Impact of colchicine on mortality and morbidity in COVID-19: a systematic review

Abstract Introduction Colchicine, because of its anti-inflammatory and possible anti-viral properties, has been proposed as potential therapeutic option for COVID-19. The role of colchicine to mitigate “cytokine storm” and to decrease the severity and mortality associated with COVID-19 has been evaluated in many studies. Objective To evaluate the role of colchicine on morbidity and mortality in COVID-19 patients. Methods This systematic review was conducted in accordance with the PRISMA recommendations. The literature search was conducted in 6 medical databases from inception to February 17, 2021 to identify studies evaluating colchicine as a therapeutic agent in COVID-19. All included studies were evaluated for risk of bias (ROB) using the Revised Cochrane ROB tool for randomised controlled trials (RCTs) and Newcastle-Ottawa Scale (NOS) for case-control and cohort studies. Results Four RCTs and four observational studies were included in the final analysis. One study evaluated colchicine in outpatients, while all others evaluated inpatient use of colchicine. There was significant variability in treatment protocols for colchicine and standard of care in all studies. A statistically significant decrease in all-cause mortality was observed in three observational studies. The risk of mechanical ventilation was significantly reduced only in one observational study. Length of hospitalisation was significantly reduced in two RCTs. Risk for hospitalisation was not significantly decreased in the study evaluating colchicine in outpatients. Very few studies had low risk of bias. Conclusion Based on the available data, colchicine shall not be recommended to treat COVID-19. Further high-quality and multi-center RCTs are required to assess the meaningful impact of this drug in COVID-19. KEY MESSAGES Colchicine, an anti-inflammatory agent has demonstrated anti-viral properties in in-vitro studies by degrading the microtubules, as well as by inhibiting the production of pro-inflammatory cytokines. Colchicine has been studied as a potential therapeutic option for COVID-19, with variable results. Until further research can establish the efficacy of colchicine in COVID-19, the use of colchicine in COVID-19 shall be restricted to clinical trials.


Introduction
Emanating from the Wuhan region of China in 2019, the coronavirus disease (COVID-19) has since spread and caused a global catastrophe. As of 13th July 2021, the disease caused by the Severe Acute Respiratory Syndrome Coronavirus (SARS-CoV-2) is responsible for more than 4 million deaths across the world [1]. The treatment options are limited with only one Food and Drug Authority-approved drug (remdesivir). Some medicines have received emergency use authorisation, while there are ongoing trials on several other agents that have shown plausible efficacy in preliminary studies [2][3][4][5][6][7]. Understandably, there is a great unmet need for therapeutic options. The pathogenesis of moderate to severe COVID-19 is centred around the "cytokine storm", where the rapid upsurge in inflammatory cytokines is responsible for the multiple-organ failure and increased severity of the disease [8]. Therefore, anti-inflammatory agents like dexamethasone and tocilizumab have shown promise [2]. The known antiinflammatory properties of colchicine are utilised in several disorders including but not limited to gout, Familial Mediterranean fever (FMF), acute and recurrent pericarditis, Behcet disease, Sweet syndrome, and calcium pyrophosphate deposition disease [9]. Colchicine, an alkaloid isolated from colchicum autumnale plant (molecular formula C22H25NO6), binds to the intracellular unpolymerised protein tubulin irreversibly, forming a tubulin-colchicine complex, which prevents polymerisation of the microtubule polymer, hence arresting the microtubule growth and promoting microtubule depolymerisation. Colchicine has potential synergy in the treatment of cytokine cascade in COVID-19 at different levels. In vitro studies demonstrate the functionality of microtubules during initial cellular infection with SARS-CoV-2. By degrading the microtubules, colchicine is hypothesised to exude antiviral properties. The coronavirus spike protein utilises the cytoskeletal elements of host cells during viral entry. Colchicine hence may hamper viral entry, infection, and propagation due to the multi-faceted uses of microtubule elements. A component of the SARS-associated coronavirus called viroporin-E creates calciumpermeable ion channels and activates the NLRP3 inflammasome. Colchicine disrupts the NLRP3 inflammasome activation, which plays an important role in the development of phase 3 cytokine storm from SARS-COV 2. Furthermore, colchicine may interfere with the cytokine storm by inhibiting the production of pro-inflammatory cytokines such as IL-1b, IL-18, IL-6, and IFN-c and superoxide free radicals [8][9][10][11]. There have been several studies investigating the role of colchicine in the management of COVID-19. These have shown variable results. The objective of this study was to systematically review the available literature on the role of colchicine in the treatment of COVID-19.

Eligibility criteria
This systematic review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) recommendations [12]. Intervention: Use of colchicine for COVID-19. Comparator/Control: Other treatment modalities including standard of care as documented in the studies.
Outcomes: All-cause mortality, mechanical ventilation requirement, risk of hospitalisation, length of hospital stay, effect on inflammatory markers (C-reactive protein, ferritin, D-dimer, lactate dehydrogenase or any other inflammatory markers as mentioned in the studies), and adverse effects (gastrointestinal upset or any other adverse effects as reported in the studies).
Study design: Randomised controlled trials (RCTs), prospective and retrospective cohort studies.

Inclusion and exclusion criteria
Studies evaluating the use of colchicine for treatment of COVID-19 in adults were included for this systematic review. Exclusion criteria included the use of colchicine beyond COVID-19, animal studies, case reports, case series, review articles, meta-analyses, non-English language studies, and those without a comparator arm.
Bibliographies of identified studies and abstracts published in the annual conferences of professional medicine subspecialties including rheumatology, critical care, infectious disease, internal medicine, and emergency medicine were searched to identify additional studies.

Study selection and data collection
We used EndNote version 20.1 for data management and citation duplication assessment. Two authors independently reviewed the identified abstracts to identify articles for full-text review. The full text was also reviewed if the abstract was unavailable. Reasons for exclusion were recorded. A third author independently reviewed the results from both authors and resolved any conflicts. Relevant information from the included papers was extracted by two authors and re-examined for accuracy by a third author. Pertinent data extracted included study first author, publication date, study location, study design, study participants number and baseline characteristics, study interventions, and study outcomes.

Assessment of methodologic quality
The risk of bias for RCTs was assessed using the Revised Cochrane risk-of-bias tool for randomised trials [13]. For retrospective observational studies and cohort studies, the Newcastle-Ottawa Scale (NOS) for case-control and cohort studies was used, respectively [14]. For the NOS, a score of 6 or more was considered to be suggestive of higher study quality and study credibility [15]. One author assessed the risk of bias in the included studies, and the results were reviewed by other authors. Disagreements were resolved by group discussion and consensus. The signalling questions and quality assessment definitions are given in Appendix 2.

Review of literature/study characteristics
The initial database search identified a total of 721 citations. A total of eight studies containing 5661 patients were included in the final analysis [16][17][18][19][20][21][22][23]. A PRISMA flow diagram describing the inclusion process is mentioned in Figure 1. Of the included studies, 4 were randomised controlled trials (RCTs) and 4 were observational studies. In the RCTs, 2 were doubleblinded, 1 was open-label and 1 was single-blinded & open-label design. In the observational studies, 3 were retrospective and 1 was a prospective study. Two studies were published as preprints. One RCT was performed in Canada and recruited patients from six countries, 2 studies were conducted in the US, 1 in India, 1 in Brazil, 1 in Iran, 1 in Greece, and 1 in Italy. Only one RCT included outpatients, all other studies were performed on inpatients. The COLCORONA trial by Tardif et al. was terminated early at 75% of enrolment due to logistical and time constraints [16]. The GRECCO trial by Deftereos et al. was terminated early due to slow patient enrolment [17].

Patient characteristics
Seven studies restricted inclusion to RT-PCR confirmed cases, while one recruited clinically suspected COVID-19 patients. Baseline patient characteristics in the colchicine and control groups were separately mentioned by 7 studies ( Table 1). The mean age of patients in the colchicine and control group in these seven studies was 61 and 61.7 years respectively. The colchicine group contained 46.8% of males and the control group had 51.0% males. Only two studies reported race -In the study by Tardif et al., there were 93.3% Caucasians in the colchicine group and 93.2% Caucasians in the control group [16]. In the study by Burnetti et al., the colchicine group had 26.8% white and 48.8% Hispanics, while the control group had 26.0% whites and 48.8% Hispanics [23]. Information about patient baseline comorbidities in the studies is detailed in Table 1.

Treatment data
Standard of care (SOC-protocol to treat COVID-19 patients) was highly variable depending on physician discretion, drug availability, and institutional protocol. This included hydroxychloroquine (HCQ), azithromycin, ceftriaxone, antivirals, IL-6 inhibitor, anticoagulation, and corticosteroids in various combinations. There was no standardised approach on dosing and duration of colchicine. It was variable in each study as mentioned in Table 2. The dose was adjusted according to the weight of the patient, GFR, other treatments, and side effects severity. In the largest study by Tardif et al evaluating outpatient use of colchicine in COVID-19, colchicine was initiated within 4 h of enrolment [16]. In the studies evaluating inpatient colchicine use, the interval between hospital admission and colchicine administration was variable and ranged from within 72 h to 6.28 days after hospitalisation.

All-cause mortality
All studies assessed all-cause mortality. A statistically significant difference was observed in 3 studies. Sandhu et al reported a significant decrease in mortality in the colchicine þ SOC group (47.1% vs 80.8%: pvalue .0003) [22]. The follow-up duration is unclear. At a 28 days follow-up period, Burnetti et al reported a significant reduction in mortality after propensity matching (9.1% vs 33.3%; p-value .023) [23]. However, the results were not significant before matching (9.8% vs 22.1%; p ¼ .077). Colchicine was associated with a significant reduction in mortality after adjustment for age, comorbidity index, and c-reactive protein (odds ratio, 0.21; 95% confidence interval, 0.06-0.71; p ¼ .012). Similarly, Scarsi et al reported a significant decrease in mortality in the colchicine þ SOC as compared to the placebo þ SOC group (16.3% vs 37.1%; p < .001) at 21 days follow-up [21]. On performing a cox proportional hazards regression survival analysis, a lower risk of death was independently associated with colchicine treatment (HR ¼ 0.151 (95% CI 0.062-0.368), p < .0001). There was no death observed in both groups in the study by Salezadeh et al. [19] In the study by Tardif et al, results were not significant at 30 days follow-up period (mean value) [16]. Deftereos et al observed a decrease in death rate in the colchicine group (1.8% vs 8.0%; p-value not mentioned) [17]. None of the patients recruited to the colchicine group died in the study by Lopes et al (0.0 vs 5.6; pvalue not mentioned) and Mahale et al (28.2 vs 26.3; p-value not mentioned) observed higher mortality in the colchicine group [18,20]. Event-free survival (duration from the primary clinical endpoints) was reported by Deftereos et al. [17]. The mean event-free survival time was increased in the colchicine group (20.7 days vs 18.6 days: p-value .03).

Inflammatory markers
The effects of an intervention on inflammatory markers (CRP, Lactate Dehydrogenase, Ferritin, Ddimer) was reported in 50% of included studies. Lopes et al. reported that both groups had similar levels of  Lopes et al. [18] Tardif et al. [16] Mahale et al. [20] Scarsi et al. [21] Sandhu et al [22] Deftereos et al. [17] Brunetti et [ [17,20,23]. The need for mechanical ventilation was not reported in 2 studies [19,21]. The study performed by Salehzadeh et al was only performed in non-ICU patients and did not evaluate the need for mechanical ventilation [19]. Lopes et al reported that the median need for supplemental oxygen was decreased in the colchicine group (4 vs 6.5 days; p < .001) [18]. At day 2, 67% vs 86% of patients maintained the need for supplemental oxygen, while at day 7, the values were 9% vs 42%, in the colchicine and the placebo groups, respectively (logrank; p ¼ .001)

Risk of hospitalisation
The risk of hospitalisation was assessed in the only outpatient study by Tardif et al. [16] The risk was reduced in the colchicine group (4.5% vs 5.7%) but was not statistically significant (p-value .08). When groups were subdivided into only PCR positive patients (2075 vs 2084), there was a significant decrease in the hospitalisation risk (4.5% vs 6%; OR 0.75, 95% CI, 0.57-0.99) in the colchicine group.

Length of hospital stay
The results were statistically significant in favour of colchicine in 2 studies. Salehzadeh et al reported a significant decrease (6.28 vs 8.12 days; p-value .001) in the length of hospitalisation in the colchicine þ HCQ group as compared to the HCQ þ placebo group [19]. The median time of hospitalisation was 7.0 vs 9.0 days (p ¼ .003) in the colchicine þ SOC vs placebo þ SOC as reported by Lopes et al. [18] In 2 studies, no statistically significant difference in median hospital duration was observed (12 vs 13 days in Deftereos et al. and 10.5 vs 11 days in Sandhu et al.) [17,22]. Length of hospitalisation was 12.7 vs 11.9 days (aggregate mean) in the study by Mahale et al, however, significance was not reported [20]. Patients discharged on day 28 were approximately five times more in the colchicine group as reported by Burnetti et al (90.9% vs 66.7%: p-value .023) [23]. Duration of hospitalisation was not reported in 3 studies.  [21]. New or worsened diarrhoea was more frequent in the intervention group (17% vs 6%; p-value .26) and was controlled with the prescription of an antisecretory agent in the study by Lopes et al. [18]

Risk of bias (ROB)
The risk of bias for the RCTs was mixed, with one RCT with high risk, two with some concerns, and one with low risk of bias, with the predicted direction of bias favouring the experimental arm. Overall, the cohort and case-control studies fared poorly in terms of comparability except for the study by Brunetti et al, which had the best quality rating of all the non-RCT studies. Detailed results of the ROB assessment are provided in Table 3, and detailed scoring of individual domains is mentioned in Appendix 2.

Discussion
In this systematic review (SR) of 4 RCTs and 4 observational studies, we evaluated the effect of colchicine on mortality, need for mechanical ventilation, change in the inflammatory markers, and adverse effects in COVID-19 patients compared to those who did not receive colchicine. The study designs and clinical outcomes are variable and inconsistent to establish the clinical efficacy of colchicine treatment in COVID-19 for reported outcomes. The certainty of evidence regarding all-cause mortality in the colchicine treatment group was low, as statistically significant low mortality was reported in less than 50% of the studies, and the treatment follow-up period was inconsistently reported. Even the high-quality RCT by Tardif et al. did not show statistically significant benefit on the primary endpoint (death or hospitalisation) in the primary analysis, was underpowered, investigated the clinical outcomes in the outpatient setting, and has the limitation of including patients diagnosed with COVID19 based on clinical criteria which means around 7.3% were not confirmed by PCR [16]. Taken together, the presented data is suggestive of some benefit on patient-related clinical outcomes which need further research. Colchicine is an anti-inflammatory agent that may have a role to prevent hyperinflammatory states as indicated by a reduction in inflammatory markers in patients receiving colchicine in studies evaluating laboratory parameters. Although a decrease in length of stay was reported by 2 studies (Salehzadeh et al and Mahale et al), the presented evidence is of low quality (one study is retrospective, and the RCT is underpowered) [19,20]. This systematic review elucidates that only one study showed a lower rate of intubation in the colchicine group, but this study had many inherited weaknesses with potential performance bias and confounding from additional medications [22]. At this point, it is reasonable to suggest that data is weak and of low quality to suggest a decrease in need for mechanical ventilation with colchicine treatment.
Colchicine has a well-documented safety profile; however, it can cause GI-related side effects (nausea, vomiting, diarrhoea) and has been rarely reported to cause hematological side effects. It may be implied that colchicine's potent anti-inflammatory properties should decrease coagulability, however, on the contrary, the colchicine group had a significantly higher number of PE in the largest study performed by Tardiff et al, whose mechanism is not well studied [16]. It is well documented that COVID-19 is a hypercoagulable state and the anti-inflammatory effects of colchicine could have been negated by the underlying pro-thrombotic state. However, rates of PE were significantly higher in the colchicine group, which is surprising and needs further validation, as earlier studies have not found an increased risk of thromboembolism associated with colchicine. Heterogeneously reported outcomes and variable study designs need further inquiry. Colchicine was used with SOC in most of the studies, but the SOC varied in almost all the studies based on hospital policies and physicians' discretions. Further, several medications including some antibacterials and antivirals can increase the plasma levels of colchicine by inhibiting CYP3A4 and P-glycoprotein, thus increasing the risk of colchicine adverse effects such as GI-related adverse effects [24]. Inclusion of such medications (e.g. azithromycin) in the SOC by some studies could have enhanced the toxicity reported in those studies. Variable dosing and timing protocols were used for treatment. Adequately powered randomised clinical trials with consistent clinical outcomes, in different clinical settings (hospitalized and out-patient) with a adequate follow-up, are warranted to establish the efficacy of colchicine in COVID-19 treatment. Results from high-quality RCTs are necessary to further prove a treatment benefit of colchicine in COVID-19 patients. The National Institutes of Health and the National Institute for Health and Care Excellence have recommended against the use of colchicine in the management of COVID-19 [25,26].

Limitations
Our study has several limitations. There were no uniform outcome measures and insufficient reporting of statistical significance of the outcomes which make the precise conclusions challenging. Studies were too variable to perform a meaningful statistical analysis for major outcomes, limiting our study to be a descriptive analysis rather than a meta-analysis. The studies were also heterogeneous in terms of severity of patient population, inclusion criteria, colchicine dosing protocols, comparison group protocols, which makes generalisability and implications of the results unclear.
Overall, the quality of evidence was low. Some studies may have been missed as our review was restricted to English language studies.

Conclusion
Treatment with colchicine in COVID-19 should not be recommended until more evidence is available to support positive outcomes. Based on the available data, judicious and cautious use of colchicine shall be recommended only in clinical trials. Further well-performed clinical trials can assess the efficacy and safety of this drug in COVID-19.

Acknowledgements
There was no funding for the work associated with this publication. None of the authors have been paid by any agency or pharmaceutical company to write this article. All authors have full access to the manuscript and all the data in the study, and the corresponding author has the final responsibility for the decision to submit for publication. All authors contributed to the study design, critically reviewed the first draft, approved the final version, and agreed to be accountable for the work.

Disclosure statement
No potential conflict of interest was reported by the author(s). What is the predicted direction of bias arising from the randomisation process?