Exposure to Environmental and Occupational Particulate Air Pollution as a Potential Contributor to Neurodegeneration and Diabetes: A Systematic Review of Epidemiological Research

It has been hypothesised that environmental air pollution, especially airborne particles, is a risk factor for type 2 diabetes mellitus (T2DM) and neurodegenerative conditions. However, epidemiological evidence is inconsistent and has not been previously evaluated as part of a systematic review. Our objectives were to carry out a systematic review of the epidemiological evidence on the association between long-term exposure to ambient air pollution and T2DM and neurodegenerative diseases in adults and to identify if workplace exposures to particles are associated with an increased risk of T2DM and neurodegenerative diseases. Assessment of the quality of the evidence was carried out using the GRADE system, which considers the quality of the studies, consistency, directness, effect size, and publication bias. Available evidence indicates a consistent positive association between ambient air pollution and both T2DM and neurodegeneration risk, such as dementia and a general decline in cognition. However, corresponding evidence for workplace exposures are lacking. Further research is required to identify the link and mechanisms associated with particulate exposure and disease pathogenesis and to investigate the risks in occupational populations. Additional steps are needed to reduce air pollution levels and possibly also in the workplace environment to decrease the incidence of T2DM and cognitive decline.


Introduction
Globally, there is an estimated 2.3 million deaths every year attributed to work, with 86.9% of these attributed to work-related diseases [1]. In Britain, deaths from workplace exposure to chemicals, dusts, and fibres have increased over the last 40 years, with currently around 13,000 premature deaths each year from occupational lung disease and cancer [2]. The non-occupational environment can also have detrimental impacts on health. More specifically, individuals are exposed to many risk factors in their homes and in the environment such as air pollution, chemicals, ionising and non-ionising radiation, and noise. Exposure to air pollution is a particular concern to the public; with 91% of the world's population living in areas where air quality is poor and exceeds World Health Organisation (WHO) guideline limits [3].
Overall occupational and environmental exposures account for around 12.5% of the total global mortality each year. Within this, it is estimated that air pollution is responsible for around 3.2% of the disease burden, which, according to WHO, accounts for an estimated 4.2 million deaths per year worldwide due to stroke, heart disease, lung cancer, and chronic respiratory diseases [3,4]. Outdoor air pollution contributes about 40,000 early deaths each year in the UK [5].
Ambient particulate air pollution, such as PM 10 , has been a public health concern in Britain since the 'London Smog' in 1952, when increased mortality and morbidity were reported soon after the event [6]. Similar incidents have occurred in Belgium (1930) and in Pennsylvania (1948) [7]. PM 10 is particulate matter 10 micrometres or less in diameter and has a complex composition, with the ultrafine particle component held principally accountable for eliciting much of the toxicity associated with PM 10 exposure [8,9]. Epidemiological studies indicate links between PM 10 and PM 2.5 with both short-term and long-term health effects [10]. Furthermore, both short-term (e.g., hours) and long-term exposure (e.g., months, years) can increase morbidity and mortality [11]. The link between PM 10 and adverse health effects was first established in the 1990s, where epidemiological studies found a positive association between the level of particulate air pollution in cities and increased morbidity and mortality rates in both adults and children, with adverse health effects being manifested predominantly in susceptible individuals who had pre-existing pulmonary or cardiovascular disease [12,13]. It is now widely accepted that exposure to PM 10 may cause or exacerbate allergic respiratory diseases (e.g., asthma) [14], chronic obstructive pulmonary disease, pneumonia, and cardiovascular disease [15] and is also defined as a definite carcinogen by International Agency for Research on Cancer (IARC) [16]. The cellular and molecular mechanisms underlying the toxicity of inhaled particles has been extensively investigated, and hypothesised to be driven by the stimulation of inflammation and/or oxidative stress [9].
The WHO estimates that 422 million people worldwide have diabetes, 47 million people are living with dementia and more than 10 million people are living with PD with increasing incidence expected over coming years.
As the world's population is ageing, dementia has become a major public health concern. Most cases of neurological diseases are sporadic, and thus identification that particulate air pollution may be an important environmental risk factor may help better understand the aetiology of neurodegenerative disorders.
Diabetes is a metabolic disease, with different types (type 1 and type 2). Type 1 is often diagnosed in childhood and cannot be controlled without taking insulin, whereas type 2 is usually diagnosed in over 30 year olds and can generally be controlled with medication, or diet and exercise. Increasing evidence indicates that environmental exposures may cause diabetes and more recently T2DM [32].
In addition to particle exposure via the environment, workplace exposures to particles can also occur. Workplace exposures can include a range of fine dusts and diesel exhaust. Accordingly, it is possible that exposure to particles in an occupational setting may also increase risk of T2DM and neurodegenerative disease.
The aim of this review is to systematically explore the association between long-term exposure (years/decades) to airborne particulate matter, both from ambient environmental pollutants and workplace exposures, and T2DM and cognitive impairment; earlier onset of greater than age-related decline and/or neurodegeneration in adults. The information from the review is discussed in relation to occupational exposure to particulate matter and fumes, and risk of dementia and diabetes.
We hypothesise that exposure to ambient fine airborne particles is associated with an increased risk of neurodegenerative conditions and T2DM, and that these risks also apply to workplace exposures to particulate matter.

Evaluation of the Systematic Review and Quality of the Studies
The internationally recognised GRADE system for the scientific evaluation of the quality of the identified studies was used to appraise the quality of the studies identified [33]. This system was developed for assessing interventions in health care contexts, and it has also been adapted to epidemiological studies [34,35]. In our case, an environmental/occupational factor or exposure can replace the clinical intervention [35]. The GRADE system classifies the quality of evidence as high, moderate, low, and very low and according to GRADE, evidence based on randomised controlled trials begins as 'high quality', and observational studies start as 'low quality'. All of the cohort and case-control studies epidemiological studies identified for this study were observational. A welldesigned observational study can be upgraded and provide a quality grading similar to a randomized controlled trial (RCT), and an RCT can be downgraded if certain aspects are missing. GRADE looks at studies individually, but also considers the whole evidence base.

Evaluation of the Systematic Review and Quality of the Studies
The internationally recognised GRADE system for the scientific evaluation of the quality of the identified studies was used to appraise the quality of the studies identified [33]. This system was developed for assessing interventions in health care contexts, and it has also been adapted to epidemiological studies [34,35]. In our case, an environmental/occupational factor or exposure can replace the clinical intervention [35]. The GRADE system classifies the quality of evidence as high, moderate, low, and very low and according to GRADE, evidence based on randomised controlled trials begins as 'high quality', and observational studies start as 'low quality'. All of the cohort and case-control studies epidemiological studies identified for this study were observational. A well-designed observational study can be upgraded and provide a quality grading similar to a randomized controlled trial (RCT), and an RCT can be downgraded if certain aspects are missing. GRADE looks at studies individually, but also considers the whole evidence base.
The following aspects of quality were considered in this study [36]:

1.
Representativeness and size of the study sample, as well as the duration of follow-up period. High number of participants, many years of follow-up, no attrition and lost during follow-up and participants that were members of a bigger cohort, could upgrade the quality of the studies.

2.
Publication bias or other kinds of bias, such as selection bias, misclassifications, selective reporting, or conflict of interest diminish the quality of a study. The quality may be downgraded if authors presented incomplete or selectively reported the tested hypothesis, compared to their aim and objectives. If there was no declaration of funder and involvement of the funding body in the research or if the authors had indicated a conflict of interest the quality was also downgraded. 3.
Potential confounders such as age, sex, physical activity, education level, alcohol intake, smoking status, socioeconomic status (SES) should have been considered. There is a need of an adequate control of confounding and adjustment in the statistical analysis. If more appropriate potential confounders were taken into account, then the quality of the study could be upgraded.

4.
For both exposure and outcome assessment, standardized and validated methods were required. Well established methods enable the comparison between studies and as a result they can contribute to higher quality rating.
This systematic review follows the PRISMA guidelines (Appendix A).

Results
The available research uses many different methodological designs to explore the associations between environmental or occupational particle exposures and the health-related outcomes. The approaches mostly quantify particle exposures using various measures of environmental pollution, and the outcome measures are quite disparate.

Air Pollution and Neurodegeneration
The characteristics of all the included studies focused on investigating the relationship between air pollution exposure and dementia are summarized in Tables 1 and 2. Of these, 10 studies were cohort studies, 4 studies were cross-sectional, 1 case-control, 2 nested case-control studies, and 1 was a time-series analysis. These were all recent studies, conducted after 2008, in North America (US, n = 10, Canada, n = 1), Europe (Germany, n = 3 and UK, n = 1), and Asia (Taiwan, n = 3). For the cohort studies the range of the follow up periods ranged from 5 to 30 years.
Most studies (17 out of 18) examined airborne particulate matter, four studies also included NO x and two studies considered both PM and traffic noise or proximity to a major road. Five studies quantified black carbon (BC) as a measure of PM. Exposure estimates-assessments based on measurement data-were mostly based on data from air monitors at fixed monitoring sites, using either the participant's postcode [26,37], address of residence [38][39][40][41][42], county [43] or the community [44]. Half of the studies (9 out of 18) considered traffic-related air pollution (BC, NO x , NO 2 , traffic noise, proximity to a major road, traffic PM), and their findings supported an association between air pollution exposure and dementia-related outcomes. Furthermore, one study supported a positive association of mild cognitive impairment (MCI) with road traffic noise [45] and a second study supports an association between shorter distance to a very busy road (10,000 cars per day) and poorer performance in a general assessment of cognition and a selective attention test [40].
Two studies used neuroimaging to show an adverse association between PM exposure and neurodegeneration (PD and white matter (WM) loss) [38,46]. One study supported an adverse association between PM and performance on a general ability test (working memory and orientation) [47], another supported an adverse association with visuo-spatial ability [41], another supported an adverse association with reasoning, short term memory, and verbal fluency [37]. Other studies identified adverse associations between cognitive decline and performance of mini mental state examination (MMSE) [42,48] and an adverse association between PM and hospitalization for neurodegenerative diseases [44]. Four studies found no association between PM exposure and the health outcomes of interest based on examining medical records or participant's reports [43,[49][50][51].
There was a wide range of techniques used to assess neurodegeneration and here we discuss the validity of different approaches. There is no single test that can provide a definitive diagnosis of Alzheimer's or other neurodegenerative disease in life. Diagnosis considers a range of cognitive abilities and functions and there is a need to look at them longitudinally. Physical and neurological examination, mental status tests (cognitive test scores, tests within a battery), neuroimaging, and medical history may all provide an insight into dementia-related endpoints. Information from laboratory tests or a physical examination can help identify health issues that can cause symptoms of dementia. Imaging technologies have revolutionized the understanding of the structure and function of the living brain to better track and diagnose the progress of Alzheimer's disease. Each method of detecting the dementia-related outcome has advantages and disadvantages. Cognitive test scores are informative and easy to use in epidemiology studies [52], whereas using a battery of tests offer considerable advantage over traditional measures (MMSE) "for predicting and discriminating stable from deteriorating forms of suspected dementia at an individual level" [21]. Furthermore, neuroimaging plays a crucial part in the diagnosis of dementia and provides a perception of the underlying pathologic process. Neuroimaging is important not only to exclude non-AD pathologies, but also to indicate biological markers to support the AD diagnosis [53].
Chen and co-workers [46] used a neurological examination and imaging to assess cognitive decline and Chen et al. [38] included tomography scans as an outcome measure, along with a battery of validated cognitive tests. Tomography scans and especially WM loss measurement is quantitative and specific to identify cognitive decline. Jung and colleagues used both history examination and neuroimaging [26]. However, most of the studies use a variety of different cognitive tests to measure cognitive decline [37,[39][40][41][42][43]45,47,48], medical records [49][50][51], or other self-reports [29].
As mentioned above, the size of the study sample, as well as the years of follow-up period were considered in the quality rating and high number of participants, many years of follow-up, and participants that were members of a larger cohort, can upgrade the studies. Kioumourtzoglou et al. [44] had the largest sample size, which included 9,817,806 adults, both men and women. The follow up period of the study was about 10 years and the study showed positive associations with all the three outcomes (PD, AD, and dementia). Chen et al. [54] had the second largest sample size, which again included men and women, 20-85 years, and had a long follow-up period (12 years). Both studies of Palacios et al. included thousands of participants and had more than 15 years of follow-up. Weuve et al. [42] had 19,409 female participants, who were part of the Nurses' Health Study Cognitive Cohort, and a follow up period of approximately 10 years. Ranft et al. [40], which was based on the SALIA cohort, studied proximity to a major road and had a 20 year of follow up period. Jung et al. [26] performed a study of 95,690 participants over 10 years, by using history, examination, lab tests and MRI to examine the associations between air pollution exposure and the health outcome. Studies of Kirrane et al., Tonne et al. and Power et al. [29,37,39] had adequate sample size and follow up period of 17 years, 5 years, and 10 years accordingly. Ranft  There is also one small case control study from Wu et al. comprised of 871 participants and only three years of follow-up. All of the other studies found statistically significant results, as mentioned above, except for three that found no statistically significant associations between particulate air pollution and PD.
Most of the reviewed studies adjusted for age, sex, smoking status, and physical activity, although several studies failed to adjust for sociodemographic factors, such as education or SES, which may potentially be confounders [26,29,43,47]. Lower education and low annual income have been associated with a greater risk for dementia [55][56][57] and it is plausible that people with low income live in areas with high levels of pollution. Although it is clear that people with cardiometabolic diseases tend to be more vulnerable to air pollution and that cardiometabolic diseases are a risk factor for cognitive dysfunction, only a few studies adjusted for such conditions [38,39,44,54]. Moreover, exposure to high levels of noise impairs cognitive abilities [54], and a large fraction of ambient pollution is traffic generated, making noise a possible confounder; only two studies adjusted for environmental noise exposure [39,45].
Bias, misclassification, and selective reporting can diminish the study quality. From our 18 studies, two mentioned selection bias [40,45], where people who were cognitively impaired were less likely to participate. Five studies [37,39,[49][50][51] mentioned misclassification of personal exposure levels (exposure misclassification), because the exposures were estimated from residential address, and one study [44] mentioned outcome misclassification; the measures of the diseases were based on hospital admission records, with the possibility of misclassification of diagnosis.
Taken together, the results from existing epidemiological studies suggest that there is a relationship between PM exposure and cognitive decline.

Air Pollution and Diabetes (T2DM)
The characteristics of all the epidemiology studies which have assessed the link between PM exposure and T2DM are summarized in Tables 3 and 4. These mostly comprise cohort studies, except for five cross-sectional studies [17,[58][59][60][61]. They were conducted in North America (Canada, n = 2 and US, n = 6), Europe (UK, n = 1, Switzerland, n = 2, Denmark, n = 1 and Germany, n = 3), and Asia (Korea, n = 1 and China, n = 2). For the cohort studies, the length of the follow up ranged from 3 to 22 years.
Most studies of environmental exposure (17 out of 18) examined associations with airborne particulate matter, although 12 studies also considered NO x , or traffic noise and proximity to a major road, and two studies considered annual mean residential BC concentration. Exposure estimates were mostly based on air monitors at fixed locations. There was one publication that investigated occupational exposure to environmental pollution, which was amongst traffic police. This study estimated exposure from PM monitors at fixed locations in different areas with various intensities of vehicle traffic [62].
A previous comprehensive meta-analysis of 12 studies, which also reviewed the study of Brook et al. [63] has shown that ambient PM and exposure to air pollutants, such as NO 2 and O 3 , were significantly associated with an increased risk of diabetes mortality [64]. Therefore, the findings from this meta-analysis add evidence for the adverse effect of particulate air pollution on diabetes-associated mortality and although diabetes is a risk factor for many other conditions (e.g., cardiovascular disease), this meta-analysis shows what it is already established; that higher levels of air pollutants, such as PM 10 , PM 2.5 , and NO 2 are associated with T2DM.
There are two studies ( [63,65]) in our review that examined diabetes mortality. Other studies investigated this association by assessing the incidence of T2DM from hospital admissions [66], or hospital diagnosis [67], or self-reports [20]. Diabetes status can be measured directly from biomarkers [68], such as fasting blood glucose (FBG), but also because insulin resistance (IR) and inflammation are important hallmarks of T2DM, biomarkers of inflammation, such as C-reactive protein (CRP), are also relevant. Most of the studies in this review used questionnaires and interviews, that collected information on treatment, blood glucose measurements, insulin, leptin, HbA1c, and CRP levels [69][70][71][72], but some studies carried out only blood tests of the above biomarkers [17,58,73], or used both methods [24,59,74].
In five studies [24,59,69,71,74] the outcome was measured using both questionnaires, completed by the participants and blood samples taken by them. There were two studies [66,67] based on hospital admission and diagnosis, four studies [32,58,61,73] that relied on samples taken by health specialists and one [60] based on a physical examination. Finally, there were three studies [20,70,72] that only used questionnaires, which may result in misclassification because of the use of self-reported diagnosis, and could not identify undiagnosed cases of diabetes.
In terms of study sample size and follow-up, Brook et al. [63] had the largest sample size (2.1 million adults) and a 10-year follow-up period. Pope and colleagues [65] also had a very large sample size (669,046 adults) and a 22-year follow-up. Puett and co-workers [72] [17,[58][59][60][61] are all cross-sectional studies and Liu et al. [58] and both O'Donovan et al. [61] and Wolf et al. [59] had one and three years of follow-up, respectively. All of the studies found statistically significant results, as mentioned above, except for Chen et al. [73], who found that exposure to PM 10 was associated with an increased level of FBG in the univariable analysis, whereas the results in the multipollutant model were not significant and Puett et al. who found no strong evidence between PM and T2DM.
It is known that an unhealthy diet can lead to obesity, one of the biggest risk factors of diabetes. Moreover, some studies indicate that the effects of PM and NO 2 are more noticeable in females [75]. Almost all the studies adjusted for BMI, sex, age, education, and physical activity. Some also adjusted for family history of diabetes [20,32]; or genes and genetic predisposition already known as risk factors of T2DM. It is hypothesised from animal studies and limited human subjects' studies, that air pollutants can decrease the normal synthesis of insulin [62], therefore in studies investigating environmental pollution it is important to adjust for occupational exposure to vapours, gases, dusts, and fumes, that are similar to urban air pollutants; only two studies did so [60,70]. Moreover, exposure to physical agents, such as noise -both indoor and outdoor-, can also modify the levels of the insulin, therefore adjustment for noise could be important, as Eze and colleagues [17] did.
All the studies mentioned limitations that can downgrade study quality, particularly information bias. Ten studies mentioned exposure misclassification [32,[58][59][60]63,[67][68][69]71,74] and two outcome misclassification [70,72], due to self-reporting of diagnosis. Two studies [63,65] measured the disease outcome from a mortality database and death certificates, possibly underestimating the true prevalence of T2DM. Underestimation of the effect, because of a selected sample of study subjects, was mentioned in two studies [24,65]. All the studies declared no conflict of interest and were funded by national agencies.
Taken together the results from existing epidemiological studies suggest that there is a relationship between PM exposure and T2DM.           In male traffic police mean plasma insulin levels were significantly lower compared with controls (p = 0.000).
In female were also significantly lower (p = 0.002).
Plasma insulin level was altered in traffic police who are exposed to chemical and physical stressors

Rating the Quality of Evidence According to GRADE
GRADE considers quality, publication bias, consistency, directness, and effect size and we discuss each in turn. In this system, observational studies start as low-quality evidence, but they can be rated down in a case of high risk of publication bias. There may be some doubt when the evidence derives from many small studies and especially if these studies have been commercially funded [76]. The evidence in this review comes from large observational studies and none of the studies was funded by industry. We have found some studies with statistically significant results and other 'null studies' of similar size; there is little evidence of publication bias.
In the GRADE scheme, the quality of evidence decreases when essential differences occur between the populations studied, or the outcomes measured, particularly if the outcomes are indirectly related to the disease [77]. For example, in our systematic review, the population of interest in all the 37 studies was adults, but some studies were conducted in women or men only [38][39][40][41][42]49,50,[67][68][69], which might influence the generalisability of the reported results, as studies of both sexes do not always examine risks separately for males and female and as the two genders may have different exposures because of different occupations for example, or because, as mentioned above, some effects are more noticeable in women. Also, there is a possibility that the desired outcome may be different from the measured outcome. The use of surrogate endpoints instead of 'patient-important outcome of interest' can be a source of indirectness [77]. For example, in diabetes-related studies the patient-important outcomes may be the hospital admission, diabetic symptoms, or complications because of diabetes and the surrogate outcome measures blood glucose or HbA1c concentration. In our review, there are studies that assess T2DM by looking at hospital admissions [66,67], or questionnaires [70,72], but there are also studies that measure the fasting blood glucose and other blood biomarkers [32,58,61,73]. In the dementia-related studies, the patient-important outcome may be the patient's function and behaviour, whereas the surrogate outcomes may be measures of cognitive function. In most of our dementia studies the measurement of the health outcome was cognitive function not a patient-important outcome.
GRADE also examines the consistency of the evidence; are there conflicting results in groups between studies, for example do the majority shows associations, do the studies have similar results, or is there a lack of agreement between studies? Existing evidence from meta-analysis indicates an association between air pollutants and both T2DM [75] and cognitive decline [78]. The studies that examined neurodegeneration-related outcomes mostly reported positive associations of particulate air pollution exposure and only three studies found no association [49,51,79]. We judge there is convincing evidence for consistency in the association of particulate ambient air pollution, cognitive decline, and dementia (AD), although there are conflicting results for an association between air pollution and PD. More specifically, seven studies indicated associations between fine particulate and traffic related air pollution and cognitive decline and impairment [37,[39][40][41][42]45,80], two studies identified association between air pollution and AD and dementia [48,54], two studies evaluated the association between air pollution exposure and only PD positively [29,46] and one study used a time-series-like approach to link particulate matter pollution with both PD, AD, and dementia [44]. Generally, a causal role of particulate matter can be concluded from many studies and also traffic emissions and living near busy roads can be a health threat. Increased daily levels of air pollution are related to higher hospital admissions and adverse health conditions [81,82].
Exposure to air pollution has already been suggested as a contributing factor to the increasing incidence and prevalence of diabetes [83,84] and all of the diabetes-related studies but one [72], reported positive associations between particulate air pollution exposure and diabetes-related outcomes.
It is also important to look at the effect of all possible confounding factors. In our systematic review, not all the possible confounders have been accounted for. In some cases, it is hard to exclude the possibility of social confounders. Moreover, most of the studies do not address indoor sources of air pollution, which cannot be calculated using geospatial models of air pollutants, therefore personal monitoring should take place at future studies to give us a clearer picture of the impact of different sources of pollutants. Another potential confounding variable that should be better explored in future studies is noise pollution, because of the possible association with memory loss.
Effect size, according to the GRADE approach, is based on the reported odds ratio, or relative risk or hazard ratio (OR/RR/HR) for comparison. However, it is somewhat unreliable to assess the odds ratios, due to different methodological approaches in the studies, but for most studies the observed risks were all modestly increased. There are no studies with an RR or OR of 2 or more, although most of them were statistically significant (except for [49][50][51]).
The GRADE evidence is discussed further in the following section.

Discussion
T2DM and dementia are common multi-causal conditions. An unhealthy diet, obesity, stress, culture, physical inactivity, and genetic predisposition may lead to cardio-metabolic diseases, such as T2DM [25]. Furthermore, both environmental and genetic factors play a crucial role in the aetiology of neurodegenerative diseases (e.g., AD) [26], depression, delirium, stroke, traumatic brain injury, ageing, and family history [85]. There are some familial cases (that are associated with genetic abnormalities) and some environmental factors have been identified as risk factors for neurological disease (e.g., pesticide exposure), but most of these cases are idiopathic [86].
Epidemiological evidence suggests that long term exposure to particulate air pollution is a risk factor for T2DM and dementia. The mechanism underlying this is currently unknown, but it may involve the translocation of inhaled particles to the target site (e.g., CNS), or the release of (inflammatory) mediators from the lung which impact on the function of extrapulmonary organs [10,30,31] (Figure 2). In the lung, it is established that PM stimulates inflammatory and oxidative responses, which drive its toxicity (reviewed in [10,87,88]). Accordingly, both these processes are likely to be important in mediating the detrimental outcomes of PM in other organs.
More specifically, inflammation and oxidative stress are implicated in the pathogenesis of neurodegenerative diseases and T2DM. Therefore, exposure to PM may contribute to the initiation of disease pathogenesis or accelerate disease development potentiating existing responses. Impacts of inhaled particles on the CNS may emerge due to (i) particle translocation (via neurones or blood) to the CNS following inhalation or (ii) the release of systemically acting factors from the lung which impact on neurone function. Examples of the clinical impacts of inhaled particles at different target sites (lung, extrapulmonary organs, and CNS) are summarised in blue boxes. (B) The cellular and molecular events underlying particle toxicity to the lungs have been extensively investigated and hypothesised to involve the stimulation of inflammation and oxidative stress. More specifically, it is hypothesized that inhaled ultrafine particles interact with pulmonary cells (e.g., epithelial cells, alveolar macrophages) to stimulate an increase in intracellular ROS and Ca 2+ concentration which leads to the expression of pro-inflammatory genes (e.g., cytokines) via the activation of transcription factors (such as NFκB). BBB = blood brain barrier. COPD = chronic obstructive pulmonary disease.
Chronic inflammation has been found to be associated with the development of T2DM in humans [89]. Furthermore, a chronic inflammatory response is associated with obesity [90], one of the main risk factors for T2DM. Insulin resistance (IR) is one of the most important hallmarks in the pathogenesis of T2DM. IR is directly linked with a variety of inflammatory responses and these responses play a crucial role in the development of the condition [91]. Indeed, quantification of inflammatory mediators in blood is used as a biomarker for T2DM [92]. Individuals with T2DM have elevated cytokine levels in blood compared to people without diabetes and there is mounting evidence that supports that diabetes and particulate air pollution are associated with inflammatory dysregulation [22,93,94]. In addition, there is evidence that increased exposure to PM is significantly associated with increased fasting blood glucose (a biomarker for diabetes) in humans [73,95]. Whilst epidemiological evidence suggests that PM exposure is a risk factor for T2DM development, further Impacts of inhaled particles on the CNS may emerge due to (i) particle translocation (via neurones or blood) to the CNS following inhalation or (ii) the release of systemically acting factors from the lung which impact on neurone function. Examples of the clinical impacts of inhaled particles at different target sites (lung, extrapulmonary organs, and CNS) are summarised in blue boxes; (B) The cellular and molecular events underlying particle toxicity to the lungs have been extensively investigated and hypothesised to involve the stimulation of inflammation and oxidative stress. More specifically, it is hypothesized that inhaled ultrafine particles interact with pulmonary cells (e.g., epithelial cells, alveolar macrophages) to stimulate an increase in intracellular ROS and Ca 2+ concentration which leads to the expression of pro-inflammatory genes (e.g., cytokines) via the activation of transcription factors (such as NFκB). BBB = blood brain barrier. COPD = chronic obstructive pulmonary disease.
Chronic inflammation has been found to be associated with the development of T2DM in humans [89]. Furthermore, a chronic inflammatory response is associated with obesity [90], one of the main risk factors for T2DM. Insulin resistance (IR) is one of the most important hallmarks in the pathogenesis of T2DM. IR is directly linked with a variety of inflammatory responses and these responses play a crucial role in the development of the condition [91]. Indeed, quantification of inflammatory mediators in blood is used as a biomarker for T2DM [92]. Individuals with T2DM have elevated cytokine levels in blood compared to people without diabetes and there is mounting evidence that supports that diabetes and particulate air pollution are associated with inflammatory dysregulation [22,93,94]. In addition, there is evidence that increased exposure to PM is significantly associated with increased fasting blood glucose (a biomarker for diabetes) in humans [73,95]. Whilst epidemiological evidence suggests that PM exposure is a risk factor for T2DM development, further research is required to better understand the cellular and molecular events underlying this as few studies have investigated the mechanism of toxicity, to date.
Numerous studies have established the role of neuroinflammation in both AD and PD pathology [28,96]. When inflammation is activated, neurones, and microglia cells release pro-inflammatory mediators (e.g., cytokines) which can stimulate an inflammatory response which damages neuronal cells and ultimately brain tissue [97,98]. The stimulation of inflammation is key to the pathogenesis of dementia. For example, the degree of inflammation correlates with brain atrophy and the severity of dementia in early AD [99]. There are also markers of inflammation, such as C-reactive protein levels (CRP) and interleukin-6 (IL-6) that are elevated in the blood plasma of patients with AD and vascular dementia [100]. An increase in the levels of the inflammatory markers in blood is associated with an increase in the risk of all types of dementia [101]. Moreover, it is already known that higher levels of inflammatory markers are associated with greater brain atrophy than expected for age [102].
Exposure to particulate air pollution, has been linked also with brain inflammation [103]. The first indications that inhaling polluted air could cause neurodegeneration came from an experiment with demented dogs in Mexico City. In this experiment, canines were exposed to significant concentrations of ozone, PM, and other pollutants. The researchers identified inflammation in the brains of the dogs along with endothelial damage, which prompted the hypothesis that the initial inflammatory source was the respiratory tract [104]. There is now evidence from several in vivo studies that PM can stimulate inflammatory and oxidative responses in the CNS [104][105][106][107][108][109][110]. Furthermore, Hullmann et al. [111] demonstrated that diesel exhaust particles accelerated the development of hallmarks of Alzheimer's Disease in a mouse disease model. In addition, Finch and Morgan [112] showed in mice that inhalation of particulate polluted air activated the brain's microglia, which stimulated an inflammatory response that was linked to memory loss and the pollution-exposed mice showed signs of brain damage. They suggested that the fine airborne particles might travel from the nasal cavity to the brain. Interestingly, few in vitro studies [48,109,[113][114][115] have investigated the response of neurones to PM 10 despite evidence from epidemiology and in vivo studies that PM 10 can cause neurotoxicity.
Exposure to environmental toxicants (such as air pollutants) increases the risk of Parkinson's disease [116] and there is evidence that neuroinflammation is the etiopathogenesis not only for Alzheimer's disease, but also of Parkinson's disease [28,117]. Long-term exposure to particulate air pollution may cause damage to dopaminergic neurons and lead to chronic brain inflammation to accelerate AD and PD development [29].
Epidemiological evidence has identified that diabetes and neurodegenerative diseases are linked, for example it is known that people with a metabolic syndrome, such as diabetes, are at higher risk of developing cognitive impairment [118] and Alzheimer's disease. IR links obesity with pre-diabetes and diabetes and is associated with an increased risk for cognitive decline [119] and age-related memory impairment and AD [120]. Metabolic syndrome is considered as an independent risk factor for pre-AD syndrome and AD [121][122][123][124] and generally diabetes is a known risk factor for cognitive dysfunction and all-cause dementia [125]. T2DM is similar to AD in many ways. They are both associated with impaired glucose uptake, increased oxidative stress, inflammation, ageing, brain atrophy and they may both cause impaired cognition and dementia. This suggests that these two diseases share many factors in terms of pathophysiology and clinical outcome and due to the similarity and the pathophysiological bridge between them, AD is often referred as "type 3 diabetes" [126,127]. There is also a link between diabetes and Parkinson's disease [128]. Individuals with T2DM are at increased risk of developing PD [129,130].
The relationship between diabetes and neurodegenerative disease may arise as a consequence of a common inflammatory mechanism [131]. For example, among individuals with metabolic syndrome, those with a higher level of inflammation are at higher risk of developing cognitive impairment compared to those with low inflammation [132]. Whilst there is evidence that inflammation is linked to both T2DM and neurodegenerative diseases, it is uncertain whether T2DM is a prerequisite to develop neurodegenerative disease from this inflammatory mechanism, or whether inflammatory processes act independently to cause these diseases. Furthermore, the hormone insulin plays an important role in memory and in brain function [120] and thus dysregulation of insulin signalling has been linked to metabolic and neurodegenerative disorders [133,134]. Accordingly, several conditions, including T2DM, activate a range of inflammatory, metabolic, and oxidative changes that might contribute to deleterious effects on the brain and other metabolic changes that might potentially drive neurodegenerative processes [135].
In this systematic review, we considered 37 studies and both levels and assessment of exposure and outcome definitions varied. The majority of the studies (31 out of 37) indicated a positive association between particulate traffic-related air pollution and cognitive impairment and/or T2DM. However, the quality of the epidemiological evidence is considered 'poor' because of the observational nature of the studies and there are not RCT studies that would increase the quality of the evidence. The quality of the evidence base is also poor either because it is often based on self-reports rather than on objective cognitive tests, clinical examination or neuroimaging. Furthermore, diabetes-related studies that were based on mortality data, could have underestimated the true prevalence of the disease.
Noise exposure is closely correlated with traffic related air pollution, and noise exposure is an independent risk factor for neurodegenerative diseases [136,137]. Moreover, when investigating noise or traffic proximity, which is an indicator of traffic noise, it is important to consider hearing impairment with ageing in the adult population. It is important that epidemiological studies adjust for noise exposure, although the majority of existing studies did not. It is also important that studies consider established risk factors of the disease such as obesity, nutrition, and active and passive smoking. The majority of the studies adjusted for smoking status, but only few adjusted for certain food consumption; such as fish consumption, which is associated with the risk for dementia [138]. A healthy survivor effect may also be important in some studies, because with ageing, some diabetes or dementia patients could die prematurely and no longer participate. Other lifestyle factors or other health status covariates should also be considered, such as activities of everyday life-i.e., computer experience or depression symptoms-because they play a crucial role in cognitive performance and only few studies considered them. Lack of information on these could lead to a risk of bias. Moreover, family history of cardiovascular disease can also be important and should be considered, but some studies only mentioned cardiovascular risk factors or already existing cardiovascular disease.
We were not able to perform a meta-analysis because of the variety of different outcomes that were described in the studies. However, previous review studies on T2DM have done meta-analysis for different exposure subgroups and all support an association of air pollutants with an increased risk for T2DM [18,75,139]. In practice a meta-analysis would be feasible with any of the morbidity or mortality endpoints provided there were sufficient contributing studies with consistency in the definition of the outcomes.
There are many occupations that are exposed to airborne particles, but there is only one epidemiological study that is informative about possible risks of T2DM and dementia from such exposure [62]. In workplaces, ultrafine particles are found in metal and polymer fumes and both can induce acute inflammation responses in the lung upon inhalation [140]. Many other occupational aerosol exposures can cause chronic lung inflammation-some closely related to air pollution, such as diesel engine exhaust particulate and others, such as respirable crystalline silica, that cause inflammation because of chemical or surface properties. Carbon black workers may show lung function reduction along with pro-inflammatory cytokines secretion [141]; exposure to dust can cause pulmonary reactions in dairy farmers, cotton workers, and wood workers; and inhalation of irritants can demonstrate pulmonary inflammation [142]. Also ultrafine particles can affect the nasal epithelium and produce inflammation that damages the brain [31].
As we have seen, inflammation is the key biological process linked to T2DM and neurodegeneration. By analogy, we judge that occupational particulate exposure that causes inflammation is a 'possible' cause of T2DM and neurodegenerative diseases due to the relatively high exposures that are likely to occur in the work environment. However, further research is needed to clarify whether there is a risk from workplace particulate exposures.

Conclusions
Despite the various studies investigating the association between air pollution and T2DM and cognitive impairment and neurodegeneration, the role of air pollution in the causation of these disorders is not fully understood and remains unclear. Available evidence indicates a positive association of ambient particulate air pollution and both T2DM and neurodegeneration risk, but corresponding evidence for similar workplace exposures is lacking. However, it is plausible that such an association between fine aerosols in the workplace are associated these diseases. Further research is required to identify the link and toxicological mechanisms associated with particulate exposure and T2DM and neurodegenerative disease. Future studies could fill key evidentiary gaps and thereby lead to additional steps to decrease air pollution levels and improve policies in the workplace environment to decrease the incidence of T2DM and cognitive decline, creating a healthier and more sustainable future.
Author Contributions: E.D. served as the primary author performing initial database searches and drafting the initial manuscript. J.W.C. also performed initial database search. J.W.C. conceived the review and was the research supervisor, assisting with database searching, quality appraisal and providing critical feedback and is the corresponding author. G.S. and H.J.J. provided guidance and critical feedback and editing on the manuscript. All authors read and approved the final manuscript for submission.
Funding: E.D. thanks the Colt Foundation for financial support through a personal fellowship.

Conflicts of Interest:
The authors declare no conflict of interest. Specify study characteristics (e.g., PICOS, length of follow-up) and report characteristics (e.g., years considered, language, publication status) used as criteria for eligibility, giving rationale. 3 Information sources 7 Describe all information sources (e.g., databases with dates of coverage, contact with study authors to identify additional studies) in the search and date last searched. 3 Search 8 Present full electronic search strategy for at least one database, including any limits used, such that it could be repeated. 3 Study selection 9 State the process for selecting studies (i.e., screening, eligibility, included in systematic review, and, if applicable, included in the meta-analysis). 3 Data collection process 10 Describe method of data extraction from reports (e.g., piloted forms, independently, in duplicate) and any processes for obtaining and confirming data from investigators. 3

Data items 11
List and define all variables for which data were sought (e.g., PICOS, funding sources) and any assumptions and simplifications made.

8-25
Risk of bias in individual studies 12 Describe methods used for assessing risk of bias of individual studies (including specification of whether this was done at the study or outcome level), and how this information is to be used in any data synthesis.

Study selection 17
Give numbers of studies screened, assessed for eligibility, and included in the review, with reasons for exclusions at each stage, ideally with a flow diagram. 4

Study characteristics 18
For each study, present characteristics for which data were extracted (e.g., study size, PICOS, follow-up period) and provide the citations.

8-25
Risk of bias within studies 19 Present data on risk of bias of each study and, if available, any outcome level assessment (see Item 12).

8-15, 17-24
Results of individual studies 20 For all outcomes considered (benefits or harms), present, for each study: (a) simple summary data for each intervention group (b) effect estimates and confidence intervals, ideally with a forest plot.  Describe sources of funding for the systematic review and other support (e.g., supply of data); role of funders for the systematic review. 31