Clin Exp Immunol 1999; 115:294300

Genetic markers in clinically well dened patients with ulcerative colitis (UC)
G. BOUMA, J. B. A. CRUSIUS, M. A. GARCI A-GONZA LEZ, B. U. G. A. MEIJER, H. P. R. HELLEMANS, R. J. HAKVOORT, G. M. Th. SCHREUDER, P. J. KOSTENSE*, S. G. M. MEUWISSEN & A. S. PEN A Department of Gastroenterology, Academic Hospital Vrije Universiteit Amsterdam, *Department of Epidemiology and Biostatistics, Medical Faculty, Vrije Universiteit Amsterdam, Amsterdam, and Department of Immunohaematology and Bloodbank, Leiden University Medical Centre, Leiden, The Netherlands

(Accepted for publication 13 October 1998)

SUMMARY Results of genetic association studies in UC are conicting. We propose that the power of candidate gene studies will increase when disease heterogeneity is taken into account. Phenotype frequencies of molecularly dened HLA-DR alleles, polymorphisms in the tumour necrosis factor-alpha (TNF-a), lymphotoxin-alpha (LT-a), IL-1 receptor antagonist (IL-1Ra) and IL-1b genes were determined in 98 clinically well characterized UC patients with a mean period of follow up of 10 years, and ethnically matched healthy controls (HC). The alleles HLA-DRB1*0103 (phenotype frequency 6% versus 02%; P 00002; odds ratio (OR) 276) and DRB1*15 (41% versus 26%; P 0001; OR 20, compared with HC) were associated with overall disease susceptibility. Subgroup analysis revealed that DRB1*15 was only increased in females (53% versus 24%; P < 00001; OR 35), but not in males. With regard to disease localization, all DRB1*0103 patients had extensive disease (P < 0002; OR 335), and DRB1*15 was found in 59% of females with extensive colitis (P < 00001; OR 44). DRB1*0103 was signicantly increased in patients undergoing colectomy (P < 00002; OR 84). No association between overall disease susceptibility and the cytokine gene polymorphisms were found. Subgroup analysis revealed several signicant associations, but most did not retain signicance when corrected for multiple comparisons. However, a noticeable nding was that haplotype TNF-C was signicantly associated with progression in extent of disease (P 0003, OR 204). This study provides additional evidence for the role of DRB1 alleles in the susceptibility to UC, and supports the hypothesis that these alleles may determine the severity of the disease. The cytokine gene polymorphisms evaluated in this study do not seem to be strong risk factors for the overall disease susceptibility in UC, but may be involved in determining the severity of the disease. Keywords HLA ulcerative colitis heterogeneity cytokines polymorphism

INTRODUCTION There is substantial evidence that genetic factors play a role in the predisposition to inammatory bowel disease (IBD), i.e. UC and Crohns disease (CD) [14]. However, the relevant genes have not been identied. Genes of the MHC are excellent candidate genes because of their role in the immune response, and the strong associations that exist in other immune-mediated diseases. Studies
G.B. and J.B.A.C. contributed equally to this work. G.B. present address: National Institutes of Health, Mucosal Immunity Section, Building 10, Room 11N238, 10 Center Drive, Bethesda, MD 20892, USA. Correspondence: Professor A. S. Pena, MD, PhD, Laboratory of Gastrointestinal Immunogenetics, Faculty of Medicine Vrije Universiteit. Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands.

on HLA associations in IBD, however, have yielded inconclusive results. An increased prevalence of HLA-DR2 (DRB1*15) in UC has been observed in several studies, especially from Japan [59]. No conclusive data on this association in European Caucasian patients exist, however. We recently observed an increased prevalence of DRB1*15 in a small group of Dutch UC patients [10], and an association with this allele was also found in a study from Spain [11]. A study from the UK, however, showed that the alleles DRB1*0103 and DRB1*12, but not the DRB1*15 allele, were increased in their patient population [12]. In favour of a role for the HLA alleles in disease susceptibility, Satsangi et al. reported linkage of UC to the HLA-DRB1 locus using non-parametric linkage analysis [12]. Cytokines play a key role in the initiation, regulation and effectuation of the immune response. The genes that encode
1999 Blackwell Science

294

Genetic markers in UC
these proteins might therefore be another group of candidate genes for UC. Manseld et al. found that allele 2 of the IL-1 receptor antagonist (IL-1Ra) is increased in patients with (extensive) UC [13]. Although we found no signicant association between disease in general and this allele, we could conrm that allele 2 was signicantly increased in those patients with extensive colitis [14]. Interestingly, it was subsequently found that in IBD, carriers of the IL-1Ra allele 2 are frequently non-carriers of allele 2 of the IL-1b TaqI polymorphism, suggesting that an imbalance in the IL1Ra/IL-1b ratio may contribute to disease susceptibility in IBD [15]. Tumour necrosis factor-alpha (TNF-a) and TNF-b or lymphotoxin-alpha (LT-a) are cytokines that play a central role in the immune reaction. The location of their genes within the MHC class III region has prompted much speculation on the role of these cytokines in HLA-associated diseases. The central role of TNF-a in IBD has recently been shown by the clinical benet of anti-TNF treatment in CD patients [16]. We previously studied four polymorphisms in the TNF region. A small, but signicant decrease of the uncommon allele 2 at position 308 in the promoter region of the TNF-a gene was found in UC [17]. Louis et al. found this allele to be decreased, albeit non-signicantly, in patients with CD, but not in UC patients [18]. Several explanations exist for the different and inconclusive results from studies regarding associations with HLA alleles and cytokine gene polymorphisms. Small sample size, controls that were not ethnically matched, and different HLA typing techniques (serological versus molecular) may all have contributed to the inconclusive and conicting results. However, one of the most important explanations might be that most of these studies did not take into account disease heterogeneity. From the clinical point of view, UC is a heterogeneous disease with marked differences in clinical behaviour, response to treatment and prognosis. Moreover, during the course of their disease, individual patients vary considerably in clinical behaviour. It has been observed that > 50% of patients will have further progression of their disease during follow up [19]. As these different clinical subgroups of patients may have a different genetic susceptibility, or alternatively, if the clinical behaviour may be inuenced by a different genetic background, it is not surprising that the various studies have yielded conicting results, as in most studies only unselected groups of patients were investigated. We propose that the power of genetic association studies may increase when disease heterogeneity is taken into account. This assumption is supported by recent ndings [12,20]. Thus, no association with the DRB1*15 allele was found in an unselected group of patients from the UK [12], but subsequently a signicantly increased prevalence of this allele was found in a subgroup of patients from the same hospital undergoing colectomy [20]. The authors also observed an association with the DRB1*0103 allele. The strength of the association with DRB1*0103, as estimated by the odds ratios, was stronger in their selected study population than in their unselected population. To emphasize the importance of disease heterogeneity in genetic association studies, we can add another observation. We previously studied HLA-DR associations in an unselected group of patients with CD and found no signicant association with any HLA allele tested [10]. However, when we investigated in a subsequent study these alleles in a selected population of patients with stulizing disease, we observed that the DRB1*03 allele was almost absent [21]. The aim of the present study was two-fold. In rst instance we studied the relation between HLA and cytokine gene poly-

295

morphisms and detailed clinical phenotype. Subsequently, we investigated the possible interplay between the various genetic markers previously found to be associated with UC. MATERIALS AND METHODS UC patients Ninety-eight randomly selected UC patients (51 females, 47 males) were included in this study. Part of this study population had been typed previously for some of the markers under investigation [10,14,15,17]. All patients were unrelated Dutch Caucasians. The diagnosis of UC was based on conventional clinicopathological criteria as described by Lennard-Jones [22]. The localization of gut involvement was dened as proctitis, left-sided (up to the splenic exure), or pancolitis (beyond the splenic exure), based on endoscopy and histology. Clinical data were collected retrospectively at two time points: when patients visited the Academic Hospital Vrije Universiteit for the rst time (at a mean time after diagnosis of 36 years, range 030 years), and at their latest visit (after a mean follow up of 102 years, range 032 years). Mean age at diagnosis was 33 years (range 879 years). At the time of diagnosis, 32 patients had distal disease, whereas 66 had extensive colitis (46 left-sided colitis and 20 pancolitis). At follow up, 17 patients had distal disease and 81 extensive disease (47 left-sided colitis and 34 pancolitis). Eighteen patients showed an increase in extent of disease during follow up (15 from proctitis to left-sided and three from proctitis to pancolitis). At follow up, 24 patients had been colectomized. Fifteen UC patients had a positive family history for IBD, among whom ve had a rst degree relative with IBD. With regard to extra-intestinal manifestations, two patients developed sacro-iliitis, two pyoderma gangrenosum, one uveitis, one aphthous stomatitis, and one patient had erythema nodosum. Two patients also suffered from primary sclerosing cholangitis. Control group A group consisting of 2400 unrelated healthy Dutch Caucasian blood donors serologically typed for HLA served as controls. Controls were recruited from the same region as the patients. Part of this control group was also typed for the subtypes of DR1 (DRB1*0101/2/4 and 0103), DR2 (DRB1*15 and DRB1*16), DR5 (DRB1*11 and DRB1*12) and DR6 (DRB1*13 and DRB1*14). In particular, HLA-DRB1*15 was determined in 1576 healthy controls (HC), and the DRB1*0103 allele was studied in 420 HC. A group of 98 HC was typed for polymorphisms in the genes encoding TNF-a, LT-a, IL-1b and the IL-1Ra. DNA isolation, polymerase chain reaction amplication, and dot-blotting for HLA typing Genomic DNA was extracted from EDTA anti-coagulated peripheral blood using a standard proteinase K digestion and phenol/ chloroform extraction procedure. Amplication of the second exon of HLA-DRB1 genes was performed by polymerase chain reaction (PCR). Dot blot analysis of amplied DNA was carried out using the procedures [23] and biotin-labelled sequence-specic oligonucleotide probes (SSO) as previously described [24]. Cytokine gene polymorphisms A single-stranded conformation polymorphism method was used for the detection of bi-allelic polymorphisms at both positions 308 and 238 in the TNF-a gene, as previously described [25]. PCR amplication with primers located in the 50 untranslated

1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 115:294300

296

G. Bouma et al.
The phenotype frequency of DRB1*0103, which is extremely low in the Dutch population, was signicantly increased in UC patients (6% versus 02%, P 00002; OR 276; 95% CI 33232). We previously observed in a small group of UC patients that the DRB1*15 allele is increased [10].We therefore rst conrmed that this allele was increased in the new group of patients as well. We then calculated the phenotype frequency of this allele in the total group of 98 patients, and observed that this allele was signicantly increased when compared with controls (41% versus 26% in HC; P 0001; OR 20; 95% CI 1331). Focusing on DRB1*15, when comparing non-carriers, carriers and homozygous carriers, we found a signicant trend across these categories (x2 for trend 19; P < 00001), conrming previous observations in a small group of patients [10]. The association with DRB1*15 was found to be very signicant in female patients. Twenty-seven (53%) of female UC patients were carriers of the DRB1*15 allele versus 24% of female HC (P < 00001; OR 35; 95% CI 2060). In contrast, no association was found in male UC patients. Thirteen male patients (28%) were carriers of this allele versus 27% of male HC. The homogeneity test revealed that the difference between males and females is signicant (P 001). TNF gene polymorphisms. We previously studied polymorphisms in the TNF genes in a small group of patients and controls [17]. In the present study, this population was extended to 98 patents and 98 controls. In accordance with our previous observations, none of the TNF haplotypes was found to be associated with disease in general, as shown in Table 2. In the present group of patients, TNF-C haplotype was somewhat increased in UC patients, but the difference did not reach signicance (36% versus 26%; P 02; OR P6; 95% CI 0930). The TNF haplotypes were equally distributed among females and males. IL-1Ra and IL-1b gene polymorphisms. In previous studies, we found no signicant association between disease in general and carriage of either allele 2 of the IL-1b gene or allele 2 of the IL-1Ra gene [14,15]. In the present cohort of patients, the phenotype frequency of the IL-1Ra allele 2 was 50% in UC patients versus 42% in HC (P 03; OR 16; 95% CI 0824). The distribution among males and females was equal. The IL-1b allele 2 was not signicantly increased either, although we observed the previously described imbalance in the distribution of the alleles of the IL-1Ra and IL-1b gene polymorphisms in this group of UC patients (results not shown) [15]. Interaction between the genetic markers in UC. Several genetic markers have previously been found to be associated with UC. To our knowledge, however, it has never been investigated whether these genes act in conjunction in the susceptibility to the disease. Both DRB1*15 and IL-1Ra allele 2 are found in approx. 45% of patients. We calculated the phenotype frequencies of IL-1Ra allele 2 in the DRB1*15-positive and -negative subgroups and vice versa. The frequency of carriers of IL-1Ra allele 2 did not differ between DRB1*15-positive and -negative patients (60% versus 44%). Similarly, the frequency of DRB1*15 in carriers and non-carriers of the IL-1Ra allele 2 was not different (49% versus 333%). With regard to the TNF haplotypes and the IL-1b allele 2, only the TNF-P haplotype, which is in linkage disequilibrium with the DRB1*15 allele, was decreased in the DRB1*15 subgroup. HLA-DRB1 and cytokine gene polymorphisms in relation to clinical characteristics Extent of disease. Seventeen patients (12 females, ve males)

region and third intron of the LT-a was used for typing the NcoI restriction fragment length polymorphism (RFLP) [26] as well as the AspHI RFLP [27]. We previously observed that the alleles of these four bi-allelic polymorphisms occur in a restricted number of combinations denoted haplotypes TNF-C, -E, -H, -I and -P (Fig. 1) [17,28]. PCR amplication and TaqI digestion of the polymorphic site within exon 5 of the IL-1b gene were performed and analysed as previously described [15]. The region within the second intron of the IL-1Ra gene that contains variable numbers of an identical tandem repeat (VNTR) of 86 base pairs was amplied by PCR and analysed as previously described [1315]. Statistical analysis Statistics were calculated using Instat version 2.02 (Graphpad Software, San Diego, CA). Phenotype frequencies were compared between the study groups by means of 2 2 table analysis using x2 (with Yates correction) or Fishers exact test. We also tested the possibility of doseresponse relationships across the categories non-carrier, heterozygous carrier, and homozygous carrier using the x2 test for trend. In order to investigate interaction, i.e. to assess whether differences between estimated odds ratios were larger than expected by chance alone, we performed an exact homogeneity test, using the StatXact software package [29]. To correct for multiple comparisons in various subgroups, the P values were corrected using the Bonferroni correction where appropriate. On subgroup analysis, we tested those markers that were increased in the UC population in general: DRB1*0103, DRB1*15, haplotype TNF-C, and the IL-1Ra allele 2. For correction of the number of comparisons, the following parameters were taken into account: disease localization, progression in extent of disease, need for operation, medication, family history and extra-intestinal manifestations. Taking into account these parameters, a P value < 0005 was considered statistically signicant. To show the strength of the associations, we have included P values, odds ratios (OR) and the 95% condence intervals (CI) in the text. RESULTS Associations with overall disease HLA-DRB1. The phenotype and allele frequencies of the HLA-DR alleles are given in Table 1. From one patient, no HLAdata were available. Two alleles were found to be signicantly associated with disease in general: DRB1*0103 and DRB1*15.
TNF- LT-

Haplotype TNF-C TNF-E TNF-H TNF-I TNF-P

238 G G A G G

308 G(1) A(2) G(1) G(1) G(1)

AspHI 2 2 2 2 1

Ncol 1(aa26=asn) 1(aa26=asn) 2(aa26=thr) 2(aa26=thr) 2(aa26=thr)

Fig. 1. Schematic representation of the tandemly arranged TNFA and LTa (TNF-b) genes. A, Untranslated portions of the exons; B, translated parts. The arrow indicates the direction of transcription. Positions of four bi-allelic polymorphisms are indicated.

1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 115:294300

Genetic markers in UC
Table 1. HLA-DR phenotype and allele frequencies in 97 patients with UC and 2400 controls

297

UC HLA-DRBI* 01 (0101/2/4) 0103 15 16 03 04 11 12 13 14 07 08 09 10 HC (%) 20 0.2 26 2 25 28 14 5 28 5 19 5 2 4 (n) 21 6 40 1 21 19 20 6 21 5 12 1 0 3 (%) 22 6 41 1 22 20 21 6 22 5 12 1 0 3 UC versus control OR (95% CI) 1.0 (0.61.7) 27.6 (3.3232) 2.0 (1.33.1) 0.5 (0.14.0) 0.8 (0.51.4) 0.6 (0.41.0) 1.6 (0.92.6) 1.3 (0.63.1) 0.7 (0.41.1) 0.9 (0.42.4) 0.6 (0.31.1) 0.2 (0.031.4) 0.2 (0.013.4) 0.7 (0.22.3) HC AF (%) 10 0.1 14 1 13 15 7 2 15 3 10 3 1 2 (N) 22 6 49 1 24 20 22 6 22 6 12 1 0 3

UC AF (%) 11 3 25 0.5 12 10 11 3 11 3 6 0.5 0 1.5

P 0.0002. P 0.001. HC, Healthy controls; UC, ulcerative colitis; n, number of individuals; N, number of alleles; AF, allele frequency; OR, odds ratio; CI, condence interval.

had distal disease, whereas 81 (39 females, 42 males) had extensive disease at a mean follow-up time of 102 years. The DRB1*0103 allele was signicantly increased in patients with extensive colitis when compared with controls. In fact, all carriers of this allele had extensive disease at follow up (P < 00001; OR 335; 95% CI 4283 compared with HC). In contrast, none of the patients with proctitis was carrier of this allele (P 10; OR 01; 95% CI 000532). The phenotype frequency of the DRB1*15 allele was 29% in patients with proctitis, and did not differ signicantly from HC. Of those patients with extensive colitis, 44% were DRB1*15 (P 0007; OR 22; 95% CI 1435). This association was highly signicant in females (59%; P < 00001; OR 44; 95% CI 2386), but not in males (Fig. 2). Two patients (12%) with distal colitis were carriers of the TNFC haplotype (P NS compared with HC). In extensive colitis, 41% of patients were carriers of this haplotype (P < 004; OR 2; 95% CI 1138), but this association did not retain signicance after correction for the number of comparisons.

The phenotype frequency of the IL-1Ra allele 2 was somewhat higher in patients with extensive colitis (53%) compared with patients with distal colitis (35%) and controls, although the difference did not reach statistical signicance. However, when we investigated the relation between carriership of allele 2 and the trend over the localizations proctitis, left-sided colitis and pancolitis, the association was signicant (x2 for trend 65; P 001). Progression in extent of disease. At follow up, 18 patients had an increase in their extent of disease (15 from proctitis to left-sided, and three from proctitis to pancolitis). Fourteen patients had only proctitis and the disease did not progress during the same follow-up period of observation. The DRB1*15 allele was found in ve patients with stable proctitis and eight with progressive disease (P NS). It must be noted, however, that the follow-up time in those ve patients was signicantly shorter than in the eight patients with progressive disease (36 versus 105 years; P < 005). With regard to the TNF haplotypes, 11 patients with progressive disease were carriers of the TNF-C haplotype, whereas only

Table 2. Phenotype and haplotype frequencies of the ve tumour necrosis factor (TNF) haplotypes in 98 UC patients and 98 controls

Phenotype frequency HC TNF haplotype C E H I P n 25 33 10 34 65 % 26 34 10 35 66 n 35 26 5 39 61 UC % 36 27 5 40 62 UC versus control OR (95% CI) 16 (0930) 07 (0413) 05 (0214) 12 (0722) 08 (0515) N 27 37 10 40 82 HC

Haplotype frequency UC % 14 19 5 20 42 N 39 30 5 43 79 % 20 15 3 22 40

HC, Healthy controls; UC, ulcerative colitis patients; n, number of individuals; OR, odds ratio; CI, condence interval; N, number of haplotypes. 1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 115:294300

298
70

G. Bouma et al.
for CD [2], and chromosomes 12, 7 and 3 for both UC and CD [3]. These data should be interpreted with great caution, as linkage studies investigate solely patients with multiple affected family members. As a positive family history is found in only 1020% of patients, this group might actually represent a subgroup of patients, with a specic genetic susceptibility. As has recently been argued, in complex heterogeneous diseases like IBD, association studies in families have far greater power [30]. We have evaluated HLA-DR alleles as well as polymorphisms in the genes for TNF-a, LT-a, IL-1b and IL-1Ra in clinically well dened patients, both at rst visit to our hospital and at follow up. This novel approach introduces more certainty of disease susceptibility and behaviour of the disease. Using this approach, we have observed the following: 1 We conrm, for the rst time, the association between DRB1*0103 and UC. This is of interest, as this allele is extremely rare in the Dutch population. Satsangi et al. observed that this allele was increased in their unselected population of UC patients from the UK [12]. Moreover, this allele was highly increased in patients with extensive colitis undergoing colectomy [20]. In agreement with these as well as preliminary results from the USA [31], we found that all our DRB1*0103 patients had extensive disease, and four out of six patients had been colectomized at follow up. These ndings support the observation that this allele may predict a more severe form of UC. 2 The DRB11*15 allele was found to be signicantly associated with UC in general, and was strongly associated with female UC. The OR for UC in females was 35 compared with HC. We previously described this association in a limited number of Dutch UC patients [10]. This is the rst report from Northern Europe showing a strong association between UC and DRB1*15 in a large group of Caucasians. To our knowledge, no other reports exist on HLA associations in UC and the female sex. This observation is of particular interest in view of recent ndings in multiple sclerosis. In this disease, which has some remarkable clinical and immunogenetic similarities with UC, a signicant association has been found between relapsing-remitting form of disease and HLA-DRB1*1501 in females, but not in males [32]. With regard to HLA-DR2 associations in UC, the different studies have had conicting results. Studies from Japan and California (where a mixed Jewish/non-Jewish population was studied) have shown signicant associations with DRB1*15. As has been suggested, a possible explanation for the conicting results might be that the DRB1*1502 is the responsible susceptibility allele, as this allele is the most frequent subspecicity of DRB1*15 in the Japanese and Jewish populations. In the Dutch population, DRB1*1501 is the most frequent allele, accounting for > 95% of the DRB1*15 alleles. In our patient population, the large majority of DRB1*15 individuals is DRB1*1501, which makes it unlikely that DRB1*1502 is the susceptibility allele for UC. We suggest that a shared epitope on the DRB1*15 molecule, or a gene in linkage with both DRB1*1501 and DRB1*1502, may determine the disease susceptibility. On subgroup analysis, we found a very signicant association between extensive colitis (in females) and DRB1*15, supporting previous observations from Japan [6] and the UK [20]. In the latter, no association between disease in general and this allele was found [12], but a signicant association was found in patients undergoing colectomy, thus showing the importance of proper patient classication in genetic association studies.

yyyyyy yy yyyyyyy yy y y yy yy y
60 50 40 30 20 10 0 * DRB1*15 (%) ** 0 Proctitis Extensive HC Females Males

Fig. 2. Phenotype frequencies of DRB1*15 in patients with proctitis, extensive colitis and healthy controls (HC). *P < 00001; OR 44; 95% CI 2386; **P 0007; OR 22; 95% CI 1435.

one patient with stable proctitis was a carrier of this haplotype (P < 0003; OR 204; 95% CI 22193). The phenotype frequency of IL-1Ra allele 2 was 67% among patients with progressive disease, and was somewhat increased compared with patients with stable proctitis (43%), although the difference did not reach statistical signicance. Need for operation. Twenty-four patients (nine females, 15 males) underwent colectomy after a mean time after diagnosis of 7 years (range 027 years). The phenotype frequencies of DRB1*0103 in operated patients was 17%, and was signicantly increased over controls (P < 00001; OR 84; 95% CI 9785). In non-operated patients, the phenotype frequency was 3%, and did not differ signicantly from the controls (P 006; OR 12; 95 CI 1130). Both DRB1*15 and IL-1Ra allele 2 were found more often in patients that underwent colectomy, but the P values did not retain signicance when the number of comparisons was taken into account. Medication. During the course of their disease, 53 patients were treated with systemic steroids, and 16 with immunosuppressive drugs (azathioprine or cyclosporin). Twenty-three patients were treated with neither glucocorticosteroids nor immunosuppressive drugs. No association was found between the need for these drugs and any of the markers studied. Family history and extra-intestinal manifestations. In those patients with a positive family history, DRB1*15 was found in 50%, TNF-C in 40%, and allele 2 of the IL-1Ra in 60%, and did not differ with statistical signicance from those without affected family members. Interestingly, none of the patients with a positive family history was DRB1*0103. Extra-intestinal manifestations were found in nine patients. This group was too small for statistical analysis. The three patients with skin manifestations, as well as the two patients with primary sclerosing cholangitis (PSC), were all DRB1*15. DISCUSSION Much effort has already been made to delineate genomic regions containing genes involved in the disease process underlying IBD. Whole genome screening family-based linkage studies have recently revealed the rst regions of interest in chromosome 16

yy yyyyyy yyyyyy yyyyyyyyyyyyy yyyy yyyyyy

Females + males

1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 115:294300

Genetic markers in UC
3 Apart from the signicant association between DRB1*0103 and DRB1*15 and extensive colitis, the TNF-C haplotype as well as the IL-1Ra allele 2 were predominantly found in those patients who developed extensive colitis. Moreover, both DRB1*0103 and DRB1*15 alleles and the IL-1Ra allele 2 were increased in those patients undergoing colectomy. However, most associations did not retain signicance after correction for the number of comparisons. We could conrm our previous observation of a signicant association between carriership of IL-1Ra allele 2 and the trend over the categories proctitis, left-sided colitis and pancolitis [15]. A striking nding was that in the group of proctitis patients > 90% of carriers of the TNF-C haplotype had progression from proctitis to extensive colitis at follow up. Taken together, these ndings support the hypothesis that multiple genes may be responsible for a more severe course of the disease, the HLA alleles being the most signicant. However, based on the results from this study, we can not conclude yet whether these genes act in conjunction or act separately in determining the severity of the disease. There has been some interest in the relation between the presence of antibodies against the neutrophil granulocyte (pANCA) and DRB1*15. Yang et al. found that p-ANCA patients had a signicantly increased prevalence of DR2 compared with ANCA controls [33]. We have studied p-ANCA for > 5 years in a large group of patients, and have observed strong uctuations in pANCA positivity at follow up. A considerable number of patients show seroconversion during their course of disease (unpublished results). However, in those patients that were consistently pANCA (22%), 47% were DRB1*15, suggesting that p-ANCA is not a subclinical marker for the DRB1*15 subgroup in our population. One of the explanations for the association between several immune-mediated diseases and specic HLA alleles is that the HLA alleles are not directly involved in disease susceptibility but are markers for other closely linked genes. TNF-a and LT-a are cytokines with key functions in the immune response. Their genes are located in the central region of the MHC, between the class I and class II genes. In this regard it is of interest that several authors have shown a relation between the presence of HLA-DR2 and decreased production of TNF-a [34,35]. The exact mechanism for this decreased secretion has not been unravelled yet. We previously studied TNF-a secretion by peripheral blood mononuclear cells (PBMC) in relation to polymorphisms in the TNF genes, and found that the TNF-C haplotype is associated with decreased TNFa secretion, whereas the TNF-E haplotype is associated with high secretion of this cytokine [36]. These haplotypes differ only at position 308 in the promoter region of the TNFA gene. Recent results from another group indicate that this position is involved in the regulation of TNF-a transcription [37]. These ndings suggest that individuals carrying the DR2 allele or the TNF-C haplotype may be genetically predisposed to secrete low levels of TNF-a upon an antigenic stimulus. Various studies have investigated whether TNF-a secretion is altered in patients with UC. TNF-a levels in UC patients have been measured in serum, in stools, in the mucosa of the gut, and in isolated PBMC. The results of these studies are highly inconsistent. Interestingly, two studies have shown a decreased secretion of TNF-a in UC patients. Mazlam et al., when studying TNF-a secretion by peripheral blood monocytes after stimulation with lipopolysaccharide (LPS), found a signicantly decreased production of TNF-a in UC patients compared with CD patients and HC [38]. We studied TNF-a and

299

LT-a secretion by PBMC after T cell activation and found that the secretion of biologically active TNF was decreased in UC patients compared with HC [39]. In contrast to these ndings, no studies exist on decreased TNF-a secretion in CD. In the latter, several clinical and laboratory observations suggest a pivotal role for this cytokine in the pathogenesis of CD. Recent clinical studies using anti-TNF-a MoAbs have underlined this [16,40,41]. In our study population, a large number of UC patients with extensive colitis is carrier of a low secretor TNF-a genotype (either DRB1*15 or TNF-C). This may suggest that UC patients have a different immunological disturbance compared with CD patients, and may therefore not benet to the same extent as CD patients from antiTNF-a treatment. Preliminary results from clinical trials support this concept, as this form of treatment was effective in only one of three studies [4244]. Therefore, to unravel further the mechanism of action of this form of treatment, it will be of paramount importance to investigate the genetic background of these patients when evaluating this form of treatment in UC patients. In conclusion, this study provides further evidence for the role of DRB1*0103 and DRB1*15 in disease susceptibility to UC. The DRB1*0103 and the DRB1*15 alleles (in females) predict a more aggressive form of disease, as they are associated with extensive colitis and an increased risk for colectomy. Neither the IL-1Ra nor the TNF genes are associated with overall disease susceptibility, but may play a modest role in determining the severity of the disease. The TNF-C haplotype may be a marker for progressive ulcerative colitis. ACKNOWLEDGMENTS The authors would like to acknowledge the nancial support received from Tramedico B.V., The Netherlands, and the Falk Foundation, Germany. REFERENCES
1 Satsangi J, Jewell DP, Rosenberg WMC, Bell JI. Genetics of inammatory bowel disease. Gut 1994; 35:696700. 2 Hugot JP, Laurent-Puig P, Gower-Rousseau C et al. Mapping of a susceptibility locus for Crohns disease on chromosome 16. Nature 1996; 379:8213. 3 Satsangi J, Parkes M, Louis E et al. Two stage genome-wide search in inammatory bowel disease provides evidence for susceptibility loci on chromosomes 3, 7 and 12. Nature Genetics 1996; 4:199202. 4 Ohmen JD, Yang HY, Yamamoto KK et al. Susceptibility locus for inammatory bowel disease on chromosome 16 has a role in Crohns disease, but not in ulcerative colitis. Hum Mol Gen 1996; 5:167983. 5 Sugimura K, Asakura H, Mizuki M et al. Analysis of genes within the HLA region affecting susceptibility to ulcerative colitis. Hum Immunol 1993; 36:1128. 6 Masuda H, Nakamura Y, Tanaka T, Hayakawa S. Distinct relationship between HLA-DR genes and intractability of ulcerative colitis. Am J Gastroenterol 1994; 89:195762. 7 Futami S, Aoyama N, Honsako Y et al. HLA-DRB1*1502 allele, subtype of DR15, is associated with susceptibility to ulcerative colitis and its progression. Dig Dis Sci 1995; 40:8148. 8 Toyoda H, Wang SJ, Yang HY et al. Distinct associations of HLA class II genes with inammatory bowel disease. Gastroenterology 1993; 104:7418. 9 Caruso C, Palmeri P, Oliva L, Orlando A, Cottone M. HLA antigens in ulcerative colitis; a study in the Sicilian population. Tissue Antigens 1985; 25:4749. 10 Bouma G, Oudkerk Pool M, Crusius JBA et al. Evidence for genetic heterogeneity in IBD. HLA genes in the predisposition to suffer from

1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 115:294300

300

G. Bouma et al.
haplotypes in the TNF region in celiac disease. Eur Cytok Network 1994; 2:168. Mehta C, Patel N. StatXact-Turbo. Statistical software for exact nonparametric inference. User manual. Cambridge MA: CYTEL Software Corporation, 1992. Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science 1996; 273:15167. Duerr RH, Chensy LJ. Associations between HLA-DR alleles and subsets of ulcerative colitis dened by extent of colitis [Abstract]. Gastroenterology 1997; 112:A963. De la Concha EG, Arroyo R, Crusius JBA et al. Combined effect of HLA-DRB1*1501 and interleukin-1 receptor antagonist gene allele 2 in susceptibility to relapsing/remitting multiple sclerosis. J Neuroimmmol 1997; 80:1728. Yang H, Rotter JI, Toyoda H et al. Ulcerative colitis: a genetically heterogeneous disorder dened by genetic (HLA class II) and subclinical (antineutrophil cytoplasmic antibodies) markers. J Clin Invest 1993; 92:10804. Bendtzen K, Morling N, Fomsgaard A et al. Association between HLADR2 and production of tumour necrosis factor alpha and interleukin 1 by mononuclear cells activated by lipopolysaccharide. Scand J Immunol 1988; 28:599606. Jacob CO, Fronek Z, Lewis GD, Koo M, Hansen JA, McDevitt HO. Heritable major histocompatibility complex class II-associated differences in production of tumor necrosis factor alpha: relevance to genetic predisposition to systemic lupus erythematosus. Proc Natl Acad Sci USA 1990; 87:12337. Bouma G, Crusius JBA, Oudkerk Pool M et al. Secretion of tumor necrosis factor alpha and lymphotoxin alpha in relation to TNF gene polymorphisms and HLA-DR alleles. Relevance for inammatory bowel disease. Scand J Immunol 1995; 43: 45663. Wilson AG, Symons JA, McDowell TL, McDevitt HO, Duff GW. Effects of a polymorphism in the human tumor necrosis factor a promoter on transcription activation. Proc Natl Acad Sci USA 1997; 94:31959. Mazlam MZ, Hodgson HJF. Peripheral blood monocyte production and acute phase response in inammatory bowel disease. Gut 1992; 33:7738. Bouma G, Oudkerk Pool M, Scharenberg JGM et al. Differences in the intrinsic capacity of T-cells to produce the cytokines tumour necrosis factor alpha and beta in patients with inammatory bowel disease and healthy controls. Scand J Gastroenterol 1995; 30:1095100. Derckx B, Taminiau J, Radema S et al. Tumour necrosis factor antibody treatment in Crohns disease. Lancet 1993; 342:1734. Stack WA, Mann SD, Roy AJ et al. Randomised controlled trial of CDP571 antibody to tumour necrosis factor-alpha in Crohns disease. Lancet 1997; 349:5214. Baert F, DHaens G, Geboes K, Ectors N, Rutgeerts P. TNF-alpha antibody therapy causes a fast and dramatic decrease of histologic colonic inammation in Crohns disease but not in ulcerative colitis [Abstract]. Gastroenterology 1996; 110:A859. Sands BE, Podolsky DK, Tremaine WJ et al. Chimeric monoclonal anti-tumor necrosis factor antibody (cA2) in the treatment of severe, steroid-refractory ulcerative colitis (UC) [Abstract]. Gastroenterology 1996; 110:A1008. Evans RC, Clark L, Heath P, Rhodes JM. Treatment of ulcerative colitis with an engineered human anti-TNF alpha antibody cdp571 [Abstract]. Gastroenterology 1996; 110:A905.

11

12

13

14

15

16

17

18 19

20

21

22 23

24

25

26

27

28

ulcerative colitis and Crohns disease. Clin Exp Immunol 1997; 109:1759. De la Concha EG, Fernandez-Arquero M, Santa-Cruz S et al. Positive and negative associations of distinct HLA-DR2 subtypes with ulcerative colitis (UC). Clin Exp Immunol 1997; 108:3925. Satsangi J, Welsh KI, Bunce M et al. Contribution of genes of the major histocompatibility complex to susceptibility and disease phenotype in inammatory bowel disease. Lancet 1996; 347:12127. Manseld JC, Holden H, Tarlow JK et al. Novel genetic association between ulcerative colitis and the anti-inammatory cytokine interleukin-1 receptor antagonist. Gastroenterology 1994; 106:63742. Bioque G, Bouma G, Crusius JBA et al. Evidence for genetic heterogeneity in IBD. 1. The interleukin-1 receptor antagonist in the predisposition to suffer from ulcerative colitis. Eur J Gastroenterol Hepatol 1996; 8:10510. Bioque G, Crusius JBA, Koutroubakis I et al. Allelic polymorphism in IL-1b and IL-1 receptor antagonist (IL-1RA) genes in inammatory bowel disease. Clin Exp Immunol 1995; 102:37983. van Dullemen HM, van Deventer SJH, Hommes DW et al. Treatment of Crohns disease with anti-tumor necrosis factor chimeric monoclonal antibody (cA2). Gastroenterology 1995; 109:12935. Bouma G, Xia B, Crusius JBA et al. Distribution of four polymorphisms in the tumour necrosis factor (TNF) genes in patients with inammatory bowel disease (IBD). Clin Exp Immunol 1996; 103: 3916. Louis E, Satsangi J, Roussomoustakaki M et al. Cytokine gene polymorphism in inammatory bowel disease. Gut 1996; 39:70510. Langholz E, Munkholm P, Davidsen M, Nielsen OH, Binder V. Changes in extent of ulcerative colitis. A study on the course and prognostic factors. Scand J Gastroenterol 1996; 31:2606. Roussomoustakaki M, Satsangi J, Welsh K et al. Genetic markers may predict disease behavior in patients with ulcerative colitis. Gastroenterology 1997; 112:184563. Bouma G, Poen C, Garca Gonzalez MA et al. HLA-DRB1*03, but not the TNFalpha -308 promoter gene polymorphism confers protection against stulising Crohns disease. Immunogenetics 1998; 47:4515. Lennard-Jones JE. Classication of inammatory bowel disease. Scand J Gastroenterol 1989; 24 (Suppl. 170): 26. Verduyn W, Doxiadis IIN, Anholts J et al. Biotinylated DRB sequencespecic oligonucleotides. Comparison to serologic HLA-DR typing of organ donors in Eurotransplant. Hum Immunol 1993; 37:5967. Mehra NK, Verduyn W, Taneja V, Drabbels J, Singh SPN, Giphart MJ. Analysis of HLA-DR2-associated polymorphisms by oligonucleotide hybridization in an Asian Indian population. Hum Immmol 1991; 32:24653. DAlfonso S, Momigliano Richiardi P. A polymorphic variation in a putative regulation box of the TNFA promoter region. Immunogenetics 1994; 39:1504. Messer G, Spengler U, Jung MC et al. Polymorphic structure of the tumor necrosis factor (TNF) locus: a NcoI polymorphism in the rst intron of the human TNF-beta gene correlates with a variant amino acid in position 26 and a reduced level of TNF-beta production. J Exp Med 1991; 173:20919. Ferencik S, Lindemann M, Horsthemke B, Grosse-Wilde H. A new restriction fragment length polymorphism of the human TNF-beta gene detected by AspHI digest. Eur J Immunogenet 1992; 19:42530. Crusius JBA, Bing X, Mulder CJJ, Mearin ML, Pena AS. Relevance of

29

30 31

32

33

34

35

36

37

38 39

40 41

42

43

44

1999 Blackwell Science Ltd, Clinical and Experimental Immunology, 115:294300