The Ondansetron Premedication Trial in Juvenile Idiopathic Arthritis

1. Purpose:

   To evaluate if routine pre-medication with the anti-emetic ondansetron reduces methotrexate intolerance and increases the proportion of children with JIA able to continue taking methotrexate, resulting in a better quality of life and more cost-effective medication use.

2. Hypothesis:

   Prophylactic prescription of ondansetron with methotrexate will increase the proportion of children that remain on methotrexate and intolerance free one year after starting methotrexate, relative to prescription of ondansetron after intolerance symptoms develop.

3. Justification:

   Methotrexate intolerance is thought to be largely the result of Pavlovian conditioning secondary to previous exposure and symptoms can be triggered by associated stimuli, such as a yellow liquid similar in color to methotrexate or the smell of alcohol cleansing swabs. This intolerance leads to poor adherence to optimal dosing or to stopping the medication altogether. Children who cannot adhere to optimal dosing or stop methotrexate may require expensive biologic medications to control their JIA. Methotrexate intolerance often leads to use of anti-emetic medications which, unfortunately, rarely reverse conditioned responses. In oncology anti-emetics are used prophylactically as premedication in all subjects receiving chemotherapy to prevent the establishment of conditioned intolerance.

   By conducting a registry-based pragmatic adaptive superiority randomized clinical trial the investigators will include most subjects in whom the treatment may be used under the usual conditions of practice to demonstrate effectiveness under real word circumstances. Patients enrolled in a pragmatic trial are more representative because eligibility criteria are less strict. More so, a registry-based pragmatic RCT can also be much cheaper than a traditional RCT.

4. Objectives:

The investigators will assign patients starting low-dose methotrexate for JIA 1:1 online using the CAPRI Registry and block randomization by Canadian region and patient weight to one of two groups:

Intervention: Routine premedication with oral ondansetron (2, 4 or 8 mg for patient weights <15Kg, 15-30Kg, >30 Kg; 3 doses a week).

Control: Oral ondansetron at the same dosing, prescribed only to those patients who develop methotrexate-induced nausea/vomiting during usual care.

The investigators will compare:

1. The proportions of patients remaining on methotrexate with no intolerance between the two groups one year after starting methotrexate (primary objective).

2. Safety and tolerability

3. The cumulative incidence of:

   1. methotrexate intolerance

   2. attainment of inactive disease

   3. biologic medication initiation

4. The mean quality of life scores and methotrexate intolerance severity scores between the two groups 4-8 months after starting methotrexate.

   The investigators will also collect information on impacts on quality of life and medication utilization that will enable a future cost-effectiveness analysis.

5. Research Design:

   The investigators will conduct a CAPRI JIA Registry-based pragmatic adaptive superiority RCT as per the following methods.

   Registry-based: The trial will use infrastructure already in place for the CAPRI JIA Registry.

   Pragmatic: Broad inclusion criteria and simple outcome assessments compatible with usual care.

   Adaptive: This is a Planned Sample-Size Re-Estimation Adaptive Trial as described by Bhatt & Metha [34].

   Superiority: Designed to test superiority (as opposed to non-inferiority) of ondansetron premedication.

   RCT: Treatments assigned by block randomization and analyses adjusted for post-randomization imbalances

6. Statistical Analysis:

The primary outcome statistic is the ratio of two proportions (relative risk, RR), the proportion of children remaining on methotrexate with no intolerance in the intervention group divided by the same in the control group. This choice allows flexibility for the frequency of follow-up visits to be set-up as per clinical need, obviating the need for study-specific visits, increasing feasibility and decreasing costs. It is also not impacted by the times at which methotrexate intolerance or discontinuation occur which leads to simplicity in interpretation and analysis.

The sample size calculation is based on expressing the RR in the log scale and using established formulas for standard errors. With p<0.05, power of 90% and expected methotrexate continuation with no intolerance of 50% in the control group and 75% in the intervention group the result is 79 evaluable subjects per group. The investigators will aim to recruit 176 subjects total to allow for up to a 10% dropout rate, although they expect a dropout rate of <5% based on previous CAPRI studies.

A preliminary analysis after recruitment of 90 subjects will consist of a two-sided confidence interval (CI) for the RR of continuing on methotrexate with no intolerance, where the confidence level is adjusted for information accrual as per Schoenfeld.

The Data and Safety Committee will assess preliminary results according to the following guidance:

If the CI is entirely below a relative risk of 1.2, the study will be stopped for futility.

If the CI is entirely above a relative risk of 1.2, the study will be stopped and superiority will be claimed.

If the CI includes a relative risk of 1.2, the trial will continue until the re-calculated full target sample is attained.

The RR of 1.2 has been selected on clinical grounds. In a prospective study of 142 children with JIA starting methotrexate followed for one year, Van Dijkhuizen et al reported that 59 patients developed intolerance and 11 discontinued methotrexate for other reasons, for a total of 72 (50.7%) remaining on methotrexate with no intolerance. An increase in this proportion to <60% is deemed too small to justify the additional costs and risks of adding prophylactic ondansetron.

The final analysis will be an intention to treat analysis conducted after the final sample size is achieved or at the time the study is stopped. All subjects enrolled in the study at that time will complete a one-year follow-up and will be included in the final analysis. Logistic regression models will be used to adjust effect estimates for post-randomization imbalances in the two groups (intervention, control). A secondary per-protocol analysis will be conducted on those subjects who received the assigned intervention for at least 3 months.