The Ocular Surface Tear Film as a Biomarker for Systemic Health

Abstract

The tear film is a complex structure with rich interactions with the human body. A growing body of evidence suggests that measuring changes in protein, lipid, or other metabolite concentration in the tear film can be used to help detect disease. Particularly in the era of precision medicine, the tear film serves as a promising source of non-invasive insights into systemic health for early diagnosis and treatment. This paper analyzes the latest research in tear film biomarkers for systemic diseases. The review was conducted through PubMed and Embase databases using the PRISMA protocol and includes 54 articles. This paper first reviews the anatomy and physiology of tear film, as well as the latest proteomic analysis techniques on the tear film. We then provide a disease-by-disease review on the tear film as a biomarker including 5 articles related to Alzheimer’s Disease, 10 articles related to Cancers, 1 article related to Cystic Fibrosis, 1 article related to Migraines, 4 articles related to Multiple Sclerosis, 15 articles related to Parkinson’s Disease, 7 articles related to Rheumatoid Arthritis, and 11 articles related to Thyroid Disease. This paper highlights the promising results of these studies yet also reviews the challenges with limited sample sizes, reproducibility, and biological understanding of biomarkers. We conclude this paper with insights for future work to ensure clinical validity and generalizability. Ultimately, the tear film is a clinically accessible, complex structure that provides a wealth of information that may contribute to a more comprehensive understanding of systemic health.

1. Introduction

In the era of precision medicine, there is an emphasis on identifying clinically accessible biomarkers to further understand systemic health. Particular goals in precision medicine include predicting disease risk prior to symptoms, as well as monitoring treatment response to tailor and optimize outcomes and safety [1]. Oculomics, a branch of precision medicine when it relates to the eye, is the study of ocular biomarkers and how they intersect with systemic diseases [2]. With advancements in the field, there is strong evidence to suggest oculomics can be used as a non-invasive approach for detecting cardiovascular disease and diabetes [3]. However, other branches of oculomics such as cancer related biomarkers in tear fluid are still in much earlier stages of research [3]. Notably, in October 2024, the U.S. Department of Health and Human Services Advanced Research Projects Agency for Health (ARPA-H) announced a stated goal of developing a continuous tear monitoring device through the OCULAB initiative which focuses on developing tear-based biomarkers for continuous health monitoring [4]. Blood and urine tests have been staples in modern medicine for capturing disease, however, these gold standard tests have limitations in accessibility. The goal of OCULAB is to provide accessible, continuous monitoring with the tear film biomarkers to provide a more comprehensive understanding of health [4]. Although an exciting area of study, tear film-based oculomics is a relatively new field, thus warranting further investigation, particularly in the context of tear film and its intersection with systemic diseases.

The tear film, being the outermost layer of the eye, serves an important role in ocular health: both as the protector and the lubricator [5]. Various systemic diseases have effects on the tear film, often manifesting as dry eye disease (DED). These diseases include Sjögren’s syndrome, facial rosacea, rheumatoid arthritis, and peripheral artery disease and have been significantly associated with DED [6]. There continues to be growing evidence suggesting that in patients of certain diseases (with or without DED), there are corresponding changes in their tear film biomarkers. In this review, we aim to examine the current literature on ocular proteomics, with a focus on the tear film’s role as a biomarker for systemic diseases.

2. Methods

This systematic review was conducted in accordance with the PRISMA protocol. A comprehensive literature search was performed using the PubMed and Embase databases to identify relevant articles examining the relationship between biomarkers in the ocular surface tear film and a predetermined list of diseases. The diseases chosen were primarily non-ocular: Alzheimer’s disease, cancers, cystic fibrosis, migraine, multiple sclerosis, Parkinson’s disease, rheumatoid arthritis, and thyroid disease. Diabetes was not included due to the extent of literature on the topic and the large overlap with ocular manifestations of systemic diabetes. Additionally, an initial search found over fifty research articles on tear biomarkers in diabetes, suggesting an independent meta-analysis would be needed to thoroughly cover the depth of work on the disease.

The PubMed query was as follows:

(“tear film” OR “tear biomarkers” OR “tear proteomics” OR (“tear” AND “biomarkers”) OR (“tear” AND “biomarker”)) AND (“Alzheimer” OR “cancer” OR “Cystic Fibrosis” OR “headache” OR “Multiple Sclerosis” OR “Parkinson” OR “Rheumatoid” OR “Thyroid”)

For Embase, the query used was:

(‘tear film’:ab,ti OR ‘tear biomarkers’:ab,ti OR ‘tear proteomics’:ab,ti OR (‘tear’:ab,ti AND ‘biomarkers’:ab,ti) OR (‘tear’:ab,ti AND ‘biomarker’:ab,ti)) AND (‘alzheimer’:ab,ti OR ‘cancer’:ab,ti OR ‘cystic fibrosis’:ab,ti OR ‘headache’:ab,ti OR ‘multiple sclerosis’:ab,ti OR ‘Parkinson’:ab,ti OR ‘rheumatoid’:ab,ti OR ‘thyroid’:ab,ti) AND [english]/lim

Studies were included if they met the following criteria: investigated molecular biomarkers in the tear fluid, compared patients with a listed disease against a control group, and provided original research data. Studies were excluded if they were review articles, only measured aggregate property of tears (such as osmolarity or tear breakup time), focused solely on animal models, were not published in English, or if only an abstract was published and not a full manuscript.

Articles from PubMed and Embase were collected using the queries provided and duplicates removed, leaving 431 articles for review. Two independent reviewers screened the titles and abstracts for relevance based on the inclusion and exclusion criteria. After reviewing titles and abstracts, 75 articles remained under consideration for inclusion. After full-text screening, 54 articles were selected for final analysis. Figure 1 shows the flowchart of article selection and the breakdown of articles included by disease. Figure 2 shows a summary of biomarkers found: more detail is provided in each corresponding section of this paper.

Figure 1:

Study selection according to the PRISMA protocol

Figure 2:

Summary of potential biomarkers in the tear film for systemic diseases. Central tear film graphic reprinted with permission from Shan et al. [7] under Creative Commons License. All other images are royalty-free vector graphics.

3. Environmental and Systemic Impact on Tear Film Homeostasis

The tear film is a very intricate system that is supplied from various regions of the eye such as the lacrimal gland or the meibomian glands. However, the layer is constantly in flux as fluid is being lost through various mechanisms such as blinking or evaporation. These are not the only mechanisms that affect the amount of fluid in the tear film. Other than evaporation, there are other environmental factors (e.g., humidity, air quality) that can play a role in the homeostasis of the tear film. Conversely, there are certain systemic factors (e.g., hydration, autoimmune diseases) that, too, can affect tear film homeostasis.

There have been many studies exploring the effect of environmental factors on tear film stability. Environmental factors are divided into two main categories: air quality and weather conditions [8].

1. Weather Conditions: Across various weather conditions, some that have been found to have the most effect on the tear film and ocular surface are humidity, altitude, temperature, sunlight, and wind. Each of which showcases independent effects on the tear film. Many of these studies were divided to be specific to indoor or outdoor effects, however, the main findings across most studies found that high temperatures, wind, and altitude play a significant role in worsening dry eye symptoms whereas high humidity was found to be more protective [8].

2. Air Quality: This category is subdivided into two smaller categories. Particulate matter (defined as particles of various sizes and compositions floating around in the air) and gaseous material. In terms of the particulate matter (PM), there is a general link between oxidative stress and worsening of dry eye symptoms and diagnosis, despite confounding evidence of indoor vs outdoor PM. The gaseous pollutants studied by various researchers were nitric dioxide (NO₂), sulfur dioxide (SO₂), carbon monoxide (CO), ozone (O₃), and volatile organic compounds (VOCs) (formaldehyde, toluene, acetone, ethanol, and more). These gases were also generally found to have effects on inflammation and irritation of the ocular surface and worsened symptoms of dry eyes. However, regarding specific gases themselves, the evidence was inconsistent [8].

Various systemic factors are also involved in changes in tear film secretion. Those include hormonal stimulation and nervous regulation.

1. Androgens: These play a key role in tear film stability, which they are able to do via influence of meibomian and lacrimal glands [9].

   1. Androgens promote the function of meibomian glands via regulation of lipid production and secretion. Patients who are either androgen deficient or on some form of antiandrogen therapy are found to have increased debris in their tear film as well as decreased lipid content [9].
   2. Similarly, androgens play an important role in the growth of the lacrimal gland as well as the secretions of lacrimal gland fluid [9].
2. Estrogen: In the past, it was believed that estrogen deficiency worsened tear film stability and dry eye disease since menopausal women have worse dry eye disease. However, hormone replacement therapy (HRT) in this population was found to actually worsen dry eye symptoms compared to those not on HRT [10]. More recent evidence points to estrogen having a negative effect on tear film stability [9]. Similar to androgens, it exerts its effect on both the meibomian and lacrimal glands [9].

   1. In the meibomian gland, estrogen causes a decrease in lipid production and secretion. However, even with estrogen’s antagonistic effect, it does not play as big a role in dry eye symptoms in comparison to androgen deficiency [9].
   2. In the lacrimal gland, estrogen’s role is still a topic of research and debate. Some studies point towards minimal effect while some state that it has adverse effects on the lacrimal gland and can lead to acinar cell necrosis [9].

3. Thyroid hormone: Patients with any thyroid problems, generally, are at an increased risk of dry eye disease [11]. Some mechanisms suggest that the lacrimal gland is involved as there is a general reduction in tear secretion associated with thyroid issues. Patients with thyroid eye disease have elevated rates of meibomian gland dysfunction (MGD) which can lead to increased evaporation of tears [11].

4. Vitamin D: Vitamin D has been shown to play a role in the treatment of dry eye disease [12]. It has been shown to improve osmolarity in tears and reduce inflammation at the superficial side of the eye and margins of the eyelid [9,12].

5. Autoimmune disease: The most common symptom related to the eyes that patients with autoimmune disease complain of is dryness. This is often due to their various effects on the lacrimal and meibomian glands. However, beyond those two, they also affect the cornea, conjunctiva, and lacrimal duct all aiding in the worsening of dry eye symptoms [13].

As alluded to in the previous section, there are various factors that play a role in the health and stability of the tear film. Often when one or more of those systems fail, it leads to a collection of diseases called dysfunctional tear syndrome (DTS). Dysfunctional tear syndrome is an umbrella term that includes various diseases such as aqueous deficiency, blepharitis/MGD, goblet cell deficiency, and exposure-related DTS [14].

Interaction between tear film abnormalities and systemic health.

Often, these abnormalities in the tear film are secondary to broader systemic issues and can be traced back to another process in the body. When comparing the layers of the tear film, each layer has specific disease processes associated with it [15].

• The lipid layer can often be affected by any disease process that causes MGD [15]. As discussed above, androgen deficiencies as well as anti-androgen therapy are a common cause of MGD. Other systemic causes of MGD include type 2 diabetes mellitus, use of tobacco smoke, migraines, chronic blepharitis, atopy, demodicosis, insufficient protein intake in bariatric patients, Sjögren’s syndrome, Steven-Johnson syndrome, psoriasis, trachoma, contact lens wearers, as well of use of specific medications such as systemic antibiotics, isotretinoin, antidepressants, topical epinephrine, topical beta-blockers, or topical prostaglandin analogs [15–17].

• The aqueous layer becomes dysfunctional often secondarily to lacrimal dysfunction. It is divided between Sjögren and non-Sjögren dry eye [15]. The non-Sjögren dry eye can have various etiologies including systemic rheumatologic/autoimmune diseases, depression, age-related atrophy, Mikulicz disease, contact lens use, antidepressant use, and hydrochlorothiazide usage [15,17].

• Less commonly, dry eyes can be secondary to a defect in the mucin layer. Most often, it happens due to a vitamin A deficiency, which in and of itself, can have multiple etiologies. Less commonly, it can be due to Steven-Johnson syndrome or a manifestation of severe burns [15].

4. Proteomic Analysis of the Tear Film

Advances in molecular biology, lab techniques, and data analysis pipelines have led to the rise of proteomics as a paradigm for understanding human health. Proteomics, the study of the structure, function, and interaction of proteins, provides a more dynamic assessment of cellular state compared to standard genomics [18]. It is also more complicated, however, as protein expression varies by cell type and over time. Sophisticated lab techniques are needed to detect and measure proteins, as well as differentiate between different conformational states and post-translational modifications. There has been increasing interest in proteomics as the bridge between genetics and clinical manifestation of disease, with an emphasis on diagnostics, therapeutics, and prognostication [19].

Human tears have a rich proteome, with over 1500 expressed proteins identified so far [20]. The protein concentration ranges from about 6 to 10 μg/mL [21]. The tear proteome is distinct in terms of both which proteins are present and how much they are expressed compared to other fluids in the body [20]. Due to the tear film’s unique protein characteristics and ease of access, proteomic analysis of the tear film has emerged as an exciting avenue of research for the discovery of novel biomarkers.

The most common methods for collecting tears are Schirmer strips and glass microcapillary tubes (Figure 3) and Schirmer strips (Figure 4), though other devices are also used [22]. The methods vary in terms of the amount of liquid produced with Schirmer strips generally producing the most for non-dry eyes, while cellulose sponges more reliably extract tears from dry eyes [23]. Lower fluid yield is one of the technical difficulties with tear proteomics analysis compared to other biological fluids [24]. Additionally, each collection method may introduce its own impact on the proteomic profile of the collected fluid due to interactions between proteins and the collection device [25,26]. This variability presents challenges when comparing results across different studies.

Figure 3:

Tear fluid collection using plastic capillary tubes. Tear fluid is collected from the lower conjunctival fornix (a) using an end-to-end plastic capillary tube with a capacity of 10 μl. The capillary is transferred to a punctured (b) 0.5 ml tube (c). It is centrifuged into a 1.5 ml tube (d) to retrieve the fluid. Reprinted with permission from Bachhuber et al. [22] under Creative Commons License.

Figure 4:

Tear fluid collection using Schirmer strips. The Schirmer strip (a) is bent at the 0 line to an angle of 120 degrees (b). The bent part is placed beneath the inferior eyelid (c) near the midpoint and left to soak for five minutes (d). A picture 0.5 ml tube (e) is used to hold the soaked part of the Schirmer strip (f). A solution of 0.9% NaCl is added in a fixed multiple of the collected tear fluid. After 2 hours of incubation, the fluid is centrifuged into a 1.5 ml tube (h). Reprinted with permission from Bachhuber et al. [22] under Creative Commons License.

Fluid samples stored at room temperature lose protein over time due to the presence of catabolic enzymes such as hydrolases. Keeping the sample at −80°C shows low protein loss even up to a month out and so has become the gold standard for tear storage for proteomic analysis [27]. A recent study found comparable protein concentrations when stored at −20°C and −80°C, suggesting that −20C can be appropriate when a colder freezer isn’t available [28].

A variety of techniques exist to detect and measure proteins. To measure total protein concentration, the general approach is to apply a reagent creating a hue in the sample and measuring absorption with a spectrophotometer. Commonly used assays include the Bradford assay, bicinchoninic acid (BCA) assay, or Lowry assay [29].

To detect individual proteins, the traditional approach is to begin with a separation technique such as gel electrophoresis, most commonly SDS-PAGE or DIGE, or some variant of chromatography such as thin layer chromatography (TLC) or high-performance liquid chromatography (HPLC) [30]. Next, antibody-based detection with methods such as Western blot, ELISA, multiplex bead immunoassays, or antibody microarray are used to detect and measure specific proteins [31].

More recently, mass spectrometry (MS) has become the method of choice for high throughput analysis of proteins [32]. Termed “shotgun proteomics”, the goal is to detect all proteins in a sample rather than measure the concentrations of a previously determined set of proteins, as is the case with antibody-based approaches. Many variants of MS exist with differing ability to detect certain classes of proteins. As a general overview, the stages of MS involve sample preparation, optional peptide labeling, optional sample separation, ionization, mass analysis, and data processing. Different pipelines may rearrange the order of steps or include additional steps.

1. Sample preparation involves pretreating the sample and breaking down proteins into peptides using a protease.

2. Peptide labeling, using a method such as iTRAQ, applies isobaric tags to pool multiple samples into the same experiment, improving accuracy and reproducibility [33].

3. Sample separation is most often done with liquid chromatography (LC). Reverse phase or and strong cation exchanges can be used with LC to give the RP-LV or SCX-LC variants respectively [34]. Recently “nano-LC”, using low flow rates through narrow columns, has shown higher sensitivity in smaller volume samples [35].

4. Ionization is traditionally done with electrospray ionization (ESI), but alternative methods such as matrix-assisted laser desorption/ionization (MALDI) or surface-enhanced laser desorption/ionization (SELDI) can be used [36]. MALDI or SELDI can also be used for intact protein separations, rather than on peptides.

5. Mass analysis is the stage where peptides are detected. Various options provide different resolving power and mass accuracy. The available mass analyzers include orbitrap, quadruple time-of-flight (Q-TOF), triple quadruple (QqQ), time-of-flight (TOF), and Fourier transform ion cyclotron resonance (FT-ICR) [37]. A subset of mass analyzers that confine selected ions to a region of space are known as ion traps, such as the quadrupole ion trap and linear trap quadrupole (LTQ). Tandem mass spectrometry (MS/MS) involves a second mass analysis step using the ions from the ion trap.

6. Finally, signals from the mass analysis are compared against reference databases and used to come up with protein measurements in the data analysis stage.

Antibody based methods can reliably detect major proteins in tear fluid, but the findings have been generally limited to detecting under 100 unique proteins. In 2005, Li et al used gel-based and solution-based approaches with LC-ESI and LC-MALDI MS and MS/MS to confidently identify 54 proteins in the tear fluid [38]. In 2006, de Souza et al used SDS-PAGE followed by hybrid linear ion trap-Fourier transform (LTQ-FT) and a linear ion trap-Orbitrap (LTQ-Orbitrap) tandem mass spectrometry to identify 491 proteins in the tear fluid [39]. In 2012, Zhou et al improved the number of proteins to 1543 using SCX chromatography, followed by nano-RPLC-MS/MS with a high speed TripleTOF system for mass analysis [20].

5. Systemic Diseases

In this section, we discuss various systemic diseases that have been studied with the tear film as biomarkers. As the diseases are diverse and there are a multitude of biomarkers, we organize this based on disease and first describe the disease and the individual biomarkers found rather than organize by individual tear biomarker. Various diseases included do have ocular findings, but the pathophysiology is systemic. Of note, although diabetes is a systemic disease, we have not included diabetes in the discussion below since multiple diabetic ocular changes can be identified through clinical exam without the need for additional biomarkers.

5.1. Alzheimer’s Disease

Alzheimer’s disease (AD) is a neurodegenerative disorder resulting in cognitive decline across a range of functions. There are approximately 50 million people living with dementia, and this number is estimated to triple by 2050. AD is the most common cause of dementia and is thought to account for up to 80% of cases [40]. Early detection of AD not only provides patients with an opportunity to develop advanced care plans and support, but diagnosing AD in the early stages could save up to $7 trillion in healthcare burden [40]. It may also lead to early treatment and prolonged preservation of patient’s cognitive function [40]. Though the exact etiology of AD is unknown, the pathophysiology may involve the production and accumulation of misfolded proteins such as beta-amyloid peptides and protein tau. These toxic substrates lead to the appearance of hallmark neurofibrillary tangles and plaques. The clinical presentation exists on a spectrum with other entities such as subjective cognitive decline (SCD), mild cognitive impairment (MCI), and dementia. Alzheimer’s is diagnosed primarily through clinical presentation and exclusion of other diagnoses, but suggestive imaging findings can play a role as well. The search for a fluid-based biomarker remains open. A summary of the reviewed articles relating to Alzheimer’s disease is presented in Table 1.

Table 1.

Tear film biomarker studies for Alzheimer’s Disease.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Schirmer strip	Assays specific for Ab and tau protein	23 patients with SCD, 22 with MCI, 11 with dementia, 9 controls	In a subset analysis of 23 patients for whom CSF was also collected, total tau protein was elevated in dementia compared to SCD as well in patients with neurodegeneration according to the A/T/N staging system compared to those without.	Gijs et al (2021) [41]
Microsponge	ELISA for b-amyloid Western blot for phosphorylated tau protein (p-tau)	11 patients with MCI, 10 with AD, 14 controls	Ab1-42 was significantly lower in MCI and AD compared to control. No differences were found with p-tau.	Gharbiya et al (2023) [42]
Capillary tube	SDS PAGE, LC-MS/MS Selected reaction monitoring (SRM)	14 patients, 9 controls	The differentially expressed proteins were identified as components of the chemical barrier of the eye. A combination of lipocalin-1, dermcidin, lysozyme-C, and lacritin was suggested as a possible AD biomarker.	Kalló et al (2016) [43]
Schirmer strips	SDS-PAGE, RP-LC-MS/MS miRNA assay	9 patients with AD, 8 with MCI, 15 controls	Twelve proteins were differentially expressed in AD, one of which was elongation initiation factor 4 (eIF4E). Levels of microRNA-200b-5p were higher in AD compared to controls.	Kenny et al (2019) [44]
Polyester fiber rod	LC-MS/MS Self-assembled nanoparticle-mediated amplified fluorogenic immunoassay	Discovery cohort (21): 7 patients with MCI, 7 patients with AD, 7 controls Validation cohort (39): 15 patients with MCI, 10 patients with AD, and 14 controls.	75 differentially expressed proteins were identified, with adenylyl cyclase-associated protein 1 (CAP1) suggested as a biomarker due to its significant difference in expression and protein function. Testing on a validation cohort showed sensitivity 90% and specificity 100% with an appropriately chosen threshold.	Lee et al (2023) [45]

An initial investigation for tears begins with looking for amyloid peptides and tau protein. Two studies sought to compare these molecules in patients with cognitive impairment vs controls. Gijs et al [41] compared levels of amyloid peptides and tau protein in tears between controls, patients with SCD, patients with MCI, and patients with dementia. They did not find significant differences for either protein using the data of all patients. Among a subset of patients for whom CSF was collected, they found that total tau was elevated in dementia (n=3) compared to SCD (n=9) and that it was correlated with neurodegeneration with the A/T/N classification system for Alzheimer’s disease. In another study, Gharbiya et al [42] found significantly lower amyloid peptide Ab1-42 with MCI and AD patients compared to controls. They did not find any difference in phosphorylated tau-protein.

In the more general search for protein biomarkers in tears for AD, Kalló et al [43] compared the expression of proteins with LC-MS/MS between three randomly selected AD patients and two controls among their study population. The differentially expressed proteins were identified as components of the chemical barrier of the eye. They selected ten proteins to measure for their entire cohort of 23 patients using selected reaction monitoring (SRM). They found that a combination of four proteins lipocalin-1, dermcidin, lysozyme-C, and lacritin could be a potential biomarker, with an 81% sensitivity and 77% specificity on their cohort for the diagnosis of AD.

Kenny et al [44] also performed shotgun proteomic analysis with LC-MS/MS, adding a microRNA assay to the data collected. They found twelve differentially expressed proteins, with a GO term analysis showing that eleven fell into the category of metabolic process. They found elongation initiation factor 4 as the most significant protein, given it appeared in 55% of the AD samples but none of the controls. They were unable to verify its presence in the AD tear fluid using a Western blot afterward. Their microRNA analysis identified microRNA-200b-5p as a potential biomarker. MicroRNA represents a relatively unexplored potential source of biomarkers, existing in the space between genomics and proteomics.

Lee et al [45] used LC-MS/MS to perform proteomic analysis of a discovery cohort of 21 patients, where they found 75 differentially expressed proteins between AD and controls (Figure 5). Upon further analysis, they selected adenylyl cyclase-associated protein 1 (CAP1) as a potential biomarker due to the magnitude of the change in expression as well as its role in neuronal actin dynamics and functioning of growth cones. They then designed an enzyme-free nanoparticle-based immunoassay to detect CAP1, which was then used on a verification cohort of 39 patients. Their test area under the curve (AUC) values for MCI and AD were 0.7619 and 0.9714 respectively, with p values less than 0.0001. With an appropriately chosen threshold, their test distinguished AD from healthy controls with sensitivity 90% and specificity 100%, showing promise as a biomarker for AD.

Figure 5:

Flow of differentially expressed protein discovery using proteomic methods for a study investigating Alzheimer’s disease. A) Schematic of tear collection and processing to obtain potential protein biomarkers in Alzheimer’s disease (AD) or minor cognitive impairment (MCI) compared to healthy controls (HC). B) Heatmap of differentially expressed proteins obtained by comparing the relative expression levels in patients with MCI or AD to the HC group (MCI/HC or AD/HC). A minimum ratio of 1.5 defines upregulated (top) and a maximum ratio of 0.67 was used for downregulated (bottom). C) Protein interaction network generated for differentially expressed proteins. The inner and outer gray circles represent the relative protein expression (fold change) of each of the AD and MCI groups compared to the HC group, with values close to 1.5 and −1.5 on the log2 fold change scale shown in red and blue, respectively. Reprinted with permission from Lee et al. [45] under Creative Commons License.

Though not specifically looking for biomarkers, and thus not part of the articles included for review, two other studies measured proteins in the tear fluid relating to Alzheimer’s. Sampani et al [46] performed an agreement validation study comparing the concentration of various neurodegenerative proteins in eye fluids relative to the blood in 79 patients undergoing vitrectomy, all without formal dementia diagnoses. In this cognitively healthy cohort, they found generally higher levels of the proteins considered in eye fluids compared to the plasma, as well as a significant correlation between AB40 in plasma and tears. In another study on a healthy cohort, van der Thiel et al [47] found elevated tear total-tau correlates with enlarged perivascular spaces in the brain as seen on MRI. These studies highlight the ability to even use healthy cohorts to investigate the physiological regulation and role of potential biomarkers.

5.2. Cancer/ Malignancy

Cancer represents an inhomogeneous group of disorders with various pathological regulatory pathways leading to abnormal tissue growth. While the United States spends a significant portion of healthcare cost on cancer screening, the life expectancy is lowest compared to other industrialized countries [48]. Given the need for better and earlier screening, there have been tremendous effort to identify serum biomarkers, which can aid in diagnosis, prognostications, and monitoring of treatment or recurrence [49]. The tear fluid presents a novel avenue for biomarkers. Most work so far has investigated tear biomarkers in breast cancer, but lymphoma, colorectal cancer, and head and neck cancers have also been studied. A summary of the reviewed articles relating to cancers is presented in Table 2.

Table 2.

Tear film biomarker studies for malignancies.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Glass microcapillary tubes	Gel electrophoresis	8 breast cancer patients, 6 lung cancer patients, 5 colon cancer patients, 3 ovarian cancer patients, 1 prostate cancer patient, 5 controls.	Higher percentage of cancer patients had lacryglobin detected in the tear fluid compared to controls, though the study size was too small to assess significance.	Evans et al (2001) [50]
Schirmer strip	SELDI-TOF-MS	Study 1: 10 breast cancer patients, 10 controls Study 2: 50 breast cancer patients, 50 controls	Statistically significant differences in protein patterns were found between breast cancer patients and controls.	Lebrecht et al (2009) [51] and Lebrecht et al (2009) [52]
Schirmer strip	MALDI-TOF-MS	25 breast cancer patients, 25 controls	27 proteins were identified as increased or decreased by a factor of at least two in breast cancer patients compared to controls.	Böhm et al (2012) [53]
Schirmer strip	Reverse-transcription polymerase reaction (qRT-PCR), Western blot	5 metastatic breast cancer patients, 8 controls	Increased expression of breastcancer-specific miR-21 and miR-200c were found in the breast cancer patients compared to controls.	Inubushi et al (2020) [54]
Schirmer strip	LC-MS/MS ELISA	Initial: 51 breast cancer patients, 51 controls Validation: 75 breast cancer patients, 96 controls.	14 proteins were identified as possible biomarkers with data from the initial group. Three were selected for ELISA measurement on the validation group, with S100A8 and S100A9 showing significantly higher expression in breast cancer patients.	Daily et al (2022) [55]
Schirmer strip	ELISA	Cohort 1: 391 samples (87 confirmed breast cancer). Cohort 2: 456 samples (21 breast cancer).	Following up on the above, logistic models were built using Cohort 1 data with SN100A8 and SN100A9 as inputs. Evaluation across both cohorts showed sensitivity ranging from 52 to 90% and specificity from 31 to 79% depending on model choice and diagnostic threshold.	Daily et al (2022) [58]
Glass microcapillary tubes	Cytokine multiplex panel using immunofluoresc ence	21 patients with extra nodal marginal B-cell lymphoma of the ocular adnexa (OA-EMZL), 14 controls	The concentration of IL-1RA and IL8 were significantly increased in OA-EMZL while FGF-2, IL-2, and IL-4 were significantly decreased.	Xiao et al (2022) [59]
Syringe	LC-MS/MS	Low risk (6 benign hyperplastic polyps, 9 controls), High risk (13 with tubular adenomatous polyp, 16 with colorectal cancer)	80 proteins identified as significantly differentiated between the two groups.	Kaufmann et al (2022) [60]
Schirmer strip	LC-MS	29 radiated head and neck cancer patients, 21 controls	Over 90 proteins were identified as increased or decreased by a factor of at least two in cancer patients compared to controls.	Hynne et al (2022) [61]

An early study in 2001 by Evans et al [50] measured lacryglobin in human tears using electrophoresis. Lacryglobulin was selected due to its high sequence homology to mammaglobins, proteins upregulated in breast cancer. Across a total of 28 patients, they found lacryglobin was present in the tears for 88% of breast cancer patients (n=8), 83% of lung cancer patients (n=6), 100% of colon cancer patients (n=5), 33% of ovarian cancer patient (n=3), 100% of prostate cancer patients (n=1), and 60% of controls (n=5). The study was not powered to assess for statistical significance, but it was ahead of its time in its use of tear fluid for biomarker discovery.

Several studies specifically investigate breast cancer. Lebrecht et al [51] used SELDI-TOF-MS to perform shotgun proteome measurement. Multi-discriminant analysis was used to select significantly different peaks between the breast cancer and control group. A neural network trained on the data set was able to classify breast cancer from controls with an AUC of 0.90 and sensitivity of 90%. Lebrecht et al followed up with a second study [52] increasing sample size from 10 breast cancer patients and 10 controls to 50 of each. Their second study followed the same methodology as the first, and it resulted in the neural network having an AUC of 0.75, corresponding to a sensitivity and specificity around 70% after selecting an appropriate threshold. Böhm et al [53] used MALDI-TOF-MS to measure proteins and found 27 proteins that were increased or decreased by a factor of at least two in breast cancer patients compared to controls. Many of these proteins were involved in either metabolism or immune response. Inubushi et al [54] searched the tears for breast cancer specific oncogenic miRNA in exosomes. They found that two of the five miRNAs for which they searched, miR-21 and 200c, were significantly higher in their cohort of patients with metastatic breast cancer compared to healthy controls.

Daily et al [55] conducted a two-tier approach with separate data for the identification of potential biomarkers and their validation. They used 102 samples (half from breast cancer patients and half from controls) along with LC-MS/MS to identify 14 possible protein biomarkers. They selected three for validation using ELISA on validation samples (75 breast cancer patients and 96 controls). Two of their putative biomarkers, S100A8 and S100A9, showed significantly higher expression in the breast cancer patients using the validation group with p < 0.0001 for each. Of note, S100 proteins are thought to play a role in the development of various cancers, and specifically for breast cancer through association with non-functional BRCA1 [56,57]. Daily et al [58] then followed up with a larger study, building a first cohort of 391 samples and a second cohort of 456 samples. They used the first cohort’s data to create logistic regression models with S100A8, S100A9 and other measurements such as age as inputs. They evaluated their models across both cohorts, observing sensitivity ranging from 52 to 90% and specificity from 31 to 79% depending on model and diagnostic threshold chosen.

Tear-based biomarkers have also been investigated for other cancers. Xiao et al [59] measured cytokine expression in the tears of patients with extra nodal marginal B-cell lymphoma of the ocular adnexa (OA-EMZL) compared to healthy controls. They analyzed 27 cytokines using a multiplexed assay and found that two (IL-1RA and IL-8) were significantly increased in OA-EMZL patients, while three (FGF-2, IL-2, and IL-4) were significantly decreased. Kauffman et al [60] conducted a shotgun proteomic analysis for low-risk and high-risk groups for colorectal cancer based on colonoscopy. The low-risk group was defined as either no finding or hyperplastic polyp, and the high-risk group included tubular adenomatous polyps and colorectal cancer. They identified 80 proteins as significantly differentiated between the two groups, of which 9 are associated with pathways demonstrated to be altered in colorectal cancer. Hynne et al [61] used LC-MS to measure proteins in the tears of head and neck cancer patients who previously underwent radiation to compare to controls. Their study also compared the proteins of the saliva of the patients with a third group consisting of patients with primary Sjögren’s syndrome. In terms of tears, they found over 90 proteins differentially expressed in the tears between the two groups for tear analysis.

5.3. Cystic Fibrosis (CF)

Cystic fibrosis is the most common, life-threating genetic disease among Caucasians, occurring approximately 1 in 3000 in the United States [62]. Historically CF was known as the cause of “salty skin” symptoms, leading to much speculation and misunderstandings about the disease. However, with modern science more of its pathophysiology has been established [63]. It is an autosomal recessive mutation in the CFTR (cystic fibrosis transmembrane conductance regulator) gene on chromosome 7 [63,64]. CFTR functions as a regulator of chloride ion transport and mucosal secretions [65]. There are many different possible mutations however the most common is a mutation at codon 508 leading to a deletion of phenylalanine (ΔF508) [64]. Symptomatically it’s primarily known by a classic triad of recurrent pulmonary and sinus infections, steatorrhea, and malnutrition [66]. Early detection of CF, such as through the adoption of newborn screening, has shown to improve life expectancy as treatment may be initiated earlier [62]. The CFTR channel is also used in the secretory epithelia in the eye [67]. For this reason, it can cause ocular manifestations such as conjunctival and corneal xerosis and DED [65].

In our review, we found one study from Mrugacz et al. that explored the human tear film and any changes it may have in its biomarkers. The purpose of their study was to determine if the inflammatory cytokines (IL-8 and IFN-γ) in tear fluid had any correlation with systemic inflammation and disease severity in CF [67]. The study is summarized in Table 3.

Table 3.

Tear film biomarker study for cystic fibrosis.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Microcapillaries	ELISA	24	Il-8 and IFN-γ were significantly increased in tears of CF patients when compared to the control populations.	Mrugacz et al. (2006)[67]

Mrugacz et al. collected basal tears from 24 CF patients (15 male and 9 female, mean age 14.21) as well as 24 controls (14 male and 10 female, mean age 14.85). Tears were collected using microcapillaries from the conjunctival cul-de-sac. IL-8 and IFN-γ levels were analyzed via ELISA using high-sensitivity kits. Tears were pooled over 2–3 days, frozen at −80°C, and thawed before analysis. Both Il-8 and IFN-γ were significantly increased in patients with CF when compared to the control populations. They both also showed positive correlation with severe systemic symptoms of CF and negative correlation with mild systemic symptoms of CF. Interestingly, high IFN-γ also showed positive correlation with dry eye symptoms but IL-8 did not have such a finding.

Mrugacz et al. [67] hypothesized the following mechanism: Initially walking through the systemic causes of elevated inflammatory cytokines. IFN-γ causes the secretion of IL-8 when produced in excess. This process is key for inflammatory diseases as it “induces antiviral and antiproliferative activities, stimulates macrophages, and controls the expression of several adhesion molecules and surface cell receptors, several cytokines, and major histocompatibility complex (MHC) class I and II molecules” [67]. This activation also occurs in the eye with the activation of IL-8 leading to attraction of leukocytes and worsening inflammation contributing to worsened dry eye symptoms.

While Mrugacz et al. [67] doesn’t explicitly mention any limitations some factors must be remembered. The study was limited by a small sample size, conduction of the study at a single medical center, and no mention of presence of confounders and how they were managed.

5.4. Migraine

Migraines constitute a primary headache disorder that is diagnosed clinically through characteristic symptoms, triggers, and duration. The burden of migraine on patients spans a spectrum, with episodic migraine defined as less than fifteen occurrences per month and chronic migraine as more than fifteen [68]. It is estimated to affect more than 1 billion individuals each year globally, and can impact an individual’s work and school productivity, social relationships, and mental health [69]. The exact mechanism of migraine is not known but is thought to involve inflammatory vasodilation with waves of neuronal depolarization. Calcitonin-related gene peptide (CGRP) has been shown to play a role as a neuropeptide involved with vasodilation, and therapeutics blocking the action of CGRP have shown success in preventing the onset of migraines.

Kamm et al [70] compared CGRP in the tear fluid of 48 patients with episodic migraine, 45 with chronic migraine, and 48 controls using ELISA. The study is summarized in Table 4. They found higher CGRP levels in migraine patients not currently experiencing a migraine (“interictal”) compared to controls with a mean of 1.10 ng/ml compared to 0.75 ng/ml (p < 0.022). Subgroup analysis on the migraine patients didn’t show a difference between episodic and chronic migraine patients. They also collected tears from patients with migraines at the time (“ictal”) which showed further elevation in unmedicated ictal patients at a mean 1.92 ng/ml compared to interictal, while medicated ictal patients were indistinguishable from controls. Their study also quantified plasma CGRP and, besides having lower levels in the plasma compared to the tears, plasma levels were indistinguishable in all comparisons between patients and controls.

Table 4.

Tear film biomarker studies for migraines.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Plastic capillary tube	ELISA	48 patients with episodic migraine, 45 with chronic migraine, and 48 controls	CGRP significantly elevated in interictal migraine patients compared to controls. CGRP further elevated in unmedicated ictal patients.	Kamm et al (2019) [70]

5.5. Multiple Sclerosis

Multiple sclerosis is a demyelinating disorder of the central nervous system characterized by lesions separated by time and space. Multiple sclerosis is a chronic condition with autoimmune inflammation, though the exact cause remains unknown. It is often diagnosed between the age of 20 and 50, and is estimated to affect more than 2 million people worldwide [71]. One of the common presentations is optic neuritis. Given that most patients present with intermittent, relapsing-remitting disease course, a diagnosis may be delayed until years after the first onset. Early detection of multiple sclerosis, particularly for secondary progressive multiple sclerosis, is necessary for prompt initiation of treatment [72]. The interpretation of tears as an intermediate fluid between serum and CSF suggests its potential role as a source of biomarker for neurodegenerative diseases. A summary of the reviewed articles relating to MS is presented in Table 5.

Table 5.

Tear film biomarker studies for multiple sclerosis.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Schirmer strip	Affinity chromatograp hy to protein G, isoelectric focusing, silver staining	59 patients (known or suspected multiple sclerosis)	The concordance between oligoclonal IgG bands in the tears was limited, at 39% and a high rate of only marginal pattern in tears.	Hümmert et al (2019) [73]
Glass microcapillary tubes	Cytokine screening panel	84 patients, 70 controls	Lower level of cytokines found in multiple sclerosis, with significant differences in IL-1b, IL-6, and IL-10. No correlation with time since onset of disease or presence of optic neuritis.	Adamczyk-Zostawa et al (2024) [74]
Glass microcapillary tube	SDS-PAGE, LC-MS/MS Western blot	30 patients, 25 controls	Three proteins were found with consistent differences between multiple sclerosis and control: alpha-1 antichymotrypsin increased by 1.62 – 2.48x in multiple sclerosis; zymogen granule protein 16 homolog B decreased by 0.54 – 0.58x in multiple sclerosis; proline-rich-protein 4 decreased by 0.19 – 0.44x. Alpha-1 antichymotrypsin increase was confirmed by Western Blot and found to be consistent across serum and CSF.	Salvisberg et al (2019) [75]
Schirmer strip	LC-MS/MS Direct infusion multiple sclerosis	12 patients, 21 controls	32 phospholipids with differential expression were identified, with generally lower abundance in multiple sclerosis. Distinct metabolite compositions using carnitine and amino acids were found between the two groups.	Cicalini et al (2019) [76]

MS has a standard clinical marker with immunoglobulins in the cerebrospinal fluid (CSF), which are called oligoclonal bands (OCBs). Hummert et al [73] collected tears and CSF of 59 patients with known or suspected multiple sclerosis. They measured the presence of oligoclonal bands in each fluid, looking for concordance. They were unable to get a sufficient sample in 13 patients and found only a 39% concordance, leading to the authors not recommending replacing CSF OCB detection with tear OCB detection.

Reasoning about the autoimmune nature of multiple sclerosis, Adamczyk-Zostawa et al [74] specifically measured cytokines in 84 multiple sclerosis patients which they compared to 70 controls. They found a robust decrease in the overall level of cytokines in the tears of multiple sclerosis patients. They theorized that the reduced level of cytokines may represent anti-inflammatory mechanisms responding to the chronic inflammatory state of multiple sclerosis but note that the balance is delicate. They did not find that initiation of treatment or type of treatment explained their finding.

Salvisberg et al [75] conducted a general shotgun proteomic analysis of the tears of 30 MS patients, comparing against 25 controls. They conducted three independent experiments using tandem mass tagging with LC-MS/MS. Three proteins were found to be consistently different between patients and controls in the three experiments, of which alpha-1 antichymotrypsin was consistently elevated in multiple sclerosis. They validated the increased amount of alpha-1 antichymotrypsin in tear samples with direct Western blot measurement. They also found that it was increased in serum and CSF samples among multiple sclerosis patients compared to controls, leading to their assessment of it as a promising potential biomarker.

Cicalini et al [76] investigated nonprotein biomarkers such as lipids and amino acids in the tears of MS patients. They used LC-MS/MS to analyze lipids containing choline and direct infusion MS for the targeted measurement of metabolites such as carnitine, acylcarnitine, and amino acids. They found 32 lipids with differential expression between their 12 patients and 21 controls of which 15 were phosphatidylcholines, 6 lysophosphatidylcholines, and 11 sphingomyelins. They also found distinct tear metabolite compositions between MS and controls using raw metabolites measurements and statistical clustering. Besides contributing novel lab pipelines, this study emphasizes the potential role for non-protein biomarkers.

5.6. Thyroid Disease

When considering systemic diseases that affect the eye, thyroid disorders are one of the most common. Approximately 40% of patients with Graves’ disease have thyroid eye disease (TED) [77]. However, it is not only caused by hyperthyroidism as around 6% of patients with Hashimoto’s thyroiditis also have some level of TED [78]. TED can lead to significant facial disfigurement, visual disability, and even blindness [79]. Notably, the onset of TED can occur before the patient has experienced or been diagnosed with any thyroid problems [80]. Because of this we sought to search and see if ocular proteomics can be used as a diagnostic aid of Thyroid disorders. In our search criteria, we found 11 papers that met our inclusion criteria. Their findings are summarized in Table 6.

Table 6.

Tear film biomarker studies for thyroid disease.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Schirmer’s strips	iTRAQ and SWATH-MS	72	S100A4 was significantly downregulated as the severity of TED increased. PIP was significantly increased with TED severity	Chng et al. (2018) [81]
Schirmer’s strips	TMTsixplex, LC-MS/MS, western blot, and ELISA	53	Cystatin C and Alpha 1-antichymotrypsin concentrations significantly elevated in the tears of TAO patients, Retinal dehydrogenase 1 significantly less abundant in tears of TAO patients. Alpha 1-antichymotrypsin also showed significantly higher levels in patients with CAS ≥ 3 in comparison to CAS < 3.	Kishazi et al. (2018a) [82]
Schirmer’s strips	10-plex panel and ELISA	38	IL-10, IL-12p70, IL-13, IL-6, and TNF-α were significantly higher in TAO patients	Kishazi et al. (2018b) [83]
Schirmer’s strips	MALDI-TOF	120	28 proteins were significantly different across at least one of the four groups. Compared to dry eye: PROL1, UGDH, S10A8, SMCA4, annexin, cystatin, HSP27, and galectin were significantly downregulated in TAO. Compared to healthy controls: PROL1, PRP4, S10A8, and SMCA4 were downregulated in TAO. Midasin and POTE-ankyrin domain family member I were upregulated in TAO compared to controls.	Matheis et al. (2015) [84]
Schirmer’s strips	ELISA	43	IL-6 and IL-10 were significantly higher in TED patients compared to the control and No-TED groups	Nivean et al. (2025) [85]
Schirmer’s strips	SELDI-TOF-MS	60	A panel of biomarkers that were not precisely identified showed significant differences between TED patients and controls. With this biomarker panel a trained artificial neural network was able to discriminate and diagnose TED patients from controls. It had a sensitivity and specificity of over 90% each.	Okrojek et al. (2009) [86]
microcapillaries for basal tears and nasal stimulation for reflex tears	multiplex bead analysis	40	IL-1β was significantly higher in active TAO. IL-6 and IL-17A were both highest in active TAO and were also elevated in inactive TAO. TNF-α was higher in both active and inactive TAO. Lastly IL-7 was highest in inactive TAO.	Huang et al. (2014) [87]
microcapillary pipettes	Cytokine & Growth Factor Array with a biochip array analyzer	43	Graves’ hyperthyreosis had significantly higher levels of IL-6, VEGF, MCP-1 while IL-10 and IL-1α were lower compared to healthy controls.	Mandic et al. (2018) [88]
Passive tear collection	LC-MS/MS	12	107 proteins were found to be significantly altered. 62 of which were upregulated and 45 were downregulated when they were compared to the controls.	Jiang et al. (2021) [89]
capillary flow	multiplex bead array	48	GO patients had significantly higher levels of IL-1β, IL-6, IL-13, IL-17A, IL-18, TNF-α, and RANTES in tears compared to controls.	Ujhelyi et al. (2012) [90]
Schirmer strips	QTRAP 5500 and high-performance liquid chromatography	65	Active TAO had increased polyamine spermine and ornithine decarboxylase (ODC), propionyl carnitine and butyryl carnitine. ornithine, glycine, serine, citrulline, and histidine were decreased in active TAO	Billiet et al. (2022) [91]

Chng et al. [81] aimed to determine if tear biomarkers would change quantitatively and help establish a correlation with an increasing severity of TED in patients with autoimmune thyroid disease (AITD). The study consisted of 72 patients divided evenly among four groups: AITD without TED, mild TED, severe TED, and healthy controls. Tear protein analysis was performed using iTRAQ (Isobaric Tags for Relative and Absolute Quantification) in the discovery phase and SWATH-MS (Sequential Window Acquisition of All Theoretical Fragment Ion Spectra) in the validation phase. The study identified two potential TED biomarkers, S100A4 (Calcium binding protein A4) and PIP (prolactin induced protein). S100A4 was downregulated as the severity of TED increased (p=0.029) and the converse was true for PIP as it was upregulated with TED severity (p=0.023). Another key finding was that tear protein composition differed significantly with TED severity, showing that these biomarkers could reflect progression of the disease. The main limitation to this study is its small sample size.

Kishazi et al. (2018a) [82] investigated the tear fluid in patients with Thyroid-Associated Orbitopathy (TAO) to identify possible biomarkers. The study consisted of 28 TAO patients and 25 controls. Of the 28 TAO patients 54% had active disease which was defined as a clinical activity score (CAS) ≥ 3. They collected the tears of the patients with Schirmer strips that were then stored in −80°C. Proteomics Analysis was done through various steps. Firstly, TMTsixplex^™ (Tandem Mass Tag) was used to quantify protein differences between TAO and control tears. This step included reduction, alkylation, digestion, and fractionation. After, LC-MS/MS (Liquid Chromatography-Mass Spectrometry) was used for protein identification. Then, western blot and ELISA were used to validate the selected proteins. Lastly, Statistical validation was performed using Geometric Mean (GeoMean) and Coefficient of Variation (GeoCV) calculations. They found a total of 712 tear proteins. Within those 712, 221 proteins were consistently detected. They then quantified the 5 proteins that were the most upregulated along with the top 5 that were the most downregulated. Due to limited sample volume availability the top 3 that showed significant differences were chosen. Those being: Cystatin C, Alpha 1-antichymotrypsin, and Retinal dehydrogenase 1. Cystatin C (p=0.0082) and Alpha 1-antichymotrypsin (p=0.05) were found to have higher concentrations in the tears of TAO patients compared to controls. Retinal dehydrogenase 1 (p=0.012) showed the opposite, being much less abundant in tears of TAO patients. Also, Alpha 1-antichymotrypsin was significantly higher in patients with CAS ≥ 3 in comparison to CAS < 3 (p=0.014). This study’s limitations included: small sample size, monocentric, and low volume per sample.

Kishazi et al. (2018b) [83] investigated the levels of cytokines and soluble Il-6R in the tears of patients with TAO. The goal was to determine if the cytokines could serve as possible biomarkers for early TAO diagnosis. This study included 20 TAO patients and 18 controls. Of the experimental group 70% had a more active form (defined as CAS ≥ 3). 9 of the experimental group were tobacco consumers and 6 of the controls were tobacco consumers. Tears were collected via Schirmer’s test. The cytokines were analyzed using a 10-plex panel. For IL-6R an Invitrogen human sIL-6R ELISA kit was used. To summarize their findings, it was found that IL-10, IL-12p70, IL-13, IL-6, and TNF-α were significantly higher in TAO patients than in healthy controls. IL-10, IL-12p70, and IL-8 were elevated in all TAO patients, regardless of disease severity. IL-13, IL-6, and TNF-α were significantly higher in patients with moderate-to-severe TAO (CAS ≥ 3). None of these cytokines were able to distinguish mild TAO (CAS < 3) from controls. sIL-6R levels were able to distinguish mild TAO patients from controls but not able to differentiate severe cases from controls. IL-13, IL-6, and IL-8 showed a strong positive correlation with CAS. In terms of limitations to the study, there was a small sample size, technological variability, and the inability to address smoking as a confounder. Non-smoking TAO patients had significantly increased IL-10, IL-12p70, IL-13, IL-6, IL-8, and TNF-α compared to non-smoking controls. These results suggest that cytokine profiling in tears could aid in TAO diagnosis and disease activity assessment, though larger studies are needed to confirm these findings.

Matheis et al. [84] aimed to differentiate TAO from DES using proteomic analysis of tear fluid. The purpose was because TAO often presents with DES, if able to establish a specific protein marker, then there would be improvement in diagnostic accuracy and monitoring. The study was conducted with 120 participants. 30 patients had TAO with dry eye, 30 had TAO without dry eye, 30 had DES, and 30 were healthy controls. Their tears were collected using Schirmer’s test and the proteins were analyzed through MALDI-TOF (matrix-assisted laser desorption ionization mass spectrometry). A total of 69 proteins were identified, of which 28 were significantly different across at least one of the four groups. 18 between TAO and dry eye, 8 between TAO and controls, and 11 between dry eye and controls. Compared to dry eye, PROL1, UGDH, S10A8, SMCA4, annexin, cystatin, HSP27, and galectin were significantly downregulated in TAO. Compared to healthy controls, PROL1, PRP4, S10A8, and SMCA4 were downregulated in TAO. Midasin and POTE-ankyrin domain family member I were upregulated in TAO compared to controls. Overall, TAO showed an upregulation of inflammatory proteins and a downregulation of protective proteins. There was also a correlation with smoking as PROL1, PRP4, and UGDH were negatively correlated, meaning that higher smoking exposure was associated with lower levels of these proteins. In terms of limitations this study identified smoking as a confounder, small sample size, and lack of longitudinal data.

Nivean et al. [85] sought to investigate tear fluid biomarkers in patients with TED in a south Indian population. With the goal to find a biomarker to aid in early detection and disease monitoring in TED. A total of 43 patients were involved in the study. This included 18 TED patients, 11 of which had dysthyroid optic neuropathy (DON), 11 patients with thyroid dysfunction but no TED(No-TED), and 13 healthy controls. Their tears were collected using Schirmer’s strips and levels of various biomarkers were measured via multiplex ELISA using flow cytometry. Nivean et al. found that only IL-6 and IL-10 were significantly higher in TED patients compared to the control and No-TED groups. They also found that IL-6 was significantly elevated in patients with inactive TED, while IL-10 was also elevated it wasn’t significant. No significant differences were found in IL-2, IL-13, TNFα, IFNγ and PDGF-BB. They also found that IL-6 and IL-10 were not significantly associated with DON which they attributed to their steroid regimen. In terms of limitations this study states it had a sex distribution imbalance, the steroid treatment presented as a confounding variable, and lack of longitudinal follow up.

Okrojek et al. [86] investigated proteomic patterns in the tear fluid of TED patients. These proteins could then help in better understanding the autoimmune pathological processes that affect the eye as well as serving as biomarkers for diagnosis and follow up. The study consisted of 60 patients total. 45 of which had TED with varying levels of severity, the other 15 were healthy controls. Schirmer strips were used to collect the tears and they were stored at −80°C. Protein analysis was done with SELDI-TOF-MS (Surface-Enhanced Laser Desorption/Ionization Time-of-Flight Mass Spectrometry) and different chromatographic surfaces to identify protein expression differences. A panel of biomarkers that were not precisely identified showed significant differences between TED patients and controls. Most of the proteins were downregulated in TED patients and only a few being overexpressed. While no specific biomarkers were named the authors suggested that defensin and cathelicidin family proteins could be potential candidates due to their roles in immune response. With this biomarker panel a trained artificial neural network was able to discriminate and diagnose TED patients from controls. It had a sensitivity and specificity of over 90% each.

Huang et al. [87] aimed to investigate lacrimal gland changes as well inflammatory cytokines in tears of patients with TAO. The study included 24 TAO patients who were further divided into active and inactive, as well as 16 controls. MRI scans were done on patients to measure their lacrimal gland size, tear and serum analysis was done to look for serum cytokines IL-1β, IL-6, IL-7, IL-17A, IFN-γ, and TNF-α. They found that the lacrimal gland was significantly larger in both active and inactive TAO which was more so affecting the width rather than the length. When measuring the inflammatory cytokines of reflex tears they found IL-1β was significantly higher in active TAO compared to inactive and the controls. IL-6 and IL-17A were both higher in active and inactive TAO in comparison to controls, they were also higher in the active TAO population when compared to the inactive TAO population. While TNF-α was higher in both active and inactive TAO in comparison to controls. Lastly IL-7 was highest in inactive TAO when compared to active and controls. It was also lower than the controls in active TAO. In serum, IL-1β, IL-6, IFN-γ were significantly elevated in active TAO. However, IL-17A and TNF-α which were elevated in tears did not have significant differences.

Mandic et al. [88] investigated the contents and ratios of 12 cytokines and growth factors in human tears in patients with graves hyperthyroidism (GH) without TAO. They then compared that to healthy controls to determine if there are tear cytokine changes that could be indicators of subclinical TAO in patients with Graves. The study included 43 participants, 21 of which had GH and 22 being healthy controls. Their tears were collected using glass microcapillary pipettes and. The 12 cytokines and growth factors that were focused on were IL-2, IL-4, IL-6, IL-8, IL-10, IL-1α, IL-1β, IFN-γ, TNF-α, MCP-1, VEGF, and EGF. IL-6 was significantly much higher in the GH group compared to the controls. VEGF and MCP-1 were also significantly higher in the GH group. IL-10 and IL-1α were significantly higher in the control group. The ratio of all cytokines when compared to IL-10 were significantly elevated.

Jiang et al. [89] investigated the proteomic changes in tears of inactive TAO. With a goal of identifying biomarkers that are more prevalent in an inactive stage as many studies on this topic focus on the active stage. The study included 12 total participants: 6 with inactive TAO (CAS < 3) and 6 healthy controls. The tears were analyzed with LC-MS/MS. 107 proteins were found to be significantly altered. 62 of which were upregulated and 45 were downregulated when they were compared to the controls. The tear proteins suggested that immune activity, particularly interferon and IL-1 pathways, were elevated. There were also changes in proteins linked to metabolism, and proteins related to apoptosis were altered.

Ujhelyi et al. [90] investigated the cytokine profile in the tears of patients with graves’ disease (GD) with and without orbitopathy (GO). The tear samples that were collected were 27 patients who had GO, 9 patients with GD without orbitopathy, and 12 control patients. They focused on the levels of IL-1β, IL-6, IL-13, IL-17A, IL-18, TNF-α, RANTES, and PAI-1 using multiplex bead array. GO patients were found to have significantly higher levels of IL-1β, IL-6, IL-13, IL-17A, IL-18, TNF-α, and RANTES in tears compared to controls. It was also found that IL-6 release was 2.5 times higher in GO patients than in controls. There was no significant difference in cytokine levels of GD without GO patients to those with GO. The main difference between those two groups was the significant increase in PAI-1 levels in GO patients when compared to GD patients without GO and controls.

Billiet et al. [91] investigated the metabolic composition of tears in patients with active vs inactive TAO with the goal of identifying potential biomarkers for TAO activity. Their study consisted of 21 patients with active TAO (CAS ≥ 3), 24 patients with inactive TAO (CAS < 3), and a control group of 20 individuals. They collected tears using Schirmer strips. They found that patients who had active TAO had increased polyamine spermine and ornithine decarboxylase (ODC). They also had increased short chain acylcarnitine: propionyl carnitine and butyryl carnitine. They also found the following decreased in active TAO: ornithine, glycine, serine, citrulline, and histidine.

5.7. Parkinson’s Disease

Parkinson’s disease (PD) is a neurodegenerative disorder that is estimated to affect 1% of the population over the age of 60 [92]. It is one of the fasting growing neurologic disorders globally with significant economic burden. By 2037, the economic burden, which includes direct medical costs, non-medical costs, and indirect costs, is projected to be $79 million [93].

PD is caused by the loss of dopaminergic neurons in the substantia nigra, resulting in motor slowing, resting tremor, and rigidity. Part of the pathogenesis involves characteristic abnormal protein deposits called Lewy bodies, which mainly consist of alpha-synuclein. Alpha synuclein can exist both as an unfolded monomer or folded oligomer. The oligomeric conformation is implicated in abnormal buildup leading to pathology. Though pathophysiology of PD has been partly elucidated, the diagnosis of early PD remains a challenge given that clinical symptoms may be inconclusive, and structural neuroimaging methods such as MRI may not yet show sufficient characteristic findings. A summary of the reviewed articles relating to PD is presented in Table 7.

Table 7.

Tear film biomarker studies for Parkinson’s disease.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Schirmer strip	ELISA	84 patients, 84 controls	Oligomeric alpha-synuclein and lactoferrin were significantly elevated in PD group.	Hamm-Alvarez et al (2019) [95]
Schirmer strip	Ultra-sensitive single molecule array detection system with immunoassay	75 PD patients, 31 atypical Parkinsonian patients, 75 controls	Total alpha-synuclein was significantly elevated in PD vs control group.	Maass et al (2020) [96]
Schirmer strip	ELISA	85 patients, 80 controls	Oligomeric alpha-synuclein was significantly elevated in PD vs control group, along with CC chemokine ligand 2 (CCL-2).	Lew et al (2019) [97]
Schirmer strip	ELISA	170 patients (77 early since onset, 43 intermediates since onset, 50 late since onset), 67 controls	Oligomeric alpha-synuclein was significantly elevated in PD vs control group but no difference was found on stratification by disease duration.	Lew et al (2022) [98]
Schirmer strip	Immunoassay	49 patients, 45 controls	Total alpha-synuclein and neurofilament light chain (NfL) levels were higher in PD group.	Lin et al (2022) [99]
Schirmer strip	ELISA	26 patients, 26 controls	Total alpha-synuclein was significantly elevated in PD vs control group, with positive correlation to central volume of inner nuclear layer of retina.	Camacho-Ordonez et al (2024) [100]
Polyvinyl acetal sponge	Multiplex immunobead assay	18 patients, 17 controls	Significantly higher TNF-a was found in PD patients.	Çomoğlu et al (2013) [102]
Schirmer strip	High performance LC (HPLC) assay of catecholamines and metabolites Photometer assay of a-2-macroglobulin	Study 1: 26 patients, 19 controls Study 2: 31 patients, 32 controls	Significantly higher norepinephrine and dopamine levels and lower levels of epinephrine were found in PD. Increased a-2-macroglobulin in PD group.	Kim et al (2019) [103] and Bogdanov et al (2021) [104]
Schirmer strip	Western blot	35 patients, 35 controls	No significant difference found with levels of glypican-4.	Tatenhorst et al (2024) [105]
Schirmer strip	LC MS	36 patients, 18 controls	Full proteome analysis found 31 proteins only in PD group and 7 only in control group. They found 21 proteins significantly elevated and 19 significantly decreased in the PD group compared to control group.	Boerger et al (2019) [106]
Microcapillary tubes	nLC-MS/MS	24 patients, 27 controls	Several broad proteome alternations between PD and controls were found. Three proteins showed good capability to differentiate PD and control: CATD, ASAH1 and DYHC1, all of which involve lysosomal function.	Acera et al (2022) [107]
Schirmer strip	RNA Seq	16 patients, 16 controls	Found more relatively downregulated genes (898) compared to upregulated (238) in PD vs control. Clustering of 60 DNA repair pathway genes led to separation of PD and controls with 83.9% accuracy.	Lew et al (2023) [108]
Manual meibomian gland expression with spatula collection	NMR and infrared spectroscopy	10 patients with PD, 10 patients with PD and dry eye, 29 controls	The control group has twice as much cholesteryl esters compared to the PD group.	Blinchevsky et al (2021) [109]

An initial candidate for a biomarker is alpha-synuclein given its accumulation in the brain in PD. Much work has gone into using alpha-synuclein in the CSF as a biomarker with conflicting evidence on clinical utility [94]. For tears, the consistent finding across a range of studies is that alpha-synuclein is increased in the tears of PD, though there is nuance as to whether oligomeric alpha-synuclein is measured or total alpha-synuclein. Hamm-Alvarez et al [95] used targeted ELISA to show that oligomeric alpha-synuclein was significantly elevated in the reflex tears of PD patients, along with lactoferrin. Maas et al [96] found elevated total alpha-synuclein looking at basal tears of PD patients measured with an ultrasensitive assay. They extended their analysis by including patients with related parkinsonian spectrum diseases such as PSP, CBS, and MSA, and they did not find a significant difference between the PD group and this atypical parkinsonian group. They did not find any correlation between alpha-synuclein level and PD duration or UPDRS score. Lew at al [97] used ELISA to find increased oligomeric alpha-synuclein, along with increased CC chemokine ligand 2 (CCL-2), in patients with PD. In a second study [98], the same authors designed an experiment to specifically assess correlation between oligomeric alpha-synuclein level and time since onset of PD with a bigger cohort. They redemonstrated robustly elevated alpha-synuclein in all three duration groups against controls but found no significant difference between different durations. Lin et al [99] used immunoassays and found elevated total alpha-synuclein and neurofilament light chain in PD patients. Camacho-Ordonez et al [100] used ELISA to further find the same result, but they also looked at retinal layer thicknesses and found that higher levels of alpha-synuclein correlate with the central volume of the inner nuclear layer in PD.

One potential biomarker is TNF-a, general inflammatory mediators. In the PD context, it is released by microglia in the brain and theorized to play a role in dopaminergic neuron loss. Past research has found higher levels of TNF-a in both the brain and CSF of PD patients compared to controls [101]. Comoglu et al [102] conducted a targeted study using an immunobead assay and found significantly higher TNF-alpha in the tears of PD patients compared to healthy controls. They did not find any correlation with either PD duration or UPDRS score.

Another avenue of biomarkers is to consider the level of catecholamines, given dopamine is a catecholamine and dopaminergic neuron loss is the hallmark of PD. Kim et al [103] measured catecholamines in the tears and found significantly increased levels of dopamine and norepinephrine and lower levels of epinephrine in PD patients versus a control group. They observed that the increased dopamine and norepinephrine was more pronounced on the side of the body with more severe motor symptoms. Given there were no significant differences in volume of tears collected by the side of the body, hypokinesia of eye muscles resulting in reduced tear flow was deemed less likely to be the explanation. The same lab extended their study, with Bogdonov et al [104] using a slightly larger sample size with the same finding. In the follow up study, they also assessed for a-2-macroglobulin and found it to be increased in PD tear fluid.

Due to their role as biomarkers for other diseases such as cardiovascular disease, cancer and diabetes, glypican-4 was investigated by Tatenhorst et al [105] across a range of biofluids, including serum, CSF, and tears. They were able to measure it in all fluids collected but did not find a significant difference in amount between PD and control in any fluid.

Boerger et al [106] conducted the first shotgun proteomic study of tear proteins in PD patients. Using a bottom-up LC MS approach (BULCMS), they found 31 proteins only detected in PD patients and 7 only in the control group. They found 21 proteins significantly elevated and 19 significantly decreased in the PD group compared to control group. Using protein functional annotations, they found the differences corresponded to immune response, lipid metabolism, and oxidative stress. Acera et al [107] conducted another full proteome analysis using nLC-MS/MS. They found many network level differences, along with three proteins that showed good capability to differentiate PD and control: CATD, ASAH1 and DYHC1. All three involve lysosomal function.

In the non-protein biomarker space, Lew at al [108] conducted RNA sequencing analysis of the tears. They found more relatively downregulated genes (898) compared to upregulated (238) in PD vs control. They found DNA repair pathways were downregulated and using clustering methods with 60 DNA repair pathway genes led to separation of PD and controls with 83.9% accuracy (sensitivity 93.3%, specificity 75%). Blinchevsky et al [109] investigated meibum lipid composition in PD tears. Using NMR and infrared spectroscopy, they found that PD tears contained significantly less cholesteryl esters compared to controls, along with a significantly lower cooperativity of phase transition. Their findings suggest that altered lipid composition could play a role in the susceptibility to dry eye in PD.

5.8. Sjögren’s syndrome and Rheumatoid Arthritis

Autoimmune diseases often impact the ocular surface, leading to patients dealing with dry eyes [7]. The mechanism by which they do this is often due to excess immune and inflammatory factors causing tissue damage of the lacrimal gland, meibomian glands, cornea, and other structures [7]. More specifically, Sjögren’s syndrome (SS) and rheumatoid arthritis (RA) are two of the most common autoimmune causes of dry eye [110,111]. SS is estimated to affect up to 3.1 million adults, and many of them also have RA [112]. SS is associated with significant oral dryness, which can profoundly impact the quality of life by interfering with basic daily functions including eating, speaking, sleeping, and may also accelerate periodontal diseases. Early diagnosis is essential for management of SS, though this can be challenging since symptoms may not present concurrently [113]. With this in mind, many studies wanted to assess the tear film of patients with these conditions and see if they could find biomarkers to aid in diagnosis or monitoring. Literature review of “Rheumatoid” also pulled in studies that focus on SS. A summary of the reviewed articles relating to SS and RA is presented in Table 8.

Table 8.

Tear film biomarker studies for Sjögren’s syndrome and rheumatoid arthritis.

Tear Collection	Lab Technique	Number of patients	Results	Citation
Schirmer’s strips	2D-DIGE, NANO-LC-MS/MS	129	In DES-RA it was found that SHC transforming 1 isoform, Ribonuclease P protein subunit 20, Lactotransferrin isoform 1 precursor, Protocadherin, Heterogeneous nuclear ribonucleoprotein Q isoform 6 were significantly downregulated. Ecto-ADP ribosyl transferase 5 and Rho-related GTP-binding protein were upregulated significantly	Aluru et al. (2017) [114]
Schirmer’s strips	antibody microarray	32	S100A6 and MMP9 were significantly elevated in patients with SS. CST4 was significantly decreased in SS patients when compared to controls.	Boto de los Bueis et al. (2023) [115]
Schirmer’s strips	ELISA	156	CTSS was significantly increased in SS patients compared to all the other groups. Cys C, LF, and sIgA were significantly decreased in SS patients.	Edman et al. (2018) [116]
Schirmer’s strips	Cathepsin S activity assay kit	278	SS patients had significantly elevated CTSS activity in their tears.	Hamm-Alvarez et al. (2014) [117]
Onion vapor to capillary tubes	RP-HPLC	60	Lysozyme levels were significantly reduced in RA patients. Lactoferrin and immunoglobulin levels were significantly increased in RA patients.	Sariri et al. (2010) [118]
Schirmer’s strips	Zymography	13	The major MMPs in the tears of RA patients with ocular disease were MMP-9 and an unidentified 116 kDa species	Smith et al. (2001) [119]
Not mentioned	ELISA	25	IgG anti-α-fodrin was significantly elevated in SS patients	Yavuz et al. (2006) [120]

Aluru et al. [114] aimed to investigate any potential biomarkers in the tear fluid of patients who have DES secondary to RA. Their study consisted of 129 patients who had DES as the experimental group. The experimental group also included RA- associated DES, primary Sjögren’s syndrome, and non-Sjögren’s syndrome. The control group was made up of 73 age and sex matched controls. Reflex tears were collected using Schirmer’s strips, 2D-DIGE was used to compare tear protein expression, and NANO-LC-MS/MS identified the differentially expressed proteins. DES-RA was found to have a significantly lower tear protein level compared to the control group. The primary SS group and non-SS group also had decreased protein levels when compared to the control. The RA group without DES on the other hand had normal tear protein levels. The study mentions 7 proteins were found to be statistically significant in patients with DES-RA. Those include downregulation of SHC transforming 1 isoform, Ribonuclease P protein subunit 20, Lactotransferrin isoform 1 precursor, Protocadherin, Heterogenous nuclear ribonucleoprotein Q isoform 6 and the upregulation of Ecto-ADP ribosyl transferase 5 and Rho-related GTP-binding protein.

Boto de los Bueis et al. [115] aimed to evaluate the potential of tear biomarkers in primary Sjögren’s syndrome (SS) with a goal of seeing if these biomarkers could be useful in diagnosing and monitoring SS. The study consisted of 32 people, 22 with SS and 10 healthy controls. They collected tears from the participants and using an antibody microarray quantified the numbers of S100A6, MMP9, and CST4 specifically. They found that S100A6 and MMP9 were significantly elevated in patients with SS. Conversely, CST4 was significantly decreased in SS patients when compared to controls. They also found that S100A6 had a strong correlation with rheumatoid arthritis.

Edman et al. [116] investigated the role of protease inhibition and cystatin C (Cys C) in the tears of patients with SS leads to increased cathepsin S (CTSS). Along with its effect on Lactoferrin (LF) and secretory IgA (sIgA). The study consisted of 156 female subjects which when broken down included: 33 SS patients, 33 RA patients, 31 other autoimmune disease, 35 non autoimmune dry eye patients, and 24 healthy controls. They found that CTSS activity was significantly increased in the tears of SS patients compared to all the other groups. Cys C, LF, and sIgA levels were significantly decreased in SS patients. CTSS activity correlated inversely with Cys C, LF, and sIgA levels. This study suggest CTSS is a strong biomarker for SS, with LF and CTSS together improving diagnostic accuracy for differentiating SS from non-autoimmune dry eye.

Hamm-Alvarez et al. [117] evaluated tear CTSS activity in patients with SS as a potential biomarker. The study had 278 female subjects, including: 73 SS patients, 79 RA patients, 40 systemic lupus erythematosus (SLE) patients, 10 blepharitis patients, 31 non-specific dry eye patients, 12 other autoimmune disease patients, and 33 healthy controls. A summary of their findings is that SS patients had significantly elevated CTSS activity in their tears: 4.1-fold higher than non-SS autoimmune patients, 2.1-fold higher than non-specific dry eye patients, and 41.1-fold higher than healthy controls (p < 0.0001). The study supports further investigation into CTSS as a diagnostic tool and its possible role in disease monitoring.

Sariri et al. [118] evaluated tear proteome profiles as a potential non-invasive diagnostic tool for Rheumatoid Arthritis (RA) by analyzing differences in tear proteins and enzymatic activity between RA patients and healthy controls. The study included 60 subjects, 30 with RA and 30 healthy controls. Tear samples were collected using onion vapor. They found that lysozyme levels were significantly reduced in RA patients, while lactoferrin and immunoglobulin levels were significantly increased in RA patients. This study supports the potential use of tear analysis as a non-invasive diagnostic tool for RA, but further research is needed.

Smith et al. [119] investigated the source of MMPs in the tears of patients with peripheral ulcerative keratitis (PUK) and to investigate whether MMP accumulation in tears is specific to PUK or a feature of other anterior segment diseases. The study was done by collecting tear samples from: patients with RA with and without ocular disease, patients with various anterior segment diseases (e.g., keratoconus, herpetic eye disease, dry eye), and healthy controls. They found that the major MMPs accumulating in the tears of RA patients with ocular disease were MMP-9 and an unidentified 116 kDa species. They also found that MMP accumulation was not specific to PUK and was also found in various ocular conditions such as, keratoconus patients with atopic disease, herpetic eye disease patients, and dry eye patients with and without systemic autoimmune disease. RA patients with ocular disease had significantly higher tear MMP9 levels when compared to RA patients without ocular disease and healthy controls.

Yavuz et al. [120] aimed to evaluate the levels of IgA and IgG antibodies against α-fodrin in serum, tear fluid, and saliva of SS patients and compare them with anti-Ro/SSA and anti-La/SSB antibodies. This way to aimed to assess their diagnostic potential and correlation with disease severity. The study was conducted with 25 SS patients (17 primary, 8 secondary), 8 SLE patients, 7 RA patients, and 20 healthy blood donor controls. They collected serum, tear fluid, and saliva. Their main findings were IgG anti-α-fodrin was significantly elevated in SS patients compared to controls in all three fluids, IgA anti-α-fodrin was also elevated in SS patients which was significant in serum and saliva but not tear fluid. Lastly, Anti-Ro/SSA and anti-La/SSB antibodies were more frequently detected in SS patients than in controls.

6. Discussion

Tear fluid-based biomarkers show promise for diagnostic testing for many diseases that are not primarily ocular. The ease of collection along with distinctive composition creates potential for novel findings in the tears. Building on previous work focusing on ocular disease, the diseases to consider next include those with possible ocular manifestations (such as rheumatoid arthritis, Sjögren’s syndrome, or thyroid disease) and those involving the central nervous system (such as Alzheimer’s disease, multiple sclerosis, Parkinson’s disease, and even migraines). Investigation has already extended past these, with some initial success with biomarkers for cystic fibrosis and various cancers, especially breast cancer. Diabetes was excluded from this study partially due to the presence of diagnostic clinical ocular changes as well as the large quantity of research meriting its own meta-analysis.

The biomarkers considered thus far have primarily consisted of proteins. Proteins represent a powerful class of biomarkers given that specific proteins can be linked to the pathophysiology of disease, along with the number of possible biomarkers and improving lab techniques to quantify them with standard workflows. Furthermore, proteins have complex dynamics to their structure and post-translational modifications, and in certain cases, measuring the conformational form of the protein has been shown to be useful in addition to the total amount. Other biomarkers that hold promise in the tears and have been investigated for certain diseases include peptides, amino acids, RNA, and lipids. Going forward, the search for biomarkers can benefit from leveraging the broad diversity of molecules in the tear fluid.

The approaches taken can be divided between “targeted studies,” those with candidate molecular biomarkers known ahead of time, and “broad studies,” those that quantify all potential biomarkers in a category. The targeted approach has fewer candidates from the onset and can leverage molecule specific immunoassays in the lab. These candidates generally have previous literature describing their role in the disease, leading to better understanding of the potential biomarker.

The broad approach often involves shotgun proteomics, generally with a variant of mass-spectrometry. The study size needed for statistical significance is larger given the number of molecules being compared, and the role of the potential biomarkers in the disease may not be clear. On the other hand, the broad approach may uncover the previously unknown relevance of molecules to a disease. Additionally, “composite” biomarkers can be found, where the biomarker is not the concentration of a single molecule but rather a mathematical function of the concentration of multiple molecules. These composite biomarkers can be constructed using machine learning algorithms. The flexibility of such methods can pose the concern of overfitting to data, but they can potentially find markers of disease that would otherwise be impossible to observe.

The difficulties encountered in the search for biomarkers in the tears for systemic disease largely come from sample size, replicability, and biological understanding of the biomarker. Most of the included studies consisted of under 100 subjects. Smaller samples sizes can lead to higher variation in measurements, which may limit reproducibility of the findings. Larger study populations may be needed to show a potential biomarker has statistically significant measurements in patients with disease compared to control. The population size needed goes up with the number of individual biomarkers being tested in a single study to account for false discovery rate. Recognizing this limitation, many previous studies characterize themselves as “pilot studies”, showing a proof of concept that can be conducted more rigorously with larger future studies.

The second major challenge is reproducibility. A review of the literature shows some cases where studies examining a protein as a biomarker in a disease show consistent results (such as alpha-synuclein in Parkinson’s disease), but in other cases, different studies find disjoint sets of candidate biomarkers for the same disease or are unable to find the same protein’s significance. For example, two studies [83,85] found IL-10 elevated in patients with thyroid eye disease compared to controls, whereas another found it relatively lower [88]. Some factors that may explain the differences seen are limited study sizes, differences in tear collection, and varied lab processing techniques. Analyses are typically based on the use of basal tears, but some studies use reflex tears. Tear collection is usually done with Schirmer strips or glass capillary tubes, but other methods are used as well. Protein processing is generally done with LC-MS, but many variants are used as they may obtain more complete sets of protein measurements. Each of these decisions can lead to differences in results. Future work explicitly comparing the impact of these factors on measurements and suggesting best standardized practices will help improve replicability of these experiments.

A third challenge is biological understanding of potential biomarkers. Broad approaches can uncover many candidate proteins, but lack of detailed understanding of their relation to the disease can make it difficult to select which to prioritize. Functional annotation databases provide a starting point, but basic science research can help to further elucidate the role of a biomarker in the disease. This challenge presents the possibility of synergistic work between the clinical search for biomarkers creating candidate molecules for basic science investigation to better understand the pathophysiology of disease.

A useful two-tier paradigm employed in certain studies divides the study population into a discovery group and a validation group [45,55]. A broad approach is taken to identifying potential biomarkers in the discovery group, generally employing mass-spectrometry without a predefined set of molecules. The data from this group, along with functional annotations of the proteins and existing literature on the disease, is used to create a short list of candidate biomarkers. A targeted approach using these candidates is taken with the validation group. The biomarker quantification can be done with a more robust immunoassay specifically measuring the biomarker rather than mass spectrometry. An additional advantage is that immunoassays are easier to conduct on larger groups of patients. This two-tier simultaneously conducts a broad search of biomarkers while managing to show statistical significance, replicability, and potential clinical utility.

7. Conclusion

The tear fluid represents an exciting avenue for biomarker discovery for a range of diseases including Alzheimer’s disease, cancers, cystic fibrosis, migraine, multiple sclerosis, Parkinson’s disease, rheumatoid arthritis, and thyroid disease. Much work has been done comparing the concentrations of proteins and other molecules in patients with disease and controls with many potential biomarkers identified. Current challenges include limited sample size and reproducibility. Future work will focus on larger study populations, replicable tear collection, and lab techniques, and understanding the biological correlation of the identified biomarkers. With these additions, the tear-fluid may show utility for both clinical diagnostics and the improved understanding of human disease.

Highlights:

• Human tears have a rich proteome, with over 1500 expressed proteins identified so far

• 54 articles were selected for final analysis

• Tear fluid-based biomarkers show promise for diagnostic testing for many diseases

• Current challenges include limited sample size and reproducibility