Association of Systemic Markers of Inflammation with Signs and Symptoms of Dry Eye Disease and Sjogren’s Syndrome in the Dry Eye Assessment and Management (DREAM©) Study

Abstract Purpose To evaluate the possible role of systemic inflammation in dry eye disease (DED) via systemic inflammatory marker associations with DED signs and symptoms, and an analysis of a subgroup with Sjogren’s Syndrome (SS). Methods Participant serums were analyzed using line immunoassays (LIAs) for the presence of antibodies against 34 systemic inflammatory markers. Using the 2012 American College of Rheumatology definition, the 481 participants were categorized into group 1 (SS; n = 52), group 2 (autoimmune disease not including SS; n = 66), or group 3 (control, i.e. no autoimmune disease; n = 363). Results 3 markers were positive in ≥10% of participants: Ro52 (19.3%), Scl-70 (15.0%), CN-1A (14.2%). 2 markers were positively associated with symptoms: PM-Scl100 (p = 0.02), Sm (p = 0.009). 5 markers were positively associated with signs: U2SnRNP A’, Ro52, La, DNA, Ro60. SS participants showed significantly higher positivity for 4 markers compared to participants with no autoimmune disease: PL-7 (p = 0.02), Ro52 (p < 0.0001), La (p < 0.0001), Ro60 (p < 0.0001). SS participants showed significantly higher positivity for 3 markers compared to participants with another autoimmune disease: Ro52 (p < 0.0001), La (p = 0.002), Ro60 (p < 0.0001). Conclusions This study did not show evidence of significant systemic inflammation in participants with moderate-to-severe DED, based on the markers tested. PM-Scl100 and Sm may be associated with more severe DED symptoms. U2SnRNP A’, Ro52, La, DNA, and Ro60 may be associated with more severe ocular surface disease. Ro52 and PL-7 may be diagnostic markers for SS. Future research evaluating these relationships and their clinical significance is needed.


Introduction
Dry eye disease (DED) is a chronic ocular condition that results in discomfort, fatigue, and visual disturbances. 1,2One of the most common reasons for visiting eye care specialists, 3 symptomatic dry eye disease has a prevalence of 14% in the United States and is seen at higher rates in women and older age groups. 4While DED is a heterogenous disease and its exact pathogenesis still unknown, inflammation is believed to play a key role in its development and progression. 5It is unclear if this inflammation is limited to the ocular surface or includes systemic involvement, as seen in some autoimmune diseases, 6 such as Rheumatoid Arthritis and Sjogren's Syndrome.
The Dry Eye Assessment and Management (DREAM) Study was a randomized double-masked controlled trial evaluating the efficacy and safety of oral omega-3 fatty acid supplementation in the treatment of DED. 7,8The data from the DREAM Study present an opportunity to study the systemic inflammation in DED.Using line immunoassays (LIAs) with the ability to detect several antibodies at once from a single sample, this exploratory study analyzed the serum taken from DREAM participants at the baseline examination for the presence of antibodies against 34 systemic inflammatory markers.These markers were then evaluated for their association with the clinical symptoms and signs of DED collected at the same visit.As part of a subanalysis, the markers of DREAM participants with Sjogren's syndrome (SS), a known systemic autoimmune disease, were also compared to those without SS.This analysis furthers previous studies on the association between SS and novel markers 9 by exploring a new set of systemic inflammatory markers and the potential to apply them clinically as biomarkers for DED.

Methods
The current study is a secondary analysis of the data obtained from the DREAM study.The DREAM study was conducted with approval from the Institutional Review Boards (IRB)/Ethics Committee and in accordance with the tenets of the Health Insurance Portability and Accountability Act and the Declaration of Helsinki. 7,8A description of the study's design and methods can be found in previous publications. 7,8Participants provided informed consent for this study.

Participants
Between October 2014 and July 2016, the DREAM study used screening and eligibility visits at 27 clinics in the United States to enroll a total of 535 adult participants with moderate to severe DED in the study.Inclusion and exclusion criteria used for this study are detailed in prior DREAM publications. 7Serum from 481 randomized participants was available for analysis from the baseline visit.Of the 535 randomized participants, 54 participants were not included because 41 did not have serum collected at baseline and the 13 other samples were insufficient for analysis.All serum samples were analyzed in the same laboratory, Immco Diagnostics.Based on the post-hoc power calculation for this secondary analysis of existing data from the DREAM study, this sample size of 481 participants with data on systemic markers provided at least 90% power to detect the effect size of 0.37 to 0.70 for the systemic markers with a positive rate of 20% to 5% at the type I error rate of 0.05 based on the two-sample t-test for comparing mean score of symptoms and signs between participants with positive vs. negative systemic markers.

Determination of Sjogren syndrome classification
As part of the sub analysis of SS participants performed in this study, SS status was determined based on the 2012 American College of Rheumatology (ACR) criteria, in accordance with previous DREAM publications. 9Based on these guidelines, participants were divided into three groups: group 1 (participants who met the ACR criteria for SS and were considered the "SS group"), group 2 (participants who did not meet the ACR criteria for SS, but provided a self-report of a history of other autoimmune diseases [rheumatoid arthritis, systemic lupus erythematosus, fibromyalgia, calcinosis, Raynaud phenomenon, esophageal dysmotility, sclerodactyly, telangiectasia, antiphospholipid syndrome, Raynaud disease, scleroderma, graft-versus-host disease, and sarcoidosis]), 9 and group 3 (participants who did not meet the ACR criteria for SS and did not report a history of another autoimmune disease). 9ollowing classification criteria given in past DREAM publications, 9 52 participants (10.8%) met the 2012 ACR criteria to be included in the SS group (group 1), 66 participants (13.7%) reported another autoimmune disease and were included in the "other autoimmune disease" group (group 2), and 363 participants (75.5%) reported no history of autoimmune diseases and were included in the control group, i.e. no autoimmune disease (group 3). 9

Antibody assays
Baseline serum samples collected from DREAM participants were sent to the Immco laboratory, where they were frozen and processed per standard Immco procedures.After all samples were received, laboratory researchers masked to the clinical findings of DREAM participants used two line immunoassays (LIAs) for the detection of autoantibodies to 34 systemic inflammatory LIA markers (Immco Diagnostics, Inc., Buffalo, NY, USA).The ImmcoStripe ANA Advanced LIA (Immco Diagnostics Inc., Buffalo, NY, USA) was used for the detection of antibodies against the following antigens: SSA/Ro60, SSA/Ro52, PM-Scl100, PM-Scl75, AMA-M2, DFS70, CENP-B, PCNA, Ku, Mi2, SRP54, Histones, DNA, Sm, Nucleosomes, Jo1, U1SnRNP A, U1SnRNP68, U1SnRNP C, Scl70, SSB/La, Ribo-P.The ImmcoStripe Myositis Advanced LIA (Immco Diagnostics Inc., Buffalo, NY, USA) was used for the detection of antibodies to the following antigens: PM-Scl100, PM-Scl75, Ku, Mi-2, SSA/Ro-52, Jo-1, Fibrillarin/U3 RNP, PL7, PL12, OJ, EJ, U1 SnRNP68, U1 SnRNP A, U1 SnRNP C, U2 SnRNP A' , U2SnRNP B'' , SRP54, SAE1, MDA5/CADM-140, CN-1A, MORC3, TIF1γ or TIF1a, KS.By creating separate antigen lines on a single strip, LIAs measure the reactivities of several antibodies in a single sample. 10In brief, individual LIA strips were pre-blocked for 10 min with the provided diluent in the wells of the reaction tray.Next, 15 mL of participant serum samples diluted in a 1:101 ratio was added individually into the wells of the reaction tray, where they were incubated with rocking for 60 min.Following this, the test strips were washed with wash buffer for 5 min.This wash was repeated 2 more times, after which the strips were incubated with rocking for 30 min in the pre-diluted enzyme conjugate.The wash steps were repeated, after which the strips were incubated for 10 min with 3,3′,5,5′-tetramethylbenzidine (TMB) substrate to produce signals as blue-colored lines. 11In addition to specific antigen lines, each LIA test strip has serum, conjugate, and cut-off control test lines that work as internal controls to confirm test validity.The results were read visually and assessed based on the intensity of color production on each antigen line relative to the cut-off control line. 10If no color signal (band) appeared or the signal generated was of lower color intensity than the cut-off control, that sample was assigned a value of 0 and considered normal (i.e.negative for that marker).If there was a specific signal generated with greater color intensity than the cut-off control, that sample was assigned a value of +1 (low antibody production) or +2 (high antibody production), based on the color intensity of the reaction line, and considered abnormal (i.e.positive for that marker). 10The percentages of samples showing low antibody production (+1) and high antibody production (+2) were combined to generate a single overall value of percentage of samples showing positivity for the given antibody.
The assays to measure antibodies against these proteins are CLIA/NYSDOH approved for participant testing and are routinely used for participant screening at Immco Diagnostics laboratory.

Assessment of dry eye symptoms and signs
Dry eye symptoms and signs measured at the baseline confirmation visit of the DREAM study were used in this analysis.Evaluation of DED symptoms was done using the Ocular Surface Disease Index (OSDI), a 12-item questionnaire that generates scores ranging from 0 (least symptom severity) to 100 (greatest symptom severity).
Evaluation of signs was done using the corneal fluorescein staining score, Schirmer's test results, conjunctival lissamine-green staining score, tear break-up time (TBUT), tear osmolarity, and meibomian gland dysfunction (MGD).The corneal fluorescein staining score was calculated using the National Eye Institute (NEI) scale, which grades 5 areas of the cornea on a scale of 0-3 and adds them up for each eye, 0 to 15 per eye. 12The Schirmer's test was done by placing a paper strip in the lower eyelid and measuring the length of wetting in 5 min. 12The conjunctival lissamine-green staining score was calculated by scoring the nasal and temporal areas of the eye on a scale of 0 to 3 and 0 to 6 for each eye. 12Tear break-up time was found by measuring the time in seconds between blinking and gap development in the tear film. 12The average of 3 separate measurements was used.A higher score indicates less stability of the tear film.The evaluation of MGD was done using the TearScience Meibomian Gland Evaluator at slit lamp; plugging and lid secretion were measured on a scale of 0-3, with lower scores indicating better MGD. 12 These values were then combined to generate a composite sign score of severity for each participant (defined under "Statistical Analysis").

Statistical analysis
We calculated the positive rate for each inflammatory marker, and the frequency for the total number of positive inflammatory markers.For evaluating associations between inflammatory markers and DED symptoms and signs, we compared the OSDI score and composite severity score of signs between participants with positive vs. negative inflammatory markers using analysis of variance.The composite severity score of DED signs was calculated using a method adapted from previous studies, 13 based on TBUT, Schirmer testing, corneal and conjunctival staining, tear osmolarity, and meibomian gland dysfunction.][16] The comparison of marker positivity among the three groups (SS, other autoimmune disease, and no autoimmune disease) was performed using a chi-square test followed by pairwise comparison if overall difference among the three groups was statistically significant.All statistical analyses were performed in SAS v9.4 (SAS Institute Inc., Cary, NC, USA), and two-sided p-value <0.05 was considered statistically significant.We did not correct for multiple comparisons due to the exploratory nature of this analysis.
When considering the number of positive markers per subject, 138 participants (28.7%) did not have any positive markers, 141 participants (29.3%) were positive for one marker, 17.7% of all participants were positive for two markers, and the remainder (24.3%) of participants were positive for more than two markers (Table 2).

Inflammatory marker association with DED symptoms and signs
Table 3 shows the association between DED symptoms (measured by mean OSDI score) and inflammatory marker positivity.For the majority of the markers tested, there was significant association with OSDI score.However, positivity for two markers (PM-Scl100, Sm) was significantly associated with higher (i.e. more severe) symptoms: PM-Scl100 (mean OSDI score 54.2 vs. 44.5, p = 0.02), Sm (mean OSDI score 66.7 vs. 44.7,p = 0.009).
Table 4 shows the association between severity of DED signs (composite severity score) and positivity for the markers.The presence of the following 5 markers was associated with a higher composite severity score (i.e. more severe) of DED signs: U2SnRNP A' , Ro52, La, DNA, Ro60.Furthermore, having a larger total number of positive markers in participant serum was significantly associated with more severe DED signs (Table 5).

Comparison of systemic inflammatory marker positivity among SS participants, non-SS participants with other autoimmune diseases, and Non-SS participants without other autoimmune diseases
Table 6 compares marker positivity rates among the three groups (SS, other autoimmune disease, control, that is, no autoimmune disease).Overall, participants with SS or other autoimmune diseases had significantly higher positivity than controls in the expression of U1SnRNP C (p = 0.002), U1SnRNP A (p = 0.04), PL-7 (p = 0.03), Ro52 (p < 0.0001), La (p < 0.0001), and Ro60 (p < 0.0001).When comparing the SS group and control group, the following were significantly more positive in the SS group: PL-7 (p = 0.02), Ro52 (p < 0.0001), La (p < 0.0001), Ro60 (p < 0.0001).When comparing the other autoimmune disease group and control group, the following were significantly more positive in the other autoimmune disease group: U1SnRNP C (p = 0.0007), U1SnRNP A (p = 0.01).When comparing the SS and other autoimmune disease groups, the following were significantly more positive in the SS group: Ro52 (p < 0.0001), La (p = 0.002), Ro60 (p < 0.0001).SS participants had significantly more positive markers than the control and other autoimmune disease groups (Table 7 in Supplemental Material).Specifically, negative markers were only seen in 2 (3.8%)SS participants, in 14 (21.2%)participants with other autoimmune diseases, and in 122 (33.6%) participants in the There was no statistically significant difference in age among these three groups (p = 0.60).

Discussion
In this exploratory hypothesis-generating study, serums of 481 DREAM participants with moderate-to-severe DED were analyzed for the presence of 34 systemic inflammatory markers.In summary, the following results were found: a) most markers had low positivity in participants' serums, and only 3 markers (Ro52, Scl-70, CN-1A) showed greater than 10% positivity, b) 2 markers (PM-Scl100, Sm) showed a positive association with symptom severity, c) 5 markers (U2SnRNP A' , Ro52, La, DNA, Ro60) showed a positive association with sign severity, d) 4 markers (PL-7, Ro52, La, Ro60,) were significantly higher in the SS group compared to the control group, e) 2 markers (U1SnRNP C, U1SnRNP A) were significantly higher in the other autoimmune disease group compared to the control group, f) 3 markers (Ro52, La, Ro60) were significantly higher in the SS group compared to the group of DED participants with other autoimmune diseases.When considering whether systemic inflammation is involved in the pathogenesis of DED, few systemic inflammatory markers showed positivity in DED participants.Only 3 of the 34 markers showed greater than 10% positivity: Ro52, which is not disease-specific, but found in many autoimmune diseases including myositis, systemic sclerosis (SSc), SLE, and SS; 17 Scl 70, which has been associated with SSc and SSc-related interstitial lung disease (ILD); 18 and CN-1A, which has largely been associated with inclusion body myositis (IBM), other idiopathic inflammatory myopathies, and in some cases, systemic lupus erythematosus (SLE) and SS. 19Moreover, when considering the number of positive markers in DED participants, almost a third of showed no positivity for any markers.Similarly, almost a third showed positivity for only one marker.Among the participants who were positive for more than one marker, the majority were positivity for only 2 or 3 markers.These findings may reflect the heterogenous nature of DED, 5 the possibility that the systemic inflammatory processes represented by these markers are not involved in the pathogenesis of DED, and/or the possibility there are other inflammatory responses implicated in DED for which the 34 markers tested are not representative.
When looking at the symptom severity of DREAM participants, two of the 34 markers, PM-Scl100 and Sm, were significantly associated with more severe symptom reports.PM-Scl100, a component of the exosomal mRNA decay pathway, 20 is a Systemic Sclerosis (SSc)-associated antibody 21 that has also been found in systemic lupus erythmatosus (SLE) and SS. 21Anti-Sm antibodies are highly specific for the diagnosis of SLE. 22While the association between these markers and DED symptom severity is unclear, there may be clinical utility in further studying these markers in the context of DED and whether they define a distinct group of DED participants who may have an underlying autoimmune disease represented by both antibodies.
When looking at DED sign severity, 5 markers (U2SnRNP A' , Ro52, La, DNA, Ro60) were significantly associated with higher severity scores.This association suggests that these markers may indicate greater ocular surface damage in comparison to DED participants who do not have these markers.Two of these markers (Ro60, La) have been associated with SS both in previous studies as well as in ours. 9,17Additionally, Ro52, which may also be associated with SS, 17 was also associated with sign severity.The other two markers, U2SnRNP A' and DNA, have also been associated with various inflammatory pathologies.Antibodies to U2SnRNP A' , a component of the U2 small nuclear ribonucleoprotein involved in removing introns from pre-mRNA, 23 have been associated with MCTD and Scleroderma-Polymyositis Overlap Syndrome. 24,25Anti-DNA antibodies have classically been associated with and used in the diagnosis of SLE. 26 While the associations between these antibodies and DED sign scores are currently unclear, it is possible these antibodies may indicate the presence of another underlying autoimmune disease that may be related to increased ocular surface damage.Future research focusing on the autoimmune diseases associated with these antibodies and the progression of ocular surface damage is needed.
When comparing the markers that were associated with symptom severity and those associated with sign severity, there was no overlap.This finding is consistent with past studies showing discordance between DED sign and symptom reports, 14,15 which may in part be due to variations in etiologies 14 or comorbidities that induce pain or discomfort. 15ur study found that La and Ro60 were significantly higher in the SS group than in the other autoimmune disease and control groups, which is confirmatory of previous research associating these markers with SS and their use for diagnostic purposes. 17,27In addition to confirming these findings, our study also evaluated associations between SS and other immune markers.We found that Ro52, which has been associated with diseases such as myositis, systemic sclerosis (SSc), and SS, 17 was significantly higher in the SS group compared to the control and other autoimmune disease groups.These findings suggest that Ro52 autoantibodies may also be a SS-associated marker and may have clinical utility in distinguishing SS from other autoimmune diseases and from normals.
Autoantibodies to PL-7 were also significantly higher in the SS group in comparison to the control group.This protein, also known as threonyl-tRNA synthetase, is one of the eight aminoacyl-transport ribonucleic acid synthetases 28 and plays a role in protein translation, skeletal muscle regeneration, and myoblast differentiation. 29Autoantibodies have largely been associated with anti-synthetase syndrome, an inflammatory condition that can have variable clinical presentations that may include interstitial lung disease, myositis, inflammatory arthritis, "mechanic's hands", and Raynaud phenomenon. 28,30The higher prevalence of PL-7 in the SS group suggests this marker may be implicated in the diagnosis or pathogenesis of SS.Future research should further examine this association.
When looking at the other autoimmune disease group and control group, U1SnRNP C and U1SnRNP A were significantly higher in the former.These proteins, which are subunits of the larger U1 small nuclear ribonucleoprotein particle, are involved in pre-mRNA processing and have been associated with SLE, Mixed Connective Tissue Disease (MCTD), and scleroderma. 31,32These findings may suggest the presence of these autoimmune diseases in this subgroup of participants.
While the reasons for the various associations between markers, DED signs and symptoms, and SS status may be due to underlying relationships and pathogenesis, another possible explanation is that several of the genes that predispose to autoimmune diseases will predispose to more than one or all autoimmune diseases.While all people have autoantibodies present, they are usually at such low titers that they go undetected.In the case of autoimmune diseases, the antibody amounts are increased, which may explain the elevated markers seen in some participants.
Of note, some markers, such as KS, showed negativity even in participants with autoimmune diseases.While there are various factors that play into the inflammatory response and autoantibody expression, it is important to note that in some autoimmune diseases, the injury is cell mediated and may not involve autoantibodies. 33Furthermore, in some autoimmune diseases, all of the autoantigens involved are not fully understood and thus autoantibodies for which we do not have a test may be missed.This may be a possible explanation for why some of the inflammatory markers tested in our study did not show positive expression in participants with autoimmune diseases.
Limitations in this study include the focus on a select group of inflammatory markers.While little to no signs of systemic inflammation were seen in the majority of participants with moderate-to-severe DED, this was based on the 34 markers tested through the LIAs.It is possible that other systemic inflammatory markers that were not tested may have been elevated.Second, this study only included participants with moderate-to-severe DED.Excluding participants with mild DED may limit our capability to detect differences in this population.Third, we did not measure immunoglobulin levels in participant serum in order to rule out hyper-immunoglobulin (hyper-Ig) syndromes and the potential for false positives resulting from these syndromes.However, if participants did indeed have a hyper-Ig syndrome, we would expect to see elevations in all markers.We did not see this, but rather saw a significant difference only in a small number of markers.Fourth, this study did not collect data on the systemic clinical pictures of the various inflammatory diseases.The systemic inflammatory and autoimmune diseases are solely based on participant self-report of medical history, a limitation in our study due to the risk of inaccurate and/or incomplete reporting.Moreover, while data was collected on the levels of systemic markers associated with different inflammatory pathologies, both the clinical picture and antibody levels would be needed to make a diagnosis, something that should be included in future studies.Finally, in this exploratory hypothesis-generating analysis of inflammatory markers, we did not correct for multiple comparisons from analysis of 34 systemic markers.This is because these comparisons are descriptive and hypothesis-generating in nature instead of for the purpose of decision-making in clinical care.Some significant differences from these multiple comparisons can be due to chance; thus these "significant" results need to be validated in future confirmatory studies.
In conclusion, the DREAM study provided a unique opportunity to study the presence of systemic inflammatory markers in DED and a subgroup of DED participants with SS.This is the first reported evaluation of the association of systemic inflammatory markers with the symptoms and signs of DED in a large group of moderate to severe DED participants.We found some associations between systemic markers and DED signs, symptoms, and SS status.Future research will need to evaluate these findings further to determine the clinical significance of these markers in the diagnosis and management of DED.

Table 1 .
summary of LIa markers found in participants' serum at baseline (N = 481).

Table 2 .
number of LIa inflammatory markers positive per subject.

Table 3 .
association of markers positive with dry eye symptoms at baseline.

Table 4 .
: association of marker positivity with composite severity score of the signs.

Table 5 .
association of number of markers that were positive with composite severity score of DeD signs.

Table 6 .
Comparison of marker positive rates by ss disease status.