB lx0 1 in the CD patients achieved statistical significance (p, = 0.013, RR = 0.12). Some differences in DRBl frequencies between the CD and EATCL groups were also observed. Most notably, DRB1’07 showed a significant decrease in frequency in the EATCL series (7.5%) compared with the CD group (38.5%) (p, = 0.017, RR = 0.13) and a nonsignificant decrease compared with the control series (25.2%). Conversely, there was a nonsignificant increase in DRB1*04 frequency in the EATCL group (45.0%) compared with the CD series (15.4%), whereas DRB 1” 11 was absent in the EATCL

TABLE 2 DQAl* 0101 0102 0103 0201 030 112 0401 0501 0601

HLA-DQAl

(5.5) (20.9) (3.3) (40.7) (13.2) (0.0) (92.3) (0.0)

see Table 1 for abbreviations.

DQAl genotypic fmqaencies. Both CD and EATCL patient groups showed highly significant increases in the inci-

genotype frequencies in CD and EATCL patients and in controls

CD patients (n = 91) n (%) 5 19 3 37 12 0 84 0

series compared with an incidence of 8.8% in the CD group and 15.2% in the control series (p, = NS). An analysis of the frequency of DRB 1*03,X heterozygotes (where X is the second DRBl allele; Table 4) indicated that most of the increased incidence in DRB1*04 in EATCL patients may be attributed to a significant increase in the frequency of DRB 1*03,04 heterozygotes (40.0%) compared with both CD patients (6.6%_;g, = 0.04, RR = 9.4) and controls (2.O%;p, C 10 , RR = 32.9). This analysis also revealed a significantly increased frequency of DRB 1”03,03 homozygotes (18 of 91, 19.8%) and DRB1*03,07 heterozygotes (28 of 91, 30.8%) in CD but not EATCL patients as compared with the control series (in controls, DRB1*03,03 = 4 of 151, 2.8%;~~ = 0.06, RR = 8.5;DRBlX03,07 = 8 of 151, 5.3%;~~ = 0.06, RR = 7.9). The incidence of all possible second DRBl alleles in DRB1*03 CD and EATCL patients and controls is discussed below.

EATCL patients (n = 43) n (%) 7 10 2 11 7 0 43 2

(16.3) (23.3) (4.7) (25.6) (16.3) (0.0) (100.0) (4.7)

(n =

Controls 151) n (%)

41 65 21 36 46 2 55 0

(27.2) (43.1) (13.9) (23.8) (30.5) (1.3) (36.4) (0.0)

PC1 0.02 NS NS NS NS NS
PC11 NS NS NS NS NS NS
PC111 NS NS NS NS NS NS NS NS

33

HLA-DR and -DQ in CD and EATCL patients

TABLE

3

HLA-DQBl

genotype

frequencies

in CD and EATCL patients

CD patients (n = 91) n (%;)

EATCL patients

DQBl* 0201 0301 0302 0303 0501/2 0503 0602 0603 0604

88 (96.7) 9 (9.9) 11 (12.1) 3 (3.3) 4 (4.4) 1 (1.1) 17 (18.7) 3 (3.3) 2 (2.2)

40 (93.0) 10 (23.3)

(n = 43) ?J (%j)

14 (32.6) 7 (16.3) 4 (9.3) 2 (4.7) 4 (9.3) l(2.3) 2 (4.7)

Controls (?I = 151) n (%)

65 (43.1) 45 (29.8) 50 (33.1) 8 (5.3) 33 (21.9) 4 (2.7) 56 (37.1) 18 (11.9) 8 (5.3)

and in controls PCI

PCII

PC111

c1o-6 NS NS NS 0.008 NS NS NS NS

<:10-6 NS NS NS NS NS NS NS NS

NS NS NS NS NS NS NS NS NS

See Table 1 for abbreviations.

dence of DQA lw050 1 compared with controls (92.3% in CD patients versus 36.4% in controls Cp, < 10-6, RR = 20.91 and 100% in EATCL patients Cp, < 10e6, RR > 75.1). Conversely, both patient groups showed nonsignificant decreases in the frequencies of DQA l*O 102, 0103, and 0301/2 compared with the control series. However, the CD patient group showed a nonsignificant increase in DQAl”O20 1 frequency (4 1.2%) compared with EATCL patients (25.6%) and controls (23.8%), while the CD group also showed a decreased frequency of DQA IWO10 1, which is significant when compared with the control series (5.5% vs 27.2%;~~ = 0.02, RR = 0.16).

DQB 1 genotypicfreqtiencies. Both CD and EATCL patient groups showed highly significant increases in the incidence of DQB 1#020 1 compared with controls (96.7% in CD patients vs 43.1% in controls Cp, < 10W6, RR = 38.8) and 93.0% in EATCL patients (p, < 10e6, RR = 17.71). There was also a trend toward a decreased incidence of DQBl*O501/02 and 0602 in both CD and EATCL patients compared with controls, with the decrease in DQB 1*0501/2 being significant in the CD group (p, = 0.008, RR = 0.04). However, when comparing the CD and EATCL patient groups there was a nonsignificant increased incidence of DQB 1*030 1, 0302, and DQB1*0303 in the EATCL group. Taken together, the DQA l/DQB 1 results indicate a highly significant increase in the DQA1*0501/ DQB 1*020 1 heterodimer combination in both CD and EATCL patients compared with controls (83 of 91 {91.2%] in CD Fp, < 10A6, RR = 522.2) and 40 of 43 [93.0%} in EATCL Ip, < 10W6, RR = 44.2) compared with 35 of 151 [23.2%} in controls). Results also indicate an increased incidence of DRBl*03,04 heterozygotes among the EATCL patients and a trend toward an increase in DQB1*0301, 0302, and 0303 frequency in these EATCL patients.

Incidence of DRBl , DQAl , and DQBl alIeles in patients and controls carvying at least one DRBlX03 allele or DQAl*O501 lDQBl*O201 allele combination. In an attempt to identify further features of the HLA class II genes that might distinguish between uncomplicated CD and EATCL, we next investigated the incidence of the second DRB 1, DQAl, or DQB 1 allele in patients and controls who carried at least one copy of the DRB lx03 or DQA lx050 l/DQB TX020 1 CD-associated haplotypes. A summary of the major significant and nonsignificant trends revealed by this analysis is given in Table 5. This approach showed a significant increase in the incidence of DRB 1’04 among DRB 1*03,X individuals when comparing EATCL with CD patients (16 of 38 [42.1%} vs 6 of 79 {7.6%&p, = 0.01, RR = 8.9). A nonsignificant increase in the incidence of DRB l”O4 was also observed when comparing the EATCL patients with controls (16 of 38 f42.1%] vs 3 of 33 f9.1%1), reflecting the overall increase in DRB 1*03,04 heterozygotes among the EATCL series. No other significant differences in DRB 1 “second allele” frequencies were observed between the EATCL patients, CD patients, and controls. A consideration of the frequency of DQAl “second alleles” in DQAl*050l/DQB1*0201 CD and EATCL patients, plus controls, revealed no significant difference between the CD and EATCL patients. However, in agreement with results from the overall comparison of DQAl genotype frequencies in CD and EATCL patients (Table 2), there was a nonsignificant increase in frequency of DQAl*020 1 as the second DQAl allele in 83 DQAl*050l/DQB 1+020 1 CD patients, compared with 38 DQA1*0501/DQB1*0201 EATCL patients (41.0% vs 26.3%). It is also noteworthy that no EATCL patients were homozygous for DQAl”0501, compared with 24.1% CD patients and 25.0% controls. In the case of DQB 1 “second alleles” in DQA 1*050 l/DQB 1’020 1 individuals, no EATCL patients were homotygous for DQB lx020 1. This absence of DQB 1*020 1 homotygosity in DQBl*OSOl//DQB 1*0201 EATCL patients was

34

W. M. Howell et al.

Early-onset us late-onsetCD. DRB 1, DQAl, and DQB 1 genotype frequencies were also compared in a subset of 15 CD patients subdivided according to age of onset of the disease (see patient biopsy section of Patients, Materials, and MetboaS). Although the numbers concerned are small (six early onset and nine late onset; age of onset data unavailable for the remaining 76 patients), it is of note that 3 of 6 early-onset CD patients were DQB1*0201 homozygous, compared with 0 of 9 lateonset patients. Conversely, 4 of 9 late-onset CD patients typed as DRBl”04 compared to 0 of 6 early-onset CD patients.

DISCUSSION

FIGURE 1 HLA-DRB 1, -DQAl, and -DQB 1 SSOP dot blot hybridizations of amplified DNA derived from two paraffin-embedded EATCL biopsy specimens. Deduced HLA class II types: biopsy A, DRBl*0301,1302; DQA1*0501,0102; DQB1*0201,0604; and biopsy 8, DRBlr0301,1401; DQA1*0501,0101; DQBlf0201,05031. DRBl SSOPs: 7004 detects DRB1*0301 in A and B; 3703 detects DRBl*0301 and 1302 in A and 0301 in B; 5704 detects DRB1*1401 in B; and 7007 detects DRB1*1302 in A. DQBl SSOPs: 5705 detects DQB lx0201 in A and B; 2604 and 5701 detect DQBl*O604 in A; and 2602 and 5703 detect DQBl*0503 in B. DQAl SSOPs: 3402 detects DQA1*0501 in A and B and 0102 in A; 5504 and 6903 detect DQA1’0501 in A and B; 5501 detects DQA1*0102 in A and 0101 in B; and 3401 detects DQAl*OlOl in B.

highly significant when compared with CD patients (53.0%;~~ < 10V6, RR < 0.02) and is also significant when compared with control individuals (37.5%; p, < 0.02, RR C 0.05) (Table 5). Conversely, there were nonsignificant increases in the frequency of DQB1*0301/0302 and 0303 as the second DQBl allele in the EATCL group compared with both controls and CD patients (Table 5).

This study investigated the genetic relationship between CD and EATCL by comparing HLA class II DRBl, DQAl, and DQB 1 genotypes in these two patient groups, plus healthy controls, by PCR-SSOP typing of DNA extracted from peripheral blood and both frozen and paraffin-embedded biopsy materials. Results indicate that DNA of suitable quality and yield for PCRbased HLA typing can be derived from most (but not all) routinely fixed, paraffin-embedded pathology specimens, confirming earlier results from our own and another laboratory 117, 227. Use of such archival biopsy material is of considerable advantage in retrospective analysis of HLA-disease associations. Significant increases in DRBl”03, DQAl”0501, and DQB1*0201 frequencies were observed in the CD patient series, in agreement with previous studies of HLA class II genes and CD in the British and other northern European Caucasoid populations 13, 23, 241. The EATCL series also showed these significant increases in DRB 1*03, DQA 1#050 1, DQB 1*020 1 genotype frequencies compared with healthy controls. Significant decreases in DRBl*Ol, DQAl*OlOl, and DQB1*0501/2 phenotype frequencies were also demonstrated in the CD series and observed as nonsignificant trends in the EATCL series. These data suggest that EATCL arises against the same genetic background as that predisposing to CD. This is the first study to investigate HLA class II DR and DQ polymorphism in a large series of EATCL patients by using DNA-based typing. Nevertheless, these results confirm earlier studies of small series of lymphoma patients using serologic HLA-DR typing, which revealed similar HLA-DR types in patients with CD and lymphoma 112, 251. In the present study, however, some differences between the two patient groups were also observed. First, the frequency of DRB1*03,03 homozygotes and DRB 1*03,07 heterozygotes was significantly increased in the CD but not the EATCL series, compared with control individuals. The latter result reflects a decreased incidence of DRB1*07 in the EATCL series,

35

HLA-DR and -DQ in CD and EATCL patients

TABLE

4

HLA-DRB1*03,

X genotype

CD patients (n = 91) n (%)

DRB l+

03,03

18 18 6 28

03,15 03,04 03,07 See Table 1

(19.8) (19.8) (6.6) (30.8)

frequencies

in CD and EATCL patients

EATCL patients (n = 40) n (%) 0 9 16 3

(0.0) (22.5) (40.0) (7.5)

5

Second allele DRB 1’04 DQAl*020 1 DQAl*OSOl DQBl*0201 DQBl*0301 DQBl”0302 DQB I*0303 See Table

Controls (n = 151) n (%6) 4 5 3 8

(2.6) (3.3) (2.0) (5.3)

PC I

PCII

0.006 NS NS 0.006

NS NS <1o-6 NS

PC111 NS NS 0.04 NS

for abbreviations.

which is significant when compared with the CD group and nonsignificant when compared with the control series. Secondly and more notably, DRB1*03,04 heterozygosity was significantly increased in the EATCL series compared both with CD patients and controls, reflecting an overall nonsignificant trend toward an increase of DRB1*04 in the EATCL patients. Thus, DRBlX04 (or a particular subtype of DRB1*04) is a potential marker which may distinguish those CD patients at risk of developing EATCL from uncomplicated CD. This is of note as in certain population groups (e.g., Ashkenazi Jews from Israel), the majority of CD patients not carrying the DQAl”050 l/DQB 1*020 1 alleles are positive for DRB*0402 [5]. Furthermore, DR4-associated haplotypes have also been reported in non-DR3, non-DR5/7 Spanish CD patients [ZGf. In an attempt to further investigate possible genetic differences between CD and EATCL and based on the assumption that the DQAl”050 l/DQB 1*020 1 heterodimer predisposes to both CD and EATCL as suggested by this study, the incidence of particular DRB 1, DQA 1, and DQBl alleles in patients carrying at least one DRBl”03 allele or DQA1*0501/DQB 1*0201 allele combination was then investigated. This analysis confirmed the trend toward an increased frequency of DRB 1”03,04 heterotygotes in EATCL as compared with CD, but no other significant differences in frequency of the “second” DRBl allele were seen. In addition, no significant differences in DQA 1 second allele frequencies TABLE

and in controls

were seen between CD and EATCL patients, although no EATCL patients were homozygous for DQA1’0501. More notably, no DQB l”O201 homozygotes were seen in the EATCL series. This absence of DQB 1*0201 homozygosity in the EATCL group was highly significant when compared with CD patients and was also significant when compared with the control group. It has been suggested that in individuals carrying the DQA 1*0501/ DQB 1+0201 a11e 1es, the presence of a second copy of DQB 1*0201 increases susceptibility to CD 1271, or in combination with DQA l”O50 1 is associated with earlier onset of disease C281, although this is not seen in all studies {29). In the current study, the absence of DQB 1*020 1 homozygosity in EATCL patients may also suggest that DQB 1*020 1 homozygosity in CD protects against the development of overt EATCL. Finally, in 15 CD patients for which age-of-onset data were available, a nonsignificant trend toward decreased DQB 1*020 1 homozygosity and increased incidence of DRB 1*04 in late as opposed to early age of onset CD was observed, which mirrors the findings seen in EATCL patients compared with the whole group of uncomplicated CD patients. Taken together, it can be hypothesized that if DQB1*0201 homozygosity (and/or other HLA class-II-associated factors such as DRB 1*03/07 heterozygosity) leads to more severe symptoms in CD, this may lead to the detection and diagnosis of CD at an earlier age than in non-DQB 1*020 1 homozygous individuals. The latter individuals may have subclinical

Frequency of second DRB 1, DQAl, and DQB 1 alleles in CD and EATCL patients and in controls carrying at least one copy of DRBl”03 or DQA1*/05Ol/DQB 1*0201 (summary of significant differences and nonsignificant trends) CD patients 6179 34183 20/83 44/83 6183 7/83 l/83

1 for abbreviations.

(7.6%) (41.0%) (24.1%) (53.0%) (7.2%) (8.4%) (1.2%)

EATCL patients

Controls

16/38 (42.1%) 10138 (26.3%) O/38 (0.0%) O/38 (0.0%) 8/38 (21.0%) 12138 (31.6%) 6138 (15.8%)

3/33 (9.1%) 7/32 (21.9%) 8132 (25 .O%) 12132 (37.5%) 3132 (9.4%) 3132 (9.4%) O/32 (0.0%)

NS NS NS NS NS NS NS

PCI1

PC111

NS NS NS -=0.02 NS NS NS

0.01 NS NS
36

W. M. Howell et al.

symptoms which, if left undetected and untreated, could result in an increased likelihood of developing lymphoma. Results from this study therefore suggest that while uncomplicated CD and EATCL arise in individuals with a similar genetic background (DRBl”03, DQA1*0501/ DQB1*0201 associated), certain HLA class II polymorphisms (namely, DRB 1*07 and DQBl”0201 homozygosity) may protect against the development of EATCL, whereas others (DRB 1*03,04 heterozygosity) may predispose to EATCL. In addition, the HLA class II genotypes of late-onset CD patients are more closely related to those found in EATCL patients, which may be significant, as most cases of EATCL arise in CD patients with late-onset disease. However, results clearly indicate that CD and EATCL are closely related genetically. Other studies from our laboratory using PCR-based techniques to investigate the nature of the T-cell populations in the gut of uncomplicated CD and EATCL employing molecular analysis of the T-cell. receptor y- and P-gene rearrangements support this view 1301. We conclude that while EATCL arises in individuals with the DQA lx050 1 ,DQB lx020 1 CD-predisposing genotype, additional HLA-DWDQ-associated alleles acting independently or in association with known CDassociated genotypes, may represent a risk factor for EATCL. Further molecular typing, including DRB 1*04 subtyping, of extended series of EATCL and CD patients is required to identify these polymorphisms, and may aid identification of those individuals at risk of developing lymphoma.

ACKNOWLEDGMENTS

This work was supported by a grant from the United Kingdon Coeliac Trust. Thanks are due to Dr. N. Rooney (Department of Pathology, University of Bristol), Dr. A. Wotherspoon (Department of Pathology, University College and Middlesex Hospitals School of Medicine), Dr. A. Krajewski (Department of Pathology, University of Edinburgh), and Dr. E. Connolly (Department of Pathology, University College Hospital, Galway) for supplying Patient biopsy material. Thanks are also due to Bridget Murrey and Eileen Andrews for careful preparation of the manuscript.

REFERENCES Stokes P, Asquith P, Holmes G, Mackintosh P, Cooke W: Inheritance and influence of the histocompatibility (HL-A) antigens in adult coeliac disease. Gut 14:627, 1973. Tiwari JL, Terasaki PI: HLA and Disease Associations. New York, Springer-Verlag, 1985. Hall MA, Mazzilli MC, Satz ML, Barboni F, Bartova A, Brunnler G, Ciclitira PJ, Corazza GR, Ferrante P, Gerok

W, Herrera M, Kollek A, Keler E, Lanchbury JS, Mantovani V, Muser K, Petronzelli F, Roshmann E, Theiler G, Volk BA, Welsh KI, Wienker T, Albert E: Coeliac disease study. In Tsuji K, Aizawa M, Sasazaki T (eds): HLA 1991, vol 1. Oxford, Oxford University Press, 1992. 4. Sollid LM, Markussen G, Ek J, Gjerde H, Vartdal F, Thornsby E: Evidence for a primary association of celiac disease to a particular HLA-DQ d/p heterodimer. J Exp Med 169:345, 1989. 5. Tighe MR, Hall MA, Barbado M, Cardi E, Welsh KI, Ciclitira PJ: HLA class II alleles associated with celiac disease susceptibility in a southern European population. Tissue Antigens 40:92, 1992. 6. Lundin K, Sollid H, Ek J, Thorsby disease associated beta heterodimer.

L, Quigstad E, Markussen G, Gjertsen E: T lymphocyte recognition of a celiac cir- or tnzns-encoded HLA-DQ alpha/ J Immunol 145:136, 1990.

7. Lundin KEA, Scott H, Hansen T, Paulsen G, Halstenson TS, Fausa 0, Thorsby E, Sollid LM: Gliadin-specific HLA-DQ (al*0501,l3 1*0201) restricted T cells isolated from the small intestinal mucosa of celiac disease patients. J Exp Med 178:187, 1993. 8. Harris OD, Cooke WT, Thompson H, Waterhouse JAH: Malignancy in adult coeliac disease and idiopathic steatorrhea. Am J Med 42:899, 1967. 9. Holmes GKT, Stokes PL, Sorahan TM, Prior P, Waterhouse JAH, Cooke WT: Coeliac disease, gluten-free diet and malignancy. Gut 17:612, 1976. 10. Isaacson PG, O’Connor NTJ , Spencer J , Bevan DH, Connolly CE, Kirkham N, Pollock DJ, Wainscoat JS, Stein H, Mason DY: Malignant histiocytosis of the intestine: a T cell lymphoma. Lancet 2:688, 1985. 11. Spencer J, MacDonald TT, Diss TC, Walker-Smith JA, Ciclitira PJ, Isaacson PG: Changes in intraepitheliod subpopulations in coeliac disease and enteropathy associated T cell lymphoma (malignant histiocytosis of the intestine). Gut 30:339, 1989. 12. Swinson CM, Slavin G, Coles EC, Booth C: Coeliac disease and malignancy. Lancet 1: 111, 1983. 13. Cooper BT, Read AE: Coeliac disease and malignancy. J Med 63:269, 1987.

Q

14. O’Farrelly C, Feighery C, O’Brian DS, Connolly CE, McCarthy C, Weir DG: Humoral response to wheat proteins in patients with coeliac disease and enteropathy associated T cell lymphoma. Br Med J 293:908, 1986. 15. Wright DH, Jones DB, Clark H, Mead GM, Hodges E, Howell WM: Is adult-onset coeliac disease due to a lowgrade lymphoma of intraepithelial T lymphocytes? Lancet 1:1373, 1991. 16. Alfsen GC, Beiske K, Bell H, Marton PF: Low-grade intestinal lymphoma of intraepithelial T lymphocytes with concomitant enteropathy-associated T cell lymphoma: case report suggesting a possible histogenetic relationship. Hum Path01 20:909, 1989. 17. Bateman AC, Leung ST, Howell WM, Roche WR, Jones

HLA-DR and -DQ in CD and EATCL patients

37

DB, Theaker JM: Detection of specimen contamination in routine histopathology by HLA class II typing using the polymerase chain reaction and sequence specific oligonucleotide probing. J Path01 173:243, 1994.

clitira PJ: A family study confirms that the HLA-DP associations with celiac disease are the result of an extended HLA-DR3 hapiotype. Hum Immunol 31:100, 1991.

18. Howell WM, Evans PR, Devereux SA, Sage DA, Smith JL, Haegert DG: Absence of strong HLA-DR/DQ-DP linkage disequilibrium in the British and French Canadian Caucasoid populations. Eur J Immunogenet 20:363, 1993.

25. O’Driscoll BR, Stevens FM, O’Gorman TA, Finnegan P, McWeeney JJ, Little MPG, Connolly CE, McCarthy CF: HLA type of patients with coeliac disease and malignancy in the west of Ireland. Gut 23:662, 1982.

19. Kimura A, Sasazuki T: Eleventh International Histocompatiblity Workshop reference protocol for the HLA DNAtyping technique. In Tsuji K, Aizawa M, Sasazuki T (eds): HLA 1991, Vol. 1. Oxford, Oxford University Press, 1992. 20. Nevinny-Stickel C, Hinzpeter M, Andreas A, Albert ED: Non-radioactive oligotyping for HLA-DRl-DRWlO using polymerase chain reaction, digoxigenin-labelled oligonucleotides and chemiluminescence detection. Eur J Immunogenet 18:323, 1991. 21. Svejgaard A, Ryder LP: HLA and disease associations: detecting the strongest association. Tissue Antigens 43: 18, 1994. 22. Shibata D: Identification of mismatched fixed specimens with a commercially available kit based on the polymerase chain reaction. Am J Clin Path01 100:666, 1993. 23. Hall MA, Lanchbury JSS, Bolsover WJ, Welsh KI, Ciclitira PJ: Celiac disease is associated with an extended HLA-DR3 halplotype which includes HLA-DPwl. Hum Immunol 27:220, 1990. 24. Bolsover WJ, Hall MA, Vaughan RW, Welsh KI, Ci-

26. SpurkIand A, Sollid LM, Polanco I, Vartdal F, Thorsby E: HLA-DR and DQ genotypes of celiac disease patients serologically typed to be non-DR3 or non-DRSf7. Hum Immunol 35: 188, 1992. 27. Ploski R, Ek J, Thorsby E, Sollid LM: On the HLADQ@ lx050 1, fl1*020 1)-associated susceptibility in celiac disease: a possible gene dosage effect of DQB 1*020 1. Tissue Antigens 41: 173, 1993. 28. Congia M, Cucca F, Frau F, Lampis R, Melis L, Clemente MG, Cao A, De Virgiliis S: A gene dosage effect of the DQA 1*05Ol/DQB 1*020 1 allelic combination influences the clinical heterogeneity of celiac disease. Hum Immunol 40: 138, 1994. 29. Petronzelli F, Multari G, Ferrante P, Bonamico M, Rabuffo G, Campea L, Mazzilli MC: Different dose effect of HLA-DQaP heterodimers in insulin-dependent diabetes mellitus and celiac disease susceptibility. Hum Immunol 36:156, 1993. 30. Murray A, Cuevas EC, Jones DB, Wright DH: Study of the immunohistochemistry and T-cell clonality of enteropathy-associated T-cell lymphoma. Am J Path01 146: 1, 1995.