Association of Dry Eye Disease with Dyslipidemia and Statin Use

Introduction

Dry eye disease (DED) affects a significant proportion of the general population, with an estimated prevalence of 7–33%.^1–3 The incidence and severity of DED increase with age and female gender.¹ This multifactorial condition is characterized by decreased tear production and/or increased evaporation and can lead to symptoms of ocular discomfort, visual disturbances, and a diminished quality of life.^4,5 Meibomian gland dysfunction (MGD) is the most common cause of evaporative dry eye,⁶ and involves abnormalities in the quantity and/or composition of tear film lipids, including excess free cholesterol.⁷ The relationship between systemic lipid abnormalities and those of the tear film has not been clearly established.⁸ However, prior studies, though conflicting, suggest an association between MGD/DED and dyslipidemia.^9–13

Dyslipidemia is a significant risk factor for cardiovascular disease, affecting an estimated 12% of adults.¹⁴ It is most often treated with “statin” medications.¹⁵ This drug class is comprised of HMG-CoA reductase inhibitors, which block the rate-limiting step in the biosynthesis of cholesterol.¹⁶ HMG-CoA reductase expression has been identified within the sebaceocytes of meibomian glands in human eyelid tissue.¹⁶ Thus, it is possible that statin use could alter cholesterol synthesis and lipid homeostasis within the meibomian glands, leading to destabilization of the tear film and subsequent DED.

To investigate a potential association between DED and each of a history of statin use or dyslipidemia, we conducted a study of a large number of patients seen at University of North Carolina (UNC)-affiliated healthcare facilities over a 10-year period.

Methods

This was a retrospective case-control study with approval obtained from the institutional review board of UNC. All methods described herein adhered strictly to the tenets of the Declaration of Helsinki and Health Insurance Portability and Accountability Act regulations. The dataset was acquired from the Carolina Data Warehouse for Health (CDWH), a repository of de-identified patient information collected from patient visits at UNC-affiliated hospitals and outpatient clinics. Using an online interface linked to the CDWH, 72,931 unique patients older than 18 years of age, with prior lipid panel results, and seen at UNC ophthalmology clinics from May 1, 2008 to May 31, 2018, were identified.

Queries were performed to identify unique individuals amongst this group carrying a diagnosis of DED (ICD-9 and ICD-10 codes 370.33, 375.15, H04.12x, and H16.22x). Additional queries were carried out to identify patients with a history of low-intensity, moderate-intensity, and high-intensity statin use as well as those with dyslipidemia. Low-intensity statins included fluvastatin 20–40 mg daily, lovastatin 20 mg daily, pitavastatin 1 mg daily, pravastatin 10–20 mg daily, and simvastatin 10 mg daily. Moderate-intensity included atorvastatin 10–20 mg daily, fluvastatin 40 mg twice a day or 80 mg daily, lovastatin 40 mg daily, pitavastatin 2–4 mg daily, pravastatin 40–80 mg daily, rosuvastatin 5–10 mg daily, and simvastatin 20–40 mg daily. High-intensity included atorvastatin 40–80 mg daily and rosuvastatin 20–40 mg daily.¹⁷ Abnormal lipid values were total cholesterol >200 mg/dL, HDL <40 mg/dL, LDL >130 mg/dL, and triglycerides >150 mg/dL.¹⁸ Statin categories were constructed to be unique and not double count individuals.

Upon data acquisition, exclusion of individuals with confounding factors well known to be associated with DED was performed. Individuals were excluded if they used specific medications associated with DED (tricyclic antidepressants, antihistamines, or diuretics), had a history of rheumatoid arthritis, Sjogren’s disease, or lupus (ICD-9 and ICD-10 codes 710.0, 710.2, 714.0, M32.x, M35.x, M05.79, M05.89, M06.09, and M06.89), or a history of cataract or refractive surgery (CPT codes 66984, S0800, and S0810). Odds ratios (ORs) and their associated 95% confidence intervals (CIs) were calculated between DED and each of the above parameters and further stratified by age. All data were analyzed using SAS software version 9.4 (SAS Inc., Cary, NC).

Results

A total of 72,931 patients were considered for inclusion in the study. After excluding individuals with the confounding factors identified above, the analyzed cohort consisted of 39,336 individuals, of which 53.9% were female. Demographic characteristics of the analyzed population are outlined in Table 1. In total, 3,399 patients (8.6%) carried a diagnosis of DED. Low-intensity, moderate-intensity, and high-intensity statin regimens were used by 751 (1.9%), 2,655 (6.8%), and 1,036 (2.6%), respectively. Lipid abnormalities were found in the following numbers of patients: Total cholesterol >200 mg/dL, 4,558 (11.6%); HDL <40 mg/dL, 2,078 (5.3%); LDL >130 mg/dL, 2,756 (7.0%); and TGs >150 mg/dL, 2,881 (7.3%).

Table 1.

Demographics of the analyzed study population.

Demographic	Number (%) of Patients (n = 39,336)
Sex
Male	18,149 (46.1)
Female	21,187 (53.9)
Race
White	22,701 (57.7)
African American	6,862 (17.4)
Asian	1,194 (3.0)
Native American or Alaskan	177 (0.4)
Other or Unknown	8,402 (21.4)
Age Group, years
18–34	11,259 (28.6)
35–54	10,334 (26.3)
55–64	6,070 (15.4)
≥ 65	11,673 (29.7)

Tables 2 and 3 display the calculated odds ratios and their associated 95% confidence intervals for this cohort. The odds of carrying a diagnosis of DED given the presence of low-intensity, moderate-intensity, and high-intensity statin use were 1.39 (1.13,1.72), 1.47 (1.30,1.65), and 1.46 (1.21,1.75), respectively, when compared to patients not taking statins. The odds of carrying a diagnosis of DED given the presence of total cholesterol >200 mg/dL, HDL <40 mg/dL, LDL >130 mg/dL, and TGs >150 mg/dL were 1.66 (1.52,1.82), 1.45 (1.26,1.67), 1.55 (1.39,1.74), and 1.43 (1.27,1.61), respectively.

Table 2.

Calculated odds ratios and corresponding confidence intervals between statin therapy and dry eye disease.

Exposure	Age Group (years)
	18–34	35–54	55–64	≥ 65	All Ages
Low-intensity Statin therapy	N/A*	0.85 (0.44,1.62)	1.00 (0.62,1.63)	1.71 (1.33,2.21)	1.39 (1.13,1.72)
Moderate-intensity Statin Therapy	6.96 (2.63,18.42)	1.13 (0.81,1.58)	1.37 (1.08,1.74)	1.57 (1.35,1.82)	1.47 (1.30,1.65)
High-intensity Statin Therapy	17.50 (5.99,51.13)	1.15 (0.69,1.90)	1.28 (0.89,1.85)	1.54 (1.21,1.95)	1.46 (1.21,1.75)

^*Note: queries for males and females aged 18–34 returned with sample sizes too small to properly calculate odds ratios in this age range.

Table 3.

Calculated odds ratios and corresponding confidence intervals between specified serum lipid levels and dry eye disease.

Exposure	Age Group (years)
	18–34	35–54	55–64	≥ 65	All Ages
Elevated Total Cholesterol	1.44 (0.82,2.55)	1.38 (1.15,1.66)	1.28 (1.06,1.54)	2.16 (1.89,2.47)	1.66 (1.52,1.82)
Low HDL	1.01 (0.47,2.16)	1.30 (1.00,1.70)	1.62 (1.21,2.15)	1.52 (1.22,1.89)	1.45 (1.26,1.67)
High LDL	1.11 (0.45,2.73)	1.48 (1.18,1.85)	1.18 (0.94,1.48)	1.92 (1.63,2.26)	1.55 (1.39,1.74)
Elevated Triglycerides	1.88 (1.08,3.26)	1.41 (1.13,1.76)	1.21 (0.96,1.54)	1.55 (1.31,1.85)	1.43 (1.27,1.61)

Discussion

We report a retrospective analysis of a large, diverse population of patients, analyzing the association of DED with statin use and dyslipidemia. There was an approximately 40% greater odds of a diagnosis of DED in patients on statin regimens of all intensities, 60% greater odds for patients with cholesterol >200, and 40–50% greater odds with TG >150, HDL <40, or LDL >130. While previous studies of smaller, select populations support an association between dyslipidemia and MGD or DED, none have specifically evaluated the association of statin use/intensity and a clinical diagnosis DED.

Previously reported case-control studies support the association of elevated total cholesterol,^9–11 triglycerides^9–11 and LDL^10,11 with MGD or DED. As well, one study found an association of elevated LDL/triglycerides with more severe MGD.¹² However, in these studies, individuals on lipid-lowering therapy were specifically excluded, limiting this potential confounder but disallowing its evaluation. Another population-based study of 5,627 Korean adults found an association of elevated total cholesterol and LDL with DED in women only.¹⁹ Conflicting results have been found in regard to HDL levels and MGD/DED.^9–11,19,20 In contrast, a population-based study of 3,280 Malay persons in Singapore showed an inverse relationship between LDL and MGD and no significant association with other lipid parameters.²⁰ A recent analysis of data from subjects in the Blue Mountains Eye Study found no association between hypercholesterolemia and DED; however, oral statin use was associated with moderate to severe DED symptoms based on a questionnaire.⁸

The exact mechanism(s) underlying the relationship between statin use and/or dyslipidemia and MGD/DED is unclear. It is well established that meibomian glands secrete meibum into the tear film, slowing its rate of evaporation.^21–23 Studies have shown that the meibum of individuals with MGD has an increased concentration of cholesterol esters, likely resulting in a higher melting point, increased viscosity, and a propensity toward meibomian gland obstruction.^24,25 While it is plausible that systemic lipid abnormalities may be associated with abnormalities of tear film lipids, no studies have specifically examined this relationship in detail. One study of two subjects failed to identify an association between plasma and tear cholesterol.²⁶ An alternative hypothetical mechanism relates to the proinflammatory characteristics of LDL^27,28 as inflammation is known to contribute to the pathogenesis of MGD and DED.⁶

Statins exert their effects via inhibition of HMG-CoA reductase, an enzyme which catalyzes the rate-limiting step for the synthesis of sterols and isoprenoids in human meibomian gland epithelial cells.^16,29,30 Thus, it has been postulated that statins may disrupt essential cholesterol synthesis in the meibomian glands.¹⁶ In our study, the risk of DED was similar for low, moderate, and high-intensity statin regimens, suggesting that any effect of statin use may not be dose dependent. Alternatively, it is also possible that statin use is only a surrogate marker for treated dyslipidemia and the association of statin use with DED could be spurious. As nearly all patients with dyslipidemia were treated with statins in our cohort, we were unable to separate the effects of statin use and dyslipidemia on the risk of DED in this study. Further prospective analyses evaluating the effects of statin use on meibomian gland secretions would provide insights into any influence on meibum quantity and composition in the tear film. Interestingly, topical statin use has been demonstrated to be beneficial in treating dry eye associated with blepharitis.^31,32 The mechanism of this response has not been established although statins are known to also exert anti-inflammatory effects which could be involved.³¹

To reduce the number of potentially confounding factors, we excluded patients with conditions or medications that are well known to be associated with DED. We excluded patients with antihistamine, diuretic, and tricyclic antidepressant use in our analyses because these classes of medication may be associated with the development or worsening of DED.^33,34 Refractive surgery and cataract surgery can also precipitate or worsen DED.^35–37 Additionally, autoimmune diseases such as lupus, rheumatoid arthritis, and Sjögren syndrome are strongly associated with DED.^38–40 However, because many other systemic conditions and treatments may have mild potential associations with DED or lipid profiles, there could be other confounding factors that were not controlled for in this study. For example, we were unable to assess for over the counter supplement use, such as that of omega-3 fatty acids or niacin, which have known effects on serum lipid profiles.

Our study possesses several other limitations inherent to retrospective studies, including an inability to establish whether temporal or duration-based associations exist between statin use or dyslipidemia and DED. As well, we were unable to stratify by cause of DED. As the code for MGD (ICD10, H02.88) was not present in ICD-9, its evaluation could not be assessed prior to October 2018 due to not having its own code until that point.⁴¹ As such, DED of all potential causes were included. Data from the CDWH database were only available in aggregate, which limited our ability to examine trends in statin use and DED rates based on individualized variables, such as the presence and severity of MGD, tear film breakup time, or Schirmer testing results. The CDWH database does not allow for determination of the exact timing of diagnoses or laboratory values. In addition, the use of diagnostic codes to select patients for our analysis could have introduced classification bias into our study because no specific criteria for the diagnosis of DED could be used. However, previous literature suggests a high degree of consistency between ICD-9 and ICD-10 codes and findings present within medical records.^42,43

Despite these limitations, we report the largest retrospective analysis to date examining DED as associated with statin use and dyslipidemia. Our study supports an association between statin use, dyslipidemia, and DED, particularly amongst elderly patients. Further research is needed to determine any separate effects of dyslipidemia and statin use on DED, to elucidate pathogenic mechanisms involved, and to identify any potential therapeutic targets.