Patients with Crohn's disease (CD) or ulcerative colitis (UC) are frequently placed on antiinflammatory and immunosuppressive therapies, including aminosalicylates, corticosteroids, 6-mercaptopurine (6-MP), azathioprine, methotrexate, calcineurin inhibitors, and antitumor necrosis factor (anti-TNF) antibodies. Recently, important safety concerns have been raised regarding the use of thiopurines and biologics to treat these two diseases. Studies have suggested that these medications are associated with an increased risk of lymphoma.1–4 One meta-analysis postulated a 4-fold increased risk of lymphoma in patients with inflammatory bowel disease (IBD) treated with thiopurines, but could not conclude whether it was a result of the severity of the disease, treatment effects, or a combination of the two.5
While some studies have suggested an increased risk of non-Hodgkin's lymphoma in IBD patients on immunosuppressive agents, the precise risk of lymphoma in this population remains unclear. One study in adults suggested that the increased risk of lymphoma in adults with IBD is almost exclusively due to infection with Epstein–Barr Virus (EBV).2 More recently, a rare, aggressive, and often fatal type of lymphoma, hepatosplenic T-cell lymphoma (HSTCL), has been reported in children and young adults with IBD, particularly those treated with combination thiopurine and infliximab.6, 7 While these case reports have raised considerable concern among parents of children with IBD, no published pediatric studies have attempted to estimate risk by calculating the number of incident cases of lymphoma as a ratio to the years of medication exposure.
To date, there remains a lack of primary data regarding the precise risk of lymphoma in children with IBD. In order to provide further information regarding the risk of lymphoma in children with IBD, we conducted a single-center retrospective study involving over 1300 patients seen in our IBD center over a 30-year period.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
This was a single-center retrospective study involving the review of 1560 records of children and young adults with IBD who were cared for at Children's Hospital Boston (CHB) from August 1, 1979 to January 1, 2009. Patients were included in the study if they were cared for at Children's Hospital Boston for more than 3 months. The diagnosis of CD, UC, or IBD unclassified (IBDU, i.e., “indeterminate colitis”) was established utilizing standard clinical, radiographic, and endoscopic criteria. Patients were excluded from the review if they did not have a confirmed case of IBD, if they were a second-opinion patient not followed at CHB for at least 3 months, or if medical records were unavailable. Data were abstracted from hospital and outpatient charts utilizing a standardized case report form. Our case report form included the patient's birth dates, gender, date of diagnosis, IBD type (CD, UC, or IBDU), and phenotype as determined by the Montreal Classification System. For phenotype, the phenotype at the last period of follow-up was utilized.
The total time of follow up (in months) was calculated from the date the patient was first seen at CHB to either the date they left the practice or until the date the chart review was completed (for patients currently followed at CHB). To track medication use we ascertained if the patients had ever received the medication and calculated the duration (in months) they had received the medication during the period of follow-up. We obtained information about medication usage for the following therapies: aminosalicylates, methotrexate, azathioprine or 6-MP, cyclosporine A, tacrolimus, and anti-TNF antibodies (infliximab, adalimumab, or certolizumab pegol). The total duration of medication used for a specific therapy during the study period was calculated in months. If a patient was continually on a medication from the initiation of therapy until the end of the study period the months of medication exposure were calculated from the beginning to the end of the study period. However, if a patient discontinued a treatment for a period of time and then restarted that therapy later, only the months of medication exposure were included in the analysis. Because of particular interest in the potential role of combination therapy as a risk factor for lymphoma we also calculated the number of months patients were concomitantly using a thiopurine and anti-TNF antibody.
Because our hospital is a tertiary care oncology center and sister institution to the Dana-Farber Cancer Institute, any IBD patients with suspected malignancies are diagnosed and treated by our hospital staff. Thus, we have access to medical records for all IBD patients who have been diagnosed with lymphoma during their period of follow up at CHB. In patients diagnosed or suspected with lymphoma we reviewed the record to ascertain the type of lymphoma, any IBD medications that preceded the lymphoma diagnosis, and the therapy for the tumor.
Patient years of exposure to various medications, total time of follow up in patient years, and the ratio of lymphomas to years of exposure were all observed and recorded. In order to assess drug therapy and risk we calculated two different “denominators” of drug exposure (as shown in Table 3). “Patient-years taking medication” reflects the aggregate total number of years patients were actually treated with a therapy. In contrast, “Patient-years observed” reflects the aggregate total duration of follow-up in any patient ever exposed to that therapy. For example, a 15-year-old male diagnosed at age 11, treated for 20 months with 6-MP, then changed to methotrexate and followed for 36 additional months would be recorded as having 20 months of “Patient-years taking 6-MP,” and 36 months of “Patient-years taking methotrexate,” but 56 months of total “Patient-years observed” in both the 6-MP and methotrexate columns.
Differences in prevalence of demographics across IBD diagnoses were examined using the Pearson chi-Square test or analysis of variance. Patient-years of exposure to various medications, total time of follow-up expressed in patient-years, and the ratio of number of lymphomas to years of exposure were calculated from data abstracted from patient medical records. These data were analyzed with SAS (SAS/STAT v. 9.2 of the SAS System for Windows, SAS Institute, Cary, NC). Relative rates were calculated as standardized incidence ratios (SIR) by comparing the observed lymphoma rate to the expected (Hodgkin's and non-Hodgkin's) lymphoma rate of 0–19 year old children from the Surveillance Epidemiology and End Results (SEER) database using the STATA “IR” command (STATA 11.2, College Station, TX).
- Top of page
- MATERIALS AND METHODS
The data presented above represent our attempt to ascertain the lymphoma risk in our population of children and young adults with IBD treated over a 30-year period. In our cohort of 1374 patients who received a total of 6624 patient-years of follow-up, we identified two cases of lymphoma, for an overall risk of 3.0 lymphomas per 10,000 patient-years. In comparing our observed lymphoma incidence rate with the National Cancer Institute SEER database for Hodgkin and non-Hodgkin lymphoma in males and females 0–19 years of age, there is a suggestion that thiopurines may increase the incidence of lymphoma. The absolute incidence rate of lymphoma for patients treated with thiopurines was 4.5 per 10,000 patient years compared to the expected rate in SEER of 0.58 per 10,000 patient years with an SIR of 7.51 (95% CI 0.74–41.98). These results suggest that there may be an increased risk in children receiving thiopurines, but the risk did not reach statistical significance in this large cohort.
How different risk factors contribute to the increased prevalence of lymphoma in IBD remains unknown. Lymphomas occur when immune dysregulation or antigen driven B lymphocyte proliferation result in a monoclonal population of B or T lymphocytes. EBV, which immortalizes B cells in vitro, is clearly a lymphoma risk factor, particularly in patients immunosuppressed with thiopurines or calcineurin inhibitors.2 Lymphomas have been seen both with primary EBV infection and on occasion with EBV reactivation; in these cases, EBV genome may be identified within the tumor.8, 9 In some inflammatory disorders (including systemic lupus and Sjogren syndrome), the inflammatory state itself may be a predisposing factor to lymphoma. However, a large meta-analysis by Lewis et al10 failed to demonstrate an association between IBD itself and lymphoma. Other potential contributing factors may include genetic mutations in lymphocytes that predispose to lymphoma (as seen in X-linked lymphoproliferative syndrome), and ionizing radiation utilized in the diagnosis and therapy of IBD patients.11, 12
Studies in adults with IBD have shown varying results regarding the lymphoma rate in association with certain IBD treatment therapies.13–15 The primary studies show conflicting results, with some studies demonstrating no increase in lymphoma, and others demonstrating an increase.2, 3, 16, 17 A meta-analysis of a subset of these studies by Kandiel et al5 suggested an increased risk of lymphoma in patients receiving thiopurines (pooled relative risk 4.18, 95% CI 2.07–7.15). A separate meta-analysis by Siegel et al4 demonstrated a relative risk of 6.1 per 10,000 patient years in patients who had received biologics, but the majority of these patients also had exposure to immunomodulators. The most definitive study associating thiopurines with an increased risk of lymphoma was by Beaugerie et al,18 which evaluated 19,486 patients for a follow-up of 49,713 patient years. In that study 23 lymphomas were identified, and 15 occurred in patients treated with thiopurines. However, the study was largely restricted to adults and only four of the lymphomas described occurred in patients under 35 years of age. The authors concluded that patients receiving thiopurines had an increased hazard risk of 5.28 (95% CI 2.01–13.9) compared to patients not treated with thiopurines.
While the prevalence of lymphoma in our study did not reach statistical significance, the risk reported in our study for patients taking thiopurines is similar to both the Kandiel et al meta-analysis and the Beaugerie et al cohort. In contrast, a large matched case–control study from the Netherlands failed to show an increased prevalence of lymphoma, once lymphomas triggered by EBV were excluded.19 A rare and severe form of lymphoma, hepatosplenic T cell lymphoma, has also been reported primarily in adolescents and young adults with IBD treated with a combination of thiopurines and biologics.6, 7, 20 We did not identify any such cases in our patient population despite several thousand patient-years of observation, emphasizing the rarity of this life-threatening malignancy.
Hemophagocytic lymphohistiocytosis (HLH), a lymphoproliferative disorder seen in immunocompromised patients, has been associated with a number of infections, including cytomegalovirus, parvovirus, and EBV.21, 22 Blank et al23 have recently reported an association between thiopurines and the development of HLH in children with IBD. Although HLH was not the focus of our study, we identified a similar case in our population. Based on these data, we would encourage the temporary discontinuation of thiopurines (if feasible) in patients with active EBV infection until the infection has resolved.
This study is the largest reported pediatric cohort addressing the question of malignancy in IBD. Strengths of this study include the fact that we have a large sample size (particularly for a pediatric study), and a high number of patient-years of follow-up. Our hospital is also a pediatric tertiary care center specializing in both IBD and oncology and is nationally and regionally recognized. Therefore, since patients diagnosed with lymphoma stay at our center for diagnosis and treatment, this reduces the likelihood that patients we followed were diagnosed and treated elsewhere. Even though we did find two lymphomas associated with thiopurine use in our cohort, the rarity of this complication should be reassuring to patients who receive these agents, and to physicians who prescribe them.
However, our study has some limitations. The primary limitation of the study is that even with over 6000 patient-years of follow-up, the study may still be underpowered to detect such a rare event as lymphoma. Thus, even though our results raise the question of an increased risk of lymphoma in this population, the finding did not reach statistical significance. However, we hope that additional centers analyzing their pediatric cohorts and comparing them to ours might promote further research in this area, and perhaps facilitate a large meta-analysis in the future. This study did not assess adherence to medication, which may be another confounder of drug toxicity and efficacy. Another limitation is our lack of follow-up on the observed patients once they left our center. It is therefore possible that some patients may develop lymphoma after transitioning their care to another institution, or to adult providers. However, the fact that both of our patients developed lymphoma within 5 years of starting thiopurine therapy argues against a long latency period before risk is increased.
In summary, the overall risk of lymphoma in children with IBD is low, even in those receiving thiopurines. However, our study suggests children receiving thiopurines may have a slightly increased risk, comparable to that reported in studies of adults. Published research suggests that “hard data” regarding risks and benefits is of value to patients who need to make difficult decisions about treatment.24 The investigators hope that this study will encourage other large pediatric centers to conduct similar studies, so that physicians caring for children with IBD can more clearly communicate medication risks to their patient families.