Air pollution and telomere length in adults: A systematic review and meta-analysis of observational studies
*Mohammad Miri
a,*, Milad Nazarzadeh
b,c, Ahmad Alahabadi
a,
Mohammad Hassan Ehrampoush
d, Abolfazl Rad
e, Mohammad Hassan Lot fi
f,
Mohammad Hassan Sheikhha
g, Mohammad Javad Zare Sakhvidi
h, Tim S. Nawrot
i,j, Payam Dadvand
k,l,m,**aCellular and Molecular Research Center, Department of Environmental Health, School of Public Health, Sabzevar University of Medical Sciences, Sabzevar, Iran
bThe George Institute for Global Health, University of Oxford, Oxford, UK
cThe Collaboration Center of Meta-analysis Research (ccMETA), Torbat Heydariyeh University of Medical Sciences, Torbat Heydariyeh, Iran
dEnvironmental Science and Technology Research Center, Department of Environmental Health, School of Public Health, Shahid Sadoughi University of Medical Sciences, Yazd, Iran
eCellular and Molecular Research Center, Sabzevar University of Medical Sciences, Sabzevar, Iran
fDepartment of Epidemiology, School of Public Health, Shahid Sadoughi University of Medical Sciences, Yazd, Iran
gResearch and Clinical Center for Infertility, Yazd Reproductive Sciences Institute, Shahid Sadoughi University of Medical Sciences, Yazd, Iran
hOccupational Health Research Center, Department of Occupational Health, School of Public Health, Shahid Sadoughi University of Medical Sciences, Yazd, Iran
iCentre for Environmental Sciences, Hasselt University, Hasselt, Belgium
jDepartment of Public Health&Primary Care, Leuven University, Leuven, Belgium
kISGlobal, Barcelona, Spain
lPompeu Fabra University, Barcelona, Catalonia, Spain
mCiber on Epidemiology and Public Health (CIBERESP), Madrid, Spain
a r t i c l e i n f o
Article history:
Received 15 March 2018 Received in revised form 20 September 2018 Accepted 26 September 2018 Available online 8 October 2018
a b s t r a c t
Telomere length (TL) has been suggested to be a surrogate for cellular ageing, and a record of cumulative inflammation and oxidative stress over life. An emerging body of evidence has associated exposure to air pollution to changes in TL. To date there is no available systematic review of literature on this association.
We aimed to systematically review and conduct meta-analysis of published studies on the relationship between air pollution and TL in adults. Electronic databases were systematically searched for available English language studies on the association between air pollution and TL published up to 1 July 2018.
Meta-analyses were conducted following MOOSE guidelines. The heterogeneity in the reported associ- ations was assessed using Cochran's Q test and quantified as I2index. Publication bias was assessed using Egger's regression. Our search identified 19 eligible studies including 11 retrospective and eight pro- spective studies of which, four had excellent quality, ten had good quality andfive had fair quality. Meta- analysis result of two studies on long-term exposure to PM2.5showed an inverse association between these exposures and TL (for 5mg/m3PM2.5e0.03 95% CI;0.05,0.01). Meta-analysis of short-term exposure to PM2.5with three studies and Polychlorinated Biphenyls (PCBs) with two studies revealed a direct association between these exposures and TL (0.03 95% CI; 0.02, 0.04 and 0.10 95% CI; 0.06, 0.15 respectively). No statistically significant relationship between exposure to PM10and polycyclic aromatic hydrocarbons (PAHs) exposure and TL were observed. We observed suggestive evidence for associations between air pollution and TL with potentially different direction of associations for short- and long-term exposures.
©2018 Published by Elsevier Ltd.
*This paper has been recommended for acceptance by Haidong Kan.
*Corresponding author. Sabzevar University of Medical Sciences, PO Box 319, Sabzevar, Iran.
**Corresponding author. ISGlobal, Doctor Aiguader, 88, 08003 Barcelona, Spain.
E-mail addresses:[email protected](M. Miri),[email protected](P. Dadvand).
Contents lists available atScienceDirect
Environmental Pollution
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m/ l o ca t e / e n v p o l
https://doi.org/10.1016/j.envpol.2018.09.130 0269-7491/©2018 Published by Elsevier Ltd.
Environmental Pollution 244 (2019) 636e647
1. Introduction
Telomere is a ribonucleoprotein complex that caps the end of chromosomes. It protects chromosomes from degradation and end- to-end fusion to ensure genome stability and to prevent the loss of genetic information (Blackburn, 1991). Telomere length (TL) pro- gressively shortens as cells undergo division because of the inability of DNA polymerase to replicate the lagging DNA strand to its terminus (Allsopp et al., 1992;McCracken et al., 2010). Hence, TL has been suggested to be a marker of cellular aging (Levy et al., 1992). Shortened TL has also been proposed to be a marker of oxidative stress and systemic inflammation (Mather et al., 2007;
Samani and van der Harst, 2008) and been associated with age- related diseases, such as cardiovascular disease (Haycock et al., 2014) and cancer (Wentzensen et al., 2011).
Air pollution is the major environmental contributor to morbidity and mortality worldwide (Nawrot et al., 2011;Miri et al., 2016a;Miri et al., 2016b;Provost et al., 2016;Sheehan et al., 2016).
Systemic inflammation and oxidative stress have been suggested to be two main mechanisms underlying adverse health effects of air pollution (Chuang et al., 2007;Colicino et al., 2017). An emerging body of evidence has associated exposure to air pollution to changes in TL (Hou et al., 2012;Bassig et al., 2014;Mitro et al., 2016;
Pieters et al., 2016;Walton et al., 2016). Such an association is of importance because it could lead to better understanding of the pathways through which air pollution affects health. Currently, however, a systematic review or meta-analysis of the available evidence on this association is non-existent. We therefore aimed to systematically review and conduct meta-analysis of published studies on the relationship between air pollution and TL in adults.
2. Methods 2.1. Search strategy
The review was conducted according to MOOSE guidelines (Stroup et al., 2000). Electronic bibliographic databases including Embase and PubMed were applied to search for articles on the association between air pollution and telomere length that were published until 1 July 2018. Search strategy included terms corre- sponding to air pollution in relation to TL without time and lan- guage restrictions based on the keywords detailed in Table 1.
Finally, manual searching was carried out based on the reference list of articles found in thefirst step to identify those articles that were not found through aforementioned search. All studies on the association between air pollution and TL conducted in adults from general population or occupational studies were included in our review. The studies in children were exclude from our review. In addition, we considered all the observational study design including cohort, cross-sectional and case-control studies.
The following information was extracted using a data extraction
form: year of publication, geographical location, mean age and standard deviation (SD), exposure definition, sample size, type of pollutants, concentration of pollutant, study design, exposure length, assay methods, the mean and SD of TL, assay coefficient of variation, correlation coefficients, measure of association (odds ratio (OR), relative risk (RR), etc), and confounding factors (Table 2).
The quality of individual articles included in the review was assessed using an 11 criteria framework (Supplementary Table S1) adapted from previous systematic reviews (Lachowycz and Jones, 2011;de Keijzer et al., 2016;Gascon et al., 2016). For each of the 11 criteria, a study could obtain a point between 0 and 2 with the possible maximum total point for each study being 22. Accordingly, the sum of the points for each study was converted to the per- centage of the possible maximum total point obtained by that study (Lachowycz and Jones, 2011;de Keijzer et al., 2016;Gascon et al., 2016). We used the following categorization to classify the quality of the studies based on their percentage:>81% as excellent quality, 61%e80% as good quality, 41%e60% as fair quality, 21%e40% as poor quality, and<20% as very poor quality (Lachowycz and Jones, 2011;
Gascon et al., 2015;de Keijzer et al., 2016). The quality assessment was carried out independently by two reviewers and the final agreement was made by consensus.
2.2. Statistical analysis
We converted different units of air pollutants to the same unit for every pollutant. The heterogeneity between studies was assessed using Cochran's Q test and quantified applying I2(Higgins and Thompson, 2002;Huedo-Medina et al., 2006;Ioannidis et al., 2007). We used meta-analysis to abstract the combined estimates for association between each air pollutant and TL. Given the limited statistical power of Cochran's Q test, as conservative approach, we decided to conduct random effect meta-analysis for all evaluated associations. Publication bias was assessed using Egger's regression (Egger et al., 1997; Peters et al., 2006). Statistical analyses were performed using Stata software version 14 (Stata Corp LP, College Station, Texas). When possible, we ran our analysis based on long and short-term exposure. The exposure time less than 30 days was considered as short-term and more than it was taken as long-term exposure (EPA, 2011).
3. Results
Our initial search in PubMed and Embase identified 264 unique articles (Fig. 1). After screening the titles, 68 articles remained for abstract screening of which 30 full-text articles assessed for eligi- bility. After review of full-texts, 19 relevant articles were included in the review and meta-analysis (Hoxha et al., 2009;McCracken et al., 2010;Pavanello et al., 2010;Shin et al., 2010;Dioni et al., 2011;Li et al., 2011;Hou et al., 2012;Bassig et al., 2014;Shan et al., 2014;
Wong et al., 2014a; Wong et al., 2014b; Bijnens et al., 2015;
Table 1 Search strategy.
PubMed, Search until 1 July 2018
(“T/S ratio”OR“telomerase Expression”OR“telomeric”OR“Bimolecular Marker”OR“Telomerase Methylation”OR“telomerase”OR“Telomere activity”OR“Telomere attrition”OR“Telomere shorting”OR“Telomere length”OR“telomere”[Mesh](AND(“Volatile organic compounds”OR“Benzene”OR“BTEX”OR“Carbon monoxide”OR
“ozone”OR“dust”OR“SO2”OR“NO2”OR“PCB”OR“PAH”OR“poly colour biphenyl”OR“Poly aromatic hydrocarbon”OR“black carbon”OR“Motor vehicle”OR“diesel exhaust”OR“PM10”OR“PM2.5”OR“Particulate matter”OR“vehicle emissions”OR“Traffic related”OR“Outdoor exposure”OR“Air Pollution”[Mesh])
EMBASE, Search until 1 July 2018
(“T/S ratio”OR“telomerase Expression”OR“telomeric”OR“Bimolecular Marker”OR“Telomerase Methylation”OR“telomerase”OR“Telomere activity”OR“Telomere attrition”OR“Telomere shorting”OR“Telomere length”OR“telomere”).afAND(“Volatile organic compounds”OR“Benzene”OR“BTEX”OR“Carbon monoxide”OR
“ozone”OR“dust”OR“SO2”OR“NO2”OR“PCB”OR“PAH”OR“poly colour biphenyl”OR“Poly aromatic hydrocarbon”OR“black carbon”OR“Motor vehicle”OR“diesel exhaust”OR“PM10”OR“PM2.5”OR“Particulate matter”OR“vehicle emissions”OR“Traffic related”OR“Outdoor exposure”OR“Air Pollution”).af
M. Miri et al. / Environmental Pollution 244 (2019) 636e647 637
Table 2
Main characteristics of studies included in the systematic review.
Authors (date)
Location Study population Sample size
Age (yr) Definition of exposure
Exposure, concentration in mg/m3
Study design and Exposure length
Telomere length (TL)
Measure of association type
Measure of association (95% CI)
Interpretation
General population exposure McCracken
et al., 2010
Massachusetts, USA
Normative Aging Study (NAS):
Population of males, (all never smokers)
165 73.6±7.1 Estimated based on a
spatiotemporal model calibrated with BC measurements from 82 locations within the study area
Annual black Carbone (BC):
0.32±0.2mg/m3
Cohort, (365 days before each blood drawn)
TL¼1.25±1.42 Relative unit
Percentage of regression coefficient
BC¼ 7.6%
(12.8%,2.1%)
1 IQR (BC)¼0.25mg/
m3
Shin et al., 2010
South Korea General healthy population
84 56.2±7.0 Serum levels of POPs
POPs Cross- sectional
(6 months)
TL¼2.02±0.32 Relative unit
Correlation coefficient
Correlation coefficients from about 10.25 to 10.35
Not reported
Shan et al., 2014
Sichuan, China Rural women 25 59 (38e85) Monitoring of personal, indoor and home- outdoor levels of PM2.5and BC
24-h personal PM2.5in low vs high exposed: 39±11 vs 101±37mg/m3 24-h personal BC in low vs high exposed:
2.6±1.8 vs 14.9±11.2mg/
m3
Cross-sectional (24-h exposure)
Low vs high exposed group TL¼5.4±4.5 vs 3.3±2.3 Relative unit
Mean difference
Low vs high exposed group for PM2.5: 43 (113 to 28) and for BC not reported
Between the low and high PM exposure group
Bijnens et al., 2015
Belgium Twins of Caucasian origin born between 1975 and 1982
211 27.5 (19e40) Distance to main roads
Traffic-related air pollution
Prospective, TL¼12.7±6.7 Kbp (absolute)
Regression coefficient
Distance to road¼5.32%
(1.90%, 8.86%) increase TL
Per doubling in the distance to major roads
Mitro et al., 2015
USA General population
aged between 20 and 60 years
1330 20e60 Serum levels of
PCBs, Dioxins and Furans
PCBs Dioxins furans
Cross-sectional (1 year)
TL¼1.17±0.05 Relative unit
Percent of regression coefficient
PCB¼3.74% (2.10%, 5.40%) increase in TL
Per doubling of exposure Pieters et al.,
2015
Flanders, Belgium
Non-smokers elderly
166 70.6±4.7 The annual exposure levels of PM2.5 were estimated using Kriging combined with a dispersion model
Annual PM2.5:
21.1±1.76mg/m3 Cross-sectional, (last year, last month and last week exposure)
TL¼1.04±0.05 relative unit
Regression coefficient
PM2.5(1yr)¼ 0.032 (0.053,0.011), PM2.5(1 month)¼0.028 (0.015, 0.040), for last week exposure no significant association were observed.
Per 5mg/m3increase in average PM2.5
Scinicariello et al., 2015
USA General adult
population
2431 42.81±0.52 Serum levels of PCBs, PCDDs and PCDF
SPCBs¼82.13±2.55(ng/
m3)
SPCDDs¼370.9±14.2(pg/
g)
PCDF¼7.70±0.29
Cross-sectional (Two periods:
1999e2000 and 2001e2002)
TL¼1.06±0.021 relative unit
Percent of regression coefficient
SPCBs¼11.63% (95% CI:
6.18e18.53) longer LTLs.
No significant associations reported for the sum of PCDD.
difference % Quartile 4 to quartile 1
Xia et al., 2015
Shangha, China Patients with Type 2 diabetes
35 65±8 Monitored
ambient level of size-fractionated PM (0.25 e10mm), SO2, NO2, CO and O3
PM1: 39.6±21.7mg/m3, PM2.5: 6.6±4.5mg/m3 PM10: 21.1±18.5mg/m3 SO2: 12.4±14.8mg/m3 NO2: 46.4±23.4mg/m3 CO: 768.9±252.9mg/m3 O3: 76.8±41.3mg/m3
Cohort, (Lag 0, 1, 2, 3 and 4e7 days prior blood drawn)
TL¼0.95±0.041 relative unit
Regression coefficient
PM2.5¼0.11 (- 0.97, 1.19),
SO2¼0.48 (1.01, 1.97), NO2¼0.25 (0.76, 1.26), CO¼0.02 (1.16, 1.20), O3¼0.28 (0.76, 1.32)
IQR (PM2.5)¼5.0 IQR (SO2)¼11.1 IQR (NO2)¼26.2 IQR (CO)¼494.2 IQR (O3)¼51.3
M.Mirietal./EnvironmentalPollution244(2019)636e647638
Ling et al., 2016
China College students 666 21.4±1.2 Urinary level of PAHs metabolites
PAHs metabolites:
2-OHFlu: 0.413±1.838mg/
g, 1-OHPyr:
0.020±4.065mg/g, 1-OHNap:
0.068±7.195mg/g, 2-OHNap:
0.650±2.099mg/g, 1-OHPhe:
0.084±3.986mg/g, 2-OHPhe:
0.175±1.838mg/g, 3-OHPhe:
0.190±1.856mg/g, 4-OHPhe:
0.007±4.393mg/g, SPAH metabolites:
1.885±1.841mg/g
Cohort, (1 year)
Sperm TL¼ 0.951±1.324 Relative unit
Regression coefficient
1-OHPyr¼ 0.414 (0.774,0.054) 1-OHNap¼ 0.081 (0.149,0.014) For other pollutants, not significant relationship were observer, and regression coefficient were not reported
Per 1 unit increase in metabolite of PAHs
Callahan et al., 2017
Alabama, USA Anniston Community Health Survey (ACHS) participants
559 >40 40e60
<60
Serum levels of PCBs
SPCBs:
<1403.61 ppt 1403.61e3141.40 ppt 3141.40e7345.35 ppt
>7345.35 ppt
Cross-sectional (2 years)
Relative unit Regression coefficient
White participants:
SPCBs¼0.0809 (0.0119, 0.145)
African American participants: PCBs were associated with longer relative LTL among those over age 64 only
Per doubling of exposure
Occupational exposure Hoxha et al.,
2009
Milan, Italy Exposed: street traffic officers and control: office workers
134 <30 y:
39%
30-40 y:
37%
40 y: 24%
Personal passive samplers
Airborne benzene:
Referents:
13.0 (2.0e115.1)mg/m3 Traffic officers: 31.8 (9.0 e315.7)mg/m3 Airborne toluene:
Referents: 43.4 (6.0 e368.0)mg/m3
Traffic officers: 128.7 (24.4 e1710.7)mg/m3
Case-control (One work shift 7 h)
TL in Exposed¼1.10 (95%CI 1.04e1.16) TL in
Control¼1.27 (95%
CI 1.20e1.35) Relative unit [p<0.001]
Present of regression coefficient
Benzene: 6.39%
(10.39%,2.09%), .
Toluene: 6.2%
(1.70%,10.4%),
IQR(benzene)¼11.2 IQR(Toluene)¼25.7
Pavanello et al., 2010
Poland Exposed: Coke oven workers,
Nonplused: control matched group
92 Median (range) Expoused¼36 (20e59) Control¼38 (21e58
Urinary levels of PAHs metabolites
PAHs (metabolites of 1- Pyrenol)
Median (range) Espoused¼3.09 (0.41 e7.48) (lmol/mol creatinine)
Non exposed¼0.09 (0.01 e0.40) (lmol/mol creatinine)
Exposed and control compare results (after at least 3 consecutive working days)
Median(range) -TL in Exposed¼0.99 (0.31e3.00) relative unit TL in Control¼1.20 (0.43e2.12)
Regression coefficient
Control group vs Coke oven workers:
0.002 (0.345) vs 0.013 (0.042)
Per 1 unit increase in metabolite
Dioni et al., 2011
Brescia, Italy Steel workers, 63 44(25e55) Monitored ambient levels PM10and PM1
PM10¼262±272mg/m3 PM1¼8.0±7.7mg/m3
Cross -sectional, PM10and PM1
exposure for 3 days
TL based line:
1.23±0.28 relative units
TL post exposure:
(1.43±0.51;
p<0.001) relative units
Regression coefficient
PM10: 0.01 (0.003, 0.017),
PM1: 0.26 (0.009, 0.51),
Per 10mg/m3 increase in average PM10and PM1
Li et al., 2011Sweden Rubber industry workers
157 38 (19e65) Personal samplers of N- nitrosamines (in the air) and
N-nitrosamines: 1.3 (0.1 e22) ng/l
PAH (1 -hydroxypyrene):
0.14 (0.0002e0.85)mmol/
Cross-sectional (1 year)
Median (range):
TL¼0.71±0.16 e1.3,
Regression coefficient
N-nitrosamines:¼ 10, (17,1.9)
PAH: 11, (168,146)
Per 1 unit increase in metabolite
(continued on next page)
M.Mirietal./EnvironmentalPollution244(2019)636e647639
Table 2(continued) Authors (date)
Location Study population Sample size
Age (yr) Definition of exposure
Exposure, concentration in mg/m3
Study design and Exposure length
Telomere length (TL)
Measure of association type
Measure of association (95% CI)
Interpretation
measurement urinary level PAHs and CS2 metabolite
mol cratinine
CS2 (2-thiothiazolidine):
24(1.7e690)mmol/mol creatinine
CS2:0.21, (0.41,0.0086)
Hou et al., 2012
Beijing, China Espoused: Truck Drivers
Unexposed: indoor office workers
120 Truck driver:
33.5±5.7 Office worker:
30.27±7.96
personal monitoring of PM2.5and elemental carbon, monitoring of Ambient PM10
levels
Personal PM2.5for truck drivers' vs office workers:
126.8±68.8a vs 94.6±64.9mg/m3 Personal EC for truck drivers' vs office workers:
17.3±6.7a vs 13.1±4.0mg/m3 Ambient PM10for truck drivers' vs office workers:
Examination day:
116.7±50.2 vs 123.5±50.1mg/m3 - 1 day: 121.5±47.8 vs 119.5±51.2mg/m3 - 1e2 days:121.6±38 vs 119.3±40.3mg/m3 - 1e5 days: 119.5±26.9vs 118.2±25.6mg/m3 - 1e14 days: 119.9±18.7 vs 121.7±17.8mg/m3
Espoused econtrol, Personal PM2.5
and EC exposure for 8- h.
Ambient PM10exposure:
Examination day, 1 day, 1e2 days, 1e5 days, 1e7 days, 1e10 days, 1e14 days
TL in Office Worker¼0.79 (0.67, 0.93) TL inTurk Driver¼0.87(0.74, 1.03)
Relative unit
Mean difference And Regression coefficient
Compare TL mean; truck driver vs office worker:
0.87 (0.74; 1.03) vs 0.79 (0.67; 0.93); P¼0.002 And PM2.5: 5.2% (1.5%
e9.1%), p¼0.007 for each IQR increase.
Personal EC: 4.9% (1.2%
e8.8%), p¼0.01 for each IQR increase.
PM10:
- Examination day: 7.7%
(3.7%e11.9%), p<0.001 - 1-day: 8.4% (4.0%
e13.0%), p<0.001 - 1e2 days: 8.1% (3.1%
e13.3%), p¼0.002 - 1e5 days: 1.3% (6.3%
to 4.1%), p¼0.64 - 1e14 days: 9.9%
(17.6%,1.5%),p¼0.02
Per 5mg/m3increase in PM2.5and 10mg/
m3increase in PM10
Bassing et al., 2014
Shanghai, China
Exposed: Worker in a factory that used benzene Control: Two workplaces in the same geographic area that did not use benzene or other chemicals associated with bone marrow toxicity
43 35±5 Organic vapor
passive dosimetry badges
Benzene (ppm): overall:
62.7±70.8
Low exposed and high exposed vs control 14.0±68.9 and 109.2±673.0 vs n/a
Cross-sectional, (full work shift onfive separate days exposed to benzene
TL¼ Exposed:
1.32±0.20 Control:
1.26±0.17 Relative unit
Mean difference
Exposed vs control 1.32±0.20 vs 1.26±0.17 Workers exposed to
>31 ppm of benzene exposure had an increase in TL (1.37±0.23) compared with unexposed workers.
Adjusted estimate for age and sex (P- value¼0.21).
Exposed vs non- exposed and worker with highly exposed (more than 31 ppm of benzene exposure) vs low exposed
Wong et al., 2014
Massachusetts, USA
Male Boilermaker workers
87 42.7±13.1 Toenail heavy metal concentrations (as marker of long-term particulate matter exposure)
Cumulative PM exposure, Toenail metal
concentrations: (mg/g) Pb¼0.67±1.01 As¼0.18±0.14 Cd¼0.01±0.06 Mn¼1.35±2.14
cohort, (18 months)
TL¼0.63±0.17 Relative unit
Regression coefficient
No statistically significant association was found between the cumulative PM exposure construct, with telomere length
Wong et al., 2014
Massachusetts, USA
Male Boilermaker workers
48 39.3±12.8 Monitored ambient level of PM2.5
Career cumulative PM2.5:713.7±1457.9 mg/
m3hr
Year cumulative PM2.5: 74.0±77.3 mg/m3hr Month cumulative PM2.5: 8.5±10.2mg/m3hr
Cohort, career, 1 yr, and 1 mo prior to each blood draw
TL¼0.65±0.16 Relative unit
Regression coefficient
Career:b: 0.105 (0.24, 0.03),
Year:b: 0.11 (0.265, 0.045),
Month:b: 0.2 (0.4, 0.005)
Per 5mg/m3increase in cumulative PM2.5
M.Mirietal./EnvironmentalPollution244(2019)636e647640
Scinicariello and Buser, 2015;Xia et al., 2015;Ling et al., 2016;Mitro et al., 2016;Pieters et al., 2016;Callahan et al., 2017;Xu et al., 2018).
Table 2 presents the summary characteristics of the included studies. Overall, 10 studies were based on population-based sam- ples (McCracken et al., 2010;Shin et al., 2010;Shan et al., 2014;
Bijnens et al., 2015;Scinicariello and Buser, 2015;Xia et al., 2015;
Ling et al., 2016;Mitro et al., 2016;Pieters et al., 2016;Callahan et al., 2017) and eight were based on occupational samples (Hoxha et al., 2009;Pavanello et al., 2010;Dioni et al., 2011;Li et al., 2011;Hou et al., 2012;Bassig et al., 2014;Wong et al., 2014a;Wong et al., 2014b;Xu et al., 2018). Eleven studies had retrospective (4964 Participants) (Hoxha et al., 2009;Pavanello et al., 2010;Shin et al., 2010;Dioni et al., 2011;Li et al., 2011;Bassig et al., 2014;Shan et al., 2014;Scinicariello and Buser, 2015;Mitro et al., 2016;Pieters et al., 2016;Callahan et al., 2017) and eight had prospective design (1395 participants) (McCracken et al., 2010;Hou et al., 2012;Wong et al., 2014a;Wong et al., 2014b;Bijnens et al., 2015;Xia et al., 2015;Ling et al., 2016;Xu et al., 2018). Six studies were conducted in the Unite State of America, seven in Asian countries (mainly China), and the rest were conducted in Europe. No study was conducted in Africa or Latin-America.
The quality of the studies included in the review ranged be- tween 45.4% and 90.9% (seeSupplementary Tables S2 and S3). Of studies included in our meta-analyses, two were classified as excellent quality (Hou et al., 2012;Ling et al., 2016), seven as good quality (Wong et al., 2014a;Scinicariello and Buser, 2015;Xia et al., 2015;Mitro et al., 2016;Pieters et al., 2016;Callahan et al., 2017;
Xu et al., 2018) and two as fair quality (Dioni et al., 2011;Li et al., 2011).
A wide range of air pollutants including PM10, PM2.5, PM1, NO2, NO, NOx, PAHs, PCBs, benzene, toluene, POPs, black carbon (BC) and elemental carbon (EC) were evaluated by the reviewed studies.
Four studies investigated association between PM10 and TL including one based on long-term (Walton et al., 2016) and three based on short-term exposure (Dioni et al., 2011;Hou et al., 2012;
Xia et al., 2015). In six studies the relationship between PM2.5and TL was investigated of which three evaluated long-term (Wong et al., 2014a; Pieters et al., 2016;Walton et al., 2016) and three investigated short-term associations (Hou et al., 2012;Shan et al., 2014;Xia et al., 2015). Four studies evaluated the relationship be- tween PAHs and TL, from which three were based on short-term (Li et al., 2011;Ling et al., 2016;Xu et al., 2018) and one was based on long-term exposures (Pavanello et al., 2010). Three studies used 1- hydroxypyren (Li et al., 2011;Ling et al., 2016;Xu et al., 2018) while the other one used 1-Pyrenol (Pavanello et al., 2010) to assess exposure to PAHs. PCBs were investigated by three studies on short-term effects (Scinicariello and Buser, 2015;Mitro et al., 2016;
Callahan et al., 2017). Other air pollutants were investigated just by one study. TL measurement technique for all studies was quanti- tative polymerase chain reaction (qRT-PCR) technique. Two studies used buccal cell for measuring TL (Shan et al., 2014;Walton et al., 2016), one (Bijnens et al., 2015) used placental tissue, one (Ling et al., 2016) used sperm, and the rest used Leukocytes for measuring TL. A relative unit (Telomere/Single gen ratio) for TL measurement was used in all of the papers exceptBijnens et al., (2015) study in which absolute unit was used. The study by Bijnens et al., (2015) was not included in meta-analysis, because it was the only study that evaluated distance to major roads in rela- tion with telomere length.
3.1. Exposure to PAHs
Based on Cochran's Q test, heterogeneity between studies were observed (P-value¼0.03) with an I2of 71.1%. Meta-analysis of the three studies using 1-hydroxypayren (Li et al., 2011; Ling et al., Xuetal., 2018SwedenExposed:Male asphaltworkers, Unexposed:control group
167Asphalt workers:43 (23e59) Control:46(24 e62) Measurement urinarylevel PAHsmetabolite
Exposed: Pre-working1-OH-PYR (mmol/molcreatinine): 0.041(0.017e0.13) Pre-working2-OHePH (mmol/molcreatinine): 0.13(0.063e0.52). Post-working2-OHePH (mmol/molcreatinine): 0.21(0.076e0.69). Post-working1-OH-PYR (mmol/molcreatinine): 0.076(0.021e0.27) Control: Pre-working1-OH-PYR (mmol/molcreatinine): 0.028(0.0087e0.091) Post-working1-OH-PYR (mmol/molcreatinine) 0.028(0.0092e0.091) Espoused econtrol,pre- workingon Monday morningand post-working onThursday afternoonafter 4days working
TL¼1.1(0.73e1.7) RelativeunitRegression coefficient1-OH-PYR:b:0.067 (0.0038,0.14) 2-OH-PYR:b:0.075 (0.0046,0.15) Per1unitincrease inmetabolite
M. Miri et al. / Environmental Pollution 244 (2019) 636e647 641
2016;Xu et al., 2018) did not show any statistically significant as- sociation between exposure to this pollutant and TL (Fig. 2).
3.2. Exposure to PCBs
There were no indication of between-study heterogeneity in the reported associations for PCBs (Cochran's Q test P-value of 0.56 and I2of 0.0%). An increase of 1 ng/m3in PCBs was associated with 0.10 (95% CI; 0.06, 0.15) increase in TL (Fig. 2).
3.3. Exposure to PM2.5
3.3.1. Combined exposure
Meta-analysis of the association between combined short- and long-term exposure to PM2.5and TL included four studies (Hou et al., 2012; Wong et al., 2014a; Xia et al., 2015; Pieters et al., 2016). We observed indications of heterogeneity between studies (Cochran's Q test P-value¼0.00) with an I2of 81.2%. Meta-analysis did not show any statistically significant association between combined short- and long-term exposure to PM2.5and TL (Fig. 2).
3.3.2. Short-term exposure
Three studies (Hou et al., 2012;Xia et al., 2015;Pieters et al.,
2016) were included in the meta-analysis for the short-term exposure to PM2.5that did not show any indication of heteroge- neity (Cochran's Q test p-value of 0.49 and I2of 0.0%). An increase of 5mg/m3in short-term exposure to PM2.5was associated with 0.03 (95% CI; 0.02, 0.04) increase in TL (Fig. 3).
3.3.3. Long-term exposure
Two studies (Wong et al., 2014a;Pieters et al., 2016) with no indication of heterogeneity (Cochran's Q test p-value of 0.33 and I2 of 0.0%) were included in the meta-analysis for long-term exposure to PM2.5 (Fig. 3). An increase of 5mg/m3 in this exposure was associated with0.03 (95% CI;0.05,0.01) decrease in TL.
3.3.4. Sensitivity analysis
For the short-term PM2.5exposure, there were two population- based and one occupational study. We repeated the meta-analysis for this exposure by removing the occupational study (Hou et al., 2012) study. The results of the meta-analyses based on two population-based studies (Xia et al., 2015;Pieters et al., 2016) were consistent with those of the main analysis including all three studies (seeFig. S1of Supplemental material).
Fig. 1.Searchflow diagram.
M. Miri et al. / Environmental Pollution 244 (2019) 636e647 642
Fig. 2.Combined estimate of TL associated with 1mg/m3increase in exposure to polycyclic aromatic hydrocarbons (PAHs) and polychlorinated biphenyls (PCBs) and 5mg/m3in- crease in short- and long-term exposure to PM2.5.
M. Miri et al. / Environmental Pollution 244 (2019) 636e647 643
Fig. 3.Combined estimate of TL associated with 5mg/m3increase in short-term and long-term exposure to PM2.5and 10mg/m3increase in short-term exposure PM10. M. Miri et al. / Environmental Pollution 244 (2019) 636e647
644
3.4. Exposure to PM10
The relationship between shot-term exposure to PM10and TL were investigated based on two studies (Dioni et al., 2011; Hou et al., 2012) (Fig. 3) that showed heterogeneity between their re- ported associations (Cochran's Q test P-value¼0.00 and I2of 90%).
According to Fig. 3, no statistically significant association was observed between short-term exposure to PM10and TL. There was not any study for long-term exposure to PM10and TL in adults.
3.5. Studies not included in meta-analyses
Studies investigating a pollutant that was not evaluated any other study (McCracken et al., 2010; Bassig et al., 2014; Bijnens et al., 2015), those that did not reportb(Shin et al., 2010; Shan et al., 2014;Wong et al., 2014b) and had overlap in terms of par- ticipants with other studies (Mitro et al., 2015) were excluded from our meta-analysis.McCracken et al. (2010)(McCracken et al., 2010) reported an inverse association between BC and TL only in males (7.6% (CI:95%;12.8%,2.1%).Shin et al. (2010)(Shin et al., 2010) reported correlation coefficients ranging from 10.3 to 10.4 between persistent organic pollutants (POPs) and TL.Shan et al. (2014)(Shan et al., 2014) reported mean difference of43 (CI:95%,113 to 28) in TL between the low (39±11mg/m3) and high (101±37mg/m3) PM2.5exposure group.Bijnens et al. (2015) (Bijnens et al., 2015) investigated relation between traffic-related air pollution and TL and reported that with doubling increase distance to nearest major roads, the percentage change in TL was 5.32% (CI:95%, 1.90%, 8.86%).
A study byBassing et al. (2014)(Bassig et al., 2014) reported a direct association between exposure to benzene and TL (exposed vs.
control: 1.32±0.20 vs. 1.26±0.17). AlsoHoxha et al., 2009(Hoxha et al., 2009) reported that exposure to airborne benzene (6.39%;
95% CI: 10.39%,2.09%) and toluene (6.2%; 95% CI: 1.70%,10.4%) was associated with shorter TL.Wong et al. (2014)(Wong et al., 2014b) did notfind any association between the estimated cumu- lative PM2.5exposure based on toenail levels of a number of metals (Pb, As, Cd, Mn) and TL in boilermaker male workers.Mitro et al.
(2015)(Mitro et al., 2015) reported that exposure to PCBs (0.04;
95% CI: 0.02, 0.05) was associated with longer TL.
3.6. Publication bias
Because of small number of studies for each air pollutant, we included all studies in the same Egger's model. As presented in Fig. 4, the symmetry of funnel plot supported the absence of pub- lication bias in our review (beta for egger's regression¼0.89, standard error¼0.54, P-value¼0.11).
4. Discussion
To our knowledge, this is the first systematic review and meta-analysis on the association between air pollution and TL in adults. It identified 19 articles published between 2009 and July 2018 on this association of which, four had excellent quality, ten had good quality andfive had fair quality. We conducted meta- analyses for estimating combined associations for PAHs, PCBs, PM2.5(short-term and long-term exposure) and PM10exposure.
We observed a negative association between exposure to long- term PM2.5exposure and TL. In contrast, for exposures to PCBs and short-term PM2.5exposure, we observed direct associations with TL. The combined associations between PM10 (short-term exposure) and PAHs exposure and TL were not statistically significant.
We conducted meta-analysis only for PM2.5, PM10, PCBs and PAHs, because other air pollutants were examined only in one study, investigated in children, or did not reportbestimates (see Table 2). Because of the small number of studies available for air pollutants, the power to detect associations was likely limited and further studies are required to support ourfindings.
We did not observe any statistically significant association be- tween PAHs metabolites and TL based on three studies. Exposure to PAHs generally occurs as mixtures that can be absorbed through the skin, respiratory tract, and gastrointestinal tract (Boogaard, 2008). PAHs, through a series of metabolic processes, form active metabolites that can react with DNA to produce DNA adducts. PAH metabolites may also trigger an elevation in reactive oxygen species (Kalmbach et al., 2013;Ling et al., 2016).
Our meta-analyses showed a positive association between PCBs and TL. Exposure to PCB compounds has been associated with increased risk of cancers of skin, liver, intestines, and biliary tract (Atsdr, 2000). Other POPs than PCBs such as dioxins and furans, have also been associated with increase in TL (Senthilkumar et al., 2012;Mitro et al., 2016).Mitro et al. (2016) (Mitro et al., 2016). Considering that many dioxin-associated cancers are also associated with longer TL, these results may provide insight into the mechanisms underlying PCB and dioxin- related carcinogenesis.
Air pollution could potentially influence TL through two main mechanisms: increasing the replication rate of cells and enhancing the extent of telomere loss during each replication. Epidemiologic studies, human exposure experiments, animal models, and in vitro studies have provided evidences about inflammatory effects of short-term and long-term exposures to air pollution (Zhang et al., 2013;Eze et al., 2015; Martens and Nawrot, 2016;Khreis et al., 2017). Systemic inflammation induced by air pollution could in- crease the numbers of leukocytes (Chen and Schwartz, 2008). There is also supportive evidence that diesel exhaust exposure can result in oxidative DNA damage (Vattanasit et al., 2014). Human panel studies have shown that guanine oxidation, in particular, is asso- ciated with particle exposure (McCracken et al., 2010). Oxidative stress mediated by PM could arise from direct contact with reactive oxygen species on the surface of particles deposited in the lungs, soluble compounds such as transition metals or organic com- pounds entered to the bloodstream, or activation of inflammatory cells generating reactive oxygen and nitrogen species (Martens and Nawrot, 2016;Walton et al., 2016). The GGG triplets on telomeres are highly sensitive to hydroxyl radicals and oxidative stress could be a major cause of telomere shortening independent of shortening due to incomplete end replication when stem cells are divided to produce new leukocytes (Martens and Nawrot, 2016). A causal interpretation of the association between atmospheric pollutants and TL, therefore, relies on the plausibility that these pollutants could be involved in at least one of these two mechanisms.
Fig. 4.Funnel plot of all studies that investigated air pollution and TL.
M. Miri et al. / Environmental Pollution 244 (2019) 636e647 645
Long and short-term exposure to air pollution can change the TL in different directions. Some studies have reported that long-term exposure to particulate matter (PM) is associated with reduce TL (Wong et al., 2014a;Pieters et al., 2016). However, experimental studies show that during acute inflammation, telomere length in inflammatory cells undergoes a transient increase, which is thought to contribute to the proliferative capacity and clonal expansion necessary to build up an efficient inflammatory response (Weng et al., 1997). Moreover,Dioni et al., 2011(Dioni et al., 2011) and Hou et al., 2012(Hou et al., 2012) reported that short-term exposure to PM is associated to fast increase in blood TL. Taken together, these observations suggest that short-term PM exposures might produce an acute increase in TL, which may participate in sustaining the inflammatory mechanisms associated with PM health effects. Consistently, our meta-analysis of short-term exposure indicated a positive significant association between short-term exposure to PM2.5 and TL, whereas for long-term exposure to PM2.5a negative association was observed. For short- term exposure to PM10, we did not observe any association with TL. However,Dioni et al. (2011)(Dioni et al., 2011) andHou et al.
(2012)(Hou et al., 2012) reported significant association between short-term exposure to PM10and TL. Particulate matters can carry other species like PAHs, benzene, volatile organic compounds and PCBs that are known as toxicological agents (Riddle et al., 2007;
Miri et al., 2017;Nikoonahad et al., 2017).
Our observed heterogeneity in reported associations between air pollution and TL by different studies could have raised from differences in the population exposed to air pollutants (occupa- tional exposure and population-based exposure), type of exposure (short- vs. long-term), exposure assessment methods (models, personal monitoring, and biomarkers), tissues used to extract telomere (buccal cell, placental tissue, sperm, leukocytes), and constituents of particulate air pollutants in different areas. Because of the small number of published studies for each pollutant, it was not possible to stratify the meta-analyses based on these differ- ences or use meta-regression methods to evaluate the contribution of these factors to our observed heterogeneity.
4.1. Public health relevance of ourfindings of long-term PM2.5and telomere length
As the combined studies for long-term PM2.5exposure used a real-time PCR method we are not able to provide absolute values of telomere lengths to estimate the effects of the combined decline based on absolute values as measured, for instance, using terminal restriction fragments. Nevertheless, an estimation can be based on available data in the literature. In young adulthood telomeres are on average 8 kb (Nawrot et al., 2004) and the annual telomere loss in adult leukocytes is between 32.2 and 45.5 bp (Muezzinler et al., 2013). Our combined estimates indicated that for each increment of 5mg/m3long-term exposure to PM2.5, telomeres were 3% shorter.
This reduction of 3% corresponds to a reduction of 240 kb indicating that this effect-size of 3% shortening is equivalent to a loss of 5.3e7.4 years (based on telomere attritions of 32.2e45.5 bp per year) which can have important public health implications.
4.2. Future directions
Further studies are required to evaluate the association between TL and air pollutants in different ethnicities, sexes, and ages, while disentangling the effects of short- and long-term exposures. For the short-term effects, different exposure lags should be tested.
Moreover, logitudinal studies with repeated measures of TL are warranted. There is also need for clinical,in vivoandin vitrostudies to identify the different mechanisms of the effect of air pollutants
on TL. Moreover, air pollutants should be reported in comparable scales and models that permit to direct verdict about its health effects.
5. Conclusion
We conducted a systematic-review and meta-analysis of the available evidence on the impact of air pollution on TL. In general, existing evidence is still limited but suggestive for a potential negative association between long-term exposure to PM2.5and TL and a positive association between short-term exposure to PM2.5, and PCBs and TL. For PM10, and PAHs exposure the results were not conclusive.
Acknowledgments
This study was supported by Shahid Sadoughi University of Medical Science. Payam Dadvand is funded by a Ramon y Cajal fellowship (RYC-2012-10995) awarded by the Spanish Ministry of Economy and Competitiveness.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.envpol.2018.09.130.
References
Allsopp, R.C., Vaziri, H., Patterson, C., Goldstein, S., Younglai, E.V., Futcher, A.B., Greider, C.W., Harley, C.B., 1992. Telomere length predicts replicative capacity of humanfibroblasts. Proc. Natl. Acad. Sci. Unit. States Am. 89, 10114e10118.
Atsdr, 2000. Toxicological Profile for Polychlorinated Biphenyls (PCBs). U.S.
Department of Health and Human Services, Public Health Service, Atlanta, GA.
Bassig, B.A., Zhang, L., Cawthon, R.M., Smith, M.T., Yin, S., Li, G., Hu, W., Shen, M., Rappaport, S., Barone-Adesi, F., Rothman, N., Vermeulen, R., Lan, Q., 2014. Al- terations in leukocyte telomere length in workers occupationally exposed to Benzene. Environ. Mol. Mutagen. 55, 673e678.
Bijnens, E., Zeegers, M.P., Gielen, M., Kicinski, M., Hageman, G.J., Pachen, D., Derom, C., Vlietinck, R., Nawrot, T.S., 2015. Lower placental telomere length may be attributed to maternal residential traffic exposure; a twin study. Environ. Int.
79, 1e7.
Blackburn, E.H., 1991. Structure and function of telomeres. Nature 350, 569.
Boogaard, P.J., 2008. Urinary biomarkers in the risk assessment of PAHs. Occup.
Environ. Med. 65, 221e222.
Callahan, C.L., Pavuk, M., Birnbaum, L.S., Ren, X., Olson, J.R., Bonner, M.R., 2017.
Serum polychlorinated biphenyls and leukocyte telomere length in a highly- exposed population: the Anniston Community Health Survey. Environ. Int.
108, 212e220.
Chen, J.-C., Schwartz, J., 2008. Metabolic syndrome and inflammatory responses to long-term particulate air pollutants. Environ. Health Perspect. 116, 612e617.
Chuang, K.-J., Chan, C.-C., Su, T.-C., Lee, C.-T., Tang, C.-S., 2007. The effect of urban air pollution on inflammation, oxidative stress, coagulation, and autonomic dysfunction in young adults. Am. J. Respir. Crit. Care Med. 176, 370e376.
Colicino, E., Wilson, A., Frisardi, M.C., Prada, D., Power, M.C., Hoxha, M., Dioni, L., Spiro, A., Vokonas, P.S., Weisskopf, M.G., Schwartz, J.D., Baccarelli, A.A., 2017.
Telomere length, long-term black carbon exposure, and cognitive function in a cohort of older men: the VA normative aging study. Environ. Health Perspect.
125, 76e81.
de Keijzer, C., Gascon, M., Nieuwenhuijsen, M.J., Dadvand, P., 2016. Long-term green space exposure and cognition across the life course: a systematic review. Curr.
Environ. Health Rep. 3, 468e477.
Dioni, L., Hoxha, M., Nordio, F., Bonzini, M., Tarantini, L., Albetti, B., Savarese, A., Schwartz, J., Bertazzi, P.A., Apostoli, P., Hou, L., Baccarelli, A., 2011. Effects of short-term exposure to inhalable particulate matter on telomere length, telo- merase expression, and telomerase methylation in steel workers. Environ.
Health Perspect. 119, 622e627.
Egger, M., Smith, G.D., Schneider, M., Minder, C., 1997. Bias in meta-analysis detected by a simple, graphical test. BMJ (Clinical Research Ed.) 315, 629e634.
EPA, U., 2011. Vocabulary Catalog List DetaileIntegrated Risk Information System.
IRIS) glossary.
Eze, I.C., Hemkens, L.G., Bucher, H.C., Hoffmann, B., Schindler, C., Künzli, N., Schikowski, T., Probst-Hensch, N.M., 2015. Association between ambient air pollution and diabetes mellitus in Europe and North America: systematic re- view and meta-analysis. Environ. Health Perspect. 123, 381e389.
Gascon, M., Triguero-Mas, M., Martínez, D., Dadvand, P., Forns, J., Plasencia, A., Nieuwenhuijsen, M.J., 2015. Mental health benefits of long-term exposure to M. Miri et al. / Environmental Pollution 244 (2019) 636e647
646
