Associations of pre‐ and postnatal exposures with optic nerve status in young adults

We aimed to explore the effect of multiple pre‐ and postnatal exposures on optic nerve status in young adults due to this critical period for development.


| I N T RODUC T ION
The human foetus is vulnerable to toxic agents transferred from the mother to the foetus through the placenta.In vivo optical coherence tomography (OCT) studies of humans have found an association between maternal smoking during pregnancy and children having a thinner peripapillary retinal nerve fibre layer (RNFL) (Ashina et al., 2017;Lee, Mackey, et al., 2021;Pueyo et al., 2011).The prevalence of maternal smoking during pregnancy in Europe is 8.1%, and hence the scope of the problem is considerable (Lange et al., 2018).The concern about exposure of children to tobacco smoke is aggravated by a study that found thinner RNFLs in children aged 6-8 years who had been exposed to environmental tobacco smoke during childhood (Li et al., 2021).A thin RNFL is likely to make the optic nerve more vulnerable to glaucoma, which is characterised by accelerated loss of retino-optic nerve fibres and retinal ganglion cells (Ashina et al., 2017;Lee, Mackey, et al., 2021;Pueyo et al., 2011).Other observations of interest include air pollution being associated with a thinner RNFL (Chua et al., 2019) and physical activity with a thicker RNFL (Lee, McVeigh, et al., 2021).
The impact of early-life exposures on health problems later in life (Brustad et al., 2020;Gürdeniz et al., 2021;Schoos et al., 2019Schoos et al., , 2020Schoos et al., , 2022;;Sunde et al., 2022) indicates that more intense promotion of foetal and childhood health could lead to considerable gains in general health.
In this study, we have analysed RNFL and macular thickness in relation to multiple pre-and postnatal exposures, including smoking and air pollution, to validate and expand previous findings.This was done in young adults aged 18 years who participated in the Copenhagen Prospective Studies on Asthma in Childhood 2000 (COPSAC 2000 ) mother-child cohort.

| Study participants
The Copenhagen Prospective Studies on Asthma in Childhood 2000 (COPSAC 2000 ) cohort consists of 411 children of mothers with a history of asthma (Hallas et al., 2019).Eye examinations were part of the 18-year follow-up of the cohort.
The study was approved by the Copenhagen Ethics Committee (KF01-289/96) and the Danish Data Protection Agency (2015-41-3696), and the study was conducted in accordance with the Declaration of Helsinki (World Medical Association, 2013).Informed written consent was obtained from the participant (if aged ≥18 years) or his or her parents prior to the examinations.

| Smoking related exposures
Maternal smoking was registered 2 weeks after giving birth and was categorised as (1) smoking during pregnancy and (2) not smoking during pregnancy.If the mothers smoked during pregnancy, they were asked how many cigarettes that they smoked per week and in which trimesters they smoked.Cotinine concentrations in blood were measured in the child within 10 days after birth.Exposure to passive smoking at home during pregnancy was categorised as (1) yes or (2) no.Nicotine and cotinine concentrations were measured in hair at the 1year follow-up visit.Passive smoking during childhood (environmental tobacco smoke during childhood) was categorised as (1) the child having been exposed to passive smoking at home during the first 7 years of life and (2) the child not having been exposed to passive smoking at home during the first 7 years of life.Smoking at 18 years of age was categorised as (1) smoking at least 1 cigarette during the last 4 weeks or (2) not having smoked during the last 4 weeks.

| Air pollution
The indoor concentration of PM2.5 was measured with a system consisting of a pump, cyclone, and Teflon filter.Measurements were carried out by trained personnel in the participant's bedroom three times during the first 3 years of life.The concentration of PM2.5 was calibrated according to air flow and season (Raaschou-Nielsen et al., 2010).The outdoor concentration of PM2.5 was assessed using the Operational Street Pollution Model (OSPM) (Berkowicz, 2000; Appendix S1).

| Anthropometrics
Information on birth weight was extracted from the Danish Medical Birth Registry.At the 18-year followup, height without shoes and bodyweight data were collected.

| Steroid use
Inhaled plus intranasal corticosteroid treatment was categorised as (1) had not received inhaled or intranasal corticosteroid treatment ever, (2) had at least once in childhood received inhaled or intranasal corticosteroid continuously for a duration under 365 days, (3) had at least once in childhood received inhaled or intranasal corticosteroid continuously for a duration between 364 days and 730 days, (4) had at least once in childhood received inhaled or intranasal corticosteroid continuously for a duration between 729 days and 1095 days, and (5) had at least once in childhood received inhaled or intranasal corticosteroid continuously for a duration of 1095 days (3 years) or more.Participants who received cutaneous corticosteroid were included, but such exposure was not considered as corticosteroid exposure in the analyses (Appendix S1).

| Genetics
Filaggrin genotyping was performed and categorised as (1) having a mutation in either R501X, 2282del4, R2447X or S3247X or (2) not having any mutation in the locations (Appendix S1).
Genetic risk scores (GRS) were constructed based on recent and relevant genome-wide association studies of smoking and asthma (Ahluwalia et al., 2020;Privé et al., 2022).GRS for each trait were calculated based on all SNPs from the summary statistics and in turn calculated for each subject as a weighted average.The per-allele effect estimate (beta value) was multiplied with the number of risk alleles for each SNP and summed, resulting in a normally distributed GRS.This score was scaled and centred with mean 0 and with a unit in standard deviation (SD) (Appendix S1).

| Eye examination
Eye examination was performed by trained personnel following a standardised protocol.Noncycloplegic refraction, uncorrected, and best-corrected visual acuity were determined using an autorefractor (Auto Ref/ Keratometer ARK-1s, NIDEK).For visual acuity, acceptance of a line of letters having been read required correct identification of at least three out of five letters per line.Ocular axial length was measured using an interferometric device (IOL Master, Carl Zeiss Meditec).Five measurements were made, and the composite value calculated by the machine was used.
RNFL thickness was measured using OCT (Cirrus HD-OCT, Carl Zeiss Meditech, Jena, Germany) in the form of 6 mm × 6 mm optic disc cube 200 × 200 lines scan.The average and four sectorial peripapillary thicknesses (superior, inferior, nasal, and temporal) were extracted by automated computer analysis.Macular thickness was measured using the 6 mm × 6 mm macular cube 512 × 128 lines scan.Macular thickness was measured as the average distance between the internal limiting membrane and the retinal pigment epithelium within the cube (overall macular thickness).An internal fixation target was used for alignment of the eye.Scans were only accepted if they had been fully completed, were well centred, had a signal strength of at least 6 on a scale up to 10, and no blinking or motion artefacts.

| Statistical analysis
Mean and SDs was calculated for continuous variables; median and interquartile range was calculated for skewed distributions (age and body weight).Associations between RNFL or macular thickness and exposure variables were assessed using a general linear regression model.Sex, smoking, alcohol use, inhaled plus intranasal corticosteroid use, mode of delivery, antibiotic use, type of day-care, aeroallergen sensitisation, and filaggrin mutations were entered as categorical data.Age, birth weight, fat mass, lean mass, height, body weight, BMI, nicotine, cotinine, PM2.5, social circumstances, GRSs, 25(OH)D, CMPF, respiratory tract infections, hs-CRP, u-EPX, blood eosinophils, total IgE, axial length, and spherical equivalent refractive error were entered as continuous data.Spherical equivalent refraction was calculated as the sum of the sphere and half of the cylinder.Data for nicotine, cotinine, indoor PM2.5, CMPF, hs-CRP, u-EPX, blood eosinophils, and total IgE were log-transformed.Assumptions of linearity, variance homogeneity, normality of distribution of residuals underlying the statistical model, and missing data were validated by visual inspection of plots.Analyses were adjusted by including relevant variables in a general linear model, and estimates are presented with 95% CIs.The level of significance was set at a two-sided p < 0.05.Statistical analyses were made using the R software package, version 4.0.3(R Foundation for Statistical Computing, www.r-proje ct.org).

| R E SU LT S
The final analysis included 269 out of 350 participants with eye examination data out of the total cohort of 411 participants.The left eye was the default eye used in data analyses, except in 53 cases where scans were deemed to be of substandard quality, hence data from the right eye were used instead.For 28 participants, no scan of acceptable quality was available.One participant was undergoing current work-up for retinitis pigmentosa and was excluded from the study.For description of excluded participants, see Appendix S1.Of the 269 participants with RNFL data of adequate quality, 246 participants (91.4%) had acceptable macular scans.

| Retinal nerve fibre layer
Univariate analyses of exposures found significant RNFL deficits related to maternal smoking during pregnancy, passive smoking during childhood, indoor air pollution (higher PM 2.5 concentrations), lower birth weight, a more myopic refraction, and longer ocular axial length (Table 2a,b).RNFL thickness was not associated with alcohol use, inhaled or intranasal corticosteroid use, or other exposures investigated (Tables 2a,b and Appendix S1).
In a multivariate analysis, only maternal smoking during pregnancy, lower birth weight and longer ocular axial length remained significant (Table 2a,b).
The impact of smoking was investigated further by a subgroup analysis (Figure 2), where the results showed that participants exposed to smoking during pregnancy combined with passive smoking during childhood had the thinnest mean peripapillary RNFL thickness (mean difference −8.6 μm; 95% CI −12.6 to −4.6 μm, p < 0.0001).This result remained significant after adjusting for age, sex, birth weight, axial length, alcohol use, and corticosteroid use (adjusted mean difference −9.6 μm; 95% CI −13.4 to −5.8 μm; p < 0.0001).There was no interaction between the p-values for maternal smoking during pregnancy and passive smoking during childhood; however, the p-value of interaction was close to significance (p = 0.07).Participants who smoked at age 18 years did not have a measurably thinner peripapillary RNFL than non-exposed participants (mean difference − 1.7 μm; 95% CI −4.1 to 0.7 μm; p = 0.18, Table 2b).

| Macular thickness
In univariate analyses, participants whose mothers smoked during pregnancy had a lower mean overall macular thickness than the offspring of non-smoking mothers (280.3 μm vs. 285.4μm, −5.0 μm; 95% CI −8.6 to −1.4 μm; p = 0.007; Table 3), despite offspring of mothers with a higher genetic risk of smoking having thicker maculae (Table 3).Mothers smoking during pregnancy had higher GRS for smoking (p = 0.01).Participants exposed to higher indoor concentrations of PM2.5 had a lower mean overall macular thickness (−2.7 μm; 95% CI −5.3 to −0.1 μm; p = 0.04; Table 3).There was no interaction between the p-values for indoor concentrations of PM2.5 and passive smoking during childhood (p = 0.51).Participants exposed to maternal antibiotic use during pregnancy had a lower mean overall macular thickness than non-exposed participants (−3.3 μm; 95% CI −6.5 to −0.1 μm; p = 0.048; Table 3).No other exposure in Table 3 was associated with macular thickness deviation in univariate analyses.
In the multivariate analysis, only maternal smoking during pregnancy and having a mother with a higher GRS for smoking remained significantly associated with macular thickness deviation (Table 3).

| DI SC US SION
This study of young adults found that exposure to maternal smoking during pregnancy was associated with a thinner peripapillary RNFL.The largest RNFL deficit was found in those who had been exposed to tobacco smoke both during foetal life and childhood.Our findings suggest a dose-response relationship between number of cigarettes smoked during pregnancy and RNFL deficit.Further, exposure to maternal smoking during pregnancy was associated with a macular thickness deficit.We also found that lower birth weight and longer ocular axial length were associated with a thinner peripapillary RNFL.
The study collected data prospectively from early pregnancy and throughout childhood, covering a broad range of exposures and outcomes.The use of selfreported maternal tobacco consumption in the study has previously been validated by comparison with plasma and/or urine cotinine concentration (Klebanoff et al., 2001).In this study, significant correlation has been found between self-reported maternal tobacco consumption and blood cotinine concentrations in the offspring shortly after birth (p < 0.001).The use of nicotine  and cotinine in hair as an indicator for exposure to environmental tobacco smoke is supported by another study (Sørensen et al., 2007).
The main limitation of the study is the population size, but the study nevertheless reproduced earlier findings of RNFL deficits related to maternal smoking during pregnancy, passive smoking during childhood, lower birth weight, and longer ocular axial length (Ashina et al., 2017;Lee, Mackey, et al., 2021;Li et al., 2021).The principal new observation is that maternal smoking during pregnancy and passive smoking during childhood work in conjunction to produce additive, statistically independent detrimental effects on the inner retina and, by anatomical extension, on the optic nerve.
The 4.6 μm RNFL deficit attributable to maternal smoking during pregnancy found in this study is comparable to the 4.8 μm deficit found in a considerably larger birth cohort from Australia (Lee, Mackey, et al., 2021) and the 5.7 μm deficit found in an equally large cohort from Denmark (Ashina et al., 2017).Passive smoking during childhood was associated with a deficit of 4.4 μm in the Hong Kong Children Eye Study, which excluded c Body mass index.
d Spherical equivalent refractive error.
g Principal component analysis of mother's age, education, mother's income, and household income by time of birth.
*Significant at a two-sided p < 0.05.Number of data (n) available and included in the analysis is given in the table.
children exposed smoking in utero found evidence of a dose-response effect (Li et al., 2021).The Australian study found no effect of passive smoking during childhood after adjusting for in utero exposure to smoking (p = 0.69).This is in line with our finding; in contrast, our p-value was close to significance (p = 0.06).
The lack of an effect of active smoking at 18 years suggest that prenatal life and early childhood may be particularly critical for ocular health.This insight underlines the importance of avoiding exposure to tobacco smoke during pregnancy and childhood.
A thinner RNFL, which forms the conducting element of the optic nerve, is expected to be more susceptible to glaucomatous damage later in life, given that glaucoma is characterised by progressing focal loss as well as universal loss of retinal nerve fibres and their ganglion cells (Alasil et al., 2014;Bowd et al., 2000;Kwon et al., 2009).
Previous studies have shown that maternal smoking during pregnancy is associated with deficits in cerebral cortical thickness in the offspring (Ekblad et al., 2010;Rivkin et al., 2008).The common susceptibility of the retina and the brain to damage from smoking shows how ophthalmic investigations may be of value of the study of central nervous system toxicology (London et al., 2013).
Also, previous studies have found higher intraocular pressure (IOP) in patients treated with inhaled and intranasal steroids compared with healthy individuals (Shroff et al., 2018;Zain et al., 2019).We hypothesised that participants who at least once in childhood had been treated continuously for a long period of time with inhaled or intranasal corticosteroid would have thinner RNFLs than participants because of the IOP increase that can result from such treatment.This cohort is ideal for investigating this association.The absence of corticosteroidrelated RNFL deficits in the present asthma-high-risk cohort is reassuring and in agreement with other studies, (Dereci et al., 2015) but the idiosyncratic nature of the corticosteroid-induced intraocular hypertension

Atopic dermatitis diagnosis
Present (n = 125) vs. none (n = 121) 0.9 (−2.1 to 3.9) 0. response means that continued awareness of this side effect is pertinent.Our finding of a lower birth weight being associated with a thinner RNFL complements the results obtained by another Danish study (Ashina et al., 2017).The decrease of the RNFL with increasing ocular axial length that we demonstrated in our study and previous studies (Ashina et al., 2017;Dhami et al., 2016;Eckmann-Hansen et al., 2021;Kang et al., 2016) may be a simple effect of eyeball elongation and distention of the retina.
The finding of a 4.7 μm deficit in macular thickness in participants exposed to maternal smoking during pregnancy corroborates our RNFL findings.It is also consistent with a study of newborn rats that found ganglion cell and inner plexiform layer deficits after gestational nicotine exposure (Evereklioglu et al., 2003).The selectivity of tobacco damage for the retinal ganglion cells is supported by a recent study of 1296 adolescents that found normal parafoveal cone density in the offspring of mothers who smoked during pregnancy (Eckmann-Hansen et al., 2021).
This study of young adults found that exposure to maternal smoking during pregnancy and passive smoking during childhood were associated with RNFL thickness deficits and that maternal smoking during pregnancy was associated with a thinner macula.These distinct and high-reproducible signs of damage to the central nervous system in relation to early-life exposure to tobacco smoke underlines the relevance of a continued effort to control and limit the spread of this toxic agent.Future follow-up examinations may reveal whether such RNFL thickness deficits predispose to glaucoma or other types of optic nerve disease.

F
The mean of retinal nerve fibre layer (RNFL) thickness in relation to total number of cigarettes smoked during pregnancy among the 60 participants whose mothers had smoked during pregnancy.Crude analysis.*Significant at a two-sided p < 0.05.F I G U R E 2 The mean of retinal nerve fibre layer (RNFL) thickness in relation to different subgroups of smoking.Crude analysis.*Significant at a two-sided p < 0.05.(1) No exposure to smoking during pregnancy, passive smoking during childhood, or smoking at 18 years of age (n = 91) [reference group] (2) Exposure to smoking during pregnancy but no exposure to passive smoking during childhood or smoking at 18 years (n = 4) (3) Exposure to passive smoking during childhood but no exposure to smoking during pregnancy or smoking at 18 years (n = 44) (4) Exposure to smoking during pregnancy and passive smoking during childhood but no exposure to smoking at 18 years of age (n = 30) (5) Exposure to smoking during pregnancy and passive smoking during childhood plus exposure to smoking at 18 years of age (n = 20).T A L 3 Macular thickness investigated variables in young adults aged 17-19 years.