Association of Ocular Surface Immune Cells With Dry Eye Signs and Symptoms in the Dry Eye Assessment and Management (DREAM) Study

Abstract

Purpose: Dry eye disease (DED) is a multifactorial, heterogeneous disease of the ocular surface with one etiology being ocular surface inflammation. Studies using animal models demonstrate the role of ocular surface immune cells in the inflammatory pathway leading to DED, but few have evaluated humans. This study described the white blood cell population from the ocular surface of patients with DED and assessed its association with DED signs and symptoms in participants of the Dry Eye Assessment and Management (DREAM) study.

Methods: Participants were assessed for symptoms using the Ocular Surface Disease Index, signs via corneal staining, conjunctival staining, tear break-up time, and Schirmer test, and Sjögren's syndrome (SS) based on the 2012 American College of Rheumatology classification criteria. Impression cytology of conjunctival cells from each eye was evaluated using flow cytometry: T cells, helper T cells (Th), regulatory T cells (Tregs), cytotoxic T cells, and dendritic cells.

Results: We assessed 1049 eyes from 527 participants. White blood cell subtype percentages varied widely across participants. Significant positive associations were found for Th and conjunctival staining (mean score of 2.8 for 0% Th and 3.1 for >4.0% Th; P = 0.007), and corneal staining (mean score of 3.5 for 0% Th and 4.3 for >4.0% Th; P = 0.01). SS was associated with higher percent of Tregs (median 0.1 vs. 0.0; P = 0.01).

Conclusions: Th were associated with more severe conjunctival and corneal staining, possibly indicating their role in inflammation leading to damage of the ocular surface. There is no consistent conclusion about Tregs in SS, but these results support that Tregs are elevated in SS.

Dry eye disease (DED) is a common eye condition affecting approximately 5% to 50% of the world's population, and up to 75% of adults over the age of 40.¹ Approximately 16.4 million people in the United States have been diagnosed with DED.² Patients with DED experience a decreased quality of life and increased interference with their activities of daily living.^3,4 

As a heterogeneous disease with several likely pathologic mechanisms,^5,6 DED was defined by the International Dry Eye Workshop in 2017 as: “a multifactorial disease of the tears and ocular surface, that results in symptoms of discomfort, visual disturbance, and tear film instability with potential damage to the ocular surface. It is accompanied by increased osmolarity of the tear film and inflammation of the ocular surface.”⁷ Immune-mediated inflammation has been recognized as playing a prominent role in the pathogenesis of DED.^8–13 Studies using animal DED models have been crucial in identifying and understanding the role of the immune system on the ocular surface in the pathogenesis and progression of DED.^12,13 

The Dry Eye Assessment and Management (DREAM) Study (Clinicaltrials.gov identifier NCT02128763)¹⁴ is a multicenter placebo-controlled, double-masked, randomized clinical trial that enrolled a large population of well-characterized patients with moderate-to-severe DED. The study included many exploratory endpoints including impression cytology (IC) to study the cells on the ocular surface. The purpose of this paper is to evaluate the distribution of the types of white blood cells (WBCs) on the ocular surface and their associations with the severity of DED as defined by various signs and symptoms, and other demographic characteristics including presence of Sjögren's syndrome (SS). 

Methods

The DREAM study was a multicenter, double-masked, placebo-controlled randomized clinical trial testing omega-3 supplementation as a treatment for DED (ClinicalTrials.gov number NCT02128763). Participants were recruited from 27 private and academic ophthalmology and optometry practices across the United States. The study used minimally restrictive inclusion and exclusion criteria to include a wide variety of typical patients with DED seeking treatment of their symptoms. Subjects were enrolled if they were 18 years or older, had moderate-severe DED symptoms as defined by an Ocular Surface Disease Index (Allergan, Madison, NJ)¹⁵ score of 21 or greater, and repeatable signs of DED on clinical examination,¹⁶ with two out of four of the following signs in one eye on examination: tear film breakup time of 7 seconds or less; corneal fluorescein staining score of 4 or more on a scale of 0 to 15; conjunctival lissamine green staining score of 1 or more on a scale of 0 to 6; and Schirmer test with anesthesia measurement of 1 or more to 7 or less mm/5 min. The same qualifying signs had to be present in the same eye at both the screening and eligibility confirmation visits. The tear film breakup time was measured 30 seconds after instillation of 5 µL of 2% fluorescein-containing solution. Time between the last blink and appearance of the first discontinuity in the tear film was noted and repeated twice. Corneal fluorescein staining was scored using the National Eye Institute/industry-recommended scoring guidelines,¹⁷ and conjunctival staining was scored using a modified version of the National Eye Institute/industry-recommended guidelines where the entire temporal and nasal sections of each eye were graded on a scale from 0 to 3 (from 0 [no staining] to 3 [severe staining]).¹⁶ Schirmer's test was performed by inserting the standardized strip at the junction of the lower lid for 5 minutes after instillation of a topical anesthetic. 

Subjects were classified as having SS using the 2012 American College of Rheumatology criteria.¹⁸ Those criteria are (1) positive for traditional SS antibodies (SSA or SSB or [rheumatoid factor and antinuclear antibody ≥ 1:320]), (2) ocular staining score of 3 or more, and (3) labial salivary gland biopsy exhibiting focal lymphocytic sialadenitis with a focus score of 1 focus/4 mm². Labial salivary gland biopsy was not done in the DREAM study. We considered a sum of 3 from the corneal and conjunctival staining scores to be equal to an ocular staining score of 3, thus fulfilling that criterion.^18,19 

The study protocol was approved by the institutional review board of each respective clinical center or a centralized institutional review board (University of Pennsylvania). The DREAM Study was in compliance with the Health Insurance Portability and Accountability Act and adhered to the principles of the Declaration of Helsinki. Informed consent was obtained from all participants. 

Sample Collection and Processing

Ocular surface cell samples were collected from both eyes of participants via IC using an established standard operating procedure.^20,21 IC samples were not collected until at least 20 minutes after the final clinical test. Eyes were anesthetized with 0.5% proparacaine hydrochloride ophthalmic solution. One-half of a sterile 0.20-µm, 13-mm polyether sulfone filter membrane (Supor 100, 13-mm diameter, pore 2 µm; Gelman, Pall Sciences, Fort Washington, NY) was placed onto the superotemporal area and then placed into a tube containing 2 mL of cold sterile PBS with 0.01% paraformaldehyde. This was repeated with the second half of the filter paper for the inferotemporal area of the bulbar conjunctiva of the same eye. Both pieces of filter paper from the same eye were placed into the same tube. This procedure was then repeated for the other eye with the filter papers placed in a new and separate tube. Each eye had a separate tube to allow for comparison with that specific eye's clinical examination scoring of dry eye signs. After completion of the last sample, a drop of prophylactic antibiotic (0.5% moxifloxacin hydrochloride) was administered to the patient's eyes. The tubes were stored and shipped at 4°C and processed within 30 days of collection. Cell processing was completed at the Ocular Biomarker Laboratory located at the Icahn School of Medicine at Mount Sinai, New York, New York. 

Ocular surface cells were harvested from filter paper by shaking tubes at 400 rpm for 20 minutes at 4°C, and then vortexed for 20 seconds. An additional 2-mL sterile PBS with 0.5% BSA (buffer) was added to each tube. Filter papers were then removed and discarded. Next, the tubes were centrifuged at 1000 rpm for 10 minutes. Supernatant was aspirated leaving behind 100 µL of liquid. Next, 50 µL of antibody cocktail containing phycoerythrin-labeled anti-EpCAM (epithelial cell marker) antibody (BD Biosciences, Franklin Lakes, NJ, USA), pacific orange-labeled anti-CD45 (panwhite blood cell marker) antibody (Invitrogen, Waltham, MA, USA), Texas Red-labeled anti-CD8 antibody (BD Biosciences), PE-Cy5-labeled anti-CXCR3 antibody (BD Biosciences), PE-Cy7-labeled anti-CD127 antibody (BD Biosciences), pacific blue-labeled anti-CD3 antibody (BD Biosciences), brilliant violet 650-labeled anti-CD25 antibody (BD Biosciences), FITC-labeled anti-CCR6 antibody (Biolegend, San Diego, CA, USA), AlexaFluor-labeled anti-CD4 antibody (Biolegend), APC-Cy7-labeled anti-CD11c antibody (Biolegend), brilliant violet 605-labeled anti-CCR4 antibody (Biolegend), and additional antibodies not discussed in this paper. CD45⁺/CD11c⁺/CD3^– represented dendritic cells (DCs),²² CD45⁺/CD3⁺ were T cells,²³ CD45⁺/CD3⁺/CD8⁺ were cytotoxic T cells (Ttox),²⁴ CD45⁺/CD3⁺/CD4⁺ were helper T cell (Th),²⁵ CD45⁺/CD3⁺/CD4⁺/CD25⁺(high)/CD127^– were regulatory T cells (Treg),^26,27 CD45⁺/CD3⁺/CD4⁺/CXCR3⁺/CCR4⁻/ CCR6⁻ were Th1,²⁸ and CD45⁺/CD3⁺/CD4⁺/CCR4⁺/CCR6⁺/CXCR3⁻ were Th17.²⁹ 

Antibody concentrations used in the cocktail for a particular antibody lot were based on prior titrations performed with peripheral blood mononuclear cells to determine optimal signal-to-noise ratio. This strategy minimized the noise from nonspecific binding of the antibodies to low-affinity targets and achieved the best signal with the lowest background. Titrations were performed for each new antibody lot. The tubes were then gently vortexed and incubated at room temperature for 20 minutes. The cells were washed with a 2-mL buffer by centrifuging at 1000 rpm for 10 minutes and resuspended in 170 µL of buffer for flow cytometry. 

Flow Cytometry

Samples were analyzed using a BD LSRFortessa cell analyzer. Each sample was analyzed using a common configuration established at the beginning of the study. Eight peaked Rainbow Calibration Particles (BD Biosciences) were used to track cytometer performance and compensation was calculated using single antibody-stained beads (BD CompBead) to eliminate spectral overlap associated with simultaneous usage of multiple fluorochrome-labeled antibodies. Fluorescence Minus One controls were also performed to ensure antibody binding specificity as well as to demarcate gating areas for positive cell populations. This required the preparation of multiple cocktails, each one missing one specific antibody from the panel and staining pooled IC samples (from one individual) separately with each of these minus-one antibody cocktails. Flow cytometer outputs were analyzed using the FCS Express 6 data analysis software (De Novo Software, Pasadena, CA, USA). Analysis was performed using a hierarchical gating strategy developed at the Ocular Biomarker Laboratory (Fig. 1). The two personnel conducting the analysis periodically analyzed and compared data with a common set of samples to reduce user-based variations. 

Figure 1.

 

Flow cytometry gating strategy. Flow cytometer analysis. Dot plots of representative sample from the study. Total analyzable cells (B) were gated out from a scatter plot of forward scatter-width (FCS-W) vs. side scatter-area (SSC-A). (A) Sequential gating of CD45⁺ expressing cells (C) from the total population yielded total WBC from the sample. CD3⁺ and CD3^– (D) were gated from the total WBCs. CD3^– cells were gated for CD11c (E) to give total DCs. CD3⁺ cells were gated for CD8⁺ CD4^– to give Ttox, and CD4⁺ CD8^– for Th (F). CD4⁺ cells were gated for CD25⁺ and CD127^– for Tregs (G). CD4⁺ were further gated to determine Th subtypes (H). CD4⁺ CXCR3^– CCR4⁺ CCR6⁺ were gated for Th17 (I), and CD4⁺ CXCR3⁺ CCR4^– CCR6^– were gated for Th1 cells (J).

Statistical Analysis

We summarized continuous measures using mean, SD, median and interquartile range, and summarized categorical measures using frequency count and percentage. For evaluating association between each immune cells with DED symptoms and signs, we categorized eyes into groups using immune cell quartiles (owing to the skewed distribution of immune cell measures), and compared their mean scores of symptoms and signs using generalized linear regression model. In all these analyses, the intereye correlation was accounted for using generalized estimating equations, and linear trend P value was used for statistical significance. 

For evaluating the association of immune cells with age, sex, and SS, the average of two eyes was taken for each immune cell measure and its comparison between patient groups was made using Wilcoxon rank sum test. All statistical comparisons were performed in SAS v9.4 (SAS Institute, Inc., Cary, NC, USA) and a two-sided P value of 0.05 or less was considered to be statistically significant. 

Results

Among 535 DREAM participants, 527 subjects had a total of 1054 IC samples collected separately from the right and left eye. Five samples from 5 subjects were excluded for having fewer than 1000 total fixed cells (epithelial and WBC) collected from the ocular surface of one eye, below the needed limit per our standard operating procedures (SOP).²⁰ Therefore, 1049 IC samples were used for this analysis. 

At baseline, the mean age of the subjects was 58.2 ± 13.1 years with 428 females (81.2%). There were 51 subjects (9.7%) with SS and 69 (13.1%) with a reported history of other autoimmune diseases (Table 1). 

Table 1.

 

Baseline Characteristics of Participants Analyzed in the Study

 

The mean value was 42.0 ± 15.5 for the Ocular Surface Disease Index score. The mean value of all eyes was 2.9 ± 1.5 for conjunctival staining score, 3.8 ± 3.0 for corneal staining score, 3.1 ± 1.7 seconds for tear film breakup time, and 9.6 ± 1.7 mm for the Schirmer test (Table 1). 

The median (first quartile, third quartile) of total gated cells per IC sample was 14,274 (9200, 20,284), with a range of 1061 to 88,097 cells. The median (first quartile, third quartile) of total WBCs (CD45⁺) per sample was 131 (73, 245) with a range of 1 to 2791 , which is 1.1% (0.6%, 1.8%) of total cells gated (Table 2). 

Table 2.

 

Descriptive Statistics of Number of Cells Per Eye at Baseline (N = 1049 Eyes)

 

Of the total WBCs from each sample, the median percentage (first quartile, third quartile) of subgroups was DCs 3.0% (0.4%, 11.1%), T cells 76.7% (55.4%, 89.2%), Ttox 9.6% (4.2%, 16.0%), Th 0.9% (0.0%, 4.0%), and Treg 0.0% (0.0%, 0.6%). There were 441 eyes (42.0%) that measured 0% Th and 730 eyes (69.6%) that measured 0% Treg. There were 946 eyes (90.2%) that had no Th1, and the maximum percentage of Th1 in a sample was 5.6%. There were 969 eyes (92.4%) that had no Th17, and the maximum percentage of Th17 in a sample was 8.2% (Table 3). 

Table 3.

 

Descriptive Statistics of All the Immune Cells % at Baseline (N = 1049 Eyes)

 

Th showed a significant association with corneal staining (P = 0.01) (Table 4, Fig. 2) and conjunctival staining (P = 0.007) (Table 4, Fig. 3). Total WBCs, DCs, Ttox, and Tregs showed no significant association with any DED symptoms and signs (Table 4). WBCs were not significantly associated with demographic characteristics such as age and sex. However, SS was significantly associated with association with higher % of Treg (P = 0.01) (Table 5, Fig. 4). 

Table 4.

 

Associations of Immune Cells and DED Symptom and Signs at Baseline

 

Figure 2.

 

Association between corneal staining score and Th. The 75th and 25th percentiles make up the upper and lower boundaries of the boxes. Within the box, the diamond represents the mean and the horizontal line in the box represents the median. The vertical lines issuing from the box extend to the highest value within 1.5× the interquartile range of the 75th percentile and the lowest value within 1.5× the interquartile range of the 25th percentile. Each dot outside the upper and lower fence corresponds with an outlier value.

Figure 3.

 

Association between conjunctiva staining score and Th. The 75th and 25th percentiles make up the upper and lower boundaries of the boxes. Within the box, the diamond represents the mean and the horizontal line in the box represents the median (the median is the same as the first quartile for the 0% group). The vertical lines issuing from the box extend to the highest value within 1.5× the interquartile range of the 75th percentile and the lowest value within 1.5× the interquartile range of the 25th percentile. Each dot outside the upper and lower fence corresponds to an outlier value.

Table 5.

 

Association of Immune Cells With Age, Sex and SS

 

Figure 4.

 

Association between Sjögren's syndrome and Treg cells. The 75th and 25th percentiles make up the upper and lower boundaries of the boxes. Within the box, the diamond represents the mean and the horizontal line in the box represents the median. The vertical lines issuing from the box extend to the highest value within 1.5× the interquartile range of the 75th percentile and the lowest value within 1.5× the interquartile range of the 25th percentile. Each dot outside the upper and lower fence corresponds with an outlier value.

Discussion

DREAM is the first study of a large, well-characterized DED population to describe the WBC population of the ocular surface and assess their association with demographics, signs, and symptoms of DED. The DREAM study enrolled a population of real-world subjects with moderate to severe DED who had both symptoms and signs of ocular surface disease at baseline, despite ongoing DED treatment. Evaluation of DREAM subjects thus is likely to be generalizable to DED seen in routine eyecare in the US.¹⁶ As an exploratory endpoint, cells of the ocular surface were sampled using IC and analyzed by flow cytometry to describe the WBC population and its association with signs and symptoms of DED.^16,20,30 

Analyzable samples with more than 1000 cells were obtained from all eyes except for 5 eyes (<0.5%) thanks to extensive training provided to clinical site personnel and strict adherence to study SOP.¹⁶ Samples with fewer than 10,000 cells (30% of our samples) were not excluded, as done in other studies.^31–34 Previous studies in our laboratory support the inclusion of low cell number samples.³⁵ This emphasizes the importance of SOPs for sample collection and processing in multicenter clinical trials to maximize number of analyzable cells.²⁰ 

Past studies, primarily in animal models of DED, suggest that inflammation of the ocular surface is a core mechanism behind the pathogenesis of DED.^8,36–41 Immune cells such as DCs, Th, Th17, Th1, Ttox, and Treg have all been implicated in the inflammatory process.^{12,13,41–60} Few studies have evaluated the role of ocular surface WBCs in human subjects. Most studies done in humans only examined one cell type, a specific disease population (i.e., only SS) and/or used a small patient population.^59,61–65 However, these studies generally do support the role of immune cells, such as DCs, Th, Ttox, and Treg in human DED subjects as seen in animal models. This study focused on inflammatory cells, WBCs, and subgroups to better understand their relationship with the severity of signs and symptoms in moderate to severe, well-characterized DED subjects. 

No other study using human subjects has presented a detailed description of the WBCs found on the ocular surface of patients with DED. Our results showed a wide variety of WBC percentages across our population (Table 3). DCs had a range of 0.0% to 92.0%; Ttox 0.0% to 54.3%; Th 0.0% to 42.0%; and Tregs 0.0% to 22.6%. Although WBCs and subgroups that are thought to be important in DED inflammation were analyzed, we found only rare Th17 and Th1, and, therefore, could not analyze for their association with signs and symptoms of DED. The heterogenous nature of DED and the ability to only analyze surface cells by IC may explain the wide range of WBC percentages seen. However, even in subgroup analyses of known systemic inflammatory disease, such as SS, large number of WBCs were not observed nor a high number of Th17, which have been reported to be associated with autoimmune diseases.¹² 

CD4⁺ Th are considered to have an important role in DED and disrupting the corneal and conjunctival surface.^12,66 Th have previously been demonstrated in both animals and humans to be the predominant WBC infiltrating the ocular surface and playing a pathogenic role in all types of DED owing to their production of proinflammatory cytokines.^41–46,66 Corneal barrier disruption in DED leads to increased infiltration of Th into the conjunctiva and cornea leading to exacerbation of the disease.⁶⁶ Our study supports this finding, because Th were found to have a significant positive association with conjunctival staining score (P = 0.007) and corneal staining score (P = 0.01) (Table 4). The DREAM analysis showed Th to be associated with ocular surface changes in DED and, therefore, supports their role in the pathogenesis of DED. 

However, we were unable to detect Th subtypes, Th1 and Th17 in significant amounts, and thus unable to associate with signs and symptoms of DED. Recently, the Th subtypes of Th17 and Th1 have been shown to play a major pathogenic role in animal models of DED through their production of proinflammatory cytokines, such as IL-17, and IFN-gamma.^{12,47–50,66} However, human studies for tear cytokines, not cells, have not yet shown association of high levels of IL-17 and/or IFN-gamma with the severity of DED.⁶⁷ Another report from the DREAM study focusing on tear cytokines found a weak negative correlation between IL-17A and IFN-gamma with the severity of DED clinical signs, and no correlation with symptoms.⁶⁷ Our study of cells on the ocular surface did not detect Th17 or Th1 in more than 90% of the samples collected from patients suffering from moderate to severe DED and was, therefore, unable to assess their association with the signs and symptoms of DED. 

The role of DCs in humans with DED remains unclear. DCs were previously found to facilitate the inflammatory response as an antigen-presenting cell and increased levels have been demonstrated with increased severity of DED in animal models.^12,13 Several studies in humans have shown that DCs are present on the ocular surface in greater quantities than control subjects.^62,64 Pflugfelder et al.⁶² showed an approximate range of 10% to 50% of DCs collected using IC from the conjunctiva in humans with SS-associated DED, much higher than their control group. Our study also found a wide range of DC percentage present on the ocular surface of DED subjects. However, we did not find any association with DED signs, symptoms, or SS. 

The role of CD8⁺ Ttox in humans with DED also remains unclear. Ttox traditionally play an important role in viral infections, chronic inflammation, and autoimmune disease.^60,68,69 They have also been implicated in the pathogenesis of DED, especially SS, and shown to be increased with increased severity.^59,60 However, we did not detect a significant association between Ttox and clinical findings of DED or SS. 

Tregs were found to be higher in subjects with SS, but had no association with any of the signs or symptoms of DED. Tregs typically serve as an immunoregulatory cell that suppresses excessive inflammation.^12,51,52 In DED models, Tregs are unable to suppress IL-17 production by Th, despite a normal number of Tregs.^53,54 Studies on ocular surface Tregs in mouse models of DED suggest that Tregs are upregulated in response to inflammation in an attempt to reduce the inflammatory response and complete removal of Tregs from the ocular surface contribute to the development of more severe disease owing to an unchecked inflammatory response.^41,55–58 Tregs on the ocular surface have not been studied in humans with DED. Tregs on the ocular surface were rare in our study population with almost 70% of eyes having no Tregs detected (Table 3). Although we found that Tregs had no association with any clinical signs or symptoms of DED, Tregs were present in significantly greater numbers in patients with SS as compared with those without SS (P = 0.01) (Table 5). However, this study was only able to collect Tregs on the ocular surface of the conjunctiva and there may have been different responses in other areas of the eye such as the cornea, lacrimal glands, meibomian glands, and other deeper tissues. More studies are needed to better understand the role of Tregs in inflammatory conditions such as DED and SS. 

SS is an autoimmune disease typically associated with DED.⁷⁰ Studies focused on the role of WBC in SS found that T cells play an important role and become activated in response to environmental triggers. Th and Ttox were found to be decreased in the serum of patients with SS as compared with normal controls.⁷¹ However, increased activated T cells were found in affected glands of SS, with most of the subtypes being Th, with Th17 predominating.⁵⁸ Treg numbers and function are still unclear in SS.⁷² There are numerous studies, including a review article, that report Tregs being higher, similar, and lower in the serum of patients with SS and mouse models as compared to healthy controls.^72–94 

DREAM data on other inflammatory markers have not shown a clear-cut picture of the type and/or extent of inflammation in moderate to severe DED. A wide distribution of HLA-DR% in conjunctival cells likely reflects the heterogeneity of mechanisms leading to DED.³⁵ Inflammatory cytokines in tear samples showed no definitive expression patterns and only some (IL-17A, IL-10, IFN-gamma) were weakly associated with signs of DED.⁶⁷ 

A strength of the DREAM study is the large number of well-characterized DED subjects that were included from a large geographical distribution in the United States. All subjects had symptoms and signs of DED at a screening and then baseline visits. The study was carefully conducted following protocols specifically designed to maximize reproducibility of data across all clinical centers. Standardization of IC collection using the same materials and method across all centers and the methodology for the assessment of clinical characteristics were well-established and widely accepted.¹⁶ SOPs were established to standardize processing and analysis of IC samples to minimize variability.²⁰ 

There are some limitations with this study. The DREAM study did not include subjects with mild DED or normal, non-DED subjects to be compared with the DREAM subjects with moderate to severe DED. Factors that could contribute to variable results in the DREAM subjects include variability in IC, such as the number of cells collected, storage, processing, and flow cytometry analysis via gating. Additionally, flow cytometry of IC samples is not an absolute metric. The collection of cells using a minimally invasive technique such as IC will have variability and the number of cells collected is limited because only cells from the conjunctival surface are collected and may impact the ability to correlate with clinical signs and symptoms. This factor was evident in the relatively low number of Tregs, Th1, and Th17 that were detected. Additionally, the low number of cells does not allow for duplicate testing of the same sample. Flow cytometry also has variability owing to its subjective nature because each grader may differ when analyzing via gating, although laboratory procedures tried to decrease variability by having duplicate analysis on random samples throughout the analysis timespan.^35,95 In general, there is a lack of standardized methods for flow cytometry and analysis. This point has been acknowledged as an issue where the immunophenotyping is necessary for diagnosis, detection, or treatment of disease.^96,97 However, efforts are underway to standardize sampling and flow cytometry and gating.^98,99 

In conclusion, this study provided the first comprehensive description of WBCs on the ocular surface in a large, well-defined cohort of patients with moderate to severe DED. DED is a multifactorial disease with inflammation and WBCs playing an important role.^5,7,8,36 There was a wide variety of WBC percentages found between our subjects and few significant associations with clinical signs and symptoms were noted. Th showed a significant positive association with conjunctival and corneal staining score, possibly indicating their role in inflammation that leads to damage of the ocular surface. Tregs were significantly positively associated with SS; however, there is still no consensus on the expected role and levels of Tregs in inflammatory conditions such as SS.^58,72,94 Future research is needed using well-characterized subjects to confirm the role of ocular surface WBCs in the pathogenesis of DED and correlation with severity of DED as measured by symptoms and/or signs. 

Acknowledgments

The authors thank Yi Wei and Neeta Roy for their work and contributions to the DREAM Study as co-directors of the Biomarker Laboratory at the Icahn School of Medicine at Mount Sinai. 

Supported by grants U10EY022879, U10EY022881, and R21EY031338 from the National Eye Institute and supplemental funding from the Office of Dietary Supplements, National Institutes of Health. 

Disclosure: E.J. Kuklinski, None; Y. Yu, None; G.-S. Ying, None; P.A. Asbell, None