Abstract
Background:
Ambient air pollution has been associated with Alzheimer’s disease (AD) in the elderly. However, its effects on AD have not been meta-analyzed comprehensively.
Objective:
We conducted a systematic review and meta-analysis to assess the associations between air pollution and AD incidence.
Methods:
We searched PubMed and Web of Science for indexed publications up to March 2020. Odds risk (OR) and confidence intervals (CI) were estimated for particulate matter (PM)10 (PM10), PM2.5, ozone (O3), nitrogen dioxide (NO2), sulfur dioxide (SO2), and carbon monoxide (CO). The subgroup analysis was conducted based on the pollution levels.
Results:
Nine studies were included in the meta-analysis and review. The OR per 10μg/m3 increase of PM2.5 was 1.95 (95% CI: 0.88–4.30). The corresponding values per 10μg/m3 increment of other pollutants were 1.03 (95% CI: 0.68–1.57) for O3, 1.00 (95% CI: 0.89–1.13) for NO2, and 0.95 (95% CI: 0.91–0.99) for PM10 (only one study), respectively. Overall OR of the five air pollutants above with AD was 1.32 (95% CI: 1.09–1.61), suggesting a positive association between ambient air pollution and AD incidence. The sub-analysis indicated that the OR (2.20) in heavily polluted regions was notably higher than that in lightly polluted regions (1.06). Although AD risk rate data related to SO2 or CO exposure are still limited, the epidemiologic and toxicological evidence indicated that higher concentration of SO2 or CO exposure increased risks of dementia, implying that SO2 or CO might have a potential impact on AD.
Conclusion:
Air pollution exposure may exacerbate AD development.
INTRODUCTION
Alzheimer’s disease (AD), contributing to more than 60% of all cases of dementia, has been one of the most prevalent neurodegenerative disorders. It is characterized by cognitive decline, accumulation of amyloid-β peptide in the brain, hyperphosphorylation of tau protein, and formation of plaques and neurofibrillary tangles.
In 2019, nearly 50 million people had AD worldwide [1], and AD cases are predicted to increase to 135 million people by 2050 [2]. The annual socioeconomic cost per AD patient was 19,144.36 dollars, and total expenses were 167.74 billion dollars in 2015. The annual total costs are predicted to reach 1.89 trillion dollars in 2050 [3]. Based on the statistical report of the World Health Organization (WHO), AD appeared more threatening in developed countries and became one of the major causes of death when the mortality related to AD presented among the top ten [4]. Therefore, AD has become a serious social and public health problem in the world now.
The main risk factors for the development of AD are age, diet, genetics, and environmental factors [5–8]. Of note, epidemiologic studies revealed the etiologic role of air pollution exposures [9–11].
Aging is a widespread public health problem worldwide. It is a crucial risk factor for AD [6]. The consequences of aging associated with adverse changes in neuropathological features and cognition may lead to AD [12].
A number of components are found in air pollution, mainly including particulate pollutant (PM) such as PM10 (aerodynamic diameter≤10μm) and PM2.5 (diameter≤2.5μm), and gaseous pollutants such as ozone (O3), nitrogen dioxide (NO2), sulfur dioxide (SO2), and carbon monoxide (CO). Pollutant sources can be complex. For PM, NO2, SO2, and CO, the pollution sources are majorly from the direct emission of fossil fuel burning, while O3 is mainly from secondary pollutant emissions.
Air pollution has been linked to human health and has caught worldwide attention. According to the 2019 State of Global Air Report, it caused the most death among all risk factors worldwide. It was only second to diet, hypertension, smoking, and high fasting blood glucose [13]. On a global scale, air pollution can diminish life expectancy for nearly two years [13]. With the development of the economy and society, air pollution has been notably lessened in developed countries while it still contributed to approximately five million deaths in 2017, especially in developing countries [13, 14].
The increased age-standardized mortality rates associated with air pollution was found stronger in Asia and Africa than other continents [14]. In 2017, the air pollution-induced global attributable deaths in China and India both exceeded one million cases [13].
The consensus that air pollution is strongly associated with harming human health has been widely built in many research areas. The morbidity and mortality link to air pollution has a wide range at a time, especially in some severe diseases such as cardiovascular and respiratory diseases, cancer, and stroke [14, 15].
Different components of air pollution may come from various sources. For example, PM mainly comes from coal- and fossil- and biomass burning, industry, and traffic. Among all types of air pollutions, PM was considered to be the most threatening, leading to respiratory diseases, cardiovascular diseases, and even neurological disorders [16–18]. In 2017, the Health Effects Institute found PM2.5 itself contributed to nearly 3 million deaths, which could be due to cardiovascular diseases [13, 14]. A time-series based meta-analysis with mortality in 652 cities worldwide exposed to air pollution reported that the short-term exposure of PM10 and PM2.5 could visibly increase the risks of death when PM showed a statistical significance relating to overall mortality, and mortality of cardiovascular and respiratory diseases [19].
Compared to PM, the major sources of O3 is the precursor substances produced by human activities. It contains nitrogen oxide, and compounds containing carbon and volatile organic compounds produced by plants. The photochemical reactions under UV light from sunshine produces O3. O3 could be the cause of cardiovascular and respiratory diseases [16,20, 16,20]. NO2 primarily comes from coal-fired power plants and motor vehicle exhaust. SO2 was majorly produced by coal and another fossil burning, and the emission from the metallurgical industry. O3, NO2, and SO2 can stimulate the eyes and respiratory tract. Besides, CO mainly comes from incomplete coal burning. A high concentration of CO in closed environments may induce dyspnea, asphyxia, and even death. NO2, SO2, and CO are considered to be closely linked to adverse cardiopulmonary events [20, 21].
In 2017, the population living in the area under the WHO guideline for PM2.5 (≤10μg/m3) was less than 8% [13]. With the severe air pollution situation, people paid more and more attention to diseases related to air pollutants. Despite cardiovascular and respiratory diseases, it is necessary to look inward at the associations between air pollution and other health hazards and inner mechanisms. For example, a recent study found there were strong associations between PM2.5 and multiple neurological disorders [18]. In the pathological research of air pollution-induced brain injury, some toxicological studies revealed that air pollutants (such as PM, O3, NO2, SO2, and CO) might cause brain injury and neurotoxicity [22–25] through regulation roles related to mitochondrial damage, oxidative stress, and inflammation, which are involved in AD pathology [26, 27]. However, the evidence of air pollutant-induced AD from epidemiologic studies is still limited.
Many previous studies reported their research on the relationship between PM and AD. Jung et al. (2015) demonstrated that the odds ratio (OR) for PM2.5 and AD was 2.38 (95% CI: 2.21–2.56) based on a cohort study, while Wu et al. (2015) found that for PM10 and AD, it was 4.17 (95% CI: 2.31–7.54) based on a case-control study [9,28, 9,28]. Tsai et al. (2019), through a meta-analysis by estimating three related studies, revealed the pooled hazard ratio (HR) between PM2.5 and AD was 4.82 (95% CI: 2.28–7.36) [29]. Up to now, people have still made in-depth studies for providing a new basis for comprehensively revealing the correlations between PM2.5 and AD.
Likewise, PM having AD as the outcome, Jung et al. (2015) also reported that the risks for the elderly would increase 212% when atmospheric O3 was increasing per 10.91 ppb [9]. For NO2, the onset risks of the elderly would rise to 138% when the atmospheric NO2 was increasing per 10μg/m3 [30]. It remains unclear if the effects of SO2 and CO on AD with a risk estimate.
Meta-analysis is a practical assessment method of estimating the risks of specific impact factors on public health. This study aims to comprehensively provide evidence on the associations of multiple air pollutions and AD via a systematic review and meta-analysis. It would contribute to the publication of health guidebooks and policymaking in the meantime.
MATERIALS AND METHODS
Data sources and search strategy
We conducted a systematic review and meta-analysis of studies indexed in PubMed and Web of Science and published until March 18, 2020, that reported on the chronic effects of ambient air pollutants on AD incidence. Keywords for searching were set by combining the name of air pollutant and AD, such as “Air pollution, Alzheimer’s disease”, “PM2.5, Alzheimer’s disease”, “PM10, Alzheimer’s disease”, “particulate matter, Alzheimer’s disease”, “O3, Alzheimer’s disease”, “Ozone, Alzheimer’s disease”, “SO2, Alzheimer’s disease”, “Sulfur dioxide, Alzheimer’s disease”, “nitrogen dioxide, Alzheimer’s disease”, “NO2, Alzheimer’s disease”, “Carbon monoxide, Alzheimer’s disease” (the detailed search strategy can be found in the Supplementary Material). The articles were screened and excluded if they 1) did not publish in English; 2) did not have AD as an endpoint; 3) did not contain the concentration of target air pollutant; 4) did not conclude its results with a calculated risk estimate such as OR, relative risk (RR), or HR in the abstract. The full text and the reference list of the screened articles were further reviewed to find any additional available data.
Risks of incidence were extracted and stratified by AD and duration of air pollutant exposure. For uniformity, all risk estimates were converted to the standard unit of “per 10μg/m3 increase in PM2.5, PM10, O3, or NO2 concentrations” (the conversion can be found in the Supplementary Material).
Data analysis
Data from each extracted article were stratified and recorded with a standard order, including the name of the study, authors, publication year, risk estimates, location, percentage of gender, mean age, population, study design (time series, case-crossover analyses, etc.), study period, the local concentration of targeted air pollutant, and adjust models. We kept the age range if a mean age was not be provided and left it blank for other lack of information.
A score list was performed to evaluate the quality of each included article, which was modified from the original version of the Agency for Healthcare Research and Quality and followed by our previous study [18, 31]. The full score for each publication was ten, and it will be deducted one point if it did not provide the required information. The detailed score list can be found in the Supplementary Material.
Subgroup analysis was processed based on the concentration level of the pollutants. According to the WHO Air Quality Guidelines [32], the average annual concentrations for Interim Target (IT)-2 of PM2.5 and PM10 were 25 and 50μg/m3, respectively, when that of NO2 was 40μg/m3. The guideline for 8 h-concentration of O3 was 100μg/m3. For each of pollutant within the limitation, it was selected as lightly polluted countries (or regions) while it was considered as heavily polluted countries (or regions) if exceeding the guideline limit. We analyzed the risk estimated for lightly- and heavily-polluted countries of the pollutants including PM2.5, PM10, O3, and NO2 to evaluate the risk effects depending on polluted levels. Although the majority of the included studies provided a positive association, only the study by Cerza et al. (2019) provided data with all negative results for each category of air pollutants including PM2.5, PM10, O3, and NO2. Thus, we repeated the meta-analysis by eliminating Cerza et al. (2019) to investigate its influence on the results [33].
All meta-analysis was conducted by the software STATA 14.0 (STATA Corp). Sensitivity analysis was performed by the meta-inf functions to evaluate the influence of each publication by removing it from the meta-analysis. The I-squared value was used for the heterogeneity analysis. When the I-squared value was more than 50%, we chose a random model; otherwise, a fixed model was used [34]. We managed the Egger’s test to evaluate the publication bias and performed meta-regression to discuss the potential heterogeneity source.
RESULTS
Our search identified 1,423 studies (563 from PubMed and 860 from Web of Science). After excluding irrelevant studies, we identified 9 studies for use in our meta-analysis and review [9–11, 35–37].
Characteristics from the 9 articles selected to study the relationships between air pollution and AD are shown in Table 1. The table describes the information about the author, year, location, study period, study population, study design, sexuality (ratio of male to female number), age, local concentration, relationships (HR, RR, OR), and adjusted factors of models.
Characteristics from 9 articles selected to study the relationships between air pollution and AD for meta-analysis and review
Table 1 shows that the time range in this study is from 1993 to 2013. The population size exceeded ten million distributed in the United States (US), the United Kingdom (UK), Canada, Spain, Italy, Sweden, and Taiwan. The age range in this study was from 30 to 85 years of age. The study design included five cohort studies, two longitudinal studies, one ecological time-series study, and one case-control study. The included articles were high quality, with 8.89 average scores out of 10.
As Fig. 2 shows, there were six articles conducted to meta-analyze for PM2.5, four for NO2, three for O3, and one for PM10. One Taiwan study was excluded for meta-analysis as its HRs were unconvertable to the standard unit of per 10μg/m3 even after communication with the author [11]. Two studies assessed the association of PM10 with AD [28, 33]. Wu et al. (2015) found that PM10 (≥49.23 mg/m3) had strong associations with AD (OR = 4.17) when comparing the highest to lowest PM10 tertile [28]. However, Cerza et al. (2019) revealed a negative association between PM10 and AD (OR = 0.95), while it converted to a positive association when estimating the risks of PM10 on dementia [33]. Only one data [33] was to undertake a pooled or meta-analysis of the association between exposure to PM10 and AD in this study. Three studies were synthesized for assessing the association of O3 with AD, while four studies were assessed for the relationship between NO2 and AD. We did not include one study from Taiwan reported by Li et al. (2019) into our meta-analysis since the OR from that study could not be reliably converted to units of per 10μg/m3 PM2.5 [11]. A study showed that the highest tertile of O3 (≥21.56 ppb) exposure was associated with increased AD risk (highest versus lowest tertile of O3: adjusted OR = 2.00) [28], but this study was excluded as OR value was not able to be converted for calculation of per 10μg/m3 O3.
Flow diagram.
Meta-analysis of the association between air pollutions and AD.
No article was used for CO due to a lack of related studies between CO and AD with a calculatable risk estimate. However, many studies discussed the relationship between CO and dementia (Table 2), indicating that exposure to CO was associated with a higher risk of dementia [38–42]. For example, Chang et al. (2014) found that long-term exposure to CO was associated with a 61% risk of increase of dementia in the Taiwanese population via a retrospective cohort study [38].
Key findings and results of CO related to dementia
There was no calculable data for SO2 affecting AD for inclusion in the meta-analysis. Nevertheless, one study stated that the adjusted OR of vascular dementia in association with SO2 exposures in 5 years before vascular dementia diagnosis was 0.95 (95% CI: 0.83–1.08) per 1-ppb SO2 increase [42].
As the I-squared value for meta- and subgroup analysis exceeded 50%, random models were used for each analysis. The risk estimate for all included pollutant types was 1.32 (95% CI: 1.09–1.61) (Fig. 2). The risk estimates for each type of pollutants, namely PM2.5, PM10, O3, and NO2 were 1.95 (95% CI: 0.88–4.30), 0.95 (95% CI: 0.91–0.99), 1.03 (95% CI: 0.68–1.57), and 1.00 (95% CI: 0.89–1.13), respectively (Fig. 2). Specifically, the result for NO2 (1.00, 95% CI: 0.89–1.13) displayed a stronger association (1.06, 95 % CI: 0.89–1.26) when eliminating the data from Cerza et al. (2019) (Supplementary Figure 1) [33].
In the subgroup analysis, according to the pollution concentration levels, the overall risk estimate remains 1.32 (95% CI: 1.09–1.61) (Fig. 3). The risk estimates of heavily polluted countries were 2.20 (95% CI: 1.01–4.79), while that of lightly polluted countries was 1.06 (95% CI: 0.96–1.16).
Meta-analysis of the association between air pollutions and AD in lightly and heavily polluted countries.
The results were robust as the sensitivity analysis did not find a significant difference by eliminating every included study individually. The Egger’s test suggested that there might be potential publication bias. The meta-regression examined the type of pollutants, study designs, population size, or age range. We found that the pollutant types might be the source of heterogeneity, while the other three may not be.
DISCUSSION
Mechanisms of the association between PM and AD
In the present study, we included over 10 million adult subjects aged >30 years from six studies about PM2.5 in the US, UK, Spain, Canada, Italy, and Taiwan for meta-analysis, shown in Table 1. The results of our meta-analysis suggested that exposure to a 10μg/m3 increase in PM2.5 was significantly and positively associated with dementia (pooled HR = 1.95 95% CI: 0.88–4.30). As 60–70% dementia was AD, the existing evidence confirmed that exposure of PM2.5 is a potential determinant of AD. Additionally, the two studies of the association of PM10 and AD exhibited opposite results [28, 33]. The PM10 concentrations during the experiment period in Rome and Taiwan were 36.9μg/m3 and 47.06μg/m3, respectively, between IT-2 (30μg/m3) and IT-3(50μg/m3) [32]. The risk of AD declined in Rome while it increased in Taiwan [28, 33]. One reason for such opposite results may because of the different levels of concentration. Thus, the association needs to be investigated further.
Studies in past decades revealed that the major mechanisms of PM-induced neurodegenerative disorders referred to oxidative stress, neuroinflammation, disorders of metal homeostasis and proteostasis, and compromised blood-brain barrier [43]. PM has two major ways of inducing nervous system injury. First, it can pass the olfactory system crossing the olfactory bulb or lung-blood-brain barrier getting into the brain [44, 45]. The inflammatory response and oxidative stress induced by PM could lead to amyloid-β (Aβ) accumulation, tau protein phosphorylation, and synaptic dysfunctions and finally contribute to AD formation [46, 47]. Second, when entering of the lungs, PM can stimulate the whole lung system and induce local and systemic inflammatory response syndrome. Inflammatory mediators and cytokines produced during the process may get into the brain via the blood circulation system, and cause nervous system injury [48–51].
Our study found that the risks of AD had a 95% increase when per 10μg/m3 PM2.5 raised, suggesting that the association between PM2.5 and AD was strong. PM with diameter ≤10μm normally situated in the upper respiratory tract, while PM with diameter ≤2.5μm can get deep into the respiratory tract and deposit in the lungs. Tinier PM could even get deep down into the bronchiole and pulmonary alveolus. As inhalable particles, PM2.5 has more toxicity than PM10. The larger specific surface area of PM2.5 provides more possibility of absorbing toxic contaminants and as its tiny inhalable particulate can affect not only lung but other important organs such as the brain [52]. More seriously, the contamination of PM2.5 has spread global health problems, according to the WHO guideline; only 8% of populations worldwide lived in the safe level of PM2.5 (≤10μg/m3) in 2017. 54%, 67%, and 82% of populations lived in the areas exceeding IT-1 to IT-3 (35μg/m3, 25μg/m3, 15μg/m3) [13]. As a result, PM2.5 caught more attention than PM10 due to its more harmful characteristics.
The average PM2.5 concentration in the six meta-analyzed studies was from 4.1 to 34.4μg/m3. In Canada with the 4.1μg/m3 of PM2.5 mean concentration, less than the IT-3 of PM2.5 (<15μg/m3) [32], it was found that the risk for PM2.5-related AD declined when the OR = 0.90 for per 1.54μg/m3 of PM2.5 concentration increase. However, the result in Taiwan (34.4μg/m3), which was almost IT-1 (<35μg/m3), has OR = 2.38 for per 4.34μg/m3 of PM2.5 concentration increase. It clearly reflected that the higher risks of PM2.5-induced AD would happen with higher PM2.5 concentration.
Interestingly, from our result, the risks of PM2.5 on AD had a nonnegligible decline since 2019. With per 5μg/m3 PM2.5 increase, the OR of AD declined 9%, when another indicated that with per 1.54μg/m3 PM2.5 increase, the OR of AD dropped 10% [33, 37]. We speculated it might result from the low PM2.5 concentrations in these two countries (regions) (4.1 and 19.7μg/m3) and benign health care services. Simultaneously, it could be an illustration evidencing that the significance of improving air quality and health care services for enhancing public health.
Mechanisms of the association between O3 and AD
Jung et al. (2015) pointed that per 10.91 ppb increased in the concentration of O3, one of the toxic atmosphere gases, can raise the risks of AD for 212% on the elderly [9]. Wu et al. (2015) discovered that elderly with the highest tertile (>21.56 ppb) O3 has double risks to get AD compared to those with lower tertile (<20.20 ppb) [28]. On the contrary, Cerza et al. (2019) and Carey et al. (2018) indicated that O3 exposure was negatively related to AD [33, 35]. Combining and meta-analyzing these three reports, our result suggested that O3 was one of the significant factors inducing AD with an increased risk (1.03, 95% CI: 0.68–1.57) for per 10μg/m3 elevation in O3.
There are two possible mechanisms for O3 inducing AD. Firstly, O3 is a powerful oxidizing molecule and is soluble in water. When inhaled, O3 could transfer from the respiratory system to the brain via the bloodstream [53]. Besides, the nose and olfactory pathway is another portal of entry into the brain [22]. Secondly, O3 inhalation can produce reactive oxygen species (ROS) and cause oxidative stress, inflammation, consecutive neuronal dysfunction, and cell apoptosis in the brain [54]. Due to the oxidative stress, O3 could lead to the degeneration of hippocampal neurons and the deduction of brain recovery ability [22, 55]. Additionally, the amyloidogenic pathway may be activated by O3 and cause Aβ42 overproduction and mitochondrial dysfunction in the brain [56, 57].
Another important study provided more mechanisms explaining O3 itself may not induce AD directly, yet it can intensify the AD pathophysiology in genetically predisposed populations by synergizing with genetic risk factors [58]. This is of concern and merits further investigation of the exact association of O3 with AD incidence and the toxicological mechanisms.
Mechanisms of the association between NO2 and AD
The result of this study showed that exposure to a 10μg/m3 increase in NO2 had no significant association with AD (OR = 1.00, 95 % CI: 0.89–1.13). Although the association in the result was not obvious, some other studies stated positive results. Oudin et al. (2016) discovered that with per 10μg/m3 increased concentration of O3 could raise the 5% risk for the elderly [30]. Besides, Carey et al. (2018) also found that such risks increase for the elderly would be up to 23% when the concentration of O3 raised for per 7.5μg/m3 [35].
It is worthy to notice Cerza et al. (2019) reported all AD risks decline for PM10, PM2.5, and NO2 with the increasing concentration levels for each pollutant [33]. Compared with other studies, the sources of such negative results might not be a coincidence. We agree with Cerza et al. (2019) stating in the article that the potential sources could come from the uneven distribution of air pollution in Roma, the residual confounding for socioeconomic status, and the differences during using health services [33]. In fact, we repeated the meta-analysis by eliminating the data from Cerza et al. (2019) and find the overall OR for all air pollutions was 1.56 (95% CI: 1.15–2.15). The risk estimates for PM2.5, O3, and NO2 on AD were 2.31 (95% CI: 0.80–6.68), 1.03 (95% CI: 0.42–2.57), and 1.06 (95% CI: 0.89–1.26), respectively (Supplementary Figure 1). Compared with the results in Fig. 2, we found remarkable facts that NO2 generated a positive risk estimate, suggesting an important relationship between NO2 and AD. The deeper mechanism for such an association requires further investigation.
NO2 is an irritating gas and is soluble in water. Like the ways of O3 entry into the brain, NO2 may enter the brain through the respiratory tract-lung-bloodstream-brain pathway and the nose and olfactory pathway. NO2 inhalation decreased the ratio of brain to body weight, caused mild brain pathology, increased neuronal apoptosis, and altered antioxidants (superoxide dismutase, glutathione peroxidase, and nitric oxide) activity in rats [59]. NO2 exposure also increased the production of ROS and caused structural damage and dysfunction of mitochondria in rat brain [60]. Particularly, NO2 inhalation deteriorates spatial learning and memory in APP/PS1 mice and promoted amyloid deposition, for monoacylglycerol lipase (MAGL) disruption suppresses cyclooxygenase-2 (COX-2)-derived prostaglandin E2 (PGE2) and results in cognitive deterioration in response to NO2 inhalation [61]. These might be a key modulator in NO2-mediated aggravation of AD progression.
Mechanisms of the association between CO and AD
Four studies have indicated that CO poisoning is associated with an increased risk of dementia [11, 41]. For instance, Chang et al. (2014) pointed out that the adjusted HRs of dementia for all participants in quartiles Q2, Q3, and Q4 CO concentrations (196.2–241.6 ppm, 241.7–296.9 ppm,>296.9 ppm) compared to Q1 (<196.2 ppm) were 1.07 (95% CI, 0.92–1.25), 1.37 (95% CI, 1.19–1.58), and 1.61 (95% CI, 1.39–1.85) [38]. Lai et al. showed that the adjusted HRs of dementia associated with CO was 1.50 [40].
Toxicology research showed that AD patients were noted to have plasmatic coagulation kinetic and thrombus ultrastructural changes consistent with exposure to CO [39]. CO inhalation has a pleiotropic effect on cellular mitochondrial respiration, cellular energy utilization, inflammation, and free radical generation in the brain [62]. A higher expression of tau protein was found in patients who had lost consciousness during CO exposure [63]. As a result, the neurotoxic effects could be the toxicological mechanisms of CO-induced AD. With the target of above neurotoxic effects, related cohort and toxicological studies are still in demand.
Mechanisms of the association between SO2 and AD
SO2 is an important air pollutant relating to a great extent of diseases. Apart from the respiratory system, it can likewise cause neurological disorders. SO2 was uncovered being associated with stroke in which the risk of hospital admissions for ischemic stroke would raise 1.37% (1.05% –1.70%) with per 10μg/m3 increase of SO2 concentrations [64]. A recent study did not find significant effects on vascular dementia (OR = 0.95 for per 1 ppm SO2 concentration increased) [42]. However, for the long-term case-control study design ranging from 2005 to 2013, it provided some evidence of SO2-induced brain damage and neurotoxicity [42]. Another study discovered that the SO2 could cause neuroinflammation and signal transduction abnormality in neuronal cytoplasm and gene transcription in the nucleus, and lead to neurological damage in rat cerebral cortex [65].
Neurodegenerative injury such as tau phosphorylation overexpression, Aβ42 overproduction, and proinflammatory cytokine level increases could be consequent to the synergy effects of PM2.5 and SO2 [24,66, 24,66]. In a study with a model combing PM2.5 and SO2, the adjusted HRs for PM2.5 exposure to AD was altered with the influence of SO2 (2.34, 95% CI: 2.17–2.52) [9]. The situation remains when combining O3 and SO2, and the adjusted HR was changed to 2.34 (95% CI: 2.17–2.52) [9]. These findings suggested that exposure to air pollution, no matter with PM2.5 or O3, SO2 played an important role in leading to a higher risk of AD.
The differences between lightly- and heavily-polluted countries
With the stratification of data, we, for the first time, meta-analyzed the most typical air pollutants and found that the risks of AD due to air pollution in heavily polluted countries were significantly higher than that in lightly polluted countries (or regions). The double risks in heavily polluted countries uncovered the necessity to improve the air condition since people who live in those countries tend to be much more susceptible to AD. Besides, not only the direct influence of each pollutant but the synergistic effect and genetic toxicological disorders could be more popular for the long-term exposure and heredity. This finding was not by coincidence. It has been revealed that individuals living in severely polluted areas show a remarkable increased COX-2 expression and Aβ42 accumulation [67]. Neuritic plaques and neurofibrillary tangles induced by the neuronal dysfunctions due to such abnormal accumulation may worsen the AD development [67].
Limitations
Still, some limitations remain and need to be pointed out. First, the measurements and concentration levels of air pollution differed from studies and countries. Air pollutants increased AD risks and their toxicological mechanisms of the brain injury should consider the exposure levels of air pollutants and exposure time duration. The combined effects of co-exposure of air pollutants should be studies in-depth. It may have an influence on AD; however, it was hard to conduct due to the lack of raw data. Second, the amount of some air pollutants was still limited, especially those of SO2 and CO. For PM10, we kept it in the meta-analysis, although there was only one study included. Besides, the uneven distribution of air pollution, differences of uneven air pollution distribution, age, and habituation for using healthcare services might affect the included data. We only hope that with the larger included studies, it can provide more precise results regarding the association between air pollution and AD. Third, during the subgroup analysis by concentration levels, we found that the studies were mainly located in the lightly polluted countries such as the US, Canada, and Sweden. Although we had included heavily polluted countries in our study, there was still a lack of some countries with more typical air polluted conditions such as China and India. Further investigations are still required.
Conclusions
The present results from meta-analysis provided positive proof for the influence of PM2.5 and O3 on AD. With the subgroup analysis based on the pollution levels of air pollutants including PM2.5, PM10, O3, and NO2, it revealed that the individuals who lived in heavily polluted countries had double risks of AD onset compared with those in lightly contaminated countries. Although our study suggested a weak correlation between PM10 or NO2 and the incidence of AD, there still are potential pathological threats. The potential risks of SO2 and CO on AD deserve close attention, because of the positive relations of SO2 or CO with dementia and its pathological effects on AD animals. For the associations between AD onset and PM10, SO2, NO2, and CO, more epidemiology and toxicology studies need to be investigated. The governments around the world should fully notice the health concerns regarding the link between air pollution and AD and make policies to control air pollution emissions.
