Safety and Efficacy of Proprotein Convertase Subtilisin-Kexin Type 9 Inhibitors After Acute Coronary Syndrome: A Meta-analysis of Randomized Controlled Trials

We included all published randomized controlled trials (RCTs) investigating the effect of PCSK9 inhibitors in patients diagnosed with ACS or who had a history of an ACS disease over the past 12 months with no restrictions on age, ethnicity or country. Patients with acute coronary syndrome [ST-elevation myocardial infarction (STEMI)/non-ST-elevation myocardial infarction (non-STEMI)/unstable angina] who failed to reach the target lipid levels (according to ESC/EAS 2019 guidelines) [16] despite other lipid-lowering drugs, who were randomized to PCSK9i as an intervention group and placebo and/or other lipid-lowering agents as a control group, and who reported outcomes of interest in an intention-to-treat analysis were excluded. We excluded any observational studies, non-randomized clinical trials, review articles, book chapters, press releases, non-English studies, animal studies and duplicates. Moreover, studies with only an abstract, unavailable full text or overlapped data and studies without a comparison group were also excluded.

The efficacy outcomes were the mean change in low-density lipoprotein cholesterol (LDL-C), total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C), non-HDL-C, lipoprotein (a) [Lp (a)], triglycerides (TG), apolipoprotein A1 (Apo-A1) and apolipoprotein B (Apo-B). Moreover, the safety outcomes were the incidence of any adverse events, all-cause mortality, cardiovascular specific mortality, myocardial infarction (MI), cerebrovascular events [stroke/transient ischemic attack (TIA)], coronary revascularization or coronary heart disease event (CHD) which is defined as death from coronary heart disease, nonfatal myocardial infarction, unstable angina requiring hospitalization.

We systematically searched the following databases: PubMed, Scopus, Web of Science (WOS) and Cochrane Library to include the relevant published RCTs from inception until 2 December 2022. The search terms used were as follows: (alirocumab OR SAR236553 OR SAR-236553 OR REGN-727 OR “monoclonal antibody REGN727” OR “REGN727 monoclonal antibody” OR “Monoclonal Antibody” OR “Anti- Cholesterolemic” OR REGN727 OR praluent OR “PCSK9 Inhibitors” OR PCSK9 OR Evolocumab OR Repatha OR AMG-145 OR AMG 145 OR inclisiran OR leqvio) AND ((acute coronary syndrome) OR STEMI OR (non-STEMI) OR NSTEMI OR (unstable angina*) OR (myocardial infarction*) OR (Myocardial Infarct*) OR (Heart Attack*) OR (Unstable Angina Pectoris) OR (Pre-infarction Angina) OR (ST-elevation myocardial infarction) OR (Non ST-elevation myocardial infarction)). Furthermore, we performed manual research by backward citation analysis and checking cited articles in Google Scholar to identify all relevant articles.

Duplicates were removed using Endnote software. Two reviewers blindly screened the retrieved references using Rayyan [17] in two phases: the first phase was the title and abstract screening of all extracted articles, and the second phase was a further full-text screening of all eligible abstracts. Any discrepancies were resolved through a third reviewer.

Two independent reviewers extracted the data from the eligible studies into a homogeneous data-extraction sheet, cross-checking was applied and all discrepancies were discussed.

The following data are extracted to a uniform data extraction sheet that includes the following three domains: study characteristics (study design, country, duration, sample size, population, study arms and follow-up period), population characteristics (age, sex, BMI, index ACS event, active smoking, hypertension, diabetes mellitus, cerebrovascular events, family history of CAD, peripheral arterial disease, previous myocardial infarction, previous CABG, previous PCI, statin therapy, β-blockers, ACE inhibitors, ezetimibe, LDL-C and other atherogenic markers), and the outcomes mentioned in the inclusion criteria.

The quality of the included studies was evaluated by two blinded investigators using the Cochrane Collaboration’s Risk of Bias tool (version 2, RoB2) [18]. This tool uses a composite score of five main domains: randomization process, deviation from the intended interventions, missing outcome data, measurement of the outcome and selection of the reported result. For each of these domains, the judgments of the investigators are categorized as ‘low risk’, ‘some concerns’ or ‘high risk’ of bias. Disagreements were reanalysed and solved by a third investigator. The summary and graph of the RoB2 tool were generated on the Review Manager software (RevMan, version 5.4) for Windows [19].

Efficacy outcomes were evaluated through either LDL-C, HDL-C, TC, non-HDL-C, Lp (a), TG, Apo-A1 and/or Apo-B outcomes and reported as mean [standard deviation (SD)], while safety outcomes, which were any adverse events, all-cause mortality, cardiovascular specific mortality, myocardial infarction (MI), cerebrovascular events (stroke/TIA), coronary heart disease (CHD) and/or coronary revascularization, were expressed as an event and total.

Dichotomous data were pooled as an odds ratio (OR) with its 95% confidence interval (CI), while continuous data were pooled as a Cohen’s d with its 95% CI. STATA software 17 for Mac (StataCorp, 2021) [20], was used for all statistical analyses and for forest plot generation. In any study that reported data in the median (range or interquartile range), the Wan et al. method [21] was applied to convert it to mean (SD).

DerSimonian Laird random effects model was adopted to calculate the pooled effect size for all outcomes, which assigns more weight to a small number of studies on the cost of larger studies, to provide pooled estimates with a larger standard error to accommodate for any inconsistent effect sizes. Therefore, possible inconsistency in our computed meta-analysis effect sizes must be taken into consideration as those estimates were conservative. We reported the outcomes of the last follow-up point, in case studies reported the outcomes in multiple timepoints.

Statistical heterogeneity between studies was evaluated by visual inspection of the forest plots and measured by a chi-square test (Cochrane Q test) and Higgns and Thompson I² using the following equation: I² = ((Q − df)/Q) × 100%. Heterogeneity was considered to be significant if the p value of the chi-square test was less than 0.1 and to be low, moderate and high if I² was < 25%, from 25–75% or > 75%, respectively [22].

We used funnel plots to examine the publication bias according to Egger and colleagues [23, 24] using Egger’s test [23], while the Galbraith plot was used to visualize the heterogeneity between studies. To adjust for the potential bias and test the robustness among the included studies, we conducted a sensitivity analysis, excluding studies with a baseline LDL-C below 100 mg/dl [25].

After searching through databases, we retrieved a total of 5664 records. Of which, 1596 records were identified as duplicates by Rayyan [17]. We screened records to include only RCTs that met our inclusion criteria, and finally 11 studies were included in this review as shown in Fig. 1.

We included 11 RCTs of 24,732 patients, grouped as 13,111 patients receiving PCSK9i and 11,621 patients as the control group. The included studies varied in terms of number of participants (from 20 to 18,924) sittings (most of them in European countries) and in follow-up. A summary of the included studies and baseline characteristics of the patients included are summarized in Tables 1, 2 and 3.

We assess the risk of bias of 11 RCTs included using ROB2, of which seven studies [26‐32] had an overall low risk of bias, while the other studies showed an overall unclear bias due to the following reasons: (1) their method of randomizing patients, nearly 20%, and (2) missing outcomes reporting data in some studies, nearly 30%, as shown in Fig. 2.

The incidence of MI and cerebrovascular events were significantly lower in the PCSK9i group compared with the placebo group; the pooled OR was 0.87 [95% CI (0.78–0.97), p = 0.01] for MI and 0.71 [95% CI (0.52–0.98), p = 0.04] for cerebrovascular events. The pooled studies were homogeneous for the two reported outcomes [(I² of 0%, p = 0.63) and (I² of 2%, p = 0.31), respectively], as shown in Supplementary Figs. 5 and 6. There was no significant difference between the two studied groups in terms of any adverse events, all-cause mortality, cardiovascular-specific mortality, CHD and coronary revascularization, respectively. The pooled studies were homogeneous [(I² of 13%, p = 0.71), (I² of 14%, p = 0.46), (I² of 0%, p = 0.82), (I² of 24%, p = 0.21) and (I² of 18%, p = 0.20), respectively], as shown in Supplementary Figs. 7, 8, 9, 10 and 11. The results of safety outcomes are summarized in Fig. 9.

The results for ‘HDL-C,’ ‘triglycerides’ and ‘Apo-A1’ analyses were consistent. On the other hand, the rest of the efficacy outcomes for ‘LDL-C’, ‘total cholesterol’, ‘non-HDL-C’, ‘Lp (a)’ and ‘Apo-B’ all exhibited considerable heterogeneity, which could be attributable to the high risk of bias in some of the included trials [25, 33‐35]. Furthermore, conducting sensitivity analysis on these outcomes by the exclusion of Ako et al. [25] illustrated a marked reduction of heterogeneity. Ako et al. [25] showed an extreme advantage of PCSK9i making the analysis appear inconsistent, which could be related to the low levels of LDL-C at baseline among its participants, with means of 97.9 mg/dl, and 95.9 mg/dl for PCSK9 inhibitors and control groups, respectively.

The mechanism by which PCSK9i obtains their action is by increasing the number of LDL receptors on the surface of the liver by inhibiting PCSK9, which stimulates LDL receptor degradation, resulting in increased LDL absorption by liver cells and reduced LDL levels in the blood [36]. The reduction in LDL-C levels is independent of statin activity, which inhibits HMG-CoA reductase, an enzyme involved in cholesterol production [37].

The total cholesterol (TC) in the blood comprises both LDL and HDL. When LDL levels are reduced, TC levels are often reduced as well. Apo-B is a protein present on LDL particles that helps in their binding to cell receptors. When LDL cholesterol levels fall, so does Apo-B [38]. Triglycerides are another type of blood lipids that can contribute to the development of heart disease. When LDL levels are reduced, triglyceride levels can also be reduced [39]. The rationale for the PCSK9i-induced drop in Lp (a) levels is unknown, as Lp (a) has never been thought to have a substantial role in the metabolization of the LDL receptor. The rise in HDL and Apo-A1 levels may be related to an increase in LDL-C absorption by the liver, which results in a decrease in very low density lipoprotein (VLDL) generation and an increase in HDL production [40, 41].

In our meta-analysis, the bulk of the population was evaluated in two of the included trials [32, 34]. Both studies were large multicentre trials that took place between 2011 and 2017, with a median follow-up time of 2.8 years in Schwartz et al. [32] and 11.1 months in Sabatine et al. [34]. The study by Schwartz et al. [32] used alirocumab SC (75 mg) in the experimental group and placebo in the control group in their trial which included 18,924 patients from 1315 different centres. While the study by Sabatine et al. [34] included 4465 patients from 305 centres, evolocumab was used in the experimental group and standard statin therapy was used in the control group. The findings of the two studies matched our efficacy and safety findings.

The main strength point of this review is that it is the first meta-analysis to assess the safety and efficacy of PCSK9 inhibitors among patients with ACS. Multiple reviews were performed on the same topic [42, 43]. However, none of them included more than three trials or performed a meta-analysis. A large meta-analysis [44] evaluated the safety and efficacy of the same type of drugs, however, they enrolled patients with broader criteria. Their participants were adults with dyslipidaemia and/or established atherosclerotic cardiovascular diseases, in addition, they did not consider the lipid markers during their meta-analysis, unlike the current review. Furthermore, we limited the included studies whose populations are patients with ACS or who experienced recent episodes.

The previous meta-analysis [44] studied the efficacy and safety of PCSK9 inhibitors in a slightly different manner, as they considered all-cause mortality, cardiovascular-specific mortality, MI, coronary revascularization and ischemic stroke as efficacy outcomes. While their safety endpoints were drug discontinuation, neurocognitive adverse events, liver enzymes elevation, allergic reactions, haemorrhagic stroke, rhabdomyolysis, new onset of diabetes mellitus and injection site reactions.

Our findings matched with those of the previous meta-analysis [44] in terms of all-cause mortality, cardiovascular-specific mortality and MI. According to the study, PCSK9 inhibitors did not improve both all-cause and cardiovascular-specific mortality while significantly reducing MI incidence. They did, however, observe a considerable improvement in coronary revascularization, which contradicts the findings of the current analysis. This could be due to the several differences between the two studies stated above.

Another study strength is the vast number of trials included (11 RCTs), with a very powerful sample size (24,732 patients), from various locations and durations. This may demonstrate the actual effect of these medications despite the different settings of included trials.

The results of our study indicate that PCSK9 inhibitors could be prescribed for patients with ACS in addition to the standard regimen regarding their vast effect on lipid biomarkers, including, but not limited to, a reduction in LDL-C and TC levels and an increase in HDL-C levels. Moreover, PCSK9 inhibitors has a safe profile with no major cardiovascular events and a favourable effect in reducing the incidence of MI and cerebrovascular events. Together, these effects highlight the potential pharmacological benefits of PCSK9 inhibitors in patients with ACS. Regarding clinical research, we recommend further RCTs to address the current issue with longer follow-up time, since 37.6% of the control group in Koskinas et al. achieved target LDL cholesterol values. So, we need an update on how much we expect PSCK9i will benefit from additional LDL-C lowering effect. Long-term studies are lacking in current literature.

There are several important limitations of our study that we must address. First, the current findings are constrained by the intrinsic limitations of the included RCTs in terms of study design, sample size, definitions, risk of bias and event detection. Second, the inclusion, exclusion criteria, doses of intervention and periods of follow-up were not homogeneous throughout the included RCTs. Third, although we justified them, the elevated values of heterogeneity that were detected between the included RCTs, particularly in terms of efficacy outcomes, are considered a limiting factor. Fourth, whereas the majority of patients in this meta-analysis received high-intensity statin therapy, some received other lipid-lowering medications due to statin intolerance or other factors. Fifth, we observed a scarcity of data that discusses the safety outcomes among the majority of included RCTs.