In our study, one of the largest pediatric cohorts reported to date, the cumulative incidence of the development of complicated disease in pediatric patients with CD was 13.0%, 26.7%, and 37.9% at 1, 5, and 10 years, respectively, from the time of diagnosis of disease. Our study illustrates that evolution to complicated disease in a pediatric cohort is common.
Few reports describe the evolution of disease in the pediatric population.7–12
It is difficult to make direct comparisons of these reports to our study because of different disease classification schemes and approaches to statistical analysis. Regardless, taken together, our studies indicate that the development of stricturing and penetrating disease are important complications of CD diagnosed during childhood and adolescence.
Studies focusing on primarily adult populations also illustrate that evolution of disease is common in CD.3,4,13
Cosnes et al14
reported that the probability of having inflammatory disease (without complications) at 5 and 20 years after diagnosis was 48% ± 1% and 12% ± 1%, respectively. In contrast, in our pediatric cohort the cumulative probability of remaining free of stricturing or penetrating disease at 5 and 10 years after diagnosis was 73.3% (95% CI = 69.5%–76.8%) and 62.1% (95% CI = 55.3%–68.2%), respectively. In the Cosnes et al cohort, at 5 and 20 years after diagnosis, the risk of stricturing disease alone (not associated with a penetrating complication), was 12% and 18%, respectively. The cumulative incidence of stricturing disease in our pediatric cohort was 13.6% and 20.5% at 5 and 10 years after diagnosis, respectively. It is difficult to make direct comparisons of the development of stricturing disease in these 2 cohorts, as we did not exclude patients from this group who also had a penetrating complication. In the Cosnes et al population, the cumulative incidence of penetrating disease at 5 and 20 years after diagnosis was 40% and 70%, respectively. In our pediatric population, the cumulative incidence was 17.1% at 5 years and 24.5% at 10 years. The cumulative incidence of development of penetrating complications of disease in the Cosnes et al study cohort was higher than our pediatric cohort, possibly reflecting differing phenotypes of disease diagnosed in adulthood compared with childhood and adolescence.
Louis et al3
reported that 78% of patients with stricturing disease remained stricturing (median follow-up 6.5 years; range 1–32 years). We found that in patients with stricturing disease, patients who had penetrating complications prior to first surgery developed penetrating disease within 2 years of the stricturing complication. The cumulative incidence of penetrating disease prior to first surgery was 13.3% at 2 years from the time of diagnosis of stricturing disease. Furthermore, Louis et al3
reported that the rate of change from nonstricturing, nonpenetrating disease to stricturing or penetrating disease was stable at approximately 25%–33% every 5 years. In contrast, the cumulative incidence of development of complicated disease in our cohort increased from 26.7% at 5 years to 37.9% at 10 years from the time of diagnosis of disease.
Studies have reported associations between small bowel disease and stricturing complications13,14
and between colonic disease and penetrating complications.3,15
Others have reported an association between stricturing/penetrating disease complications and small bowel CD and between inflammatory disease and colonic CD.4,12
Smith et al4
reported that progression to more severe disease was associated with small bowel disease (irrespective of associated colonic lesions), while, conversely, colonic disease was associated with the least progression to more severe disease. Similarly in our cohort, the cumulative incidence at 1, 5, and 10 years of complicated disease was lower in patients with colonic disease compared with patients with small bowel disease or small bowel disease combined with colonic disease.
In a cohort of 169 patients with CD diagnosed between 1970 and 1993, Schwartz et al16
found the cumulative incidence of at least 1 fistula (any location) after 1 year was 21%, after 5 years was 26%, after 10 years was 33%, and after 20 years was 50%. The cumulative incidence at 10 years of any fistula in our cohort was 18.7%, lower than the Schwartz et al population. Schwartz et al reported that the cumulative risk of at least 1 perianal fistula was 12% after 1 year, 15% after 5 years, 21% after 10 years, and 26% after 20 years. In contrast, in our cohort the cumulative incidence of perianal fistula was 2.2% at 1 year, 5.0% at 5 years, and 9.4% at 10 years. These observed differences in the incidence of any fistula and perianal fistula in the Schwartz et al population compared with ours may reflect that the Schwartz et al population included both adult and pediatric patients, while our population was a defined pediatric cohort. Schwartz et al reported that 9% of fistulizing episodes were rectovaginal fistulas and 3% were enterovesical. Although we cannot make direct comparisons to our study because of different approaches to statistical analysis, the cumulative incidence of rectovaginal fistula and rectovesical fistula in our cohort was low: 1.4% and 1.0%, respectively, at 10 years from the time of diagnosis.
Classification of Disease
The Montreal classification system allows for disease diagnosed in patients less than age 16 years to be classified as a separate category.17
Studies in pediatric patients have suggested early onset disease (age 0–5 years) may represent a different disease phenotype.5,18,19
In a previous study, we found that children who are older at diagnosis (6–17 years) are at higher risk for the development of complicated disease compared with children who are younger at diagnosis (0–5 years).2
Therefore, adjusting the classification system further to create an age category for patients diagnosed at 0–5 years may be beneficial.
A hierarchy for disease behavior exists for stricturing and penetrating complications.3
If a penetrating complication is present, disease is classified as penetrating even in the presence of coexisting strictures. Reports on the association between disease behavior and future complications are conflicting.4
Our data show that development of stricturing disease commonly follows development of penetrating disease. It may be helpful to revise the classification system such that hierarchical rules are not in place, and classification as stricturing or penetrating disease is not mutually exclusive.
The main limitations of this study include those that are inherent to conducting a retrospective, secondary database analysis in tertiary care referral centers and affiliated centers. Although our study is not population-based, almost all pediatric patients with chronic disease including CD are followed by pediatric specialists in major regional medical centers such as those represented by the Consortium. Inaccurate codings are inherent in data collection and storage and can influence the results; however, these limitations will most likely bias our results toward the null hypothesis secondary to nondifferential misclassification. Complete data on disease location are missing for some of our patients. Although our findings regarding associations between the development of disease complications and initial disease location are similar to previous reports in the literature, prospective longitudinal studies that minimize missing data are needed. While the total number of subjects declines with time, primarily due to transition to adult care, the statistical models take this into account, emphasizing the earlier part of the models when the numbers are larger. Furthermore, the sample size of our cohort is one of the largest populations of pediatric patients with CD reported to date.