We included 28 patients who were diagnosed with CD, and had been followed for at least 36 months, at the Samsung Medical Center between March 1998 and January 2009. The patients were given infliximab to treat conventional therapy-resistant CD and severe active CD for induction of remission were identified. A retrospective chart analysis was conducted. Biopsies were obtained by endoscopy in all eligible patients. CD was diagnosed in accordance with the European Society for Pediatric Gastroenterology, Hepatology, and Nutrition-Porto criteria [17]. The efficacy of treatment was assessed by comparing the pediatric Crohn's disease activity index (PCDAI) scores. Infectious diseases, such as tuberculosis, were ruled out by taking a detailed family history, reviewing imaging studies that had been performed, and confirming the presence of a negative, purified-protein derivative test result with a negative PCR hybridization for Mycobacterium tuberculosis on biopsy tissue. The study was approved by the institutional review board of our institution (2012-04-051).

Enrolled patients were divided into two groups according to the treatment regimen. Ten patients refractory to conventional therapy, including eight with steroid-resistant CD and two with steroid dependent CD, were included in the 'step-up' group for infliximab treatment. Eighteen patients with moderate-to-severe CD were assigned to the 'top-down' group. In the 'step-up' group, oral corticosteroids (prednisolone, 1-2 mg/kg/day) were used for induction therapy. Mesalamine (50-80 mg/kg/day) or azathioprine (2-3 mg/kg/day) was provided for the conventional treatment as maintenance therapy. Infliximab (5 mg/kg) was administered by intravenous infusion at weeks 0, 2, and 6 in combination with daily azathioprine, and this course was repeated every 8 weeks for 10-12 months thereafter.

The 'top-down' group received infliximab and azathioprine for induction and maintenance therapy for the first year, and were treated with azathioprine after 2 years. The group treated with early infliximab had not been previously treated with other medications, such as corticosteroids or immunomodulators. All patients were followed for at least 36 months.

We defined disease remission as a PCDAI score of less than 10 points and relapse as a score greater than 10 points [18]. The relapse rate was defined as the rate of the presence of relapse, more than once, after a remission was achieved with treatment. Moderate to severe disease was defined as having a score greater than 30 points. The two groups were compared with regard to baseline characteristics, adverse events, clinical status (including the PCDAI score) and relapse rates at 1, 2, and 3 years.

Adverse events and laboratory results were investigated to identify potential side effects. Patients treated with mesalamine underwent evaluations for hypersensitivity, rash, alopecia, anorexia, headache, elevated liver enzymes, and pancreatitis. For azathioprine, patients were evaluated for pancytopenia, pancreatitis, hepatotoxicity, rash, alopecia, anorexia, and arthralgia. Patients receiving infliximab were assessed for anaphylaxis, dyspnea, rash, headache, nausea, elevated liver enzymes, pancreatitis, pancytopenia, and serious infections. Evaluation of well-known adverse effects of prednisolone was not included in the analysis.

Peripheral blood samples of 5 mL each were collected from patients with at least three previous infliximab infusions. A commercially available enzyme-linked immunosorbent assay (ELISA) kit (Matriks Biotek Laboratories, Ankara, Turkey) was used to detect antibody to infliximab (ATI). The assay was based on a highly specific double antigen binding assay principle. In this procedure, the specific antibody in serum, namely ATI, binds to the infliximab-coated well with one of its Fab arms and also binds to the biotinylated infliximab with the aid of its other Fab arm. The ELISA kit demonstrates the presence of ATI, and the intensity of the color developed is proportional to the amount of specific antibody [19].

Analyses comparing the study groups were performed using the χ² test for variables. Fisher's exact test was used to compare each of the measurements. Logistic regression analysis was used to compensate for difference in baseline characteristics. A p<0.05 was regarded as statistically significant.

The study population at diagnosis included 24 males (85.7%) and four females (14.3%) with a mean age of 12.9±3.9 years. The location of disease was the terminal ileum in two patients (7.1%), the colon in four patients (14.2%) and the ileocolon in 24 patients (78.6%). The average total number of infliximab infusions for each patient was 10.3±3.1. The mean PCDAI score at initial diagnosis was 40.5±11.3 (Table 1).

PCDAI scores at initial diagnosis, before infliximab treatment, and the total number of infliximab infusions were not significantly different the two groups. The age at diagnosis was 10.2±5.4 years in the 'step-up' group and 14.4±1.3 years in the 'top-down' group. The duration from the initial diagnosis to infliximab infusion was 49.6±5.2 weeks and 1.8±2.4 weeks for the 'step up' and 'top-down' groups, respectively (Table 2).

The relapse rate of two sex groups was not significant different (p=0.295); 66.7% (16 of 24 patients) for male patients and 100% (four of four patients) for female patients. There was no significant difference in the relapse rate between the different age groups (p=0.295); 100% (four of four patients) for patients less than 12 years of age and 66.7% (16 of 24 patients) for those more than 12 years of age. The relapse rate in patients who only had colon involvement was 50.0% (two of four patients). The relapse rate in patients with terminal ileum only involvement was 100% (two of two patients), and for patients with ileocolonic involvement the relapse rate was 72.7% (16 of 22 patients). There was no difference in the relapse rate between patients with mild disease, with a PCDAI score less than 30 points at diagnosis (two of four patients, 50%), and moderate to severe disease, with a PCDAI score more than 30 points (19 of 24 patients, 64.3%) (Table 3).

The baseline characteristics of age and duration from the initial diagnosis to infliximab infusion (week) were significantly different between the two groups. These variables were confounding factors, and the relapse rate of the 'top-down' group was lower than that of the 'step-up' group.

The relapse rate at 1 year was 16.7% in 'top-down' group (3 out of 18 patients), 50% in 'step-up' group (five of 10 patients). At 1-year, the absolute difference between two groups was 33.3%. There was not a significant difference between the groups (p=0.091). At the 2-year follow-up, the relapse rate in 'top-down' group was 50% (9 out of 18 patients) and 90% in the 'step-up' group (nine of 10 patients). At 2 years, the absolute difference between two groups was 40%. There was a significant difference between the groups (p=0.048). At the 3-year follow-up, the relapse rate of each group was 61.1% and 90%. There was no significant difference (p=0.194) (Fig. 1).

Ten patients were checked for the presence of ATI. We divided the patients into two groups according to PCDAI at sampling time: 'maintained response to infliximab' and 'loss of response to inflixmab'. ATI was not detected in either group. Table 4 summarized ATI detection results.

Two patients treated with mesalamine had anorexia. Eleven patients treated with azathioprine experienced adverse events (leukopenia in nine and pancreatitis in two). For patients treated with infliximab, only one patient in the 'step-up' group had an adverse event, consisting of dyspnea and tachycardia that occurred after the third infusion.