The effects of lipid-lowering therapy on coronary plaque regression: a systematic review and meta-analysis

To assess the influence of lipid-lowering therapy on coronary plaque volume, and to identify the LDL and HDL targets for plaque regression to provide a comprehensive overview. The databases searched (from inception to 15 July 2020) to identify prospective studies investigating the impact of lipid-lowering therapy on coronary plaque volume and including quantitative measurement of plaque volume by intravascular ultrasound after treatment. Thirty-one studies that included 4997 patients were selected in the final analysis. Patients had significantly lower TAV (SMD: 0.123 mm3; 95% CI 0.059, 0.187; P = 0.000) and PAV (SMD: 0.123%; 95% CI 0.035, 0.212; P = 0.006) at follow-up. According to the subgroup analyses, TAV was significantly reduced in the LDL < 80 mg/dL and HDL > 45 mg/dL group (SMD: 0.163 mm3; 95% CI 0.092, 0.234; P = 0.000), and PAV was significantly reduced in the LDL < 90 mg/dL and HDL > 45 mg/dL group (SMD: 0.186%; 95% CI 0.081, 0.291; P = 0.001).Thirty-one studies that included 4997 patients were selected in the final analysis. Patients had significantly lower TAV (SMD: 0.123 mm3; 95% CI 0.059, 0.187; P = 0.000) and PAV (SMD: 0.123%; 95% CI 0.035, 0.212; P = 0.006) at follow-up. According to the subgroup analyses, TAV was significantly reduced in the LDL < 80 mg/dL and HDL > 45 mg/dL group (SMD: 0.163 mm3; 95% CI 0.092, 0.234; P = 0.000), and PAV was significantly reduced in the LDL < 90 mg/dL and HDL > 45 mg/dL group (SMD: 0.186%; 95% CI 0.081, 0.291; P = 0.001). Our meta-analysis suggests that not only should LDL be reduced to a target level of < 80 mg/dL, but HDL should be increased to a target level of > 45 mg/dL to regress coronary plaques. Trial Registration PROSPERO identifier: CRD42019146170.

Our meta-analysis suggests that not only should LDL be reduced to a target level of < 80 mg/dL, but HDL should be increased to a target level of > 45 mg/dL to regress coronary plaques. Trial Registration PROSPERO identifier: CRD42019146170.
A previous study suggested that the prevalence of coronary heart disease (CHD) resulting in significant myocardial infarction (MI) morbidity and CHD mortality in American adults who are 20 years of age or older was 6.7% 1 . The severity of coronary atherosclerosis in patients with CHD is closely related to adverse cardiovascular events. Therefore, stabilization and regression of coronary atherosclerotic plaques by lipid-lowering therapy plays an important role in the treatment of CHD 2 .
Plaque regression, which includes the removal of lipids and the necrotic core, was shown to restore endothelial function, although the cessation of intravascular smooth muscle cell proliferation is a complex process 3 . Coronary atherosclerotic plaque regression can be detected using various imaging techniques that can measure changes in plaque volume, and intravascular ultrasound (IVUS) is currently one of the most common of such methods 4 . Total atheroma volume (TAV) and percent atheroma volume (PAV) are the indices usually used to evaluate coronary plaque volume. TAV is more sensitive and PAV is more accurate 5 . A plaque has regressed when a reduced plaque volume is detected after treatment. Recent studies have indicated that lipid-lowering therapy can lead to the regression of a coronary atherosclerotic plaque and reduce the incidence of adverse cardiovascular events 6 . A recent meta-regression analysis by Bhindi et al. 7 showed that a 1% reduction in mean PAV was induced by dyslipidemia therapies and was associated with a 20% reduction in the odds of major adverse cardiac events (MACE).
Statins are the cornerstone of lipid-lowering therapy, but other lipid-lowering drugs include bile acid sequestrants, ezetimibe, and PCSK9 inhibitors 8 . These drugs can reduce blood lipid levels through different mechanisms, including lowering total cholesterol (TC), triglycerides (TG), and low-density lipoproteins (LDL) and increasing www.nature.com/scientificreports/ high-density lipoproteins (HDL) to some extent. Although there have been a number of meta-analyses concerning lipid-lowering therapy and coronary plaque volume in recent years, most studies have been conservative in the drug interventions selected for inclusion in their studies. For example, they only analyzed TAV but not PAV or they only considered the effect of LDL on plaque regression but not HDL. Therefore, we performed a meta-analysis to assess the influence of lipid-lowering therapy on coronary plaque volume (TAV and PAV) in this study, and to identify the LDL and HDL targets for plaque regression to provide a comprehensive overview.

Results
Selection of sources of evidence. Our search strategy yielded a total of 10,985 studies. There were 4989 studies after the repeated studies were excluded and 885 studies did not meet the inclusion criteria. We then excluded 57 studies because of insufficient data, 32 studies because they were animal studies, and 6 studies because they were duplicate reports of the same study population. Therefore, 31 studies (with 4997 patients in the lipid-lowering therapy group and 769 patients in the control group) that measured TAV or PAV at baseline and follow-up were included in our final analysis   (Fig. S1).
Critical appraisal within sources of evidence. The quality of randomized controlled trials was assessed by the Jadad quality scale, and the quality of non-randomized controlled trials was evaluated by the Newcastle Ottawa scale (NOS). The details were shown in Table S1 and S2.
Publication bias can influence the results of a meta-analysis. Therefore, a funnel plot and Egger's and Begg's tests were used to evaluate the potential publication bias in the included studies. The assessment of the symmetry of the funnel plots for the TAV or PAV showed little publication bias in our results (Fig. S2). Egger's test (TAV: P = 0.315; PAV: P = 0.272) and Begg's test (TAV: P = 0.398; PAV: P = 0.209) both confirmed this finding.
A sensitivity analysis was conducted by performing additional meta-analyses after deleting individual studies one by one. The results of the sensitivity analysis showed that none of the studies influenced the pooled SMD, which indicated that our meta-analysis was statistically stable (Fig. S3).

Results of individual sources of evidence.
The relevant data of each included study were presented in Table 1.

Discussion
A total of 31 studies with 4997 enrolled patients who received lipid-lowering therapy were included in our metaanalysis. The changes in coronary plaque volume were measured by IVUS, and the results showed significant coronary plaque regression in patients after receiving lipid-lowering therapy. The subgroup analysis indicated that TAV was significantly reduced when LDL at follow-up was less than 80 mg/dL and HDL was greater than or equal to 45 mg/dL, and PAV was significantly decreased when LDL at follow-up was less than 90 mg/dL and HDL was greater than or equal to 45 mg/dL. These findings were also confirmed by sensitivity analysis. Regression analysis showed that LDL levels at follow-up significantly influenced our results.
To better understand the link between lipid-lowering therapy and plaque regression, a meta-analysis was conducted to explore the changes in TAV and PAV in patients after receiving treatment, and the results showed a significant reduction in TAV and PAV at follow-up with some heterogeneity. We performed a subgroup analysis of the different types of drug administration in patients to explore the source of heterogeneity. The heterogeneity in the subgroups decreased significantly, suggesting that different drug regimens may be potential sources of heterogeneity.
In recent studies, LDL has been shown to accumulate abnormally in the vascular wall due to the dysfunction of endothelial cells. Moreover, LDL can be converted into ox-LDL, which can damage endothelial cells and smooth muscle cells, thereby causing abnormal activation of the endothelial cells, producing foam cells and eventually promoting plaque progression 40 . According to the latest guideline for the management of blood cholesterol and dyslipidemias, experts recommended that patients with a very high risk of arteriosclerotic cardiovascular disease (ASCVD) reduce LDL levels to below 70 mg/dL, which can delay the progress of risk factors and reduce the incidence of adverse events 8 . Our subgroup analysis showed that TAV was significantly reduced when the LDL levels were less than 80 mg/dL at follow-up, and PAV showed a significant decrease when the LDL levels were less than 90 mg/dL at follow-up.
In previous studies, HDL was shown to play an important role in the regression of coronary plaque by reverse cholesterol transport (RCT) 3 . HDL is mainly synthesized by apoAI and apoAII, which can clear or reuse cholesterol through lipid metabolism pathways, thereby reducing the progressive accumulation of cholesterol in plaque and promoting the regression of plaque 41 . A rise in HDL levels can reduce the incidence of cardiovascular adverse events. In the latest guideline for the management of dyslipidemias, HDL is the class I recommendation for lipid analyses in cardiovascular disease risk estimation 42 . In recent research, a rise in HDL level was shown to promote regression of coronary plaque and reduce the occurrence of MACE when LDL was greater than or equal to 70 mg/dL in patients receiving statin therapy 43 . Our subgroup analysis demonstrated a significant reduction in both TAV and PAV when HDL levels were greater than or equal to 45 mg/dL after lipid-lowering therapy. www.nature.com/scientificreports/ Plaque regression is affected by various factors. In a study by Nicholls et al., a rise in HDL and reduction in TG slowed the progression of coronary atherosclerotic plaque 44 . A previous study also suggested that diabetes and hypertension can damage vascular endothelial function and promote the progression of coronary atherosclerotic plaque 45 . Changes in plaque volume at follow-up can also be affected by factors such as drug dose, method of observation, and location of the plaque. Therefore, we conducted a meta-regression analysis to assess other factors that could influence outcomes. The results indicated that LDL at follow-up affected both TAV and PAV, and gender only affected TAV, while dose of drugs and TG levels at baseline only affected PAV. Other risk factors did not influence the results.
Our analysis suggests that patients with CHD require an LDL level below 80 mg/dL and HDL above 45 mg/dL at follow-up for regression of coronary plaques to occur. TAV and PAV exhibited different target levels of LDL for plaque regression in our analysis, which may be due to differences in the number of studies included. The results have considerable significance for current CHD patient management and further research on coronary plaque regression. A recent study suggested that the regression of coronary atherosclerotic plaque in patients with stable CHD is closely related to myocardial infarction and vascular revascularization, but not significantly associated with MACE 46 . However, in the study by Hirohata et al., patients with atheroma progression displayed more adverse events than patients with no progression 47 . Therefore, combined with our research, these results suggest that for patients with CHD controlling LDL at follow-up below 80 mg/dL and HDL above 45 mg/dL can have a positive effect, improving patient prognosis. At the same time, the regression analysis also suggests the important role of LDL in plaque regression, which can provide new ideas for research on plaque regression in the future.

Limitations
This study also had several potential limitations. Most importantly, some of the studies included in the metaanalysis had a small sample size. Furthermore, some subgroup analysis included limited studies; therefore, more studies are needed to support the results. Finally, it is important to assess heterogeneity among studies, and although it may not be possible to identify all possible sources of heterogeneity, the stability of our outcomes was confirmed after adjusting for potential publication bias.

Conclusions
In general, recent meta-analyses have only considered the effect of LDL on plaque regression, whereas our metaanalysis indicates not only that LDL should be reduced to a target level of < 80 mg/dL, but also that HDL should be increased to a target level of > 45 mg/dL to regress coronary plaque.

Methods
This meta-analysis strictly abided by the PRISMA guidelines 48 .
Protocol and registration. The review protocol was developed according to PRISMA guidelines , and was registered in PROSPERO. The registration number was CRD42019146170.
Eligibility criteria. Studies were included if they met the following inclusion criteria: (1) the study design was a prospective clinical cohort study; (2) the impact of lipid-lowering therapy on coronary plaque volume was investigated, including quantitative measurement of plaque volume by IVUS; (3) sufficient information on blood lipids and IVUS findings at baseline and at the end of the study were presented; (4) lipid-lowering therapy was administered for at least 6 months; and (5) primary or secondary outcomes included change in total atheroma volume or percent atheroma volume.
Studies were excluded if they were: (1) non-clinical studies, observational studies, or retrospective studies; (2) duplicate reports or secondary or post hoc analyses of the same study population; or they contained (3) insufficient information on plaque volume and blood lipids (mean, SD, and sample sizes).

Information sources.
The review searched studies based on PICOS (populations: CHD patients; interventions: lipid-lowering therapy; comparisons: before lipid-lowering therapy; outcomes: change in plaque volume as the first or secondary outcome; study design: prospective clinical cohort study) strategy in online databases (PubMed, EMBASE, Cochrane Library, and Web of Science) up to 15 July 2020 were systematically searched.
Search. The  Selection of sources of evidence. Two reviewers (Yingrui Li and Songbai Deng) extracted data from included studies independently. When there was a disagreement on studies, the two reviewers reached a consensus through negotiation. The data extracted from each study included the sample size, LDL, HDL, TAV, and PAV at baseline and at the end of the study. Data items. Data about article (title, authors, year, area and study design), participant characteristics (sample size, age, BMI, gender, PAV, TAV, HDL, LDL, drug administration, smoking, diabetes and hypertension), and information on the assessment used (included population, methods for measuring participation and measurement of exposure factors) were extracted from included studies.
Critical appraisal of individual sources of evidence. The Jadad quality scale was used to assess the quality of randomized controlled trials, and the NOS was used to assess the quality of non-randomized controlled trials. The results ranged from 0 to 5 and 0 to 9, respectively, with higher scores representing better methodology quality ( Table 1).

Synthesis of results.
To calculate the 95% CI of the pooled standard mean difference (SMD) or weighted mean difference (WMD), we used a fixed effects model or a random effects model to perform all statistical analyses using Stata 12.0. P values < 0.05 were considered statistically significant and all P values were two-sided. The χ 2 and I 2 statistics were used to evaluate the heterogeneity between studies. If P was < 0.1 and I 2 was > 30%, a random effects model was used; otherwise, a fixed effects model was used. Considering that one of the purposes of this study was to identify the LDL and HDL targets for plaque regression and determine the potential impact of confounding factors on the results of the study, we performed subgroup analyses of drug administration regimens and LDL and HDL levels at follow-up. Meanwhile, to examine the influence of individual studies on the total merged effects, we used a sensitivity analysis to evaluate the stability of the results. We applied Begg's and Egger's tests to assess publication bias in the included studies, and we assessed possible small sample effects by analyzing the symmetry of a funnel plot. P values < 0.10 were considered statistically significant 49,50 . Taking into account the differences between the studies, all of our analyses used a random effects model. To explore the link between the dependent variable and the covariate, meta-regression is often used. We hypothesized that the included studies may have shown differences according to the age, gender, region, drugs and drug dosages, smoking, diabetes, hypertension, and blood lipids (LDL, HDL, TC, TG) of the patients. To evaluate the possible impact of these factors on the results of the meta-analysis, we established a regression model with the TAV or PAV value as the dependent variable (y) and the abovementioned covariate as the independent variable (x).
Patient and public involvement. Patients were not involved in the design or conduct of the study.

Data availability
No datasets were generated or analysed during the current study.