Development of Crohn Disease During Anti–TNF-α Therapy in a Child With Juvenile Idiopathic Arthritis

Abstract

Tumour necrosis factor alpha (TNF-α) is believed to play a key role in initiating and amplifying the inflammatory reaction in the intestinal mucosa of patients with Crohn disease (1–3). High levels of TNF-α are secreted by lamina propria T lymphocytes and monocytes in patients with CD, and abundant TNF-α is even detectable in the stools (3–5). TNF-α is a transmembrane protein with an intracellular tail. Once cleaved by a metalloproteinase, soluble TNF-α is secreted (6). Based on the role of TNF-α in experimental colitis and the high levels of TNF-α found in the mucosa during active disease, therapeutic strategies have been developed to block TNF activity. In fact, neutralizing anti-TNF antibodies have proved to be highly effective in otherwise treatment-resistant CD (7–9). The hypothesis that the beneficial effect of anti–TNF-α antibodies is a result of neutralization of soluble TNF-α was recently questioned. For instance, long-term remissions which sometimes occur after one single injection of infliximab cannot be explained solely by the neutralization of soluble TNF, given the pharmacodynamics of this antibody. Furthermore, different types of anti-TNF molecules with high affinity to soluble TNF-α are more or less effective in treating patients with CD(3). We report a boy with juvenile idiopathic arthritis (JIA) who experienced CD while taking anti-TNF medication (etanercept) for his arthritis and who had a severe relapse of his intestinal/perianal disease during a second treatment course with etanercept. This child is highly responsive to anti-TNF treatment with infliximab, indicating that a more profound immunoregulatory effect than simple TNF neutralization is needed to down-regulate the mucosal inflammation in CD. CASE REPORT This Caucasian boy was born at term to healthy, unrelated parents. Development during the first 2 years was normal. The family history was not contributory. There was no history of autoimmune or intestinal disorders. At 2.5 years, he presented with an oligoarticular form of JIA involving both ankles and hands, according to the Durban criteria (10). At the time of diagnosis, antinuclear antibodies were positive (1/200) with negative rheumatoid factor. There was no iridocyclitis. Treatment with nonsteroidal anti-inflammatory drugs controlled symptoms for the next 1.5 years. Subsequently the course of JIA was characterized by chronic oligoarthritis. Methotrexate therapy (MTX 0.6 mg/kg/week) was initiated at 4 years, with complete remission for several months. Several episodes of joint relapse responded well to combined therapy with oral prednisone and MTX, and intra-articular triamcinolone hexacetonide (THX) injections. At 6 years of age, he first experienced abdominal pain and episodic diarrhea and vomiting. These symptoms initially were attributed to MTX but symptoms persisted despite discontinuation of MTX. Unfortunately arthritis worsened because treatment with prednisone alone was insufficient to maintain remission. Therefore, treatment with etanercept (ENBREL®, 0.4 mg/kg twice a week) was started and there was an immediate remission of JIA (Fig. 1). The intestinal symptoms, however, worsened dramatically, with massive diarrhea, abdominal pain, weight loss, and a perianal abscess that rapidly became an anocutaneous fistula. At that time, the development of an associated intestinal inflammatory disease was suspected.FIG. 1.: Medications and clinical course of patient from age 2 to 10 years. The width of shaded areas is a qualitative indicator of severity of symptoms.Endoscopic evaluation showed ulcerations of the terminal ileum, no gross involvement of the colon, moderate gastritis, and duodenitis. A small bowel contrast radiograph confirmed the involvement of the terminal ileum. No stenosis was seen. Infectious causes of the intestinal inflammation were ruled out and Crohn disease was confirmed by the presence of epithelioid granulomas in both ileal and colonic biopsies. Given the aggravation of the intestinal symptoms during etanercept therapy, this drug was discontinued and mesalamine (5-ASA, 50 mg/kg/day) therapy was started. 5-ASA was only partially effective in controlling intestinal symptoms and not at all effective for the joint symptoms, which again requiring intra-articular THX injections. At 8 years, after a relapse of the anocutaneous fistula and poor control of the ileal inflammation, enteral nutrition associated with azathioprine (2 mg/kg/day) therapy was started. This therapy was effective for the intestinal and perianal symptoms, allowing weight gain and even a limited catch-up growth. However, it was insufficient to prevent several relapses of JIA, which were treated with increasing doses of prednisone. Given the steroid-dependency and the high number of joint relapses, etanercept 0.4 mg/kg twice a week was re-introduced, and remission of the joint symptoms was again observed. At that point, enteral nutrition, which had been in use for more than 12 months, was discontinued because the child refused to accept further naso-gastric feedings. Despite the immunosuppressive medication with azathioprine and etanercept, bloody diarrhea, perianal fistula, and weight loss recurred rapidly. Laboratory evaluations included: erythrocyte sedimentation rate, 48 mm/hour; C-reactive protein, 100 mg/L; and albumin, 32 g/L. Immunoglobulin A anti-Saccharomyces cerevisiae antibodies (ASCA) were weakly positive (26 U/mL) as was the pANCA. Antinuclear antibodies remained positive (1/400) and the rheumatoid factor was negative. Endoscopic evaluation at 9.5 years confirmed the diagnosis of CD with gastric, duodenal, ileal, and perianal involvement. Re-introduction of enteral nutrition was refused by the child, and etanercept therapy was stopped. A new treatment strategy equally effective for the intestinal and joint inflammatory disease had to be chosen. At this time, the patient’s TNF-α levels were elevated in serum (24–99.5 pg/mL; normal, <20 pg/mL) and stool (135 pg/g; normal, <30 pg/g). Infliximab was introduced. The response to the initial induction treatment, consisting of three injections (5 mg/kg) at 0, 2, and 6 weeks, was excellent and equally effective for joint and digestive/perianal symptoms. Stool levels of TNF-α rapidly dropped to normal (<30 pg/g). This patient has now been on maintenance therapy with infliximab for more than 6 months (injection of 5 mg/kg every 8 weeks). To date, the clinical response of both digestive and articular disease has been excellent, with complete remission after each injection. TNF-α levels remain within the normal range in stool and serum. DISCUSSION We report a child who developed CD during etanercept therapy of a chronic form of oligoarticular JIA. This observation is particularly interesting in light of the role of TNF-α as a major mediator and amplifier of the intestinal inflammation in CD (3,11,12) and the joint inflammation in JIA (1,13,14). Various animal models of RA have shown that TNF plays a key role in the pathogenesis of joint lesions (1,13). Soluble TNF-α levels in the synovial fluid in patients with rheumatoid arthritis are elevated (15,16), and neutralization of this cytokine is successful in inducing remission (13,17,18). By binding TNF-α, etanercept has been recognized to induce complete remission in patients with rheumatoid disorders as well as JIA (17,18). However, etanercept failed to ameliorate or inhibit the development of CD in our patient despite effective TNF-neutralizing doses as evidenced by the control of his JIA symptoms. Unfortunately, we do not have any tissue or stool TNF-α levels from our patient during the initial treatment period with etanercept at the time he first experienced symptoms of inflammatory bowel disease. However, given the beneficial effect on the joint inflammation, the anti-TNF effect seemed to be adequate and a systemic TNF-blocking effect can be supposed. This child’s case indicates that inhibition of the biologic effects of circulating TNF alone is not sufficient for treatment of CD as opposed to JIA. In fact, a recent double-blind, placebo-controlled trial in 43 adult patients with CD revealed that the administration of etanercept was not effective in moderate to severe CD (19). This finding is in clear contrast to the effect of the anti–TNF-antibody infliximab. Targan et al. (7) showed in the first controlled clinical trial a short-term effect of a single perfusion of infliximab in 50% to 81% of adults with otherwise treatment-resistant active CD. Similar data have been obtained when treating adult or pediatric patients with a fistulizing form of CD (8,9). The nature of etanercept and infliximab is different. Etanercept is a genetically engineered fusion protein consisting of two molecules of the natural p75 TNF receptor and the Fc domain of human IgG1, whereas infliximab is a chimeric mouse-human antibody directed against TNF-α (20). The affinity of both drugs is extremely high for soluble TNF-α; however, membrane-bound TNF is not recognized by etanercept, whereas infliximab also binds to this uncleaved form (21). The different TNF binding pattern may explain the distinct biologic effects of these anti-TNF drugs. A recent study by Van den Brande et al. (21) indicated that infliximab was effective in treating CD whereas etanercept was not. By binding to activated peripheral blood lymphocytes and lamina propria T cells or monocytes, infliximab induces apoptosis of these cells (21,22). This results in the selective elimination of activated T cells or monocytes, explaining a sustained effect of one single perfusion of infliximab for several weeks. In contrast, etanercept does not have any lymphotoxic effects. As in our patient, the sustained effect of infliximab is accompanied by the induction of remission of both gut and joint disease, whereas the elimination of TNF-α by etanercept was not effective in treating CD. In conclusion, this observation underlines the differential effect of the two anti-TNF therapies in CD.