Changes of Lacrimal Gland and Tear Inflammatory Cytokines in Thyroid-Associated Ophthalmopathy

Abstract

Purpose.: To explore changes in lacrimal gland and tear inflammatory cytokines in thyroid-associated ophthalmopathy (TAO) patients.

Methods.: Patients with TAO were divided into active and inactive TAO groups. These two TAO groups and the control completed the Ocular Surface Disease Index (OSDI), underwent thorough ophthalmologic examinations, and underwent orbital magnetic resonance scan to measure the size of the lacrimal gland. Basal tears, reflex tears induced by nasal stimulation and serum samples, were collected to analyze the concentrations of interleukin (IL)-1β, IL-6, IL-7, IL-17A, interferon γ, and tumor necrosis factor α by multiplex bead analysis.

Results.: The coronal lacrimal gland area was significantly larger in active (P < 0.000) and inactive TAO (P = 0.002) than in the control, and the axial lacrimal gland area was significantly larger in active (P < 0.000) and inactive TAO (P = 0.001) than in the control. The coronal lacrimal gland width was significantly greater in active (P < 0.000) and inactive TAO (P = 0.001) than in the control, and axial lacrimal gland width was significantly greater in active (P < 0.000) and inactive TAO (P = 0.035) than in the control. In TAO patients, the axial area was positively correlated with IL-1β and IL-17A concentrations in tears (r = 0.357, P = 0.013; r = 0.359, P = 0.012), and both coronal and axial areas were positively correlated with IL-6 concentrations in tears (r = 0.346, P = 0.016; r = 0.340, P = 0.018).

Conclusions.: Increased inflammatory cytokines play an important role in ocular surface damages, and might be associated with the inflammatory involvement of the lacrimal gland.

Introduction

Thyroid-associated ophthalmopathy (TAO), also known as Graves' ophthalmopathy, is an inflammatory autoimmune disorder of the orbit. ¹ Patients with TAO often complain of ocular surface discomforts, including photophobia, excess tearing, grittiness, and foreign-body sensations, which are ascribed to the changes in anatomic parameters of the ocular surface, especially the vertical palpebral fissure width. ^2–4 In these earlier studies, it was suggested that the change in vertical palpebral fissure width accelerates tear film evaporation, thereby increasing tear film osmolarity with resultant ocular surface damage. 

Further evidence, however, has supported lacrimal gland involvement in the pathogenesis of ocular surface damage in TAO patients. First, clinical and radiological studies reported lacrimal gland enlargement in TAO. ^5,6 Specifically, Harris ⁷ investigated 128 adult Caucasian TAO patients and concluded that the lacrimal gland is statistically significantly enlarged in TAO. Meanwhile, laboratorial evidence of lacrimal expression of the thyroid stimulating hormone receptor (TSH-R) suggests involvement of TSH-R-specific autoantibodies, which then contribute to lacrimal gland impairment and ocular surface damage. ⁸ Based on the previous pathological observation on lacrimal glands in active TAO, ⁹ this lacrimal gland involvement is actually lymphocyte infiltration and interstitial edema, which is a sign of inflammation. 

From another perspective, lacrimal gland inflammation can be reflected not only in enlarged or ill-defined imaging, but also in the increased production of proinflammatory cytokines ^10,11 and proteomic changes in tears. ^12,13 Our previous study ¹¹ showed that several inflammatory cytokines increase in the tears of TAO patients, which indicated that orbital inflammation may be involved in the ocular surface damage of TAO. The possibility of lacrimal gland involvement is even higher when considering the altered regulation of protective proteins secreted specifically from lacrimal gland. ^12,13  

Few studies have looked into the connection between anatomic changes, lacrimal gland involvement, and the increase in proinflammatory cytokines. Our study tried to investigate this possible connection, and explore the role of the lacrimal gland and tear inflammation in ocular surface damage among TAO patients. 

Methods

Study Subjects

The study involved 24 TAO patients (48 eyes) and 16 age-matched normal control subjects (32 eyes). Written informed consent and institutional review board approval were obtained. The diagnoses of TAO were based on Bartley's criteria. ¹⁴ The inflammatory activity in TAO patients was evaluated using the seven-point modified formulation of the clinical activity score (CAS), as previously published. ¹⁵ A CAS ≥ 3/7 was defined as active TAO, and CAS < 3/7 was defined as inactive TAO. All patients were in a euthyroid state with normal serum concentrations of fT3, fT4, and TSH. Patients were excluded if they had suffered from additional diseases that could affect tear constitution and ocular surface condition, such as dry eye syndrome, allergic ocular surface disease, pterygium, diabetes, rheumatologic or respiratory diseases, and any other significant diseases. They were also excluded if they had received medications or treatments such as eye drops, contact lenses, steroids, radiation therapy, and ophthalmic surgery. 

Clinical Examination

Patients with TAO and control subjects underwent ophthalmologic examinations, including visual acuity, slit-lamp examination, fundus examination, tear film break-up time, and the Schirmer test with topical anesthesia. Ocular Surface Disease Index (OSDI) symptom questionnaires were calculated as previously reported, with scores ranging from 0 to 100. ¹⁶ After TAO patients gave data on their chief complaint, detailed present history, and history of treatment and other diseases, they underwent measurements of vertical palpebral fissure width and lagophthalmos using a ruler vertically placed in the middle point of the lids, examinations of lid lag and trichiasis, and evaluations of CAS, ¹⁵ and fluorescein staining (Fl) scores. ¹⁷ An Fl score was the sum of grading scores in five corneal regions, ranging from 0 to 15 scores. The division of the corneal surface and the standardized grading system of 0 to 3 was shown in Figure 1. ¹⁷  

Figure 1

 

The measurement of Fl. A standardized grading system of 0 to 3 is used for each of the five areas on each cornea.

Orbital Magnetic Resonance Examination and Measurement

Magnetic resonance imaging (MRI) of the orbit was performed on a 1.5-tesla unit system (Echostar Centauri; AllTech Medical Systems, Co. Ltd., Chengdu, China). Both coronal and axial spin-echo T1-weighted images (TR/TE = 530–653/12 ms, 1–2 excitations) were obtained. The field of view was 200 mm, the matrix was 256 × 256, the slice thickness was 3 mm, and the slice gap was 0 mm. Images were imported into a professional Digital Imaging and Communications in Medicine viewer (Onis 2.4; DigitalCore Co., Ltd., Tokyo, Japan) for measurements. 

Measurement of proptosis was performed precisely on axial T1-w sequences referring to Kirsch's method. ¹⁸ Measurement of the lacrimal gland was performed on both coronal and axial T1-w sequences with reference to Harris' and Tamboli's methods. ^7,19 As the palpebral and orbital lobes are difficult to distinguish on MRI, the lacrimal gland was treated as one structure. In both the coronal and axial series, the image in which the lacrimal gland appeared the largest was preliminarily chosen. This image, together with the images before and after it, were used to measure the area of the lacrimal gland by manually enclosing it using the polygon tool in the software. The largest measurement results from these three images determined the final chosen image. 

Beside the largest lacrimal gland areas, two other measurements were made on the bilateral sides of the final chosen image. In the coronal images, the length of the lacrimal gland was measured from the superior tip to the inferior tip of the lacrimal gland area. The width of the lacrimal gland was measured from the lateral edge to the medial edge of the lacrimal gland at its widest point perpendicular to the length line (Fig. 2). In the axial images, the length of the lacrimal gland was measured from the most anterior tip to the most posterior tip. The width was measured from the lateral edge to the medial edge of the lacrimal gland at its widest point perpendicular to the length line (Fig. 3). Thus, there were six measurements performed on each eye, including three measurements each on the coronal and axial images. 

Figure 2

 

Coronal MRI showing the area of the lacrimal gland measured using the polygon tool, the length measured from the superior tip of the area to the inferior tip, and the width measured at the widest bilateral points perpendicular to the length.

Figure 3

 

Axial MRI showing the area of the lacrimal gland measured using the polygon tool, the length measured from the most anterior tip to the most posterior tip, and the width measured at the widest bilateral points perpendicular to the length.

Tear and Serum Sample Collection

Basal tear samples were collected from remaining tears in conjunctival sac as previously described ²⁰ using disposable 2-μL microcapillaries (Microcaps 2 μL; Drummond Scientific, Broomall, PA, USA). Care was taken to minimize ocular surface contact during basal tear sample collection. A total of 6 μL of tears was obtained and put into 0.5-mL tubes (Eppendorf, Fremont, CA, USA), and stored at −80°C after centrifugation (9,245g, 5 minutes). 

After basal tear collection, reflex tearing was induced by nasal stimulation using a dry cotton-tipped applicator as described by Solomon et al. ²¹ One minute after stimulation, the reflex tear fluid was collected and stored using the same method as for basal tear sample collection. The tear samples were never pooled together during the study. Peripheral blood samples were also collected, and the separated serum samples were stored in aliquots at −80°C until assayed. 

Multiplex Analyses of Cytokines in Tears and Serum

For the multiplex analyses, tear samples were diluted 10 times with assay buffer. The concentrations of cytokines (interleukin [IL]-1β, IL-6, IL-7, IL-17A, interferon [IFN]-γ, and tumor necrosis factor [TNF]-α) in tears and serum were analyzed using a commercial assay system of immunoassay kits and panels (MILLIPLEX Human Cytokine/Chemokine Panel [MPXHCYTO-60K]; Millipore Corp., Billerica, MA, USA). The standard curves of known concentrations of recombinant human cytokines were used to convert fluorescence units to concentrations (pg/mL). To calculate the molecular concentrations in tears and serum samples, we analyzed the median fluorescent intensity data using a five-parameter logistic. The cytokine concentrations in each original tear sample were 10 times the concentrations in the diluted sample. 

Statistical Analysis

The normality of all the numerical data was checked using the Shapiro-Wilk test (significance level α = 0.10). The comparisons involving clinical examinations, measurements of lacrimal gland, and cytokine concentrations in tears and serum among active TAO, inactive TAO, and control subjects were performed using one-way analysis of variance followed by the LSD post hoc test, or the Kruskal-Wallis test followed by the Mann-Whitney test, depending on the test of normality and homogeneity of variance of the data. The Student's t-test or Mann-Whitney test was used when comparing data between two independent groups, such as active and inactive TAO, and the paired t-test or paired Mann-Whitney test was used when comparing data between two related groups, such as the basal tears and reflex tears of each eye, according to the test results for the normality of the data. Correlations among clinical examinations, measurements of lacrimal gland, and cytokine concentrations in tears and serum were determined using Pearson or Spearman rank correlation, as well as the test of normality. A value of P < 0.05 was considered to be statistically significant. 

Results

Demographic and Clinical Data

The study included 24 TAO patients with a mean age of 40.50 ± 11.096 years (eight males, 16 females) and 16 control subjects with a mean age of 40.75 ± 11.947 years (eight males, eight females). The demographic and clinical data for the two TAO groups are presented in Table 1. Measurements of MR proptosis, Fl scores, and lagophthalmos were significantly higher in active than inactive TAO. The results of the ocular surface examinations for the three groups are presented in Table 2. The OSDI scores were higher and the Schirmer test scores were lower in both active and inactive TAO than the control, as well as in active TAO in comparison with inactive TAO. Tear film break-up time scores were lower in both active and inactive TAO than the control. 

Table 1

 

Comparisons of Demographic and Clinical Data Between the Two TAO Groups

	Active TAO	Inactive TAO	P
Sum of eyes, n (right/left)	27 (13/14)	21 (11/10)	\
Age, y	42.69 ± 11.593*	37.91 ± 10.406*	0.303
	42.71 ± 12.238†	37.40 ± 8.947†	0.256
Sex, male/female	4/9*	4/7*	1.000
	5/9†	3/7†	1.000
CAS	4 (3–7)	1 (0–2)	\
MR proptosis, mm	18.04 (13.53–28.72)	15.51 (12.26–25.43)	0.019
Fl score	3 (0–8)	0 (0–4)	0.003
Palpebral fissure width, mm	10 (7–13)	10 (8–15)	0.966
Lagophthalmos, mm	2 (0–6)	0 (0–3)	0.013
Lid lag, ±	21/6	14/7	0.516
Trichiasis, ±	8/19	4/17	0.510

 The slashes indicate no comparability.

*  Demographic data on age and sex were determined with data from right eyes.

†  Demographic data on age and sex were determined with data from left eyes.

Table 2

 

Comparisons of Ocular Surface Examinations Among the Three Groups

	OSDI Score	TBUT, s	Schirmer Test, mm/5 min
Active TAO	48.38 ± 19.82	1.58 (0.77–4.07)	5.5 (1.0–17.0)
Inactive TAO	26.32 ± 11.84	1.93 (0.88–4.82)	8.0 (3.0–14.0)
Control	6.88 ± 1.85	12.52 (10.09–19.16)	12.0 (10.0–20.0)
P	<0.000*	<0.000†	<0.000‡

*  Active versus inactive TAO: P < 0.000; active TAO versus control: P < 0.000; inactive TAO versus control: P < 0.000.

†  Active versus inactive TAO: P = 0.205; active TAO versus control: P < 0.000; inactive TAO versus control: P < 0.000.

‡  Active versus inactive TAO: P = 0.004; active TAO versus control: P < 0.000; inactive TAO versus control: P < 0.000.

Features of the Lacrimal Gland Measurements

We compared the lacrimal gland measurements on the axial images and the coronal images and found that the values of the area (P < 0.000) and the length (P < 0.000) in the coronal images were higher than in the axial images; meanwhile, the widths in the coronal images and the axial images showed no difference in the TAO group and the control (P = 0.442). The bilateral measurements showed no difference except the axial widths in the control group, which showed a marginally significant difference between the bilateral sides (P = 0.05). 

In the TAO group, negative correlations were found between age and lacrimal gland length on the two different scan planes of both sides (coronal right side r = −0.451, P = 0.027; coronal left side r = −0.407, P = 0.049; axial right side r = −0.601, P = 0.002; axial left side r = −0.510, P = 0.011). However, in the control, no correlation was found between the age and lacrimal gland measurements on the two different scan planes for both sides (P > 0.05). 

Lacrimal Gland Measurements in Three Groups and Correlative Factors

The coronal lacrimal gland area was significantly larger in active (P < 0.000) and inactive TAO (P = 0.002) than the control; meanwhile, it was larger in active than inactive TAO (P = 0.018). The axial lacrimal gland area was significantly larger in active (P < 0.000) and inactive TAO (P = 0.001) than the control; meanwhile, it was larger in active than inactive TAO (P = 0.001). Coronal lacrimal gland width was significantly wider in active (P < 0.000) and inactive TAO (P = 0.001) than in the control. Axial lacrimal gland width was significantly wider in active (P < 0.000) and inactive TAO (P = 0.035) than the control; meanwhile, it was wider in active than inactive TAO (P = 0.044). No significant difference was found in lacrimal gland coronal lengths (P = 0.068) and axial lengths (P = 0.281) among the three groups (Figs. 4, 5). 

Figure 4

 

Comparisons of the lacrimal gland measurements among active TAO, inactive TAO, and control. Areas are expressed in cm². Length and width are expressed in mm. *P < 0.05.

Figure 5

 

Examples of the lacrimal gland areas of active TAO, inactive TAO, and control.

In TAO patients, the coronal area, axial area, and axial width were positively correlated with the CAS (Fig. 6; r = 0.409, P = 0.004; r = 0.585, P < 0.000; r = 0.368, P = 0.010) and the OSDI score (r = 0.466, P = 0.001; r = 0.631, P < 0.000; r = 0.418, P = 0.003), and negatively correlated with the Schirmer test score (r = −0.361, P = 0.012; r = −0.361, P = 0.012; r = −0.294, P = 0.043). 

Figure 6

 

The coronal area, the axial area, and the axial width were positively correlated with the CAS (Spearman's rank correlation).

Cytokine Concentrations in Tears and Correlative Factors

There was no difference between basal tear samples and reflex tear samples in the concentrations of IL-1β (P = 0.620), IL-6 (P = 0.103), IL-7 (P = 0.637), IL-17A (P = 0.622), IFN-γ (P = 0.161), or TNF-α (P = 0.053). The cytokine concentrations in reflex tears were used for statistical comparisons among the three groups. The tear levels of cytokine concentrations for the three groups are shown in Figure 7. The IL-1β concentration in reflex tears was significantly higher in active than in inactive TAO (P < 0.000) and the control (P < 0.000). Concentrations of IL-6 and IL-17A in reflex tears were significantly higher in active (P < 0.000; P < 0.000) and inactive TAO (P < 0.000; P = 0.028) than in the control; meanwhile, they were higher in active than inactive TAO (P < 0.000; P = 0.002). Concentration of IL-7 in reflex tears was highest by a significant margin in inactive TAO among the three groups (P < 0.000); meanwhile, it was lower in active TAO than in the control (P < 0.000). Concentration of TNF-α was significantly higher in both active (P < 0.000) and inactive TAO (P = 0.001) than in the control. 

Figure 7

 

Comparisons of reflex tear inflammatory cytokine concentrations among active TAO, inactive TAO, and control. Cytokine values are expressed in pg/mL. *P < 0.05.

In TAO patients, reflex tear IL-1β, IL-6, and IL-17A concentrations were positively correlated with the CAS, OSDI, and Fl scores, and reflex tear IL-7 concentrations were negatively correlated with the CAS, OSDI, and Fl scores (Table 3; Fig. 8). Reflex tear IL-1β, IL-6, and TNF-α concentrations were negatively correlated with the Schirmer test score (Table 3). 

Figure 8

 

Tear IL-1β, IL-6, and IL-17A concentrations were positively correlated with the CAS. Tear IL-7 concentrations were negatively correlated with the CAS (Spearman's rank correlation).

Table 3

 

Correlations of Tear Cytokine Concentrations With Clinical Examinations

	IL-1β	IL-6	IL-7	IL-17A	IFN-γ	TNF-α
CAS	r = 0.671, P < 0.000	r = 0.659, P < 0.000	r = −0.699, P < 0.000	r = 0.439, P = 0.002	r = 0.138, P = 0.351	r = 0.050, P = 0.734
OSDI score	r = 0.569, P < 0.000	r = 0.534, P < 0.000	r = −0.422, P = 0.003	r = 0.436, P = 0.002	r = 0.133, P = 0.368	r = 0.015, P = 0.917
Fl score	r = 0.547, P < 0.000	r = 0.391, P = 0.006	r = −0.411, P = 0.004	r = 0.366, P = 0.011	r = −0.012, P = 0.935	r = 0.291, P = 0.045
Schirmer test score	r = −0.436, P = 0.002	r = −0.463, P = 0.001	r = 0.269, P = 0.065	r = −0.208, P = 0.156	r = −0.041, P = 0.783	r = −0.292, P = 0.044
Lagophthalmos	r = −0.250, P = 0.086	r = 0.289, P = 0.046	r = −0.239, P = 0.101	r = 0.226, P = 0.123	r = −0.176, P = 0.230	r = 0.102, P = 0.490
Trichiasis	r = 0.282, P = 0.052	r = −0.003, P = 0.981	r = −0.170, P = 0.248	r = 0.364, P = 0.011	r = −0.056, P = 0.705	r = 0.116, P = 0.431

In the lacrimal gland measurements of TAO patients, the axial area was positively correlated with IL-1β and IL-17A concentrations in reflex tears (Fig. 9; r = 0.357, P = 0.013; r = 0.359, P = 0.012). Both coronal and axial areas were positively correlated with IL-6 concentrations in reflex tears (Fig. 10; r = 0.346, P = 0.016; r = 0.340, P = 0.018). The coronal width was positively correlated with IFN-γ and TNF-α concentrations in reflex tears (r = 0.310, P = 0.032; r = 0.373, P = 0.009). 

Figure 9

 

The axial area was positively correlated with IL-1β and IL-17A concentrations in reflex tears.

Figure 10

 

Both coronal and axial areas were positively correlated with IL-6 concentrations in reflex tears (Spearman's rank correlation).

Cytokine Concentrations in Serum and Correlative Factors

There were five patients in the TAO group whose bilateral eyes were in different stages of activity, which made the grouping unfeasible; thus, their data were excluded when comparing the serum cytokine concentrations among the three groups. The serum level of cytokine concentrations for the three groups is shown in Figure 11. Concentrations of IL-1β and IFN-γ in serum were significantly higher in active TAO than inactive TAO (P = 0.047) and the control (P = 0.005). Furthermore, IL-6 concentration in serum was significantly higher in active (P = 0.004) and inactive TAO (P = 0.035) than the control. 

Figure 11

 

Comparisons of serum inflammatory cytokine concentrations among active TAO, inactive TAO, and control. Cytokine values are expressed in pg/mL. *P < 0.05.

The mean values of basal tear cytokine concentrations, reflex tear cytokine concentrations, and CAS of both eyes were used to analyze the correlative factors of serum cytokine concentrations in the TAO group. Serum IL-1β concentrations were positively correlated with basal and reflex tear IL-1β concentrations (r = 0.471, P = 0.020; r = 0.426, P = 0.038). Serum IFN-γ concentration was positively correlated with CAS (r = 0.449, P = 0.028). No correlation was found between other serum cytokine concentrations and CAS (P > 0.05). 

Discussion

In our study, we found that changes in tear inflammatory cytokines in TAO patients are accompanied by an enlargement of the lacrimal gland. As early as 1981, Trokel ²² had reported lacrimal gland enlargement in Graves' disease. So far, few studies have involved accurate lacrimal gland measurement to demonstrate its enlargement. ^23,24 For the first time, we measured the lacrimal gland on MRI referring to Harris' and Tamboli's methods on CT images, ^7,19 and found no difference in terms of age, sex, or between bilateral eyes in the control, which was inconsistent with previous studies ^19,23,25 due to the small sample size in the control group. 

The results of lacrimal gland measurements in the three groups and correlative factors analysis demonstrated that the lacrimal gland is larger in TAO patients than in the controls, and even larger as the TAO inflammatory activity heightens and ocular surface discomfort enhances. This enlargement has obvious features that are mainly displayed in terms of width. That is to say, the lacrimal gland of TAO patients is enlarged laterally toward both sides of the head, which is different from the lengthwise changes with age. ²⁶ Since it is hard to obtain histological evidence of lacrimal gland inflammation in TAO patients, the use of imaging tests is almost the only approach to visualizing lacrimal gland lesions. Enlargement of the gland usually occurs with inflammation or tumors, and in our case, inflammation is the most likely cause. 

Previous researchers focusing on the ocular surface of TAO patients usually collected basal tear samples, for analyzing both tear film osmolarity ^2–4 and the tear protein component. ^11–13 We found several drawbacks in basal tear collection and analysis in our study. First, the ocular surface of TAO patients is usually covered by mixed tears with reflex tears in it caused by ocular surface damage; meanwhile, basal tears in TAO patients are very difficult to collect because the tear fluid volume is rather small and susceptible to environment and stimulation. Second, reflex tears are more a reflection of lacrimal gland secretion than basal tears due to evaporation and possible secretion from ocular surface by mitogen-activated protein kinase intracellular signaling pathways. ²⁷ Thus, we collected not only basal tears, but also reflex tears with nasal stimulation, and found no significant differences in six cytokine concentrations between two samples, which means that evaporation or possible secretion from the ocular surface was not observable in our study. 

The changes in six cytokine concentrations in tears were similar to those found in previous studies. ^11,28,29 Tear IL-1β, IL-6, and IL-17A levels were higher as CAS increased, indicating that it may be possible to use these cytokines as markers of disease activity. The results also indicated that Th1 cytokines, Th2 cytokines, and Th17 cytokines were involved in the course of TAO. The reason for increased IL-7 concentration in the inactive phase was unclear, but might have to do with fibroblasts proliferating in the chronic stage. ²⁹ An influence of serum on tear cytokine concentrations was not observable in the serum analysis results. 

The possible causes of increased cytokines in the tears of active TAO patients have not been satisfactorily explained. We cannot attribute it to evaporation or possible secretion from ocular surface cells because neither mechanism was observable in our study. Since lacrimal gland inflammation is the most probable cause of its enlargement, combined with experimental study of lacrimal gland impairment in TAO, ⁸ we put forward the following hypothesis: Autoantibodies might bind to TSH-R, and this antigen-specific immune response may engender inflammation and inhibit lacrimal gland function. The inflammatorily swelling lacrimal gland may be able to produce inflammatory cytokines that are secreted into tears and cause ocular surface damage. However, the levels of tear cytokines may not be in accordance with the degree of lacrimal gland enlargement, as our results showed, which may be affected by the course of disease, amount of autoantibodies and TSH-R, mean target of orbital inflammation, and so on; however, the influence of serum could not be observed in our results. 

So far, there is no ideal animal model of Graves' disease, especially with orbital pathological changes. ³⁰ Looking into the future, an ideal animal model would help us to obtain histological evidence of lacrimal gland inflammation in TAO, avoiding the limitations of imaging tests. Furthermore, other cytokines and chemokines may also play an important role in inflammatory disease, and should be analyzed in future study. 

In conclusion, we have shown that TAO patients have higher levels of inflammatory cytokines in tears and larger lacrimal gland than control subjects. The enlargement of the lacrimal gland in TAO manifested mainly in an extension toward both sides of the head. The changes in the inflammatory cytokine concentrations existed at the very beginning of tear secretion from the lacrimal gland. We concluded that increased inflammatory cytokines play an important role in ocular surface damages, and might be associated with the inflammatory involvement of the lacrimal gland. 

Acknowledgments

Supported by the Guangdong Technology Project Foundation of China (Nos. 2010B031100012 and 2011B061300068). 

Disclosure: D. Huang, None; Q. Luo, None; H. Yang, None; Y. Mao, None 

Footnotes

 DH and QL contributed equally to the work presented here and should therefore be regarded as equivalent authors.