• Crohn's disease;
  • serologic markers;
  • fecal diversion


  1. Top of page
  2. Abstract
  6. References

Objectives: Fecal diversion is occasionally indicated in patients with advanced perianal or colorectal Crohn's disease (CD). Because CD may result from an aberrant immunologic response to bacteria within the gut lumen, fecal diversion should be effective in managing these complications. However, not all patients achieve a clinical response after fecal diversion. CD patients can be characterized by their antibody responses against Pseudomonas fluorescens (I2), E.coli outer membrane porin C (OmpC), oligomannan (anti-Saccharomyces cerevisiae antibodies [ASCA]), and antinuclear antigens (perinuclear antineutrophil cytoplasmic antibodies [pANCA]). This study examines the association between clinical features and seroreactivity to these microbial and auto-antigens in predicting a clinical response to fecal diversion. Methods: Twenty-seven consecutive CD patients undergoing fecal diversion were included. Sera were drawn and tested for anti-I2, anti-OmpC, ASCA, and pANCA in a blinded fashion. Response was assessed using clinical parameters. Results: Seventeen (63%) patients underwent fecal diversion for medically resistant proctocolitis and 10 (37%) for severe perianal disease. Median follow-up was 41 months. Seventeen (63%) patients achieved a clinical response. No preoperative clinical or surgical factor predicted response to diversion. Clinical response after fecal diversion was seen in 15 of 16 (94%) patients who were I2 positive compared with only 2 of 11 (18%) patients who were I2 negative (P = 0.0001). Seroreactivity to OmpC, ASCA, or pANCA was not associated with a clinical response to diversion. Conclusion: Expression of I2 antibodies against a bacterial antigen of Pseudomonas fluorescens was highly associated with clinical response to fecal diversion in CD patients.

Indications for fecal diversion in the surgical management of colorectal and perianal Crohn's disease (CD) remain incompletely defined. Although previously considered as a procedure performed in patients too debilitated to undergo a major resection or to facilitate a limited resection in patients with diffuse colorectal disease,1,2 fecal diversion is now used selectively in cases of Crohn's colitis refractory to medical treatment,3 to allow healing of perianal lesions,4 and to avoid growth retardation in children with Crohn's colitis requiring steroids.5 The rationale for using fecal diversion to manage these complications rests on observations that CD may result from an aberrant immunologic response to bacteria within the gut lumen.6–8 However, fecal diversion for perianal or colorectal CD is not uniformly successful. Success rates in prior reports of fecal diversion for CD vary between 20% and 100%.1,3,9 Unfortunately, it has not been possible to preoperatively identify those CD patients who will respond to fecal diversion and those who will continue to be symptomatic.10

Serologic immune markers have been the subject of extensive research in patients with inflammatory bowel disease (IBD). Perinuclear antineutrophil cytoplasmic antibodies (pANCA) are found in 60% to 80% of ulcerative colitis (UC) patients.11–13 pANCA is also present in 15% to 25% of patients with CD, where it is associated with a “UC-like” clinical picture.14 Anti-Saccharomyces cerevisiae antibodies (ASCA) are found in approximately 60% of CD patients.15,16 Although serum pANCA and ASCA are the best studied serologic markers for IBD, antibodies against the outer membrane porin C (anti-OmpC) of Escherichia coli and a CD-related protein from Pseudomonas fluorescens (anti-I2) are also found commonly in 50% of patients with CD and to a lesser degree in UC patients.17–19 Our group has previously shown that CD patients can be classified into specific “immunologic” subgroups on the basis of their antibody response patterns to these microbial and auto-antigens.19

We hypothesized that CD patients with immune responses to microbial antigens would respond best to fecal diversion. This prospective study examined a group of CD patients undergoing fecal diversion to investigate the association between preoperative clinical and serologic factors and the response to fecal diversion. Our study shows that the only preoperative variable associated with a clinical response to fecal diversion in CD is the expression of anti-I2 antibodies.


  1. Top of page
  2. Abstract
  6. References

The clinical outcome of fecal diversion in CD patients was prospectively assessed at Cedars-Sinai Medical Center over the 10 year period ending October 2003. All research-related activities were approved by the Cedars-Sinai Medical Center Institutional Review Board (IRB Study # 3358-03b). The diagnosis of CD was based on standard clinical, endoscopic, histologic, and radiographic features.

Preoperative and operative clinical variables were tabulated, including patient age and sex, duration and site(s) of disease, presence of extraintestinal manifestations of CD, preoperative use of steroids or other immunomodulators, family history of IBD, smoking history, and indication for operation. Serum indices of disease, including albumin, hematocrit, white blood cell (WBC) count, bands, platelets, and erythrocyte sedimentation rate (ESR) were also recorded. Operative details recorded included type of fecal diversion, operative technique of stoma creation, and whether any synchronous bowel resection was performed.

Study patient sera were drawn immediately before surgery, coded, and stored for future analysis by investigators blinded to the clinical response to fecal diversion. Analysis of pANCA and immunoglobulin (Ig)G and IgA ASCA were performed at Cedars-Sinai Medical Center or Prometheus Laboratories (San Diego, CA). All assays for anti-OmpC and anti-I2 were performed at Cedars-Sinai Medical Center. Antibody levels were determined and results expressed as enzyme-linked immunosorbent assay (ELISA) units (EU/mL). Some of the patients in this cohort had serologic assessment at the time of surgery and once at least 6 months after surgery.

Serum ANCA Determination and Subtype Characterization

Serum ANCA expression and ANCA subtype characterization were assessed using ELISA and indirect immunofluorescence, as previously described.11 Sera with circulating anti-neutrophil cytoplasmic IgG antibody exceeding the normal reference range value were termed “ANCA+”. Numeric values that were below the normal reference range were termed “ANCA−”. ANCA+ sera were further subtyped by way of indirect immunofluorescence staining to determine the ANCA neutrophil binding pattern. Sera exhibiting perinuclear highlighting that then lost this characteristic staining pattern with DNAse treatment were termed “pANCA+”.

Serum ASCA Determination

Serum samples were analyzed for ASCA expression by ELISA, as previously described.15,16 Serum expressing ASCA reactivity (IgG or IgA) exceeding the normal reference range (IgA greater than 20 EU/mL or IgG greater than 40 EU/mL) were termed “ASCA+”.

Serum Anti-I2 and Anti-OmpC Antibody Measurement

Human IgA antibodies that bind OmpC or I2 were assessed by direct ELISA, as previously described.17,18 Microtiter plates (Greiner, USA Scientific, Ocala, FL) were coated with glutathione-S-transferase (GST) alone and I2-GST (5 μg/mL) or OmpC (0.25 μg/mL) in borate buffered saline, pH 8.5. Unbound antigen was washed away, and then nonspecific binding was blocked with bovine serum albumin (0.25%) in phosphate-buffered saline. Next, control and coded sera were added at a 1:100 dilution, followed by detection with alkaline phosphatase conjugated goat anti-human IgA (alpha-chain specific, Jackson ImmunoResearch, West Grove, PA) at a dilution of 1:1000. Substrate solution containing P-nitrophenol phosphate was then added, and color was allowed to develop for 1 hour, at which time, the plates were read at 405 nm. Nonspecific binding of sera to GST alone (typically < 0.1) were subtracted from raw values of I2-GST binding to obtain I2 specific absorbances. Levels were determined relative to a standard consisting of serum obtained from a well-characterized OmpC+ or I2+ CD patient. Patients with IgA for anti-OmpC greater than 20.9 EU/mL were termed “OmpC+”, whereas patients with IgA for anti-I2 greater than 15.8 EU/mL were termed “I2+”.

Evaluation of Clinical Response

Assessment of clinical response to diversion was primarily determined based on subjective improvement in the patient's overall medical condition. Particular attention was focused on improvements in urgency, abdominal and perineal pain, decreased anal fistula drainage, and weight gain. Determination of clinical response was performed by an experienced clinical investigator (P.R.F.) blinded to the results of the patients' serologic analysis. Endoscopic response to diversion was compared with preoperative mucosal findings and was assessed as improved, unchanged, or worse. Laboratory indices evaluating disease activity were drawn at the time of surgery and 3 months after fecal diversion.

Statistical Analysis

Chi-square or Fisher's exact tests (if expected cell counts were less than 5) were used to determine the association of categorical clinical variables and antibody positivity with response to fecal diversion. For continuous covariates, medians were compared with the use of the nonparametric Wilcoxon's test. A multiple logistic regression was used to determine the primary association between response to fecal diversion and quantitative antibody levels for IgA ASCA, IgG ASCA, OmpC, pANCA, and I2. Nonparametric paired t tests were performed to examine whether the difference between preoperative and postoperative antibody levels was significantly different from 0. All statistical analyses were performed using the Statistical Analysis Software 8.02 (SAS Institute, Inc., Cary, NC) for Windows.


  1. Top of page
  2. Abstract
  6. References

Study Group Characteristics

Twenty-seven patients comprised the study group. Patient demographics and clinical characteristics are shown in Table 1. A majority of patients were male. Twenty-two (81%) patients were being treated with systemic steroids at the time of diversion. Immunomodulators were being used in most patients and included 6-mercaptopurine (22 patients), cyclosporine (11 patients), infliximab (10 patients), and methotrexate (7 patients). Four (15%) patients had extraintestinal manifestations of their disease, two patients with arthritis, one patient with erythema nodosum, and one patient with both ocular disease and arthritis.

Table TABLE 1. Study Group Characteristics
Clinical Variablen
  1. IBD = Inflammatory bowel disease.

Median age (yr)43
Median duration of disease (mo)67
Preoperative medication (%) 
Family history of IBD2
Cigarette smoking7
Extraintestinal disease (%)4
Surgical Indication 
 Medically refractory disease17
 Perianal disease10

Indications for surgery and the different types of fecal diversion are shown in Figure 1. Seventeen of the 27 (63%) study patients underwent diversion for medically refractory colorectal disease. Fourteen of these 17 (82%) patients had refractory proctocolitis, and the remaining 3 (18%) patients had resistant proctitis. Nine of the 27 (33%) study group patients underwent a synchronous bowel resection. Six (22%) patients with refractory proctocolitis had a synchronous bowel resection, including five patients having an abdominal colectomy and one patient having a sigmoid colectomy. Of the 10 patients undergoing diversion for perianal disease, 3 had a synchronous bowel resection (ileocolic resection, 1; small bowel resection, 1; left hemicolectomy, 1).

thumbnail image

Figure FIGURE 1. Flowchart showing indications for surgery and different forms of fecal diversion.

Download figure to PowerPoint

Fecal diversion was accomplished using an ileostomy in 17 (63%) patients and a colostomy in 10 (37%) patients. Although the majority of stomas were made using a laparotomy, 10 (37%) stomas were constructed using laparoscopic techniques. There were 10 loop-ileostomies and 7 end-ileostomies. Seven of the 10 colostomies were loop in nature, and 3 were end-stomas.

Response to Fecal Diversion

There was complete follow-up in 20 (74%) patients. Five (19%) patients were lost to follow-up after a median period of 27 (range, 6-46) months after surgery. Two additional patients died during follow-up period. One patient died of lymphoma 35 months after fecal diversion, and the other died of a myocardial infarction 37 months after surgery.

Median follow-up was 41 (range, 6-122) months. Clinical response was achieved in 17 of the 27 (63%) study patients. Median time to response was 1 (range, 1-2) month. All responders were successfully weaned off steroids. Seventeen of the 27 (63%) study group patients underwent endoscopic evaluation of the colon or rectum after diversion. Endoscopic response was considered improved in eight (47%) patients, unchanged in eight (47%) patients, and worse in one (6%) patient. Although endoscopic improvement was observed more commonly in the group of patients who responded clinically (60%) than in the group of patients who did not respond clinically (28%), this trend was not statistically significant (P = 0.23).

Laboratory characteristics of the responding and nonresponding patient groups before and after surgery are shown in Figure 2. These patients were markedly ill before surgery, as reflected by marked hypoalbuminemia, anemia, bandemia, and high ESR. Fecal diversion was accompanied by an improvement in these indices of chronic disease. Serum albumin, hematocrit, and ESR improved significantly in the responder group after fecal diversion. Although there were trends seen in improvements in platelet count (P = 0.08), WBC count (P = 0.94), and bandemia (P = 0.25), these changes did not reach statistical significance.

thumbnail image

Figure FIGURE 2. Laboratory findings before and after surgery in group of patients who responded to fecal diversion (-) and group of patients who did not respond to fecal diversion (- - - -). Changes in albumin (P = 0.004), hematocrit (P = 0.0005), and erythrocyte sedimentation rate (ESR) (P = 0.01) were all significant in the responder group. ESR was the only significant change (P = 0.03) after surgery noted in nonresponder group.

Download figure to PowerPoint

None of the 17 patients who initially responded to fecal diversion had disease recurrence after fecal diversion. Two (12%) of these patients however required re-operation, one for small bowel obstruction and the other for repair of an incarcerated parastomal hernia. Six of the seven (86%) patients who did not achieve initial clinical response to fecal diversion had additional surgery, including four patients who underwent total proctocolectomy and two who required a number of local anorectal procedures to control perianal sepsis.

Serologic Analysis

Initially, we determined the serum reactivity to each of these antigens in our cohort of CD patients. Figure 3 shows the scatter graph of the serologic responses for each antigen in this cohort. Serum anti-I2, anti-OmpC, ASCA, and pANCA were detected in 16 (59%) patients, 13 (48%) patients, 7 (27%) patients, and 9 (33%) patients, respectively. There were no significant differences in clinical characteristics in patients who were serologically positive compared with patients who did not express these markers (data not shown).

thumbnail image

Figure FIGURE 3. Scatter graphs of level of serum reactivity toward microbial and auto-antigens in this patient cohort. In each panel, gray zone indicates negative values. White bullets represent nonresponders to fecal diversion, and black bullets represent responders to fecal diversion.

Download figure to PowerPoint

Clinical, Surgical, and Serologic Factors Predicting Response to Fecal Diversion

The association between various clinical and surgical features and response to fecal diversion is shown in Tables 2 and 3. In confirmation of the results of another study,10 no preoperative clinical or surgical factor was associated with a clinical response to fecal diversion in these patients.

Table 2. Preoperative Clinical Factors were not Associated with Response to Fecal Diversion
FactorTotal No.Responders, n(%)Nonresponders, n(%)
  1. P = not significant for all.

 Male1712 (71)5 (29)
 Female105 (50)5 (50)
Median age (yr) 4436
Median duration of 7252
disease (mo)   
Preoperative medication   
 Steroids2213 (59)9 (41)
 Immunomodulators2314 (61)9 (39)
Cigarette smoking75 (71)2 (29)
Extraintestinal disease42 (50)2 (50)
Table 3. Surgical Factors were not Associated with Clinical Response to Fecal Diversion
FactorTotal No.Responders, n (%)Nonresponders, n (%)
  1. P = not significant for all.

Surgical indication   
 Medically refractory179 (53)8 (47)
 Perianal disease108 (80)2 (20)
Type of diversion   
 Ileostomy1711 (65)6 (35)
 Colostomy106 (60)4 (40)
Synchronous bowel96 (67)3 (33)

The interaction between serologic marker expression and response to fecal diversion is shown in Figure 4. Clinical response after fecal diversion was seen in 15 of 16 (94%) patients who were I2+ compared with 2 of 11 patients (18%) who were I2 negative. This difference in response based on I2 serologic expression was highly significant (P = 0.0001). However, expression of anti-OmpC, ASCA, or pANCA was not predictive of clinical response to diversion (P = 0.23, P = 0.67, and P = 0.48, respectively). The association between I2 expression and response to fecal diversion was observed in patients who required surgery for refractory proctocolitis. Although this trend was also observed in patients with perianal disease, this trend was not statistically significant (Fig. 5). The sensitivity of I2 expression in predicting a clinical response to fecal diversion was 88% (95% confidence interval [CI], 66-98%) and the specificity was 90% (95% CI, 60-99%).

thumbnail image

Figure FIGURE 4. Response to fecal diversion based on serology status. Response to fecal diversion was significantly higher in patients who were I2 positive compared with patients who were I2 negative (*P < 0.0001 compared with I2 negative).

Download figure to PowerPoint

thumbnail image

Figure FIGURE 5. Response to fecal diversion based on I2 expression in patients with refractory proctocolitis (RP) or perianal disease (PAD). In RP patients, response to fecal diversion was significantly higher in patients who were I2 positive compared with patients who were I2 negative (*P < 0.001 compared with I2 negative). Although this trend was also observed in PAD patients, it did not reach statistical significance (P = 0.1).

Download figure to PowerPoint

Although there appears to be an association between qualitative antibody responses toward I2 and the response to fecal diversion, we then examined the association between the level of antibody response to microbial and auto-antigens and the response to fecal diversion. Logistic regression multivariate analysis demonstrated a significant correlation between quantitative level of serum I2 and likelihood of response to fecal diversion (Table 4). This relationship was not found to be significant with the other three serologic markers.

Table 4. Association between Quantitative Level of Serologic Markers and Response to Fecal Diversion
Serologic MarkerSignificance Level (P)
  1. ASCA = anti-Saccharomyces cerevisiae antibodies; Ig = immunoglobulin; pANCA = perinuclear antineutrophil cytoplasmic antibodies; OmpC = outer membrane porin C.

ASCA IgA0.44
ASCA IgG0.63

Effects of Fecal Diversion on Microbial and Auto-Antigen Expression

In an effort to determine the influence of surgery on these antibody responses, we analyzed the level of the antibody responses toward microbial and auto-antigens in 14 (52%) patients after surgery. As demonstrated in Figure 6, there were no significant changes in antibody responses toward any microbial or auto-antigen after surgery.

thumbnail image

Figure FIGURE 6. Changes in serologic responses to microbial and auto-antigens after fecal diversion in Crohn's disease. In each panel, gray zone indicates negative values.

Download figure to PowerPoint


  1. Top of page
  2. Abstract
  6. References

In selected CD patients with severe colorectal or perianal disease, fecal diversion can be used to promote mucosal healing and resolution of perianal disease. More importantly, it might improve a patient's clinical symptoms and provide a general sense of “well-being”. These attractive outcomes must be tempered by the fact that not all patients having a diverting stoma improve after surgery.1,3,9 Unfortunately, there does not appear to be a preoperative factor or factors that can predict a response to fecal diversion.10 In this report, however, we found that preoperative expression of the serologic marker anti-I2 was associated with clinical response to fecal diversion in CD patients with medically refractory disease.

Surgical options in patients with severe CD involving the colon alone include segmental resection with colocolonic anastomosis or total abdominal colectomy with ileorectal anastomosis.9 However, surgical approaches to patients with Crohn's colitis also involving the rectum are limited and usually involve a colostomy or ileostomy. Because the role of surgery in CD is to control symptoms while preserving the bowel, fecal diversion is an attractive alternative to conventional resection in selected CD patients. The use of diversion is indicated to achieve mucosal healing in patients who have failed medical treatment, facilitate limited resection in severely debilitated patients or in the presence of diffuse disease, and to avoid growth retardation in children with diffuse colitis requiring steroids.1–4 Patients with severe perianal disease who have failed local anorectal procedures are also candidates for this procedure.3 An additional indication for fecal diversion is to enable the patient with severe disease needing a permanent stoma to adjust to life with an ileostomy or a colostomy.

The rate of clinical response to diversion in our study (63%) is comparable with that previously reported in the literature (Table 5). A majority of patients had subjective improvement of their disease, and many of the objective parameters of disease activity we measured also improved significantly. Judging the effectiveness of operative interventions in CD can be difficult and can be based on subjective criteria or objective parameters such as laboratory, endoscopic, or histologic findings. For a number of reasons, in this study, we chose to use subjective criteria to assess response to fecal diversion. Subjective methods have been the most common to assess response in prior reports of fecal diversion.1-4,21,31,33,34 In addition, these observations were clinically relevant because they were collected by an experienced surgical investigator. Furthermore, laboratory indices of disease activity do not always improve with clinical improvement, as demonstrated previously by others2,21,31 and as confirmed by our data. Finally, in many instances, it is virtually impossible when assessing CD patients to know whether endoscopic or histologic changes in the defunctionalized colon and rectum are caused by ongoing CD or diversion colitis.32,35,36 In fact, many patients who appear to be well clinically have ongoing inflammation in the diverted CD segment.3,21,35 Our study is further strengthened by noting that assessment of response was performed prospectively and was blinded to the results of the patient's serologic profile. Studies to assess response to fecal diversion using validated quality of life instruments are underway in our center.

Table 5. Reported Effectiveness of Fecal Diversion in Crohn's Disease
First Author [Reference]YearnLocationPercent of Response
  1. PC = refractory proctocolitis; PA = refractory perianal disease.

Howel-Jones20196621PC, PA10
Oberhelman4196813PC, PA100
Burman21197129PC, PA97
McIlrath3197113PC, PA92
Lee5197542PC, PA86
Zelas2198079PC, PA91
Harper241982102PC, PA93
van Dongen2819869PA44
Winslet31199344PC, PA71
Edwards32200073PC, PA86
Current series200427PC, PA63

There are a number of clinical and animal studies suggesting the importance of the fecal stream in the induction and propagation of CD. Intestinal lesions predominate in distal parts of the gastrointestinal tract where the intestinal microflora is most abundant. Changing the fecal stream using either an elemental diet with or without nonabsorbable oral antibiotics37,38 or using total parenteral nutrition39,40 have been shown to be effective in some patients with CD. In addition, many CD patients improve using oral antibiotic or probiotic therapy.41–43 In surgical patients, diversion of the fecal stream can prevent recurrence of CD after terminal ileal resection.44 In addition, treatment with metronidazole for 3 months after resection halves the relapse rate of CD at 1 year after ileal resection.45 Surgical diversion of the fecal stream (even without concomitant resection) has been successful in obtaining at least an early response in many, but not all, patients.5,9 Furthermore, reintroduction of small bowel effluent into the colon of patients with Crohn's colitis treated by split ileostomy can induce inflammation of the isolated bowel segment.46,47 Animal studies have shown that specific cytokine knock-out mice develop CD when their bowel becomes colonized with the normal bacterial flora but do not do so when reared in a germ-free environment.6 Although these studies clearly suggest a role for one or more of the components of the fecal stream in the maintenance and possible pathogenesis of the inflammatory process in CD, the factor(s) in the fecal stream responsible for these observations remains elusive.46,47

Expression of antibodies has been an area of active research in IBD. ASCA represents an antibody against mannose epitopes from the yeast Saccharomyces cerevisiae and is found in 60% of CD patients.15 Although more commonly found in UC, pANCA is detected in approximately 20% of CD patients.15 A study of a large number of CD patients from our institution revealed that ASCA was directly associated and pANCA was inversely associated with the presence of small bowel disease.48 Most of the patients in this study cohort had colonic disease, which accounts for the relatively high observed incidence of reactivity to pANCA (33%) and relatively low incidence of ASCA reactivity (27%).

I2 is a sequence of microbial DNA found in lamina propria of mononuclear cells isolated from lesions of active CD and is not present in noninflammed areas of the same patient.18 Further investigation has revealed that this I2 sequence is found in Pseudomonas fluorescens, a normal commensal bacterium in the gastrointestinal tract.49 Antigens of other commensal organisms, including Escherichia coli and streptococcal species, have also been identified in active CD.50 Commensal bacteria are found in most humans and appear to be important in the pathogenesis of CD. Rodents that are genetically susceptible to the development of IBD, when raised under germ-free conditions, no longer develop colitis. When the commensal flora is reintroduced, these mice again develop colitis.51 In addition, CD4 T cells reactive with antigens of enteric bacteria produce mucosal inflammation similar to CD after adoptive transfer to immunodeficient hosts.52 The results of our study revealed that an immune response to a normal inhabitant of the intestinal tract is associated with a unique clinical phenotype and further suggest that antigens of commensal bacteria can initiate and perpetuate IBD.

Antibodies to a genetic component of Pseudomonas fluorescens (anti-I2) are much more common in CD than UC or control inflammatory conditions.18 Although the I2 sequence is expressed in the normal ileum, it is also found in high concentrations in the colon of CD patients, particularly in patients with active Crohn's colitis. In addition, I2 expression has not been correlated with the development of perianal disease in CD.18 These observations might explain why patients with refractory colorectal disease (but not perianal disease) who were anti-I2+ had a higher response to fecal diversion than patients who did not express antibody to I2.

Because the clinical response to fecal diversion is not universal, patients with severe colorectal or perianal CD requiring fecal diversion face the daunting possibility of not improving after having stoma surgery. Factors that could predict the outcome of diversion surgery would be of great clinical value. Unfortunately, our data showed that no preoperative clinical factor was predictive of response, confirming the results of previous studies.10 However, serologic expression of anti-I2 was highly associated with a response. Interestingly, expression of antibodies to other microbial and auto-antigens was not associated with a clinical response. This observation suggests that an immunologic response to selective microbial antigens, and not enhanced global immunoreactivity, is important in defining subsets of CD characterized by anatomic location of the disease.


  1. Top of page
  2. Abstract
  6. References