Disease location is highly variable at diagnosis and is not fixed over time. Data from a large multicenter European registry found the initial disease location of 582 children with Crohn disease to be widely distributed according to the Paris classification [10], with disease location L1 in 16%, L2 in 28%, L3 in 53%, and isolated L4a + L4b in 4% [11]. In a report from Scotland, at diagnosis extensive disease including the ileum, colon, and upper GI tract (disease location characterized as L3 + L4 by the Montreal classification [12]) was found in 31% of children [13]. However, among a subgroup of 149 children with less extensive disease at diagnosis who were followed at least 2 years after diagnosis, extension of CD was noted in 39% [13].

Disease behavior also evolves over time. At initial diagnosis, the vast majority of children have an inflammatory disease phenotype. However, as time goes on, an increasing proportion expresses a changing phenotype, characterized as either stricturing or penetrating. In a systematic review of the literature from years 1966 to 2010 evaluating 3505 pediatric-onset CD patients with at least 5 years of follow-up, development of stricture occurred in 24–43%, fistulae in 14–27%, and perianal disease in 25–30% of patients was found [14]. Similar disease behavior has also been documented clearly in data derived from the pooled observations from three multicenter North American pediatric IBD registries [15]. Among 796 children followed prospectively from diagnosis, 96 (12%) presented with a stricturing or penetrating CD phenotype. Among the 700 who had an inflammatory phenotype at presentation, 140 (20%) developed stricturing or penetrating disease after a mean of 32 months of follow-up [15], a finding strikingly similar to the 24% rate of complicated CD behavior described after 4 years in a pediatric study from Scotland [13]. Similar observations over extended periods of time have been reported in population-based studies in adults from both France [16] and New Zealand [17] (Fig. 7.2). In the latter study, a comparison of 630 subjects with adult-onset disease and 85 children diagnosed before age 17 years revealed no difference in the rate of progression from inflammatory to either stricturing or penetrating disease phenotype [17].

Racial differences may affect the frequency of complicated CD, as a study from Baltimore has demonstrated more frequent stricturing and penetrating disease in black children compared to white children seen in the same university-based practice [18]. The risk for phenotypic change may also be associated with the presence of specific genetic allelic variants. Earlier reports suggested an increased risk of fibro-stenosis complications for patients with NOD2/CARD15 variants [19, 20], while those with abnormalities in the IBD5 gene may be more likely to develop perianal fistulae [21]. However, more recent studies only implicate NOD2/CARD15 mutation in risk of ileal disease location (which may be more likely to stricture compared to colonic disease) and not independently with increased risk of stricture [22, 23]. Crohn disease in children who are homozygous carriers of ATG16L1 mutation may have significantly increased stricturing behavior and have lower risk of perianal disease compared to wild-type patients [5]. In a small study of children and adults in Taiwan, homozygous mutation in the risk candidate gene SLCO3A1 was significantly associated with perforating disease compared to those with wild type who had more inflammatory disease [24]. Children at risk for stricturing or internal penetrating complications have also been shown to be more likely to have increased immune responses to microbial antigens, characterized by the presence of high-titer antibodies such as anti-ompC and anti-I2 [15, 25].

For a significant subgroup of children with CD, growth impairment is an important characteristic of the disease’s natural history. While acute weight loss commonly is present in children with both ulcerative colitis and CD, impairment in linear growth is primarily a problem in the latter condition. At the time of initial diagnosis, about a third of children with CD have already dropped two or more major growth channels from their pre-illness growth percentiles [26, 27]. More dramatically, 88% have delayed height velocity at diagnosis [28]. Over time, periods of significantly impaired growth can be seen in about 60% of children and adolescents [27]. While catch-up growth is often possible, 7–35% of children diagnosed during the 1970s and 1980s had final adult heights that were significantly shorter than expected [27]. As a group, young adults who develop CD as children have adult heights skewed toward the lowest percentiles. In reports from both Chicago and New York, ~50% of young adults with childhood-onset CD have final adult heights less than the 10% for the general population, and ~25% have adult heights less than the 5% [26, 27]. While therapies including enteral nutrition [29], methotrexate [30], and infliximab [31, 32] have improved growth in the short to medium term, current therapies have not yet been demonstrated to provide a long-term reversal of growth impairment.

An important characteristic of Crohn disease in children as well as adults is the tendency to develop corticosteroid dependence. Population-based studies in adults from both Olmstead County, Minnesota [33], and Copenhagen County, Denmark [34], demonstrate similar findings; acute response to corticosteroid therapy in adults with CD is reasonably good (complete remission in 48–58%, partial remission in 26–32%, and no response in 12–20%). However, long-term response is less optimal, with rates of corticosteroid dependence of 28–36% at 1 year [33, 34].

A similar risk for corticosteroid dependence is evident in children. As in adults, acute response to a course of corticosteroids is good. In data derived from a multicenter North American observational registry, among 109 newly diagnosed children with moderate-severe CD activity treated with corticosteroids, 60% have a complete and 24% a partial clinical response by 3 months after initiation of treatment [35]. However, despite concomitant use of immunomodulators in many of these children, 31% are corticosteroid dependent at 1 year. In fact, without infliximab, only 46% of the children in this study maintained a corticosteroid-free remission to 1 year following an initial course of corticosteroids [35]. Similarly, in a French population-based study of 404 patients diagnosed before age 17 years, 24% were corticosteroid dependent, and 5% were corticosteroid resistant at 1 year from diagnosis [36]. In this cohort, corticosteroid use had a hazard ratio for intestinal surgery of 2.98 (p <0.01).

The need for surgery represents another important aspect of the natural history of Crohn disease in children. Table 7.1 summarizes published rates for surgery in children from a variety of different countries. Data from Denmark estimate a mean yearly operation rate of approximately 13%. The cumulative probability of surgery in this Danish cohort at 20 years was estimated to be 47% [6]. A multicenter pediatric experience from the USA estimates the cumulative incidence of surgery to be 6% at 1 year, 17% at 5 years, and 28% at 10 years after diagnosis [41]. Similarly, a pediatric study from Scotland noted resection rates of 20% at 5 years and 34% at 10 years [13]. The presence of variant NOD2/CARD15 [19, 20] and ATG16L1 [5] alleles appear to increase the risk for surgery, presumably due to the known association of these genetic polymorphisms with the development of fibrostenotic ileal disease. The presence of anti-Saccharomyces cerevisiae antibodies also appears to be associated with increased risk for surgery [15, 41].

The effect of immunomodulatory therapy on the need for surgery remains an open question. An analysis from France evaluated a series of successive 5-year adult CD cohorts [42]. Although there was a significant increase in the use of immunomodulatory therapy over time, there was no associated change in the rate of surgery [42]. By contrast, multivariate analysis from a similar series of 5-year adult CD cohorts from the UK identified the early use of thiopurines (within 3 months of diagnosis) to be associated with a marked reduction in the rate of surgery [43, 44].

The studies evaluating infliximab therapy in decreasing surgical rates are also mixed. In a Spanish retrospective assessment of infliximab therapy used in a “step-up” fashion, no significant decrease in surgical rates could be identified in patients receiving infliximab compared to those not receiving the treatment [45]. However, other studies reach the opposite conclusion. For instance, in a study utilizing data from a combined Danish and Czech collaboration, surgical rates in adults 40 months after starting infliximab were 20–23% in infliximab responders compared to 76% in nonresponders [46]. In the ACCENT I [47] and ACCENT II [48] trials of adults with moderate to severe CD, and fistulizing CD, respectively, response to maintenance infliximab was associated with decreased surgery. Similar findings in children have been reported, with surgical rates 50 months after starting infliximab of 10% in patients maintained on the biologic compared to 70% in infliximab failures [49]. Further, in children with a favorable initial response, development of antibodies to infliximab led to loss of response and increased risk of surgical resections [50]. It remains to be determined if the recent trend toward early initiation of biological therapy (“top-down approach”) will prove to result in decreased rates of surgery over time.

Although there are little pediatric data published to document clinical experience, following surgery, the natural history of Crohn disease is to recur both endoscopically and symptomatically. The natural progression of CD following ileocolonic resection has been previously described by Rutgeerts and colleagues, with five levels of disease severity (i0–i4) found endoscopically. Postoperatively, disease appears to evolve from normal mucosa (i0) to the initial appearance of a few aphthous ulcers (i1–i2), followed by progressively more and/or deeper ulcerations until an area of confluent inflammation, large ulcers, or stricturing develop (i3–i4) [51]. CD recurrence is defined by an endoscopic score of i2, i3, or i4, while postoperative remission is defined by a score of i0 or i1.

In retrospective adult studies, symptomatic recurrence of CD following so-called curative resection (complete resection of all visibly evident disease) is reported to be 20–30% within the first year after surgery, with increasing likelihood in each subsequent year [52]. One or more additional surgeries are required in 15–45% of adults within 3 years, 26–65% in 10 years, and 33–82% in 15 years [34]. Controlled trials document severe endoscopic recurrence after placebo treatment in 43–79% of adult subjects by 1 year after surgery and in 42–85% of subjects after 2 years [53–58].

In children, the overall rate of clinical recurrence is estimated to be 50% at 5 years after initial resection [39]. However, the site and extent of preoperative CD can affect the recurrence-free interval such that it is estimated that 50% of children with extensive ileocolitis recur within 1 year, compared to a 50% recurrence rate after 5 years in children with ileocecal disease and a 50% recurrence rate after 6 years if preoperative disease is confined to the small bowel [39]. In a retrospective review of 30 children, multifocal preoperative disease, with perianal or upper gastrointestinal tract involvement, was a risk factor for postoperative recurrence after localized ileal or ileocecal resection [40]. Altering the natural history of postoperative CD and preventing recurrence have become an integral part of CD management. Use of mesalamine or thiopurines appears to have limited benefit [58, 59], while anti-TNF therapy may be highly effective [60, 61] in preventing CD recurrence.

Whether children with CD are at increased risk for malignancy over their lifetime is unknown. No data derived from a population with childhood-onset CD has been reported. In a recent retrospective multinational study from 2006 to 2011, including 20 European countries and Israel, 12 children with CD were identified with a malignancy prior to age 19 years [62]. Since the total number of children with CD during the study period could not be identified, the malignancy risk could not be reported. The majority of cancers were lymphoma or leukemia, and only three patients were not on immunomodulator therapy within 3 months of cancer diagnosis. The effect of CD independent of immunosuppression +/− Epstein-Barr virus infection in developing a malignancy could not be calculated, although the risk appeared to be small. Studies in adults, however, suggest that patients with CD do have an excess of malignancies compared to the general population. In a population-based cohort from the Uppsala region of Sweden, there was an increased relative risk of colorectal cancer of 2.5 (95% confidence interval (CI) 1.3–4.3) in patients with CD [63]. Duration of illness and gender did not affect risk, but those subjects with colonic disease had a greater risk of colorectal cancer than those with only small bowel involvement. Of note, however, among those subjects with any colonic involvement diagnosed with CD before the age of 30 years, the relative risk of colorectal cancer increased to 20.9 (95% CI 6.8–48.7) [63]. By contrast, a similar population-based study from Denmark identified a relative risk of colorectal cancer of only 1.1 (95% CI 0.6–1.9), and no risk differences were noted in different subgroups of patients [64]. A similar modest increase in colorectal cancer risk (1.9; 95% CI 0.7–4.1) was found in a population-based study from Olmstead County, Minnesota [65].

By contrast, the risk of small bowel cancer consistently appears increased in patients with CD. In part because the rate of small bowel cancer in the general population is very low (estimated to be 0.005% at 5 years and 0.03% at 25 years according to data cited in reference 62), there is a significantly elevated relative risk for small bowel cancers in patients with CD. In the Danish study cited above, the relative increased risk for small bowel cancer was 17.9 (95% CI 4.8–42) [64]. In Olmstead County, the relative risk was found to be 40.6 (95% CI 4.4–118) [65]. Duration of CD did not appear to influence risk of developing a small bowel cancer. Adenocarcinoma, carcinoid, leiomyosarcoma, and primary intestinal lymphoma have all been reported. The effect of age at CD onset on the risk of developing small bowel cancer has not been reported.

There may also be a slight increase in risk of developing lymphoma, although data is mixed and not always controlled for risk associated with therapeutic agents. In a single-center, retrospective study between 1979 and 2008 that included 791 children with CD followed in Boston, MA, one non-Hodgkin’s lymphoma occurred in a patient receiving thiopurines; the overall cohort lymphoma risk did not meet statistical significance [66]. In a large population-based retrospective study of adults living in the UK between 1988 and 1997, seven patients with lymphoma were reported among 6605 patients with CD, and 0/7 were exposed to thiopurines. The risk of lymphoma in this cohort was not increased compared to the control population (relative risk 1.39; 95% CI 0.50–3.40) [67]. Among 454 adults living in Olmsted County who developed CD between 1950 and 1993, one developed a non-Hodgkin’s lymphoma, resulting in a slight increase in relative risk (2.4; 95% CI 0.1–13), although not statistically significant compared to the general population [68]. In this report, however, the referral practice of the same group of investigators revealed that development of lymphoma was associated with treatment with immune modifiers in about 5% of cases [68].

A meta-analysis utilizing data from six studies estimates the risk of lymphoma in inflammatory bowel disease (IBD) patients treated with azathioprine or 6-mercaptopurine to be increased about fourfold (4.18; 95% CI 2.07–7.51) [69]. Infliximab in combination with thiopurine may add an additional element of risk, especially for children and young adults, in whom the development of hepatosplenic T-cell lymphoma has been described [70]. Whether these risks to children with CD are due to the nature of the illness in children, their frequent need for potent immune modifier and biological therapy, or both, is not known.

In addition to imposing significant physical morbidity, CD in childhood imposes potentially dramatic psychosocial burdens as well. Health-related quality of life (HRQOL) scores, as measured by the IMPACT questionnaire (a validated, pediatric IBD health-related quality of life questionnaire) [71], correlate with physician’s global assessment of disease severity, such that children with moderate-severe activity have the poorest HRQOL scores [72]. Over the first year after diagnosis, age also appears to be an independent factor affecting HRQOL scores, with older children reporting poorer IMPACT scores [72]. While treatment results in significant improvement in IMPACT scores in the first year after diagnosis, it is unknown whether further improvements in HRQOL occur over time. One by-product of increased disease severity is increased parental stress. The effect of parental stress was recently studied and found to partially contribute to lower HRQOL in children with active disease [73]. Children frequently report being bothered by having a chronic illness, having to undergo tests, and feeling tired. Over time, they also report feeling that their chronic illness is unfair and that they experience problems revolving around having to keep their illness a secret from others [72].

Other studies have noted that children with CD experience frequent absences from school, that they frequently require home tutoring, and that they commonly cannot participate fully at all in physical education classes [74, 75]. Children express fears concerning everyday childhood activities, schooling, and ability to get a job [76]. Fifty-seven percent of a cohort was reported to have had an absence from school of at least 2 months duration, and in this same cohort, 8% were involuntarily unemployed as young adults [77]. Similar impairments are described in adult CD populations, with 15% of a Danish population on disability by 15 years after diagnosis, 25% reporting some inability to work in any given year of follow-up, and 50% reporting one or more years during first decade of disease with at least some inability to work [7]. These latter studies [7, 74–77] report on patients treated in the prebiologic era, and the use of anti-TNF therapies appears to have significantly improved patient quality of life measures [78, 79].