Perioperative statin therapy in cardiac and non-cardiac surgery: a systematic review and meta-analysis of randomized controlled trials

Background The effects of perioperative statin therapy on clinical outcome after cardiac or non-cardiac surgery are controversial. We aimed to assess the association between perioperative statin therapy and postoperative outcome. Methods Electronic databases were searched up to May 1, 2018, for randomized controlled trials of perioperative statin therapy versus placebo or no treatment in adult cardiac or non-cardiac surgery. Postoperative outcomes were: myocardial infarction, stroke, acute kidney injury (AKI), and mortality. We calculated risk ratio (RR) or odds ratio (OR) and 95% confidence interval (CI) using fixed-effects meta-analyses. We performed meta-regression and subgroup analyses to assess the possible influence of statin therapy regimen on clinical outcomes and trial sequential analysis to evaluate the risk of random errors and futility. Results We included data from 35 RCTs involving 8200 patients. Perioperative statin therapy was associated with lower incidence of postoperative myocardial infarction in non-cardiac surgery (OR = 0.44 [95% CI 0.30–0.64], p < 0.0001), but not in cardiac surgery (OR = 0.93 [95% CI 0.70–1.24], p = 0.61) (psubgroup = 0.002). Higher incidence of AKI was present in cardiac surgery patients receiving perioperative statins (RR = 1.15 [95% CI 1.00–1.31], p = 0.05), nonetheless not in non-cardiac surgery (RR = 1.52 [95% CI 0.71–3.26], p = 0.28) (psubgroup = 0.47). No difference in postoperative stroke and mortality was present in either cardiac or non-cardiac surgery. However, low risk of bias trials performed in cardiac surgery showed a higher mortality with statins versus placebo (OR = 3.71 [95% CI 1.03–13.34], p = 0.04). Subgroup and meta-regression analyses failed to find possible relationships between length of statin regimens and clinical outcomes. Trial sequential analysis suggested no firm conclusions on the topic. Conclusions Perioperative statins appear to be protective against postoperative myocardial infarction in non-cardiac surgery and associated with higher AKI in cardiac surgery. Possible positive or even negative effects on mortality could not be excluded and merits further investigations. Currently, no randomized evidence supports the systematic administration of statins in surgical patients. Electronic supplementary material The online version of this article (10.1186/s13613-018-0441-3) contains supplementary material, which is available to authorized users.


Background
Perioperative complications are still relatively frequent in patients undergoing cardiac or non-cardiac surgery, and attempts to minimize complications and deaths are crucial. Pathophysiological mechanisms behind clinical complications may include inflammatory processes after surgery.
The knowledge of statin's pleiotropic and anti-inflammatory effects has led to consider perioperative statin a potential treatment able to modulate the clinical outcome after surgery [1]. Previous studies suggested beneficial effects of perioperative statin therapy on postoperative outcome after high-risk non-cardiac surgery [2][3][4], with a previous meta-analysis advising that perioperative statin therapy decreases the perioperative incidence of mortality and myocardial infarction in this high-risk population [5].
In contrast, growing evidences on perioperative statins administration in cardiac surgery suggested neutral or even detrimental effects. In facts, two large, high-quality, trials randomizing cardiac surgery patients to receive perioperative statin or placebo were recently published [6,7]. Perioperative statins did not prevent postoperative atrial fibrillation or perioperative myocardial damage, but acute kidney injury (AKI) was more common in patients receiving statin. All this evidence supported a potential neutral or even negative effect of perioperative statin in cardiac surgery.
We conducted a systematic review and meta-analysis of randomized controlled trials (RCTs) to provide clinical guidance of administration of perioperative statin therapy in the perioperative period. We aimed to assess the effects on postoperative myocardial infarction, stroke, AKI, and mortality in adult cardiac and non-cardiac surgery in patients treated with perioperative statins.

Methods
We performed a systematic review and meta-analysis of RCTs, in compliance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [8] and the Cochrane methodology [9], and according to a pre-published protocol (PROSPERO database, CRD42018093997) [10]. A PRISMA checklist is available in the supplement (eTable 1, Additional file 1). The authors had no conflicts of interests.

Search strategy
Two trained investigators (AP and CS) independently searched PubMed, the Cochrane Central Register of clinical trials, and EMBASE (last updated on May 1, 2018) for appropriate articles. The search strategy for PubMed is reported in the supplement (eMethods 1, Additional file 1) and was designed to include any RCT ever published with perioperative statin therapy compared to control in adult humans undergoing surgery. In addition, we employed backward snowballing (i.e., scanning of references of pertinent articles). No language restriction was enforced.

Study selection
References, obtained from database and literature searches, were examined first at an abstract level independently by 2 investigators (AP and CS), with eventual divergences resolved by consensus, and then, if potentially pertinent, were retrieved as complete articles. Eligible studies met the following PICOS criteria: (1) Population: adult cardiac and non-cardiac surgery patients; (2) Intervention: perioperative administration of statin therapy; (3) Comparison intervention: placebo or no active intervention as control; (4) Outcome: any outcome of the present systematic review (see below); (5) Study design: RCT. The exclusion criteria were: overlapping populations and pediatric studies. Two authors (AP and JP or AB) independently assessed selected studies for the analysis, with divergences resolved by consensus with a third author (GL).

Data abstraction and study endpoint
Baseline characteristics, procedural, and outcome data were abstracted by one author (AP) extracted relevant information from each selected study, and these data were checked by a second author (JP or AB). Specifically, we extracted potential sources of significant clinical heterogeneity, such as study design, clinical setting/ indication, statin dose, and control treatment, as well as primary study outcomes.
The per-protocol primary outcomes were: postoperative MI, postoperative stroke, postoperative AKI, and mortality at the longest follow-up available. Post hoc secondary outcomes were AKI requiring postoperative renal replacement therapy (RRT) and AKI not requiring RRT. The outcomes were reported as per-author definition. We extracted data following the intention-to-treat basis whenever possible. Corresponding authors of all eligible articles were contacted in case of missing data on outcomes of interest.

Risk of bias assessment
We used the Cochrane methodology to evaluate the methodological quality of each included trial [9]. Each trial was finally judged to be of low, unclear, or high risk of bias (eMethods 2, Additional file 1). Publication bias was assessed by visually inspecting funnel plots for pooled analyses containing > 10 studies [9].

Statistical analysis
The analysis was stratified according to cardiac or noncardiac surgery setting. For each outcome, we calculated the odds ratio (OR) with 95% confidence intervals (CI). For common events, pre-defined as frequency of the event in the control group > 10%, we calculated risk ratio (RR) with 95% CI [9]. A p value equal or less than 0.05 was considered significant. In case of statistical significant results, we calculated the number needed to treat (NNT) or number needed to harm (NNH) and 95% CI. Heterogeneity was explored by the Cochran Q statistic and characterized with I 2 . We used a fixed-effects model in the absence of significant heterogeneity, defined as p value > 0.10 and I 2 < 50%. In case of significant heterogeneity, we employed the random-effects model except if few trials dominate the available evidence or if significant publication bias was present, since random-effects metaanalysis, in these contexts, can give inappropriate high weight to smaller studies [9].
According to Cochrane methodology [9], we performed subgroup analyses for each outcome in order to assess the influence of trials' risk of bias, including only low risk of bias trials or only trials with unclear or high risk of bias. Sensitivity analyses were performed changing summary statistic (OR, RR, risk difference) and according to possible conflicts of interests/funding. Meta-regression was employed to examine the possible influence of statin therapy regimen on clinical outcomes. Subgroup analyses were performed on trials that included only statin-naïve patients or trials not employing only statinnaïve patients (trials enrolling a population of chronic statin therapy or a mixed population). Subgroup differences were tested using Chi-square statistics [9]. The meta-analysis was performed using Review Manager (RevMan [Computer program], version 5.3. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration, 2014). All the analyses were pre-defined [10].

Trial sequential analysis
To control risks of random errors due to sparse data and repetitive testing of cumulative data, we performed a perprotocol fixed-effects trial sequential analysis (TSA). TSA is a methodology that combines an information size calculation, representative of the cumulated sample sizes of all included trials, with a threshold for a statistically significant treatment effect and a threshold for futility of the intervention. In particular, TSA pools the required information size with trial sequential monitoring boundaries which adjust the confidence intervals and decrease type I errors [11][12][13]. In TSA, the inclusion of each trial in the meta-analysis is regarded as an interim meta-analysis and TSA permits to control the risk for type I and type II errors and helps to clarify whether additional trials are needed. Conclusions made using TSA show the potential to be more consistent than those using traditional metaanalysis techniques [11][12][13]. We conducted TSA with the purpose to maintain an overall 5% risk of type I error and a 10% risk of type II error, at a power of 90%. We assumed a relative risk reduction (RRR) or relative risk increase (RRI) of 15%, and we derived the control event proportion from low risk of bias trial. The resulting required information size was further diversity (D 2 )-adjusted; in case of D 2 = 0, we performed a sensitivity analysis assuming a D 2 = 25%. We used the TSA software (TSA Viewer [Computer program], version 0.9.5.5 Beta, Copenhagen Trial Unit, Centre for Clinical Intervention Research, Rigshospitalet, 2016).

Study characteristics
The literature search yielded 8376 references of which 35 RCTs (8200 randomized patients) [2, 3, 6, 7, 14-44] met the eligibility criteria and were included in the analysis (Fig. 1). Major exclusions are presented in the supplement (eTable 2, Additional file 1). Characteristics of the included trials are reported in Table 1 and in eTable 3 in the Additional file 1.
Twenty-four trails included exclusively statin-naïve patients (4101 patients), 3 trials included patients on chronic statin therapy (1080 patients), and 5 trials had a mixed population (2810 patients). All trials administered statins preoperatively and 19 trials postoperatively. Length of preoperative statin treatment ranged from 1 to 28 days (median 7 days), and the total duration of therapy varied from 2 to 45 days (median 7 days). Atorvastatin (10-80 mg) was administered in 22 trials, simvastatin (20-80 mg) in 6 trials, rosuvastatin (10 and 20 mg) in 5 trials, pravastatin (40 or 80 mg) in 1 trial, and fluvastatin 80 mg in 1 trial. Twenty-six trials administered placebo as control, and 9 trials administered no intervention as control.

Myocardial infarction
The rate of postoperative MI was lower in non-cardiac surgery patients randomized to statins (OR = 0. 44 18.27% of the information size accrued) suggesting the need of further trials for a firm conclusion (eFigure 6, Additional file 1). The results were similar when limiting the analysis to low risk of bias trials (OR = 0.28 [95% CI 0.13-0.64], 2 trials and 300 patients) and at sensitivity analyses (eTables 4 and 5, Additional file 1).

Acute kidney injury and renal replacement therapy
Perioperative statins were associated with higher incidence of AKI when compared to control in cardiac surgery patients ( (Fig. 4), but only 3 trials (239 patients) reported the outcome, with all the events coming from 1 trial [19].

Mortality
The administration of perioperative statin therapy was associated with no significant difference in mortality in both study population, at a median follow-up of 30-days (Fig. 6). However, when assessing only lower risk of bias trials, perioperative statins were associated with increased mortality in the cardiac surgery population (OR = 3.71 [95% CI 1.03-13.34], NNH = 181 [95% CI 97-1187]) (eTable 4, Additional file 1). Trial sequential analysis suggested no firm evidence for a RRI = 15% and the need of further randomized trials.

Clinical outcomes and statin regimen
In statin-naïve patients undergoing cardiac surgery, perioperative statins are associated with no significant differences in all the assessed outcomes (eTable 6, Additional file 1). Trials enrolling chronic statin users or a mixed population showed that perioperative statins are In non-cardiac surgery, lower myocardial infarction is evident in both statin-naïve (OR = 0.49 [95% CI 0.30-0.81]) and chronic statin patients (OR = 0.37 [95% CI 0.21-0.67]), and no differences in other outcomes (eTable 6, Additional file 1). No significant between-groups differences were found between statin-naïve trials and other trials enrolling patients on chronic statin therapy or a mixed population (p groups > 0.05) (eTable 6, Additional file 1).
Meta-regression analysis failed to find possible relationships between length of pre-or postoperative statin regimen and clinical outcomes (eResults 2, Additional file 1).

Discussion
Our systematic review and meta-analysis suggests that perioperative statin therapy could be protective against postoperative myocardial infarction in non-cardiac surgery but associated with an increased risk of acute kidney injury in cardiac surgery. Statins were associated with an increase in hospital mortality in cardiac surgery in low risk of bias trials. However, the quality and quantity of randomized evidence are still insufficient to allow a firm conclusion on the topic. This is the largest and most comprehensive meta-analysis of RCTs of statins in surgery performed so far, including 35 randomized trials and 8200 patients, evaluating the effect of statin therapy on outcomes in both non-cardiac and cardiac surgery using separate analysis. In noncardiac surgery, a previous meta-analysis that evaluated 5 RCTs in vascular surgery did not show any detrimental or favorable effect of statins on postoperative outcomes [45]. De Waal et al., in a meta-analysis of 16 trials, only included statin-naïve patients undergoing surgery and showed that starting statin therapy, in this cohort of patients who were not already on long-term statin treatment, reduces perioperative mortality, myocardial infarction, atrial fibrillation, and decreases length of hospital stay [5]. Our study is consistent with previous findings regarding prevention in postoperative myocardial infarction, but we failed to find any other possible beneficial effects, even in the statin-naïve subgroup. On the other hand, a recent meta-analysis evaluated 23 trials in cardiac surgery and showed no effect of statin on postoperative incidence of myocardial infarction, infection, stroke, and atrial fibrillation, and a higher occurrence of AKI [46].
A potential beneficial effect of perioperative statins on postoperative MI in non-cardiac surgery was found, in contraposition to the lack of effects in cardiac surgery patients. Regarding non-cardiac surgery, a randomized study suggested that a short-term treatment with atorvastatin significantly reduces the incidence of major adverse cardiovascular events after vascular surgery [2]. In contrast, a more recent and larger trial found neutral results and did not demonstrate a reduction in major cardiovascular complications after a short-term perioperative course of statin in statin-naïve non-cardiac surgery patients, even in the vascular surgery subgroup [19]. Chopra and colleagues, in a meta-analysis published in 2012 including 15 RCTs with 2292 patients undergoing cardiac and non-cardiac surgeries, found that perioperative statin treatment reduced the risk of myocardial infarction [4]. Nowadays, the growing quality and quantity of data can allow a more thoughtful analysis, confirming the potential benefits of statin in non-cardiac surgery in terms of MI prevention, but not in cardiac surgery. In fact, in the largest RCTs performed so far in cardiac surgery patients, no difference was found in postoperative MI and myocardial injury [6,7]. Perioperative myocardial ischemia in non-cardiac surgery occurs mostly as consequence of endogenous catecholamine release and acute inflammatory response, resulting in an increased oxygen demand, heart rate, and contractility. Inflammation also leads to changes in atheromatous plaque features, culminating in the rupture of lipid-laden vulnerable lesions, thrombocytes activation with development of platelet-rich thrombi, acute vessel occlusion, and ischemia [47,48]. Perioperative statins might reduce the risk of perioperative infarction due to its properties of endothelial modulation resulting in vasodilation in addition to its anti-inflammatory effects, reducing plaque instability [49]. In cardiac surgery, myocardial infarction occurs mainly due to an ischemic event arising from either a failure in graft function, an acute coronary event involving the native coronary arteries, or inadequate cardioprotection during cardiopulmonary bypass. We might postulate that the different pathophysiology mechanisms could explain our findings of no protection of statins against MI after cardiac surgery. In addition, our findings showed no effect of perioperative statin in postoperative stroke, another crucial cardiovascular complication. However, further investigations are warranted in patients at high risk of cerebrovascular events.
Our systematic review, including 25 studies in cardiac surgery, of whom 5 with low risk of bias, confirms the previous findings of possible negative effects of statins on renal function of cardiac surgery patients. For many years, perioperative statins were considered an attractive therapy for preventing AKI following cardiac surgery, hypothesis based mainly on the positive results from retrospective series or high risk of bias RCTs [50]. Nowadays, high-quality RCTs support the lack of a kidney-protective effect [6,7,30]. The largest RCT showed that rosuvastatin therapy resulted in a significantly higher rate of AKI and plasma creatinine compared to placebo at 48 h after cardiac surgery [7]. Similarly, the second largest RCT published so far showed a non-significant trend in favor of placebo and a possible harmful effect of perioperative statins in the small subgroup of statin-naïve patients with chronic kidney disease [6]. The exact mechanisms of increased occurrence of AKI related to statin use need to be elucidated. There are possible effects of statin in kidney function, including mitochondrial dysfunction leading to overall cellular energy imbalance [50][51][52]. Further research is needed, especially in the non-cardiac surgery setting, where evidence is still limited, not allowing any conclusion on possible renal effects of statin in this population.

Strengths and limitations
Our study has some limitations, which most of them are characteristics of all aggregate data meta-analyses [53]. Different statin dosages and formulations were used as statin therapy in the included studies. We did not perform subgroup analysis of different types of perioperative statin regimens, since most of the trials administered different statin dose and formulation for different length of time, and the analysis would be underpowered and mainly driven by results of small and higher risk of bias trials. Another limitation is that only few high-quality randomized trials have been published so far in non-cardiac surgery setting. To assess the effects of methodological biases on results [54], we assessed subgroup analyses according to trial risk of bias. A major strength of this meta-analysis is that we assessed the most important clinical outcome, mortality. We decided not to assess surrogate outcomes commonly related to statin administration, such as myalgia or creatinine kinase elevation, due to the heterogeneity in definitions and spurious reporting. Finally, most of the included trials were focused on perioperative period and did not report clinical outcomes at longer follow-up.
Further high-quality trials should systematically evaluate the relationship between postoperative outcomes and patients' variables (e.g., statin-naïve, chronic kidney disease), to the surgery, and to the statin regimen (e.g., therapy duration).

Conclusions
Our results suggest that perioperative statins appear to be protective against postoperative myocardial infarction in non-cardiac surgery and associated with an increased risk of acute kidney injury in cardiac surgery. There are still insufficient randomized data for firm conclusions on perioperative statin therapy, and possible positive or even negative effects on mortality could not be excluded. No randomized evidence supports the systematic administration of statins in surgical patients, especially in statinnaïve patients. Further RCTs should evaluate the safety profile, possible beneficial effects on patients' outcome, particularly in non-cardiac surgery, and assess the more appropriate time-point for eventual statin discontinuation before surgery in patients under chronic statin therapy.

Additional file
Additional file 1: Table 1. PRISMA 2009 checklist. Table 2. Major exclusions. Table 3. Further characteristics of the included trial. Figure 1. Risk of bias graph.     Figure 6. Trial sequential analysis for postoperative myocardial infarction. Table 4. Primary and secondary analyses. Table 5. Sensitivity analyses. Table 6. Postoperative outcomes in statin-naïve trials.

Abbreviations
AKI: acute kidney injury; CI: confidence interval; NNH: number needed to harm; NNT: number needed to treat; OR: odds ratio; RR: risk ratio; RRT : renal replacement therapy.