Chua et al - Association of ambient air pollution with age-related macular degeneration and retinal thickness in UK Biobank

Clinical science
Br J Ophthalmol: first published as 10.1136/bjophthalmol-2020-316218 on 25 January 2021. Downloaded from http://bjo.bmj.com/ on January 26, 2021 at Swets Subscription Service
Association of ambient air pollution with age-related
macular degeneration and retinal thickness in
UK Biobank
Sharon Y L Chua ‍ ‍,1 Alasdair Warwick,2 Tunde Peto ‍ ‍,3 Konstantinos Balaskas,1,4
Anthony T Moore,5 Charles Reisman,6 Parul Desai ‍ ‍,7 Andrew J Lotery,8
Baljean Dhillon,9,10 Peng T Khaw,1,7 Christopher G Owen ‍ ‍,11 Anthony P Khawaja,1,7
Paul J Foster,1,7 Praveen J Patel ‍ ‍,1,7 on behalf of The UK Biobank Eye and Vision
Consortium
►► Additional material is ABSTRACT of the retinal pigment epithelium (RPE), photo-

published online only. To view Aim To examine the associations of air pollution with receptor loss and retinal degeneration.2 By 2020,
please visit the journal online
(http://d x.doi.o rg/10.1136/ both self-reported age-related macular degeneration the global projected number of people with AMD
bjophthalmol-2020-316218). (AMD), and in vivo measures of retinal sublayer is approximately 200 million, increasing to nearly
thicknesses. 300 million by 2040.3 Well- known risk factors
For numbered affiliations see Methods We included 115 954 UK Biobank include older age, smoking and genetic factors.1
end of article. A constellation of adverse factors (both risk geno-
participants aged 40–69 years old in this cross-
sectional study. Ambient air pollution measures included types, smoking and body mass index (BMI) ≥25)
Correspondence to
Paul J Foster, UCL Institute of particulate matter, nitrogen dioxide (NO2) and nitrogen together increases the risk 19- fold.4 As smoking
52709191. Protected by copyright.

Ophthalmology, 11-43 Bath oxides (NOx). Participants with self-reported ocular tobacco is a risk factor, it is plausible that ambient
Street, EC1V 9EL, London, UK; conditions, high refractive error (< −6 or > +6 diopters) air pollution may also be a modifiable risk factor.
p.foster@u cl.ac.uk and poor spectral-domain optical coherence tomography Air pollution is one of the world’s most important
(SD-OCT) image were excluded. Self-reported AMD was environmental health risks. It is associated with
PJF and PJP are joint senior
authors. used to identify overt disease. SD-OCT imaging derived increased mortality and morbidity.5 Exposure to
photoreceptor sublayer thickness and retinal pigment air pollution is associated with pulmonary and
Received 3 March 2020 epithelium (RPE) layer thickness were used as structural cardiovascular disease6 and eye diseases including
Revised 4 November 2020 biomarkers of AMD for 52 602 participants. We examined glaucoma7 and AMD.8 The mechanisms of air
Accepted 7 December 2020
the associations of ambient air pollution with self- pollution-induced health effects may likely involve
reported AMD and both photoreceptor sublayers and oxidative stress and inflammation.9 The retina is
RPE layer thicknesses. one of the highest oxygen-consuming tissues in the
Results After adjusting for covariates, people who human body and resides in an environment that
were exposed to higher fine ambient particulate is primed for the generation of reactive oxygen
matter with an aerodynamic diameter <2.5 µm (PM2.5, species and resultant oxidative damage.10 Oxida-
per IQR increase) had higher odds of self-reported tive damage increases with age, resulting in retinal
AMD (OR=1.08, p=0.036), thinner photoreceptor dysfunction and cell loss. Rapid, non- invasive
synaptic region (β=−0.16 µm, p=2.0 × 10−5), thicker optical coherence tomography (OCT) imaging of
photoreceptor inner segment layer (β=0.04 µm, the retina is now commonly used by community
p=0.001) and thinner RPE (β=−0.13 µm, p=0.002). opticians and hospital eye clinics and to assess
Higher levels of PM2.5 absorbance and NO2 were retinal structural changes associated with AMD,
associated with thicker photoreceptor inner and outer and to guide its management.11
segment layers, and a thinner RPE layer. Higher levels of If air pollution has an adverse effect on AMD
PM10 (PM with an aerodynamic diameter <10 µm) was risk, this may offer a new range of interventions for
associated with thicker photoreceptor outer segment controlling this important condition. We examined
and thinner RPE, while higher exposure to NOx was data from UK Biobank (UKBB), a large community-
associated with thinner photoreceptor synaptic region. based cohort study. The aim of our study was to
Conclusion Greater exposure to PM2.5 was associated evaluate the relationship between ambient air pollu-
with self-reported AMD, while PM2.5, PM2.5 absorbance, tion, AMD status and OCT imaging derived struc-
© Author(s) (or their PM10, NO2 and NOx were all associated with differences tural features of the disease: photoreceptor sublayer
employer(s)) 2021. No in retinal layer thickness. and RPE layer thickness.
commercial re-use. See rights
and permissions. Published
by BMJ. METHODS
Study population
To cite: Chua SYL,
INTRODUCTION UKBB is a very large community-based cohort of
Warwick A, Peto T, et al.
Br J Ophthalmol Epub ahead Age-related macular degeneration (AMD) is the 502 656 UK residents registered with the National
of print: [please include Day leading cause of irreversible blindness in adults 50 Health Service and aged 40–69 years at enrolment.
Month Year]. doi:10.1136/ years and above in high- income countries.1 Dry Baseline examinations were carried out between
bjophthalmol-2020-316218 AMD is characterised by progressive dysfunction 2006 and 2010 at 22 study assessment centres. The
Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi:10.1136/bjophthalmol-2020-316218 1
Clinical science
North West Multi-centre Research Ethics Committee approved
the study in accordance with the principles of the Declaration of
Helsinki. The overall study protocol (http://www.ukbiobank.ac.
uk/resources/) and protocols for individual tests (http://biobank.
ctsu.ox.ac.uk/crystal/docs.cgi) are available online. Participants
answered a wide-ranging touch-screen questionnaire covering
demographic, socioeconomic, lifestyle, systemic and ocular
diseases information. Definition of hypertension was based
on self-reported data. Physical measures included height and
weight. BMI was defined as weight divided by height squared.
Ocular assessment
Ocular assessment was introduced as an enhancement in 2009
for six assessment centres which are spread across the UK.12
Habitual visual acuity (VA) was measured using a logarithm of the
minimum angle of resolution (LogMAR) chart (Precision Vision,
LaSalle, Illinois, USA) on a computer screen under standard illu-
mination.12 13 Refractive error was measured using an autore-
fractor (Tomey RC 5000, Nagoya, Japan).14 High-resolution
OCT imaging was performed using the Topcon 3D OCT 1000
Mk2 (Topcon Inc, Oakland, New Jersey, USA) in a dark room, Figure 1 Flowchart of participants included in the study. AMD,
without pupillary dilation using the 3D macular volume scan age-related macular degeneration; BMI, body mass index; D, Diopters;
(scan settings: 512 horizontal A scans per B scan; 128 B scans NO2, nitrogen dioxide, NOx, nitrogen oxide; OCT, optical coherence
in a 6×6 mm raster pattern). The Topcon Advanced Boundary tomography; PM2.5, particulate matter (aerodynamic diameter of less
Segmentation (TABS) Algorithm (V.1.6.1.1)15 was used to detect than 2.5 µm); PM2.5 absorbance, particulate matter (a measurement of

retinal layer boundaries and measure the thickness of the RPE16 the blackness of PM2.5 filter—a proxy for elemental or black carbon);
and photoreceptor sublayers (online supplemental figure 1). The PMcoarse, particulate matter (aerodynamic diameter between 2.5 and
TABS segmentation algorithm has been validated previously 10 µm); PM10, particulate matter (aerodynamic diameter of less than
showing a high degree of precision and reproducibility compared 10 µm); RPE, retinal pigment epithelium; SER, spherical equivalent
with manual segmentation methods.15 Strict quality control was refraction.
implemented to exclude images of poor quality as described in
detail previously.17 OCT scans with image quality score (signal
strength) <45 were excluded. Several segmentation indicators Estimates of air pollution
were calculated to identify poor scan quality or segmentation The air pollution estimates were provided by the Small Area
failures. Participants with the poorest 20% of images for each of Health Statistics Unit (http://www. sahsu.
org/) as part of the
these indicators were also excluded. These indicators included BioSHaRE-EU Environmental Determinants of Health Project
an inner limiting membrane (ILM) indicator, a validity count (http://www. bioshare.eu/), and were linked centrally to the
and motion indicators. The ILM indicator was a measure of assessment data by UKBB analysts (http:// biobank.
ctsu.
ox.
ac.
the minimum localised edge strength around the ILM boundary uk/crystal/docs/EnviroExposEst.pdf). Detailed estimates of
across the entire scan. It is useful for identifying blinks, scans air pollution parameters have been published.20 The annual
that contain regions of severe signal fading and segmentation average concentration of PM2.5 (aerodynamic diameter of less
errors. The validity count indicator is used to identify scans with than 2.5 µm), PMcoarse (aerodynamic diameter between 2.5 and
a significant degree of clipping in the OCT scan’s z-axis dimen- 10 µm), PM10 (aerodynamic diameter of less than 10 µm), PM2.5
sion. The motion indicators use both the nerve fibre layer and absorbance (a measurement of the blackness of PM2.5 filter—a
the full retinal thicknesses, from which Pearson correlations and proxy for elemental or black carbon), nitrogen dioxide (NO2)
absolute differences between the thickness data from each set of and nitrogen oxides (NOx) were calculated centrally by the
consecutive B-scans are calculated. The lowest correlation and UKBB using a land use regression model developed by the
the highest absolute difference in a scan serve as the resulting European Study of Cohorts for Air Pollution Effects (ESCAPE)
indicator scores and identify blinks, eye motion artefacts and project (http://www.escapeproject.eu/).21 By using the predictor
segmentation failures. The image quality score and the afore- variables obtained from the Geographic Information System
mentioned indicators usually are highly correlated.18 such as traffic, land use and topography, the land use regression
models calculate the spatial variation of annual average air pollu-
tion concentration at participants’ residential addresses given at
Definition of AMD status baseline visit. NO2 annual concentration data were available for
Definition of AMD status was based on self-reported data. AMD 4 years (2005, 2006, 2007 and 2010), while PM10 data were
status was determined as those who selected ‘macular degener- available for 2007 and 2010. We averaged the values to obtain
ation’ from a predefined list of eye disorders to the question the mean estimate. All other particulate matter and nitrogen
‘Has a doctor ever told you that you have any of the following pollutants had the exposure data for a single year (2010).
problems with your eyes?’ We also carried out a validation of
self-reported AMD status by carrying out masked grading of Inclusion and exclusion criteria
the retinal OCT and fundus images for features of AMD based A uniform set of exclusion criteria was applied in the analysis of
on the Beckman AMD classification on a random subset of age- AMD status, photoreceptor layer and RPE thickness (figure 1).
matched participants.19 We excluded data from: (1) participants who withdrew consent;
2 Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi:10.1136/bjophthalmol-2020-316218
Clinical science
or (2) had self-reported diabetes-related eye disease, eye injury
Table 1 Demographic, systemic and ocular characteristics of
resulting in vision loss or other serious eye conditions; high
participants with availability of data on self-reported AMD and retinal
refractive error (< −6 diopters [D] or > +6D) or (3) partici-
layers
pants who had poor OCT image scans using TABS software.16 22
These participants were excluded because of the well-recognised Participants
Participants with data with data on
impact these factors have on retinal layer thickness.23
on self-reported AMD retinal layers
(n=115 954) (n=52 602)
Statistical analysis Sociodemographic factors
The present analysis was based on cross-sectional data collected Age 56.8 (8.0) 56.4 (8.1)
at one point in time. For this analysis, if both eyes of a patient Sex
were eligible for inclusion in the analysis, one eye was randomly Men 53 218 (46%) 24 753 (47%)
selected using STATA software (V.13, StataCorp LP, College Women 62 736 (54%) 27 849 (53%)
Station, Texas, USA). We examined the baseline characteristics Race
of participants included for each specific outcome (self-reported
White 105 465 (91%) 48 475 (92%)
AMD and retinal layers). Descriptive statistics for continuous
Non-w
hite 10 489 (9%) 4127 (8%)
variables are presented as mean (SD), whereas categorical vari-
Townsend deprivation index −1.1 (3.0) −1.2 (2.9)
ables are presented as number (percentage). We examined the
Clinical factors
associations of each air pollutant (independent variables) with
BMI (kg/m2) 27.3 (4.5) 27.2 (4.4)
self-reported AMD (dependent variable) using logistic multivari-
able regression models, adjusted for age, sex, race, Townsend Smoking status
deprivation index, BMI, smoking status and refractive error. Never 64 554 (56%) 29 238 (56%)
The associations of air pollutants with photoreceptor sublayers Previous 40 224 (35%) 18 421 (35%)
and RPE thicknesses (dependent variables) were adjusted for Current 11 176 (10%) 4943 (9%)
the same variables, using linear multivariable regression models. Spherical equivalent (diopters) −0.1 (2.1) 0.0 (2.0)
The effect estimates represent the change in self-reported AMD Numbers are mean (SD) or n (%), unless otherwise stated.
and retinal layers variables per IQR increment in air pollution.

AMD, age-related macular degeneration; BMI, body mass index; NO2, nitrogen
Statistical significance was set at p<0.05 for the outcomes dioxide; NOx, nitrogen oxide; PM10, particulate matter (aerodynamic diameter of
reported AMD and RPE thickness. When photoreceptor
self- less than 10 µm); PM2.5, particulate matter (aerodynamic diameter of less than
2.5 µm); PM2.5 absorbance, particulate matter (a measurement of the blackness
sublayer thickness was analysed as an outcome, statistical signif-
of PM2.5 filter—a proxy for elemental or black carbon); PMcoarse, particulate matter
icance was set at p<0.002 after Bonferroni correction as we (aerodynamic diameter between 2.5 and 10 µm).
examined six different types of air pollutants with four distinct
photoreceptor-related layers. In sensitivity analysis, we exam-
ined the associations of air pollutants with visually significant
7.6%, all p<0.001, online supplemental table 1). The distribu-
self-reported AMD. Visually significant self-reported AMD was
tion of ambient air pollution exposure of participants with data
defined as self-reported AMD participants with VA worse than
on self-reported AMD and a subgroup with retinal layer data
LogMAR 0.3 (equivalent to Snellen 20/40), while non-visually
are shown in online supplemental table 2). The mean (SD) of
significant self-reported AMD was defined as those with VA of
the various retinal layers are as follows: total length of photo-
LogMAR 0.3 or better.
receptor (142.1 µm (8.2 µm)), photoreceptor synaptic region
(80.4 µm (6.6 µm)), photoreceptor inner segment (23.8 µm
RESULTS (2.0 µm)), photoreceptor outer segment (37.9 µm (4.3 µm))
Of the 133 964 participants who completed ocular assessment, and RPE (25.6 µm (7.2 µm)). Of the 115 954 participants, 1286
24 participants withdrew their consent. Of the 133 940, we (1.1%) were diagnosed with AMD. Masked grading of OCT
excluded 13 329 participants according to the exclusion criteria and retinal fundus images from 119 participants (60 with self-
(figure 1), leaving data on 120 611 participants. There were reported AMD and 59 without self- reported AMD) showed
complete data (age, sex, race, Townsend deprivation index, that 75% of those with self-reported AMD had OCT features of
BMI, smoking status, refractive error, self-reported AMD and AMD while only 12% of those without self-reported AMD had
air pollution measures) for 115 954 participants. Of the 115 OCT features of AMD.
954, there was complete OCT imaging data on retinal layers for Participants exposed to higher levels of PM2.5 concentration
68 088 participants. We excluded 15 486 participants according were 8% more likely to have self- reported AMD (OR 1.08,
to the exclusion criteria for OCT. Hence, 52 062 participants 95% CI 1.01 to 1.16; p=0.036, per IQR increase, table 2).
were included in the analysis for examining RPE and photo- Following Bonferroni correction, higher levels of PM2.5 and
receptor layer thickness. This large number of exclusions for NOx were associated with thinner photoreceptor synaptic region
retinal layers was because of a later start for OCT imaging in (table 3). In contrast, per IQR increase in PM2.5, PM2.5 absor-
UKBB, meaning a smaller number of people were scanned. bance and NO2 were associated with a thicker photoreceptor
The characteristics of participants with data on self-reported inner segment layer. Exposure to higher levels of PM2.5 absor-
AMD and a subgroup with data on retinal layer are shown in bance, PM10 and NO2 were associated with a thicker photore-
table 1. Both groups had similar sociodemographic and clinical ceptor outer segment layer (table 3). Higher concentration of
characteristics. Compared with participants with self-reported PM2.5, PM2.5 absorbance, PM10 and NO2 were associated with
AMD, those without self-reported AMD were more likely non- a thinner RPE layer (table 4). In addition, we examined the
white (9.1% vs 7.0%; p=0.01), younger (56.8 years vs 61.6 association of smoking status with self-reported AMD. Among
years), more likely male (46.0% vs 40.9%), more likely to come participants with self-reported AMD, 510 (39.7%)/1286 and
from a more deprived area (less negative Townsend deprivation 101 (7.9%)/1286 were previous and current smokers, respec-
index, −1.1 vs −1.4) and more likely to be smokers (9.7% vs tively. After adjusting for age, sex, race, Townsend deprivation
Clinical science
Table 2 Association of ambient air pollution with self-reported
1.1×10−8
8.7×10-7
P value
BMI, body mass index; NO2, nitrogen dioxide; NOx, nitrogen oxide; PM, particulate matter; PM10, PM <10 µg/m3; PM2.5, PM <2.5 µg/m3; PM2.5 absorbance, a measurement of the blackness of PM2.5 filter—a proxy for elemental or black carbon;
AMD
0.001
0.009
0.04
0.85
Multivariate regression
Photoreceptor outer segment

OR (95% CI) P value
(−0.04 to 0.03)
(0.003 to 0.10)
Air pollution factors
(0.07 to 0.17)
(0.04 to 0.15)
(0.11 to 0.22)
(0.01 to 0.09)
PM2.5 (µg/m3) 1.08 (1.01 to 1.16) 0.036
(95% CI)
PM2.5 absorbance (µg/m3) 1.00 (0.93 to 1.07) 0.95
PM2.5–10 (µg/m3) 1.01 (0.96 to 1.07) 0.58
PM10 (µg/m3) 0.94 (0.86 to 1.02) 0.11
−0.003
NO2 (µg/m3) 0.99 (0.91 to 1.08) 0.80
0.05
0.12
0.09
0.17
0.05
NOx (µg/m3) 1.03 (0.97 to 1.09) 0.34
β
The OR represents per IQR increase in exposure variable.
Values are adjusted for age, sex, race, Townsend deprivation index, BMI, smoking
2.0×10−4
status and spherical equivalent refraction.
P value
0.001
0.001
0.004
Bold values denote statistical significance at P<0.05 level.
0.32
0.63
AMD, age-related macular degeneration; BMI, body mass index; NO2, nitrogen
dioxide; NOx, nitrogen oxide; PM10, particulate matter (aerodynamic diameter of
Photoreceptor inner segment

less than 10 µm); PM2.5, particulate matter (aerodynamic diameter of less than
(−0.02 to 0.007)
(−0.01 to 0.007)
2.5 µm); PM2.5 absorbance, particulate matter (a measurement of the blackness
(0.008 to 0.04)
(0.02 to 0.06)
(0.02 to 0.06)
(0.02 to 0.07)
of PM2.5 filter—a proxy for elemental or black carbon); PMcoarse, particulate matter
(aerodynamic diameter between 2.5 and 10 µm).
(95% CI)
index, BMI, SER and PM2.5, compared with never smoking,
previous and current smokers were not associated with self-
−0.008
−0.002
0.04
0.04
0.04
0.03
reported AMD (p>0.05). We have additionally adjusted for

hypertension in the multivariable models in view of its rela-
β
tionship with AMD24 and air pollution.25 The associations of
air pollutants with self-reported AMD, photoreceptor sublayers
2.0×10−5
P value
and RPE thickness did not differ after additional adjustment for
0.001
0.004
0.21
0.24
0.19
hypertension. Sensitivity analysis showed that participants with
higher exposure to PM2.5 was marginally associated with visually
Photoreceptor synaptic region
significant self-reported AMD (n=167, OR 1.18, 95% CI 0.98

to 1.41; p=0.08, per IQR increase) compared with participants (−0.23 to 0.09)
(−0.17 to 0.03)
(−0.08 to 0.02)
(−0.13 to 0.03)
(−0.14 to 0.03)
(−0.16 to 0.04)
with either no self-reported AMD or those with non-visually
Association of ambient air pollution with thickness of the photoreceptor sublayers
significant self- reported AMD, although it was not statisti-

(95% CI)
Values are adjusted for age, sex, race, Townsend deprivation index, BMI, smoking status and refractive error.
cally significant. None of the other air pollutants were statisti-
cally significant with visually significant self-reported AMD. In
the sensitivity analysis, we have also additionally adjusted for
smoking pack years and there was a borderline significant asso-
−0.16
−0.10
−0.03
−0.05
−0.06
−0.10
ciation between PM2.5 and self-reported AMD (OR 1.07, 95% CI

β
0.99 to 1.16; p=0.07, per IQR increase). Bold values denote statistical significance at P<0.002 level after Bonferroni correction.
P value
0.004
0.15
0.22
0.18
0.47
0.63
DISCUSSION
In this large study of UKBB participants, we have identified
novel associations between ambient outdoor air pollutant levels
(−0.16 to 0.02)
(−0.03 to 0.14)
(−0.11 to 0.02)
(−0.06 to 0.14)
(−0.09 to 0.06)
Multivariate regression
at participants’ residential addresses with self-reported AMD,

(0.04 to 0.26)
The β coefficients represent per IQR increase in exposure variable.

Total photoreceptor
and also with retinal structure (including thickness of photore-

(95% CI)
ceptor and RPE layers on OCT imaging).

Our results showed that greater ambient PM2.5 exposure was
associated with increased odds of AMD and corresponding retinal
thicknesses (specifically photoreceptor sublayer and RPE). No
−0.07
−0.04
−0.02
0.06
0.04
0.15
such significant associations were observed for PMcoarse. This may

β
be explained by differences in the sites of deposition in the respi-

PMcoarse, PM between 2.5 and 10 µg/m3.
ratory tract and the sources and chemical composition for these
different-sized PM.26 PMcoarse are primarily produced from mechan-
ical grinding, windblown dust and agricultural activities, and
PM2.5 absorbance (µg/m )
3
mainly deposit in the upper and larger airways. In contrast, PM2.5

particles are mainly from combustion process and are able to reach
PMcoarse (µg/m )
the smaller airways and alveoli and are transmitted to the blood,27
3
PM2.5 (µg/m3)
PM10 (µg/m3)
NO2 (µg/m3)
NOx (µg/m )
3
causing a cascade of physiological events associated with morbidity

and mortality.5 28 The deeper penetration of PM2.5 may account for
Table 3
the stronger associations of PM2.5 with self-reported AMD and struc-

tural biomarkers observed in our study.
Clinical science
geographic atrophy is by degeneration of RPE cells, followed by
Table 4 Association of ambient air pollution with thickness of the
loss of photoreceptor cells and choriocapillaris.33 Since the RPE
RPE layer
is involved in the turnover of photoreceptor outer segments,
Multivariate regression RPE dysfunction may lead to thickening of photoreceptor outer
RPE segments.
β (95% CI) P value Our results showed that PM2.5 and NOx were associated with
a thinner photoreceptor synaptic region. This is in agreement
with a reduction in the number of photoreceptor synaptic
PM2.5 (µg/m3) −0.13 (−0.21 to –0.05) 0.002
terminals overlying drusen in AMD.34 In contrast, PM2.5, PM2.5
PM2.5 absorbance (µg/m3) −0.09 (−0.17 to –0.008) 0.03
absorbance and NO2 were associated with thicker photore-
PMcoarse (µg/m3) −0.02 (−0.08 to 0.04) 0.50
ceptor inner segment, while PM2.5 absorbance, NO2 and PM10
PM10 (µg/m3) −0.12 (−0.21 to –0.02) 0.01 were associated with thicker photoreceptor outer segment. As
NO2 (µg/m3) −0.12 (−0.21 to –0.02) 0.01 mitochondria are prominent in photoreceptor inner segments,
NOx (µg/m3) −0.05 (−0.12 to 0.02) 0.17 oxidative stress may induce mitochondrial swelling,35 leading to
The β coefficients represent per IQR increase in exposure variable. a slight thickening in the photoreceptor inner segment. Abnor-
Values are adjusted for age, sex, race, Townsend deprivation index, BMI, smoking malities in the photoreceptor inner and outer segments have also
status and refractive error. been reported in retinal toxicity associated with hydroxychlo-
Bold values denote statistical significance at P<0.05 level.
roquine.36 Our study did not show an association between air
BMI, body mass index; NO2, nitrogen dioxide; NOx, nitrogen oxide; PM10, particulate
matter less than 10 µm in aerodynamic diameter; PM2.5, particulate matter less
pollution and average total photoreceptor layer thickness, which
than 2.5 µm in aerodynamic diameter; PM2.5 ab, (PM2.5 absorbance) a measurement may be explained by thinning of the synaptic region cancelling
of the blackness of PM2.5 filter—a proxy for elemental or black carbon; PMcoarse, out the thickening of the inner/outer segments. In a study by
particulate matter between 2.5 µm to 10 µm in aerodynamic diameter; RPE, retinal Schuman et al, although the authors reported decreased photo-
pigment epithelium.; receptor thickness over drusen, there was a lack of widespread
photoreceptor loss.37 Hence, it is possible that there was focal
loss of the photoreceptor thickness in our study but an overall
NO2 is a product of combustion, primarily from traffic and loss of photoreceptor layer was not observed.

industrial sources, and one of the most notable ambient air pollut- Cigarette smoking may also contribute to particulate matter air
ants associated with health effects.29 30 Similarly, NOx is produced pollution.38 Because of the previously recorded, very strong link
from the reaction of nitrogen and oxygen gases in the air during between AMD and smoking,39 and the plausible link between
combustion.31 NOx contributes to the formation of fine particles smoking and particulate air pollution, we examined the association
and ground level ozone. PM2.5 absorbance, a measurement of the between smoking status of participants with self-reported AMD and
blackness of PM2.5 filter—a proxy for elemental or black carbon, did not observe a significant association. This suggests that the rela-
is also an indicator of combustion particles. Since the major tionship between PM2.5 and self-reported AMD is not mediated by
source of NO2, NOx and PM2.5 absorbance is from combus- cigarette smoke. The prevalence of late AMD standardised to the
tion particles, it may explain the similar associations observed UK population aged 50 years or more and 65 years or more was
between these air pollutants with the retinal structures. A recent 2.4% and 4.8%, respectively. Prevalence of geographic atrophy
longitudinal population-based study using data from the Taiwan was 1.3% and 2.5% for the respective age groups.40 The European
National Health Insurance Programme between years 2000 and Eye Epidemiology (E3) Consortium performed a meta-analysis and
2010 included 39 819 AMD-free participants, with 1442 partic- showed that overall prevalence was 13.2% for early AMD and 3.0%
ipants developing AMD during the 11 year follow- up. AMD for late AMD for people aged 70 years or older.41 Compared with
status was defined via International Classification of Diseases, the E3 Consortium, participants in UKBB are slightly younger and
Ninth Revision, Clinical Modification (ICD-9-CM). Compared include a healthier population than the rest of UK population.42 The
with participants in the lowest exposure quartile, those in self-reported AMD cases in our study may represent AMD in the
the highest quartile of NO2 and carbon monoxide (CO) had early stages. We compared the VA between participants with and
increased risk of self-reported AMD (NO2: HR=1.91, 95% CI without self-reported AMD. Among those with self-reported AMD,
1.64 to 2.23, p<0.001 and CO: HR=1.84, 95% CI 1.50 to 2.15, there was a higher proportion of participants with visual impairment
p<0.001, respectively).8 The difference in findings between ours (VA worse than LogMAR 0.3) compared with those without visual
and the Taiwanese study may be related to the study population, impairment (1.8% vs 1.0%; p<0.001). The proportion of self-
definition and proportion of AMD cases, type and method of reported AMD (1.1%) in our study may have been underestimated
estimating the exposure of air pollutants and type of covariates and it is likely that the risk estimates may have been underestimated.
adjusted in the multivariable models. Compared with our study, In addition to the increased risk of AMD associated with
the Taiwan study included slightly older participants (mean=62 higher exposure to air pollution in the Taiwanese study, other
years vs 56 years), had a slightly higher proportion of AMD studies in the UKBB43 and China7 have reported increased odds
(3.6% vs 1.1%) and estimated a smaller number of air pollutants of glaucoma with higher exposure to PM2.5. In the UKBB study
(two air pollutants including NO2 and CO vs six air pollutants). of 111 370 participants, greater exposure to PM2.5 was associated
In addition, the participant’s living area was defined based on with both self-reported glaucoma and retinal structures associ-
the treatment venue for acute upper respiratory tract infection ated with the disease.43 Wang et al reported that higher average
in the Taiwan study. The effect of pollution on retinal structure levels of PM2.5 was associated with higher burden of glaucoma
associated with AMD were not examined in the Taiwan study. disability, using national level data.7 The New England-based
Ambient air pollution could plausibly be associated with AMD Normative Aging Study showed an association between black
through oxidative stress or inflammation. Oxidative damage carbon exposure with IOP that was greater in individuals with
induces many adverse biological effects including lipid, protein, a high oxidative stress allelic score.44 Taken together, our results
DNA oxidation, initiation of proinflammatory processes28 support published findings of increased risk of eye diseases or
and RPE apoptosis.32 Atrophic or ‘dry’ AMD, also known as association with retinal structures in participants with higher
Clinical science
4
exposure to ambient air pollution. As certain groups of individ- School of Biological Sciences, University of Manchester, Manchester, UK
5
uals including people with diabetes mellitus45 or hypertension24 Department of Ophthalmology, University of California San Francisco, San Francisco,
California, USA
may have increased risk of AMD, it will be useful to explore if 6
Topcon Healthcare Solutions Research & Development, Oakland, New Jersey, USA
these groups of individuals are at greater risk of eye disease when 7
Moorfields Eye Hospital, London, UK
exposed to air pollution in future analysis. 8
Clinical and Experimental Sciences, Faculty of Medicine, University of Southampton,
Strength of this study include its large sample size and the highly Southampton, UK
9
Centre for Clinical Brain Sciences, School of Clinical Sciences, University of
accurate and reproducible measurements of the OCT retinal thick-
Edinburgh, Edinburgh, UK
ness. Limitations of the study include the UKBB is a volunteer 10
NHS Lothian Princess Alexandra Eye Pavilion, Edinburgh, UK
cohort, and participants are likely healthier than the general popu- 11
Population Health Research Institute, St George’s, University of London, London,
lation. Outdoor air pollution was estimated using the participants’ UK
home address and do not explain all variation in indoor concentra-
tions. As most individuals spend a large amount of time indoors, Acknowledgements The authors acknowledge a proportion of our financial
support from the UK Department of Health through an award made by the
individual exposure to all forms of air pollution may differ from
National Institute for Health Research to Moorfields Eye Hospital NHS Foundation
that indicated by the ambient outdoor figures. This is most likely Trust and UCL Institute of Ophthalmology for a Biomedical Research Centre for
to be non-differential between cases and controls and will therefore Ophthalmology.
skew the associations towards the null. Another limitation of this Contributors SYLC had full access to all of the data in the study and takes
analysis was the use of self-report as the sole determinant of AMD responsibility for the integrity of the data and the accuracy of the data analysis. PJF
status rather than incorporating a qualitative analysis of the colour and PJP led conception and design of the study. SYLC, PJF and PJP contributed to the
fundus photographs and spectral-domain optical coherence tomog- data analyses, data interpretation and wrote the draft of the manuscript. All authors
reviewed the results, read and critically revised the manuscript. All authors approved
raphy (SD-OCT) imaging, though we did carry out masked grading the final manuscript. The corresponding author attests that all listed authors meet
of retinal imaging in a proportion of participants. This may result in authorship criteria and that no others meeting the criteria have been omitted.
non-differential misclassification bias and most likely bias the esti- Funding The UK Biobank Eye and Vision Consortium is supported by grants from
mates towards the null. Although we applied strict automated quality Moorfields Eye Charity, The NIHR Biomedical Research Centre at Moorfields Eye
control criteria including a manual check of SD-OCT scans with Hospital NHS Foundation Trust and UCL Institute of Ophthalmology, the Alcon
high and low outlying layer thickness,17 it was not practical to manu- Research Institute and the International Glaucoma Association (UK). SYLC, PTK,
PJF and PJP received salary support from the NIHR BRC at Moorfields Eye Hospital.
ally check all OCT scans for segmentation accuracy. Selection bias

PTK is supported in part by the Helen Hamlyn Trust. PJF received support from the
may exist: out of the 115 954 participants with data on self-reported Richard Desmond Charitable Trust, via Fight for Sight, London. APK is supported by a
AMD, 52 602 participants had measurements on outer retinal layers. Moorfields Eye Charity Career Development Fellowship. This research used data from
However, the baseline characteristics (table 1) across the two AMD- the UK Biobank Resource, under data access request number 2112.
associated outcome groups appear to be similar. The cross-sectional Competing interests CR reports employment by Topcon Healthcare Solutions,
design of our study limits the ability to determine the causality Inc, outside the submitted work. PJF reports personal fees from Allergan, Carl Zeiss,
between ambient air pollution and AMD- associated outcomes. Google/DeepMind and Santen, a grant from Alcon, outside the submitted work. PJP
reports grants from Topcon Inc, outside the submitted work.
Further research is needed to probe the relationship between prior
air pollution exposure and risk of incident disease. Patient and public involvement statement Not commissioned; externally peer
reviewed.
In this large study of an older middle-aged UK population,
higher PM2.5 exposure was associated with a higher risk of self- Patient consent for publication Not required.
reported AMD, while all pollutants except PMcoarse were associ- Ethics approval The North West Multi-centre Research Ethics Committee
ated with changes in retinal structure (in either photoreceptor approved the study (reference no., 06/MRE08/65), in accordance with the tenets of
the Declaration of Helsinki. Detailed information about the study is available at the
sublayer and/or RPE layer thickness). Overall, our findings UK Biobank web site (www.ukbiobank.ac.uk).
suggest that ambient air pollution, especially fine PM or those of
Provenance and peer review Not commissioned; externally peer reviewed.
combustion-related particles, may affect AMD risk. It is possible
Data availability statement Researchers wishing to access UK Biobank data can
that the structural features observed may be unrelated to AMD,
register and apply at https://www.ukbiobank.ac.uk/.
but associated with pollution-induced retinal toxicity. However,
Supplemental material This content has been supplied by the author(s). It
the direction of the relationships between air pollution and both
has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have
AMD and associated retinal layer thicknesses indicate higher been peer-reviewed. Any opinions or recommendations discussed are solely those
exposure to air pollution may make the cells more vulnerable of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and
and increase the risk of AMD. Our findings add to the growing responsibility arising from any reliance placed on the content. Where the content
evidence of the damaging effects of ambient air pollution, even includes any translated material, BMJ does not warrant the accuracy and reliability
of the translations (including but not limited to local regulations, clinical guidelines,
in the setting of relative low exposure of ambient air pollution. terminology, drug names and drug dosages), and is not responsible for any error
As UKBB is a very large prospective cohort, we anticipate being and/or omissions arising from translation and adaptation or otherwise.
able to explore the effect of particulate matter on future risk
of AMD. Further studies examining both outdoor and indoor ORCID iDs
Sharon Y L Chua http://orcid.org/0000-0002-2501-0948
ambient air pollution estimates on AMD and outer retinal struc- Tunde Peto http://orcid.org/0000-0001-6265-0381
tures may help to substantiate our findings and understand the Parul Desai http://orcid.org/0000-0002-7532-972X
implications for retinal disease associated with ageing. If our Christopher G Owen http://orcid.org/0000-0003-1135-5977
findings are replicated, this would support the view that air Praveen J Patel http://orcid.org/0000-0001-8682-4067
pollution is an important modifiable risk factor for AMD.
REFERENCES
Author affiliations 1 Mitchell P, Liew G, Gopinath B, et al. Age-Related macular degeneration. Lancet
1
UCL Institute of Ophthalmology, National Institute for Health Research Biomedical 2018;392:1147–59.
Research Centre at Moorfields Eye Hospital NHS Foundation Trust and UCL Institute 2 McLeod DS, Grebe R, Bhutto I, et al. Relationship between RPE and choriocapillaris in
of Ophthalmology, London, UK age-related macular degeneration. Invest Ophthalmol Vis Sci 2009;50:4982–91.
2
UCL Institute of Cardiovascular Science, University College London, London, UK 3 Wong WL, Su X, Li X, et al. Global prevalence of age-related macular degeneration
3
School of Medicine, Dentistry and Biomedical Sciences, Queen’s University Belfast, and disease burden projection for 2020 and 2040: a systematic review and meta-
Belfast, UK analysis. Lancet Glob Health 2014;2:e106–16.

Clinical science
4 Seddon JM, Francis PJ, George S, et al. Association of CFH Y402H and 25 Ibald-Mulli A, Stieber J, Wichmann HE, et al. Effects of air pollution on blood pressure:
LOC387715 A69S with progression of age-related macular degeneration. JAMA a population-based approach. Am J Public Health 2001;91:571–7.
2007;297:1793–800. 26 Wilson WE, Suh HH. Fine particles and coarse particles: concentration relationships
5 Cohen AJ, Brauer M, Burnett R, et al. Estimates and 25-year trends of the global relevant to epidemiologic studies. J Air Waste Manag Assoc 1997;47:1238–49.
burden of disease attributable to ambient air pollution: an analysis of data from the 27 Brown DM, Wilson MR, MacNee W, et al. Size-Dependent proinflammatory effects
global burden of diseases study 2015. Lancet 2017;389:1907–18. of ultrafine polystyrene particles: a role for surface area and oxidative stress in the
6 Lelieveld J, Evans JS, Fnais M, et al. The contribution of outdoor air pollution sources enhanced activity of ultrafines. Toxicol Appl Pharmacol 2001;175:191–9.
to premature mortality on a global scale. Nature 2015;525:367–71. 28 Brook RD, Rajagopalan S, Pope CA, et al. Particulate matter air pollution and
7 Wang W, He M, Li Z, et al. Epidemiological variations and trends in health burden of cardiovascular disease: an update to the scientific statement from the American heart
glaucoma worldwide. Acta Ophthalmol 2019;97:e349–55. association. Circulation 2010;121:2331–78.
8 Chang K-H, Hsu P-Y, Lin C-J, et al. Traffic-Related air pollutants increase the risk for 29 Gaffin JM, Hauptman M, Petty CR, et al. Nitrogen dioxide exposure in
age-related macular degeneration. J Investig Med 2019;67:1076–81. school classrooms of inner-city children with asthma. J Allergy Clin Immunol
9 Lodovici M, Bigagli E. Oxidative stress and air pollution exposure. J Toxicol 2018;141:2249–55.
2011;2011:1–9. doi:10.1155/2011/487074 30 World Health Organization. Air quality guidelines—global update 2005. Available:
10 Jarrett SG, Boulton ME. Consequences of oxidative stress in age-related macular http://www.who.int/p he/health_topics/outdoorair/outdoorair_aqg/en/ [Accessed 20
degeneration. Mol Aspects Med 2012;33:399–417. Feb 2019].
11 Puliafito CA, Hee MR, Lin CP, et al. Imaging of macular diseases with optical 31 Icopal. nitrogen oxide (NOX) pollution. Available: http://www.i copal-noxite.co.uk/nox-
coherence tomography. Ophthalmology 1995;102:217–29. problem/nox-pollution.aspx [Accessed 15 Apr 2020].
12 Chua SYL, Thomas D, Allen N, et al. Cohort profile: design and methods in the eye and 32 Jiang S, Moriarty-Craige SE, Orr M, et al. Oxidant-Induced apoptosis in human retinal
vision Consortium of UK Biobank. BMJ Open 2019;9:e025077. pigment epithelial cells: dependence on extracellular redox state. Invest Ophthalmol
13 Cumberland PM, Rahi JS, UK Biobank Eye and Vision Consortium. Visual function, Vis Sci 2005;46:1054–61.
social position, and health and life chances: the UK Biobank study. JAMA Ophthalmol 33 Sarks JP, Sarks SH, Killingsworth MC. Evolution of geographic atrophy of the retinal
2016;134:959–66. pigment epithelium. Eye 1988;2 (Pt 5:552–77.
14 Cumberland PM, Bao Y, Hysi PG, et al. Frequency and distribution of refractive 34 Johnson PT, Brown MN, Pulliam BC, et al. Synaptic pathology, altered gene expression,
error in adult life: methodology and findings of the UK Biobank study. PLoS One and degeneration in photoreceptors impacted by drusen. Invest Ophthalmol Vis Sci
2015;10:e0139780. 2005;46:4788–95.
15 Yang Q, Reisman CA, Wang Z, et al. Automated layer segmentation of macular OCT 35 Wilson JD, Bigelow CE, Calkins DJ, et al. Light scattering from intact cells reports
images using dual-scale gradient information. Opt Express 2010;18:21293–307. oxidative-stress-induced mitochondrial swelling. Biophys J 2005;88:2929–38.
16 Ko F, Foster PJ, Strouthidis NG, et al. Associations with retinal pigment epithelium 36 Rodriguez-Padilla JA, Hedges TR, Monson B, et al. High-Speed ultra-high-resolution
thickness measures in a large cohort: results from the UK Biobank. Ophthalmology optical coherence tomography findings in hydroxychloroquine retinopathy. Arch
2017;124:105–17. Ophthalmol 2007;125:775–80.
17 Patel PJ, Foster PJ, Grossi CM, et al. Spectral-Domain optical coherence tomography 37 Schuman SG, Koreishi AF, Farsiu S, et al. Photoreceptor layer thinning over drusen in

imaging in 67 321 adults: associations with macular thickness in the UK Biobank eyes with age-related macular degeneration imaged in vivo with spectral-domain
study. Ophthalmology 2016;123:829–40. optical coherence tomography. Ophthalmology 2009;116:488–96.
18 Khawaja AP, Chua S, Hysi PG, et al. Comparison of associations with different macular 38 Repace JL, Lowrey AH. Indoor air pollution, tobacco smoke, and public health. Science
inner retinal thickness parameters in a large cohort: the UK Biobank. Ophthalmology 1980;208:464–72.
2020;127:62–71. 39 Tomany SC, Wang JJ, Van Leeuwen R, et al. Risk factors for incident age-related
19 Ferris FL, Wilkinson CP, Bird A, et al. Clinical classification of age-related macular macular degeneration: pooled findings from 3 continents. Ophthalmology
degeneration. Ophthalmology 2013;120:844–51. 2004;111:1280–7.
20 Aung N, Sanghvi MM, Zemrak F, et al. Association between ambient air pollution and 40 Owen CG, Jarrar Z, Wormald R, et al. The estimated prevalence and incidence of late
cardiac morpho-functional phenotypes. Circulation 2018;138:2175–86. stage age related macular degeneration in the UK. Br J Ophthalmol 2012;96:752–6.
21 Eeftens M, Beelen R, de Hoogh K, et al. Development of Land Use Regression 41 Colijn JM, Buitendijk GHS, Prokofyeva E, et al. Prevalence of age-related macular
models for PM(2.5), PM(2.5) absorbance, PM(10) and PM(coarse) in 20 degeneration in Europe: the past and the future. Ophthalmology 2017;124:1753–63.
European study areas; results of the ESCAPE project. Environ Sci Technol 42 Fry A, Littlejohns TJ, Sudlow C, et al. Comparison of sociodemographic and
2012;46:11195–205. health-related characteristics of UK Biobank participants with those of the general
22 Cruz-Herranz A, Balk LJ, Oberwahrenbrock T, et al. The APOSTEL recommendations population. Am J Epidemiol 2017;186:1026–34.
for reporting quantitative optical coherence tomography studies. Neurology 43 Chua SYL, Khawaja AP, Morgan J, et al. The relationship between ambient
2016;86:2303–9. atmospheric fine particulate matter (PM2.5) and glaucoma in a large community
23 Vujosevic S, Midena E. Retinal layers changes in human preclinical and early clinical cohort. Invest Ophthalmol Vis Sci 2019;60:4915–23.
diabetic retinopathy support early retinal neuronal and Müller cells alterations. J 44 Nwanaji-Enwerem JC, Wang W, Nwanaji-Enwerem O, et al. Association of long-term
Diabetes Res 2013;2013:1–8. ambient black carbon exposure and oxidative stress allelic variants with intraocular
24 Hyman L, Schachat AP, He Q, et al. Hypertension, cardiovascular disease, and age- pressure in older men. JAMA Ophthalmol 2019;137:129–37.
related macular degeneration. age-related macular degeneration risk factors Study 45 Choi JK, Lym YL, Moon JW, et al. Diabetes mellitus and early age-related macular
Group. Arch Ophthalmol 2000;118:351–8. degeneration. Arch Ophthalmol 2011;129:196–9.

BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance
Supplemental material placed on this supplemental material which has been supplied by the author(s) Br J Ophthalmol
Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

Supplementary Table 1. Comparison of characteristics between participants with self-reported AMD and
without self-reported AMD
No self-reported AMD Self-reported AMD
(N=114,668) (N=1,286) P-value
Sociodemographic factors
Age 56.8 (8.1) 61.6 (5.9) <0.001
Sex
Men 52,692 (46.0%) 526 (40.9%)
Women 61,976 (54.0%) 760 (59.1%) <0.001
Race
White 104,269 (90.9%) 1,196 (93.0%)
Non-white 10,399 (9.1%) 90 (7.0%) 0.01
Townsend deprivation index -1.1 (3.0) -1.5 (2.9) <0.001
Clinical factors
Body mass index (kg/m2) 27.2 (4.5) 27.4 (4.3) 0.18
Smoking status
Never 63,879 (55.7%) 675 (52.5%)
Previous 39,714 (34.6%) 510 (39.7%)
Current 11,075 (9.7%) 101 (7.8%) <0.001
Spherical equivalent (diopters) -0.08 (2.1) -0.03 (2.3) 0.40
AMD status was classified based on self-reporting and hospital episode statistics data (ICD10).
Numbers are mean (SD) for continuous variables and no. (%) for categorical variables.
AMD= Age-related macular degeneration

Supplementary Table 2. Distribution of PM2.5, PMcoarse, PM10, NO2 and NOX of participants with availability of
data on self-reported AMD and retinal layers
Self-reported AMD Retinal layers
(N=115,954) (N=52,602)
Median (IQR) Range Median (IQR) Range
PM2.5 (µg/m3) 9.91 (1.07) (8.17, 19.69) 9.88 (1.12) (8.17, 19.69)
PM2.5 absorbance (µg/m3) 1.22 (0.33) (0.83, 4.05) 1.22 (0.33) (0.83, 3.71)
PMcoarse (µg/m3) 6.19 (0.75) (5.57, 12.82) 6.21 (0.77) (5.57, 11.30)
PM10 (µg/m3) 19.37 (2.67) (13.04, 29.67) 19.33 (2.77) (13.38, 29.30)
Nitrogen dioxide (NO2) (µg/m3) 31.75 (12.08) (9.44, 102.75) 31.25 (12.63) (9.44, 86.65)
Nitrogen oxide (NOX) (µg/m3) 43.66 (14.38) (19.74, 263.96) 43.17 (14.97) (19.74, 263.96)
AMD = Age-related macular degeneration, IQR = Interquartile range, PM2.5= Particular matter (aerodynamic diameter of less than 2.5µm), PM2.5
absorbance= Particulate matter (a measurement of the blackness of PM2.5 filter – a proxy for elemental or black carbon), PMcoarse = Particulate
matter (aerodynamic diameter between 2.5 and 10µm, PM10= Particulate matter (aerodynamic diameter of less than 10µm), NO2= Nitrogen
dioxide, NOx= Nitrogen oxide


Supplementary Table 1. Comparison of characteristics between participants with self-reported AMD and
without self-reported AMD
No self-reported AMD Self-reported AMD
(N=114,668) (N=1,286) P-value
Sociodemographic factors
Age 56.8 (8.1) 61.6 (5.9) <0.001
Sex
Men 52,692 (46.0%) 526 (40.9%)
Women 61,976 (54.0%) 760 (59.1%) <0.001
Race
White 104,269 (90.9%) 1,196 (93.0%)
Non-white 10,399 (9.1%) 90 (7.0%) 0.01
Townsend deprivation index -1.1 (3.0) -1.5 (2.9) <0.001
Clinical factors
Body mass index (kg/m2) 27.2 (4.5) 27.4 (4.3) 0.18
Smoking status
Never 63,879 (55.7%) 675 (52.5%)
Previous 39,714 (34.6%) 510 (39.7%)
Current 11,075 (9.7%) 101 (7.8%) <0.001
Spherical equivalent (diopters) -0.08 (2.1) -0.03 (2.3) 0.40
AMD status was classified based on self-reporting and hospital episode statistics data (ICD10).
Numbers are mean (SD) for continuous variables and no. (%) for categorical variables.
AMD= Age-related macular degeneration

Supplementary Table 2. Distribution of PM2.5, PMcoarse, PM10, NO2 and NOX of participants with availability of
data on self-reported AMD and retinal layers
Self-reported AMD Retinal layers
(N=115,954) (N=52,602)
Median (IQR) Range Median (IQR) Range
PM2.5 (µg/m3) 9.91 (1.07) (8.17, 19.69) 9.88 (1.12) (8.17, 19.69)
PM2.5 absorbance (µg/m3) 1.22 (0.33) (0.83, 4.05) 1.22 (0.33) (0.83, 3.71)
PMcoarse (µg/m3) 6.19 (0.75) (5.57, 12.82) 6.21 (0.77) (5.57, 11.30)
PM10 (µg/m3) 19.37 (2.67) (13.04, 29.67) 19.33 (2.77) (13.38, 29.30)
Nitrogen dioxide (NO2) (µg/m3) 31.75 (12.08) (9.44, 102.75) 31.25 (12.63) (9.44, 86.65)
Nitrogen oxide (NOX) (µg/m3) 43.66 (14.38) (19.74, 263.96) 43.17 (14.97) (19.74, 263.96)
AMD = Age-related macular degeneration, IQR = Interquartile range, PM2.5= Particular matter (aerodynamic diameter of less than 2.5µm), PM2.5
absorbance= Particulate matter (a measurement of the blackness of PM2.5 filter – a proxy for elemental or black carbon), PMcoarse = Particulate
matter (aerodynamic diameter between 2.5 and 10µm, PM10= Particulate matter (aerodynamic diameter of less than 10µm), NO2= Nitrogen
dioxide, NOx= Nitrogen oxide

Chua et al - Association of ambient air pollution with age-related macular degeneration and retinal thickness in UK Biobank

Uploaded by

Copyright:

Available Formats

Chua et al - Association of ambient air pollution with age-related macular degeneration and retinal thickness in UK Biobank

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Chua et al - Association of ambient air pollution with age-related macular degeneration and retinal thickness in UK Biobank

Uploaded by

Copyright:

Available Formats

Clinical science

►► Additional material is ABSTRACT of the retinal pigment epithelium (RPE), photo-

52709191. Protected by copyright.

52709191. Protected by copyright.

52709191. Protected by copyright.

Photoreceptor outer segment

Photoreceptor inner segment

52709191. Protected by copyright.

significant self-­reported AMD (n=167, OR 1.18, 95% CI 0.98

significant self-­ reported AMD, although it was not statisti-

ciation between PM2.5 and self-­reported AMD (OR 1.07, 95% CI

at participants’ residential addresses with self-­reported AMD,

The β coefficients represent per IQR increase in exposure variable.

and also with retinal structure (including thickness of photore-

ceptor and RPE layers on OCT imaging).

such significant associations were observed for PMcoarse. This may

be explained by differences in the sites of deposition in the respi-

mainly deposit in the upper and larger airways. In contrast, PM2.5

causing a cascade of physiological events associated with morbidity

the stronger associations of PM2.5 with self-­reported AMD and struc-

52709191. Protected by copyright.

52709191. Protected by copyright.

6 Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi:10.1136/bjophthalmol-2020-316218

52709191. Protected by copyright.

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi:10.1136/bjophthalmol-2020-316218 7

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

Median (IQR) Range Median (IQR) Range

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

Median (IQR) Range Median (IQR) Range

Chua SYL, et al. Br J Ophthalmol 2021;0:1–7. doi: 10.1136/bjophthalmol-2020-316218

You might also like

significant self-reported AMD (n=167, OR 1.18, 95% CI 0.98

significant self- reported AMD, although it was not statisti-

ciation between PM2.5 and self-reported AMD (OR 1.07, 95% CI

at participants’ residential addresses with self-reported AMD,

the stronger associations of PM2.5 with self-reported AMD and struc-