Withdrawal of drug therapy for patients with quiescent Crohn's disease
Abstract
This is a protocol for a Cochrane Review (Intervention). The objectives are as follows:
The primary objective of this review is to assess the feasibility and safety of discontinuing immunosuppressant or biologic drugs, administered alone or in combination, in patients with quiescent CD.
Background
Description of the condition
Crohn's disease (CD) is a chronic, relapsing and remitting, immune‐mediated disease of the gastrointestinal tract that if left untreated can cause significant morbidity and disability through progressive bowel damage. Typical symptoms include abdominal pain, chronic diarrhoea, weight loss and fatigue. The exact etiology of CD remains unknown, however a complex interplay between environmental factors, genetic susceptibility and intestinal microbiota likely cause the abnormal immune response and compromised epithelial barrier function that characterize the condition (Haag 2015).
Young adults are primarily affected, with the highest incidence of diagnosis reported during the second or third decades of life (Molodecky 2012). Epidemiological studies show a steady increase in the worldwide incidence of CD (Molodecky 2012). The highest incidence rates are reported in Australia (29.3 per 100,000 person‐years; Wilson 2010), Canada (20 per 100,000 person‐years; Bernstein 2006), New Zealand (16 per 100,000 person‐years; Gearry 2006) and Northern Europe (10 per 100,000 person‐years; Thompson 1998).
CD most commonly affects the terminal ileum and colon, yet any segment of the GI tract may be involved. Extra‐intestinal manifestations may occur in up to half of all patients (Harbord 2016). Inflammatory disease is the most frequent phenotype at diagnosis, however approximately 50% of patients eventually develop stricturing or penetrating disease which often necessitates surgery (Peyrin‐Biroulet 2012). Over 90% of CD patients develop endoscopic disease recurrence within 15 years of the first bowel resection (Buisson 2012).
Description of the intervention
Current treatment guidelines recommend a sequential 'step‐up' approach with initial use of first‐line drugs such as oral corticosteroids escalating to immunosuppressant and biologic agents as monotherapy or in combination if necessary (Colombel 2010a). Conversely, the 'step‐down' approach involves initially treating high‐risk patients and those with severe disease at diagnosis with immunosuppressant and biologic medications. Evidence in support of the step‐down strategy has accumulated over the last decade (D'Haens 2010). For example, in the SONIC trial (The Study of Biologic and Immunomodulator Naive Patients in Crohn’s Disease), 56.8% of patients treated with combination therapy achieved steroid‐free remission at week 26, as compared with 44.4% of patients receiving biologic alone, or with 30% of patients receiving azathioprine alone (Colombel 2010b). Findings from D'Haens 2008 also suggest that early combination therapy may be more effective than conventional management for induction of remission in Crohn's disease. At week 26, 60% of patients receiving combination infliximab and azathioprine achieved clinical remission compared to 23% of those who received corticosteroids followed in sequence by azathioprine and infliximab (D'Haens 2008). As a consequence, in this era of early intervention, there is an increasing number of patients being exposed to immunomodulators and biologics early in their disease course.
The feasibility of de‐escalation of therapy once remission is achieved is a common question encountered in clinical practice, driven by patient and clinician concerns around safety, adverse events, cost and national regulations. Long‐term exposure to thiopurines has been associated with the development of lymphoproliferative disorders (Beaugerie 2009), and cancer of the skin and urinary tract (Bourrier 2016; Peyrin‐Biroulet 2011). Tumor necrosis factor‐alpha (TNF‐alpha) antagonist therapies may increase the risk of infections (Billioud 2013), and conflicting evidence about a potential increase in the risk of melanoma exists (Andersen 2014; Long 2012). Patients receiving combination therapy may have an increased risk of developing hepatosplenic T cell lymphoma (Subramaniam 2014), non‐Hodgkin’s lymphoma (Siegel 2009), and infections. The potential long‐term consequences of exposure to biologic agents remains unknown. It is estimated that TNF‐alpha antagonists account for 64% of the direct costs of CD treatment (van der Valk 2014).
Questions have been raised as to whether discontinuing immunosuppressant and biologic therapies in patients with quiescent CD may limit adverse events and reduce the high healthcare costs associated with these therapies while maintaining remission.
How the intervention might work
Withdrawal of immunosuppressant and biologic drugs in patients with quiescent CD may limit adverse events and reduce healthcare costs. Alternatively, ceasing these drug therapies may result in negative outcomes such as disease relapse, drug desensitization, bowel damage and need for surgery.
Why it is important to do this review
A systematic appraisal of the literature is required to highlight the potential benefits and risks of withdrawing immunosuppressant and biologic drug therapy in CD patients who have achieved remission.
Objectives
The primary objective of this review is to assess the feasibility and safety of discontinuing immunosuppressant or biologic drugs, administered alone or in combination, in patients with quiescent CD.
Methods
Criteria for considering studies for this review
Types of studies
Randomized controlled trials (RCTs), controlled clinical trials, and prospective cohort studies will be considered based on the inclusion criteria. Included studies must have followed patients for a minimum duration of six months after drug discontinuation.
Types of participants
Adults (age > 18 years) with CD (as defined by conventional clinical, endoscopic or histologic criteria) who achieved remission (as defined by the study) while receiving immunosuppressant or biologic drugs administered alone or in combination.
Types of interventions
Discontinuation of immunosuppressant or biologic drugs, administered alone or in combination, following a period of maintenance therapy of at least six months. The comparison will be usual care.
Types of outcome measures
Primary outcomes
The primary outcome will be the proportion of patients who relapse (as defined by the included studies) following discontinuation of immunosuppressant or biologic drugs, administered alone or in combination.
Secondary outcomes
Secondary outcomes will include (where available): 1) the proportion of patients who respond to the reintroduction of immunosuppressant or biologic drugs, given as monotherapy or combination therapy; 2) the proportion of patients who require surgery following relapse; 3) the proportion of patients who require hospitalization for CD following relapse; 4) the proportion of patients who develop new CD‐related complications (e.g. fistula, abscesses, strictures) following relapse; 5) the proportion of patients with elevated biomarkers of inflammation (CRP, fecal calprotecin) in those who stop and those who continue therapy; 6) the proportion of patients with anti‐drug antibodies and low serum trough drug levels; 7) time to relapse and 8) the proportion of patients with adverse events, serious adverse events and withdrawal due to adverse events.
Search methods for identification of studies
Electronic searches
We will search the following electronic databases:
1. MEDLINE (Ovid, 1946 to present);
2. EMBASE (Ovid, 1984 to present);
3. The Cochrane Central Register of Controlled Trials; and
4. The Cochrane IBD Group Specialized Register.
The search strategies are listed in Appendix 1. The search will not be limited by language, year of publication or type of publication.
Searching other resources
We will also search the reference lists of potentially relevant trials and papers to identify additional studies. Conference proceedings from the last five years of Digestive Disease Week, European Crohn's and Colitis Organisation annual meeting, and United European Gastroenterology Week will also be searched to identify studies reported in abstract form only.
Data collection and analysis
Selection of studies
Two authors (JT and RB) will independently screen titles and abstracts identified by the literature search to determine eligibility based on the inclusion criteria described above. Any disagreement among authors will be resolved through discussion until consensus is reached. If consensus cannot be reached, a third author (JFC) will be consulted to resolve the disagreement.
Data extraction and management
A data extraction form will be used to collect information from the relevant studies. Two authors (JT and RB) will independently extract data. Disagreements will be resolved through consensus. In the case of disagreement, a third author (JFC) will be consulted. The following data will be retrieved from included studies:
1) general information: title, journal, year, publication type;
2) study information: study type, study design, setting, inclusion/exclusion criteria, methods of randomization, concealment of allocation and blinding, study duration, and definitions of remission and relapse;
3) population characteristics: number of participants recruited, total number of patients screened and randomized, total number of participants followed, baseline characteristics (e.g., age, sex, race, disease phenotype, disease duration, concurrent medications; prior medications)
4) intervention characteristics: type and dose of medication discontinued, duration of remission prior to discontinuation, duration of therapy prior to discontinuation
5) follow‐up (length of follow‐up, withdrawals, number of patients lost to follow‐up); and
5) outcomes: primary and secondary outcomes, as described above.
Assessment of risk of bias in included studies
The methodological quality of each included RCT will be independently assessed by two authors (JT and RB) using the Cochrane risk of bias tool (Higgins 2011a). Any disagreements will be resolved by consensus with input from a third author (JFC). Factors to be assessed will include:
1) sequence generation (i.e., was the allocation sequence adequately generated?);
2) allocation sequence concealment (i.e., was allocation adequately concealed?);
3) blinding (i.e., was knowledge of the allocated intervention adequately adequately prevented during the study?);
4) incomplete outcome data (i.e., were incomplete outcome data adequately addressed?);
5) selective outcome reporting (i.e., are reports of the study free of suggestion of selective outcome reporting?); and
6) other potential sources of bias (i.e., was the study apparently free of other problems that could put it at high risk of bias?).
Studies will be assigned a low risk of bias, high risk of bias, or unclear risk of bias for each category.
The quality of non‐randomized, non‐controlled studies will be assessed using the Newcastle‐Ottawa scale (NOS) (Wells 2017).
The quality of the total body of evidence for the primary outcomes of interest will be assessed using the GRADE criteria (Schünemann 2011). Randomized trials are considered high quality and will be downgraded due to:
1) risk of bias;
2) indirect evidence;
3) inconsistency (unexplained heterogeneity);
4) imprecision; and
5) publication bias.
The overall quality of evidence for each outcome will be classified as high quality (i.e. further research is very unlikely to change our confidence in the estimate of effect); moderate quality (i.e. further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate); low quality (i.e. further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate); or very low quality (i.e. we are very uncertain about the estimate).
Measures of treatment effect
Review Manager (RevMan 5.3.5) will be used to analyze data. All data will be analyzed on an intention‐to‐treat (ITT) basis. The risk ratio (RR) and corresponding 95% confidence interval (CI) will be calculated for dichotomous outcomes. For continuous outcomes, the mean difference (MD) and corresponding 95% CI will be calculated. For studies that report the log hazard ratio and standard error for time‐to‐event outcomes, we will calculate the hazard ratio (HR) and corresponding 95% CI using the generic inverse‐variance method.
Unit of analysis issues
When studies report multiple observations for the same outcome, the outcomes will be combined for fixed intervals of follow‐up (e.g., clinical remission at eight weeks). Cross‐over trials will be included if data are available from the first phase of the study (i.e., before any cross‐over). Where studies allocated participants to more than one treatment arm, then these arms will be pooled for the primary analysis. When possible, subgroup analyses will be performed to compare efficacy and safety among different doses of drugs. Although some studies may report more than one efficacy or safety event per patient, the primary analysis will consider the proportion of patients who experienced at least one event.
Dealing with missing data
When possible, the original study authors will be contacted in the case of unclear or missing data. Data that remains missing will considered to be a treatment failure, in accordance with the ITT principle. Where appropriate, sensitivity analyses will be conducted to assess the impact of including unclear data on the effect estimate.
Assessment of heterogeneity
Heterogeneity will be assessed using the Chi2 test (a P value of 0.10 will be considered statistically significant) and the I2 statistic. An I2 value of 25% indicates low heterogeneity, 50% indicates moderate heterogeneity and 75% indicates high heterogeneity (Higgins 2003). We will use sensitivity analyses to explore potential explanations for heterogeneity. If the I2 statistic shows a moderate to high degree of heterogeneity and the Chi2 test is statistically significant, forest plots will be visually inspected for obvious outliers and sensitivity analysis will be performed excluding the outlier to see if this explains the heterogeneity.
Assessment of reporting biases
Potential reporting bias will be evaluated by comparing outcomes listed in protocols to published manuscripts. If the protocols are not available, we will compare outcomes listed in the methods section of published manuscripts to those described in the results section. If a sufficient number of studies are included (i.e. > 10) in the pooled analyses, we plan to investigate potential publication bias using funnel plots (Egger 1997).
Data synthesis
Data from individual trials will be combined for meta‐analysis when the interventions, patient groups and outcomes are sufficiently similar (as determined by consensus). The pooled RR and 95% CI will be calculated for dichotomous outcomes. For continuous outcomes, we will calculate the pooled MD and corresponding 95% CI. The standardized mean difference (SMD) and 95% CI will be calculated when difference scales are used to measure the same outcome. If studies report estimates of log hazard ratios and standard errors obtained from Cox proportional hazards regression models, we will pool study results using the generic inverse‐variance method to obtain the pooled HR and corresponding 95% CI. If a mixture of log‐rank and Cox model estimates are obtained from the studies, we will convert the log‐rank estimates into log hazard ratios and standard errors (Higgins 2011b). A fixed‐effect model will be used to pool data unless heterogeneity exists between the studies. A random‐effects model will be employed if heterogeneity exits (I2 50 to 75%). We will not pool data for meta‐analysis if a high degree of heterogeneity (I2 ≥ 75%) is detected.
Subgroup analysis and investigation of heterogeneity
Planned subgroup analyses include: studies investigating children and studies investigating adults; different drug doses and routes of administration; and pregnancy.
Sensitivity analysis
Sensitivity analyses will examine the impact of the following variables on the pooled effect: random‐effects versus fixed‐effect modelling; low risk of bias only versus unclear or high risk of bias; and relevant loss to follow‐up (> 10% versus < 10%).
