Efficacy and Safety of Celecoxib on the Incidence of Recurrent Colorectal Adenomas: A Systematic Review and Meta-analysis
Cancer Management and Research
This study was performed as part of a systematic review that has been previously registered (PROSPERO CRD42015025849)27 and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses.28
Search strategy and study selection
We identified relevant studies through a systematic search of Medline and EMBASE until April 2018. In addition, we searched published systematic reviews for additional studies. The search strategy is provided in Table S1. Studies included were randomized controlled trials (RCTs) and long-term follow-up of RCTs that met the following inclusion criteria: participants were adults (aged ≥18 years) at an increased risk due to a previous history of adenomas who underwent polypectomy and with a documented clean colon before randomization; intervention was celecoxib at any dose; the comparator was placebo or control; and the outcome was the proportion of subjects who developed colorectal neoplasia, described as either colorectal adenomas or advanced adenomas.
Outcomes of interest
Primary efficacy outcomes of interest were the incidence of recurrent colorectal adenomas (advanced adenomas and any adenomas). Advanced adenoma was defined by one or more of the following features: 1 cm or larger, with villous or tubulovillous histology, with high-grade dysplasia, and/or with intramucosal carcinoma or invasive cancer.29 Any adenomas include both advanced and nonadvanced adenomas (defined as one or two small [<1 cm] tubular adenomas or serrated polyps without cytologic dysplasia)29 and invasive cancers.
Safety outcomes were the incidence of CRC, mortality due to any causes, serious adverse events, serious CV events, and renal and hypertensive disorders reported on any follow-up after randomization. Serious adverse events were defined as events resulting in death, hospital admission because of an adverse event, severe gastrointestinal bleeding, CV or non-CV complications, or discontinuation of intervention due to an adverse event or events that were defined as serious or severe by study authors. Serious CV events were defined as the composite of CV death, myocardial infarction, stroke, heart failure, thromboembolic event, or defined as serious CV event by the study investigators. Renal and hypertensive disorders included reports of elevated serum creatinine levels, fluid retention and edema, hypertension, proteinuria, and renal failure. We also evaluated the posttreatment effects of celecoxib on the incidence of recurrent colorectal neoplasia after discontinuing the intervention for >2 years.
Data extraction and quality assessment
Two reviewers (SKV and KGL) screened the relevant publications and then extracted data on the study, participants, and treatment-related characteristics onto a standardized form, and discrepancies were resolved by another author after group discussion. Data on efficacy outcomes were extracted with modified intention-to-treat analysis (ie, subjects who received at least one dose of celecoxib at any dose and had at least one colonoscopy after randomization). Data on safety outcomes were extracted by intention-to-treat principle, using the initial number of randomized participants allocated to each trial arm. Participants who were lost to follow-up were considered free of adverse events.
Previous evidence suggested that the effects of NSAIDs on adenoma recurrence may not be sustained after treatment cessation.30–32 Hence, we abstracted efficacy outcomes measured cumulatively at two time points, within 1 year of discontinuing intervention (primary efficacy analysis) and ≥2 years after discontinuing intervention (posttreatment effect analysis). Two reviewers (SKV and SMC) independently assessed the risk of bias (ROB) in the context of the primary outcome by using the revised Cochrane risk of bias tool (RoB 2.0).33
Data synthesis and statistical analysis
Meta-analysis was performed with DerSimonian and Laird random-effects model to estimate pooled risk ratios and 95% confidence intervals incorporating heterogeneity within and between studies, with Stata version 14.0 (StataCorp, College Station, TX, USA).34 Statistical heterogeneity between trials was assessed for primary outcomes using I2 statistics, with values >50% indicating substantial levels of heterogeneity.35,36 Publication bias could not be assessed due to the small number of included studies, which limited the power to distinguish between finding by chance and real asymmetry.37Subgroup analyses were performed for different dosings of celecoxib38 including 200 mg twice daily (400 mg/day), 400 mg once daily, and 400 mg twice daily (800 mg/day). Sensitivity analyses were performed based on the use of surveillance colonoscopy per protocol completer analysis (outcomes included only those subjects who underwent colonoscopy surveillance at the prespecified time period per protocol and excluded subjects who underwent a colonoscopic surveillance assessment before the expected surveillance interval), fixed-effect model, and trials with low ROB.
The Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach was used to rate the quality of evidence (high, moderate, low, and very low) of estimates derived from meta-analyses using GRADEpro version 3.6.1 (McMaster University, 2014).39
Risk–benefit integrated analysis
Similar to approaches used in previous meta-analysis,8 we used risk–benefit integrated analysis to review the potential benefits (prevention of recurrent advanced adenomas) and risks (serious adverse events and CV events) of celecoxib at 400 and 800 mg/day. To understand potential benefits at the population level, we used risk ratios derived from the placebo comparisons of celecoxib in the meta-analysis to estimate absolute risk of advanced adenomas with intervention. We used published pooled estimates from the National Cancer Institute pooling project to estimate population-level risks of advanced adenomas as assumed control risk (7.4% in low-risk group or population with history of nonadvanced adenomas and 16.3% in high-risk group or population with history of advanced adenomas).40 Similarly, to understand the potential risks, we used the pooled risk of serious adverse events in placebo groups in the meta-analysis (estimated using metaprop command in STATA) as a measure of baseline risk. We then used risk ratios derived from the placebo comparisons of celecoxib in the meta-analyses for serious adverse events and CV events to estimate absolute risk associated with celecoxib. We then presented excess benefit and risk of serious adverse events (over placebo) per 1,000 individuals who received treatment. Estimates of absolute risk were generated with the GRADEpro version 3.6.1 (McMaster University, 2014).39
We identified 391 records in which 11 potentially eligible articles were reviewed in full text. Of these, six articles were excluded mostly due to the lack of eligible population. Therefore, a total of five studies were included in our review. Among these five studies, three trials23,24,41 met the eligibility criteria for the quantitative analysis of primary outcomes. Post-trial results from three studies30,31,41were included for the analysis of posttreatment effect on the incidence of recurrent adenomas. The PRISMA flow diagram depicting the search and selection process for the primary outcomes is displayed in Figure S1 (see original version of this article).
Characteristics of the included studies
Table 1 describes the characteristics of three RCTs23,24,41 which reported the incidence of recurrent colorectal adenomas. A total of 4,420 participants with a previous history of adenomas who underwent polypectomy and with documented clean colon before randomization were included in the analysis. All trials were double-blinded and placebo-controlled. The treatment duration was 3 years in two trials23,24and 1 year in one trial.41 Postrandomization colonoscopy was performed within 1 year of discontinuing intervention in all trials. The dose per day of celecoxib used in two trials was 400 mg once daily,23,41and the remaining trial24 tested both 400 mg (200 mg twice daily) and 800 mg (400 mg twice daily) doses. A detailed description of ROB assessment among included RCTs is presented in Table S2 (see original version of this article). Among three RCTs, one trial41 was judged to be at high ROB and the remaining two trials23,24 were judged to be at low ROB in all domains.
(To view a larger version of Table 1, click here.)
Post-trial results from three studies30,31,41 were available to investigate the effect of celecoxib withdrawal on incidence of recurrent colorectal adenomas. Table 2 describes the identified studies. A total of 2,159 participants who completed follow-up colonoscopy after discontinuing intervention for 2–4 years were included in the analysis.
(To view a larger version of Table 2, click here.)