Smoking and apolipoprotein levels: A meta-analysis of published data

Abstract

Background

Apolipoproteins are major components of lipoproteins such as high-density lipoprotein (HDL) and very-low-density lipoprotein and are considered nontraditional markers in the risk assessment for cardiovascular disease. The goal of this review was to quantify the effects of smoking and smoking cessation on serum levels of apolipoproteins AI, AII, and B and the ratio of apolipoproteins B and AI.

Methods

PubMed and Scopus were searched up to June 2021 to identify publications that reported the levels of apolipoproteins AI, AII, and B and the apolipoprotein B/AI ratio in smokers and nonsmokers as well as articles reporting the effect of smoking cessation on the same endpoints. Meta-analyses were performed when a sufficient number (n ≥ 3) of articles evaluating the same outcome were available.

Results

Forty-nine studies had assessed apolipoprotein levels in smokers and nonsmokers. The meta-analyses comparing the levels of apolipoproteins AI and AII showed decreased levels in smokers relative to nonsmokers. On the other hand, the apolipoprotein B levels and apolipoprotein B/AI ratio were increased in smokers relative to nonsmokers. Insufficient publications were available on which to perform meta-analyses on the effects of smoking cessation on apolipoprotein levels.

Conclusions

Smoking is associated with reduced levels of apolipoproteins AI and AII (in line with reduced levels of HLD-cholesterol) and increased apolipoprotein B levels and apolipoprotein B/AI ratio, thereby confirming the negative impact of smoking on lipid metabolism, which contributes to increased cardiovascular risk. More data are needed to elucidate the effects of smoking cessation on these cardiovascular risk endpoints.

Highlights

• •Apolipoproteins are considered nontraditional markers for cardiovascular disease.
• •Smoking negatively alters the lipid profile in current smokers versus nonsmokers.
• •Apolipoproteins AI and AII levels are higher in nonsmokers than smokers.
• •Apolipoprotein B and Apo B:Apo AI ratio are lower in nonsmokers than smokers.
• •More data are needed to assess the effects of smoking cessation on lipid profile.

1. Introduction

Cigarette smoking has been implicated as a major risk factor in chronic pulmonary disease, carcinogenesis and atherosclerotic arterial disease leading to cardiovascular disease (CVD) [1], [2], [3]. The mechanisms through which smoking increases the risk of atherosclerosis and CVD include oxidative damage and the alteration of lipid levels [4], [5]. For instance, it has been shown that cigarette smoking decreases high-density lipoprotein-cholesterol (HDL-C) [6].

Even though low-density lipoprotein-cholesterol (LDL-C) is recognized as the primary lipid risk factor for CVD [7], other lipid measurements have been associated with an increased or decreased risk of CVD, such as apolipoproteins, which are the protein components of lipoproteins [8]. Two major types of apolipoproteins have been described: apolipoprotein A (I and II) and apolipoprotein B. Apolipoprotein AI, the major constituent of HDL, comprises roughly 70% of the HDL protein mass [9]. It mediates the efflux of cholesterol from the peripheral cell membranes and activates lecithin cholesterol acetyltransferase, which is an important enzyme in the reverse transport of cholesterol from peripheral tissues to the liver [10]. Apolipoprotein AII is the second most abundant protein in HDL, accounting for about 20% of total HDL [11]. Apolipoprotein B is an essential component of LDL and very-low-density lipoprotein (VLDL-C). It facilitates cholesterol delivery to peripheral tissues and promotes LDL-C accumulation by acting as a ligand for LDL-C receptors [1]. Apolipoprotein B is the major protein not only of LDL but also of all atherogenic particles, e.g. VLDL and intermediate-density lipoprotein (IDL). Because only one apolipoprotein B protein is present per particle [12], the total apolipoprotein B value reflects the total number of potentially atherogenic particles [13]. Due to the close association between apolipoproteins and serum lipids, their measurement has been proposed as an important factor in predicting the risk of cardiovascular diseases [8], [10]. Additionally, population-based studies have suggested that the apolipoprotein B:AI ratio could be used as an indicator of the balance between atherogenic and atheroprotective cholesterol transport, as an increase in this ratio predicts cardiovascular risk more accurately and strongly than the use of either apolipoprotein B or apolipoprotein AI individually, or any of the other cholesterol indexes [14], [15], [16].

Studies assessing the association of smoking status and smoking cessation with lipid parameters are abundant in the literature. A review of the association of HDL-C levels and smoking status was performed by Maeda et al. [17], demonstrating increased levels of HDL-C after smoking cessation. To date, no meta-analyses have been published comparing the effects of smoking status or smoking cessation on the levels of apolipoproteins, which is the objective of the present review.

2. Methods

Searches were performed through PubMed and Scopus databases to identify publications that evaluated the relationship between smoking or smoking cessation and plasma levels of apolipoproteins, including AI, AII, and B, and the ratio of apolipoprotein B to apolipoprotein AI in humans. The last search was performed on June 25, 2021. The following query was used in PubMed: ("apolipoproteins"[MeSH Terms] OR "apolipoproteins"[All Fields] OR "apolipoprotein"[All Fields]) AND (("smoking"[MeSH Terms] OR "smoking"[All Fields]) OR ("tobacco"[MeSH Terms] OR "tobacco"[All Fields] OR "tobacco products"[MeSH Terms] OR ("tobacco"[All Fields] AND "products"[All Fields]) OR "tobacco products"[All Fields]) OR cessation [All Fields] OR quitting [All Fields]). In Scopus, the following query was used: apolipoprotein AND (smoking OR tobacco OR cessation OR quitting). The reference lists of the publications obtained through the original search were checked for additional articles.

2.1. Study selection

Case-control, cohort, or interventional studies in healthy adult subjects reporting any measurements of apolipoproteins AI, AII, and B levels and/or apolipoprotein B/AI ratio by smoking exposure (smokers vs. nonsmokers and baseline vs. time after smoking cessation) were included if they provided the following: mean or median values by group, standard deviation (SD) or standard error (SE), sample size per group, or had enough information to allow for the calculation of mean and SD. Review articles, case reports, editorials, and duplicates of publications of data that were part of a more recent publication were not included.

2.2. Data extraction

The following information was extracted from each study independently by three researchers: first author’s name, year of publication, country where study was performed, study design and population characteristics, number of participants per exposure group, mean, SD or SE, and smoking status definition. Apolipoprotein levels were converted from mg/dL to g/L as needed.

2.3. Statistical analysis

All analyses were implemented using the ‘meta’ [18] and ‘dmetar’ [19] packages in R 4.0.5 [20].

To quantify the effects of smoking and smoking cessation on apolipoprotein levels, pooled mean differences between smokers and nonsmokers (when assessing effects of smoking on apolipoproteins) or differences between baseline and follow-up measures after smoking cessation (when assessing the effects of quitting smoking on apolipoprotein levels) and 95% confidence intervals (95% CI) were calculated using the fixed-effects model in the ‘metamean’ function [18]. The ‘metamean’ function uses the inverse variance method for pooling, giving studies with small variance relatively higher weight and studies with larger variance relatively smaller weight [18]. Meta-analyses were carried out using the ‘metacont’ function when at least three measures of a single estimate were available. The degree of heterogeneity between the study results was assessed by the I² statistic. Funnel plot symmetry and Egger’s regression test were used to evaluate publication bias [21]. Statistical significance was assessed at α = 0.05.

To explore possible sources of heterogeneity, the meta-analyses were performed using the random-effects model and sensitivity analyses were conducted using the ‘InfluenceAnalysis’ function, to eliminate studies that contributed the most to heterogeneity. This function conducts an influence analysis using the "Leave-One-Out" paradigm and produces data for four influence diagnostics (Baujat plot, Viechtbauer-Cheung influence characteristics plot, and forest plots for the leave-one-out analysis, sorted by effect size and I²). The Baujat plot and forest plot sorted by I² were used to identify the studies with high heterogeneity contribution and low influence on the overall results. These studies represent outliers and were removed to reduce the amount of between-study heterogeneity. Additionally, subgroup analyses were performed based on smoking status, geographical region, study design, period of publication, and sex.

The reason for using different models in the meta-analyses is that, while the fixed-effects model assumes that the estimated effects from the component studies in a meta-analysis come from a single homogeneous population, the random-effects model seeks to account for the fact that the study effect estimates are often more variable than assumed in the fixed-effects model [22].

3. Results

3.1. Identified studies

A Prisma flow diagram detailing the retrieval process of articles from the different sources used can be found in Fig. 1. For the analyses of smoking status and its association to apolipoprotein levels, a total of 66 publications that assessed the effect of smoking status on any of the following parameters were identified: 39 studies on apolipoprotein AI [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], ten studies on apolipoprotein AII [23], [24], [26], [32], [34], [36], [43], [47], [53], [55], 40 studies on apolipoprotein B [23], [24], [27], [28], [29], [30], [32], [34], [35], [37], [38], [39], [40], [41], [42], [44], [45], [46], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], and 31 studies on the ratio of apolipoprotein B to AI [23], [24], [27], [28], [29], [30], [32], [34], [35], [37], [38], [39], [40], [41], [42], [44], [45], [46], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [62], [70]. These publications compared the levels of the aforementioned lipid parameters in smokers vs. nonsmokers. Out of the 66 publications, 49 studies were included in the analyses and their characteristics can be found in Table 1. Additionally, the forest plots can be found in Appendix 1. The other 17 studies were not included because they involved diseased populations [71], [72], had incomplete information [73], [74], [75], [76], [77], [78], [79], [80], [81], reported inaccurate or inconsistent measurement units [82], [83], the populations studied included children [84], the publication was a meta-analysis [85], or because the nonsmoking group included smokers [86], [87]. Table 2.

Fig. 1.

Prisma flow diagram – article retrieval process.

Table 1.

Study characteristics of selected studies of smoking status and apolipoprotein levels.

Study	Reference	Country	Study design	Study participants	Smoking definition	Estimate number	Measurement	Subgroup			Adjustment
								Smokers Mean ± SD	Nonsmokers Mean ± SD	Mean difference Δ (95% CI)
1	Adachi et al. [23]	Japan	Cross-Sectional	169 healthy men with an average age of 54.7 ± 8.5 years	Stratified by CPD	1.1 1.2 1.3 1.4	Apolipoprotein AI Apolipoprotein AII Apolipoprotein B Ratio Apo B/Apo AI	Men 20 + CPD 1.38 ± 0.27 Men 20 + CPD 0.34 ± 0.08 Men 20 + CPD 0.86 ± 0.18 Men 20 + CPD 0.63 ± 0.03	± 0.240.34 ± 0.07 0.87 ± 0.19 0.63 ± 0.03	-0.01 [− 0.11, 0.09] 0.0 [− 0.03, 0.03] -0.01 [− 0.08, 0.06] 0.0 [− 0.01, 0.01]	None
2	Ashavaid et al. [24]	India	Cross- Sectional	470 men and 281 women aged 25–65 years	None	2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8	Apolipoprotein AI Apolipoprotein AII Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.22 ± 0.20 All Women 1.24 ± 0.22 All Men 0.28 ± 0.05 All Women 0.25 ± 0.03 All Men 0.99 ± 0.19 All Women 0.76 ± 0.09 All Men 0.83 ± 0.18 All Women 0.63 ± 0.10	1.23 ± 0.18 1.32 ± 0.18 0.28 ± 0.04 0.28 ± 0.04 0.98 ± 0.21 0.91 ± 0.21 0.80 ± 0.19 0.69 ± 0.18	-0.01 [− 0.05, 0.03] -0.08 [− 0.30, 0.14] 0.0 [− 0.01, 0.01] -0.03 [− 0.06, 0.0] 0.01 [− 0.03, 0.05] -0.15 [− 0.24, − 0.06] 0.03 [− 0.01, 0.07] -0.06 [− 0.16, 0.04]	None
3	Assmann et al. [25]	Germany	Cohort	3509 men and 1648 women, adults with no information on age.	Stratified by CPD	3.1 3.2	Apolipoprotein AI	Men 20 + CPD 1.34 ± 0.25 Women 20 + CPD 1.46 ± 0.24	1.38 ± 0.22 1.50 ± 0.24	-0.04 [− 0.07, − 0.01] -0.04 [− 0.10, 0.02]	None
4	Berg et al. [26]	Sweden	Cross-Sectional	97 men aged 40–42 years	Stratified by CPD	4.1 4.2	Apolipoprotein AI Apolipoprotein AII	Men 10 + CPD 1.09 ± 0.17 Men 10 + CPD 0.29 ± 0.04	1.24 ± 0.30 0.3 ± 0.04	-0.15 [− 0.24, − 0.06] -0.01 [− 0.03, 0.01]	None
5	Casasnovas et al. [27]	Spain	Cohort	572 young men with an average age of 19.9 ± 0.9 years	None	5.1 5.2 5.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.39 ± 0.23 All Men 0.54 ± 0.11 All Men 0.39 ± 0.01	1.34 ± 0.34 0.53 ± 0.15 0.40 ± 0.02	0.05 [− 0.01, 0.11] 0.01 [− 0.02, 0.04] -0.01 [− 0.01, − 0.01]	None
6	Chen et al. [28]	China	Cohort	84 men aged 40–80 years	10 + CPD for at least 10 years	6.1 6.2 6.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.45 ± 0.23 All Men 0.78 ± 0.28 All Men 0.54 ± 0.04	1.48 ± 0.30 0.73 ± 0.35 0.49 ± 0.07	-0.03 [− 0.19, 0.13] 0.05 [− 0.14, 0.24] 0.05 [0.01, 0.09]	None
7	Chu et al. [29]	Taiwan	Cross- Sectional	781 men aged 18–24 years	Stratified by CPD	7.1 7.2 7.3 7.4 7.5 7.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.45 ± 0.30 Men 40 + CPD 1.31 ± 0.35 All Men 0.97 ± 0.30 Men 40 + CPD 0.93 ± 0.35 All Men 0.67 ± 0.06 Men 40 + CPD 0.71 ± 0.11	1.46 ± 0.27 1.46 ± 0.27 0.95 ± 0.29 0.95 ± 0.29 0.64 ± 0.05 0.65 ± 0.05	-0.01 [− 0.06, 0.03] -0.15 [− 0.27, − 0.03] 0.01 [− 0.04, 0.06] -0.02 [− 0.14, 0.10] 0.03 [0.02–0.04] 0.06 [0.02, 0.10]	Age and body weight
8	Cuesta et al. [30]	Spain	Cross- Sectional	391 men aged 20–65 years	Stratified by CPD	8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	Men 20–29 20 + CPD 1.58 ± 0.40 Men 30–39 20 + CPD 1.60 ± 0.32 Men 40–49 20 + CPD 1.60 ± 0.32 Men 20–29 20 +CPD 1.03 ± 0.31 Men 30–39 20 + CPD 1.10 ± 0.26 Men 40–49 20 + CPD 1.12 ± 0.23 Men 20–29 20 + CPD 0.65 ± 0.07 Men 30–39 20 + CPD 0.69 ± 0.05 Men 40–49 20 + CPD 0.70 ± 0.04	1.64 ± 0.37 1.68 ± 0.39 1.70 ± 0.39 0.88 ± 0.24 0.94 ± 0.21 1.02 ± 0.28 0.54 ± 0.04 0.56 ± 0.03 0.60 ± 0.05	-0.06 [− 0.20, 0.08] -0.08 [− 0.24, 0.08] -0.10 [− 0.29, 0.09] 0.15 [0.05, 0.25] 0.16 [0.05, 0.27] 0.10 [ − 0.04, 0.24] 0.11 [0.09, 0.13] 0.13 [0.11, 0.15] 0.10 [0.08. 0.12]	None
9	Dallal et al. [31]	USA	Cross- Sectional	146 women and 92 men of Chinese origin, aged 60–96 years. (no female smokers)	None	9.1	Apolipoprotein AI	All Men 1.18 ± 0.32	1.18 ± 0.35	0.00 [− 0.22, 0.22]	None
10	Dallongeville et al. [32]	France	Cohort	180 men aged 45–65 years	5 + CPD	10.1 10.2 10.3 10.4	Apolipoprotein AI Apolipoprotein AII Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.50 ± 0.28 All Men 0.39 ± 0.11 All Men 1.30 ± 0.32 All Men 0.87 ± 0.07	1.54 ± 0.31 0.37 ± 0.08 1.24 ± 0.29 0.81 ± 0.06	-0.04 [− 0.13, 0.05] 0.02 [− 0.01, 0.05] 0.06 [− 0.03, 0.15] 0.06 [0.04, 0.08]	None
11	Dedonder-Decoopman et al. [33]	France	Cross-Sectional	206 men and 271 women, no information on age.	Stratified by CPD	11.1 11.2	Apolipoprotein AI	Men 15 + CPD 1.26 ± 0.26 Women 15 + CPD 1.32 ± 0.28	1.39 ± 0.26 1.46 ± 0.24	-0.13 [− 0.22, − 0.04] -0.14 [− 0.31, 0.03]	None
12	de Parscau & Fielding [34]	USA	Cross- Sectional	20 men, 10 smokers and 10 nonsmokers. All healthy aged 20–40 years	20 + CPD	12.1 12.2 12.3 12.4	Apolipoprotein AI Apolipoprotein AII Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.09 ± 0.21 All Men 0.38 ± 0.08 All Men 0.94 ± 0.20 All Men 0.86 ± 0.06	1.29 ± 0.17 0.36 ± 0.06 0.76 ± 0.19 0.59 ± 0.03	-0.20 [− 0.37, − 0.03] 0.02 [− 0.04, 0.08] 0.18 [0.01, 0.35] 0.27 [0.23, 0.31]	Weight
13	Dirican et al. [35]	UK	Cross- Sectional	All men, 27 cigarette smokers aged 41.9 ± 6.7 years and 31 nonsmokers aged 41.5 ± 6.9 years	10 + CPD for at least 7 years	13.1 13.2 13.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.49 ± 0.25 All Men 1.48 ± 0.42 All Men 0.99 ± 0.11	1.55 ± 0.28 1.43 ± 0.40 0.92 ± 0.10	-0.06 [− 0.20, 0.08] 0.05 [− 0.16, 0.26] 0.07 [0.02, 0.12]	None
14	Donahue et al. [36]	USA	Cross- Sectional	172 men and women aged 20–24 years.	None	14.1 14.2 14.3 14.4 14.5 14.6	Apolipoprotein AI Apolipoprotein AII	All Men 1.14 ± 0.23 Women OC 1.35 ± 0.34 Women no OC 1.19 ± 0.18 All Men 0.46 ± 0.13 Women OC 0.49 ± 0.12 Women no OC 0.39 ± 0.10	1.24 ± 0.27 1.44 ± 0.20 1.27 ± 0.27 0.41 ± 0.10 0.41 ± 0.10 0.37 ± 0.09	-0.10 [− 0.21, 0.01] -0.09 [− 0.31, 0.13] -0.08 [− 0.20, 0.04] 0.05 [0.00, 0.10] 0.08 [− 0.01, 0.17] 0.02 [− 0.03, 0.07]	None
15	Dullaart et al. [37]	Netherlands	Cross- Sectional	42 men aged 21–60 years	5 + CPD	15.1 15.2 15.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.44 ± 0.15 All Men 0.75 ± 0.20 All Men 0.52 ± 0.02	1.54 ± 0.15 0.74 ± 0.15 0.48 ± 0.01	-0.10 [− 0.19, − 0.01] 0.01 [− 0.10, 0.12] 0.04 [0.03, 0.05]	None
16	Foulon et al. [38]	France	Cross- Sectional	251 women and 72 men aged 20–29 years	None	16.1 16.2 16.3 16.4 16.5 16.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.19 ± 0.20 All Women 1.30 ± 0.22 All Men 0.92 ± 0.21 All Women 0.74 ± 0.19 All Men 0.77 ± 0.05 All Women 0.57 ± 0.04	1.21 ± 0.18 1.35 ± 0.19 0.77 ± 0.20 0.71 ± 0.15 0.64 ± 0.03 0.53 ± 0.02	-0.02 [− 0.12, 0.08] 0.05 [− 0.13, 0.03] 0.15 [0.05, 0.25] 0.03 [− 0.04, 0.10] 0.13 [0.11, 0.15] 0.04 [0.03, 0.05]	None
17	Frey et al. [39]	USA	Cohort	4196 never smokers and 3176 current smokers aged 35–75 years. Men and women	None	17.1 17.2 17.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All 1.39 ± 0.23 All 1.21 ± 0.28 All 0.89 ± 0.26	1.45 ± 0.25 1.12 ± 0.26 0.80 ± 0.23	-0.06 [− 0.07, − 0.05] 0.09 [0.08, 0.10] 0.09 [0.08,0.10]	Study
18	Frondelius et al. [40]	Sweden	Cohort	Men and women aged 44–74 years. Including 7080 current smokers and 9668 never smokers.	None	18.1 18.2 18.3 18.4 18.5 18.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/ Apo AI	All Men 1.44 ± 0.13 All Women 1.62 ± 0.14 All Men 1.13 ± 0.13 All Women 1.11 ± 0.14 All Men 0.81 ± 0.11 All Women 0.71 ± 0.10	1.46 ± 0.13 1.66 ± 0.13 1.07 ± 0.13 1.06 ± 0.13 0.75 ± 0.11 0.66 ± 0.09	-0.02 [− 0.03, − 0.01] -0.04 [− 0.05, − 0.04] 0.06 [0.05, 0.07] 0.05 [0.05, 0.06] 0.06 [0.05, 0.07] 0.05 [0.05, 0.05]	Age and screening date.
19	Gomo et al. [41]	Zimbabwe	Cross- Sectional	589 men aged 20–65 years	Stratified by CPD	19.1 19.2 19.3 19.4 19.5 19.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.98 ± 0.48 Men 20 + CPD 1.59 ± 0.36 All Men 0.89 ± 0.25 Men 20 + CPD 1.02 ± 0.34 All Men 0.45 ± 0.03 Men 20 + CPD 0.64 ± 0.07	2.01 ± 0.45 2.04 ± 0.54 0.99 ± 0.34 0.99 ± 0.44 0.49 ± 0.04 0.49 ± 0.08	-0.03 [− 0.15, 0.09] -0.45 [− 0.61, − 0.29] -0.10 [− 0.18, − 0.02] 0.03 [0.12, 0.18]	None
20	Haarbo et al. [42]	Denmark	Cohort	148 women aged 45–55 with natural menopause 6 months to 3 years previously	None	20.1 20.2 20.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Women 1.43 ± 0.17 All Women 1.03 ± 0.23 All Women 0.73 ± 0.16	1.53 ± 0.26 0.98 ± 0.25 0.66 ± 0.19	-0.10 [− 0.17, − 0.03] 0.05 [− 0.03, 0.13] 0.07 [0.01, 0.13]	Abdominal fat%
21	Haffner et al. [43]	USA	RCT-Cohort	33 men and 17 women, aged 38–64	None	21.1 21.2 21.3 21.4	Apolipoprotein AI Apolipoprotein AII	All Men 1.17 ± 0.21 All Women 1.29 ± 0.11 All Men 0.29 ± 0.04 All Women 0.31 ± 0.04	1.37 ± 0.21 1.39 ± 0.24 0.34 ± 0.04 0.32 ± 0.05	-0.20 [− 0.35, − 0.05] -0.10 [− 0.28, 0.08] -0.05 [− 0.08, − 0.02] -0.01 [− 0.06, 0.04]	None
22	Haj Mouhammed et al. [70]	Tunisia	Cross- Sectional	300 men and women, 138 nonsmokers aged 38.47 ± 21.91 years and 162 smokers aged 35.55 ± 16.03 years	None	22.1	Ratio Apo B/Apo I	All 0.83 ± 0.52	0.52 ± 0.15	0.31 [0.23, 0.39]	Age, gender, BMI
23	Hostmark et al. [61]	Norway	Cross- Sectional	165 men aged 40–60 years	None	23.1	Apolipoprotein B	All Men 1.71 ± 0.42	1.56 ± 0.39	0.15 [0.02, 0.28]	None
24	Hugues et al. [44]	Singapore	Cross- Sectional	418 men aged 30–69 years	None	24.1 24.2 24.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.33 ± 0.13 All Men 1.24 ± 0.26 All Men 0.93 ± 0.05	1.33 ± 0.18 1.21 ± 0.27 0.91 ± 0.06	[− 0.03, 0.03]0.03 [− 0.02, 0.08] 0.02 [0.01, 0.03]	Age, ethnic group and BMI
25	Kralova et al. [45]	Czech Republic	Cross- Sectional	39 women, 21 nonsmokers aged 51.8 ± 2.4 years and 18 smokers aged 50.5 ± 3.2 years	10 + CPD for at least 3 months	25.1 25.2 25.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Women 1.59 ± 0.29 All Women 1.10 ± 0.39 All Women 0.69 ± 0.06	1.71 ± 0.27 1.02 ± 0.33 1.00 ± 0.17	-0.12 [− 0.30, 0.06] 0.08 [− 0.15, 0.31] -0.31 [− 0.41, − 0.21]	None
26	Liao et al. [46]	China	Cross- Sectional	2505 men and women aged 35–93 years, 1254 of Jing nationality and 1251 of Mulao nationality	None	26.1 26.2 26.3 26.4 26.5 26.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	Jing Minority 1.30 ± 0.23 Mulao Minority 1.38 ± 0.41 Jing Minority 1.06 ± 0.23 Mulao Minority 1.00 ± 0.53 Jing Minority 1.02 ± 0.18 Mulao Minority 0.72 ± 0.19	1.30 ± 0.23 1.36 ± 0.39 1.08 ± 0.25 1.00 ± 0.55 0.98 ± 0.10 0.74 ± 0.21	0.00 [− 0.03, 0.03] 0.02 [− 0.03, 0.07] -0.02 [− 0.05, 0.01] 0.00 [− 0.07, 0.07] 0.04 [0.02, 0.06] -0.02 [− 0.04, 0.00]	None
27	Maeda et al. [47]	Japan	Cohort	349 males and 451 females aged 30–69 years	Everyday smokers	27.1 27.2 27.3 27.4	Apolipoprotein AI Apolipoprotein AII	All Men 1.42 ± 0.18 All Women 1.44 ± 0.17 All Men 0.34 ± 0.05 All Women 0.32 ± 0.04	1.46 ± 0.17 1.48 ± 0.19 0.36 ± 0.05 0.34 ± 0.05	-0.04 [− 0.09, 0.01] -0.04 [− 0.10, 0.02] -0.02 [− 0.03, − 0.01] -0.02 [− 0.04, 0.00]	None
28	Malczewska et al. [48]	Poland	Cross-Sectional	57 females aged 36–46 years	None	28.1 28.2 28.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Women 1.41 ± 0.32 All Women 1.26 ± 0.16 All Women 0.89 ± 0.05	1.70 ± 0.24 1.25 ± 0.28 0.74 ± 0.04	-0.29 [− 0.57, − 0.01] 0.01 [− 0.20, 0.22] 0.15 [0.11, 0.19]	None
29	Meenakshiduram et al. [49]	India	Cross- Sectional	274 men active smokers and 78 controls aged 40–59 years	Stratified by p/y	29.1 29.2 29.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.15 ± 0.25 All Men 1.27 ± 0.28 All Men 1.10 ± 0.12	1.26 ± 0.19 1.16 ± 0.21 0.92 ± 0.05	-0.11 [− 0.17, − 0.05] 0.11 [0.04, 0.18] 0.18 [0.16, 0.20]	None
30	Molgaard & Olsson [50]	Sweden	Cohort	140 men aged 45–69 years	Stratified by CPD	30.1 30.2 30.3 30.4 30.5 30.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.32 ± 0.19 Men 15 + CPD 1.32 ± 0.18 All Men 1.28 ± 0.36 Men 15 + PD 1.35 ± 0.36 All Men 0.96 ± 0.06 Men 15 + CPD 1.02 ± 0.09	1.34 ± 0.17 1.33 ± 0.16 1.21 ± 0.27 1.23 ± 0.28 0.86 ± 0.04 0.92 ± 0.06	-0.02 [− 0.09, 0.05] -0.01 [− 0.10, − 0.08] 0.07 [− 0.05, 0.19] 0.12 [− 0.04, 0.28] 0.10 [0.08, 0.12] 0.10 [0.06, 0.14]	None
31	Paoletti et al. [51]	Italy	Cross- Sectional	22 men aged 21–56 years	20 + CPD	31.1 31.2	Apolipoprotein AI Apolipoprotein B	All Men 1.40 ± 0.16 All Men 0.89 ± 0.12	1.35 ± 0.17 0.87 ± 0.12	0.05 [− 0.09, 0.19] 0.02 [− 0.08, 0.12]	None
32	Pasupathi et al. [60]	India	Cross- Sectional	100 smokers aged 40.1 ± 10.3 years and 100 nonsmokers aged 43.3 ± 9.7 years. All men.	None	32.1 32.2 32.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 0.93 ± 0.17 All Men 1.67 ± 0.15 All Men 1.81 ± 0.62	1.42 ± 0.22 1.00 ± 0.17 0.70 ± 0.03	-0.49 [− 0.54, − 0.44] 0.67 [0.63, 0.71] 1.11 [0.99, 1.23]	None
33	Periti et al. [65]	Italy	Cross Sectional	253 men aged 21–61 years	None	33.1 33.2 33.3 33.4 33.5 33.6 33.7 33.8 33.9	Apolipoprotein B	Men BMI 1 Age 1 0.99 ± 0.28 Men BMI 1 Age 2 1.12 ± 0.21 Men BMI 1 Age 3 1.19 ± 0.15 Men BMI 2 Age 1 1.17 ± 0.18 Men BMI 2 Age 2 1.20 ± 0.20 Men BMI 2 Age 3 1.34 ± 0.17 Men BMI 3 Age 1 1.29 ± 0.25 Men BMI 3 Age 2 1.32 ± 0.22 Men BMI 3 Age 3 1.51 ± 0.32	1.01 ± 0.21 1.20 ± 0.25 1.21 ± 0.33 1.07 ± 0.25 1.15 ± 0.27 1.31 ± 0.21 1.25 ± 0.16 1.18 ± 0.28 1.22 ± 0.32	-0.02 [− 0.18, 0.14] -0.08 [− 0.26, 0.10] -0.02 [− 0.25, 0.21] 0.10 [− 0.05, 0.25] 0.05 [− 0.11, 0.21] 0.03 [− 0.14, 0.20] 0.04 [− 0.16, 0.24] 0.14 [− 0.04, 0.32] 0.29 [0.05, 0.53]	Analysis stratified by BMI and Age tertiles
34	Rodenas et al. [69]	Spain	Cross- Sectional	203 men aged 30–55 years	Stratified by CPD	34.1	Apolipoprotein B	Men 20 + CPD 1.21 ± 0.24	1.15 ± 0.23	0.06 [− 0.05, 0.17]	None
35	Roggi et al. [52]	Italy	Cross- Sectional	510 men and women aged 20–79 years.	Stratified by CPD	35.1 35.2 35.3 35.4 35.5 35.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	Men 20 + CPD 1.38 ± 0.33 Women 20 + CPD 2.18 ± 0.87 Men 20 + CPD 1.40 ± 0.26 Women 20 + CPD 0.91 ± 0.05 Men 20 + CPD 1.01 ± 0.09 Women 20 + CPD 0.42 ± 0.08	1.53 ± 0.41 1.86 ± 0.52 1.29 ± 0.72 1.13 ± 0.34 0.84 ± 0.27 0.61 ± 0.06	-0.15 [− 0.42, 0.12] 0.32 [− 0.54, 1.18] 0.11 [− 0.13, 0.35] -0.22 [− 0.29, 0.15] 0.17 [0.08, 0.25] -0.19 [− 0.27, − 0.11]	None
36	Sakurabayashi et al. [53]	Japan	Cross- Sectional	1018 men and 1167 women aged 20–69 years	Any cigarette use during the previous year	36.1 36.2 36.3 36.4 36.5 36.6 36.7 36.8	Apolipoprotein AI Apolipoprotein AII Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.38 ± 0.17 All Women 1.47 ± 0.18 All Men 0.31 ± 0.05 All Women 0.31 ± 0.04 All Men 0.92 ± 0.18 All Women 0.86 ± 0.16 All Men 0.68 ± 0.16 All Women 0.60 ± 0.16	1.38 ± 0.16 1.43 ± 0.17 0.31 ± 0.04 0.30 ± 0.04 0.97 ± 0.16 0.91 ± 0.16 0.71 ± 0.150.15	0.00 [− 0.04, 0.04] 0.04 [0.00, 0.08] 0.00 [− 0.01, 0.01] 0.01 [0.00, 0.02] -0.05 [− 0.09, − 0.01] -0.05 [− 0.09, − 0.01] -0.03 [− 0.06, 0.00] -0.05 [− 0.09, − 0.01]	None
37	Sanchez et al. [54]	Spain	Cross- Sectional	41 men aged 19–49 years	None	37.1 37.2 37.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 0.99 ± 0.08 All Men 1.28 ± 0.28 All Men 1.29 ± 0.14	1.18 ± 0.17 0.98 ± 0.30 0.83 ± 0.07	-0.19 [− 0.27, − 0.11] 0.30 [0.09, 0.51] 0.46 [0.37, 0.55]	None
38	Saxena et al. [66]	USA	Cross- Sectional	9399 never smokers and 3663 current smokers aged 21 years or older. Men and women	> 100 cigarettes in lifetime, currently smoking every day or some days	38.1	Apolipoprotein B	All 0.94 ± 0.36	0.91 ± 0.41	0.03 [0.02, 0.04]	None
39	Sharma et al. [67]	India	Cross- Sectional	100 men aged 30–40 years	5 + CPD	39.1	Apolipoprotein B	All Men 1.21 ± 0.19	0.77 ± 0.12	0.44 [0.38, 0.50]	None
40	Siekmeier et al. [55]	Germany	Cross Sectional	68 men and women, 34 smokers with average age of 48 ± 7 years and 34 nonsmokers with average age 49.8 ± 5.8 years	None	40.1 40.2 40.3 40.4	Apolipoprotein AI Apolipoprotein AII Apolipoprotein B Ratio Apo B/Apo AI	All 1.39 ± 0.33 All 0.48 ± 0.24 All 0.91 ± 0.26 All 0.65 ± 0.06	1.58 ± 0.36 0.45 ± 0.12 0.84 ± 0.22 0.53 ± 0.03	-0.19 [− 0.35, − 0.03] 0.03 [− 0.06, 0.12] 0.07 [− 0.04, 0.18] 0.12 [0.10, 0.14]	None
41	Sirisali et al. [59]	Thailand	Cross Sectional	128 smokers and 67 nonsmokers, all male aged 35–65	Stratified by CPD	41.1 41.2 41.3 41.4 41.5 41.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.00 ± 0.14 Men 20 + CPD 0.97 ± 0.14 All Men 1.22 ± 0.35 Men 20 + CPD 1.36 ± 0.36 All Men 1.22 ± 0.15 Men 20 + CPD 1.40 ± 0.35	1.32 ± 0.26 1.32 ± 0.26 1.03 ± 0.32 1.03 ± 0.32 0.78 ± 0.03 0.78 ± 0.03	-0.32 [− 0.39, − 0.25] -0.35 [− 0.42, − 0.28] 0.19 [0.09, 0.29] 0.33 [0.21, 0.45] 0.44 [0.41, 0.47] 0.62 [0.53, 0.71]	None
42	Slagter et al. [56]	Netherlands	Cross-Sectional	24389 men and 35078 women aged 18–80 years	Smoked during last month or for longer than a year and had not stopped	42.1 42.2 42.3 42.4 42.5 42.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.40 ± 0.22 All Women 1.57 ± 0.28 All Men 1.00 ± 0.25 All Women 0.93 ± 0.24 All Men 0.71 ± 0.04 All Women 0.59 ± 0.03	1.43 ± 0.22 1.61 ± 0.27 0.94 ± 0.23 0.87 ± 0.23 0.66 ± 0.04 0.54 ± 0.03	-0.03 [− 0.04, − 0.02] -0.04 [− 0.05,− 0.03] 0.06 [0.05, 0.07] 0.06 [0.05, 0.07] 0.05 [0.05, 0.05] 0.05 [0.05, 0.05]	Age
43	Steiner et al. [68]	Canada	Cross- Sectional	52 healthy men aged 45–65 years	None	43.1	Apolipoprotein B	All Men 1.15 ± 0.27	1.02 ± 0.30	0.13 [− 0.03, 0.29]	None
44	Tan et al. [64]	China	Cross- Sectional	2160 Chinese men 65 + years	At least one cigarette per day	44.1	Apolipoprotein B	All Men 1.03 ± 0.16	1.01 ± 0.21	0.02 [0.00, 0.04]	None
45	Vella et al. [57]	Spain	Cross- Sectional	309 men aged 39.9 ± 12.5 years for smokers and 43.7 ± 12.7 years for nonsmokers	Stratified by CPD	45.1 45.2 45.3 45.4 45.5 45.6	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All Men 1.39 ± 0.23 Men 20 + CPD 1.38 ± 0.22 All Men 1.19 ± 0.20 Men 20 + CPD 1.21 ± 0.21 All Men 0.86 ± 0.04 Men 20 + CPD 0.88 ± 0.04	1.47 ± 0.27 1.47 ± 0.27 1.17 ± 0.23 1.17 ± 0.23 0.80 ± 0.05 0.80 ± 0.05	-0.08 [− 0.14, − 0.02] -0.09 [− 0.15, − 0.03] 0.02 [− 0.03, 0.07] 0.04 [− 0.02, 0.10] 0.06 [0.05, 0.07] 0.08 [0.07, 0.09]	None
46	Wang et al. [113]	USA	Cross- Sectional	100 men and women subjects aged 25–84	None	46.1	Apolipoprotein AI	All 1.38 ± 0.49	1.69 ± 0.55	-0.31 [− 0.51, − 0.11]	None
47	Yasue et al. [63]	Japan	Cohort	438 men and women average age of 61.8 ± 11.0 years (smokers) and 68.2 ± 10.7 years (never smokers)	10 CPD + for the 10 years or more	47.1	Apolipoprotein B	All 1.02 ± 0.24	0.95 ± 0.21	0.07 [0.02, 0.12]	None
48	Zagozdzon et al. [62]	Poland	Cross-Sectional	1168 men and 1245 women aged 18–79 years	At least one cigarette per day	48.1 48.2	Apolipoprotein B	All Men 0.96 ± 0.28 All Women 0.92 ± 0.27	0.92 ± 0.26 0.87 ± 0.25	0.04 [0.01, 0.07] 0.05 [0.02, 0.08]	None
49	Zaratin et al. [58]	Brazil	Cross- Sectional	29 men aged 19–35 years	10 + CPD for more than a year	49.1 49.2 49.3	Apolipoprotein AI Apolipoprotein B Ratio Apo B/Apo AI	All 1.34 ± 0.27 All 0.64 ± 0.12 All 0.48 ± 0.02	1.25 ± 0.19 0.67 ± 0.15 0.54 ± 0.02	0.09 [− 0.08, 0.26] -0.03 [− 0.13, 0.07] -0.06 [− 0.07, − 0.05]	None

CPD= cigarettes per day; OC= oral contraceptives; RCT= randomized controlled trial; p/y = pack/year.

Table 2.

Meta-analyses of smoking and apolipoprotein levels (g/L). All estimates.

Compound	Analysis	Studies	Estimates	Mean difference, smokers – nonsmokers (95% CI)
				Fixed effectsa	I2 (%)	Random effectsa
Apolipoprotein AI g/L	All Studies	39	59	-0.038 [− 0.041, − 0.035]	90	-0.073 [− 0.088, − 0.058]
Apolipoprotein AII g/L	All Studies	10	16	-0.003 [− 0.007, 0.002]	66	-0.004 [− 0.014, 0.005]
Apolipoprotein B g/L	All Studies	40	63	0.056 [0.053, 0.059]	95	0.063 [0.045, 0.081]
Ratio Apo B/ Apo AI	All Studies	31	46	0.047 [0.047, 0.048]	99	0.078 [0.068, 0.087]

^aThe estimates used can be found inTable 1.

Using influence diagnostics, four studies (Casasnovas et al. [27], Pasupathi et al. [60], Sirisali et al. [59], and Sharma et al. [67]) were identified as having a large impact on between-study heterogeneity. Thus, sensitivity analyses were conducted excluding these studies.

For the analysis of smoking cessation impact on apolipoprotein levels, eight studies were retrieved. In total, five studies had complete information, with follow-up times of six weeks [88], eight weeks [89], 12 weeks [90], [91], and 90 days [92]. Of the other studies, one included patients that had followed up from two to six weeks after smoking cessation [93], one reported results from smoking reduction [94] and one was a duplicate of another publication [95]. Due to the lack of a sufficient number of studies, no meta-analysis was performed to evaluate the effects of smoking cessation on apolipoprotein levels. The characteristics of the retrieved studies can be found in Table 3, Table 4.

Table 3.

Meta-analyses of smoking and apolipoprotein levels (g/L). Sensitivity analyses with studies by Casanovas et al., Pasupathi et al., Sirisali et al., and Sharma et al. removed.

Compound	Analysis	Studies	Estimates	Mean difference, smokers – nonsmokers (95% CI)
				Fixed effectsa	I2 (%)	Random effectsa
Apolipoprotein AI g/L	Sensitivity	36	55	-0.036 [− 0.039, − 0.033]	71	-0.042 [− 0.053, − 0.032]
Apolipoprotein B g/L	Sensitivity	36	59	0.053 [0.050, 0.055]	81	0.030 [0.018, 0.041]
Ratio Apo B/ Apo AI	Sensitivity	28	42	0.050 [0.049, 0.050]	97	0.059 [0.052, 0.066]

^aThe estimates used can be found in Table 1.

Table 4.

Characteristics of studies assessing levels of apolipoprotein levels after smoking cessation.

Study	Country	Study design	Study participants	Treatment	Findings
Eliasson et al. 1997[89]	Finland	Cohort	17 non-obese healthy male smokers aged 49.3 ± 7.1 years.	Subjects did not receive any therapy and were not allowed to use nicotine replacement therapy.	After eight weeks of abstinence there were statistically significant increases in apolipoprotein AI (from 1.23 ± 0.05 g/L to 1.39 ± 0.03 g/L).
Haziza et al. 2020[92]	USA	Interventional	160 healthy male and female smokers aged 37.7 ± 11.45	Subjects were randomized to mTHS, mCC, or smoking abstinence for 5 days in confinement and 86 subsequent ambulatory days. Subjects in the smoking abstinence group were not allowed to use nicotine replacement therapy.	After 90 days of abstinence, apolipoprotein AI and B levels increased (from 1.41 ± 0.19–1.46 ± 0.25 g/L and from 0.82 ± 0.20–0.83 ± 0.29 g/L, respectively), although neither difference was statistically significant. Switching to mTHS for 90 days resulted in an increase in apolipoprotein AI levels from 1.40 ± 0.21–1.49 ± 0.24 g/L and a decrease in apolipoprotein B levels from 0.88 ± 0.29–0.85 ± 0.29 g/L.
Iwaoka et al. 2014[90]	Japan	Interventional	86 consecutive subjects 20 + years of age	Subjects received 12 weeks of cessation therapy with varenicline.	Serum apolipoprotein AI levels significantly increased from baseline to 12 weeks in the group of successful quitters (1.52 ± 0.28 vs. 1.59 ± 0.27 g/L, p < 0.01).
Masarei et al. 1991[88]	Australia	Interventional	64 male and female smokers aged 35.9 ± 2.3 years	Subjects received 6 regular counsellings at 2-weekly intervals and telephone support.	Serum apolipoprotein AI and AII levels increased from baseline to 6 weeks in successful quitters (1.20 ± 0.15 vs. 1.24 ± 0.25 g/L and 0.51 ± 0.09 vs. 0.57 ± 0.1 g/L, respectively), although only the difference in apolipoprotein A-II was statistically significant.
Takata et al.2014[91]	Japan	Interventional	32 smokers 27–64 years of age	Varenicline or transdermal patch as part of a 12 week smoking cessation program.	Serum apolipoprotein AI levels increased from baseline to 12 weeks in the success group (1.16 ± 0.21 vs. 1.21 ± 0.13 g/L, p < 0.22), although the difference was not statistically significant.

3.2. Effects of smoking status on apolipoprotein AI levels

The pooled mean levels of apolipoprotein AI were 1.48 ± 0.009 g/L in smokers and 1.54 ± 0.007 g/L in nonsmokers. The meta-analysis for apolipoprotein AI levels between smokers and nonsmokers was performed using data from 39 studies including 59 estimates [23], [24], [25], [26], [27], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61]. The fixed-effects model mean difference was − 0.038 (95% CI: −0.041, −0.035 g/L, I² =90%). Because of the large heterogeneity, sensitivity analyses were conducted, removing the studies that were driving the most heterogeneity. Data from 36 studies, including 55 estimates, were included in the sensitivity analysis. The fixed-effects model mean difference was − 0.036 (95% CI: −0.039, −0.033 g/L, I² =71%). To address the large heterogeneity further, exploratory subgroup analyses were performed. The results of these analyses can be found in Table 5. Stratification by region explained up to 28% points of the I² value when considering only studies run in the Americas (mean difference= −0.062, 95% CI: −0.073, −0.051 g/L, I² = 43%), while most of the heterogeneity originated from Asian and African studies (I² =73%). Stratification by study design explained up to one percentage point (mean difference = −0.034, 95% CI: −0.039, −0.029 g/L, I² = 70% in the cohort subgroup). The stratification by period of publication explained up to 59% points (mean difference= −0.010, 95% CI: −0.013, 0.033 g/L, I² = 12% in the 2000 s subgroup) and the stratification by smoking definition explained up to 20% points (mean difference= −0.074, 95% CI: −0.134, −0.014 g/L, I² = 51% in the 15 + CPD subgroup). The stratification by sex explained up to 30% points (mean difference = −0.040, 95% CI: −0.044, −0.036 g/L, I² = 41% in the women-only subgroup).

Table 5.

Subgroup analyses of smoking and apolipoprotein AI levels (g/L).

Stratification	Studies	Effect estimates	Mean differences in apolipoprotein AI levels (g/L) (smokers – nonsmokers)
			Fixed effects [95% CI]	I2 (%)	Random effects [95% CI]
Region
Americas	7	10	-0.062 [− 0.073, − 0.051]	43	-0.094 [− 0.147, − 0.041]
Europe	20	30	-0.035 [− 0.038,- 0.031]	64	-0.044 [− 0.054, − 0.033]
Asia & Africa	12	19	-0.054 [− 0.065,- 0.042]	96	-0.105 [− 0.170, − 0.041]
Study Design
Cross Sectional	28	44	-0.040 [− 0.045, − 0.035]	92	-0.098 [− 0.123, − 0.072]
Cohort	11	15	-0.036 [− 0.040, − 0.033]	76	-0.038 [− 0.053, − 0.024]
Period of Publication
1970–1980 s	9	17	-0.066 [− 0.086, − 0.046]	62	-0.103 [− 0.146,- 0.060]
1990 s	17	24	-0.061 [− 0.075, − 0.047]	87	-0.088 [− 0.131, − 0.045]
2000 s	4	6	-0.065 [− 0.086,- 0.044]	98	-0.078 [− 0.250, 0.093]
2010 s	9	12	-0.035 [− 0.038, − 0.032]	84	-0.036 [− 0.047, − 0.024]
Smoking Definitiona
No Definition/Other	24	35	-0.036 [− 0.039, − 0.033]	93	-0.060 [− 0.077, − 0.044]
> 5 CPD	2	2	NA	NA	NA
> 10 CPD	6	6	-0.122 [− 0.170, − 0.074]	56	-0.095 [− 0.173, − 0.017]
> 15 CPD	2	3	-0.074 [− 0.134,- 0.014]	51	-0.082 [− 0.172, 0.008]
> 20 CPD	9	13	-0.077 [− 0.097, − 0.057]	87	-0.118 [− 0.193, − 0.044]
Sexa
Men	31	39	-0.033 [− 0.037, − 0.028]	93	-0.091 [− 0.116, − 0.066]
Women	14	15	-0.040 [− 0.044, − 0.036]	41	-0.039 [− 0.051, − 0.026]
Combined	5	5	-0.052 [− 0.062,- 0.042]	86	-0.049 [− 0.105, 0.007]
Total	39a	59	-0.038 [− 0.041, − 0.035]	90	-0.073 [− 0.088, − 0.058]

^a: sum of studies does not add up to 39, as some studies provided multiple stratum-specific estimates.

Applying the random-effects model to acknowledge the high interstudy heterogeneity resulted in a mean difference of − 0.073 (95% CI: −0.088, −0.058 g/L) for all studies and − 0.042 (95% CI: −0.053, −0.032 g/L) for studies included in the sensitivity analysis. Visual inspection of the funnel plots (Fig. 2) revealed some asymmetries, with several large studies showing a significant negative effect. Funnel plot asymmetry was confirmed using Egger’s regression test (intercept = −0.62, 95% CI: −1.17, −0.07, t = −2.21, p = 0.03) and may be an indication of publication bias.

Fig. 2.

Funnel plot of studies reporting apolipoprotein AI levels (g/L).

3.3. Effects of smoking status on apolipoprotein AII levels

The pooled mean levels of apolipoprotein AII were 0.32 ± 0.007 g/L in smokers and 0.30 ± 0.004 g/L in nonsmokers. The meta-analysis of the effects of smoking on apolipoprotein AII levels included ten studies and 16 estimates [23], [24], [26], [32], [34], [36], [43], [47], [53], [55] and found no differences in levels of apolipoprotein AII between smokers and nonsmokers (mean difference = −0.003; 95% CI: −0.007, 0.002 g/L, I² =66%). Considerable heterogeneity was found among studies; however, the studies identified through the sensitivity analysis did not report estimates for apolipoprotein AII levels. Therefore, subgroup analyses were performed using data from all studies. The detailed results of these analyses can be found in Table 6. The stratification by study design explained up to 25% points of the I² value (mean difference= 0.002, 95% CI: −0.003, 0.007 g/L, I² = 41% in the cross-sectional studies subgroup). The stratified analysis by period of publication explained 14% points (mean difference = −0.014, 95% CI: −0.023, −0.005 g/L, I² = 52% in the 1990 s subgroup). The major source of heterogeneity seemed to come from studies in the US (mean difference= −0.008, 95% CI: −0.027, 0.011 g/L I² = 73%) and studies involving only women (mean difference= 0.00, 95% CI: −0.007, 0.008 g/L, I² = 74%). Applying the random-effects model estimated a mean difference of − 0.004 (95% CI: −0.014, 0.005 g/L). The evaluation of the funnel plot (Fig. 3) did not show evidence of publication bias and the results of Egger’s regression test were not significant (intercept = 0.20, 95% CI: −1.32, 1.73, t = 0.26, p = 0.80).

Table 6.

Subgroup analyses of smoking and apolipoprotein AII levels (g/L).

Stratification	Studies	Effect estimates	Mean differences in apolipoprotein AII levels (g/L) (smokers – nonsmokers)
			Fixed effects [95% CI]	I2 (%)	Random effects [95% CI]
Region
Americas	3	6	-0.008 [− 0.027, 0.011]	73	0.012 [− 0.029, 0.052]
Europe	3	3	-0.002 [− 0.016, 0.012]	45	0.004 [− 0.021, 0.028]
Asia & Africa	4	7	-0.003 [− 0.008, 0.002]	72	-0.007 [− 0.017, 0.004]
Study Design
Cross Sectional	7	11	0.002 [− 0.003, 0.007]	41	0.002 [− 0.007, 0.010]
Cohort	3	5	-0.019 [− 0.028, − 0.010]	65	-0.017 [− 0.035, 0.0002]
Period of Publication
1970–1980 s	4	7	-0.009 [− 0.021, 0.003]	68	0.004 [− 0.024, 0.032]
1990 s	4	5	-0.014 [− 0.023, − 0.005]	52	-0.008 [− 0.024, 0.007]
2000 s	2	4	0.002 [− 0.003, 0.008]	60	-0.001 [− 0.009, 0.010]
Sexa
Men	9	9	-0.005 [− 0.010, 0.001]	66	-0.005 [− 0.016, 0.007]
Women	5	6	0.0002 [− 0.007, 0.008]	74	-0.003 [− 0.024, 0.019]
Combined	1	1	NA	NA	NA
Total	10a	16	-0.003 [− 0.007, 0.002]	66	-0.004 [− 0.014, 0.005]

^a: sum of studies does not add up to 10, as some studies provided multiple stratum-specific estimates.

Fig. 3.

Funnel plot of studies reporting apolipoprotein AII levels (g/L).

3.4. Effects of smoking status on apolipoprotein B levels

The pooled mean levels of apolipoprotein B were 1.06 ± 0.009 g/L in smokers and 0.98 ± 0.007 g/L in nonsmokers. The meta-analysis of apolipoprotein B levels in smokers versus nonsmokers included 40 studies and 63 estimates [23], [24], [27], [28], [29], [30], [32], [34], [35], [37], [38], [39], [40], [41], [42], [44], [45], [46], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [62], [70]. Increased levels of apolipoprotein B were found in smokers using the fixed-effects model (mean difference= 0.056, 95% CI: 0.053, 0.059 g/L, I² =95%). The heterogeneity was high; therefore, sensitivity and subgroup analyses were performed. The sensitivity analysis included data from 36 studies and 59 estimates. The fixed-effects mean difference was 0.053 g/L (95% CI: 0.050, 0.055 g/L, I² =81%). The results of the subgroup analysis can be found in Table 7. The stratified analyses found that four percentage points of the I² value could be explained by including only cohort studies (mean difference= 0.057, 95% CI: 0.053, 0.61 g/L, I² = 77%), 14% points could be explained by limiting the studies to those performed in Europe (mean difference= 0.055, 95% CI: 0.052, 0.059 g/L, I² = 67%), 14% points could be explained by limiting to studies run in the 1990 s (mean difference= 0.020, 95% CI: 0.003, 0.037 g/L, I² = 67%), and 17% points could be explained by limiting to studies containing only men (mean difference= 0.054, 95% CI: 0.049, 0.058 g/L, I² = 64%). Stratification by smoking definition explained up to 100% points when including only studies using 10 + CPD as their smoking definition (mean difference= 0.006, 95% CI: −0.071, 0.082 g/L, I² = 0%); however, this subgroup contained only four studies. The analyses showed that most of the heterogeneity came from studies performed in the US (I² =89%), studies containing only women (I² =92%), and studies containing both men and women (I² =92%). Studies performed in Asia and Africa (mean difference=−0.004, 95% CI: −0.016, 0.007 g/L, I² =70) and studies from the 2000 s (mean difference=−0.003, 95% CI: −0.017, 0.012 g/L, I² =83) revealed an opposite effect. The random-effects model yielded similar results to those seen with the fixed-effects model, with a mean difference of 0.063 g/L (95% CI: 0.045, 0.081 g/L) for all studies and 0.030 g/L (95% CI: 0.018, 0.041 g/L) for studies included in the sensitivity analysis. Visual evaluation of the funnel plot (Fig. 4) showed some evidence pointing toward publication bias; however, the results of Egger’s regression test were not significant (intercept = −0.62, 95% CI: −1.30, 0.06, t = −1.78, p = 0.08).

Table 7.

Subgroup analyses of smoking and apolipoprotein B levels (g/L).

Stratification	Studies	Effect estimates	Mean differences in the ratio of apolipoprotein B (g/L) levels (smokers – nonsmokers)
			Fixed effects [95% CI]	I2 (%)	Random effects [95% CI]
Region
Americas	6	6	0.064 [0.055, 0.073]	89	0.062 [0.019, 0.105]
Europe	20	37	0.055 [0.052, 0.059]	67	0.050 [0.038, 0.062]
Asia & Africa	10	15	-0.004 [− 0.016, 0.007]	70	-0.010 [− 0.035, 0.015]
Study Design
Cross Sectional	29	49	0.047 [0.043, 0.052]	81	0.023 [0.007, 0.039]
Cohort	7	9	0.057 [0.053, 0.061]	77	0.066 [0.051, 0.081]
Period of Publication
1970–1980 s	6	9	0.054 [0.016, 0.092]	68	0.074 [0.004, 0.143]
1990 s	16	29	0.020 [0.003, 0.037]	67	0.038 [0.004, 0.073]
2000 s	5	7	-0.003 [− 0.017, 0.012]	83	-0.018 [− 0.058, 0.022]
2010 s	8	11	0.056 [0.053, 0.059]	87	0.053 [0.042, 0.064]
2020 s	1	2	NA	NA	NA
Smoking Definitiona
No Definition/other	24	40	0.053 [0.050, 0.056]	82	0.056 [0.037, 0.75]
> 5 CPD	2	2	NA	NA	NA
> 10 CPD	4	4	0.006 [− 0.071, 0.082]	0	0.006 [− 0.071, 0.082]
> 15 CPD	1	1	NA	NA	NA
> 20 CPD	8	11	0.007 [− 0.021, 0.035]	86	0.073 [− 0.013, 0.159]
Sexa
Men	28	42	0.054 [0.049, 0.058]	64	0.040 [0.025, 0.055]
Women	10	10	0.051 [0.047, 0.055]	92	0.001 [− 0.025, 0.027]
Combined	5	6	0.057 [0.048, 0.066]	92	0.039 [− 0.002, 0.080]
Total	36a	59	0.053 [0.050, 0.055]	81	0.030 [0.018, 0.041]

^a: sum of studies does not add up to 37, as some studies provided multiple stratum-specific estimates.

Fig. 4.

Funnel plot of studies reporting apolipoprotein B levels (g/L).

3.5. Effects of smoking status on the ratio of apolipoprotein B to AI

The pooled mean ratios of apolipoprotein B to AI were 0.62 ± 0.002 in smokers and 0.65 ± 0.001 in nonsmokers. The meta-analysis assessing the effects of smoking on the apolipoprotein B: AI ratio included 31 studies with 46 estimates [23], [24], [27], [28], [29], [30], [32], [34], [35], [37], [38], [39], [40], [41], [42], [44], [45], [46], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [62], [70]. The fixed-effects model showed an increased ratio in smokers (mean difference= 0.047, 95% CI: 0.047, 0.048 g/L, I² =99%). The heterogeneity was high; thus, sensitivity and subgroup analyses were performed. The sensitivity analysis included data from 28 studies and 42 estimates; however, the results did not show much reduction in heterogeneity (mean difference= 0.050, 95% CI: 0.049, 0.050 g/L, I² =97%). Heterogeneity did not vary much across subgroups. Considering only studies from the 2000 s resulted in the lowest heterogeneity; however, this revealed an opposite effect (mean difference= −0.048, 95% CI: −0.060, −0.036 g/L, I² =77). Detailed results of the subgroup analyses can be found in Table 8. Despite the high heterogeneity, the random-effects model confirmed an increased ratio, with a mean difference of 0.078 g/L (95% CI: 0.068, 0.087 g/L) for all studies and 0.059 g/L (95% CI: 0.052, 0.066 g/L) for studies in the sensitivity analysis. The evaluation of the funnel plot showed (Fig. 5) no evidence of publication bias and the results of Egger’s regression test were not significant (intercept = 0.47, 95% CI: −1.65, 2.6, t = 0.42, p = 0.66).

Table 8.

Subgroup analyses of smoking and apolipoprotein B/apolipoprotein AI ratio levels (g/L).

Stratification	Studies	Effect estimates	Mean differences in apolipoprotein B/AI levels (g/L) (smokers – nonsmokers)
			Fixed effects [95% CI]	I2 (%)	Random effects [95% CI]
Region
Americas	3	3	0.044 [0.035, 0.052]	100	0.099 [− 0.038, 0.235]
Europe	16	24	0.051 [0.050, 0.051]	95	0.068 [0.062, 0.074]
Asia & Africa	9	15	0.015 [0.011, 0.019]	97	0.041 [0.013, 0.067]
Study Design
Cross Sectional	22	34	0.049 [0.049, 0.050]	98	0.056 [0.048, 0.064]
Cohort	6	8	0.057 [0.054, 0.060]	90	0.071 [0.057, 0.085]
Period of Publication
1970–1980 s	4	7	0.030 [0.023, 0.037]	99	0.102 [0.029, 0.174]
1990 s	13	19	0.047 [0.043, 0.051]	96	0.073 [0.054, 0.093]
2000 s	3	5	-0.048 [− 0.060, − 0.036]	77	-0.033 [− 0.066, 0.001]
2010 s	8	11	0.050 [0.050, 0.051]	97	0.059 [0.052, 0.066]
Smoking Definitiona
No Definition/other	18	25	0.050 [0.049, 0.051]	97	0.056 [0.048, 0.063]
> 5 CPD	2	2	NA	NA	NA
> 10 CPD	4	4	-0.042 [− 0.055, − 0.029]	96	-0.051 [− 0.149, 0.046]
> 15 CPD	1	1	NA	NA	NA
> 20 CPD	7	10	0.061 [0.055, 0.067]	98	0.085 [0.039, 0.130]
Sexa
Men	22	28	0.049 [0.048, 0.050]	98	0.073 [0.058, 0.088]
Women	9	9	0.050 [0.049, 0.051]	94	0.031 [0.016, 0.045]
Combined	5	5	0.076 [0.067, 0.085]	97	0.095 [0.039, 0.151]
Total	28a	42	0.050 [0.049, 0.050]	97	0.059 [0.052, 0.066]

^a: sum of studies does not add up to 31, as some studies provided multiple stratum-specific estimates.

Fig. 5.

Funnel plot of studies reporting apolipoprotein B:AI ratio.

4. Discussion

The meta-analyses performed comparing levels of apolipoprotein levels in smokers versus nonsmokers found that apolipoprotein AI and AII levels were higher in nonsmokers than in smokers. On the other hand, apolipoprotein B levels and apolipoprotein B:AI ratios were higher in smokers compared to nonsmokers. These findings describe an overall worse lipid profile in smokers compared to nonsmokers. A limited number of publications was found where the influence of smoking cessation on apolipoprotein levels or the ratio of apolipoprotein B to AI was assessed. For this reason, no meta-analysis was performed on the effects of smoking cessation and these endpoints. Nevertheless, the retrieved publications individually suggested that it is likely apolipoprotein AI and AII levels improve after quitting smoking.

Cigarette smoking can result in disease and disability and is the leading cause of preventable death [96]. Quitting greatly reduces the risk of developing smoking-related disease [97], but it has been proven difficult to achieve [98]. The FDA has published draft guidelines on modified risk tobacco products (MRTPs) [99]. These have led to the evaluation of risk reduction through the use of clinical risk markers [97], which should, in principle, be associated with smoking and be influenced by smoking cessation. A class of clinical risk endpoints usually assessed is lipids, but mostly with conventional measurements such as HDL-C and LDL-C [100]. Cigarette smoking is positively associated with increased levels of triglycerides and lower levels of (HDL-C) [6], [101]. However, other than for HDL-C, no meta-analysis has been published on the effects of smoking and cessation on apolipoprotein levels.

Standard lipid analyses for cardiovascular risk assessment include measuring serum or plasma total cholesterol, triglycerides, and HDL-C after an overnight fast [102]. Data from the Framingham Offspring study indicate that LDL-C, small dense LDL-C, lipoprotein(a) or Lp(a), and HDL particle measurements add significant information about CVD risk to the standard lipid profile (summarized in [102]). Furthermore, it has been shown that cardiovascular risk assessment cannot rely solely on the determination of these measurements and that other lipid parameters can be more useful and more predictive of CVD [103], [104]. For instance, the measurement of apolipoprotein AI in HDL particles by gel electrophoresis is important in assessing CVD risk and HDL functionality [102]. Evidence from epidemiologic data suggests that instead of measuring the cholesterol in LDL-C or HDL-C, measuring their respective apolipoproteins, apolipoprotein B and apolipoprotein AI, may improve coronary heart disease risk assessment, and in some observational and interventional studies, ratios of lipids and/or apolipoproteins have been better predictors of coronary heart disease risk than levels of any one lipid fraction [104], [105], [106]. According to our findings, smoking deteriorates lipid profiles by affecting the levels of all the apolipoprotein measurements.

Meta-analyses involve a two-step analysis approach, whereby in the first step, analyses of the data from the individual studies result in effect estimates and standard errors [107]. In the second step, these data are then combined. For this step, individual subject data are not generally available and effects are typically extracted from publications and then pooled [107]. Meta-analysis is a statistical method recommended by the Cochrane Collaboration [108] to combine individual results to make the best use of all available data and therefore increase the power of the analysis. Nonetheless, although meta-analyses are a robust method, they have limitations. These limitations are mostly related to the identification of studies, interstudy heterogeneity, and the availability of information [109]. For instance, in both the PubMed and Scopus searches and in the review of the reference lists, 66 publications were identified as assessing the influence of smoking status on apolipoprotein measurements, but only 49 had complete information that could be used in the analyses. Additionally, for the effects of smoking cessation, only five studies out of the eight retrieved contained useful information. Furthermore, substantial interstudy heterogeneity appeared in all analyses (I² values ranging from 66% in the apolipoprotein AII analysis to 99% in the apolipoprotein B:AI ratio analysis). The latter value is probably because ratios are usually not normally distributed, which could have added to the interstudy heterogeneity. It should be noted that the heterogeneity for apolipoprotein AI and B levels is mainly due to the size of the estimates, whereas for apolipoprotein AII levels, there is also heterogeneity in regard to the direction of the effect. The fact that the studies originated from diverse populations may have also contributed to the high interstudy heterogeneity, as many biological and environmental factors are known to influence apolipoprotein levels. For example, differences in lipoprotein metabolism [110], age [110], [111], diet [110], BMI [112], and alcohol intake [110], [112] have been shown to contribute to variation in serum apolipoprotein concentrations. For this reason, sensitivity analyses were conducted in which the studies that contributed most to the heterogeneity were removed. These studies included Casasnovas et al. [27], Pasupathi et al. [60], Sirisali et al. [59], and Sharma et al. [67]. The characteristics of the study populations in these studies may have differed from those in the other publications, resulting in increased heterogeneity. Participants in the study by Casasnovas et al. [27], for instance, had an average age of 19.9 ± 0.9 years, which is much lower than other studies. The results of the sensitivity analyses were consistent with the fixed analyses, and the interstudy heterogeneity was reduced by 19% and 14% for apolipoproteins AI and B, respectively, and by 2% for the apolipoprotein B/AI ratio. No sensitivity analysis was conducted for apolipoprotein AII as the excluded studies did not report data for this measure. Subgroup analyses and a random-effects model were implemented to account for the interstudy heterogeneity, both of which showed consistent results to the fixed analyses.

5. Conclusions

The meta-analyses show increased levels of apolipoproteins AI and AII in nonsmokers relative to smokers, while apolipoprotein B levels and apolipoprotein B/AI ratio were increased in smokers relative to nonsmokers. Based on the findings of the meta-analyses, smoking appears to negatively alter the lipid profile of current smokers relative to nonsmokers. These data need to be interpreted with caution, however, due to the high heterogeneity encountered in the analyses. Concerning smoking cessation and apolipoprotein levels, more data is needed to evaluate whether smoking cessation improves the levels of these markers of cardiovascular risk.

Funding

Philip Morris International is the sole source of funding and sponsor of this research.

CRediT authorship contribution statement

Alba Romero Kauss : Data curation, Formal analysis, Writing – original draft. Meagan Antunes: Data curation, Analysis, Validation, Writing – review & editing, Writing – original draft. Guillaume de La Bourdonnaye: Data validation, review of the original draft. Sandrine Pouly: Conceptualization, Data validation. Matthew Hankins: Review of the original draft. Annie Heremans: Resources and review of the original draft. Angela van der Plas: Conceptualization, Writing – original draft.

Declaration of Competing Interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: All authors are employed by Philip Morris International.

Appendix A. Supplementary material

Supplementary material

.