Association of fine particulate matter (PM2.5) exposure and chronic kidney disease outcomes: a systematic review and meta-analysis

Several studies have reported an increased risk of chronic kidney disease (CKD) outcomes after long-term exposure (more than 1 year) to particulate matter with an aerodynamic diameter of ≤ 2.5 µm (PM2.5). However, the conclusions remain inconsistent. Therefore, we conducted this meta-analysis to examine the association between long-term PM2.5 exposure and CKD outcomes. A literature search was conducted in PubMed, Scopus, Cochrane Central Register of Controlled trials, and Embase for relevant studies published until August 10, 2023. The main outcomes were incidence and prevalence of CKD as well as incidence of end-stage kidney disease (ESKD). The random-effect model meta‐analyses were used to estimate the risk of each outcome among studies. Twenty two studies were identified, including 14 cohort studies, and 8 cross-sectional studies, with a total of 7,967,388 participants. This meta-analysis revealed that each 10 μg/m3 increment in PM2.5 was significantly associated with increased risks of both incidence and prevalence of CKD [adjusted odds ratio (OR) 1.31 (95% confidence interval (CI) 1.24 to 1.40), adjusted OR 1.31 (95% CI 1.03 to 1.67), respectively]. In addition, the relationship with ESKD incidence is suggestive of increased risk but not conclusive (adjusted OR 1.16; 95% CI 1.00 to 1.36). The incidence and prevalence of CKD outcomes had a consistent association across all subgroups and adjustment variables. Our study observed an association between long-term PM2.5 exposure and the risks of CKD. However, more dedicated studies are required to show causation that warrants urgent action on PM2.5 to mitigate the global burden of CKD.


Searching strategy
Based on existing literature, a systematic search was implemented to search the literature on the relevance between PM 2.5 and the CKD or ESKD outcomes.Our search encompassed the PubMed, Scopus, Cochrane Central Register of Controlled Trials (CENTRAL), and Embase databases up until August 10th, 2023, to identify relevant articles.The inception date of the search strategy was June 6th, 2023.The search terms utilized were ("particulate matter 2.5" OR "PM2.5")AND ("kidney"[Mesh] OR kidney [tiab] OR "renal"[Mesh] OR renal [tiab]) in PubMed, and ("particulate matter 2.5" OR "PM2.5")AND (kidney OR renal) in Scopus, CENTRAL, and Embase.Language restrictions were not imposed during the search process.

Inclusion and exclusion criteria
The inclusion criteria of this meta-analysis comprised five points: (1) study subjects had to be adults (≥ 18 years); (2) studies had to examine long-term exposure (≥ 1 year) to fine particulate matter with an aerodynamic diameter of ≤ 2.5 μm (PM 2.5 ); (3) only observational studies, including cross-sectional and cohort studies, were accepted; (4) the outcomes had to conclude the term "chronic kidney disease" or "end-stage kidney disease" explicitly for investigation with clinical assessments (such as diagnosed by physician, using the International Classification of Disease (ICD) code, or the Kidney Disease: Improving Global Outcomes (KDIGO) guideline); (5) studies reported the effect estimates (odds ratio; OR, and hazard ratio; HR) and their 95% confidence intervals (95% CIs) of clinical outcomes with per 10 μg/m 3 increment exposure PM 2.5 concentrations were available, or sufficient data could be used to convert these results.The exclusion criteria comprised three points: (1) reviews, meta-analyses, and responses to letters; (2) studies involving non-human species; and (3) the study reporting only specific chemical components of PM 2.5 -related adverse renal outcomes.

Data extraction
The assessment of titles and abstracts for each record obtained, as well as the examination of full-text reports, was carried out independently by AB and WW.Whenever a discrepancy arose between the two reviewers, resolution was achieved through discussion involving the third author (PS).If multiple reports originated from the same cohort, the report with the largest sample size was selected.Subsequent data were extracted from each report, including the first author, year of publication, sampling period, study design type, research country, participant numbers, gender and age of participants, presence of diabetes and hypertension, mean body mass index, smoking habits, exposure assessment details, air pollutant data source, outcome and its assessment details, mean level of PM 2.5 exposure, duration of follow-up, and risk of bias score.

Assessments of quality and risk of bias
The evaluation of bias was conducted using the Newcastle-Ottawa Scale (NOS) for cohort studies and the modified NOS for cross-sectional studies 19,20 .The NOS encompasses a set of inquiries aimed at assessing the selection of study participants, the comparability of the population, and the outcomes.For cohort studies, the NOS was converted to adhere to AHRQ standards and categorized as follows: Good quality (3 or 4 stars in the selection domain AND 1 or 2 stars in the comparability domain AND 2 or 3 stars in the outcome/ exposure domain), Fair quality (2 stars in the selection domain AND 1 or 2 stars in the comparability domain AND 2 or 3 stars in the outcome/exposure domain), Poor quality (0 or 1 star in the selection domain OR 0 stars in the comparability domain OR 0 or 1 stars in the outcome/exposure domain).The adapted NOS for cross-sectional studies was designed with a maximum score of 10 points.Studies receiving 9-10 points were classified as very good, those with 7-8 points as good, those with 5-6 points as satisfactory, and those with 0-4 points as unsatisfactory.Utilizing these assessment criteria, both AB and WW conducted evaluations of the quality of each included article.Instances of differing opinions were resolved through consultation with a third author (PS).

Statistical analysis
We conducted meta-analysis to extract combined effect estimates for the association of long-term PM 2.5 exposure to CKD outcomes.The outcomes of the systematic review were classified into three categories: CKD prevalence, CKD incidence, and ESKD incidence.In each study, we extracted the adjusted effect estimates for every outcome, considering a more robust control for confounding variables.Within a subset of studies featuring analyzable and comparable data, expressing results as a standardized increment in PM 2.5 concentration (μg/m 3 ), the results were quantitatively synthesized.Odd ratios (ORs) were used as measurements of effect estimates across the included studies.If individual studies reported hazard ratios (HRs), we first converted these ratios into ORs using the method described by Shor et al. 21prior to calculating the pooled result.Random-effects model meta-analyses were performed to calculate pooled ORs for binary variables (i.e., presence versus absence of CKD outcomes) from multivariate analysis.Since the PM 2.5 increment scales used to calculate the OR value in each study are inconsistent, which cause the effect values lack uniformity and cannot be combined.To circumvent this, we standardized the effect estimates (ORs and 95%CI) by pooling them based on a uniform per 10 μg/m 3 increase in PM 2.5 concentration.The standardized OR value for each article was calculated by using the formula as follows: All pooled estimates were displayed with 95% CI.The presence of heterogeneity among the effect sizes of individual studies was assessed through the Cochrane's Q test and the I 2 index.I 2 values of 25%, 50%, and 75% or higher represent a low, moderate and high degree of heterogeneity, respectively.To explore sources of heterogeneity, we performed subgroup meta-analyses according to continents (Asia, Europe, or North America), sampling period (before 2013 or after 2013), study participants (< 10,000, 10,000-100,000, or > 100,000), mean PM2.5 level (< 25 μg/m 3 , or ≥ 25 μg/m 3 ), based on WHO defining concentrations exceeding 25 μg/m 3 as very high), pollutant data source (monitoring stations, predictive model, or machine learning), eGFR formula (Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) equation or the Modification of Diet in Renal Disease (MDRD) study equation), and exposure periods (< 10 years, or ≥ 10 years).To graphically represent this heterogeneity among the included studies, a forest plot was employed.Publication bias was assessed formally using Funnel plots and the Egger test.All of these analyses were carried out using Comprehensive Meta-Analysis version 2.0 (www.meta-analy sis.com, accessed on August 20, 2023; Biostat, Englewood, NJ, USA).

Summary of included studies
A total of 787 potentially relevant articles were initially identified through the database search.Following the removal of 311 duplicated articles, 476 article titles and abstracts underwent screening based on the inclusion and exclusion criteria, resulting in the identification of 28 full-text publications that underwent subsequent evaluation.After the full-text screening process, six articles were excluded (the reasons for exclusion are detailed in Fig. 1).Finally, a total of 22 articles were included for the systematic review and meta-analysis (Fig. 1).

Outcome assessment and exposure characteristics
The definition of CKD outcome among the included studies was consistently defined according to the KDIGO guidelines 44 .Specifically, the outcome was characterized by an eGFR lower than 60 mL/min per 1.73 m 2 , as predominantly determined by the CKD-EPI equation (n = 12; 57%) or MDRD equation (n = 6; 28.6%).Conversely, the ESKD outcome was primarily relied on the utilization of ICD codes.Of the included studies, the incidence of CKD was the most reported outcome (in 13 studies), followed by the prevalence of CKD (in 8 studies) and the incidence of ESKD (in 3 studies).Among the cohort studies, the reported incidences of CKD and ESKD ranged from 1.14% over 11.9 years to 27.3% over 17.7 years and 0.19% over 11.9 years to 1.29% over 8.52 years, respectively.In cross-sectional studies, the observed prevalence of CKD varied between 1.3 and 27.8%.

Methodological quality
Regarding the Newcastle-Ottawa Scale (NOS) for cohort studies and the NOS adapted for cross-sectional studies 19,20 , all cohort studies (n = 13; 100%) were considered to be of good quality.Likewise, all of the crosssectional studies (n = 9; 100%) were considered of good or very good quality (scores of 7-8).(see Supplementary Tables S1 and S2).
ESKD incidence.Three cohort studies 22,29,32 (3,098,054 participants) reported the association between longterm exposure to PM 2.5 and incidence of ESKD.The result showed that the combined unadjusted and adjusted ORs of ESKD incidence in the meta-analyses were 1.32 (95% CI 0.85 to 2.04; p = 0.219; 2 studies, 615,317 analyzed participants) (Table 2) and 1.16 (95% CI 1.00 to 1.36; p = 0.058; 3 studies, 3,098,054 analyzed participants) (Fig. 4, Table 2) per 10 μg/m 3 increase in PM 2.5 concentration of exposure, respectively.Owing to the small number of included studies, we did not perform heterogeneity tests (such as subgroup analysis) on this outcome.www.nature.com/scientificreports/

Investigations of heterogeneity
We found high heterogeneity in the estimated association among studies for all of the study outcomes (I 2 = 98.7% for CKD incidence, I 2 = 97.6% for CKD prevalence, and 93% for ESKD incidence).Therefore, we used the subgroup analyses to explore the potential confounding factors in the incidence and prevalence of CKD outcomes.Tables 3 and 4 detail the results of subgroup analyses examining the association between PM 2.5 body exposure (per 10 μg/m 3 increase) and risk of incidence and prevalence of CKD, respectively, as stratified by continents   In brief, all subgroups and adjustment variables considered in both the meta-analyses for incidence and prevalence of CKD consistently demonstrated that long-term exposure to PM 2.5 (per 10 μg/m 3 increase) was positively correlated with an elevated risk of both CKD outcomes (P for interaction > 0.05) (Tables 3 and 4).

Assessment of publication bias and sensitivity analysis
The funnel plot for the outcome of both incidence and prevalence of CKD in the studies included in the metaanalysis was asymmetrical (Fig. 5A,B).The results of the Egger's test suggested the presence of potential publication bias for CKD incidence outcome (p = 0.005), but not for CKD prevalence outcome (p = 0.44).The sensitivity analysis was conducted using the leave-one-out method (omitting one study at a time and recalculating the pooled effect estimate).The findings showed that the association between PM 2.5 and incidence of CKD was generally stable and not dominated by any single study, which suggested that the results of the meta-analysis are substantially reliable.

Discussion
In this systematic review and meta-analysis, we comprehensively evaluated and updated the existing epidemiologic evidence, including a total of nearly 8 million participants, on the association between long-term exposure to PM 2.5 and adverse renal outcomes (incidence and prevalence of CKD as well as incidence of ESKD).Besides well-established risk factors for CKD, our analysis suggests that air pollution, particularly PM 2.5 , is identified as one of the emerging environmental risk factors, which has detrimental effects on kidney health.Of significance is the finding that long-term exposure to PM 2.5 (per 10 μg/m 3 increase) was associated with an elevated risk of CKD incidence (adjusted OR 1.31), CKD prevalence (adjusted OR 1.31).In addition, the relationship with ESKD incidence is suggestive of increased risk but not conclusive (adjusted OR 1.16; p = 0.058; when the followup duration extends beyond 10 years) (Figs. 2, 3, 4 and Table 2).High heterogeneity was noticed in the overall meta-analysis and most subgroup analyses, which may be attributed to the country development situation, continents, sampling period, sample size, mean PM 2.5 level, pollutant data source, follow-up time, eGFR formula, temperature, and humidity.However, this finding might help explaining why CKD incidence and prevalence continued to increase on a global scale.
Table 5 illustrates the summary findings from four meta-analyses that reported the association between PM 2.5 exposure and adverse kidney outcomes.Prior to the present study, there have been three meta-analyses on this topic, with one comprising 16 studies 45 , another encompassing 7 studies 46 , and the third including 13 studies 47 .Although several studies have performed meta-analyses on the impact of PM 2.5 on CKD, none of these studies did a subgroup analysis to address the potential confounding factors that might affect this association.Of note, we found that specific subgroups (including continent, mean PM 2.5 level, and sampling period) influenced the magnitude of this correlation, albeit in the same direction (Tables 3 and 4).Furthermore, two of these reports 45,47 did not provide a clear definition of the CKD outcome, particularly in terms of distinguishing between incidence and prevalence of CKD.It is important to highlight that combining these outcomes together in the meta-analysis was not an appropriate approach.In the context of adverse renal outcomes, previous metaanalyses were limited to examining only CKD outcome; in comparison, our analysis broadened its scope by including an additional relevant outcome, which is the incidence of ESKD, with a longer observational period.Despite the consistent findings across these reports, all of which demonstrated a positive correlation between long-term exposures to PM 2.5 (per 10 μg/m 3 increase) and an elevated risk of CKD outcomes, the effect size was relatively small, ranging between 9 and 15% in terms of incremental risk.Moreover, one report by Wu et al. 45 indicated borderline statistical significance in this context probably due to the limited number of the included studies (n = 4).Distinct from our systematic review and meta-analysis, the heightened magnitude of the effect size to 31% incremental risk of CKD incidence might be attributed to the incorporation of a larger sample size compared with the previous reports (Table 5).www.nature.com/scientificreports/As a pollutant, PM 2.5 is detrimental to public health due to its physical, chemical, and biological properties 48 , with the major components comprising elemental carbon, biological substances, inorganic components, organic components, and trace elements 49 .The sources can be either natural, such as coal burning and soil dust, or anthropogenic, such as vehicle traffic and industrial emissions 50 .Although the exact mechanisms through which PM 2.5 induces kidney injury in humans remain unclear, it is hypothesized that PM 2.5 primarily disrupts normal renal homeostasis via direct and indirect pathways 51 .Currently, the majority of the evidence explaining the direct pathways of kidney damage is derived from research conducted on animals.In summary, the identified mechanisms, mainly at the cellular level, encompass oxidative stress, inflammation leading to DNA damage, endoplasmic reticulum stress, apoptosis, and the development of renal fibrosis [52][53][54] .Furthermore, dysregulation of several systemic pathways such as angiotensin/ bradykinin systems, antioxidant, immune systems, and renal vascular activities has also been observed 55,56 .Apart from direct harmful effects, there is a growing body of evidence indicating that PM 2.5 plays a substantial role in contributing to CKD through indirect pathways, primarily involving two major non-communicable diseases: hypertension 57,58 and type 2 diabetes mellitus (T2DM) 59,60 , which serve as the principal drivers of CKD.www.nature.com/scientificreports/ In subgroup analysis, we observed a notable association between PM 2.5 and CKD incidence, particularly in the Asian region, despiteits smaller number of participants included in the analysis (Table 3).However, it is essential to note that the P for interaction > 0.05 when comparing Asian to other continents, indicating no statistically significant difference in the observed effects.In consistence with the World Health Organization (WHO)'s report 61 , less-developed regions, such as Asia and Africa, suffer PM 2.5 exposures that are four to five times those of more-developed regions, including Europe and North America.The explanation behind this result lies in the rapid urbanization and economic growth observed in several Asian countries, which have led to a substantial increase in air pollution 62 .
Based on the "Air Quality Guideline" of the WHO 63 , an annual average of PM 2.5 concentrations exceeding 25 μg/m 3 is defined as a very high concentration, which can potentially have harmful effects on human health.The primary focus in terms of systemic diseases was on cardiovascular disease, respiratory disease, and lung cancers because of all linked to increased mortality risk 64 .The results from our analysis remained consistent in showing that individuals exposed to an average PM 2.5 concentration higher than 25 μg/m 3 had a greater risk of CKD outcomes in comparison to those with levels below 25 μg/m 3 (Table 3).Of particular significance, these findings emphasize that kidney diseases should be recognized as another key public health concern related to the influence of PM 2.5 .Since the WHO designated PM 2.5 as a Group 1 carcinogen in 2013, the global trend of PM 2.5 concentration has gradually decreased over time due to its reduction policy.Therefore, we also conducted a subgroup analysis of the sampling period before and after 2013.Interestingly, the result showed that long-term PM 2.5 exposure was more positively related to incident CKD and CKD prevalence in sampling periods after 2013 compared to before 2013 (Table 3).This could be clarified through the mechanism of renal injury, which involves a cumulative process requiring prolonged exposure to cause kidney damage.
Our systematic review has several strengths.This is the first systematic review and meta-analysis of observational studies that explores an association between long-term exposure to PM 2.5 and adverse renal outcomes, particularly CKD incidence and CKD prevalence.We included reports that performed multivariable analyses to account for potential confounders of these associations.Furthermore, our search encompassed studies published until August 2023.It is worth noting that in the past few years, there has been a substantial increase in publications on this topic.This has resulted in a greater number of studies, a more diverse population, and more Table 5. Summary of findings from 4 meta-analyses on the association between PM 2.5 exposure and adverse kidney outcomes.AHRQ agency for healthcare research and quality, CI confidence interval, CKD chronic kidney disease, ESKD end-stage kidney disease, GFR glomerular filtration rate, JBI-MAStARI Joanna Briggs institute meta-analysis of statistics asessment and review instrument, NOS Newcastle-Ottawa scale, PM particulate matter, RCT randomized control trial.
Wu et al. 45 Liu et al. 46 Ye et al. 47 The present meta-analysis recent data, which reduces the possibility of residual confounding factors accounting for the observed association between PM 2.5 and adverse renal outcomes.Admittedly, there are some important limitations that should be noted.First, our synthesis of the evidence was limited to observational studies, which implies that only correlation rather than causation can be demonstrated.Second, there was significant heterogeneity among the individual studies in terms of continents, sampling period, mean PM 2.5 level, pollutant data source, eGFR formula, and meteorological parameters.Although we explored the potential sources of heterogeneity by conducting sensitivity analysis and subgroup analyses, the factors examined might account for only a partial explanation of the heterogeneity.Thus, most of the original studies did not control for important covariates, such as ethnicity/ race, seasonal variations, the use of nephrotoxic agents, underlying cardiovascular disease and some unmeasured factors, which might also play a significant role in explaining the observed heterogeneity.Third, the definition of CKD also varied significantly among individual reports, and some cross-sectional studies conducted single-time tests, potentially impacting the accuracy of the diagnosis.This scenario was frequently observed in the setting of large-scale national surveys where the feasibility of repeated measurements was limited.Forth, we acknowledge the potential influence of publication bias, particularly affecting studies on CKD incidence, which may impact the robustness of our findings.Lastly, some effect estimates were not originally calculated but were converted, which might have biased the pooled result.Therefore, based on the aforementioned limitations, it is essential to interpret the results cautiously.There is an ongoing need for further high-quality prospective studies that control for significant confounding factors, identify specific populations or regions most vulnerable to the adverse effects of PM 2.5 , and define a robust outcome for accurate diagnosis in order to establish a causal relationship between PM 2.5 exposure and CKD outcomes.

Conclusion
In conclusion, our systematic review observed that long-term exposure to PM 2.5 is associated with increased risks of CKD incidence and CKD prevalence.Hence, we emphasized that air pollution, particularly PM 2.5 , might be recognized as one of the emerging environmental CKD-related risk factors, which has detrimental effects on renal function.However, more dedicated studies are required to show causation that warrants urgent action on PM 2.5 to mitigate the global burden of CKD.

Figure 2 .
Figure 2. Forest plot displaying the pooled adjusted odds ratio of CKD incidence and long-term exposure to PM 2.5 for increments of 10 μg/m 3 .

Figure 3 .
Figure 3. Forest plot displaying the pooled adjusted odds ratio of CKD prevalence and long-term exposure to PM 2.5 for increments of 10 μg/m 3 .

Figure 4 .
Figure 4. Forest plot displaying the pooled adjusted odds ratio of incidence ESKD and long-term exposure to PM 2.5 for increments of 10 μg/m 3 .

Figure 5 .
Figure 5. Funnel plot of individual studies displaying the standard error by the log odds ratio for (A) incident CKD (B) CKD prevalence outcome in long-term exposure to PM 2.5 for increments

Table 1 .
Characteristics of the studies included in the systematic review.AERMOD AMS/EPA regulatory model, AOD aerosol optical depth, CKD chronic kidney disease, eGFR estimated glomerular filtration rate, EPA environmental protection agency, CKD-EPI chronic kidney disease-epidemiology collaboration equation, ICD the international classification of disease, MDRD modification of diet in renal disease study equation, TAQMD Taiwan air quality monitoring network, ESKD end stage kidney disease, UK United Kingdom, US United States.

Table 2 .
Primary analysis examining the association between PM 2.5 body exposure (per 10 μg/m 3 increase) and risk of adverse kidney outcomes.CKD chronic kidney disease, ESKD end-stage kidney disease.

Table 4 .
Subgroup analyses examining the association between PM 2.5 body exposure (per 10 μg/m 3 increase) and risk of CKD prevalence.CKD-EPI chronic kidney disease-epidemiology collaboration equation, MDRD modification of diet in renal disease study equation.