Direct visualization of aqueous tear secretion from lacrimal gland

Study design

This study followed the principles of the Declaration of Helsinki and was approved by the Institutional Review Board of Chuncheon Sacred Heart Hospital. Twenty-six patients with non-Sjögren dry eye syndrome and 26 healthy subjects were included in the study. The inclusion criteria for the dry eye group were symptoms typical for this syndrome (i.e. dryness, itching, foreign body sensation) with a low tear film break-up time (tBUT <5 seconds), low Schirmer I score (<10 mm per 5 min without anaesthesia) and corneal punctate fluorescein staining (Oxford staining score of >1) in either eye. Exclusion criteria included a history of ocular injury, infection, non-dry eye ocular inflammation, trauma or surgery within the previous 6 months and uncontrolled systemic disease. For healthy subjects, inclusion criteria were no dry eye syndrome (no typical dry eye syndrome symptoms, tBUT >5 seconds, Schirmer I score >10 mm, Oxford staining score <1) and exclusion criteria were the same.

Examinations

Ophthalmologic examinations were performed in the following order. We first applied Schirmer's test I without anaesthesia to both eyes. We then applied fluorescein dye to both eyes to measure tBUT and scored the corneal staining according to the Oxford scoring scheme (Bron et al. 2003) using fluorescein papers (Haag Streit, Koeniz, Switzerland). We waited for 10 min between these two measurements to minimize the influence of Schirmer's test on the measurement of tBUT. To expose the palpebral lobe of the main lacrimal gland, the examiner pulled the temporal part of the upper eyelid in the superotemporal direction and directed the subject to look in the inferonasal direction. Photographs of the exposed palpebral lobes were obtained using a biomicroscope and camera system (×6.3). We applied fluorescein dye to the surface of the palpebral lobe using fluorescein paper (Haag Streit). We changed the light of the slit lamp to cobalt blue and captured the washout of fluorescein dye from the excretory openings on the palpebral lobe in photographs and movies (×10) (Fig. 1, Video S1). We investigated the location and number of excretory openings that secreted aqueous tears. In some patients, we scanned the area surrounding the opening with anterior segment optical coherence tomography (OCT; Spectralis, Heidelberg, Germany) to show that tear-secreting openings were orifices connected to the excretory ducts. We repeated these tests on the other eyes. Topical anaesthesia was not applied during the test. Tests were performed by an experienced ophthalmologist (HHS).

Tear flow rate measurement

We measured the tear flow rate as the tear volume secreted within a unit of time. Using the captured videos, we measured the tear area using imagej (version 1.38x; Wayne Rasband National Institutes of Health, Bethesda, MD, USA) 1.00 seconds after removing the fluorescein paper strip from the opening (Fig. 2). Considering that normal tear film thickness on the cornea is ~7 μm (Mishima 1965), we arbitrarily defined tear film thickness on the palpebral lobe as 10 μm (0.010 mm). Tear volume was calculated as the area covered with tears × 0.010 mm. The sum of the tear volume from each opening after 1.00 seconds was considered to equal the tear volume secreted from palpebral lobe after 1.00 seconds. The unit of tear flow rate was converted to microlitres per minute.

We compared the location, number of excretory openings and tear flow rate between the dry eye and healthy-subject groups. We analysed the correlation between the Schirmer I test value and the tear flow rate in each group.

Repeatability analysis

We performed a repeatability analysis for 17 eyes of 10 subjects in the healthy-subject group. To calculate the tear flow rate, we measured the washed-out area 1.00 seconds after removing the fluorescein paper strip from the orifice. Because we usually repeated the blocking and opening of the orifices several times during a test, we could measure the tear flow rate several times in one session. We chose 17 eyes of 10 subjects in the healthy-subject group in whom we could measure the tear flow rate more than three times during a session. We measure the tear flow rate three times at the early phase, middle phase and last phase of the tests. We defined the early phase as the first 10 seconds from the beginning, the middle phase from 10 to 20 seconds and the last phase from 20 and 30 seconds. We performed a repeatability analysis by calculating within-subject standard deviation (S_w), coefficient of repeatability (2.77 S_w), coefficient of variance (S_w: mean ratio, %) and intraclass correlation coefficient (Aramberri et al. 2012). To determine the interday variation, we performed repeatability analysis for 14 eyes of eight subjects who had a second test after an interval of >7 days.

Wolfring glands

We performed the same tests for the Wolfring glands, accessory lacrimal glands, in 13 patients with dry eye syndrome and 16 healthy subjects. To observe the tear secretion from the Wolfring glands, we everted the upper lids, applied fluorescein dye to the upper margin of the tarsus (lower margin in the photographs) and observed tear secretion under cobalt blue light, capturing it in photographs and videos (×10 or ×16). When tear secretion from Wolfring glands was observed, we photographed the same area under white light. Further, we measured the tear flow rate from the Wolfring glands using the same methods used for the palpebral lobes of the main lacrimal glands. We compared the proportion of tear secretion from the Wolfring glands between the dry eye and healthy-subject groups. We compared the flow rate from the Wolfring glands between the dry eye and healthy-subject groups.

Statistical analysis

Student's t-test or the Mann–Whitney U-test was used to compare age, Schirmer I test value, tBUT value, corneal staining, number of openings and flow rate between the dry eye and healthy-subject groups. Fisher's exact test was used to compare the sex, location of openings and proportion of tear secretions from the Wolfring glands between the dry eye and healthy-subject groups. Spearman's test was used for correlation analysis between the Schirmer I test value and tear flow rate in the dry eye and healthy-subject groups. We used spss 19.0 (SPSS Inc, Chicago, IL, USA) for the statistical analysis. Results were considered statistically significant if the p value was <0.05.

Palpebral lobe of the main lacrimal gland

Healthy subject

Subject A was a 39-year-old female without dry eye syndrome. The Schirmer I test of the left eye was 19 mm. Under cobalt blue light, aqueous tears flowed out actively from the excretory duct openings. Three openings were observed at the ridge of the palpebral lobe (Fig. 3, Video S2). The tear flow rate from each opening was 0.18, 0.45 and 0.34 μl/min. The total tear flow rate from the palpebral lobe was 0.97 μl/min. In the photographs of another subject, the excretory opening was clearly observed under the cobalt blue light, but it was not visible under the white light of the slit lamp (Fig. 4). In the OCT image of another subject, the longitudinal excretory duct connected to the opening was clearly visible (Fig. 5).

Dry eye syndrome with decreased tear secretion

Subject B was a 59-year-old female with dry eye syndrome. The Schirmer I test value of the left eye was 5 mm, tBUT was 5 seconds and the corneal stain score was 1. Under cobalt blue light, the aqueous tear flow was very slow from one or two openings (Fig. 6, Video S3). The total flow rate from the palpebral lobe was 0.05 μl/min.

Brain infarction without tear secretion

Subject C was a 62-year-old male with a history of brain infarction. His left eye expressed no natural tears. The Schirmer I test value of the left eye was <5 mm. Under cobalt blue light, tear secretion was not observed at all for 20 seconds (Fig. 7, Video S4). The tear flow rate was 0.00 μl/min.

Table 2 shows the number and location of the openings and the flow rate in the dry eye patients and healthy subjects included in the study.

Table 2. Lacrimal gland secretion data of dry eye syndrome and healthy-subject group

	Total	Dry eye syndrome	Healthy subject	p-Value
Number of openings	2.9	2.5 ± 1.6 (0–10)	3.4 ± 2.1 (0–6)	0.031a
Location of opening
Ridge (number of eyes)	80	43	37
Ridge and lateral Canthus (number of eyes)	6	2	4	0.418b
Tear secretion per palpebral lobe (μl/min)	0.67	0.45 ± 0.71	0.91 ± 1.31	0.046a
Tear secretion per opening (μl/min)	0.21	0.19 ± 0.25	0.23 ± 0.26	0.479a

• ^a Student's t-test.
• ^b Fisher's exact test.

Number of excretory openings

In some subjects or patients, we observed no openings. The maximum number of openings was 10 (Video S5). The mean number of openings was 2.5 ± 1.6 in the dry eye group and 3.4 ± 2.1 in the healthy-subject group. The difference between groups was statistically significant (p = 0.031).

Location of excretory openings

Most of the excretory openings on the palpebral lobe were observed at the ridge, the most protruding portion of the palpebral lobe. Sometimes they were located in a row, and sometimes not. Some openings were found at the lateral canthus (Fig. 8, Video S6). In two of 45 eyes in the dry eye group and four of 41 eyes in the healthy-subject group, the openings were in the lateral canthus. The difference in the location between groups, however, was not significant (p = 0.418, Fisher's exact test).

Tear flow rate

The mean flow rate from the palpebral lobe was 0.45 ± 0.71 μl/min in the dry eye group and 0.91 ± 1.31 μl/min in the healthy-subject group. The difference between groups was statistically significant (p = 0.046). The tear secretion per opening was 0.19 ± 0.25 μl/min in the dry eye group and 0.23 ± 0.26 μl/min in the healthy-subject group. The difference between groups was not statistically significant (p = 0.479). Sometimes, the tears first flowed out very slowly, and later, a large amount of tears flowed out abruptly. On the other hand, the tears sometimes initially flowed out well and then the tear flow stopped. Figure 9 shows the correlation between the Schirmer test I and tear flow rate, but the correlation was not statistically significant in either group (Spearman’ ρ = 0.263, p = 0.080 in the dry eye group; Spearman’ ρ = 0.097, p = 0.569 in the healthy-subject group).

Repeatability analysis

Repeatability analysis for three consecutive measurements for 17 eyes of 10 subjects in the healthy-subject group showed that the mean flow rate was 0.55 μl/min, within-subject standard deviation was 0.45 μl/min, repeatability coefficient was 1.26 μl/min, coefficient of variance was 83.3%, and intraclass correlation was 0.830 (95% confidence interval 0.607–0.936; p =0.000). Repeatability analysis for two visit measurements for 14 eyes of eight dry eye patients or healthy subjects showed that the mean tear flow rate was 0.39 μl/min, the within-subject deviation was 0.50 μl/min, the repeatability coefficient was 1.38 μl/min, the coefficient of variation was 128.0%, and the intraclass correlation was 0.365 (95% confidence interval 0.980–0.796; p = 0.212).

Accessory lacrimal glands: Wolfring glands

In 24 eyes of 13 dry eye patients and 25 eyes of 16 healthy subjects, we examined the Wolfring glands by everting the upper lid (Table 3). In five eyes in the dry eye group and seven eyes in healthy-subject group, we observed tearing from the Wolfring glands. The difference between the two groups was not statistically significant (p = 0.753). At the upper margin of the tarsal plate, the tears flowed out from one or two openings of the Wolfring glands (Video S7). The mean flow rate from the Wolfring glands of the dry eye group was 0.007 ± 0.019 μl/min (mean ± standard deviation) and that of the healthy-subject group was 0.009 ± 0.027 μl/min (mean ± standard deviation). The difference between groups was not statistically significant (p = 0.615). The flow rate from the Wolfring glands was 1–2% of the flow rate from the palpebral lobe. We could not find any specific findings at the same site under white light (Fig. 10).

Our results compared to Bron's study

In 1986, Bron visualized both the ductular orifices of the human lacrimal gland and lacrimal fluid secretion after the instillation of 2% sodium fluorescein (Bron 1986). The remarkable differences from the study are as follows: we quantified the tear flow rate by image analysis, compared the flow rate between healthy-subject group and dry eye syndrome patient group, analysed the relationship between Schirmer I test and tear flow rate and found the tear flow from Wolfring glands.

Limitations of Schirmer's test

The Schirmer test is the most commonly used technique for measuring tear secretion. The Schirmer I test is classically performed without anaesthesia and measures reflex tearing. Jones popularized the use of topical anaesthesia with the Schirmer test as a measurement of basal tear secretion, independent of reflex tearing (Jones 1966). The Schirmer test continues to be a simple and inexpensive method of measuring aqueous tear production. The Schirmer test, however, has been criticized for its variability and poor reproducibility. Because the Schirmer test measures standing tears in the inferior fornix, it depends not only on the tear flow rate from the lacrimal gland, but also on the lacrimal drainage system and water evaporation. If there is nasolacrimal duct obstruction and the tear meniscus is high, the Schirmer test may indicate a normal value even if the flow rate from the lacrimal glands is low. The Schirmer test value is not the actual tear flow rate (i.e. tear volume secreted during over a period of time), but it is a semiquantitative value. Using the present method, we directly observed tear secretion from the palpebral lobe of the lacrimal glands. We measured the tear flow rate quantitatively and measured the tear flow rate irrespective of the lacrimal drainage system.

Number of excretory openings

In this study, the mean number of excretory openings was 2.5 in the dry eye group and 3.4 in the healthy-subject group. A previous study reported a mean of 12 excretory openings from the main lacrimal glands in the healthy subjects (Obata 2006). Of the 12, two to five originated from the orbital lobe and six to eight from the palpebral lobe. The ductules from the orbital portion of the lacrimal gland pass through the parenchyma of the palpebral lobe before exiting into the superotemporal portion of the conjunctival fornix (Bedrossian 2012).

This visualization method does not allow us to determine whether an excretory opening on the palpebral lobe is connected to the excretory duct of the palpebral lobe or the excretory duct of the orbital lobe passing through the palpebral lobe. Because the mean tear flow rate (0.99 μl/min) of the healthy subjects in this study was comparable to the normal tear flow rate measured by other investigators (Mishima et al. 1966), we think that the tear flow from the openings on the palpebral lobes comes not only from the palpebral lobe but also from the orbital lobe. We speculate that the number of excretory openings observed in this study represents the number of ‘functioning’ openings that secrete aqueous tears rather than the number of anatomic openings. The remaining anatomic openings are ‘non-functioning’ openings that can be detected only by OCT.

Location of excretory openings

The majority of the excretory openings on the palpebral lobe were observed at the ridge. Excretory ducts coming from the palpebral and orbital lobes were previously thought to open into just the ‘superior conjunctival fornix’ (Obata 2006). In approximately 7% of the subjects, we observed openings in the lateral canthus. In another study, one or two of 12 excretory ducts opened near the lateral canthus main lacrimal gland (Duke-Elder & Wybar 1961). Sanderson and Stasior found islands of ‘peripheral’ lacrimal glandular tissue below the lateral canthus that were not associated with the main gland in 60% of cadaver specimens (Bedrossian 2012). Therefore, cosmetic lateral canthoplasty to these eyes may lead to abnormal tearing after surgery. There are some case reports of abnormal tearing after cosmetic lateral canthoplasty (Ahn et al. 2013; Leelapatranurak et al. 2013). The present test could thus be used to screen for abnormal tearing after cosmetic lateral canthoplasty.

Tear clearance and tear flow rate measurement

New methods for tear volume, tear clearance and tear flow rate measurement were recently published. Bandlitz et al. (2014) reported that a portable digital meniscometer and OCT measurements of the tear meniscus radius were significantly correlated, suggesting that the new portable digital meniscometer could be a useful surrogate for OCT in this respect. Zheng et al. (2014) introduced a new method of measuring early-phase tear clearance by anterior segment OCT. They reported that anterior segment OCT could be used as a rapid, non-invasive and quantitative method of determining early-phase tear clearance rate in a healthy population. Yamaguchi et al. (2014) investigated the changes in the tear flow velocities caused by ageing and concluded that the polymethylmethacrylate particles in a fluorescein solution technique was a simple method of examining Krehbiel flow of tears and might be used for quantitative evaluation of functional nasolacrimal duct obstruction. Our study, however, is the first to measure direct tear flow rate from the lacrimal gland with fluorescein.

Tear flow rate

We measured the quantitative tear flow rate. In some previous studies measuring the tear flow rate, fluorescein dye was applied to the eyes and fluorescein concentration in the tears was measured in the cul-de-sac (Mishima et al. 1966; Scherz et al. 1974). In one study, the normal flow rate was 1.2 μl/min (0.5–2.2 μl/min) (Mishima et al. 1966). In another study, the tear flow rate in the dry eye group was 0–0.2 μl/min (Scherz et al. 1974). In the present study, the mean flow rate from the palpebral lobe was 0.45 μl/min in the dry eye group and 0.91 μl/min in the healthy-subject group. In the present study, we measured not only total tear flow rate from the palpebral lobe, but also the flow rate from each excretory opening. The mean flow rate from an opening was 0.19 μl/min in dry eye group and 0.23 μl/min in the healthy-subject group. Previous studies could not distinguish tear flow between the main lacrimal glands (palpebral lobe and the accessory lacrimal gland; Mishima et al. 1966; Scherz et al. 1974), whereas we measured them separately.

Inaccuracy of tear flow rate measurement

In some aspects, the flow rate measured in the present study was not precise. First, we could not measure the tear film thickness on the surface of the palpebral lobe. In this study, we arbitrarily defined the thickness as 10 μm, but when there are few tears on the surface of the palpebral lobe, the thickness may be thinner than 10 μm, and when there is already a large amount of tears on the surface of the palpebral lobe, the thickness may be much >10 μm. Therefore, we may measure only the washed-out area instead of calculating the tear volume. We cannot, however, compare this area with the previous tear flow rate measured by other investigators. We calculated the volume by multiplying the same value (10 μm), so it is equivalent to measuring only the washed-out area. We are now seeking a method to measure tear film thickness around the tear-secreting orifices in real time. Second, the surface of the palpebral lobe is curved, so the measurement of the area covered with tears was not exact. We measured the quantitative tear flow rate, but in some situations, a simple grading system (e.g. no tearing, slow, normal or fast) may be more practical clinically.

Schirmer test and tear flow rate

The Schirmer test I and tear flow rate were not significantly correlated in either the dry eye group or healthy-subject group in our study. The high variability and low reproducibility of Schirmer test I might be the cause of this discrepancy. The variability is due to the different degrees of reflexes elicited by the Schirmer's test, which cannot be standardized or even controlled when anaesthesia is used. In addition, because the present test also showed low repeatability, the correlation with Schirmer's test might be even weaker. We did not compare variability and reproducibility of the Schirmer's test and the present test in this study. Further studies are needed compare the reproducibility between the two tests. Second, the Schirmer test I reflects tear secretion from the main lacrimal glands and accessory lacrimal gland, but the present test reflects tear secretion from only the palpebral lobe of the main lacrimal glands. This difference might account for the lack of a correlation between Schirmer test I and tear flow rate by this test. On the other hand, the flow rate from Wolfring glands is low compared to the flow rate from the palpebral lobes, and thus, it may not significantly affect the correlation between them. Although the Schirmer test I did not correlate significantly with the tear flow rate in the dry eye and healthy-subject groups, the values obtained in both the Schirmer test I and tear flow rate were lower in the dry eye group than in the healthy-subject group.

Role of Wolfring glands in basal secretion

The tear flow rate from the palpebral lobe was 0.45 μl/min in the dry eye group and 0.91 μl/min in the healthy-subject group. The tear flow rate from the Wolfring glands, the accessory lacrimal glands, was 0.007 μl/min in the dry eye group and 0.009 μl/min in the healthy-subject group. The Wolfring glands provided approximately 1–2% of the volume of tears to the palpebral lobe. The contributions of Wolfring glands to tear film homoeostasis may be negligible. This is the first study in which the ratio of tear flow was measured from the main lacrimal gland (palpebral lobe) and the accessory lacrimal glands. Jones termed the accessory lacrimal glands ‘basal secretors’ because they do not possess direct secretory motor fibres (Jones 1973). The reflex secretors are the main lacrimal glands. Reflex secretion provides additional secretion by peripheral sensory (5th nerve efferents, 7th nerve afferents), retinal or psychogenic stimulation. In this study, the tear flow rate from the Wolfring glands was too low compared to the normal tear flow rate, 1.2 μl/min, so we speculate that in basal secretion, tear secretion from only the accessory lacrimal gland is not sufficient, and tear secretion from the main lacrimal glands is more important. This speculation challenges the widely accepted concept that accessory lacrimal glands such as Wolfring glands are important for basal tear secretion. It is also possible that tear secretion from the conjunctiva contributes significantly to basal secretion.

Anaesthesia before test

We performed the Schirmer test and the present test without anaesthesia. Reflex tears come only from the main lacrimal gland, and basal secretion comes from the accessory gland (Jones 1973; Serin et al. 2007). With anaesthesia, reflex tearing from the main lacrimal gland is suppressed and only basal secretion from the accessory lacrimal gland occurs (Jones 1973; Serin et al. 2007). Because this test measures the function of the palpebral lobe of the main lacrimal glands, however, we did not apply topical anaesthesia before the tests. As described above, however, we speculate that in basal secretion, tear secretion from the main lacrimal glands is more important than that from the accessory gland. To decrease the variability and increase the reproducibility of this test, we think topical anaesthesia should be applied beforehand.

Applications

This test has several potential applications. First, proteins, cytokines and osmolarity of aqueous tears can be analysed directly from excretory openings on the palpebral lobe. Investigators conventionally collect standing tears from the inferior fornix. Tears in the fornix, however, may contain several substances secreted from conjunctiva, and some water from the tears may evaporate. Therefore, the tears in the fornix likely differ from the tears collected directly from the excretory openings on the palpebral lobes. In some patients with dry eye syndrome, the amount of tears in the fornix is too small to collect (Nelson & Wright 1986). If tears from the palpebral lobes are directly collected and analysed, these problems may be addressed. Tears can be easily collected from the palpebral lobes by touching the polyester fibre rod at the tear-secreting orifice after applying fluorescein dye. We are now performing this experimental study. Second, this test can be used as a Sjögren diagnosis criterion. Sjögren syndrome is an autoimmune disease that causes dry eye syndrome, dry mouth and arthritis. To diagnose Sjögren syndrome, ocular examination is necessary. According to Sjögren criteria, the Schirmer test must be <5 mm and the staining of the ocular surface according to the van Bijsterveld scoring system must be equal or higher than 4 (Vitali et al. 2002). The Schirmer test has some limitations, however, as described above. Because the present test allows for direct observation of tear secretion from the palpebral lobe and measurement of the tear flow rate, it precisely reflects main lacrimal gland dysfunction. If large-scale validation studies of this test are performed, the test could be used as a new diagnostic test for Sjögren disease.

Advantage of the test

The described visualization technique has several advantages. First, because tear secretion from the palpebral lobe of the main lacrimal glands can be observed directly, lacrimal gland function can be precisely determined. Second, flow rate from the lacrimal glands can be measured quantitatively. The flow rate can be measured from an excretory opening and from the entire palpebral lobe by capturing and analysing movies of the tear secretion. Third, flow rate can be measured from the palpebral lobe of the main lacrimal and accessory lacrimal glands allowing us to determine which region contributes more to dry eye syndrome or to basal secretion. Fourth, any ophthalmologist can easily perform this test with only fluorescein paper. For quantitative analysis, a camera connected to a biomicroscope is necessary. Videos of the test can be saved using free screen capture software, and the data can be quantitatively analysed using image analysis software such as ImageJ.

In conclusion, by applying the described method, we directly observed tear secretion from excretory openings on the palpebral lobe of main lacrimal glands and Wolfring glands. We were able to quantitatively measure the actual tear flow rate from the lacrimal glands. This test could potentially be applied clinically as well as for studies of dry eye syndrome and lacrimal gland diseases.