Efficacy and safety of bempedoic acid for the treatment of hypercholesterolemia: A systematic review and meta-analysis

Background Bempedoic acid is a first-in-class lipid-lowering drug recommended by guidelines for the treatment of hypercholesterolemia. Our objective was to estimate its average effect on plasma lipids in humans and its safety profile. Methods and findings We carried out a systematic review and meta-analysis of phase II and III randomized controlled trials on bempedoic acid (PROSPERO: CRD42019129687). PubMed (Medline), Scopus, Google Scholar, and Web of Science databases were searched, with no language restriction, from inception to 5 August 2019. We included 10 RCTs (n = 3,788) comprising 26 arms (active arm [n = 2,460]; control arm [n = 1,328]). Effect sizes for changes in lipids and high-sensitivity C-reactive protein (hsCRP) serum concentration were expressed as mean differences (MDs) and 95% confidence intervals (CIs). For safety analyses, odds ratios (ORs) and 95% CIs were calculated using the Mantel–Haenszel method. Bempedoic acid significantly reduced total cholesterol (MD −14.94%; 95% CI −17.31%, −12.57%; p < 0.001), non-high-density lipoprotein cholesterol (MD −18.17%; 95% CI −21.14%, −15.19%; p < 0.001), low-density lipoprotein cholesterol (MD −22.94%; 95% CI −26.63%, −19.25%; p < 0.001), low-density lipoprotein particle number (MD −20.67%; 95% CI −23.84%, −17.48%; p < 0.001), apolipoprotein B (MD −15.18%; 95% CI −17.41%, −12.95%; p < 0.001), high-density lipoprotein cholesterol (MD −5.83%; 95% CI −6.14%, −5.52%; p < 0.001), high-density lipoprotein particle number (MD −3.21%; 95% CI −6.40%, −0.02%; p = 0.049), and hsCRP (MD −27.03%; 95% CI −31.42%, −22.64%; p < 0.001). Bempedoic acid did not significantly modify triglyceride level (MD −1.51%; 95% CI −3.75%, 0.74%; p = 0.189), very-low-density lipoprotein particle number (MD 3.79%; 95% CI −9.81%, 17.39%; p = 0.585), and apolipoprotein A-1 (MD −1.83%; 95% CI −5.23%, 1.56%; p = 0.290). Treatment with bempedoic acid was positively associated with an increased risk of discontinuation of treatment (OR 1.37; 95% CI 1.06, 1.76; p = 0.015), elevated serum uric acid (OR 3.55; 95% CI 1.03, 12.27; p = 0.045), elevated liver enzymes (OR 4.28; 95% CI 1.34, 13.71; p = 0.014), and elevated creatine kinase (OR 3.79; 95% CI 1.06, 13.51; p = 0.04), though it was strongly associated with a decreased risk of new onset or worsening diabetes (OR 0.59; 95% CI 0.39, 0.90; p = 0.01). The main limitation of this meta-analysis is related to the relatively small number of individuals involved in the studies, which were often short or middle term in length. Conclusions Our results show that bempedoic acid has favorable effects on lipid profile and hsCRP levels and an acceptable safety profile. Further well-designed studies are needed to explore its longer-term safety.


Conclusions
Our results show that bempedoic acid has favorable effects on lipid profile and hsCRP levels and an acceptable safety profile. Further well-designed studies are needed to explore its longer-term safety.

Author summary
Why was this study done?
• Lowering low-density lipoprotein cholesterol (LDL-C) is effective for reducing cardiovascular events over time.
• A number of phase II and phase III randomized controlled trials (RCTs) are already available showing encouraging results of bempedoic acid treatment on LDL-C.
• We aimed to perform a systematic review and meta-analysis on the clinical evidence available to date to better define the efficacy and tolerability profile of treatment with bempedoic acid.
• Our results also confirmed that the therapy is overall safe and well tolerated, with no significant increase of serious adverse effects.

Introduction
Cardiovascular diseases (CVDs) are still the leading cause of disability and death in developed countries [1]. As reported by Mendelian randomization studies, a lifetime reduction of lowdensity lipoprotein cholesterol (LDL-C) of 1 mmol/l might reduce the potential risk of atherosclerotic CVDs by over 50% [2]. Controlled clinical studies successfully showed a consistent relationship between the reduction of LDL-C and cardiovascular (CV) risk decrease [3], such that lipid-lowering therapy became a cornerstone in CV risk reduction. Bempedoic acid (8-hydroxy-2,2,14,14-tetramethylpentadecanedioic acid; ETC-1002; Esperion Therapeutics, Ann Arbor, MI) is a first-in-class small-molecule inhibitor of ATP citrate lyase (ACLY), a key enzyme that supplies substrate for cholesterol and fatty acid synthesis [4]. ACLY is essential for growth and development, such that homozygous knockout (Acly − ) in mice is embryonic lethal, indicating non-redundancy during development [5]. By inhibiting ACLY, bempedoic acid induces LDL receptor upregulation and stimulates the uptake of LDL particles by the liver, which contributes to reduction of LDL-C concentration in the blood [6]. Bempedoic acid is administered orally once a day, is quickly absorbed in the small intestine, and has a half-life ranging from 15 to 24 hours [7]. It is a prodrug that is activated by verylong-chain acyl-CoA synthetase 1, an enzyme that is synthesized only in the liver [8]. Even though bempedoic acid acts on the same pathway as statins (3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors), the lack of the activating enzyme in skeletal muscle may prevent the muscular adverse effects associated with statins [8]. For this reason, bempedoic acid may represent a novel treatment to reach LDL-C goals for statin-intolerant patients [9].
A number of phase II and phase III randomized controlled trials (RCTs) are already available, showing encouraging effects of bempedoic acid treatment on LDL-C. Consequently, we aimed to perform a systematic review and meta-analysis of the clinical evidence available to date to better define its efficacy and tolerability profile.

Methods
The study is reported in accordance with the 2009 guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement (S1 PRISMA Checklist) [10], and was registered in the PROSPERO database (registration code: CRD42019129687). Due to the study design (meta-analysis), neither institutional review board approval nor patient informed consent was required.

Search strategy
PubMed (Medline), Web of Science, Google Scholar, and Scopus databases were searched, with no language restriction, using the following search terms: ("Bempedoic acid" OR "ETC-1002") AND ("Trial" OR "Study") [Search terms: (("Bempedoic acid") AND Study) OR ((Bempedoic acid) AND Trial) OR (ETC-1002 AND Study) OR (ETC-1002 AND Trial))]. The wildcard term " � " was used to increase the sensitivity of the search strategy, which was limited to studies in humans. The reference lists of identified papers were manually checked for additional relevant articles. Additional searches for potential trials included the references of review articles on bempedoic acid, and the abstracts from selected scientific conferences on the subject of the meta-analysis. Literature was searched from inception to 5 August 2019.
All abstracts were screened by 2 reviewers (FF and AFGC) in an initial process to remove ineligible articles. The remaining articles were obtained in full-text and assessed again by the same 2 researchers, who evaluated each article independently and carried out data extraction and quality assessment. Disagreements were resolved by discussion with a third party (MB).

Study selection criteria
Original studies were included if they met the following criteria: (i) were a phase II or III RCT with either multicenter or single-center design, (ii) investigated the effect of bempedoic acid on plasma lipids or high-sensitivity C-reactive protein (hsCRP), (iii) tested the safety of bempedoic acid in short-and middle-term administration, and (iv) reported all the adverse events (AEs) that occurred during the treatment.
Studies that lacked a control-treated group for comparison with bempedoic acid were excluded.

Data extraction
Data abstracted from the eligible studies were the following: (i) study registration code; (ii) first author's name; (iii) publication year; (iv) study phase; (v) main inclusion criteria and underlying disease; (vi) treatment duration; (vii) study arms; (viii) number of participants in the active and control group; (ix) age and sex of study participants, (x) baseline and outcome data of total cholesterol (TC), LDL-C, high-density lipoprotein cholesterol (HDL-C), verylow-density lipoprotein (VLDL), non-HDL-C, triglycerides (TGs), apolipoprotein (Apo) B, Apo A-1, and hsCRP; and (xi) discontinuation of treatment and AEs that occurred during the trials. Safety outcomes included: AEs, serious AEs, study-drug-related AEs, AEs leading to discontinuation of treatment, death, major adverse cardiac events, muscle-related AEs, arthralgia, gout, back pain, pain in extremity, pruritus, rash, new onset hypertension, headache, fatigue, dizziness, dyspepsia, abdominal pain, nausea, constipation, diarrhea, nasopharyngitis, sinusitis, cough, dyspnea, upper respiratory tract infection, bronchitis, urinary tract infection, vulvovaginal mycotic infection, new onset or worsening diabetes, neurocognitive disorders, vertigo, increase in blood creatinine level, decrease in glomerular filtration rate, creatine kinase (CK) elevation serum uric acid (SUA) elevation, and liver enzyme (transaminase and gamma-glutamyl transferase) elevation. All the verbatim terms for the AEs were coded to preferred term and System Organ Class with the use of the Medical Dictionary for Regulatory Activities (MedDRA).
Missing or unpublished data were sought by trying to contact authors or sponsors via email, and, in cases of no response, repeated messages were sent. Data extraction and database typing were performed by 2 authors (AFGC and FF) and reviewed by a third author (MB) before the final analysis. Doubts were resolved by mutual agreement among the authors.

Risk of bias evaluation
A systematic evaluation of risk of bias in the included studies was performed using the Cochrane tool [11]. The items used were the following: adequacy of sequence generation, blinding, addressing of dropouts (incomplete outcome data), allocation concealment, selective outcome reporting, and other probable sources of bias [12]. Risk of bias assessment was performed by 2 reviewers (FF and AFGC) independently; disagreements were resolved by a consensus-based discussion. Each item was judged as high, low, or unclear risk of bias. A trial with high risk of bias in the randomization or blinding items was judged as having high risk of bias overall.

Data synthesis
All analyses were performed with Comprehensive Meta-Analysis (CMA) version 3 software (Biostat, Englewood, NJ) [13]. Changes in continuous outcomes were calculated for each study arm by subtracting the value at baseline from the one after intervention. All values were expressed as percent change from baseline. Standard deviations (SDs) of the mean differences (MDs) were obtained as follows, per Follmann and colleagues [14]: , assuming a correlation coefficient (R) of 0.5. If the outcome measures were reported in the original articles as median and interquartile range (or 95% confidence interval [CI]), mean and SD values were obtained as described by Wan et al. [15]. In case standard error of the mean (SEM) was only reported as a dispersion measure, SD was estimated using the following formula: SD = SEM × p n, with n being the number of individuals. To handle the double-counting problem in trials comparing different treatments against a single control group, individuals within the control group were divided by the required comparisons.
Meta-analyses were conducted using a fixed-effect model or a random-effect model (using the DerSimonian-Laird method) and the generic inverse variance method based on the moderately low (<50%) or high (�50%) inter-study heterogeneity, which was quantitatively assessed using the Higgins index (I 2 ) [16]. Effect sizes for lipid and hsCRP changes were expressed as MDs and 95% CIs. For safety analyses, odds ratios (ORs) and 95% CIs were calculated using the Mantel-Haenszel method [17]. If 1 or more outcomes could not be extracted from a study, the study was removed from the analysis involving those outcomes. AEs were included in the analysis only if occurring in at least 2 of the selected clinical trials. The efficacy analysis was performed on the safety population; the analysis of safety data was based on the intention-to-treat population.
For the purpose of evaluating the influence of each study on the overall effect size, sensitivity analysis was conducted using the leave-one-out method (i.e., repeating the analysis after omitting 1 study at a time) [18]. Two-sided p-values � 0.05 were considered statistically significant for all tests.
If statistical heterogeneity was detected, attempts to identify the sources of heterogeneity and potential publication biases were made through the visual inspection of Begg's funnel plot asymmetry, and carrying out the Begg's rank correlation test and Egger's linear regression test [19]. The Duval and Tweedie "trim and fill" method was used to adjust the analysis for the effects of publication bias [20]. In case of a significant result, the number of potentially missing studies required to make the p-value non-significant was estimated by using the classical fail-safe N method as another marker of publication bias. Two-sided p-values � 0.05 were considered statistically significant.

Risk of bias evaluation
The studies reported sufficient information regarding sequence generation, allocation concealment, blinding of participants, personnel, and outcome assessment. Details of the risk of bias evaluation are reported in Table 2.
The classic fail-safe N test suggested that the following number of studies with negative results would be needed to bring the estimated effect size for each outcome to a non-significant level: 2,280 studies for TC (p < 0.001 for the test), 838 studies for HDL-C (p < 0.001 for the test), 2 studies for HDL particle number (p = 0.017 for the test), 2,004 studies for non-HDL-C (p < 0.001 for the test), 2,053 studies for LDL-C (p < 0.001 for the test), 263 studies for LDL particle number (p < 0.001 for the test), 1,308 studies for Apo B (p < 0.001 for the test), and 188 studies for hsCRP (p < 0.001 for the test). The individual analyses are included in S3 Table.
These findings were robust in the leave-one-out sensitivity analyses (S6 Fig). However, when the data from the largest study (the CLEAR Harmony trial) [23] were excluded from the meta-analysis, the effect sizes for the safety outcomes lost their statistical significance (S1 Table).
The incidence of the other AEs did not differ between groups (S4 Table). Considering the reasons for treatment discontinuation in included trials that reported all or part of them (S5 Table), it was not possible to identify the responsible reasons of the effect size of Fig 6 (S7 Fig). Visually, the funnel plot of standard error by log OR was slightly asymmetric only for risk of discontinuation of treatment. This asymmetry was imputed to 6 potentially missing studies on the right side of the funnel plot, increasing the estimated risk to 1.55 (95% CI 1.22, 1.97) (S8 Fig). The presence of publication bias for the analysis was confirmed by Egger's linear regression (p = 0.005), but not by Begg's rank correlation (p = 0.298). The classic fail-safe N test suggested that 1 study with a negative result would be needed to bring the estimated risk of CK elevation to a non-significant level (p = 0.042 for the test), and 2 studies with negative results would be needed to bring the estimated risk of transaminase elevation to a non-significant level (p = 0.021 for the test). The individual analyses are included in S6 Table.

Discussion
Inhibitors of 3-hydroxy-3-methylglutaryl coenzyme A reductase (statins) represent the firstline treatment for dyslipidemia, being able to reduce LDL-C by 30%-50% and subsequently  decrease the incidence of CV events [30]. Despite the highly favorable benefit/risk profile of statins, a large number of patients are statin intolerant or need additional lipid-lowering drugs to reach optimal LDL-C levels [3]. The current meta-analysis shows that bempedoic acid safely reduces LDL-C levels by about 23%, suggesting that it might be considered as an effective alternative or add-on therapy to statins or ezetimibe.
About 31%-49% or more of patients with hyperlipidemia do not achieve LDL-C goals with current lipid-lowering therapies [31,32], and more than half of patients stop statin treatment within 1 year of initiation [33]. Sixty percent of patients who discontinue statins report different symptoms of drug intolerance as the main reason for discontinuation [34]. Statin intolerance, usually characterized by myalgia, myositis, and/or myopathy, occurs in 2%-15% of users, the estimate being strongly variable in epidemiological and rechallenging studies [35,36]. Furthermore, large meta-analyses showed that statin treatment is associated with a 9%-13% increase in risk of developing diabetes [37]. However, scientifically unsupported concerns about statin safety spread by mass media lead to the formation of a negative image of these drugs and increase of their cessation rate [38].
Additional treatments of dyslipidemia include ezetimibe (second-line) and fenofibrate (third-line). Ezetimibe, in combination with statin therapy, lowers LDL-C by an additional 20% or so [39] and significantly reduces the risk of major adverse CV events, non-fatal myocardial infarction, and non-fatal stroke compared with statins alone, with less or no effect on fatal endpoints [40]. A simulation based on adding ezetimibe in a huge statin-treated cohort suggests that the percentage of patients with LDL-C > 1.8 mmol/l and >2.4 mmol/l would fall from 65% to 38% and from 25% to 12%, respectively [41]. Fibrates are less effective on LDL-C levels, with their main indication being moderate-tosevere hypertriglyceridemia, such that they are rarely used in cardiology settings. However, a large meta-analysis of 16,112 patients showed evidence for a protective effect compared to placebo for the primary composite outcome of non-fatal myocardial infarction, non-fatal stroke, and vascular death [42]. Besides, patients with very high baseline LDL-C level or very high or extreme global CV risk need additional lipid-lowering drugs to optimize the lipid profile [43], especially in light of the most recent international recommendations [44,45]. Monoclonal antibodies that target proprotein convertase subtilisin/kexin type 9 (PCSK9) have recently been demonstrated to dramatically reduce LDL-C level (even over 60%) in the majority of cases, while significantly reducing CV risk; however, their cost-benefit ratio is yet under discussion, and in many countries their use is limited due to strict reimbursement rules [46]. In this context, there is yet place for the development of new less-expensive, effective/safe lipid-lowering drugs.
By analyzing data from 10 phase II and phase III RCTs including a total of 3,788 patients, we confirmed that bempedoic acid significantly reduced TC (by 15%), non-HDL-C (by 18.2%), LDL-C (by 22.9%), LDL particle number (by 20.7%), Apo B (by 15.2%), and hsCRP (by 27%), while negatively affecting serum levels of HDL-C (−5.8%) and HDL particle number (−3.2%). These findings strengthen the unpowered data previously reported by Wang et al., based on only 625 patients [47]. These findings could also be quantitatively relevant, since they have usually been obtained when bempedoic acid is administered on top of an effective lipidlowering treatment, with a quite good safety and tolerability profile.
Our results also confirmed that bempedoic acid therapy is overall safe and well tolerated, with no significant increase of serious AEs. However, an increase of drug discontinuation and elevations of SUA, transaminase, and CK were observed. The detailed analysis of the reasons for discontinuation (see S5 Table) reported in the available trials does not give any clear pattern that could explain the 37% increased risk of discontinuation of bempedoic acid in comparison to placebo; this issue, however, needs to be further investigated. As for the other adverse effects possibly related to bempedoic acid, it is important to emphasize that in the 4 trials where CK increase was reported, it was observed in only 16 patients (of 1,792 investigated; 0.9%), and only single patients had a repeated and confirmed CK elevation greater than 5 times the upper limit of normal. More data with longer follow-up are also necessary to confirm the risk of SUA increase with bempedoic acid (observed in only 3 trials, where SUA increase was observed in 18/354 [5%]), as well as the risk of transaminase increase (observed in 5 trials, where transaminase increase was observed in only 1.1% of patients [20/1,823] in active-treated group). It is also worth emphasizing that bempedoic acid, due to its mechanism of action, does not increase the risk muscle-related side-effects and significantly reduces the risk of worsening or new onset diabetes by about 40% (however, based only on 2 available studies)-AEs that might be relatively often observed in statin trials, especially for high-and very-high-risk patients requiring intense therapy.
In this context, bempedoic acid seems to be an interesting option as an overall safe drug to be easily associated to statins and ezetimibe. In particular, the drug will be marketed as monotherapy or in a single pill with ezetimibe for the management of statin-intolerant patients. Considering the different mechanism of action of bempedoic acid and ezetimibe, the high safety profile of both drugs, and the lack of interaction risk between them, it is expected that this association will be a relatively effective and safe lipid-lowering treatment.
The main limitation of this meta-analysis is related to the relatively small number of patients involved in the studies, which were often short or middle term, as well as their heterogeneity (including different populations that were investigated, i.e., patients with type 2 diabetes, hypercholesterolemia, or statin intolerance). Moreover, heterogeneity of effects is moderate to large across most of the biochemical outcomes. Data on decreased CV events and mortality are lacking for bempedoic acid as well [48].
In conclusion, the current meta-analysis demonstrates an acceptable safety profile and multiple positive effects of bempedoic acid on lipid profile and hsCRP serum levels.
Further data on the cost-benefit efficacy of bempedoic acid treatment will come from the CLEAR Outcomes study, a phase III, event-driven, randomised, multicenter, double-blind, placebo-controlled trial designed to evaluate whether treatment with bempedoic acid reduces the risk of CV events. The primary endpoint of the study is the effect of bempedoic acid on major adverse CV events (CV death, non-fatal myocardial infarction, non-fatal stroke, and coronary revascularization). The enrollment ended in November 2019 [49].