In this study, we investigated the factors affecting clinical relapse and estimated the optimum duration and time of cessation of treatment in moderate to severe pediatric CD treated with combination therapy. We identified four factors related to elevated risk of clinical relapse; IFX cessation, formation of ATI, low IFX TL and low 6-TGN level. Interestingly, withdrawal of AZA was not directly associated with clinical relapse in pediatric CD patients with sustained CR of at least two years and who had achieved deep remission.
In clinical practice, physicians may encounter questions about the feasibility of therapeutic de-escalation in CD patients who have achieved deep remission, on account of safety, adverse events and cost. There is a paucity of data related to the optimum duration and the appropriate time to cease medication after achieving remission in patients with CD, particularly in pediatric patients.
It is now known that early introduction of biologics has tremendous advantages in terms of maintaining tight control of inflammation and catching the therapeutic window 16,17. A randomized controlled trial (RCT) revealed that co-treatment with IFX and AZA significantly increases the rate of CR and mucosal healing compared to AZA monotherapy in CD patients 4. In addition, recent studies have reported that endoscopic and histological mucosal healing should be treat-to-target in CD patients instead of symptomatic control to achieve better outcomes and minimize future complications18. With the development of treatment strategies, clinicians are being confronted with a crossroads decision about when to stop or reduce the therapy dose once patients achieve remission 19.
A systematic review of cessation studies for patients with inflammatory bowel disease concluded that about 50 percent of patients who withdrew anti-TNF-α agents after combination therapy with immunomodulators maintained remission for 24 months 20. Reenaers et al. reported that among patients with CD who withdrew IFX in stable CR state, twenty-one percent did not restart biologics including IFX, and sustained CR for seven years after infliximab cessation 21. Therefore, it seems reasonable to conclude that there may be a subgroup of patients who are good candidates for treatment withdrawal.
Our data that IFX cessation in patients with CD was associated with a high risk of clinical relapse is consistent with the results of other previously published studies 22,23,24,25. A recent retrospective study conducted in Korea on adults evaluated the long-term outcomes following cessation of anti-TNF-α treatment in inflammatory bowel disease (IBD) patients with CR 24. After cessation of anti-TNF-α treatment for CD patients, the cumulative relapse rates at 1, 3, and 5 years were 11.3%, 46.7%, and 62.5%. In this cohorts, mucosal healing rate before discontinuation of anti-TNF-α treatment in CD patents was 22.5%. Another recent retrospective study performed by Luca and colleagues investigated the clinical course after anti-TNF-α treatment discontinuation in selected pediatric patients with IBD who achieved deep remission similar to our study 25. Among these patients, relapse-free survival rates at 12, 24, and 36 months for CD were 83.3%, 71.1%, and 23.7%, respectively. In contrast, none of the patients with CD who maintained anti-TNF-α treatment after achieving mucosal healing experienced clinical relapse 24. Compared with previous study performed by Luca in which only 12.5% of patients were treated with combination therapy, all of our subjects were treated with combination therapy and IFX TLs at the timing of IFX withdrawal or last follow-up were found to be 3 µg/mL or higher, confirming that IFX effect was sufficient. Even with these stringent selection criteria, IFX withdrawal in pediatric CD is still inadvisable.
However, the association between discontinuation of AZA and clinical relapse is controversial. In one reported study, which was focused on CD in CR under AZA-IFX combination therapy, AZA cessation was associated with a high risk of relapse in those patients treated with combination therapy for less than 27 months 26. In contrast, two other RCTs and one meta-analysis suggested that discontinuing AZA from a combination therapy regimen may not differ in clinical relapse rates compared to continuing with combination therapy 8,20,27. However, both RCTs included subjects comprised primarily of those who had previously failed immunomodulators monotherapy. It could be argued that in this cohort, the subsequent cessation of this agent would not be expected to have a significant effect compared to a patient cohort in which patients were initially treated with combination therapy 9. Nevertheless, most studies of patients with CD who discontinued AZA after combination therapy revealed that there were no differences between AZA withdrawal from combination therapy and continuation of combination therapy in terms of clinical relapse. In addition, the European Crohn’s and colitis organization guideline provide the opinion that AZA withdrawal in patients treated with combination therapy is inappropriate in patients with high risk/refractory disease or in patients at risk of biologic failure 28. In other words, it can be said that in patients with sustained deep remission, controlled disease activity and low risk of biologic failure, azathioprine could be withdrawn.
When subjects were treated with combination therapy and maintained CR for at least two years, mucosal healing was observed in 82.7% and deep remission was achieved in 41.3% in our study. Based on these findings, we attempted drug cessation assuming that most of the patients who satisfied both conditions had received sufficient treatment and had a low risk of relapse. In our study, AZA cessation was not shown to increase the risk of clinical relapse (HR 1.078, 95% CI = 0.327–3.55, P = 0.9021). This result could be partially explained by another study which revealed that withdrawal of AZA after at least six months of combination therapy does not reduce the TL of IFX in patients with CD 29. The mean durations of AZA therapy were 38 ± 19.3 months. We also wanted to know the change in the risk of relapse according to the treatment duration with AZA, but the number of patients who discontinued AZA was small, so further statistical analysis was impossible.
According to our results and those of other reports, IFX TL and 6-TGN level affect clinical relapse 30,31. High drug concentrations within the therapeutic range could be a factor in lowering the risk of relapse. Pursuing a high therapeutic concentration may lead to concerns about adverse drug reactions. However, using therapeutic drug monitoring (TDM) for personalizing therapy for CD patients, drug concentration could be maintained within the therapeutic range. The average IFX TL and 6-TGN level during the follow-up period were 6.1 ± 5.4 µg/mL and 208.5 ± 114.2 pmol/8×108 RBC respectively. Since patients who were poor metabolizers of AZA and primary non-responders to IFX were excluded from this study, the relationship between higher drug concentrations above therapeutic range of 6-TGN and occurrence of adverse events was not addressed in this study.
In adults, the development of ATI occurs in up to 65.3% in patients with IBD 32, while ATI have been reported in 8–43% of pediatric patients with IBD 33. The results of our study are consistent with these data, as the formation of ATI was detected in 12.0% of patients receiving combination therapy. The development of ATI can neutralize IFX by direct binding of neutralizing antibodies, or accelerating the clearance of the IFX by the binding of non-neutralizing antibodies 34,35. These mechanisms lower IFX TL and can lead to loss of response during IFX administration 35. Similarly in our study, patients with ATI formation had a lower relapse free survival rate than those with negative ATI (Fig. 3C).
It is well known that concomitant use of immunomodulators reduces ATI formation, and improves the pharmacokinetics of IFX 8,37,38. However, other factors affecting ATI formation were not clearly identified. We specifically asked whether there is a relationship between the factors affecting clinical relapse and ATI formation. According to our analysis, only 6-TGN level was associated with an increased risk of formation of ATI and this result is consistent with another study 39.
Although the present study is retrospective and relatively small-sized, it sought a more analytical approach to drug cessation and drug concentration. Drug-related factors were analyzed as time-varying covariates rather than time-independent covariates. Conventional survival analysis is generally applied to the time-independent data, where the exposure variables of interest are often treated as time-fixed 40. However, values of these exposure variables can vary over time and time-fixed analysis may cause bias over time potentially altering the conclusions of the study. To the best of our knowledge, none of the studies to determine when to withdraw drugs in patients with CD have used time-varying covariates 8,9,21,23,26,28,29. The factors which are known to affect relapse such as 6-TGN level, IFX TLs, ATI formation, and timing of drug cessation are all variables that change with time after patient observation starts, therefore, the time-varying covariate approach is preferable for survival analysis in our study. This method increased the reliability of the study because it avoids biases associated with different timing of drug cessation in different patients.
The current study has a few limitations. First, this was a single-center, retrospective study and, consequently, had relatively unstructured follow-up of patients with certain limitations compared to studies using prospective design. However, all subjects received regular examinations such as endoscopy and biopsy on the same principles and IFX TLs or 6-TGN were examined at regular intervals. Therefore, the extraction of clinical disease activity using PCDAI, laboratory results, endoscopic and histologic results from medical records was possible at all outpatient visits. Although the decision to withdraw either IFX or AZA was not randomly assigned, however there was no difference in disease activity and baseline. Moreover, our findings are meaningful because these are real-world data from a single center cohort. Second, selection bias might have been introduced because the number of participating patients was relatively small and the observation period was not as long as might be desired. Also, because of adverse events such as lymphoma that may occur in male patients with long-term use of AZA, the group who withdrew AZA had a statistically larger number of male patients than the group who discontinued IFX, resulting in a selection bias of gender. Third, since multiple time-varying covariates made multivariate analysis too complicated to perform, only univariate results were presented.
Despite these limitations, our study reports the outcome of drug cessation in strictly selected cohort who sustained sufficient CR and achieved deep remission in real clinical settings. Maintaining the 6-TGN level and IFX TL high within the therapeutic range lowers the risk of clinical relapse and IFX cessation and ATI formation increase the risk of clinical relapse. In conclusion, even when applied in pediatric patients who meet strict criteria after a sufficient CR period and deep remission, IFX cessation in pediatric CD should be considered more carefully. However, withdrawal of AZA could be contemplated in selected pediatric patients with CD sustaining CR for at least two years and achieved deep remission.