Human tear serotonin levels correlate with symptoms and signs of dry eye

Abstract

Purpose

Serotonin, a neurotransmitter known to be involved in nociceptor sensitization, is present in human tears. The purpose of this study was to correlate tear serotonin levels, as a marker of nociceptor sensitization, to facets of dry eye (DE) including symptoms and signs.

Design

Cross-sectional study

Participants

Sixty-two patients with normal eyelid and corneal anatomy were prospectively recruited from a Veterans Administration Ophthalmology Clinic over 11 months.

Methods

DE symptoms (Ocular Surface Disease Index [OSDI]), signs (tear break-up time [TBUT], corneal staining, and Schirmer’s score), and clinical descriptors of neuropathic ocular pain (NOP) (sensitivity to light and/or sensitivity to wind) were assessed. For tear analysis, each patient’s tears were collected after instilling 50µl of sterile saline to the lower cul-de-sac of each eye and using capillary action microcaps to collect the ocular wash. Tear serotonin levels were measured using enzyme-linked immunosorbent assay.

Main Outcomes Measured

Correlations between tear serotonin concentrations and DE symptoms and signs.

Results

The mean age of the population was 61±14 years and 84% (n=52) of the patients were male. Serotonin concentrations negatively correlated with Schirmer’s scores (r=−0.28; p=0.02), but did not correlate with other DE parameters, such as OSDI scores, sensitivity to light or wind, TBUT, or staining. According to our hypothesis, we divided patients into groups based on both DE symptoms and aqueous tear production; serotonin concentrations were found to be significantly higher in DE group 1 (OSDI≥6 and Schirmer’s<8) compared to both DE group 2 (OSDI≥6 and Schirmer’s≥8) and controls (OSDI<6 and Schirmer’s≥8). Patients in the DE group 2 more frequently complained of sensitivity to light (64%) and wind (67%) compared to the DE group 1 (40% and 60%, respectively) and controls (8% and 17%, respectively).

Conclusion

Patients with DE symptoms and aqueous tear deficiency had higher tear serotonin levels compared to those with DE symptoms but normal tear production and those without DE symptoms.

Introduction

Dry eye (DE), a multifactorial disease of the ocular surface and tear film, affects millions of men and women in the United States.¹ It is characterized by decreased quantity and altered quality of tears, abnormal ocular surface, and bothersome symptoms. These symptoms include pain, blurry vision, and tearing, all of which adversely impact quality of life.^{2, 3} Despite their morbidity, the pathophysiology underlying DE symptoms, especially ocular pain, is not well understood.⁴ For example, DE symptoms (including ocular pain) do not correlate well with tear film parameters.^{5, 6} In a prospective study of 263 veteran men, less than 10% of the variability in symptoms was explained by measured tear film parameters.⁵ This suggests that factors beyond tear film health are likely important in driving symptoms in certain individuals. It is not surprising therefore, that some patients have persistent symptoms while on available DE therapies,⁷ most of which target the ocular surface and tear film deficiency.

A potential factor that may underlie DE-associated chronic pain is dysfunction of the ocular somatosensory nerves. There is a growing understanding that many patients diagnosed with DE describe features of neuropathic pain,⁸ defined as pain resulting from dysfunction of nerves. For example, neuropathic pain and DE share many common descriptors, including spontaneous pain in the absence of any apparent noxious stimulation, dysesthesias, hyperalgesia, and allodynia.^9–11 Neuropathic pain can result from damage or sensitization of peripheral and/or central somatosensory nerves. With respect to ocular anatomy, this would represent changes in primary afferents^{9, 12, 13} (peripheral sensitization), and/or changes in the connecting 2^nd and 3^rd order neurons in the spinal trigeminal nuclei, subcortical and cortical neuronal processes (central sensitization).^{9, 12}

Based on anatomy, it is biologically plausible that phenotypic alterations and/or peripheral sensitization of ocular somatosensory nerves could occur as a component of DE in some patients. Free nerve endings interdigitate between superficial epithelial cells and are therefore vulnerable to repeated damage from environmental exposures,^{11, 14, 15} including contact with sensitizing molecules, such as inflammatory mediators. For example, T cells, interleukins, tumor necrosis factor and matrix metalloproteinase have all been found on the ocular surface and tears of humans and animals with DE.^16–19

The presence of inflammation has been shown to affect the function of peripheral nerves in animal models both directly and indirectly, through increased expression of serotonin, its receptors, and other sensitizing neuropeptides.^20–24 In one study, rat hind paws injected with formalin induced peripheral inflammation, increased serotonin release from mast cells, and sensory sensitization.²⁵ In another model, subcutaneous serotonin injections themselves led to pain (detected by intense flinching and licking behaviors in rats), providing evidence that serotonin is an important peripheral pain mediator.²⁶ In the trigeminal system, peripheral 5-HT₃ and 5-HT_2A receptors were found to mediate pain after formalin injections into masseter muscles.^{27, 28} The subsequent occurrence of central sensitization has been well described in the setting of ongoing afferent nociceptive traffic seen with peripheral sensitization.^{29, 30}

In summary, sufficient evidence supports the notion that peripheral serotonin, acting via overexpressed serotonin receptors in primary afferents, contributes to peripheral sensitization and amplification of pain signaling.²⁰ As serotonin has been previously detected in human tears,³¹ we hypothesized that tear serotonin levels may act as a surrogate marker for corneal nociceptor sensitization, and therefore could be associated with specific DE subtypes. In this study, we tested our hypothesis that tear serotonin levels would be higher in those with abnormal versus healthy tear parameters.

Methods

Study Population

After approval by the Miami VA Institutional Review Board/Ethics Committee, patients were prospectively recruited from the Miami Veterans Administration Eye Clinic between March 2013 and February 2014. Each patient underwent a complete ocular surface examination, and patients with normal eyelid and corneal anatomy were included (n=62). Exclusion criteria included contact lenses wear, ocular medications (with the exception of artificial tears), active external ocular process, history of refractive or retinal surgery, history of glaucoma, and cataract surgery within the last 6 months. Additionally, patients with human immunodeficiency virus, sarcoidosis, graft-versus host disease, collagen vascular disease, or other inflammatory conditions were excluded. Based on these inclusion and exclusion criteria, patients with a wide range of dry eye symptoms (none to severe) and signs (none to severe) were included, including those with evidence of aqueous and evaporative tear deficiency.

Data Collection

All patients were examined by the same optometrist (ALM). Demographic and health information was collected for each patient, including age, gender, race, ethnicity, health status, and psychiatric history including diagnosed depression or anxiety disorder, use of antidepressant and antianxiety medications. DE symptoms and severity were assessed via the Ocular Surface Disease Index (OSDI).³² Clinical features of neuropathic ocular pain (NOP) were assessed by using sub-scores on the OSDI, specifically question 1 (Have you experienced eyes that are sensitive to light during the last week?) and question 10 (Have your eyes felt uncomfortable in windy conditions during the last week?). Additionally, ocular surface examination included (1) tear breakup time (TBUT) (5 µl fluorescein placed in the eye; 3 measurements taken in each eye and averaged), (2) corneal staining (National Eye Institute [NEI] scale) (5 areas of cornea assessed; score 0–3 in each), and (3) Schirmer’s strips with anesthesia.

Tear Collection

Fifty microliters of sterile saline was instilled by a pipette to the lower cul-de-sac of each eye. Tears were then immediately collected by capillary action using 1 microliter microcaps (Drummond Scientific Company, Broomall, PA, USA) applied gently to the lower temporal lid margins. A minimum of 50 microliters of tears (approximately 6 disposable micro-pipettes) was collected and pooled from the two eyes and released by bulb dispenser into 1.5-milliliter Nalgene polypropylene cryogenic vials (Sigma-Aldrich Co. St Louis, MO, USA). These vials were labeled with de-identified subject codes, and immediately placed in a -80 degree Celsius freezer.

Quantification of serotonin

Enzyme-linked immunosorbent assay (ELISA) was performed on the tear samples using an ELISA Serotonin Kit (Enzo Life Sciences, Inc. Farmingdale, NY, USA). This kit detects a minimum serotonin concentration of 0.30ng/mL, and previous research has shown measurable levels of serotonin in tears ranging from 0.12ng/ml to 50.32ng/ml with a mean of 2.74ng/ml.³¹ The negative control was sample without tears (sterile saline; no serotonin present), and the positive control was the serotonin standard (provided by the kit). Equivalent volumes of each sample were tested, and to ensure precision and reproducibility, each sample was run in duplicate and the average of the duplicates was taken. The serotonin concentration was calculated using the average optical density of each sample, and control sterile saline samples were negative for serotonin. After plotting the percent bound versus serotonin concentration for the standards, the serotonin concentration of our samples was interpolated, according to the kit’s directions. Samples were prospectively collected over a period of 11 months and thus had different storage times; regression analysis indicated that time in storage did not correlate with serotonin levels.

Statistical Analysis

All statistical analyses were performed using SPSS Version 22 (SPSS, Chicago, IL, USA) statistical package. The strength of association between tear serotonin concentrations and DE signs and symptoms were quantified using Pearson and Spearman coefficients. Analyses also included Fischer exact for nominal variables and student’s independent t-test, Mann-Whitney U, and analysis of variance (as appropriate) for continuous variables. Multivariable linear regression analyses were performed to evaluate the effect of various factors or serotonin levels. All analyses were performed using DE signs from the more severely affected eye. A p-value less than 0.05 was considered statistically significant.

Results

Study population

The mean age of the population was 61, with a standard deviation (SD) of 14 years; 84% (n=52) of the patients were male, 65% (n=40) self-characterized themselves as white, and 34% (n=21) as Hispanic. Serotonin concentrations were significantly associated with ethnicity but not with other demographic characteristics (age: Pearson r=0.14, p=0.26; Spearman rho=0.23, p= 0.07) (Table 1). Of the 62 patients included in this study, 58 had a psychiatric history available. Sixteen (28%) had depression, and of those 10 were taking antidepressants; 10 (17%) had anxiety, and of those 7 were taking anxiolytics. Mean serotonin concentrations were not different in those with and without a psychiatric history, nor were they different in those who were or were not taking antidepressant and anxiolytics. There were also no significant differences in OSDI, Schirmer’s scores, or DE subgroups among those taking and not taking antidepressants and anxiolytics.

Table 1.

Mean serotonin levels by patient demographics and clinical features

Demographics		Serotonin concentration (ng/ml), mean±SD	p-value
Race	White	1.58±0.98	0.66
	Black	1.73±0.94
	Other	2.01±1.13
Gender	Male	1.72±0.97	0.18
	Female	1.28±0.91
Ethnicity	Hispanic	1.24±0.77	0.02
	Non-Hispanic	1.86±0.99
Co-morbidities
Depression	Yes	1.74±0.95	0.51
	No	1.56±0.88
Anxiety	Yes	1.38±0.87	0.48
	No	1.64±0.90
Medication use
Antidepressants	Yes	1.74±0.92	0.61
	No	1.58±0.90
Anxiolytics	Yes	1.60±1.00	0.97
	No	1.61±0.89

Correlations between serotonin concentration and dry eye symptoms and signs

Serotonin concentrations negatively correlated with Schirmer’s scores (r=− 0.28; p=0.01), but did not correlate with OSDI scores, sensitivity to light or wind, TBUT, or staining (Table 2). A multivariable linear regression analysis considering both ethnicity and Schirmer scores found that both remained significantly associated with serotonin levels.

Table 2.

Correlations between serotonin levels and dry eye symptoms and signs (more severely affected eye)

Parameter	Pearson r	p-value	Spearman rho	p-value
OSDI total	−0.02	0.88	0.002	0.99
OSDI question 1: Light	0.007	0.96	0.008	0.95
OSDI question 10: Wind	−0.11	0.37	−0.09	0.49
Tear break-up time	0.04	0.73	0.1	0.42
Corneal staining	0.09	0.48	0.1	0.43
Schirmer score	−0.28	0.02	−0.22	0.08

Serotonin concentration by dry eye subgroups

Overall, patients with aqueous tear deficiency (ATD) (Schirmer’s score < 8) had higher tear serotonin levels compared to those without ATD (2.34 ng/ml ±1.16 versus 1.53 ng/ml ±0.89, p=0.02) (Table 3). Serotonin concentrations were significantly higher in the DE group 1 (OSDI≥6 and Schirmer’s<8) compared to both the DE group 2 (OSDI≥6 and Schirmer’s≥8) and the non-DE control group (OSDI<6 and Schirmer’s≥8) (Table 3). A multivariable linear regression analysis considering both ethnicity and the above 3 groups found that only ethnicity remained significantly associated with serotonin levels.

Table 3.

Tear serotonin concentrations, dry eye symptoms and signs (more severely affected eye) in dry eye sub-groups

	Control group (OSDI<6 and Schirmer≥8)	DE 1 (OSDI≥6 and Schirmer<8)	DE 2 (OSDI≥6 and Schirmer>8)	p value
n	12	5	45
Serotonin (mean [ng/ml] ±SD)	1.72±1.07	3.03±0.76	1.48±0.84	0.005a
Symptoms
OSDI Q1: Light≥0 (%, n)	8%, 1	40%, 2	64%, 29	<0.0005
OSDI Q10: Wind≥0 (%, n)	17%, 2	60%, 3	67%, 30	0.001
OSDI Q1: Light (mean±SD)	0.33±1.2	1.0±1.7	1.5±1.5	0.04b
OSDI Q10: Wind (mean±SD)	0.33±0.89	1.0±1.2	1.6±1.6	0.01b
OSDI total (mean±SD)	1.5±1.4	18.1±11.5	27.7±19.5	<0.0005
Signs
Tear Break-Up Time (mean [sec] ±SD)	8.4±2.9	6.3±1.7	7.4±2.8	0.47
Corneal staining (mean±SD)	2.2±2.2	4.2±3.3	3.1±2.9	0.45
Schirmer score (mean [mm] ±SD)	17.8±5.7	4±2.3	15.5±5.0	<0.0005a

DE=dry eye; OSDI=ocular surface disease index; SD=standard deviation; Q=question

^aSignificant differences between DE1 as compared to control and DE2 by least significant difference sub-analysis

^bSignificant difference between DE2 and control by least significant difference sub-analysis

NOP features by study populations

Demographic characteristics did not influence the frequency of NOP features, but patients with a diagnosis of depression and/or anti-depressant use had a higher frequency of sensitivity to wind (depression: 80% (n=12) versus 49% (n=19), p=0.04; antidepressant use 100% (n=9) versus 49% (n=22), p=0.01). Patients in DE group 2 had the highest frequency of sensitivity to wind and light, compared to DE group 1 and controls (Table 3).

Discussion

The results of this study show that patients with DE symptoms and ATD have higher tear serotonin levels compared to those with DE symptoms and normal tear production and those without DE symptoms. On the other hand, patients with DE symptoms and normal tear production more frequently endorsed wind and light sensitivity than those with DE symptoms and reduced tear production and those without DE symptoms.

We chose to measure tear serotonin concentrations on the ocular surface as a potential marker of corneal nociceptor sensitization since serotonin has been previously detected in human tears,³¹ is a known peripheral nerve sensitizer,^{20, 33} and its levels are known to increase not only with inflammation (a well described component of DE), but also with peripheral nerve abnormalities.^{20, 26, 34, 35} Using high-performance liquid chromatography with electrochemical detection, the average serotonin concentration in tears in a previous study of 22 volunteers (64% female) with an average age of 36.5±9 years was found to be 2.74±1.99 ng/ml.³¹ In our older, mostly male population, average serotonin concentrations were slightly lower using an ELISA methodology.

Serotonin produces its sensitizing effect on peripheral nerves through direct and indirect mechanisms.²⁰ Serotonin binds to G protein-coupled receptors, mobilizes downstream slow modulatory responses through second messenger signaling pathways³⁶ with activation of ligand-gated ion channels and rapid depolarizing responses in nerves.³⁷ Serotonin may also directly amplify the conductance of tetrodotoxin-resistant sodium channels in primary afferents, may shift the conductance-voltage curve towards a hyperpolarized direction, and enhance the channel kinetics.³⁸ This subsequently has the capacity to affect the threshold of activation, sensitivity, and nociceptive transmission of peripheral nerves.^{21, 22} Similarly, 5-HT₃ receptor stimulation enhances perception of pain.³⁹

To our knowledge, this study is the first to correlate serotonin levels with DE parameters. Others, however, have examined the relationship between other peripheral nerve sensitizers, such as nerve growth factor (NGF), and DE.^{40, 41} Serotonin may act synergistically together with several neuropeptides and inflammatory mediators, as a component of the “inflammatory soup,”²⁰ contributing to neuronal peripheral sensitization and hyperexcitability.²⁴ This has been shown in humans, wherein serotonin may act synergistically with bradykinin in producing pain and hyperalgesia.^{20, 42}

Understanding the role of peripheral sensitizers in DE may open up new therapeutic pathways. For example, serotonin receptor antagonists administered systemically or topically such as ketanserin (5-HT_2A antagonist) and granisetron (5-HT₃ antagonist) as well as cromolyn (prevents the degranulation of mast cells) may have therapeutic potential in some dry eye patients. Ketanserin and granisetron, for instance, have been shown to have a peripheral anti-hyperalgesic effects, the latter in trigeminal system pain in humans.^{20, 43, 44}

As with all studies, our findings must be considered in light of our study limitations. First, our study was conducted at a Veterans Affair Hospital and therefore our population consists of predominantly older males. We would argue, however, that this group is an important one in which to study DE because of the high incidence of central pain disorders and depression in this population. Second, our cross-sectional study design precludes commenting on persistence or change in DE symptoms, signs, serotonin concentration, and NOP features over time. Third, we recognize that serotonin levels and patients’ self-reported responses on questionnaires are not the same as direct assessments of ocular somatosensory nerve function. However, direct nerve conduction studies on the cornea are not currently available and therefore measures are needed to provide indirect information on this understudied parameter of DE. Fourth, it is not clear why other aspects of tear dysfunction (corneal staining and TBUT) were not significantly associated with serotonin. Further studies, with increased numbers and in other DE populations, will be needed to replicate our findings and assess which parameters of tear dysfunction, including ones not measured herein, most closely align to serotonin concentrations. Finally, we did not measure inflammation on the ocular surface, nor did we measure other sensitizers previously described in DE such as NGF. As such, we cannot evaluate the link between serotonin, inflammation, and other nerve sensitizers.

Despite these limitations, we found that patients with DE symptoms and ATD had higher levels of serotonin on the ocular surface. Previous studies have found associations between low tear levels and inflammation,^{45, 46} thus supporting our theory that ocular surface inflammation can increase local levels of neurotransmitters and thus modulate neuronal function. Interestingly, patients with DE symptoms but normal tear production had serotonin levels similar to patients without DE symptoms, but tended to describe a higher frequency of NOP complaints, suggesting the presence of central abnormalities in this sub group of patients. The implication of this study supports a disease model where there is a connection between ocular surface stress, inflammation, and peripheral sensitization of ocular somatosensory nerves in some patients with DE. Moreover, our data suggest that still other patients may have developed central sensitization with neuropathic pain like symptoms even in the absence of ongoing ocular surface abnormalities. Identifying novel biomarkers and developing new therapeutics to target the underlying pathology in each case to better manage these different DE patients will likely be an important area of research in the future.