Relation between dry eye and myopia based on tear film breakup time, higher order aberration, choroidal thickness, and axial length

The purpose of this study was to investigate the association between dry eye disease (DED) and myopia by evaluating higher order aberrations (HOAs) and choroidal thickness (CT). We recruited 72 myopic children with DED symptoms (mean age 12.8 years), measured the tear film breakup time (TBUT), corneal/intraocular/total ocular HOAs, CT, and axial length (AL), administered lifestyle questionnaires, and evaluated the relationships among TBUT, HOAs, CT, and AL. The TBUT was correlated significantly with the corneal HOAs and intraocular HOAs but not with the total ocular HOAs. Multiple regression analyses showed that the AL was associated significantly with the TBUT (β = − 0.067, P = 0.004), the intraocular HOAs, and total ocular HOAs but not with the corneal HOAs. The CT was associated significantly with the TBUT and AL (β = 9.15 and − 7.85, respectively; P < 0.001 and = 0.01, respectively). Our data suggested the association between DED and myopia might be independent of the HOAs. We showed that the TBUT was associated with the CT, which is related to the AL. Because the parasympathetic nervous system affects the lacrimal glands and CT, the parasympathetic nervous system might be a common upstream factor in the association between DED and myopia.

because we hypothesized that there is a relationship between DED and myopia, we quantified those conditions in the current study. The increased corneal and total ocular HOAs in patients with DED have been reported 22,23 and the previous studies have described the relationships between HOAs and myopia 5,12,16,17 . We also reported the association between DED and myopia in our previous study 5 . To investigate the relationship between DED and myopia, we must consider whether HOA is a confounder. Therefore, we performed all measurements, i.e., fluorescein tear film breakup time (TBUT), HOAs, and cycloplegic refraction/axial length simultaneously in the current study.
Several treatments are available for myopia progression, one of which is atropine eye drops [24][25][26] . Although the mechanism of the drop is unknown, the prevention of childhood myopia with 0.01% atropine eye drops preferentially affected subjects with a smaller mesopic pupil diameter, which means a parasympathetic tone-dominant state 26 . Therefore, there seems to be a relationship between the parasympathetic nervous system and myopia. The choroid, which is involved in the modulation of ocular elongation and control of the refractive error 27 , and the lacrimal glands 28 are both innervated by the parasympathetic nervous system; the choroidal thickness (CT) is correlated negatively with the AL [29][30][31] . In the current study, we measured the CT, AL, and TBUT, because we hypothesized that the parasympathetic nervous system may be a common upstream factor in both the TBUT and CT, so dry eye and myopia may be downstream results. In the current study, we report the relationships among the TBUT, HOAs, CT, and AL.  (Table 1). Our results showed that 94.9% of the myopic children had some symptoms of DED, and 73.6% and 54.7% of participants had TBUTs of < 10 and ≤ 5 s, respectively. Participants who had some DED symptoms and TBUT values of ≤ 5 s and those who had either severe DED symptoms or a history of clinically diagnosed DED evaluated by questionnaire 32 Table 2 shows the results of multiple regression analyses to identify the relationships between AL, cycloplegic objective refraction, and lifestyle factors including the TBUT using the questionnaires. Significant correlations were found between the AL and TBUT (β = − 0.067, P = 0.004), age (β = 0.065, P = 0.02), parental history of myopia (β = 0.497, P = 0.04), and time spent outdoors (β = − 0.027, P < 0.001). However, no significant correlations were found between cycloplegic objective refraction and any lifestyle factors including the TBUT (Table 2). Supplementary Figs. 1, 2, and 3 show the correlations and the Spearman's rank correlation coefficients (r) of the TBUT with the corneal, intraocular, and total ocular HOAs. Significant correlations were found between the TBUT and corneal HOAs (TBUT vs. corneal HOAs: SA, r = − 0.323, P = 0.02; S4, r = − 0.497, P < 0.001; THOA, r = − 0.362, P = 0.009) ( Supplementary Fig. 1) and the intraocular HOAs (TBUT vs. intraocular HOAs: S3, r = − 0.299, P = 0.04; S4, r = − 0.369, P = 0.008; THOA, r = − 0.368, P = 0.009) ( Supplementary Fig. 2). However, no significant correlations were found between the TBUT and total ocular HOAs ( Supplementary Fig. 3). Table 3 shows that multiple regression analyses did not find significant correlations between the AL, refraction, and corneal HOAs.  Table 4 shows that significant correlations were found between the AL and the intraocular S3 (β = − 21.8, P = 0.02) and the intraocular THOA (β = 21.2, P = 0.048) and between the cycloplegic refraction and the intraocular S3 (β = 96.9, P = 0.04) and intraocular THOA (β = − 112, P = 0.049). Table 5 shows the significant correlations between the AL and the total ocular S3 (β = − 41.7, P = 0.045) and the total ocular THOA (β = 43.6, P = 0.049). No significant correlations were found between the refraction and total ocular HOAs.

Discussion
The current study investigated the relationships among the TBUT, HOAs, CT, and AL. We confirmed a quantitative relationship between myopia and DED by evaluating the TBUT; the AL increased as the TBUT became shorter. The results also showed that the TBUT was correlated with the corneal HOAs and CT but not with the total ocular HOAs, and the AL was related to the total ocular HOAs and CT but not to the corneal HOAs. The relationships among the TBUT, HOAs, CT, and AL in the current study are summarized in Fig. 2. Our results Table 2. Results of multiple regression analyses to estimate the association between myopia and environmental factors. Two multiple regression models were used. Axial length or cycloplegic objective refraction was used as the outcome variable. P values less than statistically significant level (< 0.05) are marked in bold. For gender, boys were numbered 1 and girls were numbered 0. β coefficient, 95% CI 95% confidence interval, TBUT tear film breakup time, NA not applicable.  Table 3. Results of multiple regression analyses to estimate the association between myopia and corneal higher order aberrations (evaluated with natural pupillary diameters, average value φ = 6.1 mm). D diopters, SA spherical aberration, S3 3rd-order aberrations, S4 4th-order aberrations, THOA sum of the 3rd-to 6th-order aberrations, 95% CI 95% confidence interval, β coefficient. Two multiple regression models were used. Axial length or cycloplegic objective refraction was used as the outcome variable. P < 0.05 are highlighted. For gender, boys were numbered 1 and girls were numbered 0. www.nature.com/scientificreports/ Table 4. Results of multiple regression analyses to estimate the association between myopia and intraocular higher order aberrations (evaluated with natural pupillary diameters, average value φ = 6.1 mm). D diopters, SA spherical aberration, S3 3rd-order aberrations, S4 4th-order aberrations, THOA sum of the 3rd-to 6th-order aberrations, 95% CI 95% confidence interval, β coefficient Two multiple regression models were used. Axial length or cycloplegic objective refraction was used as the outcome variable. P < 0.05 are highlighted. For gender, boys were numbered 1 and girls were numbered 0.  Table 5. Results of multiple regression analyses to estimate the association between myopia and total ocular higher order aberrations (evaluated with natural pupillary diameters, average value φ = 6.1 mm). D diopters, SA spherical aberration, S3 3rd-order aberrations, S4 4th-order aberrations, THOA sum of the 3rd-to 6th-order aberrations, 95% CI 95% confidence interval, β coefficient. Two multiple regression models were used. Axial length or cycloplegic objective refraction was used as the outcome variable. P < 0.05 are highlighted. For gender, boys were numbered 1 and girls were numbered 0.  www.nature.com/scientificreports/ suggested that the TBUT is related to the CT and that the association between the TBUT and AL might be independent of the HOAs. Multiple studies have investigated the relationships between myopia and HOAs 5,12-15 . Philip et al. 13 indicated that the risk of becoming myopic increased as the total ocular SA decreased. Lau et al. 14 also reported that the amount of axial elongation decreased as the total ocular SA and total ocular THOA increased after adjusting for age, sex, and refractive error. However, Zhang et al. 15 reported that total ocular coma aberration, total ocular S3, and total ocular THOA were higher in children in whom myopia progressed rapidly (change of SE, > 0.5 D/year).
Although all of the previously mentioned studies have considered only total ocular HOAs 5,12-15 , a recent study of 64 myopic Japanese children (mean age 9.2 ± 1. 6 year, under cycloplegia) in whom the corneal HOAs and total ocular HOAs were measured showed that the corneal THOA was the variable that was most relevant to myopia progression and axial elongation among schoolchildren 35 . The current study found that the AL was related to intraocular HOAs and total ocular HOAs. Contradictions between the current and previous study 35 in which DED was not evaluated might be attributed to the consideration of DED. The current study found significant correlations between the TBUT and corneal/intraocular HOAs. Thus, we did not include the TBUT as an Table 6. Results of multiple regression analyses to estimate the association between choroidal thickness and dry eye disease. A multiple regression model was used. Choroidal thickness was used as the outcome variable. P < 0.05 are highlighted. For gender, boys were numbered 1 and girls were numbered 0. β coefficient, 95% CI 95% confidence interval, TBUT tear film breakup time.  The TBUT is associated significantly with the CT that is related to the AL. The TBUT is correlated with the corneal HOAs but not with the total ocular HOAs, whereas the AL is related to the total ocular HOAs but not to the corneal HOAs. Because the parasympathetic nervous system affects both the lacrimal glands and CT, the parasympathetic nervous system might be a common factor upstream in the association between the TBUT and the CT. www.nature.com/scientificreports/ independent variable in multiple regression analyses that examined the associations between myopia and HOAs because of the multicollinearity and found that the AL and refraction were associated with intraocular HOAs but not with corneal HOAs. Together with the previous study 35 , our data suggested that DED was a confounder in the associations between myopia and HOAs.
To date, few studies have evaluated the relationship between DED and myopia 5,36 . Ilhan et al. 36 reported that adult patients with high myopia (SE < − 6.0 D) had a higher incidence of DED. Our previous pediatric study also showed that DED assessed by questionnaires was correlated with the AL and non-cycloplegic refraction 5 . In the current study, we quantified the DED severity by measuring the TBUT and examined its relationships with AL and cycloplegic refraction. The prevalence of DED in this study (51.9%) was higher than that of childhood DED (mean age 15.2 ± 5.6 years) reported previously (0.4%) 37 . Our results indicated a quantitative relationship between the TBUT and AL, suggesting myopia may be related to at least one measure of DED.
The corneal HOAs and total ocular HOAs were elevated in DED 22,23,38,39 , especially in DED with a short TBUT 40 , due to irregular astigmatism in the tear film 41 . However, all previous studies included adults 22,23,38,39 , and the associations between HOAs and DED among children have yet to be investigated. In the current study, we showed that corneal HOAs but not total ocular HOAs increased as the TBUT shortened in children. Surprisingly, the intraocular HOAs also became elevated as the TBUT shortened in the current study. Because we evaluated the HOAs with the RMS value, we could not judge the positivity or negativity. The intraocular SA, which was not evaluated with the RMS value, had the opposite correlation tendency toward the TBUT compared to the corneal SA. This indicated that the signs of the corneal HOAs and intraocular HOAs might be opposite, and the corneal HOAs related to DED might be compensated for by the intraocular HOAs.
To the best of our knowledge, this is the first study in which the CT was associated significantly with the TBUT. Previous studies have reported the crucial role of the choroid in the modulation of ocular elongation and homeostatic control of the refractive error 27 . Parapapillary diffuse choroidal atrophy, which involves extreme thinning of parapapillary choroid 42 , is a potential precursor of pathologic myopia 43 . The CT is correlated negatively with the AL in patients with high myopia 29,30 and healthy subjects 31 Our results also showed that the CT is correlated negatively with the AL. We can suggest two plausible explanations for the association between the TBUT and CT. First, since increased parasympathetic tone enhances both the CT 27 and the amount of tear secretion from the lacrimal glands 28 , our results suggested that a "factor X", including the parasympathetic nervous system, might be a common factor upstream in the association between the TBUT (DED) and CT (myopia) (Fig. 2). Another explanation might be simpler, i.e., those who perform near work have a greater increased risk of both evaporative DED, which directly affects the TBUT, and myopia, which affects the CT.
The current study had limitations. First, because it was not a longitudinal study, it could not establish causality. Second, the sample size was small. Third, selection bias was possible because the study included only those who visited our clinic. Fourth, we cannot directly evaluate the lacrimal gland function based on the TBUT and should perform other tests such as the Schirmer's test or the phenol red test to directly evaluate secretion from the lacrimal glands in a future study.

Materials and methods
Study design and study populations. This cross-sectional study adhered to the tenets of the Declaration of Helsinki and was approved by the Keio University School of Medicine Ethics Committee (approval number: 20180189). Since ethical guidelines for clinical studies by Japanese Ministry of Health, Labor and Welfare indicate that researchers do not need to obtain written informed consent from each patient for studies not involving biologic tissue but only to review the medical records retrospectively, we displayed our ethical statement and written guidelines for the current study on our website in Japanese 44 . The requirement for obtaining written informed consent was waived by the Keio University School of Medicine Ethics Committee.
The inclusion criterion was presentation to the Keio University Hospital Myopia Clinic from July 2017 to December 2019. A total of 84 consecutive patients participated during that period and none had been treated with atropine or underwent any ocular surgery. The exclusion criteria included either hyperopia or emmetropia (n = 6), best-corrected visual acuity of < 20/20 (n = 2, amblyopia), and the absence of symptoms of either foreignbody sensation or ocular dryness (n = 4). Ultimately, 72 (85.7%) had data available for our study. www.nature.com/scientificreports/ square (RMS) values from the 3rd-to 6th-order Zernike coefficients were calculated. From these Zernike coefficients, the spherical aberration (SA), 3rd-order aberrations (S3), 4th-order aberrations (S4), and THOA were calculated. Each intraocular HOA was calculated by subtracting the respective corneal HOA from the total ocular HOA. Some patients wore contact lenses for vision correction (n = 13), and we did not measure their TBUT and HOAs and measured the AL, cycloplegic refraction, non-cycloplegic refraction, CT, UCVA, and BCVA when the contact lens were removed. The patients completed a lifestyle questionnaire that included factors such as time spent outdoors, doing near work, and sleeping; symptoms of DED; and parental history of myopia. We defined time spent outdoors as the average number of hours spent outdoors daily that was calculated using the following formula: [(hours spent on a weekday) × 5 + (hours spent on a weekend day) × 2]/7 46 . Near work included studying; reading books; using a computer, tablet. or smartphone; watching television, and playing video or portable games. The DED symptoms recorded using the Women's Health Study questionnaire 32 in Japanese 33,34 .
We defined myopia as a spherical equivalent (SE) of ≤ − 0.5 diopter (D) in the current study; only the results from the left eye are presented.
Sample size calculation. The sample size calculation indicated that a sample size of at least 48 cases would be required (the minimally acceptable correlation coefficient between CT and BUT was 0.45 [r = 0.45], so 48 cases were needed to have a 5% alpha level and 90% power).

Statistical analysis.
We analyzed the correlations among the TBUT and HOAs, TBUT and CT, and CT and AL by calculating Spearman's rank correlation coefficient. Using multiple regression analysis, we examined the associations between the outcomes (AL and cycloplegic objective refraction) and other factors including age, gender, parental history of myopia, time spent outdoors, time spent on near work, reading distance, sleeping time, and TBUT. We also investigated the associations between outcomes (AL and cycloplegic objective refraction) and other factors including age, gender and HOAs (SA, S3, S4, and THOA) of either the corneal, intraocular or total ocular HOAs by multiple regression analysis. Finally, to evaluate the associations between CT and other factors including age, gender, TBUT, and AL, we performed multiple regression analyses. In all multiple regression analyses, we evaluated the age and gender by forced entry and the other factors by stepwise analysis because age and gender have already been reported as factors associated with myopia 47 .
All statistical analyses were performed using a statistical analysis software (R version 4.0.0, R Foundation for Statistical Computing, Vienna, Austria). All P values were considered significant if < 0.05.

Conclusions
We showed that the TBUT is associated significantly with the CT, which is correlated with the AL. Because the parasympathetic nervous system affects the lacrimal glands and CT, we suggest that the parasympathetic nervous system might be a common factor upstream in the association between DED and myopia. Further study is needed to confirm this hypothesis.

Data availability
Debabrata Hazra and Erisa Yotsukura had full access to all the study data and take responsibility for the integrity of the data and the accuracy of the data analysis. The data that support the findings of this study are available from Debabrata Hazra and Erisa Yotsukura but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of Debabrata Hazra and Erisa Yotsukura.