SEARCH

SEARCH BY CITATION

Keywords:

  • Crohn's disease;
  • inflammatory bowel disease;
  • child;
  • phenotype;
  • genes;
  • colitis;
  • NOD2

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Background: Pediatric onset Crohn's disease (CD) is associated with more colitis and less ileitis compared with adult onset CD. Differences in disease site by age may suggest a different genotype, or different host responses such as decreased ileal susceptibility or increased susceptibility of the colon.

Methods: We evaluated 721 pediatric onset CD patients from 3 cohorts with a high allele frequency of NOD2/CARD15 mutations. Children with isolated upper intestinal disease were excluded. The remaining 678 patients were evaluated for interactions between age of onset, NOD2/CARD15, and disease location.

Results: We found an age-related tendency for isolated colitis. Among pediatric onset patients without NOD2/CARD15 mutations, colitis without ileal involvement was significantly more common in first-decade onset patients (P = 4.57 × 10−5, odds ratio [OR] 2.76, 95% confidence interval [CI] 1.72–4.43). This was not true for colonic disease with ileal involvement (P = 0.35), or for isolated colitis in patients with NOD2/CARD15 mutations (P = 0.61). Analysis of 229 patients with ileal or ileocolonic disease and a NOD2/CARD15 mutation disclosed that ileocolitis was more prevalent through age 10, while isolated ileitis was more prevalent above age 10 (P = 0.016). NOD2/CARD15 mutations were not associated with age of onset.

Conclusions: In early-onset pediatric CD, children with NOD2/CARD15 mutations demonstrate more ileocolitis and less isolated ileitis. Young children without NOD2/CARD15 mutations have an isolated colonic disease distribution, suggesting that this phenotype is associated with genes that lead to a specific phenotype of early-onset disease.

(Inflamm Bowel Dis 2007)

C rohn's disease (CD) and ulcerative colitis (UC) are collectively known as inflammatory bowel disease (IBD) and are thought to result from the combined effects of environmental agents and genetic susceptibility.1 Approximately 10%–20% of all IBD will present in childhood or adolescence.2 Currently, it is unclear why individuals develop IBD at varying ages (ranging from first decade of life to the elderly), or if pediatric IBD differs from adult onset IBD. Identifying subsets of patients with characteristic phenotypes is important in understanding the pathogenesis of the disease, which may vary among different subsets, and may have future therapeutic implications. It is therefore crucial to understand what dictates age of onset (AOO) of the disease and disease phenotype.

Age of onset could be a random event. However, recent data have shown that a subgroup of patients with early-onset IBD may have specific phenotypes that differ from adult onset IBD,3–6 thus suggesting that the pathogenesis of subsets of pediatric IBD and adult IBD may differ. The most consistent difference in phenotype is in disease distribution. Recent reports suggest that compared to adult CD, first-decade pediatric onset CD is associated with more colitis and less ileitis.3–8

The NOD2/CARD15 gene on chromosome 16 was the first susceptibility gene identified in patients with CD.9–13 Furthermore, patients carrying 1 of 3 well-described mutations are likely to have ileal disease.5, 14, 16, 18–19

The effect of NOD2/CARD15 mutations on AOO has been controversial, with some studies showing an effect toward a younger AOO,14, 18, 20 and others showing no effect.5, 21–23

Several theories have been raised to explain early pediatric onset, colitis-predominant disease in cohorts with a high frequency of NOD2/CARD15 mutations. These include a difference in genotype (genes predisposing to colitis only predispose to early onset), decreased ileal susceptibility in younger patients, or increased (genotype-independent) colonic susceptibility in early-onset CD. Meinzer et al7 proposed that pediatric onset colitis-predominant disease was due to a maturational delay in the development of ileal lymphoid follicles necessary for development of ileal disease, which increases with age, resulting in decreased ileal susceptibility (less involvement of the ileum) among younger children. Leshinsky-Silver et al23 proposed that NOD2/CARD15 mutations predispose an individual to disease onset primarily in the second–fourth decades, thus explaining earlier mean age of disease onset for adults with these mutations, but not for pediatric populations. Furthermore, they found that the NOD2/CARD15 mutations were less prevalent in the first decade than in subsequent decades, suggesting that first-decade onset CD may be associated with a different genetic susceptibility profile. Heterogeneity in NOD2/CARD15 genotype may add to the confusion in different studies (since low allele frequency populations are more likely to have more colitis).21, 24

In order to further our understanding of age-related phenotypes in CD, we attempted to determine if the age-related colitis pattern noticed in younger children with CD is determined by genotype, or by differences in colonic (increased) or ileal (decreased) susceptibility, or due to an age-related variability in NOD2/CARD15 mutations. We hypothesized that early onset confined to the colon is a specific phenotype that is more likely to present at an earlier age due to susceptibility genes other than NOD2/CARD15.

In order to do so, we analyzed a large cohort of children with a high allele frequency of NOD2/CARD15 mutations. We hypothesized that if increased colonic susceptibility was age- but not genotype-related, we would see an age-related effect on colonic involvement in patients with and without NOD2/CARD15 mutations (colonic involvement defined as isolated colitis or ileocolitis). In contrast, if decreased ileal susceptibility is the cause for more colonic involvement in early-onset CD, one would expect that children with NOD2/CARD15 mutations would have an age-related association with ileal involvement (i.e., less ileal and ileocolonic disease in the younger age group, more ileal and ileal colonic in the older age group).

Lastly, if a different genetic susceptibility profile is the cause for more colitis in early-pediatric onset CD, one would anticipate that differences in the prevalence of isolated colitis would be age-related and occur in the patients without NOD2/CARD15 mutations, while colonic involvement would not be affected by age.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patients were recruited from 3 well-defined and characterized pediatric onset CD (<18 years) cohorts. The methods and criteria for diagnosis (based on well-defined endoscopic, radiologic, and histologic criteria) and phenotyping of CD have been previously published.5, 8, 25–26 Patients with UC or indeterminate colitis were excluded. Crohn's colitis was evident from colonoscopy and differentiated from UC by endoscopic features (noncontiguous disease characterized by serpiginous or aphthous ulcerations) or by histological features (granulomas, focal, or patchy submucosal chronic inflammation with preserved crypt architecture and goblet cells). The pediatric onset CD cohorts were collected through tertiary care centers in the USA, Israel, and Italy. These included: 1) Wisconsin, USA (n = 311); 2) Israel (n = 210); and 3) Italy (n = 200). All 3 cohorts had been previously phenotyped using the Vienna classification27 and all had undergone full colonoscopy and small bowel imaging; most had undergone gastroscopy as well. Location was defined by macroscopic disease and colitis was defined as macroscopic colonic involvement without any evidence of small bowel disease evaluated by ileoscopy and/or SBFT. Distribution of disease was based on the maximal extent of the disease throughout follow-up. In the rare cases where ileal intubation was not performed, if other strong evidence for ileal involvement was found these were included (stricture, evidence at a later date during surgery of ileal disease, or clear-cut terminal ileal involvement by other imaging in a patient with clear-cut CD involving the colon). Age of onset was defined as age at diagnosis. Patients were genotyped for the 3 common NOD2/CARD15 mutations, R702 W, G908 R, and fs1007Ins c, using previously described methods.5, 8, 25 Gender and ethnicity/race were also recorded. All patients from the US and Italian cohorts were Caucasian and Jewish ethnicity from the Israeli cohort was subdivided into Ashkenazi or Sephardic ethnicity. Since age of onset or disease location could be confounded by ethnicity or race, patients with ethnicities characterized by a lower frequency of NOD2/CARD15 mutations (blacks and Hispanics) were excluded.26 Every analysis was corrected for ethnicity. Disease distribution was defined using the Vienna Classification (L1 = ileum only, L2 = colon only, L3 = ileocolonic disease), while the area of disease involvement was grouped into colonic (L2 and L3) or ileal (L1 and L3). Since analysis of the early-onset colitis phenotype was our objective, and in order to evaluate the role of colonic and ileal susceptibility, we excluded patients who did not have colonic or ileal involvement from the final analysis (total of 43 patients excluded).

Ethical Considerations

Data for all collection centers was obtained after informed consent and coded using a numerical code. Every center received approval for genotyping these patients through their individual local and/or national review boards.

Statistical Analysis

Haplotype Reconstruction

The NOD genotypes were phased into haplotypes using the software GERBIL.29 All haplotypes contained no more than 1 mutation, with 1 exceptional haplotype that contained 2 mutations. Based on this property and on the fact that all 3 mutations are known to be deleterious,9, 12, 13 for the association analysis we treated the 3 possible haplotypes with 1 mutation as 1 locus. Any haplotype that contained at least 1 NOD mutation was considered an “Any NOD” mutation, and a haplotype without mutations was considered an “Any NOD wildtype” allele. All association analyses were made using the “Any NOD” locus.

Association Analyses

The association between the Any NOD locus and the following phenotypes were evaluated: disease location, gender, and age of onset (AOO). The association to each phenotype was evaluated by permutation test28 as follows: For a discrete trait (e.g., disease location and gender) we used the Pearson χ2 score as the test statistic. The contingency table for the Pearson χ2 score was built based on chromosome counting (2 × 2 table). This statistic assumes the multiplicative model for penetrance. For a continuous trait (e.g., AOO) both analysis of variance (ANOVA) and Kolmogorov-Smirnov (KS) scores were used. The same statistic was calculated for 107 datasets with the same genotypes and randomly permuted labels of the phenotype. The fraction of datasets on which this score exceeded the original value was used as the P-value.

The association between disease location and AOO was evaluated using several methods:

  • 1
    ANOVA test: We tested disease distribution/involvement versus AOO by using the ANOVA score for the permutation test as described above.
  • 2
    χ2 test: By setting an age cutoff c, each individual was assigned a discrete AOO trait depending on satisfying AOO≤c or AOO>c, and the Pearson score was used for the permutation test. We tested several values of c in the range of 5–15 years, in jumps of half a year.
  • 3
    Isolated colitis and colonic involvement ratios versus AOO using a sliding window: For a given age range, we defined the colitis ratio as the fraction of patients with isolated colitis out of the total number of patients whose AOO fell inside the range. We used as the range a window of 4 years, and by sliding the window we plotted a graph of the colitis ratio as a function of the AOO. A range of 4 years was selected since it guaranteed at least 30 patients in all windows.

The association of gender with AOO and disease location was evaluated by the same methods described above.

Since there are different population groups in the study, all permutation tests were corrected as follows: The statistic score was calculated for each population separately, and the test statistic was defined to be the weighted average of these scores, where the weight of each score is the size of its corresponding population. In calculating the P-value, permutations were generated by randomly permuting the labels within each population independently. This statistic avoids the bias in the P-value that might occur due to the mixture of different populations. The Israeli population group was divided further into 2 subpopulations according to ethnicity, Ashkenazim and Sephardim. Israeli patients with mixed or unknown ethnicity were excluded from the cohort when applying the population correction.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The database with the 3 combined cohorts consisted of 721 patients. Initial evaluation of disease distribution before exclusion disclosed that colitis was significantly age-related (P = 0.0009) while ileocolitis was not (NS). Fourteen patients who had disease only above the distal ileum (L4 according to the Vienna Classification) were subsequently excluded, as well as 29 patients with missing data. The remaining 678 patients served as the final study cohort. The population distribution was as follows: 199 Caucasian patients from the Italian cohort (mean follow-up 6.8 years), 286 Caucasian patients from the Wisconsin cohort (mean follow-up 3.66 years), and 193 patients from the Israeli cohort (89 with Ashkenazi ethnicity, 50 with Sephardic ethnicity, and 54 with mixed ethnicity, mean follow-up 4.9 years). Mean age and gender of individuals with different disease location are presented in Table 1. Carriage of a NOD2/CARD15 mutation was found in 39.82% of the population and the allele frequency was 25.22% (Table 2). There were no significant differences between the 3 cohorts for homozygosity or heterozygosity.

Table 1. Mean Age and Gender Versus Disease Location
 ColonIleumIleocolonicTotal
  • **

    P = 0.00008 colitis versus ileum and ileocolonic disease.

Number (%)167 (24.6)144 (21.2)367 (54.2)678
Mean AOO ± SD11.39 ± 4.32**13.47 ± 3.2212.72 ± 3.5312.55 ± 3.75
Males %55.0963.1959.9559.44
Table 2. Minor Allele Frequency in Different Populations
 Any NOD2R702G908fs1007
  1. Differences between the 3 populations for Any NOD2 are not significant.

Italy (n = 199)26.1312.067.297.04
USA (n = 286)26.4010.315.7710.31
Israel (n = 193)22.545.7010.626.22
All25.229.517.608.19

NOD2/CARD15 and Disease Location

We found a clear association between NOD2/CARD15 mutations and disease location. NOD2/CARD15 mutations were associated with ileitis and ileocolonic disease (P = 1.1 × 10−4 using L1, L2, and L3 as 3 separate categories). Since ileal disease appears in both the L1 and L3 categories, we tested the association between ileal involvement (defined as L1+L3) and NOD2/CARD15 mutations. The P-value for ileal involvement, L1+L3 versus L2, was 1.91 × 10−5, odds ratio (OR) 2.48 (95% confidence interval [CI] 1.77–3.47).

NOD2/CARD15 Mutations and AOO

NOD2/CARD15 mutations were not associated with AOO (P = 0.9 by ANOVA, P = 0.6 by KS).

Disease Location and AOO

The mean age of onset in children differed according to the distribution of disease as defined by the Vienna Classification (P = 8.39 × 10−5; mean ages appear in Table 1). However, in order to correctly test for association between disease location and AOO we had to neutralize the strong effect of NOD2/CARD15 mutations on disease location. We subdivided the cohorts into patients carrying a NOD2/CARD15 mutation (n = 270) and those with wildtype NOD2/CARD15 (n = 408). In the NOD2/CARD15 wildtype cohort, AOO was significantly associated with disease distribution (P = 2.16 × 10−4, ANOVA test on 3 categories). This was due to a lower mean age for patients with isolated colitis, but not due to an age association with colonic involvement. The P-value of the ANOVA test for isolated colitis (L2) versus ileal or ileocolonic disease (L1+L3) was 7.13 × 10−5. The P-value of the ANOVA test for any colonic involvement (L2+L3) versus no colonic involvement (L1) was not significant (P = 0.35). We therefore clustered L3 with L1 for the analysis in this cohort. The χ2 test using age cutoffs also revealed a strong association between isolated colitis and younger AOO. All our tested cutoffs between ages 5 and 13 years yielded significant P-values, with the strength of the association decreasing after age 10 (Fig. 1). Using an AOO cutoff of 10 yielded a P-value of 4.57 × 10−5, OR 2.76 (95% CI 1.72–4.43). The effect of AOO on disease location for each genotype is presented in Tables 3 and 4.

thumbnail image

Figure 1. The association between isolated colitis and age of onse in patients without NOD2/CARD15 mutations. The graph shows the P-values of the χ2 test using different age cutoffs.

Download figure to PowerPoint

Table 3. Effect of Age of Onset (Using Age 10 as Cutoff) on Disease Distribution by Vienna Classification A. NOD2/CARD15 Wildtype Cohort
LocationAge ≤ 10Age > 10Total
  1. P = 0.0178 (χ2 test).

  2. Contingency tables of disease location (by AOO cutoff). The expected values under the null hypothesis of no association appear in parentheses.

Colitis47 (29.96)79 (96.04)126
Ileum11 (17.83)64 (57.17)75
Ileocolonic39 (49.21)168 (157.79)207
Total97311408
P = 0.00012 (χ2 test, numbers in parentheses are expected values).
B. NOD2/CARD15 Mutant Cohort
LocationAge ≤ 10Age > 10Total
Colitis14 (10.78)27 (30.22)41
Ileum8 (18.14)61 (50.86)69
Ileocolonic49 (42.07)111 (117.93)160
Total71199270
Table 4. Effect of Age of Onset (Using Age 10 as Cutoff) on Isolated Colitis A. NOD2 /CARD Wildtype
 AOO ≤ 10AOO > 10Total
  1. P = 0.65.

  2. Contingency tables of disease location (by AOO cutoff). The expected values under the null hypothesis of no association appear in parenthesis.

Colitis47 (29.96)79 (96.04)126
Ileum + ileocolonic50 (67.04)232 (214.96)282
Total97311408
P = 0.000049.
B. NOD2/CARD15 Mutant
 AOO ≤ 10AOO > 10Total
Colitis14 (10.78)27 (30.22)41
Ileum + ileocolonic57 (60.22)172 (168.78)229
Total71199270

We further investigated the association with isolated colitis in NOD2/CARD15-negative patients by plotting the isolated colitis ratio versus AOO using a sliding window (Fig. 2A). We observed that the colitis ratio decreases in the first decade and remains fairly constant in the second decade.

thumbnail image

Figure 2. The isolated colitis (L2) and colonic involvement (L2+L3) ratio by age of onset in patients with or without NOD2/CARD15 mutations (WT, wildtype). A: Isolated colitis ratio. B: Colonic involvement (isolated colitis + ileocolitis) ratio. Windows with median age below 6 were excluded since they contained fewer than 30 patients.

Download figure to PowerPoint

A similar analysis was performed among the carriers of a NOD2/CARD15 mutation. AOO was associated with disease distribution as defined by the Vienna Classification (P = 0.02, ANOVA test). The P-value of the ANOVA test for L2+L3 versus L1 was 0.01, and the P-value of the ANOVA test for L2 versus L1+L3 was 0.61. We therefore clustered L3 with L2 for the analysis in this cohort. The χ2 test using age cutoffs also revealed an association in patients with NOD2/CARD15 mutations between colonic involvement and lower AOO. All age cutoffs between 8 and 10.5 yielded significant P-values (0.014–0.026). Age 10 cutoff yielded a P-value of 0.015 (OR 3.48, 95% CI 1.57–7.71). The sliding window graph was plotted for the L2+L3 ratio (colonic involvement ratio, Fig. 2B). A similar observation as in the NOD2/CARD15 wildtype cohort was obtained: the ratio decreases in the first decade and remains constant in the second decade.

Patients with the L4 proximal intestinal disease (including those with isolated L4, who were excluded from the analysis described above) displayed a trend toward a younger AOO (P = 0.07) than patients without L4, although patients with isolated L4 disease were not younger. Since there were only 14 patients with isolated L4 location, these findings must be interpreted with caution.

Gender

Gender was not associated with AOO (ANOVA P = 0.5, age cutoff P = 0.52), nor with location (P = 0.33), nor with genotype (P = 0.86).

Colonic and Ileal Susceptibility

We wanted to test whether colonic susceptibility in young patients can be a possible mechanism that explains the observed tendency to colonic disease at young age. We took only patients who carried a NOD2/CARD15 mutation and had either ileal or ileocolonic disease (n = 229, Table 5). Patients in this cohort have ileal involvement by definition and a lower chance of having mutations in other susceptibility genes. This allowed us to control for genotype and ileal involvement. Under the null hypothesis of no association between colonic disease and AOO we expect the AOO of patients with colonic disease (ileocolonic in this cohort) to be randomly distributed. Performing the test revealed an association between low AOO and ileocolonic disease (P = 0.005 ANOVA test, P = 0.016 age 10 cutoff test).

Table 5. Effect of Age of Onset (Using Age 10 as Cutoff) on Disease Location in Patients with NOD2/CARD15 Mutations, Isolated Colitis Excluded
LocationAge ≤ 10Age > 10Total
  1. P = 0.016 (χ2 test).

Ileum8 (17.17)61 (51.83)69
Ileocolonic49 (39.83)111 (120.17)160
Total57172229

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

In this study we attempted to evaluate age-related disease location and its association with genotype. We distinguished between the distribution of the disease and intestinal involvement (small bowel involvement or large bowel involvement regardless of distribution).

We found that early-onset CD is significantly associated with a disease phenotype (isolated colitis), and that this association depends on the NOD2/CARD15 genotype. Early-onset disease was characterized by isolated colitis in patients without NOD2/CARD15 mutations. In order to exclude genotype-independent colonic susceptibility in young patients, we performed a separate analysis for colonic involvement, as opposed to isolated colitis, in the same NOD2/CARD15 wildtype cohort by age. This analysis failed to demonstrate an association with AOO, nor did we find a decrease in ileal involvement in early-onset wildtype patients. Since colonic susceptibility was not age-dependent in NOD2/CARD15 negative patients, our data suggest that early-onset colitis without ileal involvement is a specific genetic subset, and strengthens the hypothesis that the susceptibility genes associated with this colitis phenotype may present earlier.

In addition, we did not find an association between NOD2/CARD15 mutations (single or multiple) and AOO, strengthening the argument that other genetic mutations play a more important role in early-onset CD.

These differences were not due to ethnic diversity in the NOD2/CARD15 allele frequency, since all 3 cohorts had a similar allele frequency for NOD2/CARD15, and all tests were corrected for subpopulations.

The effect of AOO on phenotype was different for patients with a NOD2/CARD15 mutation. NOD2/CARD15 mutations were significantly associated with ileal involvement (P = 0.000019). However, an age-related difference was found for colonic but not ileal involvement. This finding supports previous observations that patients with colitis and ileocolonic disease tended to be younger than patients with isolated ileitis.4, 7 In order to evaluate if this phenomenon is due to decreased ileal susceptibility in younger children8 versus the result of increased colonic susceptibility, and not to genotype, we analyzed the cohort of patients with NOD2/CARD15 mutations for differences in ileal or colonic involvement. Younger children did not have a lower frequency of ileal involvement; instead, the distribution changed. Isolated ileitis was decreased at the expense of increased ileocolitis. This finding is inconsistent with the hypothesis that decreased ileal susceptibility is the cause of colitis-predominant pediatric onset CD.

Our data may suggest that disease in patients carrying NOD2/CARD15 mutations might start in the ileum. Since NOD2/CARD15 variants involve loss of function to the ileal innate immune system, it may be possible that compensatory functions that might limit damage to the ileum may not be present to the same degree in early childhood, leading to a greater extension of the disease from the ileum to the colon.

Diversity of phenotype has been shown to be associated with genotype, but it is unclear if this is solely due to variability in disease susceptibility genes, or is a result of disease-modifying genes.21–24, 30, 31

Evidence for candidate genes with mutations or single-nucleotide polymorphisms (SNPs) that are likely to present more frequently in childhood and lead to colitis-predominant CD is currently scarce, with only 1 study identifying a candidate gene for the colitis-predominant disease of childhood. Although this finding has not been substantiated in other ethnic populations, Levine et al5 described an association between a polymorphism in the TNF promoter and isolated colitis with an increased likelihood of pediatric onset disease in an Israeli cohort by comparing the prevalence of the polymorphism to an adult cohort. In addition, studies evaluating other candidate genes or loci8, 22, 31 in pediatric cohorts have failed to show a predominance of early AOO or a colitis-predominant phenotype. Despite these facts, more pediatric data are currently becoming available, and several studies are ongoing to further investigate the role of candidate genes specifically in colitis-predominant, pediatric onset phenotype. Identification of genotype–phenotype correlations in a subpopulation (such as first-decade onset disease) in genomewide screens may require a separate analysis by AOO and/or location phenotype in order to identify these associations, and are currently not standard practice. The definitive proof for early-onset genotype requires identification of the appropriate susceptibility or disease-modifying genes, with reproducible data.

We did not explore upper intestinal or gastric involvement in this study, since this was not the objective of this study, and because the definitions for involvement differ in the literature (macroscopic versus microscopic, nonspecific inflammation versus granulomas only) and were hard to control. We also did not exclude patients without a gastroscopy in this study, although all patients had small bowel imaging and the vast majority had a gastroscopy as well.

In conclusion, we found that early-onset pediatric disease differs from disease presenting during the second decade. The age-related variation in phenotype is genotype-dependent and associated with a different host response and disease distribution. Although NOD2/CARD15 mutations are associated with ileal involvement, we found increased colonic susceptibility in patients with early-onset disease carrying these mutations. The most striking finding was a significantly higher prevalence of isolated colitis, but not colonic involvement, in NOD2/CARD15 wildtype patients, which presented most commonly in early childhood and declined with age. This trend for more colitis lasted until age 11. Together, these findings demonstrate that pediatric onset CD may be characterized by different genes that predispose to early onset and isolated colitis. Further studies, therefore, are needed to specifically identify genes that predispose children to early pediatric onset disease.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The following physicians contributed to the study by providing DNA samples and clinical information on their patients: From Italy: S. Giovanni Rotondo: A. Andriulli, V. Annese, F. Bossa, M. Pastore, M. D'Altilia; Roma: O. Borrelli, C. Bascietto; Napoli: R. Berni Canani, A.M. Staiano; Padova: G. Guariso, V. Lodde; Messina: G. Vieni, C. Sferlazzas; Padova: R. D'Incà, G.C. Sturniolo; Parma: G.L. De Angelis; Bari: V. Rutigliano, D. De Venuto; Palermo: S. Accomando; Reggio Calabria: C. Romano; Foggia: A. Campanozzi; From Israel: Tel Aviv: B. Weiss, S. Reif; Haifa: R. Shaoul, R. Shamir.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES