The diagnostic significance of Fourier-domain optical coherence tomography in Sjögren syndrome, aqueous tear deficiency and lipid tear deficiency patients

Authors


Lan Gong
Department of Ophthalmology
Eye and ENT Hospital of Fudan University
NO83, Fenyang RD, Xuhui District
Shanghai 200031
China
Tel: + 86 21 64310068
Fax: + 86 21 64377134 217
Email: gonglan70@gmail.com

Abstract.

Purpose:  To determine the role of Fourier-domain optical coherence tomography (FD-OCT) in tear meniscus imaging and evaluate its diagnostic significance in Sjögren syndrome (SS), non-Sjögren’s aqueous tear deficiency (ATD) and lipid tear deficiency (LTD) patients.

Methods:  Two hundred and thirty-six dry eye patients and 174 healthy controls were enrolled in this study. All subjects were grouped as follows: group A (ATD), group B (LTD), group C (SS) and group D (normal controls). All subjects underwent dry eye questionnaire, FD-OCT scanning, tear film break-up time (BUT), corneal fluorescence staining and Schirmer I test (SIT). Tear meniscus height (TMH), tear meniscus depth (TMD) and tear meniscus cross-sectional area (TMA) were measured using FD-OCT (RTVue-100). The area under the receiver operating characteristic curve and the cut-off point were determined using a logistic regression model.

Results:  Mean TMH, TMD, TMA, BUT and SIT of dry eye patients were significantly lower than those of the controls (p < 0.05). Tear meniscus values were significantly decreased in patients with SS compared with ATD and LTD patients. Tear meniscus values were significantly correlated with clinical examination results in all groups. Accuracy of dry eye diagnosis by FD-OCT is highest in patients with SS and lowest in LTD patients. The clinical diagnostic critical points were quite different between groups.

Conclusions:  Fourier-domain optical coherence tomography could provide precise measurement of the tear meniscus with favourable repeatability. Diagnostic significance is more conspicuous in patients with SS. Tear meniscus measurement by FD-OCT is expected to become a valuable technique in ATD dry eye screening and diagnosis.

Introduction

Dry eye is a condition in which there are insufficient tears to lubricate and nourish the eye (The International Dry Eye Work Shop 2007b: the definition and classification of dry eye disease). Tears are necessary for maintaining the health of the front surface of the eye and for providing clear vision (Holly 1985). People with dry eyes may have symptoms of discomfort, visual disturbance or tear film instability with potential damage to the ocular surface (The International Dry Eye Work Shop 2007b: the definition and classification of dry eye disease). For nearly half a century, a tetrad of diagnostic tests have been universally applied to assess symptoms, tear stability, ocular surface staining and reflex tearflow (The International Dry Eye Work Shop 2007a: methodologies to diagnose and monitor dry eye disease). Schirmer I test, tear film break-up time (BUT), ocular ferning testing, fluorescein staining, tear osmolarity and tear lysozyme and lactoferrin assays are commonly used in the clinic (The International Dry Eye Work Shop 2007a: methodologies to diagnose and monitor dry eye disease); however, reproducibility is not satisfactory and there are many confounding factors (Nichols et al. 2004). Almost 90% of tear fluid is contained within the menisci (Holly 1985), and measurement of their height or curvature are indicators of tear volume. Tear meniscus parameters, such as tear meniscus height (TMH) and radius of tear meniscus curvature, have been shown to be sensitive indicators in evaluating tear-deficient dry eye (Mainstone et al. 1996). Optical coherence tomography (OCT) is increasingly being recognized as a noninvasive and precise alternative for measuring tear meniscus (Savini et al. 2006). Many studies have reported significantly decreased TMH in patients with dry eye patients (Mainstone et al. 1996; Shen et al. 2009; Ibrahim et al. 2010), and decreased TMH is positively correlated with dry eye severity. Tear meniscus values also show correlations with Schirmer I test and BUT in our previous study (Qiu et al. 2011), and the accuracy of dry eye diagnosis by Fourier-domain OCT (FD-OCT) was approximately 70% with favourable repeatability. Shen et al. (2009) also reported that OCT has good diagnostic sensitivity and specificity in non-Sjögren’s aqueous tear deficiency (ATD) patients. Therefore, OCT may not only be used to observe the tear meniscus, but also offers evidence for its use in dry eye diagnosis. However, there is currently a paucity of research on the diagnostic significance of FD-OCT in different types of dry eye. Thus, the application of OCT in different types of dry eye diagnosis is still the subject of considerable debate. The purpose of this study is to investigate a diagnostic technology that shows promise for advancing our ability to investigate, monitor or diagnose dry eye disease in future.

Methods

Patient and control population

From 2007 to 2009, 236 dry eye patients (236 eyes) diagnosed at the Eye Ear Nose and Throat Hospital of Fudan University and 174 healthy controls (174 eyes) were enrolled in this study. All dry eye patients were grouped according to dry eye classification: group A (non-Sjögren ATD dry eye), group B (lipid tear deficiency (LTD) dry eye) and group C [Sjögren syndrome (SS) dry eye]. Normal controls were enrolled in group D. Detailed group information is displayed in Table 1. The study and data accumulation were carried out with approval from the Institutional Review Board of the Eye and ENT Hospital of Fudan University before the study began. All subjects signed the informed consent. Research was carried out in accordance with the tenets of the Declaration of Helsinki.

Table 1.   Research data and clinical examination results of dry eye (Group A, B, C) and normal control group (Group D) (mean ± SD).
GroupnSex ratio (M/F)Age (years)CFS (grade 0–12)DEQ scoreBUT (S)SIT (mm/5 min)TMH (μm)TMD (μm)TMA (μm2)
  1. n = patients or normal controls’ number; CFS = corneal fluorescein staining; DEQ = dry eye questionnaire; BUT = tear film break-up time; SIT = Schirmer I test; TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area; ATD = non-Sjögren’s aqueous tear deficiency; LTD = lipid tear deficiency; SS = Sjögren syndrome.

  2. *p Values were calculated adjusting for age.

A (ATD)1020.58845.618 ± 16.2510.147 ± 0.69538.569 ± 40.24910.676 ± 0.9775.235 ± 2.507245.866 ± 81.032166.562 ± 61.70126 284.3 ± 19 837.9
B (LTD)810.21038.457 ± 16.6940.753 ± 2.01631.284 ± 11.9764.654 ± 1.59017.735 ± 6.783311.436 ± 119.890197.675 ± 65.34139 687.2 ± 29 339.5
C (SS)530.07651.679 ± 8.7245.245 ± 4.15548.566 ± 39.7662.566 ± 1.9661.906 ± 1.334138.283 ± 49.31387.761 ± 37.1938069.2 ± 5729.8
D (normal controls)1740.32835.092 ± 13.8270.017 ± 0.16923.816 ± 12.88410.661 ± 1.05620.868 ± 7.811306.977 ± 98.854208.226 ± 67.05340 775.9 ± 27 650.5
F value58.05369.690382.71177.1802571.170643.582142.327153.44581.993
p value0.0000.0000.0000.0000.0000.0000.0000.0000.000
p value (A versus D)*0.0000.0000.1860.0000.510.0000.0000.0000.000
p value (B versus D)*0.0540.1140.0000.0000.0000.0000.7480.1700.652
p value (C versus D)*0.0000.0000.0000.0000.0000.0000.0000.0000.000

Dry eye group

Patients who met the following criteria were included in each group accordingly. Group A: any of the following eye symptoms: scratchy or sandy sensation, itching, burning, redness and photosensitivity; BUT ≥ 10 seconds and SIT < 10 mm/5 min. Group B: any of the following eye symptoms, scratchy or sandy sensation, itching, burning, redness and photosensitivity; noninflamed meibomian gland dysfunction (MGD) and SIT ≥ 10 mm/5 min. The criteria of noninflamed MGD included the presence of meibomian gland dropout by transillumination through the tarsus, no or poor meibum expression in response to digital pressure, lack of active inflammation, and the presence of meibomian gland orifice squamous metaplasia or acinar atrophy (Goto et al. 2002; Lee & Tseng 1997; Shimazaki et al. 1995, 1998). Group C: patients with SS were diagnosed and enrolled according to the criteria of Fox and colleagues (Fox et al. 1986; Vitali et al. 2002). Sjögren syndrome was defined by the ocular symptoms (dry eyes, sensation of sand or gravel, use of tear substitutes), oral symptoms (dry mouth, swollen salivary glands, frequently drink liquids), ocular signs (Schirmer’s I test < 5 mm in 5 min, Rose bengal score or other ocular dye score >4 according to van Bijsterveld’s scoring system), histopathology, salivary gland involvement and autoantibodies. Any eye meeting the criteria was enrolled, but only the right eye was enrolled if both eyes met the criteria. Exclusion criteria included a history of current ocular disease (e.g. nasolacrimal obstruction, chemical injury, severe blepharitis, conjunctivitis, allergic conjunctivitis, conjunctivochalasis and lid disorders such as ectropion or entropion), ocular surgery history, contact lens use and punctal occlusion.

Control group

Enrolment criteria included the following: (i) none of the following eye symptoms: scratchy or sandy sensation, itching, burning, redness or photosensitivity; (ii) BUT ≥ 10 seconds and (iii) Schirmer I test ≥ 10 mm/5 min. Subjects with the following conditions were excluded: eye disorders, ocular surgery history, contact lens use or any allergic diseases such as allergic conjunctivitis. Only the right eye was studied in the control group.

Dry eye questionnaire

The dry eye questionnaire (DEQ) developed by McMonnies was adopted in this study (McMonnies & Ho 1987). We translated the questionnaire into Mandarin Chinese. The DEQ includes categorical scales to measure the prevalence, frequency, diurnal intensity and intrusiveness of common ocular surface symptoms, including discomfort, dryness, visual changes, soreness and irritation, grittiness and scratchiness, burning and stinging, foreign body sensation, light sensitivity and itching. The DEQ also includes questions on age, gender, affected daily activities, computer use, use of systemic and ocular medications, allergies, self-assessment and previous diagnosis of dry eye. The sum of the scores of all the questions was treated as the total symptoms score.

Tear menisci measurement

The examination was carried out by an ophthalmologist who specialized in performing the OCT examination and did not join the clinical trial. Images were acquired with an anterior segment OCT system (RTVue-100, Optovue Inc., Freemont, CA, USA) used at 830 ± 10 nm wavelength and 26 000 axial scans (A-scans) per second (axial resolution, 5 μm; wider resolution, 15 μm). The entire scanning range in air was 2–12 mm in width and 2.0–2.3 mm in depth. The lower TMH, tear meniscus depth (TMD) and tear meniscus cross-sectional area (TMA) of all patients were measured by anterior eye segment OCT using the Fourier-domain technique. The single line scanning mode by anterior segment-wide angle lens was selected (scanning line length, 3 mm; scanning direction, 90°–270°). Scanning started at the 6o’ clock position of the cornea immediately after the patient blinked. The head was stabilized by an adjustable chin rest and forehead strap. Subjects were instructed to look straight ahead and look at an external target LED in front of the eye examined. The participant was instructed to blink normally to evenly distribute tear film and minimize ocular surface dehydration. Just before the measurement, the participant was asked to hold their blink during the acquisition of the scan. Optical coherence tomography settings remained constant throughout the experiment for all subjects. Three consecutive scans were performed during each examination, each after a blink, with a scanning interval of 3–5 seconds (Fig. 1). All three measurements were recorded for data analysis.

Figure 1.

 (Left) The result of image acquisition by Fourier-domain optical coherence tomography. The length of scanning lines is 3 mm and the scanning direction is 90°–270°. Scanning started at the 6o’clock position of the cornea. (Middle) The tear meniscus height (TMH) was the straight line distance between the upper extreme and the lower extreme of the tear boundary line while tear meniscus depth (TMD) was measured as the vertical distance of the interfacing point of the cornea with the lower eyelid to the tear height line. (Right) The tear meniscus cross-sectional area (TMA) was the area of the triangle formed by the corneal anterior boundary, anterior boundary of the lower eyelid and anterior borderline line of the tear meniscus.

Tear meniscus height, TMD and TMA were determined from the OCT images with an Optovue RTVue (Optovue, Inc. Fremont, CA, USA) system. Tear meniscus height was defined as the straight line distance between the upper extreme and the lower extreme of the tear boundary line; TMD as the vertical distance from the interface of the cornea with the lower eyelid to the tear height line; and TMA as triangular area formed by the corneal anterior boundary, anterior boundary of the lower eyelid and anterior borderline line of the tear meniscus (Fig. 1).

Tear BUT and corneal fluorescein staining

Tear film stability was estimated based on tear BUT. A fluorescein-impregnated strip (Jingming, Tianjing, China) wetted with nonpreservative saline solution (20 μl by micropipette) was placed in the lower conjunctival sac. The patient was asked to blink three times and then look straight forward without blinking. The tear film is observed under cobalt blue-filtered light of the slit lamp microscope, and the time that elapsed between the last blink and appearance of the first break in the tear film is recorded with a stopwatch. Break-up time was measured in triplicate and was reported as the mean time; Fluorescein staining for corneal damage was graded on a scale of 0–3 points (none to severe) in each quadrant (total points, 0–12).

Schirmer I test

Tear production was measured by the Schirmer I test without anaesthesia. The test was performed by placing a dry Schirmer test strip (Jingming, Tianjing, China) over the lower lid margin at the junction of the middle and lateral one-third of the eye lid for 5 min. The patient was instructed to close his or her eyes during the course of the test. The strip was then removed, and the distance of wetness in millimetres was recorded as the Schirmer I test score.

Experimental Procedure

Subjects were tested between 9:00 am and 12:00 am in a consulting room in which central air conditioning and humidifiers controlled the temperature (27°C) and humidity (30–50%). To avoid reflex tearing, no additional light other than the room light was used in the consulting room. A preliminary anterior segment examination excluded current ocular diseases. Details of the lid margins and lashes, the palpebral and bulbar conjunctival surfaces, the tear film and cornea and the iris were examined. After a 15-min rest, patients answered a DEQ and then were evaluated by OCT examination. Ten minutes after OCT scanning, BUT was determined and corneal fluorescein staining was performed. After a 30-min rest, the Schirmer I test was conducted.

Data analysis

All statistical analyses were performed with spss for Windows (ver. 13.0, SPSS, Inc., Chicago, IL, USA). Data are expressed as the mean ± SD. Two-sided pairwise comparisons were conducted (p < 0.05). Analysis of variance, chi-squared test and Student’s t-test were used to determine differences between the dry eye groups and control group. Spearman correlation coefficients were calculated to assess the relationship between variables. The area under the receiver operating characteristic (ROC) curve and the cut-off point were determined using a logistic regression model. Intraindividual variation in OCT measurements is defined as the reproducibility of tear meniscus measurements by OCT. Interindividual variation is defined as the dispersion of data distribution between subjects.

Results

Clinical examination results of dry eye and control group

Comparison of the clinical examination results of three dry eye groups and those of control group revealed significant differences for all tests (Table 1). The mean TMH, TMD, TMA values were significantly lower in the dry eye group compared with the control group. Tear meniscus values were significantly decreased in patients with SS (Fig. 2).

Figure 2.

 The measurement of tear meniscus by Fourier-domain optical coherence tomography in non-Sjögren aqueous tear deficiency dry eye (ATD), lipid tear deficiency dry eye (LTD), Sjögren syndrome dry eye (SS) patients and normal controls (N).

Correlation between test results of dry eye and control group

Spearman analysis revealed significant correlations between tear meniscus values and clinical examination results (Table 1). We found a significant linear positive correlation between TMH, TMD, TMA, Schirmer I test and BUT results. Significant negative correlation was also observed between tear meniscus values, age, corneal fluorescence staining and DEQ. All clinical test results were positively or negatively correlated with age (Table 2).

Table 2.   Correlation of Clinical test results of dry eye (Group A, B, C) and control group (Group D) (Spearman).
 GenderAgeCFSDEQBUTSITTMHTMD
  1. * p = 0.159, The other p values are all below 0.05.

  2. CFS = corneal fluorescein staining; DEQ = dry eye questionnaire; BUT = tear film break-up time; SIT = Schirmer I test; TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area.

Age0.059
CFS0.1370.134
DEQ0.1470.3490.290
BUT−0.319−0.154−0.491−0.367
SIT0.042*−0.368−0.380−0.3810.149
TMH−0.080−0.204−0.356−0.2660.2180.527
TMD−0.108−0.201−0.363−0.2760.2640.5200.887
TMA−0.085−0.201−0.363−0.2570.2410.5290.9550.956

Inter- and intraindividual variation of dry eye and control group

Table 3 illustrates the wide interindividual variation in tear meniscus values between these groups. Intraindividual variability in OCT measurements was low in all groups, and thus, tear meniscus measurements were reproducible (Table 4).

Table 3.   Interindividual variation of dry eye and control group.
Interindividual variation (group)TMHTMDTMA
  1. TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area; ATD = non-Sjögren’s aqueous tear deficiency; LTD = lipid tear deficiency; SS = Sjögren syndrome.

A (ATD)0.3300.3700.755
B (LTD)0.3850.3310.739
C (SS)0.3570.4240.710
D (Normal Controls)0.3220.3220.678
Table 4.   Reproducibility of the tear meniscus measurement with FD-OCT.
Intraindividual variation (group)TMHTMDTMA
  1. TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area; ATD = non-Sjögren’s aqueous tear deficiency; LTD = lipid tear deficiency; SS = Sjögren syndrome.

A (ATD)0.0870.0860.168
B (LTD)0.0640.0800.141
C (SS)0.0790.1020.161
D (normal controls)0.0950.0860.170

Diagnostic accuracy of dry eye diagnosis by FD-OCT

In this study, we evaluated the diagnostic efficiency of FD-OCT by comparing results of patients with different dry eye types with those of a control group without eye disorders. We also studied the accuracy of tear meniscus measurement by FD-OCT. The area under the curve (AUC) was calculated by the ROC curve. Results are summarized in Tables 5–7.

Table 5.   Diagnostic accuracy of FD-OCT in dry eye diagnosis.
 TMHTMDTMA
  1. TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area; ATD = non-Sjögren’s aqueous tear deficiency; LTD = lipid tear deficiency; SS = Sjögren syndrome; TP=number of true positive specimens; FP=number of false-positive specimens; FN=number of false-negative specimens; TN=number of true negative specimens.

  2. Diagnostic accuracy = (TP + TN)/(TP + FP + FN + TN).

  3. All p values between groups were calculated by Bonferroni test.

A (ATD) (%)66.6768.1271.01
B (LTD) (%)45.4967.0646.27
C (SS) (%)91.1992.5192.51
p value (A versus C)0.0000.0000.000
p value (B versus C)0.0000.0000.000
p value (A versus B)0.0001.0000.000
Table 6.   AUC of dry eye diagnosis by FD-OCT.
AUCTMHTMDTMA
  1. AUC = the area under the ROC curve; TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area; ATD = non-Sjögren’s aqueous tear deficiency; LTD = lipid tear deficiency; SS = Sjögren syndrome.

A (ATD)0.69390.69610.7110
B (LTD)0.49620.53990.5279
C (SS)0.96500.96370.9706
Table 7.   Clinical diagnostic critical point of dry eye diagnosis by FD-OCT.
Cut-off pointTMH (μm)TMD (μm)TMA (μm2)
  1. TMH = tear meniscus height; TMD = tear meniscus depth; TMA = tear meniscus cross-sectional area; ATD = non-Sjögren’s aqueous tear deficiency; LTD = lipid tear deficiency; SS = Sjögren syndrome.

A (ATD)248.333164.00024 000.000
B (LTD)256.000142.33345 666.670
C (SS)141.333130.33313 333.333

Accuracy of dry eye diagnosis by FD-OCT is highest in patients with SS and lowest in LTD patients (Table 5). Tear meniscus cross-sectional area had better diagnostic efficiency than TMH and TMD in ATD and SS groups. The AUC values calculated by the ROC technique revealed an acceptable diagnostic efficiency in ATD and patients with SS (Fig. 3 and Table 6). The clinical diagnostic critical points were quite different between groups and could be useful in ATD and patients with SS (Table 7).

Figure 3.

 Receiver operating characteristic (ROC) graph analysis delineating the sensitivity and specificity of tear meniscus cross-sectional area measurement by Fourier-domain optical coherence tomography.

Discussion

Traditional diagnostic approaches for dry eye syndrome such as slit lamp observation, Schirmer test, BUT and fluorescein staining suffer from low diagnostic efficiency and repeatability because of low measurement accuracy or eye stimulation (Nichols et al. 2004). Noninvasive or minimally invasive techniques are advantageous because they capture data from the surface of the eye without significantly inducing reflex tearing. For this reason, techniques that gather information from the tear film by processing reflected light or images from the tear film surface are desirable and are representative of the ocular surface (Doughty et al. 2001). The availability of a noninvasive, quick and comfortable dry eye diagnostic test would be a significant advancement in diagnostic approaches. With recent technical advancements and application in the anterior segment, OCT has become a novel alternative for tear meniscus measurement (Shen et al. 2009; Ibrahim et al. 2010; Wang et al. 2006; Zhou et al. 2009; Keech et al. 2009; Shen et al. 2008; Bitton et al. 2007). Furthermore, Veres et al. demonstrated that OCT enabled a noninvasive, high resolution method of imaging, evaluating and classifying lid parallel conjunctival folds. These new classifications correlated well with the slit lamp grade and the DEQ scores, promising a new, more objective evaluation of dry eye (Veres et al. 2011). In this study, we applied the novel Fourier-domain OCT, which could provide a rapid, noncontact, noninvasive approach to produce a high-quality image and investigated its applicability in dry eye diagnosis.

The classification of dry eye could provide a guide to patients and a more elaborate diagnosis to the ophthalmologist. The major classes of dry eye are described as aqueous-deficient dry eye and lipid-deficient dry eye. Aqueous-deficient dry eye is divided into two major subgroups: Sjögren syndrome dry eye and non-Sjögren syndrome dry eye (The International Dry Eye Work Shop 2007b: the definition and classification of dry eye disease). Shen et al. (2009) reported the diagnostic application of OCT in non-Sjögren syndrome ATD patients; however, its application has not been investigated in other types of dry eye. The current study reports the first application of OCT in three major types of dry eye to determine the efficacy of FD-OCT for several types of dry eye.

Tear meniscus values measured by OCT in this study were significantly lower in dry eye patients compared with controls, especially in patients with SS. Patients with SS were mainly women and received worse clinical examination results than men. Using noninvasive interference tear meniscometry, Uchida et al. (2007) reported that the mean lower tear meniscus height was significantly smaller in ATD patients with Sjögren’s syndrome than that in the healthy control group, which was similar with Chen’s study (Chen et al. 2011).

Similar to others (Shen et al. 2009; Ibrahim et al. 2010), we found tear meniscus values were significantly correlated with clinical examination results such as BUT and Schirmer I test. In our study, the Schirmer I test correlated more significantly with the variables of tear meniscus than which in Shen’s study (r = 0.520–0.529 and r = 0.21–0.34, respectively) (Shen et al. 2009); the BUT correlated less significantly with the tear meniscus height than which in Ibrahim’s study (r = 0.218 and r = 0.53, respectively) (Ibrahim et al. 2010). However, Wang et al. (2008) found that BUT was correlated with tear meniscus measurement values, but correlation between Schirmer I test and TMH was poor. We also found OCT results of tear meniscus correlated more significantly with Schirmer I test and less with BUT, which is not in agreement with our previous study (Qiu et al. 2011). This change may be due to different types of dry eye patients enrolled or larger sample capacity in our present study. The correlation between Schirmer I test and tear meniscus was much stronger in our study, which may be a result of our avoidance of topical anaesthetics. The Schirmer I test is primarily used to quantify tear secretion and is commonly used to diagnose dry eye; however, its reproducibility is not satisfactory and many factors may interfere with the results. The topical anaesthesia applied during this test may reduce the stimulation of filter paper to the eye surface and inhibit reflective tear secretion and may influence the draining of tears, which can produce considerable variability in the test results, increasing the likelihood of false-positive or false-negative results. The Schirmer I test, which is similar to the basic secretion test but without topical anaesthesia, measures both basic and reflex tearing combined. Although this test is relatively specific, the level of sensitivity is poor. Using lower cut-off measurements increases the specificity of this test but decreases its sensitivity. Generally speaking, a decreased Schirmer value corresponds to decreased tear meniscus parameters. A dry eye patient frequently has both low tear meniscus measurement and a low Shirmer value. But sometimes because of test irritation, a patient may have decreased tear meniscus parameters but normal Schirmer value. Tear meniscus measurement using FD-OCT could better reflect the natural state of the tear secretion. Many studies have reported that OCT was a good noninvasive method for imaging and measuring tear meniscus and was able to distinguish between dry eye and nondry eye. In this study, we found that all clinical examination values were negatively correlated with age. However, the DEQ score was positively correlated with age.

Using OCT, Bitton et al. (2007) analysed tear meniscus measurement variability and demonstrated good reproducibility and relatively low inter- and intraindividual variability. In our study, the intraindividual variation of tear meniscus values had also shown favourable repeatability, which is similar to Zhou’s study (Zhou et al. 2009). Fourier-domain optical coherence tomography measures tear meniscus with higher reproducibility than previous OCT instruments. Fourier-domain optical coherence tomography may be a useful way to measure dry eye severity and treatment effectiveness.

The traditional objective diagnosis of dry eye in the clinical setting relies on the Schirmer I test, BUT and vital staining of the ocular surface. These invasive tests may cause irritation and reflex tearing and can influence the results. Often the information gained from each test is not consistent with the others. Optical coherence tomography-derived quantitative measurement of tear meniscus variables enables this noninvasive modality as a potential diagnostic tool for dry eye. Subjective examinations and physical measurements of tear meniscus were performed in our study to evaluate the diagnostic efficacy of FD-OCT in major types of dry eye patients. Diagnostic tests should combine high overall accuracy with good sensitivity. The diagnostic accuracy of FD-OCT was almost 70% in ATD patients, lower than 50% in LTD patients and higher than 90% in patients with SS. These data suggest limitations in ATD and LTD dry eye diagnosis and accuracy in SS diagnosis when using OCT tear meniscus measurements. The area under the ROC curve of TMH, TMD and TMA ranged from 0.49–0.97. According to the common grading scale (Song 1997), this technique demonstrated moderate diagnostic accuracy in ATD patients and good diagnostic accuracy in patients with SS. The area under the ROC curve in ATD patients was less than Shen’s data and similar to Ibrahim’s results (Shen et al. 2009; Ibrahim et al. 2010). Fourier-domain optical coherence tomography was likely more sensitive for severe aqueous-deficient dry eye.

Tear meniscus variables, such as height, width, cross-sectional area and meniscus curvature, have been reported to be useful in the diagnosis of dry eye (Shen et al. 2009; Ibrahim et al. 2010; Qiu et al. 2011; Zhou et al. 2009; Wang et al. 2008). Tear meniscus cross-sectional area was the best indicator of ATD and SS among the three tear meniscus variables measured. Tear meniscus cross-sectional area could eliminate some mixed factors. For example, removing the interference factor of individual palpebral aperture suggested that it may result in better efficacy for diagnosis of dry eye according to TMA. The evaluation of TMA provides valuable diagnostic information to the clinician with respect to the overall tear volume in the diagnosis of dry eye.

In this study, we obtained a series of clinical diagnostic critical points that could distinguish between patients with dry eye and healthy subjects in the different subgroups of dry eye. Because there was insufficient diagnostic accuracy of LTD patients, the critical point is not clinically relevant; however, the cut-off points of ATD and SS could be applied in clinical diagnosis. The clinical diagnostic critical points were quite different between three groups. We think that these are mainly because of the different pathomechanisms involved in the three groups. Because the diagnostic accuracy of LTD patients was quite low, the clinical diagnostic critical points were not clinical significant. Patients with SS have the features of decreased of tear secretion that is less decreased in non-Sjögren ATD patients. Therefore, the clinical diagnostic critical points were quite different between these two groups.

Our study aimed to search for a new and reliable method for different types of dry eye patients. Diagnosis of dry eye disease is made difficult by its multifactorial aetiology, by the need for a comprehensive definition, and by the use of tests that are limited and variable in their assessment of the tears and the ocular surface. A large number of tests have been used singly or in combination with diagnose the condition with variable success because of the inherent variability of most tests and their inability to specifically measure the physiological tear changes in dry eye. We applied the FD-OCT as study target to investigate a diagnostic technology that shows promise for advancing our ability to investigate, monitor or diagnose dry eye disease in future. To achieve this goal, the area under the ROC curve and the cut-off point were determined using a logistic regression model. A linear combination of specificity and sensitivity is defined (results were calculated by the sum, the total number of correct classifications or a sum weighted with respect to gains and losses), and the cut-off point that maximizes this combination in our study. To gain the best combination of specificity and sensitivity, we calculated the cut-off point in each group. The clinical diagnostic critical points are expected to be between the mean values of the controls and the patients. However, the diagnostic critical points may be occasionally lower than the mean values of dry eye patients. This could be as a result of selection bias of patient or diagnostic tests. Because the diagnostic accuracy of LTD patients (Group B) was quite low, the clinical diagnostic critical points of them were no clinical significance. That could explain the unusual higher cut-off points than the values of controls.

There are some limitations in this study. It should be noted that the diagnostic efficacy in this study may have been affected by selection bias. The ATD patients were prescreened by Schirmer test. Thus, some ATD patients may have been excluded because of the unreliable nature of this test. Also, we did not control palpebral aperture and blink rate, and location of the junctions of the tear meniscus points for the determination of the tear meniscus borders was subjective. The variation patients’ ages resulted in selection bias.

Further studies of OCT tear meniscus measurements should focus on determination of age- and gender-specific cut-off values. Moreover, these tests for the diagnosis of dry eye disease should be performed alone or in various combinations to determine their sensitivity and specificity.

Measurements of tear menisci by OCT are rapid and noninvasive. Our findings suggest acceptable sensitivity and specificity of OCT tear meniscus measurements in the diagnosis of dry eye disease. This technique was more appropriate for patients with severe aqueous-deficient dry eye. Therefore, OCT may not only be used to observe the tear meniscus but also offers evidence for its use in dry eye diagnosis. The application of OCT in dry eye diagnosis is still controversial. Although FD-OCT is not considered the gold standard for diagnosis of dry eyes, it is expected to become a valuable diagnostic technique for dry eye diagnosis in future.

Acknowledgements

This work was supported by grants from National Natural Science Foundation of China (NO. 30873287). The authors thank Dr. Naiqing Zhao for his assistance in statistical analysis (Department of Biostatistics and Social Medicine, School of Public Health, Fudan University, Shanghai, China).

Ancillary