The article by Diak et al (1) in this issue of Arthritis & Rheumatism raises a very important set of questions for the practice of pediatric rheumatology. Is there a significant increase in the risk of malignancy in children with juvenile idiopathic arthritis (JIA) treated with tumor necrosis factor α (TNFα) blockers? Do the potential risks outweigh the benefits? Answering these questions requires a careful evaluation of the data presented.
Diak et al report 48 cases of malignancies occurring in children receiving TNFα blockers (infliximab, etanercept, and adalimumab). However, 25 of these cases occurred in children with inflammatory bowel disease (IBD), and all of these 25 children had received concomitant immunosuppressants. Although some controversy continues, most pediatric gastroenterologists now accept that there is an increased risk of a previously rare intestinal T cell lymphoma in children receiving TNFα blockers and concomitant purine analogs for IBD (2). Pediatric rheumatologists must be aware of these data, but this finding is not of direct relevance to children with JIA.
Twenty-three malignancies occurred in children receiving TNFα blockers who did not have IBD. Two of these children had received in utero exposure to TNFα blockers, 1 child had sarcoidosis, and 1 child had an unknown diagnosis. Nineteen of the 48 reported cases of malignancy associated with TNFα blockers occurred in children diagnosed as having JIA (including children diagnosed as having ankylosing spondylitis or psoriatic arthritis). It is important to evaluate children with JIA separately, because we know that adults with rheumatic disease experience an increased risk of lymphoma without exposure to TNFα blockers (3–5).
Are these 19 cases reported from around the world evidence of a significantly increased risk of malignancy in children with JIA receiving TNFα blockers compared with children with JIA not receiving them? The answer to this question is obscured by the absence of knowledge regarding the background rate of malignancy in children with JIA even without controlling for disease subtype or disease severity as is done in adult studies (3–5). The 19 cases of malignancy in children with JIA described by Diak et al are from diverse countries around the world, and these cases are reported without data about the sizes of the populations from which they are drawn or the background incidences of these malignancies in those populations.
When the data are restricted to well-described cases occurring in the US, only 2 malignancies are reported in JIA patients receiving infliximab. Both were receiving concomitant methotrexate; 1 developed Hodgkin's lymphoma and 1 developed a “mast cell malignancy.” One cannot conclude any pattern from this tiny data set. Five of the 14 children who developed malignancies while receiving etanercept for JIA were from the US (in a sixth child from the US who was receiving etanercept, the indication was unknown and the malignancy was bladder cancer). There were 2 cases of Hodgkin's lymphoma, 2 cases of leukemia, and 1 case of thyroid cancer in the 5 JIA patients. Again, the numbers are very small. I am excluding from consideration the single child in the US who developed Hodgkin's lymphoma while being treated with adalimumab for JIA because the child had previously received cyclophosphamide as well as rituximab, anakinra, infliximab, and etanercept. Thus, the assumption of a meaningful association with the use of adalimumab is doubtful.
It is difficult to extract a meaningful sense of the risk of malignancy in children with JIA who are receiving infliximab by examining the data shown in Table 1 of the article by Diak et al. The authors have used the manufacturers' estimated drug usage data as the basis for total years of treatment with infliximab and etanercept. The basis for these estimates is not made clear. Although Diak et al have excluded patients with hepatosplenic T cell lymphoma in part of the analysis, the data do not distinguish those using infliximab for JIA from those with other diagnoses.
For etanercept one can assume more safely that the majority of children had a diagnosis of rheumatic disease. There were 2 patients with Hodgkin's lymphoma, both of whom were receiving concomitant methotrexate. One patient had “acute myeloid leukemia and lymphoma.” Again, there is no consistent pattern. However, if one accepts the authors' data at face value, this gives an estimated lymphoma rate of 11/100,000, compared with a background rate of 2.4/100,000 in the general population. No data are available regarding the background rate of lymphomas in children with JIA with or without concomitant methotrexate. However, case reports of lymphomas in children with JIA who had never received TNFα blockers are easily found. Indeed, Cleary et al (6) report 6 cases of non-Hodgkin's lymphoma in children with JIA who had received methotrexate therapy.
Do the TNFα blockers increase the risk of malignancy in children with JIA over the background risk in children with JIA, especially the background risk in children with JIA who are receiving methotrexate, which was for many years the standard therapy for severe disease? There is insufficient evidence to assert that this is the case. There is no evidence of an unusual pattern of malignancy comparable with the occurrence of rare hepatosplenic T cell lymphomas in the children with Crohn's disease. Taken at face value, the authors' reported rate of lymphomas in children with JIA receiving etanercept is 3/26,800 patient treatment years, or roughly 1/9,000 patient treatment years. This small risk must be balanced against the probability that JIA alone or treated with methotrexate does increase the risk of lymphoma over background.
Is this risk of 0.011% sufficient to deny the remaining 99.98% of patients a class of medications that has dramatically reduced the morbidity of JIA (7)? Given the absence of convincing data that TNFα blockers increase the risk of lymphoma in children with JIA over the background risk seen with the prior standard of methotrexate therapy, every clinician must think long and hard before denying TNFα blockers to a child with JIA. What should a pediatric rheumatologist do when questioned by parents about the warning label required for Food and Drug Administration compliance? Parents will not be aware that the bulk of the cases were in children with IBD, nor will they be familiar with the implications of selective reporting. It is the responsibility of pediatric rheumatologists not to deny the concerns of parents, but to educate them properly and to make sure that the warning label placed in the packages of these medications is fully explained and put in proper context. Given that lymphomas are known to occur in children with JIA treated with methotrexate without a TNFα blocker, and given the many other possible complications of medications, pediatric rheumatologists cannot promise complete safety with any medication. The hypothesis that TNFα blockers may be associated with an increased risk of malignancy has not been disproved; conversely, however, there is no convincing evidence that the use of TNFα blockers in children with JIA is associated with an increased risk of malignancy beyond that due to the disease alone or the disease when treated with methotrexate.