Adverse events during oral colchicine use: a systematic review and meta-analysis of randomised controlled trials

Background Colchicine is a widely used drug to treat inflammatory diseases. Due to its long historical use in medicine, controlled clinical trials have been small and there remains some caution with the use of this drug in patients with co-morbidities. The aim of the study is to systematically examine the side effect profile of colchicine in controlled clinical trials across all published indications. Methods A systematic review was conducted in accordance with PRISMA methodology. The Cochrane Library, MEDLINE and EMBASE were searched for double-blind controlled trials of oral colchicine in adult patients that reported adverse event data. Meta-analyses were used to determine the relative risk (RR) of adverse events in colchicine users compared to comparator groups. Results A total of 4915 studies were initially identified and after exclusions, 35 randomised controlled trials with placebo (n = 35) or active comparators (n = 5) were included. The most common diseases studied were gout, liver cirrhosis and pericarditis. There were a total of 8659 pooled participants, 4225 participants were randomised to receive colchicine, 3956 to placebo and 411 to an active comparator. Diarrhoea was reported in 17.9% of colchicine users versus 13.1% in comparator groups (RR 2.4, 95% confidence interval (CI) 1.6, 3.7). Any gastrointestinal event was reported in 17.6% of colchicine users and 13.1% of comparators (RR 1.7, 95% CI 1.3, 2.3). Adverse liver events were reported in 1.9% of colchicine users versus 1.1% in the comparator groups (RR 1.6, 95% CI 0.9, 3.0). Muscle events were reported in 4.2% of colchicine users and 3.3% in the comparator groups (RR 1.3, 95% CI 0.8, 1.9). Haematology events were reported in 0.6% of colchicine users and 0.4% of comparator groups (RR 1.34 (0.64, 2.82). No study reported neuropathy events. Other sensory events were reported in 1.1% of colchicine users and 1.5% of comparator groups (RR 1.4, 95% CI 0.3, 6.7). Infectious events were reported in 0.4% of colchicine users and 2.1% of comparator groups (RR 1.0, 95% CI 0.7, 1.5). No study reported death as an adverse event. Conclusion Colchicine increases the rate of diarrhoea and gastrointestinal adverse events but does not increase the rate of liver, sensory, muscle, infectious or haematology adverse events or death.


Introduction
Colchicine is an anti-inflammatory agent which is widely used for the treatment of gout and also used extensively for familial Mediterranean fever, Behcet's disease and pericarditis. Its use in the management of gout has increased due to the widespread recommendation that it be used as a gout flare prophylaxis when urate-lowering therapy is initiated [1]. It is used continuously for long periods of time in individuals with familial Mediterranean fever and Behcet's disease. However, due to its long historical use in medicine, it has not been subjected to the same registration trials that contemporary medicines require. There remains uncertainty regarding its use in certain risk groups including those with kidney and liver impairment, at higher doses, and with CYP3A4 inhibitors [2]. It has previously been used in an intravenous preparation, but this is no longer used due to the adverse safety profile of this administration method [3].
Although the adverse event profile of colchicine has been reported in various individual clinical trials and for single indications like pericarditis [4], it has not been studied systematically to our knowledge. The aim of this study was to examine the adverse events of colchicine reported in randomised controlled trials using a systemic review and meta-analysis methodology.

Methods
This study was conducted according to a pre-defined protocol using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [5].

Search strategy
Electronic databases (Cochrane Library, MEDLINE and EMBASE) were searched from inception to November 2019 using the following key words in the search term: colchicine AND ((randomised controlled trial [ NOT humans [mh])). Bibliographical references from individual included studies and review articles were also hand-searched to identify additional relevant papers. All studies generated from the search were exported to Ref-Works and screened to remove duplicates. Title and abstract screening, followed by full-text screening, was undertaken by a single reviewer (KA). Randomised controlled trials were included if they compared the effects of colchicine in patients, administered for any indication, to placebo or active comparators. Trials were included if they: had a double-blinded design, included oral colchicine in at least one of the treatment arms; involved adult participants; were published in the English language; and reported adverse event data in both the colchicine and comparator group(s) in relation to the number of participants with adverse events per group. Studies which reported the number of participants with adverse events resulting in study withdrawals and did not report the number of adverse events which occurred in the participants who remained in the study, were excluded. There was no publication date restriction. If multiple reports described the same trial, the most recent full-text publication was selected for inclusion.

Quality assessment
Quality of all included studies was assessed independently by two reviewers (SS, KY) using the 6-item modified-Jadad scale which assesses reported randomisation, blinding, withdrawals, dropouts, inclusion/exclusion criteria, adverse events and the statistical analysis [6]. The scale has a maximum score of 8 points, with low quality studies yielding scores of 0 to 3 and highquality studies yielding a score of 4 to 8. Any disagreements in the quality assessment were resolved by discussion of the two reviewers. If necessary, a third reviewer (ND) was involved to reach consensus.

Data extraction
Two reviewers (SS, KY) independently extracted data from the full-text studies using a Microsoft Excel extraction form. Any disagreements were resolved by consensus with a third reviewer (ND). Data extraction included publication details (author, year of publication, country of first author), disease state, participant characteristics (sample size, ethnicity, mean (SD) age and n % of male participants) and details of the trial (study design, length of follow up, primary outcome, interventions and dosages, intervention length). Extraction of data related to adverse events included the total number of participants with any adverse event per group and the total number of participants with each individual reported adverse event. Data for the number of deaths were extracted only if death was related to an adverse event (rather than worsening of disease).

Data-analyses
For the purpose of data analysis, adverse events were grouped under eight pre-defined categories: diarrhoea, gastrointestinal events (including diarrhoea), liver events, haematology events, muscle-related events, sensoryrelated events (including neuropathy), infectious events, death and any adverse event. In situations when studies reported the number of participants with ≥ 2 individual adverse events which both came under the same category (i.e. "nausea" and "vomiting" which both come under the gastrointestinal event category), data from the adverse event with the highest number of participants was used for that category. Only studies which reported diarrhoea as a separate event were included in the diarrhoea category. Studies which reported diarrhoea as part of a combined event (i.e. 'diarrhoea or nausea') were included under the 'gastrointestinal events' category.
Meta-analyses were undertaken to determine the relative risk of adverse events in the colchicine group compared to the comparator groups (pooled comparators, placebo and active-comparators). Relative risk was calculated based on the number of participants with adverse events. Random effects models were used for all I 2 values > 0%. As caution is recommended when colchicine is used continuously in those with liver impairment, a sensitivity analysis was undertaken excluding the studies involving participants with cirrhosis or sclerosing cholangitis [6][7][8][9][10][11]. For the 'any adverse events' category, meta-analyses were also used to determine the effects of disease indication (liver diseases, gout, Behcet's and related conditions, pericarditis and related conditions and other), duration of exposure to the intervention (with subgroups defined as ≤ 2 weeks, 1 to 2 months, 3 to 5 months, 6 to 12 months and ≥ 24 months), the average daily dose of colchicine (with subgroups defined as < 1 mg, 1 mg, > 1 < 2 mg, ≥ 2 mg) and the cumulative daily dose of colchicine (with subgroups defined as < 50 mg, 50 to < 100 mg, ≥ 100 mg to 300 mg, > 600 mg). In papers which used different colchicine doses based on participant weight categories, the highest daily colchicine dose was used to determine that study's subgroup. Subgroup comparisons were made using the Phet statistic (the P value derived from the chi-square test of heterogeneity for subgroup differences).
All meta-analyses were undertaken in Review Manager 5.3 with an alpha level of 0.05. Only studies specifically reporting an adverse event as being present or absent were included in the meta-analyses. However, as this may over-or under-estimate the true occurrence of adverse events, the proportion of participants with specific adverse events was also computed in relation to the total number of participants included in all studies (i.e. if not reported, '0' events were considered to have occurred). These data were used for descriptive purposes only and not meta-analysed.

Study characteristics
A total of 4915 studies were identified through the search following the deletion of duplicates (Fig. 1). After title and abstract screening, 70 full-text articles were assessed for eligibility. After the exclusion of 35 studies (reasons for exclusion are presented in Fig. 1), a total of 35 randomised-controlled double-blind studies were included in this review. The majority of studies were placebo-controlled (n = 30, 83%) and 5 (17%) studies were active-comparator controlled. The majority of studies were parallel-group designs and 4 studies [12][13][14][15] were cross-over designs.

Quality assessment
The results from the quality assessment are presented in Supplementary Figure 1. Overall, the modified-Jadad score indicated high quality (total score 4 to 8) for all studies. An appropriate method of randomisation and blinding was reported in 24 (69%) and 28 (80%) studies, respectively. Thirty (86%) studies provided an adequate description of withdrawals and dropouts and 33 (94%) provided a participant inclusion and exclusion criteria. The method used to assess adverse events was described by 22 (63%) studies and methods of statistical analysis by 32 (91%).

Intervention characteristics
Of the 8659 pooled participants, 4225 participants were randomised to receive colchicine, 3956 to the placebo and 411 to an active comparator. The remaining 67 participants were included in cross-over trials and received both colchicine and placebo treatments over the duration of the study [12][13][14][15].

Adverse events
Methods used in the included studies to assess adverse events are described in Supplementary Table 1. Assessment methods included self-reporting of symptoms by patients, questioning of adverse events by investigators during study visits and undertaking blood tests and laboratory analyses.

Any adverse event
The number of participants with any adverse event was reported by 27 papers (Supplementary Table 2). From this data, 21.1% (95% confidence interval (CI) 19.9, 22.4) of participants using colchicine reported any adverse event compared to 18.9% (95% CI 17.7, 20.1) of participants in comparator groups. A metaanalysis showed the overall estimated risk ratio (RR) (95% CI) of any adverse event in colchicine users compared with pooled comparator groups was 1.46 (1.20, 1.77), P < 0.001 (Fig. 2, Table 2). The difference in RR of any adverse event in colchicine users was not   Table 3).
Although the sub-group meta-analyses showed a higher relative risk for any adverse event in colchicine users with liver diseases (RR 5.92 (95% CI 2.08, 16.82)), there was no overall significant difference in the relative risk of adverse events between different disease indications (P = 0.11) (Fig. 3). Furthermore,   Figure 2), different colchicine daily dose categories (P = 0.70) (Supplementary Figure 3) or different colchicine cumulative dose categories (P = 0.09) (Fig. 4).
The proportion of participants with any adverse event computed from all 35 studies in this review (in which prevalence was considered 0% if not reported) was 20.6% (95% CI 19.5, 21.9) in colchicine users and 17.9% (95% CI 16.8, 19.1) in comparator groups.

Diarrhoea
The number of participants with diarrhoea was reported by a total of 19 papers (Supplementary Table 2). From this data, 17.9% (95% CI 16.8 19.1) of participants using colchicine reported diarrhoea compared to 13.1% (95% CI 11.9, 14.3) of participants in comparator groups. The meta-analysis showed the overall estimated RR (95% CI) of diarrhoea in colchicine users compared with pooled comparator groups was 2.44 (1.62, 3.69) (P < 0.001) ((Supplementary Figure 4, Table 2). The difference in RR between placebo and active comparator groups was not  Table 3).
The proportion of participants with diarrhoea computed from all 35 studies in this review (in which prevalence was considered 0% if not reported) was 10.8% (95% CI 9.9, 11.7) in colchicine users and 6.1% (95% CI 5.4, 6.8) in comparator groups.

Gastrointestinal adverse event
The number of participants with any gastrointestinal event was reported by 29 papers (Supplementary Table 2) and included diarrhoea, nausea, vomiting, abdominal pain, loss of appetite, bloating, constipation, melena and peptic ulcer (Supplementary Table 4). From these 29 papers, 17.6% (95% CI 16.5, 18.8) of participants using colchicine reported a gastrointestinal event compared to 13.1% (95% CI 12.1, 14.2) of participants in comparator groups. The overall RR (95% CI) of gastrointestinal events in colchicine users compared with pooled comparator groups was 1.74 (1.32, 2.30), P < 0.001 (Fig. 5, Table 2). The difference between placebo and active comparator groups was not significant (P = 0.32). After the exclusion of 6 studies involving participants with liver disease, the RR (95% CI) of any gastrointestinal event in colchicine users vs comparator groups was similar at 1.60 (1.22, 2.10), P < 0.001 (Supplementary Table 3).
The proportion of participants with any gastrointestinal event computed from all 35 studies in this review (in which prevalence was considered 0% if not reported)

Liver events
The number of participants with liver events was reported by 13 papers (Supplementary Table 2) and included increased liver enzymes, hepatitis, hepatotoxicity and hepatic abnormalities (Supplementary Table 4). Pooled data from these papers showed 1.9% (95% CI 1.2, 2.8) of participants using colchicine reported a liver event compared to 1.1% (95% CI 0.6, 1.8) of participants in comparator groups. The overall RR (95% CI) of liver events in colchicine users did not significantly differ from the pooled comparator groups: 1.61 (0.86, 3.02) (Supplementary Figure 5, Table 2). The difference between placebo and active comparator groups was also not significant. None of the included papers involved participants with liver diseases.
The proportion of participants with any liver event computed from all 32 studies in this review (in which prevalence was considered 0% if not reported) was 0.5% (95% CI 0.3, 0.7) in colchicine users and 0.3% (95% CI 0.2, 0.5) in comparator groups.

Muscle events
The number of participants with muscle events was reported by nine studies (Supplementary Table 2) and included myalgia, muscle cramps, elevated creatine phosphokinase and muscle weakness (Supplementary Table 5). Rhabdomyolysis was not mentioned in any study. All nine studies involved placebo comparator groups. Pooled data from these studies showed 4.2% (95% CI 3.0, 5.7) of participants using colchicine reported a muscle event compared to 3.3% (95% CI 2.3, 4.7) of participants in placebo groups. The meta-analysis showed an overall non-significant RR (95% CI) of muscle events in colchicine users of 1.25 (0.80, 1.93) (Fig. 6,  Table 2). None of the studies involved participants with liver diseases.
The proportion of participants with muscle events computed from all 35 studies in this review (in which prevalence was considered 0% if not reported) was 0.8% (95% CI 0.6, 1.1) in colchicine users and 0.6% (95% CI 0.4, 0.9) in comparator groups.

Haematology events
The number of participants with haematology events was reported by eight studies (Supplementary Table 2) and included anaemia, bone marrow toxicity, leukopenia and purpura (Supplementary Table 4). All studies involved placebo comparator groups. Pooled data from these studies showed 0.6% (95% CI 0.3, 0.9) of participants using colchicine reported a haematology event compared to 0.4% (95% CI 0.2, 0.7) of participants in placebo groups. The occurrence of haematology events in colchicine or comparator groups was reported by three studies [21,23,37]. The meta-analysis showed an overall non-significant RR (95% CI) of haematology events in 1.34 (0.64, 2.82) (Supplementary Figure 6, Table 2). None of the studies involved participants with liver diseases.
The proportion of participants with a haematology event computed from all 35 studies in this review (in which prevalence was considered 0% if not reported) was 0.4% (95% CI 0.2, 0.6) in colchicine users and 0.3% (95% CI 0.1, 0.4) in comparator groups.

Sensory events
No studies mentioned neuropathy-related adverse events. However, two studies involving placebo comparator groups reported other sensory events (Supplementary Table 2) which included dysthesia in the legs and paresthesia (Supplementary Table 5). From this data, the pooled prevalence of sensory events was 1.1% (95% CI 0.  Table 2). None of the included papers involved participants with liver diseases.
The proportion of participants with any sensory events computed from all 35 studies in this review (in which prevalence was considered 0% if not reported) was 0.04% (95% CI 0.0, 0.1) in colchicine users and 0.07% (95% CI 0.0, 0.2) in comparator groups.

Infectious events
Seven studies reported various infectious events (Supplementary Table 2), including urinary tract infection, parotiditis, shingles, upper respiratory tract infection, nasopharyngitis and sinus congestion (Supplementary Table 5). From these papers, 0.4% (95% CI 0.2, 0.6) of participants using colchicine reported an infectious event compared to 2.1% (95% CI 1.6, 2.7) of participants in comparator groups. The overall RR (95% CI) of infectious events in colchicine users compared with pooled comparator groups was non-significant: 1.03 (0.70, 1.51) (Supplementary Figure 8, Table 2). The difference between placebo and active comparator groups was not significant (P = 0.94). No study involved participants with liver diseases.
The proportion of participants with any infectious event computed from all 35 studies in this review (in which prevalence was considered 0% if not reported) was 2.4% (95% CI 2.0, 2.9) in colchicine users and 2.8% (95% CI 2.4, 3.4) in comparator groups.

Death
Death related to adverse events was specifically reported in three studies (Supplementary Table 2). No study reported deaths related to an adverse event.

Miscellaneous events
Miscellaneous adverse events reported by the included studies are summarised in Supplementary Table 6. These events were not meta-analysed but contributed to the 'any adverse event' category.

Discussion
This systematic review and meta-analysis of randomised controlled trials indicate that overall, colchicine increases the rate of adverse events compared to both placebo and active comparators. Analysis of individual events demonstrated an increased risk for diarrhoea and gastrointestinal events in colchicine users, but no increase in the rate of other commonly cited adverse events, including liver, muscle, haematology, sensory or infectious events.
The mechanism by which colchicine induces diarrhoea and other gastrointestinal symptoms is not exactly known, but can be attributed to an increase in prostaglandin synthesis, intestinal secretion and gastrointestinal motility with this drug [41]. Although these symptoms can be clinical features of colchicine toxicity, they are usually mild, short-lived and reversible with dose reduction [12]. Serious adverse events associated with colchicine use, including neuropathy, myotoxicity and death were not reported in any trial included in the current analysis. These events may be more readily observed in less controlled environments evident in case reports involving colchicine over-dose, chronic renal diseases, interaction with concomitant medications and intravenous administration [42][43][44][45][46][47][48][49].
Analysis of adverse events in colchicine users showed no difference across different disease indications. Although overall, adverse events were numerically higher in patients with liver diseases, this risk was not significantly different from other disease indications. Furthermore, the sub-analysis excluding participants with liver disease showed similar adverse events rates to the main analysis. Although dose reduction is generally recommended when colchicine is used continuously in those with severe renal impairment, accurate conclusions regarding adverse events in this population could not be drawn from the current analysis.
There was notable heterogeneity across the clinical trials included in this review with regards to intervention methodology, including colchicine dose and treatment duration. However, sub-group analyses concluded that differences in drug use duration, daily dose or cumulative dose categories had no effect on the risk for adverse events. This contrasts with trials assessing the treatment of acute gout which report that high-dose colchicine results in a greater risk-to-benefit ratio. The paper reporting the AGREE trial included in the current analysis by Terkeltaub et al. [38], which directly compared two different doses of colchicine, found differences in adverse event rates between low and high dose groups, with 36% and 81% of participants having any adverse event, respectively. However, the short duration of treatment (1 to 6 h) meant that the cumulative drug doses in both groups were low in the context of the other papers included in the meta-analysis, resulting in a nonsignificant effect of dose in the meta-analysis. The difference in adverse event rates between the two arms of this AGREE trial may relate to better surveillance of adverse events in this trial compared to previously reported trials, or the relatively high dose (4.8 mg over 6 h) of the high-dose colchicine group. This is the only published trial comparing two differing doses of colchicine so conclusions on the reason for this disparity are difficult to be definitive about.
The limitations of this study include the inability in assessing the occurrence of rarer adverse events when only short duration controlled clinical trials were included. Different methodology is required to assess the frequency of rarer adverse events. Furthermore, the aims of the majority of the included studies were not primarily to assess safety, resulting in limited availability of adverse event data for extraction. As only studies which specifically reported an adverse event as being present or absent were included in the meta-analyses, it is possible that the pooled results may have over-estimated the true occurrence of adverse events which were not reported in all papers. In addition, it is also possible that the pooled results may have under-estimated the true occurrence of adverse events which were not assessed (e.g. those requiring blood tests). There were few included participants with severely impaired renal function, so the ability to assess for safety in this group was limited. Clinical trials often recruit patients in a highly selective manner, including excluding those with co-morbidities, and therefore the results are not necessarily generalizable to a general patient population. In addition, the included studies spanned over 20 years and it is likely that participants in earlier studies are not representative of patients treated with colchicine in clinical practice today. Other limitations include the screening of titles, abstracts and full-texts being undertaken by a single reviewer, and the exclusion of non-English language publications.
The strengths of this study include the strict inclusion of only placebo or active comparator blinded trials which reduces the potential for bias; although the occurrence of diarrhoea in participants can lead to at the least the suspicion of being in the colchicine group. In addition, there was a wide range of included indications such as gout, familial Mediterranean fever, Behcet's disease and pericarditis, which leads to increased generalisability of the study results.

Conclusions
This meta-analysis provides reassurance that common adverse events with colchicine use are limited to diarrhoea and gastrointestinal events. Whilst these are not benign side effects in some individuals, they will settle on dose reduction or drug discontinuation. More serious adverse events during colchicine use, including liver and haematological changes, muscle toxicity, neuropathy and death are very infrequent in clinical trials.
Supplementary Table 1. Participant inclusion and exclusion criteria and adverse event assessment methods of included studies. Table 2. Frequency of any adverse event reported in colchicine and comparator groups. Table 3. Meta-analysis results showing pooled risk ratio of adverse events between colchicine and pooled comparator groups for studies not involving participants with liver diseases. Table 4. Number of participants in colchicine and comparator groups with adverse events related to gastrointestinal, liver and hematologic events. Table 5. Number of participants in colchicine and comparator groups with adverse events related to muscle, sensory, and infectious events. Table 6. Number of participants in colchicine and comparator groups with miscellaneous adverse events or death. Figure 1. Quality assessment results using the modified-Jadad score. Figure 2. Forest plot showing estimated relative risk of any adverse event during colchicine use compared to comparator groups across different durations of drug exposure. Figure 3. Forest plot showing estimated relative risk of any adverse event during colchicine use compared to comparator groups across different daily doses of colchicine. Figure 4. Forest plot showing estimated relative risk of diarrhoea during colchicine use compared to placebo and active comparator groups. Figure 5. Forest plot showing estimated relative risk of liver events during colchicine use compared to placebo and active comparator groups. Figure 6. Forest plot showing estimated relative risk of hematology events during colchicine use compared to placebo (no active comparator studies). Figure 7. Forest plot showing estimated relative risk of sensory events during colchicine use compared to placebo (no active comparator studies). Figure 8. Forest plot showing estimated relative risk of infectious events during colchicine use compared to placebo and active comparator groups.
Abbreviations CI: Confidence interval; PRISMA: Preferred Reporting Items for Systematic Reviews and Meta-Analyses; RR: Risk ratio