You are here
Evolution of disease phenotype in pediatric-onset Crohn’s disease after more than 10 years follow up—Cohort study
Digestive and Liver Disease, In Press, Corrected Proof, Available online 31 August 2016
Pediatric-onset Crohn’s disease (CD) is a heterogeneous disorder which is subjected to progression and complications in a substantial proportion of patients.
We aimed to assess the progression in pediatric-onset CD phenotype on long term follow up.
Medical charts of pediatric onset CD patients with at least 10 years follow-up were analyzed retrospectively. Disease phenotype was determined at diagnosis and during follow up at different time points. Phenotype was determined according to the Paris classification. The impact of possible predictors on phenotype progression was assessed as well as the association between different therapeutic regimens during disease course and phenotype progression.
Progression of disease location, behavior, and perianal involvement was observed in 20%, 38% and 20% of patients, respectively, after a median follow-up of 16.4 (±4.4) years. Microscopic ileocolonic disease at diagnosis was significant predictors for progression of disease extent. Treatment with anti tumor necrosis factor-ɑ agents and number of flares per years of follow-up were associated with progression of disease extent, behavior and perianal involvement.
Disease extent, behavior and prevalence of perianal disease change significantly over time in pediatric-onset CD. In our cohort, most clinical, laboratory and endoscopic parameters do not serve as predictors for long-term disease progression.
Keywords: Crohn’s disease, Long-term follow-up, Phenotype progression.
The incidence of Crohn’s disease (CD) and ulcerative colitis (UC), collectively referred to as chronic inflammatory bowel diseases (IBD), has increased over the past few decades, particularly in pediatric populations living in developed countries , , and . Approximately 25% of patients with IBD are diagnosed during childhood or adolescence . While CD in particular is characterized by a wide spectrum of phenotypes , the course of disease has often been difficult to predict based on the initial presentation. Phenotypic classification of CD may play an important role in determining the optimal treatment and may assist in predicting the clinical course of disease . According to the available literature, pediatric onset CD runs a more aggressive course, including a higher rate of complications and increasing needs for more aggressive medical therapy during disease course  and . Over time, disease phenotype (behavior and location) may progress, remain stable or regress . Adult studies have demonstrated a decreasing frequency of inflammatory behavior and an increasing frequency of stricturing or penetrating disease behavior over time , , and .
Several studies assessed the natural history of pediatric CD , , , and , though only few reports have focused on the evolution of disease phenotype over long-term follow-up , , , and . Data regarding predictive factors for phenotypic progression are limited  and . The main objective of the current study was to assess the long-term evolution of the disease phenotype in childhood-onset CD in a tertiary-based hospital cohort and to determine predictors of phenotypic progression including localization, behavior and development of perianal disease.
2. Patients and methods
The study population included all pediatric onset CD patients, diagnosed between the ages 0 to 18 years, with at least 10-years of follow up, who were evaluated at the Schneider Children’s Medical Center of Israel, a tertiary referral center which serves also as a primary center for more than half a million inhabitants in the center of Israel. These patients were followed at our institute between 1981 and 2013 and have continued their follow-up at adult gastroenterology centers. Data was extracted from patients’ medical charts (from pediatric gastroenterology institute at the Schneider Children’s Medical Center and adult gastroenterology institutes at the Rabin Medical Center and the Sheba Medical Center), as well as from a national electronic database. Data extraction was approved by both pediatric and adult institutes were patients have been followed. At our institution patients are typically transferred to an adult gastroenterologist between the ages of 18 and 20 years. Diagnosis of CD was performed according to the revised Porto criteria . The study was approved by the Institutional Review Board of our medical center.
2.2. Description of variables
The characteristics of CD were retrospectively extracted from medical charts. Age at onset, gender, ethnicity, interval between onset of symptoms and diagnosis, age at diagnosis, presence of familial IBD (first relatives), anthropometric data [weight, height and body mass index (BMI)], clinical, laboratory, endoscopic and histological findings at diagnosis and therapeutic regimens were thoroughly investigated by reviewing medical records.
Histological location was based on the presence of inflammatory changes in the mucosa (chronic gastritis/duodenitis/esophagitis, cryptitis, crypt abscesses, and granulomas), architectural abnormalities (crypt distortion, crypt branching, and crypt atrophy), and epithelial abnormalities such as mucin depletion and metaplasia.
Disease activity was assessed at diagnosis using several indices including the abbreviated pediatric Crohn’s disease activity index (abbrPCDAI) and short pediatric Crohn’s disease activity index (shPCDAI), which have been validated and are considered as reliable in pediatric CD  and . We also used the Harvey–Bradshaw index (HBI) which is an accepted tool for assessing the severity of disease in individuals with CD . In addition, for each patient, we determined disease phenotype at diagnosis according to the Paris classification  based on age at onset (A), location (L), behavior (B), perianal disease (P) and presence of impaired linear growth (G).
Definition of disease location was based on endoscopic findings, radiologic studies [including computed tomography (CT) scan and magnetic resonance imaging (MRI)] as well as surgical reports. Growth impairment was defined as height z-score significantly less than expected for age and gender by using CDC growth charts (z-score < −1).
The data collected throughout the follow-up period included: medical treatments, need for surgery, smoking habits (current, never, ex-smoker), and number of flares and hospitalizations. At the end of follow-up we assessed the current phenotype according to the Paris or Montreal classification depending on the patient age and timing of any change in phenotype.
2.3. Outcome measures
Primary outcomes were defined as changes in disease phenotype reflected by progression of extension, behavior and perianal involvement. Patients underwent additional colonoscopy during follow-up for clinical indications including: routine follow-up for disease activity evaluation, cancer surveillance or prior to treatment escalation.
We also assessed predictors of phenotypic changes over time. These predictors were classified as variables at disease onset including: age, gender, ethnicity, familial history of IBD, presence of extra intestinal manifestation (EIM), growth impairment, anthropometric indices, activity indices, duration between symptoms onset and diagnosis, laboratory and histological findings. Variables during follow-up included therapeutic regimens, smoking habits, number of flares, hospitalizations and surgical interventions.
2.4. Statistical analysis
Continuous variables were evaluated for normal distribution using histogram, Q–Q plots and Kolmogorov–Smirnov test and reported as median (interquartile range, IQR) or mean (standard deviation, SD). Categorical variables were reported as frequency and percentage. Continuous baseline characteristics were compared using independent simple T-test or Mann–Whitney and categorical baseline characteristics using Chi-square test or Fisher-exact test. Correlation between continuous variables was evaluated using Spearman or Pearson methods. Partial correlation was used to evaluate the association between continuous variables while controlling other continuous variables. Phenotype change during follow-up was described using Kaplan–Meier plot. Association between phenotype change and baseline characteristics were evaluated using uni and multivariate Cox regression. Variables which reached statistical significance in the univariate analysis (p < 0.2) or were deemed clinically relevant, were selected for inclusion in multivariate logistic regression models to identify independent characteristics at diagnosis associated with phenotype progression. Other clinical variables and different treatments during follow-up were compared between patients with and without phenotypic change by simple T-test or Mann–Whitney for continuous variables and by Chi-square test or Fisher-exact test of categorical variables. p < 0.05 was considered as statistically significant. SPSS version 22 was used for all statistical analyses.
3.1. Disease characteristics at diagnosis
Among 215 children with at least 10 years of follow-up, 3 patients were lost to follow-up. Data of 212 patients were analyzed.
The demographic and baseline characteristics of these patients are summarized in Table 1. Patients were followed for a median duration of 16.4 (±4.4) years (range 10–30.4 years). Of 212 patients, 206 (97%) had been transferred to adult care while 6 were still followed up in a pediatric medical center. The median time from the onset of symptoms and diagnosis was 4 months (IQR, 2–6 months). Of 212 patients 195 (92%) were diagnosed and evaluated at Schneider Children’s Medical Center of Israel while 17 (8%) were referred from another medical center in Israel.
|Variables at diagnosis||Variables at end of follow up|
|Diagnostic investigation, n (%)|
|Colonoscopy||212 (100)||152 (71.7)|
|Gastroscopy||170 (80.2)||45 (21.2)|
|Small bowel imaging||210 (99)||111 (52.4)|
|Surgical intervention||–||94 (44.3)|
|Age at diagnosis, n (%)|
|A1a (<10 years)||27 (12.7)|
|A1b (10 years ≤ and <17 years)||157 (74.1)|
|A2 (17 years ≤ and ≤18 years)||28 (13.2)|
|Disease location (macroscopic involvement), n (%)|
|Ileum (L1)||106 (50)||96 (45.3)|
|Colonic (L2)||35 (16.5)||25 (11.8)|
|Ileocolonic (L3)||55 (26)||88 (41.5)|
|Isolated upper GI tract (isolated L4)||16 (7.5)||3 (1.4)|
|Upper GI involvement (L4) + L1, L2 or L3||73 (34.4)||56 (26.4)|
|Disease location (microscopic involvement), n (%)|
|Ileum only||92 (43.4)||–|
|Colon only||25 (11.8)||–|
|Isolated upper gastrointestinal tract||3 (1.4)||–|
|Disease behavior, n (%)|
|Non stricturing, non penetrating (B1)||174 (82.1)||100 (47.2)|
|Stricturing (B2)||26 (12.3)||47 (22.2)|
|Penetrating (B3)||6 (2.8)||27 (12.7)|
|Stricturing & penetrating (B2B3)||6 (2.8)||38 (17.9)|
|Perianal involvement (abscess/fistula) (P1), n (%)||29 (13.7)||63 (29.7)|
|Growth impairment (G1), n (%)||54 (25.5)|
|Granulomas, n (%)||68 (32)|
|Female, n (%)||89 (42)|
|Age at diagnosis (years), median (iqr)||13.7 (11.8–15.4)|
|Familial history, n (%)|
|Crohn’s disease||27 (12.7)|
|Ulcerative colitis||4 (1.9)|
|Other autoimmune diseases||1 (0.5)|
|Median duration between symptoms and diagnosis, months, (iqr)||4 (2–6)|
|Anthropometric measurements, mean (sd)|
|Weight z score||−0.96 (1.09)|
|Height z score||−0.68 (1.12)|
|BMI z score||−0.74 (1.17)|
|HBI, mean (sd)||7 (3)|
|ShPCDAI, median (iqr)||40 (25–50)|
|AbbrPCDAI, mean (sd)||25 (10)|
IBD = inflammatory bowel disease, Iqr = interquartile range, SD = standard deviation, BMI = body mass index, HBI = Harvey–Bradshaw index, shPCDAI = short pediatric Crohn’s disease activity index, AbbrPCDA = abbreviated pediatric Crohn’s disease activity index.
Children with non-inflammatory behavior including stricturing (B2), penetrating (B3) or both (B2B3) at onset had a significantly higher rate of ileocolonic disease (L3) as compared with patients with inflammatory behavior (B1) (42.9% vs. 21.5%, p = 0.014), whereas isolated colonic disease was more prevalent in patients with inflammatory behavior (19.1% vs. 2.9%, p = 0.014). Non-inflammatory behavior was marginally associated with histological findings of granulomas at diagnosis (47.1% versus 29.5%, p = 0.051). There was no significant difference in age at diagnosis, presence of perianal disease or growth impairment between these two groups.
Males had significantly more non-inflammatory behavior (20.8% vs. 10.1%, p value = 0.048) but otherwise gender was not associated with prevalence of EIM, growth impairment, disease location, perianal disease or familial history of IBD.
Of the 212 patients, 28 (13%) did not complete comprehensive endoscopic or imaging investigations at the end of follow up, mostly due to prolonged remission and the patient’s refusal to undergo an additional colonoscopy. The phenotype of these patients (behavior and location) was considered as stable.
Ninety four patients (44%) underwent intestinal surgery during follow-up. Among them, 47 patients (50%) had ileocecal resection, 22 (23.4%) underwent distal ileal resection and right hemicolectomy, 7 (7.5%) underwent isolated small bowel resection, 5 (5.3%) fistulectomy, 6 (6.4%) colostomy, 3 (3.2%) total or subtotal colectomy and 4 (4.2%) had surgical stricturoplasty. Among these 94 patients, 19 (20%) underwent more than one surgical procedure and 5 (5%) more than 2 surgical interventions during follow-up.
3.2. Evolution of disease phenotype
3.2.1. Change of location over time
The disease extent at the time of diagnosis and at the last follow up visit is shown in Table 1. At diagnosis, 55 patients (26%) had extensive disease L3 ± upper gastrointestinal involvement (L4), while 106 (50%), 35 (16.5%) and 16 (7.5%) had L1, L2 (isolated colonic) and isolated L4 disease location respectively. Among all the 212 patients, 44 (21%) progressed to develop a more extensive (from L1, L2 or L4 to L3) disease over time. The timing to disease progression was 11.4% (n = 5), 50% (n = 22) and 38.6% (n = 17) of patients within 5, 5–10 and >10 years following diagnosis, respectively. The Kaplan–Meier survival curve of patients who remained with isolated disease (L1, L2 or L4) is shown in Fig. 1.
Phenotype progression over time.
At 25 years patients with pediatric onset CD demonstrated rates of 72% for stability of phenotypic extent, 54% for non-complicated disease and 76% non-perianal disease.
*Dotted line—Kaplan–Meier estimates of remaining free of ileocolonic disease in children with isolated disease extent at diagnosis.
*Continuous line—Kaplan–Meier estimates of remaining free of complicated behavior in children with uncomplicated behavior at diagnosis.
*Dashed line—Kaplan–Meier estimates of remaining free of perianal disease in children with P0 phenotype at diagnosis.
3.2.2. Change of behavior over time
At diagnosis, 174 (82%) patients had inflammatory disease behavior (B1). Among total 212 patients, 80 (38%) progressed to develop more non-inflammatory disease over time resulting in an increase of non-inflammatory disease from 18% to 53%. Progression of disease behavior was observed in 28.8% (n = 23), 36.2% (n = 29) and 35% (n = 28) of patients within 5, 5–10 and >10 years following diagnosis, respectively. The Kaplan–Meier survival curve of patients who remained free of stricturing or penetrating complications are shown in Fig. 1.
3.2.3. Perianal involvement progression over time
The prevalence of perianal disease at the time of diagnosis and at the last of follow up visit is shown in Table 1. At diagnosis, 29 (14%) patients had perianal disease (P1). A total of 43 patients (20%) progressed to develop P1 disease over time resulting in an increase of perianal involvement from 14% to 30%. Progression of perianal involvement was observed in 18.6% (n = 8), 46.5% (n = 20) and 34.9% (n = 15) of patients within 5, 5–10 and >10 years following diagnosis, respectively. The Kaplan–Meier survival curve of patients who remained free of perianal disease is shown in Fig. 1.
Only a small proportion of patients had phenotype progression during childhood: 24 (11%) had extent progression while 34 (16%) had behavior or perianal progression. Hence, in our cohort phenotype progression occurred mainly during adulthood.
3.2.4. Predictors of phenotype progression at diagnosis (including: extent, behavior and perianal progression)
By univariate Cox proportional hazard models, there was no significant association between most of the assessed variables at diagnosis and the 3 types of phenotype progression during follow-up (Table 2). In contrast, patients with extent progression had more ileocolonic microscopic involvement (45.5% vs. 15.9%, p = 0.001) and had higher rates of immunomodulatory treatment within the 1st year of follow-up (43.2% vs. 24.3%, p = 0.011) despite not having more active disease at diagnosis as assessed by activity indices (data not shown). In addition, patients with behavior progression had significantly lower albumin (3.56 vs. 3.7 g/dL, p = 0.026) and zinc (65 vs. 71 μg/dL, p = 0.014) levels.
Predictors of phenotype progression.
|Extent progression||Behavior progression||Perianal involvement progression|
|Among 157 patients with L1, L2 or isolated UGI at diagnosis||Among 174 patients with B1 at diagnosis||Among 183 patients with P0 at diagnosis|
|No, n = 113||Yes, n = 44||HR||P||No, n = 94||Yes, n = 80||HR||P||No, n = 140||Yes, n = 43||HR||P|
|Female n (%)||50 (44.2)||14 (31.8)||0.59||0.1||38 (40.4)||38 (47.5)||1.2||0.41||58 (41)||15 (34.9)||0.74||0.35|
|Familial history of IBD n (%)||21 (18.6)||9 (20.5)||1.02||0.97||18 (19.1)||15 (18.8)||0.93||0.79||22 (15.7)||12 (27.9)||1.83||0.08|
|EIM n (%)||21 (18.6)||7 (15.9)||0.93||0.86||17 (18.1)||19 (23.8)||1.46||0.56||29 (20.7)||9 (20.9)||1||0.96|
|Granuloma n (%)||33 (30.8)||12 (27.3)||0.87||0.67||26 (28.9)||24 (30.4)||1.04||0.88||42 (30.7)||15 (35.7)||1.2||0.58|
|Paris G1 n (%)a||27 (23.9)||9 (20.5)||0.84||0.65||18 (19.1)||23 (28.8)||1.47||0.13||35 (25)||11 (25.6)||1.1||0.77|
|Paris P1 n (%)a||15 (13.3)||5 (11.4)||0.77||0.77||12 (12.8)||10 (12.5)||0.97||0.92||–||–||–||–|
|Mild perianal disease (skin tags/fissures)||–||–||–||–||–||–||–||–||28 (20)||8 (18.6)||0.915||0.82|
|Paris B1 n (%)a||97 (86.6)||39 (88.6)||0.91||0.84||–||–||116 (83.5)||37 (86)||0.822||0.66|
|Disease location n (%)a||–||–||–||–||0.52||0.66|
|Isolated UGI (isolated L4)||–||–||–||–||7 (7.5)||7 (8.8)||1.13||14 (10)||2 (12.5)||0.51|
|Colonic (L2)||–||–||–||–||22 (23.7)||10 (12.5)||0.64||25 (17.9)||5 (11.9)||0.65|
|Ileal (L1)||–||–||–||–||46 (49.5)||43 (53.8)||1||0.52||67 (47.9)||23 (57.8)||1||0.66|
|Ileocolonic (L3)||–||–||–||–||18 (19.4)||20 (25)||1.12||34 (24.3)||12 (28.6)||1.04|
|Microscopic involvement n (%)||–||–||–||–||–||–||–||–|
|Ileal||74 (65.5)||16 (36.4)||1||0.001||–||–||–||–||–||–||–||–|
|Colonic||16 (14.2)||7 (15.9)||2.02||0.12||–||–||–||–||–||–||–||–|
|Ileocolonic||18 (15.9)||20 (45.5)||3.79||<0.001||–||–||–||–||–||–||–||–|
|Isolated UGI||5 (4.4)||1 (2.3)||1.18||0.87||–||–||–||–||–||–||–||–|
|Age at diagnosis—median (iqr)||13.8 (12.4–15)||13.6 (11.5–15.4)||1.03||0.62||27 (25–30)||27 (25–30)||1.02||0.47||13.5 (11.8–15.3)||14.17 (11.4–15.7)||1.05||0.35|
|Weight z score mean (sd)||−1 (1.04)||−0.9 (1.18)||1.11||0.51||−0.75 (1.08)||−0.79 (1.3)||0.95||0.64||−0.9 (1.07)||−1.1 (1.15)||0.875||0.37|
|Height—mean (sd)||−0.71 (1.09)||−0.72 (1.15)||1.02||0.92||6.98 (2.64)||7.4 (2.9)||1.03||0.43||−0.66 (1.21)||−0.75 (1.06)||0.93||0.624|
|BMI—mean (sd)||−0.73 (1.04)||−0.76 (1.26)||0.97||0.82||35 (25–50)||45 (32.5–50)||1.008||0.25||−0.71 (1.06)||−0.815 (1.23)||0.912||0.53|
|HBI—mean (sd)||6.82 (2.8)||7.15||1.02||0.67||23 (9.98)||25.5 (9.77)||1.01||0.26||6.82 (2.83)||7.63 (2.685)||1.09||0.123|
|shPCDAI—median (iqr)||40 (30–50)||42.5 (25–54)||0.99||0.95||4 (2–6)||4 (3–6)||1.02||0.38||40 (25–50)||45 (30–51.25)||1.01||0.31|
|abbrPCDAI—mean (sd)||23.43 (9.41)||24.62 (11.23)||1.007||0.64||11.2 (10.3–1.9)||11 (10.6–1.7)||1.01||0.92||22.1 (8.9)||25.4 (9.7)||1.03||0.08|
|Duration between onset & diagnosis—median (iqr)||4 (2–6)||4 (2–8)||1.01||0.68||10.1 (7.9–12.85)||10.6 (8.3–12)||0.97||0.3||4 (2.75–7.5)||4 (2–6)||0.983||0.6|
|1st line induction treatment n (%)
Steroids vs others
|56 (50)||25 (57)||0.785||0.68||46 (49.5)||48 (60)||0.725||0.16||73 (52.5)||26 (60.5)||0.725||0.3|
|1st line maintenance treatment n (%)
Imumomodulators vs others
|27 (24.3)||9 (43.2)||2.17||0.011||25 (26.9)||24 (30)||0.747||0.24||38 (27.5)||11 (25.6)||0.998||0.99|
a According to Paris classification, IBD = inflammatory bowel disease, EIM = extra intestinal manifestations, L1 = isolate ileal disease, L2 = isolated colonic disease, G1 = growth impairment, P1 = perianal disease (fistulas or abscess), B1 = inflammatory disease behavior, UGI = upper gastrointestinal involvement, iqr = interquartile range, SD = standard deviation, BMI = body mass index, HBI = Harvey–Bradshaw index, shPCDAI = short pediatric Crohn’s disease activity index, AbbrPCDA = abbreviated pediatric Crohn’s disease activity index.
3.2.5. Variables during follow-up associated with phenotype progression (including: extent, behavior and perianal progression)
Univariate analysis showed that phenotype progression was associated with Anti Tumor Necrosis Factor ɑ (anti-TNFɑ) treatment prior to the change in disease phenotype, higher number of flares and hospitalizations per year of follow-up. Other variable did not show a significant association (Table 3).
Variables during follow-up associated with phenotype progression.
|Extent progression||Behavior progression||Perianal involvement progression|
|Among 157 patients with L1, L2 or isolated UGI at diagnosis||Among 174 patients with B1 at diagnosis||Among 183 patients with P0 at diagnosis|
|No, n = 113||Yes, n = 44||P||No, n = 94||Yes, n = 80||P||No, n = 140||Yes, n = 43||P|
|Anti TNFa treatment prior to phenotype change n (%)||8 (7.1)||9 (20.5)||0.022||1 (1.1)||28 (35)||<0.001||0 (0)||20 (46.5)||<0.001|
|Surgery after exacerbation n (%)||51 (45.1)||17 (38.6)||0.48||21 (22.3)||54 (68.4)||<0.001||55 (39.6)||25 (59.5)||0.022|
|Smoking n (%)||12 (11.7)||7 (17.1)||0.42||9 (10.5)||13 (17.1)||0.26||14 (10)||9 (22.5)||0.058|
|Flares per years of follow-up median (iqr)||0.44 (0.18–0.71)||0.69 (0.36–1.09)||0.017||0.46 (0.19–0.74)||0.6 (0.36–1.08)||0.005||0.39 (0.18–0.68)||0.76 (0.59–1.25)||<0.001|
|Hospitalizations per years of follow-up median (iqr)||0.08 (0–0.23)||0.13 (0–0.46)||0.041||0.05 (0–0.19)||0.19 (0.09–0.4)||<0.001||0.086 (0–0.22)||0.25 (0.07–0.44)||0.005|
Anti TNFa = anti tumor necrosis factor ɑ, PPI = proton-pump inhibitors, iqr = interquartile range.
3.3. Change of phenotype by multivariate analysis
In multivariate Cox model analysis, neither BMI, activity indices, hemoglobin, CRP, albumin, initial therapy at diagnosis, gender nor duration between symptoms and diagnosis were associated with behavior progression, disease extension or the development of perianal disease. Older age at diagnosis, was associated with perianal disease development (HR 1.19, 95% CI 1.002–1.42, p = 0.048) but not with behavior or extent progression. Microscopic ileocolonic involvement was independent predictor for macroscopic ileocolonic extension (HR 4.32, 95% CI 1.93–9.67, p < 0.001).
The aim of this study was to assess the long term evolution of pediatric-onset CD phenotype in a tertiary cohort followed for more than 10 years. Our findings demonstrate that disease phenotype is dynamic and changeable even more than 10 years after initial diagnosis. As for disease location, 28% of patients with L1, L2 or isolated L4 progressed to more extensive disease (L3) over time. These results are concordant with finding demonstrated previously by Van Limbergen et al.  and are in contradiction to previous studies (mainly in adults) suggesting rates of location progression in CD of 9–16% , , and .
We also found that the prevalence of non-inflammatory disease (B2, B3 or B2B3) has increased dramatically from 18% at diagnosis to 53% at the end of follow-up. Similar rates were described by Lovasz et al.  and Margo et al.  suggesting a very high rate of non-inflammatory disease over time in childhood-onset CD patients. Timing of progression in our study is in accordance with previously reported data , , , and . The rate of perianal involvement progression in our study is also comparable to a previous pediatric cohort  thus strengthening the previously reported data. Another intriguing finding in our cohort was that the presence of anal fissures or skin tags at diagnosis was not associated with evolution of perianal abscesses or fistulas over time.
Interestingly, we found an association between non-inflammatory disease behavior and ileo-colonic distribution. These findings are in contradiction with most of the literature, which demonstrated an association between isolated ileal disease and non-inflammatory behavior  and . However, and in agreement with our findings, higher prevalence of penetrating disease had been associated with ileocolonic distribution in a previous study .
In the multivariate analyses we demonstrated that older age at diagnosis was marginally associated with the progression of perianal involvement though in agreement with 2 former reports it was not predictive of behavior progression  and .
Intriguingly and in contrast to previous studies , , and , we could not confirm that the presence of perianal disease is a prognostic factor for either behavior or extent progression, suggesting that perianal CD is a distinct phenotype not necessarily associated with penetrating disease. Moreover, disease extent had no impact on the overall risk of behavior progression. This is somewhat in contrast to previous reports , , and  in which behavior progression was more rapid in patients with isolated small bowel than those with colonic or ileocolonic disease.
In our cohort a significant proportion of patients (50%) had isolated ileal disease at diagnosis. This finding, which is discordant with previous data showing a lower rate of ileal disease in other pediatrics cohorts , , and , is comparable with higher prevalence of isolated ileal location among adult Jewish patients in Israel . This discrepancy might stem from ethnic associated variations in patient phenotypic and genotypic characteristics.
Among predictors at disease onset, we have demonstrated that presence of microscopic ileocolonic involvement predicts macroscopic extent progression. This finding has never been reported in either pediatric or adult studies and may serve as a marker for the need of more aggressive therapies.
While several studies evaluated predictors for phenotype progression in adults, studies assessing pediatric population are scarce. For example, our findings are not in accordance with a recently published study in adult CD patients, which showed that early treatment with azathioprine has an impact on the natural history of the disease . As well, in that study, there was a positive correlation between duration of treatment and lower likelihood of progression. Furthermore, we could not replicate their findings showing that upper GI involvement, male gender and corticosteroids treatment were associated with behavior progression. This discrepancy might stem from differences in the characteristics and natural history between adult onset and pediatric-onset CD  and . Additionally, early treatment with thiopurines may indeed ameliorate the natural history of the disease but on the other hand might represent a more severe phenotype requiring early treatment.
Among variables during the course of disease, number of flares and hospitalizations per year and treatment with anti TNFα agents were associated with behavior, extent and perianal progression over time, probably reflecting a more severe phenotype.
Previous studies have shown that anti TNFα treatment may prevent the development of disease complications during follow-up  and  and delay behavior progression . A possible explanation for these conflicting results lies partly in the fact that in our cohort most of the patients had not received anti TNFα treatments during the 1st year of follow-up. Therefore, earlier anti TNFα treatment in the “window of opportunity” period, as proposed by Peyrin-Biroulet et al. , could have altered our results and prevent phenotype progression.
Although the effect of smoking has been extensively investigated in IBD, there is some controversy regarding its role in disease phenotype progression. Several studies have demonstrated that smoking was associated with a greater likelihood of non-inflammatory disease , , and . However, this is not a universal finding. Similar to our findings, Aldhous et al. did not find an association between the deleterious effect of smoking and disease progression . This could be the result of differences in duration and intensity of smoking habits in teenagers and young adults or to differences in the natural history between adult and pediatric CD.
This study is however limited by several aspects: First, the retrospective design of the study did not enable complete data collection of some variables. Secondly, important data such as the timing of some of the medical therapies during the course of disease was not retrieved from medical charts. Lastly, in some patients, assessment of disease phenotype at end of follow up was not based on comprehensive endoscopic and imaging investigations. Naturally, in a retrospective study design, important data such as endoscopic/imaging indices cannot be retrieved. Nevertheless, the size of the examined population still enables us to analyze and define phenotype progression in our cohort.
In conclusion, we have demonstrated significant and sustained changes in pediatric CD behavior and extent over time. These changes are only weakly predictable. Novel clinically relevant finding of the present study is that macroscopic extent progression could be predicted by microscopic involvement at diagnosis in children with CD.
Overall, our data enrich the scarce pediatric literature on long term follow up and support a conservative approach to the validity of prognostic predictors across populations calling for caution when prognostic factors are used to predict disease progression in different geographic areas. New approaches using biomarkers to predict disease course are necessary .
Conflict of interest
The authors thank Prof. Yaron Niv, Chairman of the Gastroenterology Institute at The Rabin Medical Center, Petach-Tikva, Israel and Prof. Rami Eliakim, Chairman of the Gastroenterology Institute at The Sheba Medical Center, Tel Hashomer, Israel for enabling us to review charts of patients who were observed in their institutes.
-  M. Oliva-Hemker, S. Hutfless, E.S. Al Kazzi, et al. Clinical presentation and five-year therapeutic management of very early-onset inflammatory bowel disease in a large North American cohort. The Journal of Pediatrics. 2015;167:527-532 e3
-  G. Guariso, M. Gasparetto, L. Visona Dalla Pozza, et al. Inflammatory bowel disease developing in paediatric and adult age. Journal of Pediatric Gastroenterology and Nutrition. 2010;51:698-707 Crossref
-  N.A. Molodecky, I.S. Soon, D.M. Rabi, et al. Increasing incidence and prevalence of the inflammatory bowel diseases with time, based on systematic review. Gastroenterology. 2012;142:46-54 e42, quiz e30
-  P. Lehtinen, M. Ashorn, S. Iltanen, et al. Incidence trends of pediatric inflammatory bowel disease in Finland, 1987–2003, a nationwide study. Inflammatory Bowel Diseases. 2011;17:1778-1783 Crossref
-  B. Pigneur, P. Seksik, S. Viola, et al. Natural history of Crohn’s disease: comparison between childhood- and adult-onset disease. Inflammatory Bowel Diseases. 2010;16:953-961 Crossref
-  P.L. Lakatos. Prevalence, predictors, and clinical consequences of medical adherence in IBD: how to improve it?. World Journal of Gastroenterology. 2009;15:4234-4239 Crossref
-  B.P. Abraham, S. Mehta, H.B. El-Serag. Natural history of pediatric-onset inflammatory bowel disease: a systematic review. Journal of Clinical Gastroenterology. 2012;46:581-589 Crossref
-  C.I. de Bie, A. Paerregaard, S. Kolacek, et al. Disease phenotype at diagnosis in pediatric Crohn’s disease: 5-year analyses of the EUROKIDS Registry. Inflammatory Bowel Diseases. 2013;19:378-385 Crossref
-  J. Cosnes, A. Bourrier, I. Nion-Larmurier, et al. Factors affecting outcomes in Crohn’s disease over 15 years. Gut. 2012;61:1140-1145 Crossref
-  I.C. Solberg, M.H. Vatn, O. Hoie, et al. Clinical course in Crohn’s disease: results of a Norwegian population-based ten-year follow-up study. Clinical Gastroenterology and Hepatology. 2007;5:1430-1438 Crossref
-  E. Louis, A. Collard, A.F. Oger, et al. Behaviour of Crohn’s disease according to the Vienna classification: changing pattern over the course of the disease. Gut. 2001;49:777-782 Crossref
-  G. Vernier-Massouille, M. Balde, J. Salleron, et al. Natural history of pediatric Crohn’s disease: a population-based cohort study. Gastroenterology. 2008;135:1106-1113 Crossref
-  J. Van Limbergen, R.K. Russell, H.E. Drummond, et al. Definition of phenotypic characteristics of childhood-onset inflammatory bowel disease. Gastroenterology. 2008;135:1114-1122 Crossref
-  T.W. Eglinton, R. Roberts, J. Pearson, et al. Clinical and genetic risk factors for perianal Crohn’s disease in a population-based cohort. The American Journal of Gastroenterology. 2012;107:589-596 Crossref
-  M. Aloi, P. Lionetti, A. Barabino, et al. Phenotype and disease course of early-onset pediatric inflammatory bowel disease. Inflammatory Bowel Diseases. 2014;20:597-605 Crossref
-  H.J. Freeman. Long-term prognosis of early-onset Crohn’s disease diagnosed in childhood or adolescence. Canadian Journal of Gastroenterology. 2004;18:661-665
-  B.D. Lovasz, L. Lakatos, A. Horvath, et al. Evolution of disease phenotype in adult and pediatric onset Crohn’s disease in a population-based cohort. World Journal of Gastroenterology. 2013;19:2217-2226 Crossref
-  F. Magro, E. Rodrigues-Pinto, R. Coelho, et al. Is it possible to change phenotype progression in Crohn’s disease in the era of immunomodulators? Predictive factors of phenotype progression. The American Journal of Gastroenterology. 2014;109:1026-1036 Crossref
-  A. Levine, S. Koletzko, D. Turner, et al. ESPGHAN revised porto criteria for the diagnosis of inflammatory bowel disease in children and adolescents. Journal of Pediatric Gastroenterology and Nutrition. 2014;58:795-806
-  M.A. Shepanski, J.E. Markowitz, P. Mamula, et al. Is an abbreviated Pediatric Crohn’s Disease Activity Index better than the original?. Journal of Pediatric Gastroenterology and Nutrition. 2004;39:68-72 Crossref
-  M.D. Kappelman, W.V. Crandall, R.B. Colletti, et al. Short pediatric Crohn’s disease activity index for quality improvement and observational research. Inflammatory Bowel Diseases. 2011;17:112-117 Crossref
-  W.R. Best. Predicting the Crohn’s disease activity index from the Harvey–Bradshaw Index. Inflammatory Bowel Diseases. 2006;12:304-310 Crossref
-  A. Levine, A. Griffiths, J. Markowitz, et al. Pediatric modification of the Montreal classification for inflammatory bowel disease: the Paris classification. Inflammatory Bowel Diseases. 2011;17:1314-1321 Crossref
-  K.M. Tarrant, M.L. Barclay, C.M. Frampton, et al. Perianal disease predicts changes in Crohn’s disease phenotype-results of a population-based study of inflammatory bowel disease phenotype. The American Journal of Gastroenterology. 2008;103:3082-3093 Crossref
-  E. Louis, V. Michel, J.P. Hugot, et al. Early development of stricturing or penetrating pattern in Crohn’s disease is influenced by disease location, number of flares, and smoking but not by NOD2/CARD15 genotype. Gut. 2003;52:552-557 Crossref
-  J. Cosnes, S. Cattan, A. Blain, et al. Long-term evolution of disease behavior of Crohn’s disease. Inflammatory Bowel Diseases. 2002;8:244-250 Crossref
-  H.J. Freeman. Natural history and clinical behavior of Crohn’s disease extending beyond two decades. Journal of Clinical Gastroenterology. 2003;37:216-219 Crossref
-  J.S. Hyams. Standardized recording of parameters related to the natural history of inflammatory bowel disease: from Montreal to Paris. Digestive Diseases. 2014;32:337-344 Crossref
-  L. Peyrin-Biroulet, E.V. Loftus Jr., J.F. Colombel, et al. Early Crohn disease: a proposed definition for use in disease-modification trials. Gut. 2010;59:141-147 Crossref
-  M.D. Mandel, P. Miheller, K. Mullner, et al. Have biologics changed the natural history of Crohn’s disease?. Digestive Diseases. 2014;32:351-359 Crossref
-  E. Lindberg, G. Jarnerot, B. Huitfeldt. Smoking in Crohn’s disease: effect on localisation and clinical course. Gut. 1992;33:779-782 Crossref
-  S.S. Mahid, K.S. Minor, P.L. Stevens, et al. The role of smoking in Crohn’s disease as defined by clinical variables. Digestive Diseases and Sciences. 2007;52:2897-2903 Crossref
-  M.F. Picco, T.M. Bayless. Tobacco consumption and disease duration are associated with fistulizing and stricturing behaviors in the first 8 years of Crohn’s disease. The American Journal of Gastroenterology. 2003;98:363-368 Crossref
-  M.C. Aldhous, H.E. Drummond, N. Anderson, et al. Does cigarette smoking influence the phenotype of Crohn’s disease? Analysis using the Montreal classification. The American Journal of Gastroenterology. 2007;102:577-588l
-  G.R. Lichtenstein, S.R. Targan, M.C. Dubinsky, et al. Combination of genetic and quantitative serological immune markers are associated with complicated Crohn’s disease behavior. Inflammatory Bowel Diseases. 2011;17:2488-2496 Crossref
a Institute of Gastroenterology, Nutrition and Liver Diseases, Schneider Children’s Medical Center, Israel
b Sackler Faculty of Medicine, Tel-Aviv University, Israel
⁎ Corresponding author at: Schneider Children’s Medical Center, 14 Kaplan St., Petach-Tikva 49202, Israel. Fax: +972 3 9253014.
© 2016 Editrice Gastroenterologica Italiana S.r.l., Published by Elsevier B.V.