Introduction

Familial hypercholesterolemia (FH) and statin-intolerant patients are suffering high risk of cardiovascular disease (CVD)1, 2. Clinical guidelines advocate that low density lipoprotein cholesterol (LDL-C) is the target of CVD’s primary or secondary prevention3. Statins, the recommended first-line therapy to control lipidemia, tend not to be thoroughly effective4. In addition, of approximately 20 million statin users, an estimated 10% to 20% are statin-intolerant5, failing the goals of lowering blood lipid profile or developing statin-intolerance, such as injection-site reaction, nasopharyngitis, upper respiratory tract infections, influenza, cough, nausea, myalgia, myositis, headache, diarrhea, fatigue, abnormal pain, rectal bleeding, dehydration, arthralgia, back pain etc.6,7,8,9,10,11,12,13,14,15,16,17 FH, one of the most common genetic disorders in humans that endangers approximately 12 million people worldwide, can elevate the significant premature CVD morbidity and mortality18. For extremely high baseline LDL-C levels of FH, usually >5.2 mmol/L (200 mg/dL), even statins cannot achieve intensive LDL-C lowering targets, leading to early CVD with typical onset before age 50 in men and 60 in women19.

Proprotein convertase subtilisin/kexin9 (PCSK9) monoclonal antibodies, as an attractive therapy for lowering LDL-C levels20, can bind the LDL-receptor (LDL-R) on the surface of hepatocytes, interfering with LDL clearance in circulation, hence playing a pivotal role in regulating cholesterol homeostasis21. The human monoclonal antibodies against PCSK9 primarily include AMG145/Evolocumab and REGN727/SAR236553/Alirocumab, while RN316/bococizumab, RG7652, LY3015014, and ALN-PCSSC22,23,24,25, are now genetically validated as novel PCSK9-mAb therapies.

In the last 4 years, early clinical trials have proven that PCSK9-mAb can lower the plasma LDL-C level in FH and statin-intolerant patients. The other lipids and lipoproteins; high density lipoprotein cholesterol (HDL-C), total cholesterol (TC), triglycerides (TG), lipoprotein(a) (Lp(a)), apolipoprotein-B (Apo-B) and apolipoprotein-A1 (Apo-A1) can also benefit. Currently, there is no report to comprehensively pinpoint the applicable targets of PCSK9-mAbs—FH and statin-intolerant patients with sufficient clinical outcomes. To compare the efficacy and safety of PCSK9-mAb on FH and statin-intolerant patients, a total of 15 articles were assessed in this meta-analysis.

Results

Study selection and characteristic

A total of 178 studies were searched in our systematic literature, with 43 duplicate publications. 106 studies were unable to meet the inclusion criteria excluded; 14 studies were further ruled out, where 6 studies were articles, 2 studies had no control groups, 1 study was not RCT, 1 was open label trial and 3 were not meta-analysis with quantitative synthesis. (Figure 1) As a result, 15 studies encompassing a total of 4,288 patients were selected26,27,28,29,30,31,32,33,34,35,36,37,38,39,40. Among them, 7 trials used evolocumab (AMG 145) and 8 studies with alirocumab (SAR236553/REGN727) treatment. Baseline characteristics were detailed giving the substantially similar basic values between PCSK9-mAbs and controls. Mean age of the subjects ranged from 31 to 65 years old. All trials were published between 2012 and 2015, followed up for 8 to 78 weeks, with a low risk of bias (Table 1 and Figure 2).

Figure 1
figure 1

Preferred reporting items for systematic review and meta-analysis (PRISMA) flowchart of the process of study selection.

Full size image
Table 1 Baseline Characteristics of Trials Included in Meta-Analysis.
Full size table
Figure 2
figure 2

Risk-of-bias graph and summary table: review authors’ judgments about each risk-of-bias item presented as percentages across all included studies.

Full size image

Quantitative synthesis and heterogeneity of data

According to the efficacy outcomes of PCSK9-mAbs on FH and statin-intolerant patients, the LDL-C level dropped by the greatest margin, −49.59%, 95% CI: −57.08 to −42.10% versus control groups. There was significant heterogeneity in the comparisons (I2 = 100%). The reduction of LDL-C level in FH (−53.28%, 95% CI: −59.88 to −46.68%) was even greater than that in statin-intolerant patients (−34.95%, 95% CI: −41.46 to −28.45%). (Figure 3).

Figure 3
figure 3

Forest plot depicting the efficacy outcomes of PCSK9 monoclonal antibody on familial hypercholesterolemia and statin-intolerant patients.

Full size image

Efficacy outcomes of PCSK9-mAbs

As lipid outcomes of evolocumab, significant reduction of LDL-C level was achieved in all dosages of evolocumab (mean reduction: −51.75%, 95% CI: −53.58 to 49.92%, I2 = 92.6%), with the greatest decrease in biweekly 140 mg evolocumab (−59.30%, 95% CI: −60.75 to −57.85%), monthly 350 mg evolocumab (−43.80%, 95% CI: −44.88 to −42.72%), monthly 420 mg evolocumab (−42.19%, 95% CI: −65.62 to −18.75%) versus placebo. There was significant heterogeneity in the comparisons (I2 = 99.3%). The effect might be dose-dependent and biweekly 140 mg evolocumab tends to be the optimal administration. Compared with ezetimibe or atrovastatin controls, obvious LDL-C reduction also occurred in all dosages of evolocumab, −30.4% (95% CI: −30.95 to −29.85%) versus ezetimibe and −48.9% (95% CI: −51.27 to −46.53%), −55.90% (95% CI: −58.76 to −53.04%) versus atrovastatin controls respectively. No significant heterogeneity was detected in the comparisons (I2 = 0%), demonstrating great similar but less remarkable outcome s compared with those versus controls. (Figure 4a).

Figure 4
figure 4

Forest plots depicting the effect of Evelocumab for subgroup analysis after grouping by methods of treatment. (a) on LDL-C; (b) on HDL; (c) on TC; (d) on TG; (e) on Apo-B; (f) on Lp(a); (g) on Apo-A1.

Full size image

In addition, HDL-C level obviously increaded by 6.91% (95% CI: 6.36 to 7.46%, I2 = 84%) and Apo-A1 level by 2.39% (95% CI: 1.91 to 2.87%, I2 = 91%) versus placebo, while significant reductions of TC, TG, Lp(a) and Apo-B were generated by monthly 420 mg evolocumab treatment, which were −39.42% (95% CI: −40.17 to −38.66%), −16.65% (95% CI: −17.89 to −15.40%), −28.96% (95% CI: −29.86 to −28.07%) and −43.65% (95% CI: −44.51 to −42.78%) respectively, with levels of heterogeneity (I2 = 99%, 95%, 99% and 96%). (Figure 4b–g).

Concluding lipid outcomes of alirocumab, LDL-C level was obviously decreased by all dosages of alirocumab (mean reduction: −57.04%, 95% CI: −61.17 to −52.91%), with significant heterogeneity in the comparison (I2 = 100%). Significant reductions were achieved by biweekly 150 mg alirocumab (58.79%, 95% CI: −63.22 to −54.37%), biweekly 100 mg alirocumab (−55.90%, 95% CI: −68.43 to −43.37%) and biweekly 50 mg alirocumab (−39.70%, 95% CI: −46.93 to −32.47%) versus placebo. With the same frequency of administration, greater reduction was achieved by biweekly 150 mg than 100 or 50 mg alirocumab, adding more evidences to dose dependent effect. Compared with ezetimibe or atrovastatin controls, significant reductions of LDL-C also occurred: −30.40% (95% CI: −30.95 to −29.85%) in alirocumab versus ezetimibe; −48.90% (95% CI: −51.27 to −46.53%) in alirocumab versus placebo both under atorvastatin 10 mg once daily; −55.90% (95% CI: −58.76 to −53.04%) in alirocumab versus placebo both under atorvastatin 80 mg once daily, with no significant heterogeneity. Few studies showed largely changes of LDL-C level versus ezetimibe or atorvastatin controls, demonstrating clearly similar but less remarkable results. (Figure 5a).

Figure 5
figure 5

Forest plots depicting the effect of Alirocumab for subgroup analysis after grouping by methods of treatment. (a) on LDL-C; (b) on HDL; (c) on TC; (d) on TG; (e) on Apo-B; (f) on Lp(a); (g) on Apo-A1.

Full size image

Biweekly 150 mg alirocumab treatment significantly elevated HDL-C level by 6.65% (95% CI: 6.48 to 6.81%, I2 = 100%) and Apo-A1 level by 4.38% (95% CI: 4.24 to 4.52%, I2 = 97%) versus placebo. Substantial reductions of TC, TG, Lp(a), Apo-B were respectively achieved by −20.90% (95% CI: −21.25 to −20.55%, I2 = 0), −13.00% (95% CI: −13.30 to −12.70%, I2 = 100%), −23.32% (95% CI: −23.59 to −23.06%) and −39.50% (95% CI: −39.71 to −39.29%, I2 = 100%). (Figure 5b–g).

Safety outcomes of PCSK9-mAbs

Safety endpoints covering the common adverse events, serious events and laboratory adverse events were compared between the PCSK9-mAbs and control groups. The pooled estimation of evolocumab showed the overall incidence of common adverse events (RR: 0.87, 95% CI:0.56 to 1.34, I2 = 43%) and serious events(RR: 0.79, 95% CI: 0.31 to 2.03, I2 = 0%) versus placebo, which implied no obvious difference. No significant heterogeneity was found in hepatotoxicity risk analysis of evolocumab with levels of alanine aminotransferase (ALT) or aspartate aminotransferase (AST) no greater than three times the upper limit of normal (ULN) by RR: 1.32, 95% CI: 0.29 to 5.94, I2 = 0%. (Figure 6a) The alirocumab treatment implied no significant difference versus placebo, with the common adverse events by RR: 1.05, 95% CI: 0.81 to 1.37, serious events by RR: 0.97, 95% CI: 0.65 to 1.44 versus placebo. No significant heterogeneity was found in the analysis of abnormal liver function risk (AST/ALT ≥ 3 × ULN) in patients by RR: 2.14, 95% CI: 0.67 to 6.82). (Figure 6b) We have included an additional table for pre-specified safety end point which describes the safety events of interest, common adverse events and laboratory results of PCSK9-mAbs. No statistical differences between PCSK9-mAbs and control group except any and paraesthesia events, which were positive in the PCSK9-mAbs group. χ2 statistics was to assess the magnitude of heterogeneity and P value < 0.05 was considered to be statistically significant. (Table 2).

Figure 6
figure 6

Forest plot depicting the adverse event rates following (a) evolocumab and (b) alirocumab therapies, compared with placebo or ezetimibe controls on adverse events, serious events and ALT or AST rates.

Full size image
Table 2 Pre-specified Safety End Points.
Full size table

Discussion

As far as we know, this is the first meta-analysis using sufficient clinical outcomes comprehensively to define the applicable targets of PCSK9-mAbs—FH and statin-intolerant patients. In the present analysis, a total of 15 studies encompassing 4,288 patients with FH and statin-intolerence were included. The main point addressed is whether PCSK9-mAb treatment have the ability to improve the lipid levels of FH and statin-intolerant patients with satisfactory safety and tolerability.

At present, additional on-going large-sized multicenter randomized studies were not included in our study, such as the Phase III FOURIER, GLAGOV and GAUSS-3 trial41,42,43, which are intended to determine if PCSK9-mAbs can help to not only reduce LDL-C level obviously, but also improve cardiovascular related events such as myocardial infarction(MI), and the rates of morbidity and mortality. Nevertheless, the concept of LDL-C reduction, as a surrogate for prevention of long-term cardiovascular outcomes in high-risk patients with acute coronary syndromes, was provided by the recently published IMPROVE-IT study with the strongest clinical trial evidence. The more reduction of LDL-C, the rate of major vascular events decreases44. (Figure 7) Given the consistent effects on LDL-C reduction, PCSK9-mAbs appears to be a promising approach to cut down the risk for cardiovascular events in the patients selected.

Figure 7
figure 7

Plot of the IMPROVE-IT Trial Data and Statin Trials for Change in LDL-C versus Clinical Benefit.

Full size image

In a broader sense, LDL particles, non-HDL-C, the Apo-B/Apo-A1 ratio, and Lp(a) may be the better replacement markers for cardiovascular mortality in the foreseeable future, compared with LDL-C levels45,46,47,48. Overall, results of the study are suggestive of therapeutic efficacy of PCSK9-mAbs in curtailing the major risk factors (including LDL-C, TC, TG, Lp(a) and Apo-B) of CVD, and increasing HDL-C and Apo-A1 by detecting both evelocumab and alirocumab. It has been previously suggested that Apo-B/Apo-A1 ratio is a better risk indicator for CVD and MI than the level of lipids49. In our study, Apo-B decreased while Apo-A1 elevated in the PCSK9-mAb treatment group. The decrease in Apo-B/Apo-A1 ratio obviously indicates that PCSK9-mAb therapy greatly lowers the primary risk factors of heart disease. We also found that the reduction of LDL-C in FH patients (−53.28%, 95% CI: −59.88 to −46.68%) was greater than that in statin-intolerant patients (−34.95%, 95% CI: −41.46 to −28.45%). In terms of safety, PCSK9-mAbs may lead to many events, none of which is life-threatening; the serious events do not increase compared with the control group; more large RCTs are needed to further confirm the safety.

Recently, there have been four meta-analyses comparing the effects of PCSK9-mAbs over placebo controls, with or without considering statin therapy50,51,52,53. In these studies, AMG145/Evolocumab, REGN727/SAR236553/Alirocumab were primarily included. However, novel clinical studies are being done to evaluate the target of LDL-C, such as RN316/bococizumab, RG7652, LY 3015014, ALN-PCSSC, which has been genetically validated to lower the cardiovascular risk.

The mechanism of improving the survival of patients treated with PCSK9mAb remains obscure and its role in the observed survival benefit, for its encouraging result, should be interpreted with caution. The efficacy of these agents in reducing lipid levels (particularly LDL-C) and rates of MI due to plaque stabilization is well-evaluated. Moreover, the lower frequency of PCSK9-mAbs intake may lead to slighter hepatotoxicity. However, subcutaneous injection of PCSK9-mAb makes it an alternative for drug adherence. What’s more, studies reveal that LDL-R can act as the entry point for hepatitis C virus, which may increase the risk of infection54. Finally, it still remains unknown whether the reduced LDL-C by PCSK9-mAbs can improve the long-term clinical outcomes55.

Several limitations should be taken into consideration. First, a few studies have only been reported in abstracts or presented at meetings, leading to added detection bias. Besides, significant heterogeneities were observed in most of the reported outcomes, but we failed to reveal the heterogeneities by dividing subgroups or sensitivities methods. Some large multicenter RCTs are still under investigation. Therefore, caution should be taken in interpreting the results of the meta-analysis when combining the heterogeneous data sets. Despite these limitations, our meta-analysis covers the most applicable targets of PCSK9-mAbs—FH and statin-intolerant patients, and their clinical outcomes are sufficient enough to compensate our clinical guidelines. Hopefully, more RCTs could be carried out to provide more evidences about its long-term therapeutic efficacy, safety, and clinical outcomes.

Methods

Study searching

In order to identify appropriate RCTs, a comprehensive literature search was performed throughout PubMed, EMBASE and The Cochrane Library databases, with the following terms and key words: evolocumab, AMG 145, alirocumab, SAR236553, REGN727x, PCSK9 monoclonal antibodies, FH, statin intolerant and randomized controlled trial from it’s inception in 2012 to May 26, 2016. All the selected studies, restricted to English, were initially screened for relevance at the abstract level.

Study selection

The inclusion criteria are as follows: (1) RCTs; (2) population: FH, statin intolerant patients; (3) safety and efficacy outcomes of PCSK9-mAbs; (4) mean differences (MDs) with 95% corresponding confidence interval (95% CI) and so on. Studies that were not randomized, special-population-targeted or designed to test PCSK9-mAbs with limited number of trials (such as bococizumab, RG7652, LY3015014 and ALN-PCSSC) were excluded.

Outcomes

The primary efficacy endpoints included parameter changes following PCSK9-mAbs treatment from baseline: (1) LDL-C reduction; (2) other lipid profile changes. Therapeutic safety was evaluated by rates of the common adverse events, serious events and laboratory adverse events respectively.

Data collection

Two investigators (LJQ and YG) performed eligibility assessment with a standardized data extraction individually and another reviewer (YMZ) checked the data. Basic information was extracted as follows: study/author, year, design, diagnosis, control, drug regimen, duration, patient number, mean age (y) at baseline. As a precedence, we extracted the corresponding mean differences, 95% CI from baseline for each lipid items, such as LDL-C, HDL-C, TC, TG, Apo-B and Apo-A1 levels, before and after PCSK9-mAbs treatment as the primary outcomes. Safety endpoints covering the common adverse events, serious events and laboratory adverse events were compared between the treatment and control groups. In addition, the common adverse events included injection site reaction (e.g. generalized pruritis, hypersensitive reaction, erythema, rash, swelling, discoloration, or pain), nasopharyngitis, upper respiratory tract infections, influenza, cough, nausea, myalgia, myositis, headache, diarrhea, fatigue, abnormal pain, rectal bleeding, dehydration, arthralgia, and back pain. The serious events, fatal and life threatening, require hospital admission or prolonged stay in the hospital for the possibility of persistent or significant disability. The laboratory adverse events evaluated hepatotoxicity with ALT or AST levels ≥3 × ULN.

Quality assessment

Two reviewers (LJQ and YG) assessed the risk of bias with the Cochrane Collaboration’s tool (Review Manager version 5.3) in the included trials. For efficacy outcomes, comparisons of LDC-C reduction were performed separately on FH and statin-intolerant patients. The changes of each lipid items stratified by forms and dosages of PCSK9-mAbs were carried out versus placebo or ezetimibe treatment. We assessed the publication bias, including the risk of selection bias, performance bias, detection bias, attrition bias, reporting bias and other bias, by using the Risk-of-bias graph and summary table.

Statistical analysis

All analyses were conducted with Review Manager version 5.3. A fixed-effect model was selected if there was no unexplained statistical heterogeneity, otherwise, a random-effect model was used in the meta-analysis. Cochrane Q test to measure the heterogeneity among the included trials and I2 statistics to assess the magnitude of heterogeneity were performed separately. χ2 statistics was to assess the magnitude of heterogeneity and P value < 0.05 was considered to be statistically significant.