Association between exposure to per- and polyfluoroalkyl substances and levels of lipid profile based on human studies

Introduction

Cerebrovascular disease (CVD) is a public health epidemic, and as the leading contributor to death and disability globally, accounting for approximately 17.8 million deaths annually attributed to CVD [1]. Besides the traditional CVD risk factors, such as hypertension, obesity, smoking, unhealthy diet, diabetes and dyslipidemia, concerns of exposure to environmental chemicals and air pollution, e.g., heavy metals, ambient air pollution, persistent organic pollutants and pesticides, are growing [2].

Endocrine disrupting chemicals (EDCs) are mixture of exogenous chemicals that contain about 1,000 chemicals. It can interact with endocrine systems to affect the function of hormones causing adverse health for human [3]. Many EDCs have been confirmed to be associated with the occurrence of diabetes, obesity, immune, reproductive and behavioral complications [4]. Recently, several epidemiological studies have found that EDCs (e.g., bisphenol A (BPA), phthalates, plastic-associated chemicals (PACs) and perfluoroalkyl and polyfluoroalkyl substances (PFAS)) were thought to have association with CVD [5], 6]. Moreover, animal studies also have shown that EDCs were thought to damage the cardio-cerebral vascular system. For example, BPA exposure can increase the risk of atherosclerosis by altering the expression of cardiac microRNAs in mice [7]. In sum, these studies indicate that exposure to EDCs are an important risk factor for the pathogenesis of CVD.

PFAS as an essential category of EDCs was normally used in waterproof fabrics, nonstick cookware, food packaging and carpets [8]. The most frequently studied of all PFAS family members include perfluorooctanoic acid (PFOA), perfluorodecanoic acid (PFDA), perfluorooctane sulfonic acid (PFOS), perfluorooctane sulfonic acid (PFHxS), perfluoroundecanoic acid (PFUnDA), and perfluorononanoic acid (PFNA). Previous studies have shown that PFAS exposure affects the levels of cholesterol, low-density lipoprotein (LDL), high-density lipoprotein (HDL) and categorized PFAS as a major risk factor for the incidence of CVD [9]. Meanwhile, several animal studies also suggested that PFAS exposure may mediate the alteration of lipid profile levels (e.g., triglyceride, cholesterol, and lipoproteins) via disrupting the hepatic-intestinal circulation, thereby causing CVD [10], 11]. Lipoproteins, as a causal and independent risk factor for CVD, are the main form of transported lipid and triglycerides in the body [12]. However, previous studies remain controversial regarding the relationship between PFAS exposure and lipoproteins. For example, a cross-sectional study showed that PFOA and PFOS exposure did not disturb both HDL and TG levels [13]. However, another study found that serum PFOA, PFOS and PFNA exposure have a positive correlation with TG [14], and a negative correlation with HDL [15], 16]. Thence, we performed a meta-analysis to systematically and comprehensively evaluate the association between PFAS exposure and serum lipoproteins, and identify possible sources of heterogeneity between different studies.

Methods

Results

Study selection

A total of 3,142 studies were included in this meta-analysis from five electronic databases: PubMed (364 articles), Web of Science (696 articles), Cochrane Library (4 articles), Embase (589 articles) and Ovid (1,489 articles). After removal of duplicate studies, 1,469 studies remained. Next, 1,409 articles were excluded because they are review, comments, conference papers, animal studies or articles not related to the research topic according to the titles and abstracts of the articles. Of the remaining 60 studies, 31 were excluded due to Chinese article (1 article), irrelevant exposures and/or outcomes (29 articles) or incomplete statistics (1 article). Ultimately, 17 studies including eleven cross-sectional studies and six cohort studies were incorporated in this meta-analysis after quality assessment (excluding 12 articles due to low quality) (Figure 1).

Figure 1:

Flow chart based on the preferred reporting items for meta-analyses (PRISMA) protocol proposal.

Association between PFOA exposure and serum lipid levels

A total of 16 papers were included in the meta-analysis of the correlation between PFOA exposure and lipid levels in serum: 14 for HDL, 11 for LDL, 14 for TC, 14 for TG and four for non-HDL. Egger analysis was performed to assess publication bias (Table S4).

We found that there were no dramatically relationships between PFOA exposure and the levels of HDL (β value=0.01; 95 % CI: −0.02, 0.04), LDL (β value=0.06; 95 % CI: −0.00, 0.13), TG (β value=0.01; 95 % CI: −0.05, 0.07), non-HDL (β value=0.06; 95 % CI: −0.09, 0.22), and due to significant heterogeneity, the random-effects model was applied (Figure 2A, B, D and E). However, there was significant positive correlation between PFOA exposure and TC level (β value=0.08; 95 % CI: 0.02, 0.14). Egger analysis showed a high publication bias between PFOA and LDL, TC (Table S4). Therefore, the trim and fill method were performed to identify the stability of the results [36]. The results by the trim and fill analysis showed that PFOA had no relation to LDL (β value=0.06; 95 % CI: −0.01, 0.13), and positive correlation between PFOA and TC was observed (β value=0.07; 95 % CI: 0.00, 0.14) (Figure S1). The random-effects model was applied to the overall analysis considering higher heterogeneity (Figure 2).

Figure 2:

Pooled estimate (random-effects model) of β value with 95 % CI of relationship between PFOA and lipid profile (A, HDL; B, LDL; C, TC; D, TG; E, non-HDL) concentrations.

Correlation between PFOS exposure and serum lipid levels

Fifteen studies were incorporated into meta-analysis. The serum lipids enrolled in this study included: LDL (11 articles), HDL (13 articles), TC (14 articles), TG (13 articles), and non-HDL (4 articles). Egger analysis was performed to assess publication bias (Table S4). Our results showed that the estimated β value exhibited a significantly positive correlation between PFOS and the levels of LDL (β value=0.13; 95 % CI: 0.04, 0.21), TC (β value=0.13; 95 % CI: 0.05, 0.21) (Figure 3B and C). However, there was no remarkable relationship between PFOS and other lipids (e.g., HDL, TG and non-HDL) (Figure 3A, D and E). The random-effects model was used to assess the correlation between PFOS and lipid concentrations (e.g., HDL, LDL, TC, non-HDL) due to I ² >50 %, and the association between PFOS and lipid concentration (e.g., TG) was evaluated by a fixed-effects model according to I ² <50 %. We used the trim-and-fill method to adjust for publication bias, and the results demonstrated that the estimated β value exhibited a positive correlation between PFOS and the levels of LDL (β value=0.11; 95 % CI: 0.02, 0.21) and TC (β value=0.12; 95 % CI: 0.03, 0.21), suggesting our results are shown to be stable (Figure S1).

Figure 3:

Pooled estimate (fixed-effects model or random-effects model) of β value with 95 % CI of relationship between PFOS and lipid profile (A, HDL; B, LDL; C, TC; D, TG; E, non-HDL) concentrations.

Relationship between PFUnDA exposure and serum lipid levels

A total of five studies were enrolled to evaluate the relationship between PFUnDA and the levels of lipid, including: HDL (4 articles), LDL (4 articles), TG (5 articles) and TC (5 articles). A random-effects model was used to evaluate the pooled estimate β value due to I ² >50 %, the result showed a significant relationship between PFUnDA and the levels of LDL (β value=0.13; 95 % CI: 0.02, 0.24) (Figure 4B). Whereas there was no correlation between PFUnDA exposure and HDL (β value=0.09; 95 % CI: −0.01, 0.20), TC (β value=0.17; 95 % CI: −0.01, 0.34) or TG (β value=−0.05; 95 % CI: −0.13, 0.04) (Figure 4A, C and D).

Figure 4:

Pooled estimate (random-effects model) of β value with 95 % CI of relationship between PFUnDA and lipid profile (A, HDL; B, LDL; C, TC; D, TG) concentrations.

Discussion

In this study, we utilized multiple databases, including Embase, PubMed, Ovid, The Cochrane Library, and Web of Science, to explore the relationship between PFAS exposure and lipid profile. Our results revealed that elevated levels of LDL were notably linked to exposure to PFUnDA (β value=0.13, 95 % CIs: 0.02, 0.24) and perfluorooctane sulfonic acid (PFOS) (β value=0.13, 95 % CIs: 0.04, 0.21). Furthermore, exposure to PFOA, PFOS, and PFNA exhibited a significant association with higher total cholesterol (TC) levels, with pooled effect estimates of 0.08 (95 % CI: 0.02, 0.14), 0.13 (95 % CI: 0.05, 0.21), and 0.14 (95 % CI: 0.08, 0.20), respectively. We also performed subgroup analyses to identify the sources of heterogeneity which indicating the robustness of our results. Our results highlight PFOA, PFOS, PFNA, and PFUnDA as crucial risk factors contributing to abnormal lipid profile levels (Figure 5). These findings underscore the importance of implementing measures to reduce and control PFAS exposure, suggesting a potential avenue for preventing cerebrovascular diseases associated with unfavorable lipid profile.

Figure 5:

Cartoon picture of association between exposure to per- and polyfluoroalkyl substances and levels of lipid profile.

PFAS are a class of EDCs capable of imitating or disrupting the endocrine system, further resulting in altered biological processes in vivo, such as immunity, reproduction and metabolism [37], 38]. To date, over 4,700 PFAS have been developed and produced intentionally in the world, including PFOA, PFOS, PFUnDA, PFHxS, perfluorododecanoic acid (PFDoA) and perfluoroheptanoic acid (PFHpS) [39]. Therefore, over the past few years, the term PFAS has developed into a hallmark of environmental contamination that has attracted public, scientific and regulatory attention due to the widespread use of PFAS worldwide [40].

CVD, including hypertension, cardiometabolic disease, coronary artery disease (CAD), peripheral vascular disease (PVD), cerebrovascular disease, and congenital heart disease (CHD), has become a primary contributor to morbidity and mortality in Western developed countries and is a serious economic and medical challenge to humanity [41]. Dyslipidemia, is an abnormality of lipid metabolism characterized by elevated levels of serum LDL, TG and TC, and reduced levels of HDL, and dyslipidemia is a major determinant and critical risk factor for CVD [42]. Moreover, hypercholesterolemia is related to an elevated risk of CVD and is the most commonly diagnosed form of dyslipidemia. Over the past several decades, numerous studies have revealed that PFAS is one of the important environmental factors contributing to aberrant cholesterol levels [10], 43]. Several cross-sectional studies showed that the levels of PFOA and PFOS in serum are associated with an increase in LDL and TC concentration [28], 30]. However, other studies indicated that PFOS exposure was not associated with LDL and TC levels [24], 25]. Consequently, a systematic review is essential to establish the relationship between EDC exposure and levels of lipid profile.

In this study, we included 17 eligible articles, including 56,210 participants, and a systematical and comprehensive investigation was offered to uncover the association between PFAS exposure and the levels of lipids (e.g., HDL, LDL, TC, TG and non-HDL) in serum. When parity was not considered, our results indicated that high levels of PFOA and PFOS exposure exhibited a remarkable positive correlation with serum LDL and TC, but not on HDL, TG and non-HDL. Also, we found that PFNA exposure did not correlate with the levels of HDL, LDL and TG, while displayed a positively correlated with TC levels. Moreover, we also found that PFunDA exposure is a potentially risk factor for elevated LDL levels. However, the associations of PFHxS and PFDA exposure with lipid profile were not observed in this study.

To identify potential sources of heterogeneity, we included age, year of article publication, sample size, and study area in the subgroup analysis. We found that age was one of the key confounding factors affecting the association between PFAS exposure and lipid profile. For example, our results demonstrated that the risk of anomalous lipid levels (e.g., LDL, TG, TC) increased at ages over 60 years old than in under 60 years old when exposed to PFOA. The elderly cohort has more risk factors associated with CVD due to aging, such as degree of arterial stiffness, metabolic disorders, hypertension and other vascular diseases leading to greater sensitivity to some risk factors [44]. Moreover, the risk of anomalous lipid levels (e.g., LDL and TC) increased at ages under 20 years old when exposed to PFOS. In recent years, the excessive intake of high-energy foods and sedentary lifestyles among young people has led to a progressively younger incidence of CVD, hyperlipidemia, obesity and other disorders [45], which may be further exacerbated by exposure to environmentally harmful factors (e.g., PFOS). However, other important covariates, such as race, occupation and socioeconomic status could not be evaluated due to limited baseline data.

Recently, several PFAS exposure (e.g., PFOA, PFOS and hexafluoropropylene oxide trimer acid (HFPO-TA)) results in abnormal lipid metabolism have been confirmed by animal studies. One study revealed that the levels of LDL, TC and TG were significantly increased in Zebrafish PFOA exposure group, and RNA-seq data showed that disorder of lipid metabolism may be via altering the expression of genes related to lipid metabolism in PFOA-treated Zebrafish [46]. Moreover, several studies also showed that PFOA-treated mice reduced HDL content, while LDL level was increased [47], 48]. Meanwhile, previous studies observed that fatty acid and cholesterol contents were altered in hepatocytes of PFOS-treated mice [49]. The levels of TG were significantly elevated in PFOS-treated mice, and the expression of hepatic fatty acid translocase was altered on both mRNA and protein levels [50]. However, there are fewer animal studies on the relationship between other PFAS exposures (e.g., PFDA, PFUnDA, PFHxS) and lipid metabolism, and more animal studies should be conducted in the future to confirm the inter-relationships.

Our study also has several limitations should be considered. First, the effect of the PFAS mixture exposure on lipid profile levels was not considered in this study because humans are typically exposed to complex mixtures of environmental contaminants. Second, we did not analyze the data of follow-up for the cohort study due to the times selected for follow-up were not exactly consistent across cohorts. Moreover, not all incorporated studies were adjusted for potentially confounding factors (e.g., smoking habits, and food consumption). Thus, several important findings and complexities of the studies we enrolled in may have been ignored. Also, the small number of studies in certain subgroups limited the reliability of the findings in subgroup analyses.

In the future, to gain a more holistic understanding of the potential health impacts of PFAS exposure, it is urgent for researchers to develop and refine risk assessment models for PFAS exposure, incorporating lipid profiles as key endpoints. This gap involves integrating exposure data, biomarker measurements data, and health outcomes data to create more accurate predictive models. In addition, assessing the effects of PFAS mixtures, as individuals are often exposed to multiple PFAS compounds simultaneously. Thus, studying and developing PFAS mixtures models can provide a more realistic assessment data of real-world exposures. Lastly, conducting mechanistic studies to elucidate the underlying biological pathways through which PFAS may influence lipid profiles which provide more targeted insights and potential intervention points.

Conclusions

In conclusion, the results of this study suggested that PFOS and PFUnDA exposure have a significantly positive correlation with serum LDL concentrations. In addition, our results showed that PFOA, PFNA and PFOS exposure are risk factors for elevated TC levels. These results also provided further strong evidence that PFAS (especially PFOA and PFOS) are potential environmental risk factors for CVD pathogenesis.