American Journal of Gastroenterology ISSN 0002-9270C© 2004 by Am. Coll. of Gastroenterology doi: 10.1111/j.1572-0241.2004.04107.xPublished by Blackwell Publishing

Are pANCA, ASCA, or Cytokine Gene PolymorphismsAssociated with Pouchitis? Long-term Follow-up in 102Ulcerative Colitis PatientsJames Aisenberg, M.D., Peter E. Legnani, M.D., Naris Nilubol, M.D., Gena M. Cobrin, M.D.,Sharif H. Ellozy, M.D., Refaat A.F. Hegazi, M.D., Ph.D., Jessica Yager, B.A., Carol Bodian, Dr. P.H.,Stephen R. Gorfine, M.D., Joel J. Bauer, M.D., Scott E. Plevy, M.D., and David B. Sachar, M.D.Departments of Gastroenterology, General Surgery and Biomathematics, Mount Sinai Medical Center,New York; Department of Medicine, New York University Medical Center, New York, New York;and Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, PA

OBJECTIVE: Pouchitis is the most frequent complication after ileal pouch-anal anastomosis for ulcerative colitis.This study aims to analyze the frequency and characteristics of pouchitis in long-term follow-up in alarge population, and to determine whether a significant association exists between fiveimmunogenetic markers and pouchitis.

METHODS: From a population of over 500 ulcerative colitis patients who had undergone ileal pouch-analanastamosis 5–12 yr earlier, 102 subjects participated in the study. Using clinical data obtainedfrom interviews and chart reviews, patients were classified into three groups: no pouchitis; 1–2episodes per year; and >2 episodes per year. Coded sera from the patients were analyzed forulcerative colitis-associated perinuclear antineutrophil cytoplasmic antibodies and Crohn’sdisease-associated anti-saccharomyces cerevesiae antibodies. Interleukin-1 receptor antagonist,tumor necrosis factor (TNF), and lymphotoxin β (lymphotoxin) polymorphisms were also analyzed.

RESULTS: Pouchitis affected 49% of the study population. Antineutrophil cytoplasmic antibodies,anti-saccharomyces cerevesiae antibodies, and lymphotoxin-β polymorphisms were not associatedwith pouchitis. Carriage of interleukin-1 receptor antagonist allele 2 was significantly greater amongthose without pouchitis than those with pouchitis. Patients without pouchitis had a significantlygreater carriage rate of TNF allele 2.

CONCLUSIONS: Perinuclear antineutrophil cytoplasmic antibodies and anti-saccharomyces cerevesiae antibodiesare not correlated with pouchitis, but interleukin-1 receptor antagonist and TNF may play a role inits development. Further evaluation of these markers in pouchitis will require larger populations,long-term prospective observation, and studies that correlate polymorphisms with specificimmunologic functions.

INTRODUCTION

Pouchitis is the most frequent complication of ileal pouch-anal anastomosis (IPAA) in patients with chronic ulcerativecolitis (UC), with an occurrence in long-term follow-up stud-ies approximating 50% (1, 2). Although the causes of pouch-itis remain unknown, an interplay of environmental factorsand host susceptibility seems likely. The occurrence of pouch-itis in UC but not in familial adenomatous polyposis patients(3) and the reported association of pouchitis with certainimmunogenetic markers (4, 5) suggests an immunogeneticetiology. Also, the response of pouchitis to oral antibiotics(6), the “microbial imbalance” in the pouches in these pa-tients (7), and the prevention of relapse by the administrationof oral probiotics (8) indicate that altered bacterial flora areimportant.

The anti-saccharomyces cerevisiae antibody (ASCA) andthe UC-specific perinuclear antineutrophil cytoplasmic anti-body (pANCA) are immune markers found in the serum ofmany patients with inflammatory bowel disease (IBD), andmay distinguish different clinical phenotypes (9, 10). The cy-tokines interleukin-1 (IL-1) and tumor necrosis factor (TNF)are important mediators of human inflammation, and havebeen postulated to modulate the inflammatory response ininflammatory bowel disease (IBD) patients (4, 11).

The aims of this study were: (1) to analyze the frequencyand characteristics of pouchitis in long-term follow-up in ourpatients; (2) to determine whether either the UC-associatedpANCA or the Crohn’s disease (CD)-associated ASCA arepredictive of pouchitis; and (3) to determine whether poly-morphisms of either IL-1 receptor antagonist (IL-1ra) or tu-mor necrosis factor (TNF) or lymphotoxin β (lymphotoxin)

Are Gene Polymorphisms Associated with Pouchitis? 433

genes are associated with this disorder. There are severalwell-characterized polymorphic sites within the genes forTNF and IL-1ra, some of which, it has been suggested, influ-ence cytokine production. In particular, the IL-1ra gene hasa variable number tandem repeat (VNTR) in intron 2, withup to five variants depending on the number of repeats ofthe 86-base pair (bp) fragment. The TNF locus, which con-tains the genes for TNF and lymphotoxin is located on chro-mosome 6, within a major histocompatibility complex. Ge-netic variations within the TNF locus have been implicated inthe pathogenesis of numerous autoimmune and inflammatorydiseases, including inflammatory bowel disease. Two previ-ously described TNF locus polymorphisms were evaluatedin this study: a single nucleotide polymorphism at position–308 in the TNF gene, and a single nucleotide polymorphismwithin the first intron of the lymphotoxin gene, which allowsrecognition and cleavage of this sequence by the restrictionenzyme NcoI (restriction fragment length polymorphism, orRFLP).

MATERIALS AND METHODS

Patient EnrollmentPatients were identified from the medical records of one col-orectal surgery practice (JB, SG) specializing in IBD. Sub-jects were enrolled from a population of over 500 UC pa-tients who had undergone total proctocolectomy with IPAAbetween 1987 and 1995. All patients who underwent restora-tive proctocolectomy have been registered in our databaseprospectively since 1987. The diagnosis of UC had been es-tablished for each patient by the gastroenterologist and sur-geon utilizing accepted criteria. All patients who could becontacted by telephone, who agreed to participate, and whoprovided complete historical data and blood were included inthe study. Each patient included was assigned a unique iden-tifier number. The Institutional Review Board of the MountSinai School of Medicine approved the study, and informedconsent was obtained from all the study subjects.

Historical ReviewHistorical data were obtained from office chart review andtelephone interviews. Three physicians (one gastroenterol-ogy fellow and two surgical residents) reviewed patient chartsfor evidence of postoperative gastrointestinal illnesses. Whenpostoperative illnesses were reported, all pertinent clini-cal, endoscopic, histopathologic, and laboratory data wererecorded.

One of the three physicians then administered a telephonequestionnaire to each patient. Each patient was questionedregarding demographic and general clinical data, the his-tory of their UC prior to surgery, and their gastrointesti-nal health since surgery. All patients were asked whetheror not they had received the diagnosis of “pouchitis” fromtheir colorectal surgeons and/or gastroenterologists, experi-enced the characteristic symptoms of pouchitis, or received

oral antibiotic treatment. Patients who had been diagnosedwith pouchitis were asked to estimate how often they hadexperienced attacks: 1–2 attacks/yr or >2 attacks/yr. Thesepatients went on to complete a telephone questionnaire re-garding specific symptoms during attacks. Specifically, theywere asked whether during an attack they experienced cramp-ing or pelvic pain, loss of control of their bowel movements,fevers, or blood in the bowel movement. Patients were askedto estimate their usual stool frequency and their stool fre-quency when ill. In addition, patients were questioned abouttheir treatment for pouchitis and about the effectiveness ofthis treatment.

To ensure reproducibility in the pouchitis classificationsmade by the interviewers, five subjects, randomly selected,were interviewed blindly at the outset of the study by all threeinterviewers. There was 100% agreement in their classifica-tions. Subsequent interviews were done by one of the threeinterviewers but the interobserver-agreement exercise on thefive patients was performed for interviewer training and forassurance of consistency in the interview technique.

Patient GroupingPatients who stated they had never had pouchitis, and whohad no evidence in their medical record to suggest pouchitiswere classified as “no pouchitis.” Patients who stated theyhad pouchitis, in whom the telephone interview corroboratedthe diagnosis of pouchitis, and in whom chart review did notreveal a different etiology for their symptoms, were classifiedas “pouchitis.” Patients who had only surgical complications,such as anastomotic breakdowns, but no pouchitis, were clas-sified in the “no pouchitis” group.

Based on chart evidence and patient recollection, the pa-tients were then divided into three groups: no pouchitis; 1–2attacks/yr; and >2 attacks/yr. Patients who experienced >2attacks/yr were further categorized into those who requiredoccasional medication, those who required constant medica-tion to feel well, and those with persistent symptoms despiteconstant medications.

Laboratory AssaysTen milliliters of blood was collected from each subject.Coded sera were analyzed for the UC-specific pANCA andthe ASCA at Prometheus Laboratories (San Diego, CA), uti-lizing methods previously published (12, 13). In brief, serawere analyzed for ANCA using a fixed neutrophil ELISA, andantibody titers were measured. ANCA-positive specimensunderwent indirect immunofluorescence staining. Those witha perinuclear ANCA pattern underwent DNase digestion.Elimination of the pANCA pattern with DNase treatmentconstituted the UC-associated pANCA pattern. A cytoplas-mic ANCA pattern or retention of pANCA pattern afterDNase were considered to be negative for UC-associatedpANCA. Sera were tested for the presence of ASCA atPrometheus Laboratories using a proprietary ELISA system,which measures levels of IgG and IgA specific to a highlypurified oligosaccharide from S. cerevisiae. To be considered

434 Aisenberg et al.

positive, either IgA, IgG, or both had to exceed the normalreference range.

IL-1ra, TNF, and Lymphotoxin Genetic PolymorphismsGenomic DNA was isolated from whole blood usingQiaAMP spin columns (Qiagen). Genotyping for cytokinepolymorphisms was performed using polymerase chain re-action (PCR) based methodologies. Alleles of an 86 basepair (bp) VNTR in the second intron of the IL-1ra gene weredetermined as previously described (14) with minor modi-fications in the thermocycler protocol. PCR products wereelectrophoresed on a 2% agarose gel, and VNTRs were de-tected by ethidium bromide staining. IL-1ra VNTR analysisrevealed five alleles identified by PCR products of differentsizes. A 410 bp band corresponded to the class 1 allele (4repeats), a 240 bp band to the class 2 allele (2 repeats), a 500bp band to the class 3 allele (5 repeats), a 325 bp band to theclass 4 allele (3 repeats), and a 595 bp band to the class 5allele (6 repeats).

A transition polymorphism (G to A) in the –308 position ofthe TNF gene was detected (15). Following PCR of genomicDNA, products were purified using QIAquick PCR purifica-tion kit (Qiagen), then digested with 10 units of the Nco Irestriction enzyme overnight at 37◦C. Digest products wereelectrophoresed on a 2.5% agarose gel and detected by ethid-ium bromide staining. TNF gene RFLP analysis revealed twoalleles that produced three bands of different sizes: a set of87 bp and 20 bp bands corresponding to the TNF-1 allele (re-striction site present), and a 107 bp fragment correspondingto the TNF-2 allele (restriction site absent).

A single nucleotide polymorphism in the first intron ofthe lymphotoxin gene was detected as previously described(15). Following PCR of genomic DNA, products were puri-fied using QIAquick PCR purification kit (Qiagen), then di-gested with 10 units of the Nco I restriction enzyme overnightat 37◦C. Digest products were electrophoresed on a 0.8%agarose gel and detected by ethidium bromide staining. Asdescribed previously (15), lymphotoxin RFLP analysis re-vealed two alleles that produced three bands of different sizes:a 740 bp band corresponding to the lymphotoxin 1 allele (re-striction site absent), and a set of 555 bp and 185 bp bandscorresponding to the lymphotoxin 2 allele (restriction sitepresent).

For all genetic analyses, patients were considered to becarriers of an allele if they had that allele present in eitherone (heterozygous) or two copies (homozygous). Allele des-ignation was determined independently by two investigatorsblinded to all clinical information. There was 100% agree-ment in allele designation by the two investigators.

Statistical AnalysisStatistical analyses were performed using SPSS software (16)and StatXact software (17). Patients were classified in termsof the occurrence of pouchitis within 5 yr of surgery, forwhich complete information was available for all patients.

Statistical comparisons between the study groups ofpANCA positivity, ASCA positivity, allele frequencies, andcarriage rates were performed using Cochran-Armitageχ2 tests for trend. A p<0.05 was considered statisticallysignificant.

RESULTS

Patient Characteristics and Pouchitis PrevalenceOne hundred forty-two patients were contacted by telephone.Of these, 136 agreed to participate in the study. One hundredand two of 136 provided both complete telephone interviewsand blood. The study was conducted without any financialburden to the participants. Multiple contacts were attemptedby phone and mail for all patients in the cohort, but the major-ity had either changed phone numbers, changed residences,and/or was lost to medical follow-up by the surgery practicefrom the time of the predefined colectomy dates of 1987 to1995.

Pouchitis affected 49% of patients within 5 yr (Fig. 1).Eleven percent described 1–2 episodes/yr, while 38% re-ported >2 episodes/yr. In this last group, 10 patients requiredconstant medication to feel well, and 4 patients experiencedchronic symptoms despite constant medication.

The demographic characteristics of the 102 patientsgrouped by pouchitis attack frequency are listed in Table 1.The patients in the three groups were similar with respect toage, duration of disease, years since pouch construction, andgender. Eighty-eight percent of patients had left-sided colitiswhile in 12% the colitis was universal at the time of surgery.In addition, histologic grading of the colitis was determinedby a gastrointestinal pathologist at the Mount Sinai MedicalCenter using established criteria. Fifty percent of patients hadsevere colitis, 17% had moderate colitis, 26% had mild col-itis, and 7% had inactive colitis. There were no statisticallysignificant differences in the histologic grade of severity of

Figure 1. Classification of 102 UC patients following ileal pouch-anal anastomosis according to the occurrence of pouchitis during5–12 yr follow-up.

Are Gene Polymorphisms Associated with Pouchitis? 435

Table 1. Demographic and Clinical Characteristics

Pouchitis Group

No Pouchitis (n = 52) 1–2 Episodes/Yr (n = 11) >2 Episodes/Yr (n = 39) p All Patients (n = 102)

Age – (yr)At study (1999):

Mean 45.52 ± 11.38∗ 40.36 ± 10.23 41.74 ± 15.22 0.15 43.52 ± 12.94Range 20–68 25–54 19–84 19–84

At pouch surgery:Mean 40.35 ± 12.19 35.83 ± 9.95 34.74 ± 15.36 .21 37.72 ± 13.59Range 14–73 23–48 12–78 12–78

At time of UC Dx:Mean 28.98 ± 12.55 22.18 ± 8.26 24.56 ± 13.92 .06 26.56 ± 12.88Range 10–62 14–42 2–65 2–65

SexMale – no. (%) 22 (42) 5 (45) 17 (44) 0.99 44 (44)

Sclero. cholangitis?Yes 0 0 1 0.99 1

Surgery indication† 0.20Disease activity 43 8 37 88Dysplasia 7 3 2 12

∗± Values represent mean ± S.D.†Data not available for 2 patients.

colitis (p = 0.29) or extent of colitis (p = 0.94) at the timeof surgery between the groups who had developed pouchitisand the group who had not.

Only one patient had developed primary sclerosing cholan-gitis. This patient experienced >2 episodes of pouchitis peryear, which remitted with occasional oral antibiotic therapy.The patient was pANCA and ASCA negative, was a carrierof allele 2 of the IL-1ra and lymphotoxin genes, and was nota carrier of allele 2 of the TNF gene.

The majority of patients (85%) had undergone surgery dueto colitis disease activity rather than to the discovery of dys-plasia. Symptoms of disease activity indicating colectomyincluded refractory inflammatory diarrhea, profuse bleeding,perforation, toxic megacolon, inability to tolerate medication,and poor quality of life. Among patients who had colectomyfor dysplasia, 25% (3/12) developed pouchitis. Of those whounderwent colectomy for disease activity, 49% (43/88) devel-

Figure 2. Frequency of UC-associated pANCA among different pouchitis groups (not statistically significant by χ2 test for trend).

oped pouchitis. Although there was a trend toward a higherincidence of pouchitis in the latter group, this difference wasnot statistically significant (p = 0.15). Only 1 of the 14 pa-tients with pouchitis that required constant medications orexperienced chronic symptoms despite constant medicationshad undergone colectomy due to dysplasia.

pANCA Prevalence and TitersThe overall prevalence of UC-associated pANCA in our studypopulation was 54%. Cytoplasmic ANCA was noted in 2% ofpatients, and a pANCA that did not disappear with DNase di-gestion was noted in an additional 9%. UC-specific pANCAwas not associated with pouchitis (p = 0.83, Fig. 2). Sim-ilarly, of the 10 patients requiring constant medications tofeel well, 5 (50%) were UC-specific pANCA negative and5 (50%) were positive. All of the 4 patients with chronic

436 Aisenberg et al.

Figure 3. Frequency of ASCA among different pouchitis groups (not statistically significant by χ2 test for trend).

symptoms despite daily medication for pouchitis were UC-specific pANCA positive.

Of the 55 patients overall who were UC-specific pANCApositive, 45 had a low titer pANCA on ELISA (<40 EU/ml),9 had an intermediate-level ELISA (40–100 EU/ml), and1 had a high-level ELISA (>100 EU/ml). The one patientwith high-level pANCA suffered from mild pouchitis (1–2episodes/yr), and was not a carrier of allele 2 of IL-1ra, TNF,or lymphotoxin. Of the nine patients with the intermediatelevel pANCA, seven had no pouchitis, and two had chronicpouchitis that remitted with occasional oral antibiotics.

ASCA PrevalenceThe overall prevalence of ASCA in the study populationwas 15%. ASCA was not associated with pouchitis (p =0.570) (Fig. 3). Furthermore, neither ASCA IgA nor IgGalone was associated with pouchitis. Of the 14 patientswith chronic pouchitis requiring constant medication or withchronic symptoms despite daily medications, 2 were ASCApositive.

IL-1ra VNTR Allele 2 CarriageThe IL-1ra VNTR results are displayed in Table 2. The twomost common allelic combinations in this population were1,1 and 1,2 (Table 2). There was a higher incidence of pouch-itis in patients who carried the allelic combination 1,1 com-pared to those who carried 1,2 (67%, 26 of 39 patients vs 35%,18 of 52 patients; p = 0.01). Similarly, there was a higherincidence of the allelic combination 1,1 in the group definedas having the most “severe” pouchitis (>2 episodes/yr) com-pared to the 1,2 allelic combination (62%, 21 of 34 patients vs29%, 14 of 48 patients; p = 0.01). From this analysis, it ap-peared that the presence of allele 2 may be protective againstthe development of pouchitis. Therefore, we analyzed car-riage of allele 2 in this population. Sixty percent of patientsin the study population carried at least one copy of allele 2.There was increased carriage of allele 2 in pouchitis-free pa-

tients compared with patients who had experienced pouchitis(95% CI 0.13 and 0.67; p = 0.01) (Fig. 4). Carriage rates forallele 2 were 73% in the “no pouchitis” group, 50% in the“1–2 episodes/yr” group, and 46% in the “>2 episodes/yr”cohort. Of the 14 most symptomatic patients who either re-quired constant medication or had ongoing symptoms despitemedication, the IL-1ra allele 2 carriage rate was 50%.

No relation was noted between ASCA and carriage of al-lele 2 (57% among ASCA (+) and 61% among ASCA (–)).

Table 2. The Number of Patients with each IL-1ra, TNF, and Lym-photoxin genotype (n = 100).

IL-1 ra Genotype

1,1 1,2 1,3 2,2 2,3 2,5

No pouchitis 13 34 1 1 2 01–2 episodes/yr 5 4 0 0 1 0>2 episodes/yr 21 14 0 3 0 1

TNF Genotype

1,1 1,2 2,2

No pouchitis 35 16 01–2 episodes/yr 7 2 1>2 episodes/yr 35 4 0

Lymphotoxin Genotype

1,1 1,2 2,2

No pouchitis 22 25 41–2 episodes/yr 4 5 1>2 episodes/yr 23 15 1

Notes: The numbers shown in each category represent both the number of patientsas well as the percentage.Data not available for two patients. For IL-1ra VTNRgenotype, there was a higher incidence of pouchitis in patients who carried the alleliccombination 1,1 compared to those who carried 1,2 (67% vs 35%; p = 0.01). There isalso a higher incidence of the allelic combination 1,1 in the group defined as havingthe most “severe” pouchitis (>2 episodes/yr) compared to the 1,2 allelic combination(62% vs 29%; p = 0.01). For the TNF genotype, there was a higher incidence ofpouchitis in patients who carried the allelic combination 1,1 compared to those whocarried 1,2 (55% vs 27%; p = 0.025). There was a higher incidence of the TNF alleliccombination 1,1 in the group defined as having the most “severe” pouchitis (>2episodes/yr) compared to the 1,2 allelic combination (50% vs 20%; p = 0.025). Therewas not a significant correlation between lymphotoxin genotypes and the developmentof pouchitis (p > 0.05).

Are Gene Polymorphisms Associated with Pouchitis? 437

Figure 4. Carriage of IL-1ra allele 2 among different pouchitis groups (∗statistically significant by χ2 test for trend; p = 0.01).

Similarly, we found no association between pANCA and theIL-1ra allele 2 polymorphism. Fifty-five percent of pANCA(+) patients were carriers, while 66% of pANCA (–) patientswere carriers.

TNF RFLP Allele 2 CarriageThe TNF RFLP results are displayed in Table 2. All pa-tients but one carried the allelic combinations 1,1 and 1,2(Table 2). There was a higher incidence of pouchitis in pa-tients who carried the allelic combination 1,1 compared tothose who carried 1,2 (55%, 42 of 77 patients vs 27%, 6 of22 patients; p = 0.025). Again, there was a higher incidenceof the allelic combination 1,1 in the group defined as havingthe most “severe” pouchitis (>2 episodes/yr) compared to the1,2 allelic combination (50%, 35 of 70 patients vs 20%, 4 of20 patients; p = 0.025). From this analysis, it appeared thatthe presence of TNF allele 2 may be protective against thedevelopment of pouchitis. Therefore, we analyzed carriage ofallele 2 in this population. Twenty-three percent of patientsin the study population carried at least one copy of allele 2.

Figure 5. Carriage of TNF allele 2 among different pouchitis groups (∗statistically significant by χ2 test for trend; p = 0.02).

Fourteen percent of patients with pouchitis carried allele 2versus 31.4% of patients without pouchitis (the 95% CI 0.13and 0.96; p = 0.035) (Fig. 5). Carriage rates for TNF allele2 were 31% in the “no pouchitis” group, 30% in the “1–2episodes/yr” group, and 10% in the “>2 episodes/yr” group.Interestingly, of the 14 most symptomatic pouchitis patients,only 1 was a carrier of allele 2.

Lymphotoxin RFLP Allele 2 CarriageLymphotoxin RFLP results are also represented in Table 2.The two most common allelic combinations were 1,1 and1,2 (Table 2). Neither allelic combination significantly cor-related with the presence of pouchitis (p > 0.05). Fifty-twopercent of patients carried at least one copy of allele 2. Al-though there was less carriage of allele 2 in the group ofpatients with >2 episodes of pouchitis/yr, the difference wasnot significant (57% in the “no pouchitis” group, 60% in the“1–2 episodes/yr” group, and 41% in the “>2 episodes/yr”group) (Fig. 6). As the TNF and lymphotoxin polymorphismsare in close linkage disequilibrium, as has been previously

438 Aisenberg et al.

Figure 6. Carriage of lymphotoxin allele 2 among different pouchitis groups (not statistically significant by χ2 test for trend).

demonstrated, 100% of individuals who carried the TNF al-lele 2 also carried lymphotoxin allele 2.

DISCUSSION

Pouchitis is the most frequent complication of surgery forulcerative colitis. In worst cases, it may necessitate pouchremoval and creation of an ileostomy. Risk factors for pou-chitis in ulcerative colitis patients include extra-intestinalmanifestations of IBD, primary sclerosing cholangitis, andnonsmoking (18).

A recent study demonstrated that oral administration of ahighly concentrated mixture of probiotic bacterial strains waseffective in the prevention of relapses in patients with chronicpouchitis (8) highlighting the etiologic role of alteration of thegut flora in causing pouchitis. Fecal stasis is also important,since inflammation does not occur at end ileostomies in UCpatients.

Host immunogenetic factors also must be important, sincepouchitis occurs in UC patients but not in patients who un-dergo pouch construction for familial adenomatous polypo-sis. The present study examines the natural history of pou-chitis in a long-term follow-up of a large group of patientsfrom our center, and tests for a role for several importantimmunogenetic markers.

We enrolled patients in this study who had undergonepouch construction between 1987 and 1995, ensuring thateach patient had at least 5 yr of exposure to the onset ofpouchitis. We chose this long exposure time in order to opti-mize the accuracy of our patient grouping, since pouchitismay initially develop as late as 5 yr after ileostomy clo-sure (19). We were able to contact only a fraction of pa-tients who had restorative proctocolectomy despite multipleattempts. It would be impossible to compare those patientswho had geographically relocated to patients included in thisstudy due to insufficient data. We attempted to enroll all pa-tients with whom we established contact without selectionbias. One resulting limitation of our study is that all patients

did not receive endoscopy and biopsy to diagnose pouchi-tis, since in the late 1980s and early 1990s gastroenterolo-gists frequently made the diagnosis of pouchitis on clinicalgrounds. More detailed diagnostic tools such as the PouchitisDisease Activity Index (PDAI) (20) had not yet been devel-oped. Consequently, there are likely to be patients categorizedand treated as having “mild” pouchitis who may not have hadendoscopic criteria for this diagnosis. However, patients clas-sified as having more significant pouchitis in frequency andseverity in most cases had endoscopic and histologic con-firmation of this diagnosis as was the practice at our center.A second limitation of this analysis is that it was conductedretrospectively.

Approximately 50% of the 102 patients enrolled in ourstudy developed pouchitis within 5 yr. This finding reinforcesresults from other large series from other major centers. Forexample, a study from the Mayo Clinic in 1996 reported thatthe cumulative risk for the first episode of pouchitis was 46%(21). A recent study of 114 patients from Seattle, Washingtonreported that the incidence of pouchitis was 59%, and notedthat the incidence increased with the duration of follow-up(22).

UC disease activity (as opposed to dysplasia) was the indi-cation for surgery in 86% of patients without pouchitis, 82%of patients with 1–2 episodes/yr of pouchitis, and in 94% ofthe patients with >2 episodes/yr. Only 1 of the 14 most sickpatients had undergone colectomy due to dysplasia. Preop-erative UC disease activity was not found to correlate withpouchitis activity.

The UC-specific pANCA prevalence in our overall patientpopulation was 54%, which is consistent with previous ob-servations in UC (23). The UC-specific pANCA prevalencein our study did not correlate with the development or sever-ity of pouchitis. This finding is consistent with other series(24), although some studies have suggested that pANCA maybe a marker for pouchitis (25, 26). These divergent findingsmay reflect differences among study populations, laboratoryassays, or lengths of follow-up.

Are Gene Polymorphisms Associated with Pouchitis? 439

The rate of UC-specific pANCA prevalence in 14 of themost symptomatic pouchitis patients was similar to the ratein the overall study population. The four patients with ongo-ing symptoms despite constant medication were all pANCApositive, a finding that is of uncertain significance.

A recent study by Fleshner et al. concluded that only pre-colectomy high-titer (>100 EU/ml by ELISA) UC-specificpANCA predicts the development of chronic pouchitis af-ter surgery (27). In our study, only one patient was found tohave a high-titer UC-specific pANCA. As noted above, thepatient suffered from mild pouchitis. The paucity of high-titer pANCA in our study population may reflect the fact thatthe patients’ sera were obtained many years postoperatively,and pANCA levels may fall over the long term after colec-tomy (28). The Fleshner study did not report postoperativetiters.

Previous studies have shown elevated titers of ASCA IgGand/or IgA in 50–80% of CD patients, but in <10% of UCpatients and healthy controls (29). Approximately 15% of thepatients in our study were ASCA positive. Our study revealsno association between ASCA and pouchitis, including in thesubgroup of patients with the most refractory pouch inflam-mation. This lack of association confirms previous studies(30) discouraging the hypothesis that pouchitis may signifyoccult CD, or that ASCA is simply a marker of ileal inflam-mation.

IL-1ra is a cytokine that blocks the proinflammatory ac-tivity of IL-1 by competitively binding to IL-1 receptors andpreventing IL-1-induced signal transduction in target cells. Amucosal imbalance of IL-1 and IL-1ra has been described inpatients with UC (31, 32), suggesting that such an imbalancemay contribute to the pathogenesis of inflammatory boweldisease.

The gene-encoding IL-1ra lies on chromosome 2. The pres-ence of different copies of a VNTR of an 86 bp sequence inintron 2 comprises the different alleles of the IL-1ra polymor-phism evaluated in this study (33). Allele 2 (with two repeatsequences) has been associated with increased production ofIL-1ra in vitro (34). While several studies have demonstratedan overrepresentation of allele 2 of IL-1ra in UC (14, 35, 37)others have found no association (35, 36). Explanations forthese discrepancies include ethnic differences, disease het-erogeneity, and the possibility that the IL-1ra gene does notdirectly contribute to genetic susceptibility to UC, but ratherserves as a marker of other closely linked genes on chromo-some 2 of primary importance. The presence of IL-1ra allele2 in 60% of this study population is higher than in the gen-eral population, where carriage rates of 25–50% have beendescribed (35). In this study, carriage of allele 2 appeared tobe negatively associated with the development of pouchitis.Of the 60 patients who were carriers of allele 2, 62% didnot have pouchitis and 38% had pouchitis. This finding is incontrast to reports by Brett et al., and Carter et al., in whichcarriers of IL-1ra allele 2 had a greater chance of developingpouchitis after surgery (4, 38). These authors conclude thatthe allele is associated with a more severe inflammatory re-

sponse, and that this genetic variation may have an effect onthe level of expression of IL-1ra. However, this hypothesis re-mains to be proven, and studies to date have been conflicting.It is unclear why these studies reached opposite conclusions.It could be a result of differences in patient populations orin definitions of pouchitis. Given this important discrepancy,it will be important to evaluate this polymorphism prospec-tively in larger numbers of patients from different geographicregions to conclusively determine whether it predicts suscep-tibility to pouchitis. Furthermore, it will be critical to de-termine whether this polymorphism has functional relevancefor the inflammatory response. In our study, no relation wasfound between the expression of ANCA and IL-1ra allele 2carriage.

TNF and lymphotoxin are potent proinflammatory and im-munomodulatory cytokines. The TNF locus, which containsthe genes for TNF and lymphotoxin is located on chromo-some 6, within the major histocompatibility complex. Ge-netic variations within the TNF locus have been implicatedin the pathogenesis of numerous autoimmune and inflamma-tory diseases, including IBD. Allelic combinations of TNFmicrosatellite markers inherited on one chromosome, knownas haplotypes, were used to describe a specific associationwith CD in a case-control study (15), and subsequently usedto demonstrate transmission in families with CD (39). Fackliset al. described an association between this Crohn’s disease-associated haplotype and a subgroup of medically unrespon-sive patients with UC who are predisposed to the developmentof pouchitis (40).

Two previously described polymorphisms were evaluatedin this study: a polymorphism at position –308 in the TNFgene, and an NcoI RFLP within the first intron of the lympho-toxin gene. The carriage rate of TNF allele 2 in our study pop-ulation was 23%. This percentage was similar to that found ina UC population in previous studies (41). Fifty percent of ourpatients carried lymphotoxin allele 2. This was less than the90% reported carriage rate in healthy controls and the 92%carriage rate for UC patients described by Papo et al. (42).

In our population, carriage of TNF allele 2 appeared to benegatively correlated with pouchitis. Only one of 14 patientswith severe, chronic pouchitis carried this allele. This find-ing may, in fact, reinforce results obtained from a previousanalysis. The study by Facklis et al. had described that the5-allele CD-associated TNF microsatellite haplotype a2, b1,c2, d4, e1 may define a subgroup of UC patients with severe,medically refractory disease who go on to develop chronicpouchitis following IPAA (40). The TNF c microsatellite al-lele 2 (c2) is in close linkage disequilibrium with TNF allele1. Therefore, the extremely low carriage of TNF allele 2 in thesevere pouchitis groups may correlate with the presence ofthis CD-associated TNF microsatellite haplotype. Althoughit is controversial whether the TNF polymorphism may havefunctional significance in terms of TNF promoter activationand gene expression, elucidation of functional activity of thispolymorphism may help determine its possible role in thepathogenesis of pouchitis. Alternatively, this polymorphism

440 Aisenberg et al.

may be a marker for another closely linked gene that is di-rectly involved in pathogenesis.

ACKNOWLEDGMENTS

The authors thank Prometheus Laboratories (San Diego, CA)for performing the analyses of pANCA and ASCA. This studywas supported by a grant from the Burrill B. Crohn ResearchFoundation.

This study was presented in poster form at the DigestiveDisease Week, San Diego, CA, May, 2000.

Reprint requests and correspondence: Scott Plevy, M.D., Asso-ciate Professor of Medicine and Immunology, Division of Gastroen-terology, Hepatology, and Nutrition, Scaife Hall, Room 566, 3550Terrace Street, Pittsburgh, PA 15261.

Received May 21, 2003; accepted October 22, 2003.

REFERENCES

1. Stahlberg D, Gullberg K, Leljequist T, et al. Pouchitis fol-lowing pelvic pouch operation for ulcerative colitis. Inci-dence, cumulative risk, and risk factors. Dis Colon Rectum1996;39:1012–8.

2. Romanos J, Samarasekera DN, Stebbing JF, et al. Outcomeof 200 restorative proctocolectomy operations: the JohnRadcliffe hospital experience. Br J Surg 1997;84:814–8.

3. Salesmans J, Nagengast F, Lubbers E, et al. Post-operativeand long-term results of ileal pouch-anal anastomosis forulcerative colitis and familial polyposis coli. Dig Dis Sci1992;37:1882–9.

4. Brett PM, Yasuda N, Yiannakou JY, et al. Genetic and im-munological markers in pouchitis. Eur J Gastroenterol Hep-atol 1996;8:951–955.

5. Targan SR. The utility of ANCA and ASCA in inflammatorybowel disease. Inflamm Bowel Dis 1999;5:61–63.

6. Sandborn WJ, McLeod R, Jewell DP. Medical therapy forinduction and maintenance of remission in pouchitis: A sys-tematic review. Inflamm Bowel Dis 1999;5:33–39.

7. Ruseler-van Emdeb JG, Schouten WR, van LieshoutLM. Pouchitis: Result of microbial imbalance? Gut1994;35:658–64.

8. Gionchetti P, Rizzello F, Venturi A, et al. Oral bacteriother-apy as maintenance treatment in patients with chronic pou-chitis: A double-blind, placebo-controlled trial. Gastroen-terology 2000;119:305–9.

9. Ruemmels FM, Targan SR, Levy G, et al. Diagnostic ac-curacy of serologic assays in pediatric inflammatory boweldisease. Gastroenterology 1998;115:822–9.

10. Sendid B, Colombel JF, Jacquinot PM, et al. Specific anti-body response to oligomannoside epitopes in Crohn’s dis-ease. Clin Diag Lab Immunol 1996;3:219–26.

11. Plevy SE, Targan SR, Yang H, et al. Tumor necrosis factormicrosatellites define a Crohn’s disease-associated haplo-type on chromosome 6. Gastroenterology 1996;110:1053–60.

12. Vidrich A, Lee J, James E, et al. Segregation of pANCAantigenic recognition by DNase treatment of neutrophils:Ulcerative colitis, type 1 autoimmune hepatitis, and primarysclerosing cholangitis. J Clin Immunol 1995;15:293–9.

13. Quinton JF, Sendid B, Reumaux D, et al. Anti-saccharomyces cerevisiae mannan antibodies combinedwith anti-neutrophil cytoplasmic autoantibodies in inflam-

matory bowel disease: Prevalence and diagnostic role. Gut1998;42:788–91.

14. Tountas NA, Casini-Raggi V, Yang H, et al. Functional andethnic association of allele 2 of the interleukin-1 receptorantagonist gene in UC. Gastroenterology 1999;117:806–13.

15. Plevy SE, Targan SR, Yang H, et al. Tumor necrosis factormicrosatellites define a Crohn’s disease-associated haplo-type on chromosome 6. Gastroenterology 1996;110:1053–60.

16. SPSS 9.0 for Windows, Chicago, IL.17. StatXact 4 for Windows, Cytel Software Corporation, Cam-

bridge, MA.18. Sandborn WJ. Pouchitis: Definition, risk factors, frequency,

natural history, classification, and public health perspective.Trends Inflamm Bowel Dis Ther 1996;51–63.

19. Svaninger G, Nordgren S, Oresland T, et al. Incidence andcharacteristics of pouchitis in the Kock continent ileostomyand the pelvic pouch. Scand J Gastroenterol 1993;28:695–700.

20. Sandborn WJ, Tremaine WJ, Batts, KP, et al. Pouchitis afterileal pouch-anal anastomosis: A pouchitis disease activityindex. Mayo Clinic Proceedings 1994;69:409–15.

21. Penna C, Dozois R, Tremaine W, et al. Pouchitis after ilealpouch-anal anastomosis for ulcerative colitis occurs with in-creased frequency in patients with associated primary scle-rosing cholangitis. Gut 1996;38:234–39.

22. Simchuk EJ, Thirlby RC. Risk factors and true incidenceof pouchitis in patients after ileal pouch-anal anastomoses.World J Surg 2000;24:851–6.

23. Rutgeerts P, Vermiere S. Clinical value of the detec-tion of antibodies in the serum from diagnosis and treat-ment of inflammatory bowel disease. Gastroenterology1998;115:1006–9.

24. Yasuda N, Thomas P, Ellis H, et al. Perinulcear anti-neutrophil cytoplasmic antibodies in ulcerative colitis afterrestorative proctocolectomy do not correlate with the pres-ence of pouchitis. Scand J Gastroenterol 1998;33:509–13.

25. Sandborn WJ, Landers CJ, Tremaine WJ, et al. Antineu-trophil cytoplasmic antibody correlates with chronic pou-chitis after ileal pouch-anal anastomosis. Am J Gastroeterol1995;5:740–7.

26. Panaccione R, Sandborn WJ, Tremaine WJ, et al. pANCA isassociated with chronic pouchitis and predicts the future oc-currence of acute pouchitis following ileal pouch-anal anas-tomosis (IPAA). Gastroenterology 1999;116:A793.

27. Fleshner PR, Vasiliauskas EA, Kam LY, et al. High levelperinuclear antineutrophil cytoplasmic antibody (pANCA)in ulcerative colitis patients before colectomy predicts thedevelopment of chronic pouchitis after ileal pouch anal anas-tomosis. Gastroenterology 1999;116:A716.

28. Fleshner PR, personal communication.29. Nielsen OH, Vainer B, Madsen SM, et al. Established and

emerging biological activity markers of inflammatory boweldisease. Am J Gastroenterol 2000;95:359–67.

30. Aisenberg J, Wagreich J, Shim J, et al. Perinuclear anti-neutrophil cytoplasmic antibody and refractory pouchitis:A case-control study. Dig Dis Sci 1995;9:1866–72.

31. Casini-Raggi V, Kam L, Choung YJ, et al. Mucosal imbal-ance of IL-1 and IL-1 receptor antagonist in inflammatorybowel disease. A novel mechanism of chronic intestinal in-flammation. J Immunol 1995;154:2434–40.

32. Cominelli F, Pizarro TT. Interleukin-1 and interleukin-1 re-ceptor antagonist in inflammatory bowel disease. AlimentPharmacol Therap 1996;10(suppl 2):49–53.

33. Tarlow JK, Blakermore AIF, Lennard A, et al. Polymor-phism in human IL-1 receptor antagonist gene intron 2 is

Are Gene Polymorphisms Associated with Pouchitis? 441

caused by variable numbers of an 86 BP tandem repeat.Hum Genet 1993;91:403–4.

34. Danis VA, Millington M, Hyland V, et al. Cytokine produc-tion by normal human monocytes—Inter-subject variationand relationship to an IL-1 receptor antagonist (IL-1RA)gene polymorphism. Clin Exp Immunol 1995;99:303–10.

35. Bioque G, Bouma G, Crusius JB, et al. Evidence of geneticheterogeneity in IBD: The interleukin-1 receptor antagonistin the predisposition to suffer from ulcerative colitis. Eur JGastroenterol Hepatol 1996;8:105–110.

36. Hacker UT, Bidlingmaier C, Gomolka M, et al. Inflamma-tory bowel disease: No association between allele combi-nations of the interleukin (IL) I beta and IL-1 receptor an-tagonist gene polymorphisms. Eur J Clin Invest 1998;28:214–9.

37. Mansfield JC, Holden H, Tarlow JK, et al. Novel geneticassociation between UC and the anti-inflammatory cytokineIL-1ra. Gastroenterology 1994;106:637–42.

38. Carter MJ, Di Giovine FS, Cox A, et al. The interleukin1 receptor antagonist gene allele 2 as a predictor of pou-chitis following colectomy and IPAA in ulcerative colitis.Gastroenterology 2001;121:805–11.

39. Yang H, Plevy SE, Taylor K, et al. Linkage of Crohn’s diseaseto the major histocompatibility complex region is detectedby multiple non-parametric analyses. Gut 1999;44:519–26.

40. Facklis K, Plevy SE, Vasiliauskas ES, et al. Crohn’s disease-associated genetic marker is seen in medically unresponsiveulcerative colitis patients an may be associated with pouch-specific complications. Dis Colon Rectum 1999;42:601–5.

41. Louis E, Satsangi J, Roussomoustakaki M, et al. Cytokinegene polymorphisms in inflammatory bowel disease. Gut1996;39:705–10.

42. Papo M, Quer JC, Gutierrez C, et al. Genetic heterogeneitywithin UC determined by an interleukin-1 receptor antag-onist gene polymorphism and antineutrophil cytoplasmicantibodies. Eur J Gastroenterol Hepatol 1999;11:413–20.