Atherosclerotic coronary plaque regression from lipid-lowering therapies: A meta-analysis and meta-regression

Abstract

Background

Studies reporting collective and comprehensive data on plaque regression of different lipid-lowering therapies (LLTs) are limited.

Objectives

We evaluated plaque regression of LLTs based on multiple markers and performed subgroup analyses based on LLT type and post-treatment LDL-C levels

Methods

A literature search was performed to identify studies assessing plaque regression from LLTs. The following LLTs groups were included: High-intensity statin (HIS), HIS+ eicosapentaenoic acid (EPA), HIS + ezetimibe, Low-intensity statin (LIS), LIS + EPA, LIS + Ezetimibe, and PCSK9 inhibitors. Our primary outcomes were change in percent atheroma volume (PAV). Secondary outcomes included mean differences in total atheroma volume (TAV), lumen, plaque, and vessel volumes, fibrous cap thickness (FCT), and lipid arc (LA). Subgroup analyses were performed on LLT type and post-treatment LDL-C levels. Meta-regression was performed to control for covariates.

Results

We identified 51 studies with 9,113 adults (22 % females). LLTs reduced PAV levels (-1.10 % [-1.63, -0.56], p < 0.01), with significant reduction observed with HIS, LIS + ezetimibe, LIS + EPA, and PCSK9 inhibitors. LLTs reduced TAV levels (-5.84 mm3 [-8.64 to -3.04] p < 0.01), mainly driven by HIS (-7.60 mm3 [-11.89, -3.31] p < 0.01). LLTs reduced plaque volume and LA and increased FCT.

Conclusion

The plaque regression associated with LLTs is observed to be mainly driven by HIS, reducing both TAV and PAV. This suggest that HIS is the most effective LLT for plaque regression

Unstructured abstract

We evaluated plaque regression of LLTs from 51 studies. We found that while reduction of PAV (-1.10 % [-1.63, -0.56], p < 0.01) were present across different LLT types, reduction of TAV (-5.84 mm3 [-8.64 to -3.04] p < 0.01) was mainly driven by HIS (-7.60 mm3 [-11.89, -3.31] p < 0.01). These results suggest that HIS is the most effective LLT for plaque regression.

Central illustration

Abbreviations

LLT

lipid lowering therapy

PAV

percent atheroma volume

TAV

total atheroma volume

PV

plaque volume

LV

lumen volume

VV

vessel volume

EEM

external elastic membrane

FCT

Fibrous cap thickness

LDL-C

Low-density lipoprotein cholesterol

HDL-C

High-density lipoprotein cholesterol

HIS

High-intensity statins

LIS

Low-intensity statins

CVD

cardiovascular disease

ASCVD

Atherosclerotic cardiovascular disease

CV

Cardiovascular

1. Introduction

Cardiovascular disease (CVD) remains the leading cause of death worldwide and has contributed to loss of health and excess health system costs [1], [2], [3]. Atherosclerotic CVD (ASCVD) is mainly due to plaque that starts with endothelial dysfunction secondary to sustained exposure to unfavorable milieu such as high blood glucose in diabetes, high wall stress in uncontrolled hypertension, and the direct toxic and vasoconstrictive effects of tobacco smoking. If untreated, this can lead to plaque progression, a necessary but modifiable step between subclinical atherosclerosis and plaque rupture leading to cardiovascular (CV) events [4]. Plaque progression is traditionally defined by coronary angiography as a decrease in luminal diameter. Plaque regression, on the other hand, may occur due to lipid-lowering therapy (LLT) or non-pharmacologic modalities such as improved diet and exercise. For the past three decades, there have been tremendous advancements in treatments aimed at reducing CV events [3]. This has paralleled advancements in coronary imaging modalities[5], [6], [7] that can assess plaque volumes and composition [3]. These improvements in coronary imaging allow objective assessment of plaque modification resulting from lipid-lowering interventions [3]. Reduction in total atheroma volume (TAV) representing overall disease activity[8], has been the definition of plaque regression reflecting the initial angiographic definition [3]. However, plaque volume[9], fibrous cap thickness (FCT)[9], lumen volume[10], vessel volume[11], lipid arc[12], and external elastic membrane (EEM)[13], are also markers of plaque regression that reflect a clinically meaningful reduction in CV events. In addition to adherence to a diet low in saturated fats[14,15] and exercise[16,17], lowering low-density lipoprotein cholesterol (LDL-C) with statins among patients with coronary artery disease (CAD) has been proven in numerous trials to considerably reduce coronary plaque progression [18], [19], [20], [21]. Ezetimibe also demonstrated significant plaque regression ranging from −2.9 % to −13.9 % [22]. Proprotein convertase subtilisin kexin type 9 inhibitors (PCSK9i) have also been shown to reduce plaque burden in recent trials [23,24]. Lastly, omega-3 fatty acids, specifically eicosapentaenoic acid (EPA), also significantly reduced coronary plaque volume [25]. Individual effects of specific LLTs are therefore established, but to date, there are relatively few studies that have compared the effect of all classes of LLT on plaque regression based on the level of LDL-C and HDL-C at follow-up. Furthermore, current data on the role of duration of treatment on plaque regression from LLT remains inconclusive. Therefore, to address this gap, we examined the efficacy of all current LLTs in plaque regression based on PAV, TAV, lumen, vessel, plaque volumes, EEM, FCT, and lipid arc. Subgroup analyses were performed to elucidate potential differences in treatment effects based on the 1) type of LLT and 2) level of LDL-C at follow-up. Furthermore, regression analysis was done to ascertain the effects of follow-up LDL-C, HDL-C, and duration of treatment on the outcomes.

2. Methods

This study was reported under the Preferred Reporting Items for a Review and Meta-Analysis (PRISMA)[26], and the checklist was followed (see Supplementary Figure 1 and Supplementary Table 1). Certainty of evidence was rated using the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) framework[27]. This study was registered in the International Prospective Register of Systematic Reviews (PROSPERO)[28], with the identification number CRD42023448987.

2.1. Data sources and searches

The literature search was performed using PubMed/MEDLINE, Ovid/Embase, Web of Science, SCOPUS, and Cochrane databases from database inception until July 2023. The search Medical Subject Headings (MeSH) were “lipid lowering therapy”, “statin”, “high-intensity statin”, “Alirocumab”, “Evolocumab”, “PCSK9 inhibitor(s)”, “ezetimibe”, “eicosapentanoic acid”, “percent atheroma volume”, “plaque volume”, “total atheroma volume”, “fibrous cap thickness”, “lipid arc”, “lumen volume”, “coronary artery disease”, “optical coherence tomography”, “OCT, “intravascular ultrasonography (IVUS)”, “virtual histology,” “randomized controlled trial”, “randomization”, “clinical trials,” and “intervention studies”. We also searched the bibliographies of retrieved articles, meta-analyses, and systematic reviews. Citations of selected articles and relevant studies that evaluated plaque regression from LLTs were reviewed. After removing duplicates, records were reviewed at the title and abstract level, followed by the screening of full text based on our study criteria.

2.2. Study selection

Eligible phase II or phase III, single or double-blind randomized controlled trials (RCTs) comparing treatment with any lipid-lowering therapy (PCSK-9 inhibitors, statin, ezetimibe, EPA/DHA) with placebo or control in adult patients aged 18 years and above were included. Observational cohort studies on the efficacy of LLT on plaque regression were also included. Studies should have reported at least one of the outcomes of interest, which are PAV, TAV, vessel volume, lumen volume, plaque volume, external elastic membrane, fibrous cap thickness, and lipid arc. Studies should have a minimum follow-up of three months [29,30]. Additionally, the mean percent change in LDL-C and HDL-C from the baseline must have been reported. Studies were excluded if (1) they did not report the outcome of interest, (2) data on follow-up was expressed as percent change and not the absolute change, and (3) non-pharmacologic interventions to reduce plaque volume (diet and exercise). Review articles, case reports, letters to the editor, commentaries, proceedings, laboratory studies, and other non-relevant studies were excluded.

2.3. Data extraction

Key participant and intervention characteristics and reported data on efficacy outcomes were extracted independently by two investigators (SWC and VT) using standard data extraction templates. Any disagreements were resolved by discussion or, if required, by a third author (FBR). Data on the following variables were extracted: first author's name, year of publication, journal, study phase, interventional and control treatments, randomization method, analysis tool, number of randomized patients, and demographic and clinical data (e.g., age, sex). In case of uncertainties regarding the study data, we contacted the authors of the specific study for additional information. Quality assessment was performed independently by two review authors using the Revised Cochrane risk-of-bias tool for randomized trials and the Ottawa-Newcastle Scale for observational cohort studies.

2.4. Outcome measures

The primary endpoint of this meta-analysis and meta-regression was the weighted mean difference (MD) in PAV. Secondary endpoints include mean difference in TAV, lumen, plaque, and vessel volumes, FCT and lipid arc. Additionally, subgroup analyses were performed for applicable studies on the (1) type of LLT and (2) level of LDL-C at follow-up (either ≥70 mg/dl or <70 mg/dl).

2.5. Bias assessment

All included RCTs reported a central randomization process, and outcomes were objectively determined. The included studies reported all primary and secondary outcomes as pre-specified in their protocols, so the risk of bias for selective reporting was judged as low. Two authors (SWC and JVM) independently assessed the risk of bias based on the Cochrane Risk of Bias Tool (Supplementary Figures 2) for RCTs that fulfilled the inclusion criteria. MCV and EMC independently assessed the risk of bias on non-RCT articles using the Newcastle-Ottawa Scale (Supplementary Table 2). Disagreements between the two reviewers were resolved by consensus. In persistent disagreement, arbitration by a third reviewer (FBR) was performed.

2.6. Statistical analysis

We pooled all estimates using a random effects model based on DerSimonian and Laird method. Effect sizes were expressed using mean differences with 95 % confidence intervals (CIs). For all outcomes, the significance level was set at a p-value of <0.05 or 95 % CI not including 1. Both Cochran's Q and Higgins and Thompson's I² statistic were generated to describe the heterogeneities among the studies. The extent of heterogeneity was based on the I² statistic, wherein a value of more than 50 % was interpreted as substantial heterogeneity. Prespecified subgroup analyses were performed according to the (1) type of LLT (2) LDL-C levels at follow-up. Meta-regression analysis was performed based on LDL-C and HDL-C levels at follow-up. Stata version 18 (StataCorp, College Station, TX) was used to conduct the included studies' meta-analyses.

3. Results

A literature search was conducted through August 2023 and yielded 2540 potentially relevant references on plaque regression from LLTs (Figure S1). Of these, 94 duplicates were removed. 2345 studies with unrelated interventions, outcomes, populations, non-original data (e.g., meta-analysis or review), descriptive or observational study design, and study protocols were excluded. 94 studies were left, and 18 pooled analyses were removed for not meeting the eligibility criteria. The remaining 76 related studies were retrieved as full-text publications for detailed evaluation. Overall, 51 studies were included in the final meta-analysis. From the 51 studies, 9113 eligible individuals were included for analysis. Females were less represented in all studies, with only 22 % overall. The study characteristics are shown in Table 1.

Table 1.

Characteristics of included studies.

Author (Year)	Country	Patients (n)	Study Design	Population	Age, Yr (Mean ± SD)	Females (%)	Treatment	Comparator	Modality	Primary Outcome	F/U (mo)	LDL-C		HDL-C
												Baseline	F/U	Baseline	F/U
Ahn (2016)[42]	Korea	74	RCT	SAP + PCI	59.6 ± 9.1	14 (36.8 %)	ω−3 PUFA 3 g/d	Placebo (background statin)	IVUS	∆ AV index	12	127.00 ± 29.4	86.8 ± 51.8	45.8 ± 12.0	46.3 ± 12.1
Ako (2019)/ODYSSEY J-IVUS[24]	Japan	206	RCT	ACS	61.8 ± 10.2	19 (20.4 %)	Ali 75–150 mg q2w	Ato ≥10 mg/ d ORRos ≥5 mg/d	IVUS	∆% normalized TAV	9	98.22 ± 23.2	82.8	44.1 ± 10.5	48.7
Alfaddagh (2017) [43]	USA	285	RCT	SAP	62.5 ± 7.8	19 (15.2 %)	ω−3 PUFA	Placebo (background statin)	OCT	∆% indexed volume of noncalcified coronary plaque	30	79.1 ± 27.2	78.5 ± 27.0	47.6 ± 14.3	47.3 ± 14.2
Gao (2022)[60]	China	61	RCT	SAP/ACS	62.9 ± 8.6	15 (29.4 %)	Ali 75 mg q2w + statin	Ato 20 mg/d OR Ros 10 mg/d	CCTA	CAC progression	9	117.56 ± 30.2	51.04 ± 15.1	54.5 ± 23.6	57.2 ± 18.2
Guo (2012)[32]	China	228	RCT	SAP	58.95 ± 9.7	29 (12.7 %)	Ato 10, 20, 40, 80 mg/d	Placebo	IVUSCAG	∆ PV and plaque necrosis	3–6	109.44 ± 25.5	69.99 ± 12.4	34.8 ± 5.0	39.83 ± 8.5
Hattori (2012)[68]	Japan	42	Non-RCT	SAP + PCI	66 ± 7.8	4 (9.5 %)	Pit 4 mg/d	Dietary modification	OCTIVUS	FCT	9	134 ± 40	89 ± 23	46 ± 11	58 ± 16
Hibi (2018)[41]	Japan	128	RCT	ACS	63.2 ± 10.8	21 (20.4 %)	Pit 2 mg/d + Eze 10 mg/d	Pit 2 mg/d	IVUS	∆% PV	10	123 ± 32	64 ± 18	45 ± 14	49 ± 12
Hiro (2009)/JAPAN-ACS[21]	Japan	252	RCT	ACS	62.4 ± 10.6	46 (18.2 %)	Pit 4 mg/d	Ato 20 mg/d	IVUS	∆% PV	8–12	130.9 ± 33.3	81.1 ± 23.4	45 ± 10.1	48.8 ± 12.7
Hong (2008)[33]	Korea	30	RCT	SAP	62 ± 9	12 (40 %)	Ros 20mg	Ato 40 mg	IVUS	∆ PV	12	121 ± 45	65 ± 25	52 ± 7	56 ± 13
Hong (2009)[58]	Korea	100	RCT	SAP	59 ± 9	23 (23 %)	Sim 20 mg/d	Ros 10 mg/d	IVUS	Necrotic core volume and fibrofatty PV	12	119 ± 30	78 ± 20	43 ± 10	48 ± 12
Hong (2011)[34]	Korea	128	RCT	SAP	59 ± 10	33 (25.8 %)	Ros 20mg	Ato 40 mg/d	IVUS	∆ plaque burden	11	122 ± 37	62 ± 20	47 ± 10	47 ± 12
Hougaard (2016)/OCTIVUS[35]	Denmark	87	RCT	ACS	55.3 ± 11.0	12 (13.9 %)	Ato 80 mg/d + Eze 10 mg/d	Ato 80 mg	IVUS	∆ relative necrotic core content	12	143.1 ± 27.1	54.14 ± 30.9	42.5 ± 11.6	42.5 ± 11.6
Kawasaki (2005)[57]	Japan	52	RCT	SAP	66 ± 8.7	13 (25 %)	Pra 20 mg/d + Ato 20 mg/d	Dietary modification	IVUS	Tissue characteristics of arterial plaque	6	155 ± 22	95 ± 15	50 ± 9	56 ± 10
Kini (2013)/YELLOW Trial[49]	USA	87	RCT	SAP + PCI	64.4 ± 9.3	21 (24.2 %)	Ros 40 mg/d	Standard statin	IVUS	Lipid core burden index	1.6	79.1 ± 25.3	58.4 ± 26.3	41.6 ± 10.5	42.5 ± 10.2
Kita (2020)[64]	Japan	130	RCT	ACS	67 ± 8.2	18 (18.6 %)	Ros + EPA 1800 mg/d ORRos 930 mg/d + DHA 750 mg/d	Ros	OCT	∆ FCT	8	118 ± 34.1	82 ± 31.9	46 ± 11.9	44 ± 11.1
Komukai (2014)[66]	Japan	70	RCT	UA	63 ± 11.1	12 (20 %)	Ato 20 mg/d	Ato 5 mg/d	OCT	∆ FCT	12	127 ± 32.6	69 ± 8.2	43 ± 11.1	45 ± 9.6
Kovarnik (2011)/HEAVEN[40]	Europe	80	RCT	SAP	63.5 ± 9.3	9 (21.4 %)	Ato 80 mg/d + Eze 10 mg/d	Standard statin ORAto 10 mg for statin-naïve	IVUSVH- IVUS	∆ VH-IVUS plaque composition	12	119.9 ± 50.3	77.34 ± 30.9	46.4 ± 19.3	46.4 ± 11.6
Lee (2012)/VENUS[37]	Hong Kong	40	RCT	SAP + PCI	63.70 ± 9.8	7 (17.9 %)	Ato 40 mg/d	Ato 10 mg/d	VH-IVUS	∆ plaque lesion characteristics	6	112.4 ± 27.1	52.1 ± 12.6	42.8 ± 17.4	41.5 ± 14.1
Lee(2016)[30]	Korea	70	RCT	ACS	60.9 ± 10.9	16 (22.9 %)	Eze 10 mg + Sim 40 mg	Pra 20 mg	VH-IVUS	Reduction of absolute volume of fibro-fatty plaque	3	111.4 ± 22.0	68.1 ± 15.3	36.0 ± 8.9	37.8 ± 8.7
Masuda (2015)[22]	Japan	51	RCT	SAP	64.0 ± 7.9	16 (31.4 %)	Ros 5 mg/d + Eze 10 mg/d	Ros 5 mg/d	IVUS	∆% PV	6	131.8 ± 25.6	57.3 ± 20.2	53.1 ± 11.8	57.5 ± 15.2
Continued
Matsushita (2016)[44]	Japan	118	RCT	ACS	62.8 ± 10.2	20 (20 %)	Ato 20 mg/d OR Pit 4 mg/d	Pra 10, 30 mg/d OR Flu	IVUS	∆% PV	10	Ato 135 ± 27	Ato 72 ± 22	Ato 43 ± 10	Ato 48 ± 15
												Flu 139 ± 29	Flu 103 ± 29	Flu 48 ± 16	Flu 50 ± 15
												Pit 140 ± 20	Pit 78 ± 13	Pit 50 ± 13	Pit 50 ± 13
												Pra 152 ± 30	Pra 107 ± 23	Pra 51 ± 12	Pra 54 ± 12
Nakajima (2014)/ZEUS[53]	Japan	95	Non-RCT	ACS	63.7 ± 12.6	17 (17.9 %)	Ato 20 mg/d + Eze 10 mg/d	Ato 20 mg/d	IVUS	∆% PV	6	116.2 ± 24.7	56.8 ± 19.5	50.4 ± 13.5	48.6 ± 17.0
										∆ plaque composition	6–8	81 ± 33.3	62.5 ± 17.0	38.5 ± 10	41 ± 9.3
Nakano (2023)/CuVIC[45]	Japan	260	RCT	SAP + PCI	65 ± 10.4	4 (10 %)	Eze 10 mg/day + Statin	Statin monotherapy	IVUS
Nasu (2009)[69]	Japan	80	Non-RCT	SAP	63 ± 10	17 (21.3 %)	Flu 60 mg/d	Lipid-lowering diet	VH-IVUS	∆ plaque components	12	144.9 ± 31.5	98.1 ± 12.7	52.7 ± 12.4	53.9 ± 12.3
Nicholls (2013)[18]	USA	1039	RCT	SAP	57.9 ± 8.5	274 (26.4 %)	Ato 40 mg/d	Ros 20 mg/d	IVUS	PAV	26	119.9 ± 28.9	70.2 ± 1.0	44.7 ± 10.7	48.6 ± 0.5
Nicholls (2016)/ GLAGOV[23]	Australia	968	RCT	SAP	59.8 ± 9.2	269 (27.8 %)	Evo 420 mg	Placebo (background statin)	IVUS	∆ PAV	19.5	92.6 ± 3.6	36.6 ± 3.2	46.7 ± 1.7	51.0 ± 1.7
Nicholls (2022)/ HUYGENS[61]	Australia	161	RCT	ACS	60.9 ± 10.0	20 (25 %)	Evo 420 mg	High-intensity statin	OCTIVUS	∆ FCT	12	140.4 ± 34.0	28.1 ± 25.4	43.8 ± 10.4	51.2 ± 13.2
Niki (2016)[59]	Japan	95	RCT	SAP	68.1 ± 10.1	19 (32.2 %)	Statin + EPA at 1800 mg/d	Statin	IB-IVUS	Tissue characteristics of target coronary plaque	6	97.7 ± 20.6	91.4 ± 20.6	50.5 ± 13.9	49.4 ± 14.7
Nishiguchi (2017)/ESCORT[65]	Japan	70	RCT	ACS	66 ± 8.9	11 (20.8 %)	Early statin group(Pit 4 mg/day from baseline)	Late statin group(Pit 4 mg/d from 3 wk after baseline)	OCT	∆ FCT	0.6, 9	118 ± 19.3	76 ± 25.9	43 ± 8.9	45 ± 11.8
Nishio (2014)[63]	Japan	30	RCT	SAP	61.0 ± 12.6	2 (13.3 %)	EPA 1800 mg/d) + Ros	Ros	OCT	Morphological changes of vulnerable plaques	9	138.0 ± 35.3	80.1 ± 29.7	40.9 ± 12.0	44.9 ± 9.9
Nissen (2004)/REVERSAL[38]	USA	654	RCT	PCI-requiring	55.8 ± 9.8	292 (44.6 %)	Ato 80 mg	Pra 40 mg	IVUS	∆ PAV	18	150.2 ± 27.9	78.9 ± 30.2	42.3 ± 9.9	43.1 ± 11.3
Nissen (2006)/ASTEROID[19]	USACanadaEuropeAustralia	507	RCT	SAP	58.5 ± 10.0	147 (28.9 %)	Ros 40 mg/d	–	IVUS	∆ PAV∆ nominal AV	24	130.4 ± 34.3	60.8 ± 20.0	43.1 ± 11.1	49.0 ± 12.6
Nozue (2012)[46]	Japan	164	RCT	SAP	66 ± 9	65 (39.6 %)	Pit 4 mg/d	Pra 20 mg/d	VH-IVUS	∆ plaque components	8	126 ± 28	74 ± 22	46 ± 11	51 ± 13
Oemrawsingh (2016)/IBIS-3[36]	Netherlands	241	Non-RCT	SAP + CAG/PCI	60.4 ± 7.85	26 (15.9 %)	Ros 40 mg/d	–	RF-IVUS	∆ necrotic core volume	6, 12	96.29 ± 32.9	66.9 ± 27.4	42.9 ± 11.9	47.56 ± 14.3
Okazaki (2004)/ESTABLISH[56]	Japan	70	RCT	ACS	61.3 ± 10.1	10 (14.3 %)	Ato 20 mg/d	Lipid lowering diet	IVUS	∆% PV	6	124.6 ± 34.5	70.0 ± 25.0	45.5 ± 9.9	44.3 ± 11.2
Park (2016)91	Korea	312	RCT	SAP + CAG/PCI	62.3 ± 9.2	61 (27 %)	Ros 40 mg/d	Ros 10 mg/d	VH-IVUS	∆ VH-defined percent plaque compositionwithin target segment	12	105.3 ± 32.8	59.1 ± 22.2	44.9 ± 12.9	49.7 ± 12.3
Räber (2014)/IBIS-4[39]	Europe	103	Non-RCT	ACS	58.5 ± 9.9	6 (7.3 %)	Ros 40 mg/d	–	IVUS	∆% PAV	13	127.2 ± 27.8	73.09 ± 20.6	42.54 ± 8.6	46.4 ± 13.8
Continued
Räber (2022)/ PACMAN-AMI[31]	Europe	300	RCT	ACS	58.5 ± 9.7	56 (18.7 %)	Ali 150 mg q2w	Ros 20 mg	IVUSOCT	∆ PAV	12	154.8 ± 30.9	23.6 ± 23.8	41.3 ± 10.2	48.3 ± 11.2
Sugizaki (2020)/ ALTAIR[62]	Japan	24	RCT	ACS	–	–	Ali 75 mg q2w + Ros 10 mg/d	Ros 10 mg/d	OCT	∆ FCT	9	41.8 ± 36.7	27 ± 23.7	–	–
Sugizaki (2020)[67]	Japan	42	RCT	SAP + PCI (?)	70.8 ± 7.7	9 (21.4 %)	Ros 10 mg/d + EPA 1800 mg/d	Ros 2.5 mg	OCT	∆ native coronary plaque composition	12	90 ± 22.9	68 ± 14.8	47 ± 8.9	45 ± 10.4
Takayama (2009)/COSMOS[20]	Japan	126	Non-RCT	CAG/PCI	62.6 ± 7.7	50 (39.8 %)	Ros 2.5 mg/d	–	IVUS	∆% PV	6.3	140.2 ± 31.5	82.9 ± 18.7	47.1 ± 10.8	55.2 ± 11.7
Takayama (2016)/ALTAIR[52]	Japan	37	RCT	SAP	65.1 ± 10.1	5 (23 %)	Ros 20 mg	Ros 2.5 mg	IVUS	∆ TAV,% TAV	11	130.3 ± 25.5	61.7 ± 16.5	45.3 ± 9.7	47.7 ± 9.3
Tani (2005)[55]	Japan	75	RCT	SAP	63 ± 10	13 (25 %)	Pra 10–20 mg	Placebo	IVUS	∆% PV	6	130 ± 38	104 ± 20	48 ± 11	53 ± 13
Thondapu, et al. (2019)[50]	USA Japan Korea	80	RCT	CAG/PCI	57.5	10 (42 %)	Ros 10 mg	Ato 20 mg	IVUSOCT	∆ TAV, FCT, lipid arc	12	100 ± 21	76 ± 34	51 ± 15	52 ± 13
Tsujita (2015)/ PRECISE-IVUS[47]	Japan	246	RCT	SAP	66 ± 10	44 (21.8 %)	Ato + Eze 10 mg/d	Ato	IVUS	∆ PAV	9–12	109.8 ± 25.4	63.2 ± 16.3	41.1 ± 9.5	45.6 ± 11.9
Ueda (2017)/ ZIPANGU[77]	Japan	131	RCT	CAG/PCI	66 ± 10	22 (22 %)	Ato 10–20 mg + Eze 10 mg/d	Ato 10– 20 mg	IVUS	∆ PAV, TAV	9	101 ± 27	61 ± 17	47 ± 19	44 ± 12
Watanabe (2017)[25]	Japan	193	RCT	SAP/ACS	67 ± 10	19 (19.5 %)	Pit 4 mg/d + EPA 1800 mg/d	Pit 4 mg/d	IB-IVUS	∆ coronary plaque tissue characteristics	8	107.1 ± 34.3	76.9 ± 26.2	49.8 ± 13.0	50.3 ± 12.1
Wang (2015)[48]	China	106	RCT	SAP + PCI	63 ± 10	27 (27.6 %)	Eze + Ros 10 mg/d	Ros 10 mg/d	IVUS	New or recurrence STEMI, UA, cardiac death, stroke	12	139.9 ± 45.6	52.9 ± 32.1	43.7 ± 9.3	48.7 ± 15.9
Yano (2019)[29]	Japan	64	Non-RCT	ACS	64.6 ± 5.3	50 (22.2 %)	Evo 140 mg q2w + Ros 5 mg/d	Ros 5 mg/d	OCT	FCT	3	120.1 ± 20.2	31.1 ± 11.0	46.3 ± 14.2	49.6 ± 14.5
Yokoyama (2005)[54]	Japan	50	RCT	SAP	62.1 ± 10.2	58 (39.1 %)	Ato 10 mg/d	Dietary modification	IVUS	∆ PV	6	133 ± 13	87 ± 29	44 ± 11	49 ± 15
Zhang (2012)[51]	China	100	RCT	SAP	64.5 ± 13.8	19 (38 %)	Ato 80 mg/d	Ato 20 mg/d	IVUS	∆ PV	9	105.6 ± 22.7	62.4 ± 15.9	51.5 ± 9.7	58.5 ± 8.9

Abbreviation: ACS, acute coronary syndrome; Ali, alirocumab; Ato, atorvastatin; AV, atheroma volume; CAC, coronary artery calcification; CAD, coronary artery disease; CAG, coronary arteriography; CCTA, coronary computed tomography angiography; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; Evo, evolucumab; Eze, ezetimibe; FCT, fibrous cap thickness; FFR, fractional flow reserve; Flu, fluvastatin HDL-C, high density lipoprotein cholesterol; IB-IVUS, integrated backscattered intravascular ultrasound; IVUS, intravascular ultrasound; LDL-C, low density lipoprotein cholesterol; NIRS, near-infrared spectroscopy; OCS, observational cohort study; OCT, optical coherence tomography; PAV, percent atheroma volume; PCI, percutaneous coronary intervention; Pit, pitavastatin; Pra, pravastatin; PUFA, polyunsaturated fatty acids; PV, plaque volume; RCT, randomized controlled trials; RF- IVUS, Radiofrequency Intravascular Ultrasound; Ros, rosuvastatin; SAP, stable angina pectoris; Sim, simvastatin; STEMI, ST elevation myocardial infarction; TAV, total atheroma volume; TG, triglycerides; UA, unstable angina; VH-IVUS, Virtual Histology Intravascular Ultrasound.

3.1. Primary outcome

3.1.1. PAV regression

Overall, LLTs [18,19,[21], [22], [23],[31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48]] reduced PAV by 1.10 % (MD: −1.10 %, 95 % CI −1.63, −0.56; p < 0.001) with moderate heterogeneity (I² =39.7 %). Subgroup analysis by LLT type showed significant reductions in PAV in the HIS subgroup (MD: −1.01 %, 95 % CI −1.48, −0.54; p < 0.01; I² = 0.0 %), LIS plus ezetimibe subgroup (MD: −4.07 %, 95 % CI −7.59, −0.55; p < 0.05; I² = 61.8 %), LIS plus EPA subgroup (MD: −3.52 %, 95 % CI −5.9, −1.08, p < 0.01; I² = 6.6 %) and PCSK9i subgroup (MD: −1.39 %, 95 % CI −2.68, −0.09, p < 0.05; I² = 50.8 %), while no significant PAV reductions were observed for all other LLT subgroups (Fig. 1).

Fig. 1.

Random effects meta-analysis on the effects of LLTs on PAV based on the type of LLT. Abbreviation: LLT=lipid lowering therapy; PAV=percent atheroma volume; EPA= eicosapentaenoic acid; PCSK9i= Proprotein convertase subtilisin kexin type 9 inhibitors.

When comparing PAV reduction by LDL-C levels on follow up (<70 mg/dl vs ≥ 70 mg/dl), both subgroups showed significant PAV reductions, without any significant difference between the groups (difference: 0.57 %; p = 0.37) (Fig. 2).

Fig. 2.

Random effects meta-analysis on the effects of LLTs on PAV based on post-treatment LDL-C. Abbreviation: LLT=lipid lowering therapy; PAV=percent atheroma volume; LDL-C= low density lipoprotein cholesterol.

3.2. Secondary outcomes

3.2.1. TAV regression

Overall, LLTs [18,19,23,25,30,31,[33], [34], [35], [36], [37], [38], [39],47,49,50] significantly reduced TAV levels by 5.84 mm³ on average (MD: −5.84 mm³, 95 % CI −8.64, −3.04; p < 0.01) with no significant heterogeneity (I² = 0.0 %). Subgroup analysis by LLT type revealed that the overall reduction was mainly driven by the HIS subgroup (MD: −7.60 mm³, 95 % CI −11.89, −3.31; p < 0.01; I² = 0.0 %) while no significant TAV reductions were observed for all the other LLT subgroups (Fig. 3).

Fig. 3.

Random effects meta-analysis on the effects of LLTs on TAV based on the type of LLT. Abbreviation: LLT=lipid lowering therapy; TAV=Total atheroma volume; EPA= eicosapentaenoic acid; PCSK9i= Proprotein convertase subtilisin kexin type 9 inhibitors.

When comparing TAV reduction by LDL-C levels on follow up (<70 mg/dl vs ≥ 70 mg/dl), both subgroups showed significant TAV reduction, but the lower LDL-C (<70 mg/dl) subgroup was observed to have a slightly higher TAV reduction, but this was not statistically significant. (difference: −3.27mm³; p interaction = 0.26) (Fig. 4).

Fig. 4.

Random effects meta-analysis on the effects of LLTs on TAV based on post-treatment LDL-C. Abbreviation: LLT=lipid lowering therapy; TAV=total atheroma volume; LDL-C= low density lipoprotein cholesterol.

LLTs [22,25,30,32,41,44,[51], [52], [53], [54], [55], [56], [57], [58], [59]] significantly reduced PV (mean: −3.70 mm³, 95 % CI, −6.62 to −0.78; p < 0.05) (Fig. 5), Lipid arc [29,50,[60], [61], [62], [63], [64], [65], [66]] (mean: −12.39˚, 95 % CI, −21.69 to −3.09; p < 0.05), (Fig. 6) and significantly increased the fibrous cap thickness [29,31,50,[60], [61], [62], [63], [64], [65], [66], [67], [68]] (mean: 26.20 μm, 95 % CI, 13.22 to 39.19; p < 0.01) (Fig. 7). There was no significant difference in VV [21,33,35,40,47,[51], [52], [53], [54], [55], [56], [57],59] (mean: −4.31mm³, 95 % CI, −11.83 to 3.20; p = 0.26), (Figure S3) LV[21,22,25,30,[33], [34], [35],40,41,44,47,[51], [52], [53], [54], [55], [56], [57], [58], [59],69] (mean: −1.37mm³, 95 % CI, −6.38 to 3.64, p = 0.59) (Figure S4) and EEM [22,25,30,34,41,44,58] (mean: −4.37 mm, 95 % CI, −13.35 to 4.60; p = 0.34) (Figure S5).

Fig. 5.

Random effects meta-analysis on the effects of LLTs on plaque volume Abbreviation: LLT=lipid lowering therapy.

Fig. 6.

Random effects meta-analysis on the effects of LLTs on lipid arc Abbreviation: LLT=lipid lowering therapy.

Fig. 7.

Random effects meta-analysis on the effects of LLTs on fibrous cap thickness Abbreviation: LLT=lipid lowering therapy.

3.3. Meta-regression analysis

Meta-regression showed that TAV was influenced by LDL levels at follow-up (tau2=0, R squared=0, H2=1, p = 0.046, Fig. 8), while PAV was influenced by the duration of treatment (tau2=1.62, H2=1.88, R square=0.00, p = 0.054, Fig. 9). HDL-C levels were not associated with any outcome.

Fig. 8.

Bubble plot demonstrating the mean difference in TAV (y-axis) in comparison to the achieved LDL-C in mg/dl (x-axis) in the treatment arms of LLT trials. Bubble size is proportional to the number of patients included in the trial. Note control arms are not included in the figure.

Fig. 9.

Bubble plot demonstrating the mean difference in PAV (y-axis) in comparison to the duration of treatment (in weeks) (x-axis) in the treatment arms of LLT trials. Bubble size is proportional to the number of patients included in the trial. Note control arms are not included in the figure.

4. Discussion

This meta-analysis and meta-regression demonstrated the following important findings: 1) use of LLT achieved significant regression of both PAV and TAV; 2) plaque regression was commonly seen in HIS subgroups; 3) both LDL-C level of <70 mg/dl and ≥70 mg/dl at follow-up were associated with a greater reduction in coronary PV; 4) LLT significantly reduced PV and lipid arc, 5) use of LLT significantly increased the FCT.

Previously published pooled analyses have established the role of LLTs on plaque regression [70], [71], [72], [73]. However, most pooled analysis to date were rather conservative in the type of LLT included and did not account for the level of lipids at follow-up. Similar to recent studies, we utilized TAV and PAV as measures of plaque regression, as they are optimal variables of reporting plaque burden and are considered as strong predictors of CVD as they consider total plaque and plaque distribution throughout the coronary tree. [74,75]

In our study, LLTs significant reduced PAV consistent with the study of Li et al.[70] which reported a significant mean PAV reduction (0.123 %). However, the overall PAV reduction in our study was much higher (1.10 %), closer to the results of IBIS-4 (Integrated Biomarkers and Imaging Study)[39] which reported a PAV reduction of 0.9 % with statins. Moreover, none of the subgroups in our study showed such small reduction in PAV, with multiple LLT types (HIS, LIS plus Ezetimibe, LIS plus EPA, PCSK9i) contributing to the overall PAV reducing effect. The highest PAV reduction in our study was seen in the LIS plus ezetimibe subgroup (−4.07 %, p < 0.05) which showed substantial heterogeneity (I² = 61.8 %) while the least reduction was observed in the LIS subgroup (−0.31 %, p = 0.51; I² = 17.9 %).

The overall TAV reducing effect of LLT in our study was consistent with the meta-analysis of Li et al. [70] and Liang et al. [72], which showed a reduction in TAV from baseline (SMD: 0.123 mm³; 95 % CI 0.06, 0.19; P < 0.001 and SMD: −3.63; 95 % CI −4.44, −2.83; P < 0.001, respectively). However, both studies did not analyze the individual and pooled TAV reducing effect across multiple types of LLTs as in this study. Importantly, we established that the TAV reducing effect of LLT is primarily driven by HIS. In our pooled analysis, LLTs significantly reduced overall TAV by 5.84 mm³, with subgroup analysis revealing that this reduction was mainly driven by the HIS subgroup, the only subgroup with a significant mean TAV reduction (7.60 mm³).

The role of statins on plaque regression has been well-studied. It is established that statins can reduce TAV by up to 20 % versus a progression of 10 % in control [3]. Multiple studies have reported significant reductions in TAV with statins, similar to our findings [19,20,39]. Moreover, the efficacy of HIS in plaque burden reduction is not surprising, as the correlation of higher statin dose with greater TAV reduction has been repeatedly shown across multiple studies [21,38]. Similar to our study, a meta-analysis by Tang et al. [76] also reported a significant plaque regression with HIS after 6 months, whereas no regression was observed with LIS. However, the novelty in our finding lie in that HIS was much more effective even when compared with multiple other classes of LLTs aside from statins.

Interestingly, regimens within HIS have also been reported to have varying effects. The SATURN (The Study of Coronary Atheroma by Intravascular Ultrasound: the effect of Rosuvastatin vs. atorvastatiN) trial reported a greater reduction in TAV with rosuvastatin 40 mg compared with atorvastatin 80 mg (−4.8 % vs −3.2 %; P < 0.05) [18]. In our analysis, we were unable to examine if any particular HIS was more effective than another.

The plaque regression of HIS in our study mirrors the results of previous studies. [19,21,76] Moreover, while the addition of ezetimibe seemed to significantly increase the PAV reducing effect of LIS while decreasing the PAV reducing effect of HIS, these inconsistent results were driven by the significant heterogeneity in the LIS plus ezetimibe subgroup (τ² = 0.88, I² = 61.8 %), and possibly the limited number of included studies in the HIS plus ezetimibe subgroup. In fact, previous studies demonstrated that the addition of ezetimibe do not contribute to a significant increase in PAV reduction [22,35,40,47,77]. While the role of statins on plaque regression have been discussed above, we emphasize that HIS was the only LLT subgroup which provided significant reductions in both TAV and PAV.

We also established that PCSK9i reduced PAV. Previous trials have shown a lesser benefit of PCSK9i on plaque regression in comparison to the benefits observed in relation to LDL-C and CV outcomes. While the ODYSSEY J-IVUS (Evaluation of Effect of Alirocumab on Coronary Atheroma Volume in Japanese Patients Hospitalized for Acute Coronary Syndrome with Hypercholesterolemia) trial failed to show a statistically significant reduction in PAV at 36 weeks, [24] our findings were similar to the GLAGOV (Global Assessment of Plaque Regression with PCSK9 Antibody as Measured by Intravascular Ultrasound) trial which reported a greater PAV reduction in the evolocumab group versus placebo (difference: −1.0 % P < 0.001)[23].

Perhaps, the more significant finding of our study in PAV reduction is the role of EPA. Our findings were consistent with the CHERRY (Combination therapy of eicosapentaenoic acid and pitavastatin for coronary plaque regression evaluated by integrated backscatter intravascular ultrasonography) [78] trial and the EVAPORATE (Effect of Vascepa on Improving Coronary Atherosclerosis in People With High Triglycerides Taking Statin Therapy)[79] trials. While many previous studies have demonstrated that the addition of EPA to statin contributes to increased plaque regression, our pooled analysis showed that the addition of EPA to LIS is superior even when compared to other multiple classes of LLTs, supporting the hypothesis that EPA contributes to plaque regression in mechanisms beyond lipid-lowering, giving emphasis to the role of its anti-inflammatory effects in plaque regression [80,81]. However, the question on whether the overall results translate to a clinically meaningful reduction still remains.

FCT is the most important determinant of plaque vulnerability and risk of rupture in the coronary arteries [82]. FCT has been shown to increase in response to statin therapy and increased statin dosages, with the increase in FCT associated with plaque stabilization [66,83].. We established that LLTs collectively increased the FCT by an average of 26.20 μm. This finding is consistent with the ESCORT (Effect of Early Pitavastatin Therapy on Coronary Fibrous Cap Thickness Assessed by Optical Coherence Tomography in Patients with Acute Coronary Syndrome) trial, which randomized ACS patients to early or late pitavastatin and showed greater increases in FCT by 20 μm in the early statin treatment group versus a decrease of 5 μm among the late treatment group at 36 weeks [65]. Furthermore, the EASY-FIT (Effect of Atorvastatin Therapy on the Fibrous Cap Thickness in Coronary Atherosclerotic Plaque as Assessed by OCT) trial also demonstrated increased FCT and reduced lipid arc content measured by OCT among patients treated with more intensive statin therapy [66].

The latest AHA/ACC guidelines for managing blood cholesterol and dyslipidemias have recommended that patients at very high risk for arteriosclerotic cardiovascular disease (ASCVD) reduce their LDL levels below 70 mg/dl to reduce adverse events and decrease the progression of risk factors [84]. Another study showed that patients who achieve very low LDL-C levels of <70 mg/dl have a lower risk for major cardiovascular events than those with only a moderate decrease in LDL [85]. In our analysis, both an LDL-C level of <70 mg/dl or ≥70 mg/dl at follow-up wereassociated with a higher reduction in plaque volume. In the previous study by Li et.al where results were stratified by LDL-C level by range, TAV reduction was observed in LDL levels of up to 70–80 mg/dl and PAV reduction was observed even in levels up to 80–90 mg/dl. This suggests that the significant plaque regression in LDL-C levels of ≥70 mg/dl in our study might be due to a slightly higher threshold for plaque regression.

5. Strengths and limitation

We pooled 51 studies with more than 9100 participants and performed prespecified subgroup analysis and meta-regression to assess for factors associated with the treatment effect. However, several limitations should be noted. This is a study-level meta-analysis, and we could not access individual patient data. Since this study uses pooled data from multiple studies across different populations with varied demographics and characteristics, discrepancies may contribute to heterogeneity in some of our analyses. A more important limitation in our pooled analysis is the inclusion of different modalities to assessed plaque regression. Head-to-head comparison of CCTA to IVUS has been well established. Moreover, studies have reported that manual plaque quantification on CTCA was comparable to IVUS [86,87]. Furthermore, a recent meta-analysis established that overall plaque volumes measured using CCTA and IVUS do not differ significantly on a study sample basis [88]. However, studies have shown that changes on OCT are not significantly correlated with changes on IVUS [89,90], hence we cannot be fully certain that the assessment of plaque volume has been comparable across these different imaging modalities.

In addition, different dosages and regimens within one LLT class and further LDL-C stratification might have affected outcomes, which were inevitably unexplored due to the inherent limitation in stratification due to the nature of our data. There remains a paucity of inclusion of females in clinical trials, and as a result, there is only a small portion of women in this analysis, limiting our ability to determine if there are any particular sex differences in LLT. Publication bias may also be present, the extent of which could not fully be quantified. Nonetheless, every effort possible was made to limit bias by utilizing a robust analytical approach to adjust for potential moderators through subgroup analyses and meta-regression.

6. Conclusion

This meta-analysis showed that the significant plaque regression associated with lipid lowering therapies is observed to be mainly driven by high-intensity statins, when comparing different classes of LLTs. LLTs also significantly reduced plaque volume and lipid arc while increasing fibrous cap thickness.

CRediT authorship contribution statement

Frederick Berro Rivera: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Writing – original draft, Writing – review & editing. Sung Whoy Cha: Writing – review & editing, Data curation. Michelle Capahi Varona: Writing – review & editing, Data curation. Elaiza Marie Fernandez Co: Data curation, Writing – original draft, Writing – review & editing. John Vincent Magalong: Data curation, Writing – original draft, Writing – review & editing. John Paul Aparece: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Diana De Oliveira-Gomes: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Gurleen Kaur: Data curation, Formal analysis, Writing – original draft, Writing – review & editing. Martha Gulati: Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix. Supplementary materials

Table S1

Preferred Reporting Items for a Review and Meta-Analysis (PRISMA) checklist

Figure S1

Study selection flowchart. This flow diagram shows the process used to identify relevant records for the meta-analysis. The systematic literature search was performed from database inception to August 2023. Abbreviations: PRISMA=Preferred Reporting Items for Systematic Reviews and Meta-Analyses; RCT=Randomized controlled trial; OCS=observation cohort studies

Figure S2

Cochrane Risk of Bias Tool

Figure S3

Random effects meta-analysis on the effects of LLTs on vessel volume Abbreviation: LLT=lipid lowering therapy

Figure S4

Random effects meta-analysis on the effects of LLTs on lumen volume Abbreviation: LLT=lipid lowering therapy

Figure S5

Random effects meta-analysis on the effects of LLTs on external elastic membrane Abbreviation: LLT=lipid lowering therapy