Impact of Bariatric Surgery-Induced Weight Loss on Anterior Eye Health in Patients with Obesity

The aim of the present research was to assess the effect of bariatric surgery-induced weight loss on the tear film and ocular surface of patients with obesity. A total of 29 participants with obesity (aged 47.2 ± 10.1 years, 8 male) were measured at baseline and followed up one year after Roux-en-Y gastric bypass (RYGB) surgery. General anthropometric data, as well as serum lipid markers of cholesterol, were assessed in all individuals. Bilateral anterior eye measurements of tear meniscus height (TMH), non-invasive tear breakup time, bulbar and limbal redness and infrared meibography were captured using the Keratograph K5M (Oculus) and ocular surface damage was evaluated using fluorescein sodium and lissamine green staining. Bariatric surgery resulted in significant loss of weight (body mass index p < 0.001) and an improvement in the blood lipid profile (p < 0.01) in all participants. However, there were no statistically significant differences between the baseline and one-year follow-up for any of the measured clinical ocular surface and tear film variables (all p > 0.05). Although there were trends for a reduced TMH and a decrease in meibomian gland dropout after bariatric surgery, these differences were also insignificant (p > 0.05). In conclusion, weight reduction through bariatric surgery did not have an effect on the tear film or ocular surface in unselected patients with obesity.


Introduction
The World Health Organization (WHO) defines obesity as a body mass index (BMI) greater than 30 kg/m 2 and the international obesity task force has estimated that worldwide there are around 340 million people who meet this criterion, and counting [1].
The impact of obesity on health and its deleterious effects on the body are well known [2]. Overweight and obesity are established risk factors for the development of cardiovascular (CVD) [1] and metabolic diseases such as type 2 diabetes (T2DM) [3]. Obesity is also an important risk factor for other non-communicable diseases such as stroke, asthma and cancer [3].
In addition to its systemic complications, some studies have also established associations between obesity and various ocular pathologies, such as glaucoma [4,5], cataracts [6,7], diabetic retinopathy (DR) [8,9] and floppy eyelid syndrome (FES) [10,11]. Nevertheless, the effect of obesity on the ocular surface's health is less clear, with many conflicted reports. Indeed, although obesity is associated with abnormal lipid metabolism such as dyslipidaemia [12] which, in turn, is associated with dry eye disease (DED) [13,14], to date there is no research strongly linking obesity to DED. In addition, although high levels of body fat have been associated with dry eye symptoms in a general population, a high body mass index (BMI), which represents the main parameter considered when a person is classified as

Materials and Methods
The study was designed and conducted in accordance with the Declaration of Helsinki and the protocol received positive opinions and governance approval from the hospital and university research ethics committees prior to the commencement of the study.
Participants were recruited from the hospital trusts' weight management clinics by a specialist consultant following written informed consent. To be suitable candidates for bariatric surgery, the patients had to be between 18 and 65 years old, have a BMI > 40 kg/m 2 and be compliant with a dietary plan for at least 12 months. Participants were excluded if they had a positive diagnosis of cardio-or cerebrovascular disease and renal disease. All participants were screened for ocular disease and were excluded from the study if they had any ocular pathologies, an intraocular pressure (IOP) > 24 mmHg consistently, moderate to high cataracts and if they had a history of intraocular surgery. The existence of ocular surface damage was neither an inclusion nor exclusion criteria for participants; however, patients with symptomatic red eye (>2 grade using an Efron grading scale) [22,23] were not included in the study.
All included participants underwent RYGB surgery according to a standard technique. Baseline measurements were performed 1 month prior to the participant undergoing surgery and the follow-up was performed 12 months after the procedure. The assessments are detailed below.
Standard anthropometric measures of height and weight were recorded to determine body mass index (BMI = weight/height). Systolic blood pressure (SBP), diastolic blood pressure (DBP) and heart rate (HR) were measured using an automatic blood pressure monitor (UA-767; A&D Instruments Ltd., Abingdon, UK) to determine mean arterial pressure (MAP = 2/3 DBP + 1/3 SBP) [24]. Intra-ocular pressure (IOP) readings were obtained using non-contact tonometry (Pulsair; Keeler Ltd., Winsor, UK). In addition, blood and plasma samples drawn from the antecubital fossa vein were assessed immediately for TG, total cholesterol (T-CHOL) and HDL-C using the Reflotron Desktop Analyzer (Roche Diagnostics, Burges Hill, UK). Low-density lipoprotein cholesterol (LDL-C) values were calculated as per the Friedewald equation [25]. Measures of tear film meniscus height (TMH: average of 3 consecutive readings immediately below the pupil while in primary gaze using the built-in free-hand digital calipers) [26], objective non-invasive breakup time (NIBUT): initial and average of 3 consecutive readings while in primary gaze using the Placido disc and objective image analysis software [27] and objective bulbar and limbal hyperemia (graded against the Efron grading scale) [22,23] were obtained using the Oculus Keratoscope K5M(Optikgerate GmbH, Wetzlar, Germany) ( Figure 1). total cholesterol (T-CHOL) and HDL-C using the Reflotron Desktop Analyzer (Roche Diagnostics, Burges Hill, UK). Low-density lipoprotein cholesterol (LDL-C) values were calculated as per the Friedewald equation [25].
Measures of tear film meniscus height (TMH: average of 3 consecutive readings immediately below the pupil while in primary gaze using the built-in free-hand digital calipers) [26], objective non-invasive breakup time (NIBUT): initial and average of 3 consecutive readings while in primary gaze using the Placido disc and objective image analysis software [27] and objective bulbar and limbal hyperemia (graded against the Efron grading scale) [22,23] were obtained using the Oculus Keratoscope K5M(Optikgerate GmbH, Wetzlar, Germany) ( Figure 1).

Figure 1.
Tear film and ocular surface parameters assessed: (A) tear meniscus height (TMH) measured using digital calipers and non-invasive tear breakup time (NIBUT) determined by the objectively assessed disruption of projected infrared light Placido disc reflected from the tear film; (B) objectively graded redness of the bulbar (over the white of the eye) and limbal (around the colored iris) regions; (C) infrared light imaging of the meibomian glands under the upper (C) and lower (D) eyelid.
Meibomian gland dropout (loss) was calculated after image analysis with ImageJ software (V1.49) based on well-established grading systems [28][29][30] following infrared photography of the upper and lower palpebral surfaces with the K5M [31]. Finally, ocular surface staining was assessed using fluorescein sodium (BioFluoro, Gallowstown, Ireland) and lissamine green (GreenGlo, Hub Pharmaceuticals, Plymouth, MI, USA) ophthalmic dyes applied to the outer canthus [32]. The same measurements were obtained at the follow-up conducted 12 months post-surgery, at the same time of day (±1 h) when the initial measurements were performed for each participant. Staining images captured with the Oculus Keratograph K5m excited with blue light and imaged through a yellow cut-off filter [33] were analyzed with ImageJ software (National Institutes of Health) and the staining area outlined manually and compared to the corneal area. Meibomian gland dropout (loss) was calculated after image analysis with ImageJ software (V1.49) based on well-established grading systems [28][29][30] following infrared photography of the upper and lower palpebral surfaces with the K5M [31]. Finally, ocular surface staining was assessed using fluorescein sodium (BioFluoro, Gallowstown, Ireland) and lissamine green (GreenGlo, Hub Pharmaceuticals, Plymouth, MI, USA) ophthalmic dyes applied to the outer canthus [32]. The same measurements were obtained at the follow-up conducted 12 months post-surgery, at the same time of day (±1 h) when the initial measurements were performed for each participant. Staining images captured with the Oculus Keratograph K5m excited with blue light and imaged through a yellow cut-off filter [33] were analyzed with ImageJ software (National Institutes of Health, Bethesda, MD, USA) and the staining area outlined manually and compared to the corneal area.

Statistical Analysis
Statistical analyses were performed using Statistica ® software (StatSoft Inc., Version 13, Tulsa, OK, USA). As the data did not differ from a normal distribution (Shapiro-Wilks test), differences between the clinical investigations at baseline and follow-up were assessed by using a paired t-test. Differences between ocular measurements at baseline and follow-up were assessed by t-test or analysis of covariance (ANCOVA) where applicable. Multivariate analyses were performed to test the influence of age, clinical parameters and circulating markers on the ocular variables. Statistical significance was defined at p < 0.05. The sample size was calculated using G-power software [34]. Power calculations were based on similar studies that used the K5M system for the assessment for dry eye disease and infrared meibography [23,29,35,36]. Therefore, to provide a statistical power of 80% and medium effect size with an alpha level of 0.05, a sample size of n = 27 participants was required.

Results
A total of 40 participants were recruited for study inclusion to allow for dropout and completed all the baseline measurements; 11 participants were lost to follow-up. The remaining 29 participants (aged 47.2 ± 10.1, range 28-60 years, 11 male) attended the follow-up 12 months post-surgery to complete the study and therefore were included in the analysis.
The general characteristics of the study group at baseline and follow-up, are outlined in Table 1.  Bariatric surgery resulted in a significant mean reduction in BMI (by 10.82 kg/m 2 ; 13, Tulsa, OK, USA). As the data did not differ from a normal distribution (Shapiro-Wilks test), differences between the clinical investigations at baseline and follow-up were assessed by using a paired t-test. Differences between ocular measurements at baseline and follow-up were assessed by t-test or analysis of covariance (ANCOVA) where applicable. Multivariate analyses were performed to test the influence of age, clinical parameters and circulating markers on the ocular variables. Statistical significance was defined at p < 0.05. The sample size was calculated using G-power software [34]. Power calculations were based on similar studies that used the K5M system for the assessment for dry eye disease and infrared meibography [23,29,35,36]. Therefore, to provide a statistical power of 80% and medium effect size with an alpha level of 0.05, a sample size of n = 27 participants was required.

Results
A total of 40 participants were recruited for study inclusion to allow for dropout and completed all the baseline measurements; 11 participants were lost to follow-up. The remaining 29 participants (aged 47.2 ± 10.1, range 28-60 years, 11 male) attended the followup 12 months post-surgery to complete the study and therefore were included in the analysis.
The general characteristics of the study group at baseline and follow-up, are outlined in Table 1.  Bariatric surgery resulted in a significant mean reduction in BMI (by 10.82 kg/m 2 ; -22%; p < 0.001), as well as reductions in SBP (by 15.49 mmHg; -11%; p < 0.001), DBP (by 5.62 mmHg; -7%; p = 0.039) and HR (by 4.93 bpm; -7%; p = 0.046). In addition, blood lipid profiles also showed improvements, with statistically significant reductions in CHOL (by 0.37 mmol/L; -8%; p = 0.003), LDL-C (by 0.48 mmol/L; -16%; p < 0.001) and TG (by 0.31 mmol/L; -21%; p = 0.002), and increases in HDL-C (by 0.27 mmol/L; -21%; p < 0.001). (all Table 1) Pre-surgery, seventy-six percent of participants (n = 22) had >5% corneal or conjunctival staining and approximately half (45%, n = 13) had an unstable tear film (NIBUT < 10 s). However, there was no statistically significant change between the baseline and at 12 months post-surgery for any of the measured ocular surface characteristics (all p > 0.05, Table 2). There was a trend for the amelioration of meibomian gland dropout, a reduction 22%; p < 0.001), as well as reductions in SBP (by 15.49 mmHg; 13, Tulsa, OK, USA). As the data d test), differences between the clin sessed by using a paired t-test. Dif follow-up were assessed by t-test o Multivariate analyses were perform circulating markers on the ocular v The sample size was calculated u based on similar studies that used and infrared meibography [23,29, and medium effect size with an alp required.

Results
A total of 40 participants wer completed all the baseline measur maining 29 participants (aged 47.2 up 12 months post-surgery to com ysis.
The general characteristics of in Table 1.
Bariatric surgery resulted in -22%; p < 0.001), as well as reducti 5.62 mmHg; -7%; p = 0.039) and H profiles also showed improvemen 0.37 mmol/L; -8%; p = 0.003), LDL mmol/L; -21%; p = 0.002), and incr Table 1) Pre-surgery, seventy-six perc tival staining and approximately h s). However, there was no statisti months post-surgery for any of th Table 2). There was a trend for the 11%; p < 0.001), DBP (by 5.62 mmHg; Statistical analyses were performed using Statistica ® software (StatSoft Inc., Ver 13, Tulsa, OK, USA). As the data did not differ from a normal distribution (Shapiro-W test), differences between the clinical investigations at baseline and follow-up were sessed by using a paired t-test. Differences between ocular measurements at baseline follow-up were assessed by t-test or analysis of covariance (ANCOVA) where applica Multivariate analyses were performed to test the influence of age, clinical parameters circulating markers on the ocular variables. Statistical significance was defined at p < 0 The sample size was calculated using G-power software [34]. Power calculations w based on similar studies that used the K5M system for the assessment for dry eye dis and infrared meibography [23,29,35,36]. Therefore, to provide a statistical power of and medium effect size with an alpha level of 0.05, a sample size of n = 27 participants required.

Results
A total of 40 participants were recruited for study inclusion to allow for dropout completed all the baseline measurements; 11 participants were lost to follow-up. Th maining 29 participants (aged 47.2 ± 10.1, range 28-60 years, 11 male) attended the foll up 12 months post-surgery to complete the study and therefore were included in the a ysis.
The general characteristics of the study group at baseline and follow-up, are outl in Table 1.   Table 1) Pre-surgery, seventy-six percent of participants (n = 22) had >5% corneal or conj tival staining and approximately half (45%, n = 13) had an unstable tear film (NIBUT s). However, there was no statistically significant change between the baseline and a months post-surgery for any of the measured ocular surface characteristics (all p > 0 Table 2). There was a trend for the amelioration of meibomian gland dropout, a reduc 7%; p = 0.039) and HR (by 4.93 bpm; Statistical analyses were performed usin 13, Tulsa, OK, USA). As the data did not diffe test), differences between the clinical invest sessed by using a paired t-test. Differences be follow-up were assessed by t-test or analysis Multivariate analyses were performed to test circulating markers on the ocular variables. S The sample size was calculated using G-po based on similar studies that used the K5M s and infrared meibography [23,29,35,36]. The and medium effect size with an alpha level of required.

Results
A total of 40 participants were recruited completed all the baseline measurements; 11 maining 29 participants (aged 47.2 ± 10.1, ran up 12 months post-surgery to complete the st ysis.
The general characteristics of the study g in Table 1. Bariatric surgery resulted in a significa -22%; p < 0.001), as well as reductions in SBP 5.62 mmHg; -7%; p = 0.039) and HR (by 4.93 profiles also showed improvements, with sta 0.37 mmol/L; -8%; p = 0.003), LDL-C (by 0.4 mmol/L; -21%; p = 0.002), and increases in H Table 1) Pre-surgery, seventy-six percent of parti tival staining and approximately half (45%, n s). However, there was no statistically signif months post-surgery for any of the measure Table 2). There was a trend for the ameliorati 7%; p = 0.046). In addition, blood lipid profiles also showed improvements, with statistically significant reductions in CHOL (by 0.37 mmol/L;

Statistical Analysis
Statistical analyses were performed using Statistica ® software (StatSoft Inc. 13, Tulsa, OK, USA). As the data did not differ from a normal distribution (Shapi test), differences between the clinical investigations at baseline and follow-up sessed by using a paired t-test. Differences between ocular measurements at bas follow-up were assessed by t-test or analysis of covariance (ANCOVA) where ap Multivariate analyses were performed to test the influence of age, clinical param circulating markers on the ocular variables. Statistical significance was defined a The sample size was calculated using G-power software [34]. Power calculati based on similar studies that used the K5M system for the assessment for dry ey and infrared meibography [23,29,35,36]. Therefore, to provide a statistical powe and medium effect size with an alpha level of 0.05, a sample size of n = 27 particip required.

Results
A total of 40 participants were recruited for study inclusion to allow for dro completed all the baseline measurements; 11 participants were lost to follow-up maining 29 participants (aged 47.2 ± 10.1, range 28-60 years, 11 male) attended th up 12 months post-surgery to complete the study and therefore were included in ysis.
The general characteristics of the study group at baseline and follow-up, are in Table 1.   Table 1) Pre-surgery, seventy-six percent of participants (n = 22) had >5% corneal or tival staining and approximately half (45%, n = 13) had an unstable tear film (NI s). However, there was no statistically significant change between the baseline months post-surgery for any of the measured ocular surface characteristics (all Table 2). There was a trend for the amelioration of meibomian gland dropout, a r 8%; p = 0.003), LDL-C (by 0.48 mmol/L;

Statistical Analysis
Statistical analyses were perform 13, Tulsa, OK, USA). As the data did n test), differences between the clinica sessed by using a paired t-test. Differ follow-up were assessed by t-test or a Multivariate analyses were performe circulating markers on the ocular var The sample size was calculated usin based on similar studies that used the and infrared meibography [23,29,35,3 and medium effect size with an alpha required.

Results
A total of 40 participants were re completed all the baseline measurem maining 29 participants (aged 47.2 ± 1 up 12 months post-surgery to comple ysis.

Statistical Analysis
Statistical analyses were performed using Statistica ® software (StatSoft Inc., Ve 13, Tulsa, OK, USA). As the data did not differ from a normal distribution (Shapiro-W test), differences between the clinical investigations at baseline and follow-up wer sessed by using a paired t-test. Differences between ocular measurements at baseline follow-up were assessed by t-test or analysis of covariance (ANCOVA) where applic Multivariate analyses were performed to test the influence of age, clinical parameters circulating markers on the ocular variables. Statistical significance was defined at p < The sample size was calculated using G-power software [34]. Power calculations based on similar studies that used the K5M system for the assessment for dry eye di and infrared meibography [23,29,35,36]. Therefore, to provide a statistical power of and medium effect size with an alpha level of 0.05, a sample size of n = 27 participants required.

A
total of 40 participants were recruited for study inclusion to allow for dropou completed all the baseline measurements; 11 participants were lost to follow-up. Th maining 29 participants (aged 47.2 ± 10.1, range 28-60 years, 11 male) attended the fo up 12 months post-surgery to complete the study and therefore were included in the ysis.

Statistical Analysis
Statistical analyses 13, Tulsa, OK, USA). As test), differences betwee sessed by using a paired follow-up were assessed Multivariate analyses we circulating markers on th The sample size was ca based on similar studies and infrared meibograp and medium effect size w required.

Results
A total of 40 particip completed all the baselin maining 29 participants up 12 months post-surge ysis.

Discussion
By using participants as their own controls, this study has assessed the effect of significant weight loss after bariatric surgery on the health of the ocular surface and tear film. However, despite a significant reduction in weight and an improvement in the circulatory lipid levels, there was no significant effect of bariatric surgery on the ocular surface health in our participants, even in those with pre-existing signs of ocular surface damage. This finding is similar to an earlier report that included pre-and post-operative measurements in different groups of patients [21]. Follow-up was performed 12 months after the procedure, a period of time that is considered appropriate to allow changes in signs and symptoms to occur [37].
Our results could partially be explained by the heterogeneity of the participant cohort. Some of the participants already had a relatively good tear secretion pre-surgery, with tear stability and tear volume of the cohort on average at the borderline of normality [32], while others had various degrees of damage. Indeed, although the link between obesity and various ocular surface damage is not clear, the role of adipokines, secreted by the adipose tissue, in causing systemic chronic inflammation is well established [38]. Moreover, studies performed on animals demonstrated a link between adiponectin, a 30-kDa multimeric protein that is mainly secreted by white adipose tissue, and tear secretion in ageing mice [39]. Therefore, the presence of dry eye and/or various degrees of ocular surface inflammation in individuals with obesity is perfectly plausible. Nevertheless, the average meibomian gland loss (dropout) assessed in our obese patients was not at a level that is believed to impact the overall secretion of meibum [40].
All participants included in our study underwent a Roux-en Y gastric bypass (RYGB) procedure, which represents a mixture of restrictive and malabsorptive techniques that are key for improving circulatory CHOL (HDL, LDL and Total) and TG. However, this procedure is also known to greatly increases the risk for various nutritional deficiencies, including those for vitamin B12, folate [41] and vitamin A [42], all very important in the ocular surface health [43][44][45]. In addition, deficiency of vitamin B6, B12 and folate can also cause an increase in the level of circulating homocysteine (Hcy). High Hcy levels are linked to a higher risk for CVD, but also for the occurrence of dry eye [46] as well as of other ocular conditions, such as glaucoma and retinal artery/vein occlusions [47,48]. Indeed, there is some evidence showing that Hcy levels are modified after RYGB [49,50]. Although we did not assess the levels of vitamin B, folate or Hcy, we found that the levels of CHOL and TG were indeed improved after the procedure. If this effect counteracted the one linked to high Hcy levels and, on average, contributed to the lack of significant changes at the ocular surface level can only be a hypothesis at this stage.
The level of meibomian gland dropout in our cohort was similar to what would be expected for this age group in the general population [51] and did not change significantly with weight loss. The role of obesity on the occurrence of meibomian dysfunction is unclear, although dyslipidemia has been found to potentially play a role [52]. In addition, beside its positive impact on circulatory lipids, bariatric surgery is also known to improve insulin resistance [53] and it has been demonstrated that androgenic hormones and insulin may be necessary for the growth and normal functioning of the meibomian gland [54]. Therefore, even if a direct link between the meibomian gland and weight loss after bariatric surgery could not be demonstrated in the present study, it could be hypothesized that it is possible to have a certain beneficial effect on the meibomian gland functioning. More research is necessary to conform this hypothesis.
The main limitations of our study lie in the fact that the participants were consecutive patients with obesity, recruited from a specialist clinic without a pre-study screening for symptoms or signs of ocular surface disease. Circulatory levels of various nutrients were also not assessed. Nevertheless, our sample was a true representation of the heterogeneity of patients with obesity undergoing this procedure and, in this type of population, it can be concluded that, despite significant positive effects on the measured systemic and circulatory parameters, bariatric surgery does not have measurable effects on the ocular surface health of these individuals.  Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical considerations.