Review of cadmium exposure and smoking‐independent effects on atherosclerotic cardiovascular disease in the general population

Exposure to cadmium (Cd) via food and smoking is associated with an increased risk of atherosclerotic cardiovascular disease (ASCVD). Blood and urine levels of Cd are established biomarkers of exposure.


Introduction
Atherosclerotic cardiovascular diseases (ASCVD) encompass coronary heart disease, cerebrovascular ischemic strokes, aortic aneurysm, and lower limb ischemia. Ischemic coronary heart disease and stroke are the top two causes of death and disability in those above 50 years of age in the world population [1]. In affluent countries, the inci-dences of ischemic heart disease and stroke have been dramatically reduced by preventive measures and improved health care, but these diseases are still major causes of death and disability [2].
Atherosclerosis is a vascular disease, located in the subintimal space at predilection sites in the arterial vascular tree with the atherosclerotic plaque as the key lesion. Age is the predominant risk factor, and the majority of men aged over 60 have atherosclerosis in the coronary and carotid arteries, though the prevalence in women is lower [3]. A minority of these plaques will lead to overt clinical disease, mainly by being transformed into vulnerable lesions which may rupture or in some other way cause blood clots with ensuing ischemic damage to the end organ [4]. The major risk factors for atherosclerotic heart disease are hyperlipidaemia, hypertension, tobacco smoking, and diabetes, reflecting influences of heredity, socioeconomic factors, and lifestyle.
In the past decade, exposure to cadmium (Cd) has emerged as a candidate risk factor for cardiovascular disease (CVD), as summarized in several previous reviews [5][6][7][8]. However, it is still unclear to what extent this is independent of tobacco smoking, as the latter is both an important source of Cd and a potent cardiovascular risk factor.
The aim of this review is two fold. In Part I, we review the smoking-independent associations between Cd, measured in blood or urine and ASCVD. We also investigate the possibility of a non-linear dose-response relationship, with Cd exposure showing a threshold effect above which there is an increase in risk. In Part II, we review human, animal, and experimental studies which have investigated causal effects of Cd exposure on different stages of atherosclerotic disease. Finally, we summarize the evidence for Cd as a risk factor for ASCVD.

Cadmium
Cd is a persistent non-essential toxic metal that has adverse health effects at both occupational and environmental exposure levels. The most well-known toxic effects are kidney damage and osteoporosis/osteomalacia [9,10]. In addition, Cd is carcinogenic [10,11] and can cause emphysema, at least at occupational exposure [10].
Human exposure to Cd is ubiquitous, and the main exposure sources are food and tobacco smoke. Cd is present in agricultural soil, both as a natural background and increased by phosphate fertilizers. In non-smoking populations, various crops such as rice, wheat, vegetables, and potatoes usually account for the main part of the intake. In Europe and North America, the average intake of Cd is between 10 and 20 μg/day [10,12]. Intestinal uptake of dietary Cd is about 5%-10% and is typically higher when iron stores are low, since absorption is mediated by divalent metal transporters [10]. Dietary Cd exposure is, therefore, usually higher in menstruating than in non-menstruating people. Tobacco smoking further increases Cd exposure, as Cd in tobacco smoke is effectively absorbed (about 50%) in the lungs. Smoking of 10 cigarettes results in inhalation of 1-2 μg of Cd.
Cd accumulates mainly in the kidneys (about 50%), liver, and muscles. High concentrations are also found in erythrocytes, whilst concentrations in plasma are very low. There is no efficient excretion mechanism of Cd. Only small amounts are excreted in the urine. Elimination is therefore very slow, with a half-life of 10-40 years.
Cd in blood (B-Cd) and/or urine (U-Cd) is widely used for biological monitoring [10,13]. Since Cd excretion in urine is proportional to the Cd in the kidney, U-Cd can be used as a biomarker for the body burden of Cd [14]. B-Cd also reflects the body burden at a steady state but changes faster following altered exposure, for example after smoking cessation. Current smokers have about twice as much Cd in the kidney and urine as neversmokers, and three to four times as much B-Cd [15][16][17].

Part I: Epidemiological studies on cadmium and ASCVD
Four comprehensive systematic reviews and metaanalyses have been published on Cd exposure as a risk factor for CVD in the general population [5][6][7][8]. Some of them also presented separate results for ASCVD, as summarized in Table 1. Another recent review only included US studies [18].
A review of occupational risk factors for CVD [19] included six studies of occupational exposure to Cd [20][21][22][23][24][25]. None of them adjusted the analyses for smoking habits, and so they are not commented on here.
Since one focus in the present review is on doseresponse, we did not include two large studies on dietary exposure [26][27]. The contrast in estimated dietary intake was low, and the partial Pearson correlation coefficient between the dietary Cd estimate and measured U-Cd in a subsample of never-smokers was only 0.1 [28]. Regarding studies based on U-Cd, we only included studies that were able to adjust for diuresis. Therefore, Table 1.

Summary of risk estimates presented in four meta-analyses on cadmium (Cd) exposure and cardiovascular disease (CVD), coronary heart disease (CHD), stroke and peripheral artery disease (PAD) in lower extremities
Author ( a study in Australian women [29] based on very dilute samples not adjusted for creatinine was not considered.

Coronary heart disease
Studies with adjustment for smoking habits. Nine of the studies included in reviews of Cd and CVD in the general population reported associations between Cd exposure and coronary heart disease (CHD). All of them adjusted the risk estimates for smoking habits. Six of the studies were longitudinal [17,[30][31][32], and three were cross-sectional [33][34][35]. In the time since the reviews by Tinkov et al. [8] and Chowdhury et al. [7] were conducted, one cross-sectional study [36] had been published. The studies are summarized in Table 2.
The previous reviews and meta-analyses additionally included studies with a broader disease concept of CVD, including for example hypertension, heart failure or cardiac arrhythmia [37][38][39][40]. CVD was also investigated in a recent study by Domingo-Relloso et al. [41].
Among the four longitudinal studies, the two studies based on the National Health and Nutrition Examination Survey (NHANES) from the United States showed positive associations between CHD and increasing U-Cd [30,31], but in one of them [30] only in men. The Strong Heart study [32] reported a significant positive association between U-Cd and CHD incidence. In the Swedish Malmö Diet and Cancer Study (MDCS) data set [17], a significant positive association was found between B-Cd and acute coronary events.
Three of the cross-sectional studies [33][34][35] found a positive association between blood or urine Cd and CHD, whilst this was not the case in the study by Jeong et al. [36]. Thus, most studies have shown a positive association with the occurrence of CHD, in agreement with the recent meta-analyses [7,8].
Overall, the abovementioned studies provide strong support for associations between Cd exposure and risk of CHD after adjustment for smoking status. However, adjustment for smoking status (never, former, or current smoking) is not perfect.
Within the categories of current smokers and former smokers who had recently given up smoking, those with high consumption of cigarettes could be expected to have both higher levels of B-Cd and U-Cd and a higher risk of CHD than those with low consumption. On the other hand, adjustment for smoking might assign part of the risk of inhaled Cd to smoking. Recent reports using mediation analysis suggest that Cd in cigarette smoke can explain a substantial part of the risk of ASCVD from smoking [42,43].
Studies in never-smokers. One way of avoiding residual confounding from smoking is to perform stratified analyses in never-smokers. This was the chosen method in four of the studies reviewed above using B-Cd or U-Cd [17,32,33,35], and in a recent large Danish study including only neversmokers [44]. In three of the studies, the risk of CHD in never-smokers was increased in the highest exposure categories (Table 2).
Dose-response. Taken together, the doseresponse data based on B-Cd or U-Cd provide no support for an association between Cd exposure and risk of CHD at low-level exposure (B-Cd <0.5 μg/L or U-Cd <0.5 μg/g creatinine (μg/gC)). This was based mainly on the large Danish study in never-smokers [44] and the lack of an association at lower B-Cd categories reported in two studies [17,35].
When comparing somewhat higher levels of B-Cd and U-Cd (B-Cd >0.5 μg/L or U-Cd >0.5 μg/gC) with low-level reference categories, there was strong support for a positive association in all studies [17,[31][32][33]35]. This includes both the smoking-adjusted studies and the studies in never-smokers, with the exception of the study by Menke et al. [30], which showed an association only in men. A limitation is the relatively low numbers of never-smokers in the high exposure categories. The two Korean studies [34,36] reported overall high B-Cd. The study by Lee et al. [34] had a GM of 1.5 μg/L, but very few individuals with B-Cd <0.5 μg/L. The reference category in the study by Jeong et al. [36] included all individuals with B-Cd below the 90th percentile, which was 1.9 μg/L. Therefore, these two studies, which showed conflicting results, cannot be used in the assessment of dose-response.

Stroke
Studies with adjustment for smoking habits. The above-mentioned systematic reviews and metaanalyses on Cd and CVD [5][6][7][8] included six studies on stroke which were adjusted for smoking habits and based on B-Cd or U-Cd. Three of the studies were cross-sectional [34,35,38] and three were longitudinal [17,32,37]. In the time since the reviews by Tinkov et al. [8] and Chowdhury et al. [7] were conducted, one more cross-sectional study [36] had been published. The studies are summarized in Table 3.
Among the longitudinal studies, the Strong Heart study [32] and the Swedish MDCS study [17] found significant positive associations between U-Cd or B-Cd and stroke, whilst this was not the case in the CadmiBel study [37]. However, the latter was very small (21 cases). Three of four cross-sectional studies [35][36]38] found positive associations between B-Cd or U-Cd and stroke, whilst one study [34] found no association.
Considering that the Belgian study was very small, the studies based on B-Cd and U-Cd generally show a positive association with the occurrence of stroke, in agreement with the recent meta-analyses [7][8]. Overall, the above-mentioned studies provide strong support for associations between Cd exposure and risk of stroke after adjustment for smoking status.
Studies in never-smokers. In never-smokers (Table 3), only one of the longitudinal studies [17] showed an association between Cd exposure and stroke, whilst the Strong Heart study [32], and the large Danish study [45] showed no such association. The two cross-sectional studies [35,38] did not provide substantial support for an association. Overall, these studies provide little support for associations between Cd exposure and risk of stroke in never-smokers.

Ischemic stroke versus cerebral haemorrhage.
In studies where ischemic and haemorrhagic stroke were separated [17,45], the risk estimates were very similar for ischemic stroke and total stroke. This is to be expected since a large majority of stroke cases are ischemic.
Dose-response. Taken together, the doseresponse data based on B-Cd or U-Cd provided no support for an association between Cd exposure and risk of stroke at low-level exposure (B-Cd <0.5 μg/L or U-Cd <0.5 μg/gC). This was based mainly on the large Danish study in never-smokers [45] and the lack of an association at lower B-Cd categories in the studies by Barregard et al. [17] and Hecht et al. [35].
When comparing somewhat higher levels of B-Cd and U-Cd (B-Cd >0.5 μg/L or U-Cd >0.5 μg/gC) with low-level reference categories, there was strong support for a positive association in the smoking-adjusted studies [17,32,35,38] and the studies of never-smokers, with the exception of the study by Tellez-Plaza et al. [32]. A limitation is that there were relatively few never-smokers in the high exposure categories. For reasons mentioned in the section on CHD, the two Korean studies [34,36] could not be used in the assessment of doseresponse.

Peripheral artery disease
Studies with adjustment for smoking habits. The review by Tinkov et al. [8] included three studies of peripheral artery disease (PAD) based on the US NHANES 1999-2004 [46], the Strong Heart Study [47], and a study in Swedish women [48]. The findings from NHANES 1999-2004 have also been presented by Zhuang et al. [49] and in part (NHANES 1999(NHANES -2000 by Navas-Acien et al. [50,51]. In addition, a letter by Ujueta et al. [52] reports that in 43 clinical patients with CHD, U-Cd was higher in the 22 patients with PAD than in the 21 patients without PAD.
Overall, the above-mentioned studies, which are summarized in Table 4, provide strong support for an association between Cd exposure and risk of PAD after adjustment for smoking status.
Studies in never-smokers. In never-smokers, only the two US studies [46,47] had enough cases of PAD to allow separate analyses of never-smokers. The odds ratios were above 1, but with very wide confidence intervals. We conclude that the data in never-smokers are insufficient for an assessment of a possible association between Cd exposure and the risk of PAD.

Dose-response. Taken
together, the doseresponse data based on B-Cd or U-Cd provided no support for an association between Cd exposure and risk of PAD at low-level exposure (B-Cd <0.5 μg/L or U-Cd <0.5 μg/gC), given the lack of associations in the lower B-Cd or U-Cd categories in the above-mentioned studies. At levels of B-Cd >0.5 μg/L or U-Cd >0.5 μg/gC, the comparison with lower reference categories showed relatively strong support for associations with the risk of PAD.

Aortic aneurysm
We found only one study examining the association between Cd exposure and aortic aneurysm [53].
In a Swedish case-control study with 297 cases of abdominal aortic aneurysm (AAA) and 594 controls, after adjustment for smoking and other risk factors, the odds ratio (OR) for AAA was 2.5 (95% CI: 1.3-5.0) in the upper tertile of B-Cd (>0.3 μg/L) compared with the first tertile (<0.17 μg/L). There were only 24 cases of AAA in never-smokers and no association with B-Cd. This single study does not provide sufficient information for conclusions on Cd and aortic aneurysm.

Asymptomatic atherosclerosis in carotid or coronary arteries
The review by Tinkov et al. [8] included one study on carotid artery intima-media thickness (cIMT) [54], two studies from Sweden of carotid artery plaque [55][56], and another Swedish study of both plaque and cIMT [57]. After this review was conducted, Lin et al. [58] published a study on cIMT in young (12-30 years) Taiwanese individuals. In addition, the first study of coronary artery atherosclerosis (estimated as coronary artery calcium score [CACS]) was published recently [59].

Studies with adjustment for smoking habits.
Most of the studies with adjustment for smoking habits showed significant associations with carotid artery plaque or cIMT. Since a large cIMT may have explanations other than atherosclerosis [60], we focussed on the studies which measured plaque -the hallmark of atherosclerotic disease. Two of them showed positive associations with B-Cd [55,56] whilst one of them did not [57]. The coronary arteries were only investigated (using CACS) in one study, which showed a significant association with B-Cd in smoking-adjusted analyses [59].
Overall, these studies, which are summarized in Table 5, provide relatively strong support for associations between Cd exposure and risk of atherosclerosis in carotid and coronary arteries after adjustment for smoking status.
Studies in never-smokers. None of the studies of carotid artery plaque showed any association with B-Cd in never-smokers [55][56]. The single study of coronary arteries showed an association between B-Cd and high calcium score in never-smokers, but this was based on relatively few individuals. Therefore, studies in never-smokers provide only limited support for associations between Cd and asymptomatic atherosclerosis.
Dose-response. Taken together, the doseresponse data based on B-Cd or U-Cd provided no support for an association between Cd exposure and risk of carotid or coronary artery atherosclerosis at low-level exposure (B-Cd <0.5 μg/L), based on the lack of associations in the lower B-Cd categories in the abovementioned studies.
At levels of B-Cd >0.5 μg/L or U-Cd >0.5 μg/gC, the comparison with lower reference categories showed relatively strong support for associations with the risk of carotid or coronary artery atherosclerosis ( Table 5).

Chelation of toxic metals
The excretion of metals can be increased by various chelating agents and has been tested long ago as a treatment for lead poisoning. A review by Lamas et al. [61] included an interesting randomized clinical trial amongst patients with a previous myocardial infraction, in which treatment with edetate disodium for about 2 years reduced the incidence of cardiovascular events, including myocardial infarction and stroke, especially in patients with diabetes. This reduction may have been caused by chelation of lead and/or Cd, or by other constituents of the chelator.

Part II: Cadmium and ASCVD: Underlying mechanisms
ASCVD is preceded by the gradual development of atherosclerosis from early to complicated lesions (see Supporting Information: Atherosclerosis, Box 1 Early atherosclerosis, lipid retention for background).

Cadmium in initiation and progression of atherosclerosis
Cadmium toxicity. Cadmium inactivates sulfhydryl groups of enzymes, causes oxidative stress, epigenetic changes, damages to DNA and mitochondria leading to cell death and interferes with calcium signaling (see Supporting Information for details).
Cadmium and hyperlipidaemia. The responseto-retention hypothesis, which conjectures that the key initiating event in atherosclerosis is the retention of cholesterol-rich apoB-containing lipoproteins within the arterial wall, is also linked to hyperlipidaemia as a source of an excess of lipoprotein particles [62]. Data from human studies of general populations and cohorts with high exposures to Cd, in combination with results from animal and experimental studies, support the hypothesis that Cd exposure can contribute to hyperlipidaemia with increased concentrations of triglycerides, total and LDL cholesterol, and decreased HDL cholesterol. Cd has direct and indirect effects on cholesterologenic enzymes and hepatic lipogenesis in animal studies (see Supporting Information and Table S2). The close relationship between smoking och Cd-exposure makes it difficult to clarify the association between higher Cd doses and lipid levels because of residual confounding. But animal studies clearly show how Cd exposure is accompanied by hyperlipidaemia.
Cadmium and lipid retention. ApoB-rich lipoproteins are bound to proteoglycans in the arterial subintimal matrix (see Supporting Information Box 2 for background).
When cultured bovine vascular smooth muscle cells were exposed to Cd chloride at noncytotoxic levels (0.2 μM or less), there was an increased accumulation of the small proteoglycans biglycan and decorin in vascular smooth muscle cells [63]. Moreover, cultured bovine aortic endothelial cells showed increased numbers of heparin sulphate proteoglycan molecules after the addition of Cd chloride at 2.0 μM [64]. In apoB100 transgenic mice, intimal hyperplasia in the carotid arteries was surgically induced and a Cd-containing gel was applied at a non-toxic concentration [65]. In comparison with controls, gene expression analysis of the carotid arteries after 2 weeks showed that Cd treatment increased expression of genes encoding the proteoglycan perlecan and a proteoglycan modifying enzyme. Importantly, apoB staining was significantly increased in carotid arteries from the mice exposed to Cd compared to controls. These results provide direct experimental evidence that Cd promotes subendothelial retention of atherogenic lipoproteins.
Hence, there is clear support for the hypothesis that Cd exposure has proatherosclerotic effects on proteoglycans and retention of LDL cholesterol, but more studies are warranted.

Cadmium and oxidation of LDL cholesterol.
A key driver of inflammation in atherosclerosis is oxi-dized LDL cholesterol and oxidative stress is a key mechanism in Cd toxicity. The question is whether Cd can contribute to oxidative modifications of lipoproteins. Wistar rats exposed for 6 months to Cd in drinking water at doses corresponding to those in human exposure showed dose-dependent increases in oxidized LDL [66]. Concomitant antioxidant treatment with zinc prevented this oxidation. Similar results were obtained in another study of Wistar rats exposed to Cd, resulting in a proatherosclerotic serum lipid profile, reduced levels of antioxidant enzymes, and an increase of malondialdehyde as a measure of lipid peroxidation. Added antioxidant polyphenols attenuated the Cdinduced dyslipidaemia and indicators of oxidative stress [67].
Supporting data also exist in humans. Women with high levels of B-Cd (>5μg/L vs. <5μg/L) showed a clear indication of oxidative stress in terms of increased plasma levels of malondialdehyde [68]. In a population study, U-Cd was associated with the same and other markers of oxidative stress [69]. Hence, these data support an association between Cd exposure and oxidation of LDL cholesterol.
A tentative mechanistic link is paraoxonase 1 (PON1). This is a Ca 2+ -dependent HDL-associated enzyme that protects LDL from oxidation through hydrolysis of lipid peroxides [70][71]. A metaanalysis showed that coronary heart disease is associated with 19% lower serum PON1 activity than in controls [72]. PON1-null mice were more susceptible to atherosclerosis than wild-type littermates and had LDL that was highly susceptible to oxidation [73]. PON1 is inhibited by oxidative stress, and in vitro studies as well as human studies have shown that Cd, as a cause of oxidative stress, is associated with lower PON1 activity [74][75][76].
Cadmium accumulation in arterial wall and effects on the endothelial cell viability. As shown in Table 6, the content of Cd in the human abdominal aorta at upper-middle age and above ranges from 1.78 to 7 μM [77,78]. In symptomatic carotid plaques, Cd levels were 0.34 μM on average and correlated with blood Cd [79]. Thus, B-Cd accumulates in lesion-prone arterial walls and atherosclerotic plaques, reaching concentrations similar to those found to be associated with proatherosclerotic mechanisms in experimental studies [80,81]. Cd is known to injure vascular endothelial cells and alter vascular permeability. One underlying mechanism is that at non-cytotoxic concentrations of 10-100 nM, Cd can inhibit chemotaxis and tube formation in vascular endothelial cells. These effects on repair and angiogenesis seem to be mediated through disruption of vascular endothelial cadherin, a Ca2+-dependent cell adhesion molecule [82,83]. The disruption of vascular endothelial cadherin-cadherin bonds leads to the opening of gaps between endothelial cells, leading to immune cells having direct access to the subintimal space as shown in a mouse model [84]. Cd also induces cell death by several mechanisms, ranging from apoptosis and necrosis to autophagy, which have different causes and pathways such as DNA damage and mitochondrial dysfunction [83]. Apoptosis has been observed in human endothelial cells at a Cd concentration of 61 nM, which is found in humans at high exposure levels [85].
Hence, Cd may cause endothelial damage and cell death, thereby ameliorating the passage of immune cells into the subintimal space and atherosclerotic disease.
Cadmium and endothelial dysfunction. Endothelial dysfunction of the normal functional phenotype is of great importance for the development of atherosclerosis (see Supporting Information Box 3 for background).
A study from Thailand included three groups of women with different exposure levels to Cd (B-Cd geometric means of 1.31, 3.97, and 8.48 μg/L, respectively). Plasma and erythrocyte nitrite concentrations as measures of nitric oxide (NO) production of endothelial cells were found to be reduced in groups with higher Cd exposure, whereas plasma concentrations of asymmetric dimethylarginine, an eNOS inhibitor, were increased. Plasma levels of thrombomodulin were also elevated in the exposed groups, indicating endothelial injury. Markers of lipid and protein oxidation were increased and the antioxidant glutathione was reduced in the Cd-exposed groups. Tobacco smoking as a potential confounder was not considered, but the Cd gradient was far above that caused by smoking [68]. An increase in asymmetric dimethylarginine as a measure of eNOS inhibition has also been reported from an occupationally exposed group in comparison with controls [86].
Studies in rats using oral exposure to Cd, reflecting relatively high human exposure to Cd in heavily contaminated areas and/or in occupational conditions, support that Cd impairs NO-related vascular relaxation. In one study, circulating levels of NO were reduced, whereas lipid peroxidation increased, indicating oxidative stress and reduced NO production by the endothelium [87]. In another study, Cd exposure was associated with both a reduced relaxation of aortic rings after acetylcholine stimulation and a reduction in eNOS expression. The likely explanation is a decrease of muscarinic receptor responses to acetylcholine caused by the interaction of Cd with the thiol groups of muscarinic receptors [88]. In ApoE-/mice, Cd exposure caused reduced NO bioavailability and endothelial dysfunction [89].
These data strongly indicate that Cd exposure causes reduced NO bioavailability and endothelial dysfunction via oxidative stress.
Cadmium, endothelium, and proinflammatory changes. Activated endothelium is characterized by the expression of adhesion molecules and prothrombotic factors. In a study of middleaged women, circulating intercellular adhesion molecule-1 (ICAM-1) was associated with both blood and urine Cd and the occurrence and extent of ultrasound-assessed plaques in the carotid arteries [55]. However, experimental studies of human arterial, endothelial cells do not support a   (Table 7). In the three studies of human arterial endothelial cells there were no, or even down-regulation of proinflammatory genes [48,90,91]. Admittedly, cadmium-induced expression of adhesion molecules in human umbilical vein endothelial cells (HUVECs) [92], but these are venous cells with questionable relevance for atherosclerosis.
On the other hand, studies of rodent and bovine endothelial cells have reported proinflammatory effects of Cd exposure as well as the expression of adhesion molecules [84,[93][94][95]. Conversely, Cd may also inhibit the proinflammatory effect of lipopolysaccharides on endothelial cells [95].
When interpreting these data from the perspective of human atherosclerosis, it is important to keep in mind the extreme diversity of endothelial cells and their functional state in relation to anatomical location and types of blood vessels, and the extensive differences between arteries and veins, as well as large arteries and microvasculature [96].

Cadmium and animal models of atherosclerosis.
Studies of cholesterol-fed rabbits exposed to Cd showed both an increase and a dose-dependent reduction in atherosclerotic lesions. In studies of ApoE knock-out mice, Cd exposure was consistently associated with the progress of atherosclerosis (Table 8). Atherogenic alterations of the circulating lipid profile, oxidative stress, endothelial dysfunction and immunostaining for VCAM-1 and heat-shock protein-1 were also observed [54,82,84].
In summary, Cd causes atherosclerosis in established mouse models of this disease. In human, animal, and experimental studies, Cd exposure causes oxidative stress and endothelial dysfunction. However, none of the studies in human arterial endothelial cells showed that Cd exposure upregulated the expression of proinflammatory genes or adhesion molecules. Hence, the role of Cd in inflammation in the early stage of atherosclerosis is unclear.
Cadmium and monocytes/macrophages. Treatment of human macrophages with low concentrations of Cd (5-200 nM) resulted in a significant Abbreviations: Hsp60 = heat shock protein 60 (a protein inducing autoimmune reactions within the vessel wall), VCAM-1 = vascular cell adhesion molecule 1. † Cadmium was added to the diet. The World Health Organization has established a provisional tolerable weekly intake for cadmium at 7 μg/kg of body weight, hence a 50-fold increase in the lowest daily intake group.
reduction in the levels of several fatty acids, affecting macrophage behaviour and inflammatory state [99]. In human monocytes, Cd at micromolar doses increased MAPK phosphorylation and induced secretion of TNF-α, a prime inducer of inflammation [100]. Conversely, studies of occupationally-exposed young men with B-Cd concentrations in the nanomolar range (but still very high compared with population values in Sweden) showed no increase in TNF-α but decreases in IL-1β and IFN-γ levels, and thus anti-inflammatory effects [101].
One unresolved issue is that Cd exposure induces expression of the metal-binding protein metallothionein in several cell types. Given the high affinity for metallothionein, a large proportion of the Cd content in tissues may not be available [100]. Hence, the effects of Cd on monocyte/macrophage function in atherosclerosis are unclear.
Cadmium and vascular smooth muscle cells. Vascular smooth muscle cells (VSMCs) adopt different phenotypes which participate in different phases of atherosclerosis (see Supporting Information Box 4 for background).
In an experimental study, low levels of Cd (≤100 nM) promoted the proliferation of VSMCs through an intracellular calcium-dependent signalling pathway [102]. This may support a role for Cd in early atherosclerosis with diffuse intimal thickening. Higher Cd levels (1 μM) induce VSMC cell death [80].
The proliferation of arterial smooth muscle cells is dependent on the surrounding extracellular matrix, which includes collagen (types I, III, and IV) and laminin [103]. Interactions of the smooth muscle cells with the underlying matrix regulate cell proliferation. The effects of Cd on DNA synthesis and the proliferation of human aortic smooth muscle cells were studied during culture on fibrillar collagen type I. The results showed that Cd reduced the procollagen synthesis of aortic smooth muscle cells at a concentration of 7 μmol/kg, a concentration that is found in the human aortic media of smokers [77].
Hence, low Cd exposure may promote VSMC proliferation in early atherosclerosis, induce cell death at higher doses, and adversely affect the ability of VSMCs to synthesize collagen, thereby reducing plaque stability.
Cadmium and innate immune response. Normally, the immune system in a concerted action between its innate (see Supporting Information Box 5 for background) and adaptive responses handles infections and other threats to the organism. Atherosclerosis is one example of an inappropriate and uncontrolled response that leads to chronic inflammation and increased tissue damage. IL-6, as a link in the NLRP3 inflammasome, and part of the innate immune system has been shown to be a causal factor in ASCVD.
Cd exposure may affect different aspects of the rapid and unspecific In general, available data as described above suggest that Cd in micromolar concentrations causes upregulation of the mediators and markers of inflammation, and appears to have proinflammatory properties [81]. However, the data are not consistent. Cd has both pro-inflammatory and anti-inflammatory effects due to dose and exposure duration. More importantly, the effects of Cd may differ not only between species but also between organs and cell types [81,104,105]. Most studies have used exposure levels of Cd clearly above those observed in humans [81].
Another approach is to examine if Cd exposure is associated with C-reactive protein (CRP), which is the end product of the NLRP3 inflammasome corresponding to the interleukin (IL)-1β, IL-18, IL-6, and CRP signalling pathway. In three cohorts from the NHANES project and a large cohort of middleaged Swedish men and women, U-Cd and B-Cd were both associated with CRP after adjustment for smoking and other confounders [106][107][108][109]. The lat-ter study also examined never-smokers and found no relationship between B-Cd and CRP, probably indicating that Cd exposure has no effect until it reaches levels that are generally only seen in smokers. In addition, if there is little support for Cd effects on the arterial wall as the origin of this inflammatory signal, the stronger the findings are that Cd exposure causes human hepatocytes to synthesize the TNF-α and IL-1β which play an important role in the onset of the CRP signalling pathway [81,110].
Hence, at Cd exposure levels similar to those in humans, there are changes in macrophage fatty acid distributions compatible with changes in immune-modulatory functions. However, it has been difficult to translate these findings of basic mechanisms into consistent observations of proinflammatory effects of Cd on the atherosclerotic process.
Cadmium and adaptive immune response. The adaptive immune response is important in plaque formation, lesion stability and rupture (see Supporting Information Box 8 for background). Cd is taken up by human immune cells, mostly by T-and B-lymphocytes, and to a lesser extent by peripheral monocytes [111].
Cd exposure has been shown to increase circulating oxidized LDL cholesterol in rats [66] and to upregulate heat shock protein expression of endothelial cells from humans and mice [84,85]. A majority of other studies of adaptive immunity focussing on the splenocytes and thymus suggest that Cd has an immunosuppressive effect, but some studies report that Cd exhibits immunestimulation features [104].
Hence, although several studies have shown that Cd exposure results in oxidative stress, oxidized LDL, and heat shock proteins which have antigenic properties, there is a lack of data on how Cd affects the response from the adaptive immune system.

Cadmium and prostaglandins.
Prostaglandins have important roles in atherosclerosis (see Supporting Information Box 7 for background). As summarized by Olszowski et al., several studies have investigated the effect of Cd on COX-2 mRNA protein expression and enzymatic activity [112].
Most of them demonstrated a stimulatory effect of this metal on COX-2 in different experimental models. However, a few reports suggested Cd exerting either inhibitory action or no effect on COX-2. Their own study of human macrophages exposed to 5 nM, 20 nM, 200 nM and 2 μM of Cd for 48 h showed that Cd at the highest tested concentrations modulated COX-1 and COX-2 only at the mRNA level [112]. However, the lower tested Cd concentrations appeared to inhibit COX-1 protein expression. No convincing data of Cd effects on prostaglandins have been found.
Hence, Cd exposure seems to affect COX-2 expression, but there is a lack of consistent data regarding effects on the atherosclerotic disease process.
Cadmium, coagulation and fibrinolysis. Cd exposure has been shown to induce von Willebrand factor (vWF) expression in vascular endothelial cells in mouse lung and kidney tissues (See Supporting Information Box 10 for background). In vitro analysis showed that 1 μM Cd specifically upregulated vWF mRNA and protein expression in HUVECs, indicating that Cd targets vascular endothelial cells even at relatively low concentrations. Since vWF is a key regulator for vascular homeostasis, this may be one mechanism by which Cd can promote atherosclerotic diseases [113].
Fibrinogen is also an acute-phase protein. As such, fibrinogen biosynthesis in hepatocytes is regulated by glucocorticoids and IL-6. The latter, which is produced by fibroblasts, T-cells, endothelial cells and monocytic cells, stimulates basal levels of fibrinogen production [114]. Cd exposure increases the expression of IL-6 in many cell types [81]. There is some epidemiological indication that plasma fibrinogen levels are increased dose-dependently by Cd exposure in the general population [106].
HUVECs, human fibroblasts, and smooth muscle aortic cells were incubated in cadmium chloride (0.5, 1, or 2 μM) for 24 h. The results showed that Cd induced plasminogen activator inhibitor type 1 (PAI-1) synthesis and activity in endothelial cells without affecting tissue plasminogen activator (t-PA). Thereby Cd reduced the fibrinolytic activity of t-PA and urinary-PA in vascular endothelial cells [115][116][117].
Hence, there are several indications that Cd exposure has prothrombotic and anti-fibrinolytic effects which may be important in the later phases of atherosclerotic disease.
Smoking, cadmium and atherosclerosis. The concept that Cd, as a principal toxicant in tobacco smoke, may partially be responsible for the proathersclerotic effect of smoking, has growing support. Mediation analysis indicated that in smokers, 60% of the association between current smoking and prevalence of carotid plaques was mediated through smoking, whereas about half of the increased risk of incident coronary heart disease in current smokers was mediated via cadmium [42][43]. Data on underlying causes are lacking. But the principle of mediating causes has been exemplified in a recent DNA methylation study in the Strong Heart study. Mediation analysis supported a biological link for Cd and smoking-associated health effects, including the possibility that Cd is partly responsible for smoking toxicity through epigenetic changes [69].

Complicated lesions with plaque rupture
The final stage of the atherosclerotic disease process is when the plaque transforms into a complicated atherosclerotic lesion, with either rupture or erosion, leading to thrombosis and disrupted blood flow (see Supporting Information Box 9 for background). There is a scarcity of data on how Cd exposure may be directly involved in this final pathophysiological process. However, there are some clues. In a large population-based prospective study, proteomic analysis of plasma was performed in never-smokers in order to find proteins associated with B-Cd. Four proteins were identified, validated and found to predict ischemic stroke and/or CHD [118].
The first of these proteins was the urokinase plasminogen activator receptor (uPAR), which is mainly expressed in immune cells. uPA is a protease that by binding to uPAR plays a key role in localized proteolysis and tissue remodelling in diseases such as cancer or atherosclerosis [119]. uPAR is involved in angiogenesis, adhesion, cell migration, proliferation, cell survival, inflammation, and proteolysis. Soluble uPAR has been identified as a biomarker of prevalent and incident atherosclerotic diseases [120][121][122][123].
Studies of animal and human plaques showed that uPAR was strongly associated with severe atherosclerosis [124][125]. uPAR and macrophages accumulated within symptomatic plaques and were co-localized in the unstable parts of the plaque where rupture usually occurs [126][127].
Moreover, B-Cd was correlated with Cd levels and macrophage density in the most unstable parts of such carotid plaques [79,128], and the occurrence of uPAR and the plasminogen receptor S100A10 were correlated in the same parts of symptomatic carotid plaques [129].
These observational data raise the hypothesis that Cd is involved with the uPA/uPAR/ S100A10/annexin A2 complex in plaque tissue degradation and macrophage migration, and may participate in plaque rupture mechanisms. One possibility is that Cd induces uPAR expression, as has been found in human gastric cancer cells mediated by the ERK-1/2, NF-κB and activator protein-1 signalling pathways [130].
The three other proteins were matrix metalloproteinase-12 (MMP-12), cathepsin L (a lysosomal cysteine protease) and CX3CL1 (fractalkine), a chemokine [118]. All three promote atherosclerosis in mice models, are found in vulnerable and symptomatic plaques in connection with macrophages, and have soluble forms that are associated with prevalent or incident CHD (for detailed information, see Supporting Information). Hypothetically, these proteins may be a link between Cd exposure and ASCVD.

Complicated lesions with plaque erosion
About one-third of cases with myocardial infarctions seem to be caused by thrombi overlying intact, non-ruptured atherosclerotic plaques [131]. The underlying pathology is the erosion of luminal endothelial cells from smooth muscle and proteoglycan-rich atheromas. In contrast to plaque ruptures, endothelial erosions tend to occur on thick-capped atherosclerotic plaques with small deep-seated lipid cores and do not have to be associated with inflammation. Eroded plaques are less calcified than ruptured plaques. Risk factors for erosion are smoking and to be a premenopausal woman. Such endothelial erosion may be caused by many mechanisms. Apoptosis of endothelial cells and loss of endothelial contacts with the underlying extracellular matrix are two intimately linked processes that are believed to be important in erosion. Factors promoting endothelial apoptosis include deprivation of growth or survival factors, such as vascular endothelial growth factor, or disruption of the cell to cell contacts mediated by VE-cadherin, which interferes with signalling through the mitogen-activated protein kinase and c-Akt pathways [132]. The previously mentioned chemokine fractalkine is also known to induce endothelial cell injury [133].
The two intimately linked processes believed to be important in plaque erosion, apoptosis of endothelial cells and loss of endothelial contacts with the underlying extracellular matrix are the wellestablished effects of Cd exposure in experimental and animal studies, as summarized above [82][83][84]. Female sex and smoking seem to be risk factors for plaque erosion, and these factors are also associated with high Cd exposure in the general population [10]. Fractalkine, a chemokine that is associated with Cd exposure, is known to induce endothelial cell damage and to be expressed in endothelial cells in atheromatous lesions [134].

Conclusions
The results from the mechanistic studies can be merged into a hypothetical summary of the proatherosclerotic effects of Cd exposure (Fig. 1). The figure legend gives a more detailed overview of the chain of events over time in the development of atherosclerosis.

Discussion
Four previous meta-analyses have convincingly shown that Cd exposure is associated with CVD [5][6][7][8]. The present updated review adds a doseresponse analysis. As illustrated in Fig. 2, the available evidence indicates that Cd confers an increased risk of ASCVD above exposure levels of B-Cd >0.5 μg/L or U-Cd >0.5 μg/gC. At lower exposure levels, there is at present no evidence of increased risk. In addition, there are clear indications that Cd exposure is also associated with asymptomatic atherosclerosis in the carotid and coronary arteries above this threshold.
Smoking is an important confounder, as it is both an important source of high Cd exposure and a potent risk factor. Even if adjustment for smoking does not exclude residual confounding, several studies in never-smoking cohorts have shown associations between Cd and ASCVD, and experimental studies have demonstrated pro-atherosclerotic effects of Cd. Hence, there is strong evidence that Cd causes ASCVD.
A further finding is that that Cd accumulates in atheroprone arterial walls, reaching concentrations corresponding to those in experimental  (1) to better bind and retain LDL particles, which also become oxidized (2). At higher exposure levels, Cd also affects the liver and increases the concentration of plasma LDL cholesterol (3). As a further effect of Cd, the endothelium becomes dysfunctional, with impaired nitric oxide (NO)-related vasodilation (4), later followed by disruption of endothelial cadherincadherin bonds, which opens gaps between endothelial cells (ECs) (5). This increases the permeability and allows monocytes (5) to pass into the subintimal space, turn into macrophages, engulf LDL lipoproteins bound to proteoglycans, and differentiate to foam cells. At low concentrations, Cd promotes the proliferation of vascular smooth muscle cells (VSMCs) (6), which migrate from the media to the plaque area and strengthen the fibrous cap or turn into foam cells. At higher concentrations, Cd causes the death of not only ECs but also VSMCs (7). Increased apoptosis and insufficient removal of apoptotic cells cause expansion of a necrotic core. The synthesis of collagen is also disrupted by Cd (8). Taken, together this contributes to plaque rupture (7). In addition, Cd exposure seems to be associated with increases in urokinase plasminogen activator receptor (uPAR), matrix metalloproteinases-12 , and cathepsin L synthesis (9), which are involved in the degradation of the fibrous cap. The other type of complicated lesion is plaque erosion (7), with damage to the endothelium of the kind that is seen after Cd exposure (7) which may also include effects of fractalkine. Finally, the pro-thrombotic process is stimulated by the Cd-associated increase in von Willebrand factor (vWF) and plasminogen activator inhibitor type 1 (PAI-1) expression and production (10,11), resulting in an increased risk of a clinical event.
studies that have shown Cd to have proatherosclerotic effects. Considering the well-known cytotoxic effect of Cd, it is no surprise that Cd has dysregulatory effects on most vascular tissues: endothelial cells, smooth muscle cells, immune cells, foam cells, and collagen. Data indicate that Cd is operative both in early atherosclerosis when plaques develop and at late stages when plaques become complicated with rupture or erosion. Furthermore, Cd probably promotes thrombosis and anti-fibrinolysis in the final stage of overt ASCVD. However, as many mechanisms are hypothetical  Table S2. and data are still scarce, more studies are required to clarify the details. A still unresolved question is how Cd promotes intraplaque inflammation beyond the capacity to oxidize LDL cholesterol, the prime cause of inflammation in atherosclerosis.
Is Cd a novel risk factor for ASCVD? A pragmatic view is that Cd exposure is a proatherosclerotic causal factor for ASCVD. In never-smokers the major source is diet. In smokers, it seems increasingly plausible that Cd partly mediates the risk of smoking on ASCVD.