Electronic searches were carried out using Medline (via PubMed), the Cochrane Library Embase and Web of Science following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement [14]. We combined the following key words / MeSH terms to identify the publications in the following query: “pericarditis OR pericardial effusion AND colchicine”. The literature search was conducted using EndNote Version X7.4 (Thomson Reuters), and after exclusion of duplicates resulted in 361 records. We applied the following inclusion criteria: studies investigating ≥10 patients with clinically pericarditis or post-pericardiotomy syndrome in which colchicine was compared against placebo. We excluded publications referring to only animal experiments or in vitro experiments, human studies on ≤ 10 patients, case reports, congress reports, review articles, editorial letters, and publications written in languages other than English or German. The databases were searched by 2 independent reviewers (LLL and MN). The searches were reviewed by AGR. There were no discrepancies between the reviewers regarding the identified and classified literature. The retrieved studies and references were reviewed for relevance by using title and abstract. The full publications that met the above-mentioned criteria were retrieved and included in the analysis.

The primary aim of the study was to determine the efficacy of colchicine in the treatment and prevention of pericarditis and PPS. The secondary aim was the efficacy of colchicine in reducing the rehospitalization rate, symptom persistence after 72 h as well as the safety and adverse effects of colchicine treatment.

Data regarding study characteristics, publication year, randomization mode, allocation concealment, blinding, intention-to-treat analysis, patient demographics and clinical characteristics, treatment indication and duration, colchicine dose, study and follow-up duration, clinical outcomes including pericarditis recurrence, re-hospitalizations, persistence of symptoms after 72 h and adverse effects were recorded.

Quality and risk of bias of the included studies was assessed using the Jadad scale [15]. This tool includes assessment of randomization, blinding of participants, and recording of dropouts and thus defines a Jadad score for each included study, whose value can be interpreted in terms of the probability of bias.

Statistical analyses were performed using the software package R (The R Project for Statistical Computing; version 3.3.2) with the packages “meta” and “metafor”, which were employed for calculation of heterogeneity between the studies, Forest plots and Funnel plots [16]. Categorical variables were reported as percentages and continues variables as mean ± standard deviation (SD). A probability value of p < 0.05 was considered statistically significant. Risk ratios (RRs) with 95% confidence intervals (CIs) were used for all outcomes. Heterogeneity was assessed by means of the I² statistic [17]. The risk of bias was assessed with the function “metabias” of the “meta” package of R and presented as a funnel plot.

A total of 361 studies from 1977 until 2016 were retrieved in the original search (Fig. 1). After reviewing titles and abstracts for the inclusion and exclusion criteria, 329 were excluded as not relevant. Full text evaluation of the remaining 32 publications resulted to the selection of 10 randomized controlled trials (RCT) to be included in this meta-analysis. Eight of them were double blinded while the rest had an open-label design. Intention-to-treat analysis was included in eight studies while it was not mentioned in the rest two studies. Nine studies were multicenter while one was performed in one center. The follow-up time was different in each study and reached from one up to 24 months (mean follow-up: 13.6 months). Only five patients were lost to follow-up in one study [18]. No patients lost to follow-up were reported in the remaining studies.

The indication for colchicine therapy varied as follows: two studies included patients with the first acute pericarditis incident, three studies included patients with recurrent pericarditis, and the rest five studies described the action of colchicine in postpericardiotomy syndrome (PPS) and postoperative pericardial effusion.

The characteristics of the 10 studies included in the meta-analysis are listed in Table 1. The 10 prospective studies from 2002 to 2015 included 1981 patients (mean age of 57.6 ± 7.3 years, 62% males). The 1000 control patients (57.9 ± 8.0 years) received as standard therapy NSAID (e.g. ibuprofen) or ASA, and only rarely corticosteroids (prednisone). Instead of colchicine, the control patients received placebo. The weight-adjusted doses of colchicine were in seven studies (0.5–1.0 mg), in two studies 1.0 mg and in the rest of the included studies 1.5 mg.

The duration, adverse effects (AE) of colchicine treatment, adherence and drug withdrawal (DW) are summarized in Table 2. The duration of colchicine treatment in pericarditis patients was initiated early after clinical diagnosis and randomization, and was maintained for 3–6 months. In contrast, the duration of colchicine treatment in patients undergoing heart surgery ranged between 14 days and 4 weeks, and was initiated 48–72 h before heart surgery [11], at the 3rd postoperative day [19, 22], at day 7–30 [25], or at 3 weeks after heart surgery [18] (Table 2).

The assessment of the included studies according to the Jadad scale is presented in Table 3. The funnel plot of the studies included in the meta-analysis is shown in Fig. 2.

Colchicine was shown to reduce the overall risk of PE in PC and recurrent pericarditis, and in PPS (all 10 studies) compared with placebo (RR: 0.57; 95% CI: 0.44–0.74 (Fig. 3). Due to the different populations studied, there was a considerable heterogeneity among studies (I² = 58%, p = 0.01). In patients with pericarditis (5 studies), colchicine reduced the risk of recurrence (RR 0.46; 95% CI: 0.36–0.58) (Fig. 4, upper panel). Both patients with a first acute pericarditis (2 studies; RR: 0.40; 95% CI: 0.24–0.66) as well as patients with recurrent pericarditis (3 studies; RR: 0.48; 95% CI: 0.36–0.63 benefited from colchicine treatment (Fig. 4, middle and lower panel respectively). The reported number needed to treat (NNT) for the prevention of recurrent pericarditis was 3 and 5, respectively [10, 23]. For acute pericarditis, the reported NNT was 4 and 5, respectively [21, 24]. Colchicine did not prove superiority compared to placebo for prevention of PPS in patients after heart surgery (RR: 0.70; 95% CI: 0.48–1.03) (Fig. 5). One RCT for PPS prevention reported an NNT of 10, and a number needed to harm (NNH) of 12 [11].

Colchicine reduced the need for rehospitalization in five studies (RR: 0.33; 95% CI: 0.18–0.60;). The benefit was significant in patients with pericarditis (RR: 0.31; 95% CI: 0.16–0.60), but not for the patients after heart surgery (RR: 0.43; 95% CI: 0.07–2.53;) (Fig. 6).

Colchicine reduced the number of patients with persistent symptoms after 72 h in 5 studies with pericarditis patients (RR: 0.43; 95% CI: 0.34–0.54) (Fig. 7).

Adverse events attributable to colchicine treatment were documented in seven studies, with gastrointestinal intolerance (GI) being the most reported adverse effect of colchicine (RR: 1.42; 95% CI: 1.05–1.92) (Fig. 8). Serious adverse events (SAE) were not reported by any study. The rate of AE ranged from 7% [20] to 20% [11], with gastrointestinal intolerance (GI) being the leading cause of AE, being treated either by direct DW, or by reducing the dosage of colchicine by half, and in cases of GI symptom persistence followed by DW (Table 2). Study #10 did not report detailed AE rates. AE were the main cause of DW, which was confirmed to stop GI. The reported rates of DW ranged from 6.7% [10] to 21.7% [11] (Table 2). Noteworthy, the rates of AE and of DW were not higher in any trial as compared with controls.

Colchicine has been shown to be very effective in reducing recurrence of pericarditis. It had been already included in the first ESC Guidelines for pericardial diseases as a IIa recommendation for the administration together with an NSAID or as monotherapy for the initial attack of pericarditis and prevention of recurrence [1].

Pericarditis recurrence as well as PPS are considered to be due to a localized pericardial autoimmune activation [32, 33]. Detection of increased cytokines and further inflammatory markers in the pericardial fluid, partly with absence of increment of these markers in the blood serum, enhances this concept [34]. Therefore, colchicine can effectively prevent recurrences of pericarditis as well as PPS due to its anti-inflammatory potential. The scenario is different in acute pericarditis, where an infectious pathogenesis seems to be frequently present [35]. In this case, however, colchicine is expected to reduce the recurrence rate, besides enhancing the response of pericarditis to NSAID. It is also evident that colchicine enhances the therapeutic effect of NSAID, thus leading to significantly lower persistence of symptoms after 72 h as well as rehospitalizations due to pericarditis. The latter effect increases the cost-effectiveness of colchicine substantially in comparison to placebo and justifies its use further. In line with these insights, the reported NNT for the prevention of acute or recurrent pericarditis was low, ranging between 3 and 5 [10, 21, 23, 24]. Those effects of colchicine have warranted its inclusion in the latest guidelines as first line therapy for the treatment of acute and recurrent pericarditis [3].

Thus far, the evidence for a differential diagnosis of pericarditis and PE including virological analyses, in accordance to the diagnostic approaches for inflammatory cardiomyopathy [36], may also be important for additional immunomodulatory treatment strategies in PE patients. Whether the different categories of PE diagnosed with complex procedures such as pericardioscopy and epicardial biopsies [37] may have different response patterns for the treatment with colchicine is an issue for further investigations.

However, our meta-analysis also reveals a significant heterogeneity of the therapeutic benefit of colchicine (Fig. 3; I²: 58%; p = 0.01). Especially, the publications with the study codes 09 [25] and 10 [18] failed to show significant benefit of colchicine in patients after heart surgery (Fig. 3). This heterogeneity may be largely due to the diversity of study groups (primary and secondary prevention of PC; PPS and PE after heart surgery) included in this meta-analysis. Overall, colchicine did not prove superiority compared to placebo for prevention of PPS in patients after heart surgery (RR: 0.70; 95% CI: 0.48–1.03) (Fig. 5). The NNT reported for PPS prevention was 10 and thus higher compared with the NNT for pericarditis [3‐5], and the number needed to harm (NNH) for PPS was 12 [11]. The substantially higher susceptibility of perioperative patients for GI-symptoms may contribute to this imbalance of clinical benefit versus AE and DW this particular patient group.

Most patients (up to 80%) will develop PE to a variable degreeafter cardiac surgery, which is generally mild and rarely results in cardiac tamponade in 1–2% of the post-pericardiotomy cases [38, 39]. Owing to the presumed pericardial inflammation involved in the pathogenesis of PE after cardiac surgery, colchicine could be efficacious in reducing the rate and the size of such pericardial effusions. In the COPPS-2 trial, colchicine administration starting before surgery failed to prevent postoperative pericardial effusion in the whole study cohort but was efficacious in patients with higher CRP [11]. However, this observation is not compatible with the results of the POPE-2 trial, which showed that patients with higher CRP had no decrease of pericardial effusion size [25]. Nevertheless, the pathogenicity of PE after heart surgery is not entirely due to pericardial inflammation, such as in PPS. Hence, colchicine may be able to prevent PPS but not to reduce the rate of clinically relevant PE or to prevent complications of PE after heart surgery that may be due to other causes, such as postoperative bleeding or heart failure. As discussed in [18], one major issue may be the differentiation of the inflammatory pathogenesis of PE after cardiac surgery, since non-inflammatory factors may be prominent in a substantial proportion of patients after cardiac surgery. In patients with PE post cardiac surgery without inflammatory pathogenesis, observational approaches might be the preferred treatment option, since colchicine may be presumably not effective, and potentially harmful due to its known adverse effects, in this substantial subgroup of patients. On the other hand, colchicine was administered late (at the 8th postoperative day) in the POPE-2 study, and at even later (3 weeks after heart surgery) in study #10 [18]. These two trials focused on patients with PE documented with echocardiography after 1–3 weeks after heart surgery, and thus obviously dealt with different patient populations as that of the COPPS [22] and COPPS-2 [11] studies. We therefore conclude that the clinical use of colchicine for the setting of PPS and of postoperative PE should be investigated in further multicenter RCT.

Colchicine is associated with adverse effects in up to 10–15% of the patients that may lead to discontinuation of treatment in some patients [20, 21, 32]. The frequent gastrointestinal complaints (mainly diarrhea) may limit colchicine use, but reduction of the dosis will mostly lead to relief of the complaints, and is reversible after DW of colchicine [20, 21]. Although the overall safety of colchicine is unequivocal, use of weight-adjusted doses may decrease [40]. Furthermore, it should be highlighted that several colchicine studies have excluded patients with tuberculous, neoplastic, or purulent causes of PE, hepatic (transaminases 1.5 times the upper normal limit) or renal dysfunction (serum creatinine above 2.5 mg/dL), myopathy or current serum creatine kinase above the upper normal limit, known allergy to colchicine, blood dyscrasias with anemia, leukopenia or thrombocytopenia, and pregnant and lactating women or women of childbearing potential not protected by a contraception method [20, 21, 23]. Noteworthy, the overall rate of AE, including GI, as well as of DW due to AE were not significantly lower in placebo patients as compared with colchicine treated patients in several RCT [10, 11, 20, 21, 23, 24].