The Association Between Ocular Surface Measurements with Visual Field Reliability Indices and Gaze Tracking Results in Preperime

Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com BJO Online First, published on August 1, 2017 as 10.1136/bjophthalmol-2017-310309 Clinical science The association between ocular surface measurements with visual field reliability indices and gaze tracking results in preperimetric glaucoma Takahiro Arai, Hiroshi Murata, Masato Matsuura, Tomohiko Usui, Ryo Asaoka

Department of Ophthalmology, Abstract when the patient fails to respond to a more intense The University of Tokyo, Tokyo, Background/aims To investigate the relationship stimulus than one previously perceived at the Japan between gaze tracking (GT) results and ocular surface same location. All three VF reliability indices are condition in glaucoma. important clinical markers. Elevated FLs can mask Correspondence to 1 Dr Ryo Asaoka, Department of Method The Humphrey 24–2 visual field (VF) asw early scotoma, increased FP errors indicate ‘trig- Ophthalmology, The University measured in 34 eyes of 30 patients with open-angle ger-happy’ patients, while a high FN rate may of Tokyo, Graduate School of glaucoma without VF damage. Tear break-up time, suggest patient inattention or fatigue during the Medicine, Tokyo, 113-8655, Schirmer’s test, tear meniscus volume (TMV) and 2 Japan; r asaoka-tky@umin. ac.jp VF examination. Previous studies have reported presence of superficial punctate keratopathy (SPK) were the usefulness of these indices3; however, their Received 7 February 2017 also measured in order to describe the condition of the limitations have also been reported in more recent Revised 8 June 2017 ocular surface. Various GT parameters were calculated: studies. For instance, FLs can be increased by Accepted 2 July 2017 the average frequency of eye movements per stimulus mislocalisation of the blind spot,4 and fixational between 1° and 2° (move ), the average frequency of 1-2 instability can be experienced even in well-trained eye movements per stimulus between 3° and 5° (move 5 3- examinees. Furthermore, a high FN rate is signifi- ), the average frequency of eye movements per stimulus 5 cantly associated with the amount of field loss, more than 6° (move ), the average tracking failure ≥6 rather than threshold reproducibility,6 and there- frequency per stimulus (TFF) and the average blinking fore this index is no longer used in the HFA. frequency. The relationship between GT parameters, reliability indices and ocular surface measurements was We have investigated the usefulness of gaze tracking (GT) in assessing the reproducibility7 and investigated using linear mixed modelling. 8 9 Results SPK was positively associated with high rates overestimation/underestimation of the VF. GT is of move (coefficient=0.12 for SPK+, p=0.003) and a record of eye movement monitored throughout 3-5 the VF measurement.10 Currently, the GT results move≥6 (coefficient=0.052 for SPK+, p=0.023). High TMV was significantly related to TFF (coefficient=0.37, are represented as a printed line diagram at the p=0.023). Fixation losses, false-positives and false- bottom of the VF printout and can only be subjec- negatives were not significantly associated with any GT tively evaluated by clinicians. In our prior studies, parameters or ocular surface measurements. GT results were evaluated objectively and quanti- Conclusion SPK is associated with increased frequency tatively.7–9 However, GT uses a record of infrared video recording of the fixation stimulus (corneal of eye movements (move3-5 and move≥6). In addition, large TMV is associated with increased rate of TFF. reflex), and hence it may be largely influenced by Careful attention should be paid when interpreting GT the scattering of light associated with the condi- parameters in patients with SPK or a large TMV. tion of the corneal surface, such as superficial punctate keratopathy (SPK) and an abnormal tear film state. Indeed, examples of poor GT due to dry eye have been reported11; however, no study has Introduction quantitatively analysed the relationship between Reliable visual field (VF) results are imperative GT and dry eye. Thus, the purpose of the current for the accurate diagnosis of glaucoma and early study is to investigate the relationship between GT detection of disease progression. Fixation losses results and ocular surface condition in patients (FLs), false-positives (FPs) and false-negatives with glaucoma. The purpose of the current (FNs) are measures of reliability implemented study was to investigate the effect of dry eye to in the Humphrey Field Analyzer (HFA, Carl GT parameters in glaucomatous eyes and not in Zeiss Meditec, Dublin, California, USA). FLs are normative subjects. However, as GT parameters recorded when the patient perceives a stimulus that are significantly related to VF deterioration,8 9 is presented in the physiological blind spot. FPs were traditionally measured using catch trials, and only glaucomatous eyes — with no measurable VF a FP would be recorded when the patient reported damage — were analysed in the current study. We To cite: Arai T, Murata H, seeing a stimulus when one was not presented; recently reported that, even if a VF criterion for Matsuura M, et al. however, in the Swedish Interactive Threshold VF damage is not satisfied, VFs of glaucomatous Br J Ophthalmol Published 12 Algorithms (SITA), FPs are measured through the eyes are different from those from normal eyes. Online First: [please include 13 Day Month Year]. use of ‘listening windows’. These windows are Thus, only ‘preperimetric’ glaucomatous VFs doi:10.1136/ intervals between stimulus presentations when no were included in the current study to mitigate the bjophthalmol-2017-310309 patient response is anticipated. FNs are recorded effects of obvious VF deterioration.

Arai T, et al. Br J Ophthalmol 2017;0:1–6. doi:10.1136/bjophthalmol-2017-310309 1 Copyright Article author (or their employer) 2017. Produced by BMJ Publishing Group Ltd under licence. Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com Clinical science

average blinking frequency (BF), the average frequency of eye

movement per stimulus between 1° and 2° (denoted by move1-

2), 3° and 5° (denoted by move3-5) and more than 6° (denoted

by move≥6). Total magnitude of eye movement was calculated as the movement degree × the average frequency of eye movement (denoted by total move).

Ocular surface condition measurements Prior to VF measurements, SPK was assessed by fluorescein staining.15 The area of SPK was estimated in four gradations: 0 (no SPK), 1 (less than one-third of cornea), 2 (between one-third and two-thirds of cornea) and 3 (larger than two-thirds of cornea). The density of SPK was graded as 0 (no SPK was present), 1 (SPK was sparse), 2 (moderately dense) and 3 (SPK was high and the lesions overlapped). As the current study included only Figure 1 Illustrated gaze tracking record. An upward bar in the mild dry eye cases (only one eye had an area score larger than 2 chart indicates fixation disparity and the length of the bar represents and only three eyes had density scores larger than 2), SPK was the magnitude of disparity. A short downward bar represents tracking defined in a binary fashion: SPK− (SPK area and density scores failure, while a long downward bar indicates eyelid closure. Gaze were 0) or SPK+ (SPK area or density score was >0). Then, tracking parameters were calculated as follows: average TFF per tear break-up time (TBUT) was measured three times and the stimulus, the average frequency of eye movement per stimulus between average value was calculated. Alcon Schirmer Standard Tear Test Strips (Fort Worth, Texas, USA) were used to assess anaesthe- 1° and 2° (denoted by move1-2), 3° and 5° (denoted by move3-5) and tised tear secretion. In addition to these measurements, after the more than 6° (denoted by move≥6). BF, blinking frequency; TFF, tracking failure frequency. VF test, tear meniscus volume (TMV) was measured using anterior segment optical coherence tomography (SS-1000; Tomey, Tokyo, Japan), following our previous reports.16–18 Details of Methods the TMV measurement are described elsewhere.16–18 In short, The study was approved by the Research Ethics Committee of the TMV value was calculated automatically with the integrated the Graduate School of Medicine and Faculty of Medicine at tear meniscus area values of the consecutive central 11 images, The University of Tokyo. Written consent was given by patients where the TM area value was defined as an inferior triangular for their information to be stored in the hospital database and TM area formed by the corneal anterior boundary, the anterior used for research. This study was performed according to the boundary of the lower eyelid and the anterior borderline of the tenets of the Declaration of Helsinki. TM.

Subjects Statistical analysis Thirty-four eyes of 30 patients with open-angle glaucoma (16 The relationships between GT measurements, FLs FPs and FNs male and 14 female) were prospectively included in the study. and ocular surface measurements were first analysed using the All patients visiting the glaucoma clinic at The University of Wilcoxon test for comparing numeric values between two groups Tokyo Hospital during the period of February 2016–July 2016 and also Spearman test for analysing the relationship between who fulfilled the following criteria were included in the current two numerical values. Linear mixed modelling, whereby eyes study: (1) glaucomatous changes in the optic nerve head with or were nested in each patient, was also applied. Standard linear without a retinal nerve fibre layer defect as confirmed by a glau- regression analysis makes the assumption that all observations coma specialist (HM or RA); (2) glaucoma was the only disease causing VF damage; (3) patients were followed for at least 6 months at The University of Tokyo Hospital; and (4) no glau- Table 1 Subject demographics comatous VF defect, which was defined as a cluster of three or more points in the pattern deviation plot within a single hemi- Variables Value field (superior or inferior) with p values <5%, one of which Age (years), mean±SD (range) 62.8±12.4 (31–82) must have a p value <1%, and/or Glaucoma Hemifield Test Gender (male:female) 16:14 results outside normal limits and/or abnormal pattern SD with Eye (right:left) 25:9 p <5%.14 All VFs were measured using the HFA (24–2, SITA MD (dB), mean±SD (range) −0.5±1.5 (−4.6 to 3.9) standard program), and those with FL >50%, FP >50% or FL (%), mean±SD (range) 6.5±7.7 (0.0–28.6) 7 9 FN >50% were excluded following our previous report. FP (%), mean±SD (range) 2.1±2.8 (0.0–12.0) FN (%), mean±SD (range) 0.2±0.7 (0.0–3.0)

GT measurements move1-2 (per stimulus), mean±SD (range) 0.70±0.14 (0.15–0.92) The GT system monitors patients’ gaze position at each stim- move3-5 (per stimulus), mean±SD (range) 0.12±0.14 (0.0044–0.66) ulus presentation. As detailed elsewhere7 and summarised in move≥6 (per stimulus), mean±SD (range) 0.02±0.03 (0.0–0.11) figure 1, fixation disparity is represented by an upward bar with TFF (per stimulus), mean±SD (range) 0.02±0.03 (0.0–0.15) the length of the bar showing the magnitude of disparity. A short BF (per stimulus), mean±SD (range) 0.03±0.10 (0.0–0.59) downward bar represents tracking failure, and a long downward BF, average blinking frequency; FL, fixation loss; FN, false-negative; FP, false- bar indicates eyelid closure. Following our previous reports,7–9 positive; MD, mean deviation; move1-2, average frequency of eye movement per GT data were exported as JPEG images from the Beeline (Tokyo, stimulus between 1° and 2°, move , average frequency of eye movement per Japan) data filing system. The following GT parameters were 3-5 stimulus between 3° and 5°; move≥6, average frequency of eye movement per calculated: average tracking failure frequency per stimulus (TFF), stimulus more than 6°; TFF, average tracking failure frequency per stimulus.

2 Arai T, et al. Br J Ophthalmol 2017;0:1–6. doi:10.1136/bjophthalmol-2017-310309 Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com Clinical science p Value SE F oefficient B C p Value

Figure 2 Histograms of TBUT, Schirmer’s test result and TMV. (Top left) TBUT. (Top right) Schirmer’s test result. (Bottom right) TMV. TBUT, tear

break-up time; TMV, tear meniscus volume. average frequency of eye movement per stimulus between 3° and 5°, , SE 3-5 are independent of each other; however, in the current study, both eyes are nested within patients. Ignoring this grouping of oefficient FF T the measurements will result in the underestimation of SEs of C regression coefficients, but the linear mixed model adjusts for the hierarchical structure of the data, modelling in a way in which measurements are grouped within subjects.

All analyses were performed using the statistical programming p Value language R V.3.1.3 (The Foundation for Statistical Computing, Vienna, Austria).

Results SE Table 1 summarises the study patients’ characteristics. Twen- ty-four eyes in 21 patients had SPK and 10 eyes in 10 patients ≥6 did not have SPK. Figure 2 shows the histograms of TBUT, oefficient Move Schirmer’s test result and TMV. Thirty-one eyes in 28 patients C had a TBUT shorter than 10 s, and 17 eyes in 14 patients had a Schirmer’s test result shorter than 5 mm. TBUT was not significantly different between SPK− and p Value SPK+ groups (p=0.94, Wilcoxon-Mann-Whitney test). Simi- move average frequency of eye movement per stimulus between 1° and 2°; , larly, Schirmer’s test result and TMV were not significantly 1-2 different between SPK− and SPK+ groups (p=0.79 and 0.72, respectively, Wilcoxon-Mann-Whitney test). TBUT was significantly related to Schirmer’s test result (coefficient=0.30, SE p=0.041, respectively, Spearman test) but not TMV (p=0.91,

Spearman test). TMV and Schirmer’s test result were not signifi- 3-5 cantly associated (p=0.91, Spearman test). As shown in table 2, oefficient Move similar results were obtained with the linear mixed model; TBUT C was not significantly different between SPK− and SPK+ groups (p=0.85, linear mixed model). Similarly, Schirmer’s test result gaze tracking parameters and various false-positive and false-negative rates, and TMV were not significantly different between SPK− and p Value

Table 2 Comparison of TBUT, Schirmer’s test result and TMV in eyes with and without SPK

SPK+ (n=21) SPK− (n=10) p Value SE TBUT (s), mean±SD 6.6±2.2 (3.0–11.0) 6.8±3.6 (3.0–13.7) 0.85

(range) 1-2

Schirmer’s test result 5.3±3.4 (1–16) 5.6±4.7 (1–16) 0.82 oefficient Move C (mm), mean±SD (range) −0.0046 0.10 0.96 −0.062 0.12 0.62 −0.078 0.23 0.74 −0.28 0.44 0.53 0.04 0.13 0.78 Relationship between fixation loss, 3

TMV (mm ), mean±SD 0.034±0.067 0.016±0.0092 0.41

(range) (0.0016–0.32) (0.0016–0.029) average tracking failure frequency per stimulus. TFF, average frequency of eye movement per stimulus more than 6°; , ≥6

SPK, superficial punctate keratopathy; TBUT, tear break-up time; TMV, tear meniscus able 3 move BF, average blinking frequency; FL, fixation loss; FN, false-negative; FP, false-positive; move false-positive; FP, false-negative; FN, fixation loss; FL, average blinking frequency; BF, FL (%) FP (%) −0.058 0.03 0.09 −0.0051 0.05 0.91 0.09 0.08 0.30 −0.019 0.16 0.90 −0.0031 0.05 0.95 FN (%) 0.00 0.01 0.83 0.00 0.01 0.83 −0.013 0.02 0.49 0.02 0.04 0.53 0.00 0.01 0.72 volume. T

Arai T, et al. Br J Ophthalmol 2017;0:1–6. doi:10.1136/bjophthalmol-2017-310309 3 Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com Clinical science

SPK+ groups (p=0.82 and p=0.41, respectively, linear mixed model). TBUT was significantly related to Schirmer’s test result

p Value (coefficient=0.30, p=0.011, respectively, linear mixed model) but not TMV (p=0.72, linear mixed model). TMV and Schirm- er’s test result were not significantly associated (p=0.44, linear

SE mixed model). Table 3 represents the relationship between FL, FP and FN, and the predictors of age, gender, TBUT, Schirmer’s test result, TMV and SPK. No reliability indices were associated with these F oefficient B C parameters (p>0.05, linear mixed model). Table 4 shows the relationship between each GT param- eter and the predictors of age, gender, TBUT, Schirmer’s test p Value

, average frequency of eye movement per , result, TMV and SPK; the results are summarised here for each ≥6 outcome.

►► Move1-2: none of the predictors were significantly associated.

SE ►► Move3-5: only SPK (coefficient=0.12 for SPK+, p=0.0025) was significantly associated.

►► Move≥6: only SPK (coefficient=0.052 for SPK+, p=0.023) was significantly associated. oefficient FF T C ►► Total move: only SPK (coefficient=0.46 for SPK+, p=0.023) was significantly associated. ►► TFF: only TMV (coefficient=0.37, p=0.023) was

p Value significantly associated. ►► BF: none of the predictors were significantly associated. Table 5 shows the relationship between the outcome variables of

SE FLs, FPs, FNs and the predictor variables of age, gender, TBUT, Schirmer’s test result, TMV and SPK. None of the predictors were significantly related to any of the outcomes. Figure 3 illus- trates the GT records of four sample cases. oefficient otal move T C Discussion In the current study, TBUT, Schirmer’s test result, TMV and p Value

, average frequency of eye movement per stimulus between 3° and 5°; move average frequency of eye movement per stimulus between 3° and 5°; , SPK were measured in patients with glaucoma, but without 3-5 glaucomatous VF defects. A number of GT summary parameters were calculated and their relationships with these ocular

SE surface condition measurements were investigated. As a result,

it was suggested that SPK is positively associated with move3-5

≥6 and move≥6. Further, a high TMV was significantly related to TFF. None of the ‘standard’ reliability indices (FLs, FPs and FNs) oefficient Move C were related to any of GT parameters or ocular surface condition measurements. In glaucoma, ocular surface disease is frequently observed

p Value because patients are usually treated with a variety of ocular hypotensive drugs, often in combination. Furthermore, the prevalence of both glaucoma and dry eye increases with increasing

SE age. Many eye-drops that are used to treat glaucoma contain preservatives that can cause ocular surface disease, such as SPK.

3-5 VF measurements may exacerbate patients with dry eye, because they are required to fixate on a bright target for a considerably oefficient Move C long duration. Previous reports have suggested that the prevalence of ocular surface disease is considerably higher in patients

gender, TBUT, Schirmer’s test result, TMV and SPK, and various gaze tracking parameters and various TMV and SPK, test result, Schirmer’s TBUT, gender, with glaucoma compared with the general population, and as

p Value many as half of patients with glaucoma have ocular surface , average frequency of eye movement per stimulus between 1° and 2°; move average frequency of eye movement per stimulus between 1° and 2°; , 19 1-2 disease symptoms or signs related to dry eye. In the current study, only 29.4% (10 eyes) had SPK symptoms; only one eye had SE 0.012 1.0 −0.0033 0.0072 0.65 −0.00053 0.0044 0.9an −0.011 SPK 0.28 area 0.69score 0.001695 of 0.00163, and 0.31 only 0.0070 three 0.0083 eyes 0.41 had an SPK density score of 2 or 3. This may be because only glaucomatous eyes –6

1-2 without glaucomatous VF defects were included in the current study, and intensive intraocular pressure reduction treatments oefficient C Move −0.11 0.057 0.076 0.12 0.03585 0.0025 0.053 0.022 0.023 0.46 014 0.0035 −0.00108 0.0081 0.90 −0.031 0.041 0.45 −0.00024 0.0024 0.92 −0.0014 0.0015 0.33 −0.0010 0.00090 0.28 −0.0076 0.0059 0.21 0.00044 0.00033 0.19 0.00087 0.0017 0.61 0.0016 0.0076 0.83 −0.0073 0.0048 0.14 −0.0038 0.0029 0.21 −0.34 0.0019 0.083 −0.00173 0.0011 0.12 −0.0041 0.0055 0.46 0.025 0.053 0.64 0.00968 0.033 0.77 −0.015 0.020 0.46(such 0.0093 as combination 0.13 0.94 0.011 therapy) 0.0075 0.16are 0.050rarely 0.038prescribed 0.20 in these Relationship between age, ) −0.30 0.47 0.53 0.38 0.29 0.21 0.088 0.18 0.63 1.11 1.21 0.36 0.37 0.067 <0.0001 −0.28 0.34 0.42 3 eyes. In addition, functioning bleb due to trabeculectomy inten- sifies ocular surface disease,20 but it is rarely performed in eyes

able 4 without VF defects (no one in the current study had undergone stimulus more than 6°; SPK, superficial punctate keratopathy; TBUT, tear break-up time; TFF, average tracking failure frequency per stimulus; TMV, tear meniscus volume; total move, calculated as the movement degree × average frequency of move, tear meniscus volume; total TMV, average tracking failure frequency per stimulus; TFF, tear break-up time; TBUT, keratopathy; superficial punctate SPK, stimulus more than 6°; eye movement. BF, average blinking frequency; move average blinking frequency; BF, Presence of SPK Age (years) Schirmer’s Schirmer’s test result (cm) Gender (male) TMV (mm TBUT (s)TBUT 5.2×10 T

4 Arai T, et al. Br J Ophthalmol 2017;0:1–6. doi:10.1136/bjophthalmol-2017-310309 Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com Clinical science

Table 5 Relationship between age, gender, TBUT, Schirmer’s test result, TMV and SPK, and FL, FP and FN rates FL FP FN Coefficient SE p Value Coefficient SE p Value Coefficient SE p Value Age (years) 0.00057 0.0013 0.68 6.7×10–5 0.00046 0.88 0.00012 0.00011 0.30 Gender (male) −0.00079 0.030 0.98 0.0029 0.010 0.78 −0.0013 0.0025 0.59 TBUT (s) −0.0035 0.0065 0.59 0.0022 0.0023 0.33 −0.00040 0.00053 0.45 Schirmer’s test result 0.0023 0.0043 0.60 0.00023 0.0015 0.88 9.4×10–6 0.00035 0.98 TMV (mm3) −0.092 0.27 0.74 −0.081 0.093 0.39 0.0042 0.022 0.85 Presence of SPK −0.011 0.033 0.75 −0.011 0.011 0.32 −0.0010 0.0027 0.70 FL, fixation loss; FN, false-negative; FP, false-positive; SPK, superficial punctate keratopathy; TBUT, tear break-up time; TMV, tear meniscus volume. this procedure). Despite the relatively low prevalence of SPK in TFF were not increased with SPK, which suggests our finding is the studied population, there were obvious symptoms of dry eye, unlikely to be purely due to the scattered corneal reflex. as suggested by a high prevalence of a short TBUT (31 eyes) and BF was not associated with TBUT, Schirmer’s test result, TMV a short Schirmer’s test result (17 eyes). or SPK. Van Went et al23 investigated the corneal sensitivity in A previous study suggested that patients with dry eye exhibit patients with glaucoma or ocular hypertension, and reported that atypical VF defects, such as a defect that does not respect the intraocular pressure-lowering medications, including benzalko- horizontal meridian and/or has a rather central location in any nium chloride, were associated with decreased corneal sensi- quadrant; this is probably because these patients tend to avoid tivity. Furthermore, corneal sensitivity decreases with increasing 21 blinking during the tests. Indeed, the use of artificial tears age. This may explain why increased BF was not correlated with before VF testing improved VF results in patients with glau- ocular surface condition. In addition, Wang et al24 investigated 22 coma and dry eye. In the current study, the presence of SPK the occurrence of blinking during VF measurements, using a

was significantly associated with increased rates of move3-5 video eye-tracker system, and suggested that, for suprathreshold and move≥6, suggesting poor fixation in patients with dry eye stimuli, blinks often occurred after the presentation, although during VF measurements. However, scattered reflex light can there was no relationship to presentation time for subthreshold be induced by the presence of SPK on the cornea. As a result, presentations. The HFA GT measurement is recorded at the time increased rates of move3-5 and move≥6 could actually be an of the target presentation, and this may be another reason why artefact of the scattered corneal reflex. However, move1-2 and BF was not associated with TBUT, Schirmer’s test result, TMV or SPK. There is no well-defined reference range for TMV; however, in normal subjects it has been reported as 0.1327±0.051 mm,3 25which is much larger than observed in the current study, where TMV averaged 0.034 and 0.016 mm3 in the SPK+ and SPK− groups, respectively, and only one eye had a TMV larger than 0.1327 mm3. This difference may be due to the large difference in ages studied because it has been reported TMV decreases with increasing age26; the average age in the prior study was 36.5±6.8 years, while the average age in the current study was 62.8±12.4 years. Furthermore, TMV was measured after VF testing, which may have reduced TMV due to decreased blinking. However, in the current study, high TMV was associated with increased rates of TFF. This implies the GT record is less reliable in eyes with high TMV. We have recently Figure 3 Four example GT records with high TFF, move and move reported that a high TFF rate is associated with poor VF test– 3-5 3-5 retest reproducibility.7 High TMV is also observed in patients rates. Top: A case with high TFF (short downward bars) and TMV: TFF, 16 0.15 per stimulus; TBUT, 7.0 s; Schirmer’s test, 4 mm; TMV, 0.32 mm3; with other ocular disorders. Ohtomo et al compared TMV SPK area, 0; SPK density, 0; FL, 7.1%; FP, 1.0%; FN, 0.0%. Second from before and after dacryocystorhinostomy and showed that TMV top: A case with high TFF and TMV: TFF, 0.071 per stimulus; TBUT, 3.3 s; is decreased after the procedure. The obstruction of the lacrimal Schirmer’s test, 1 mm; TMV, 0.13 mm3; SPK area, 0; SPK density, 0; FL, drainage system is frequently seen in elderly patients and careful 0%; FP, 0%; FN, 2%. Second from bottom: A case with high move and consideration is recommended when interpreting VFs in patients 3-5 with high TMV or a high TFF. move≥6 (short and long upward bars) rates and SPK score: TBUT, 9.0 s; Schirmer’s test, 4 mm; TMV, 0.02 mm3; SPK area, 1; SPK density, 2; FL, The age range of patients in the current study was wide (31–82 years old); however, age was not related to any of the classic 0%; FP, 5%; FN, 0%. Bottom: A case with high move3-5 and move≥6 rates and SPK score: TBUT, 3.0 s; Schirmer’s test, 5 mm; TMV, 0.03 mm3; SPK reliability indices (FLs, FPs and FNs) or GT parameters, and the area, 1; SPK density, 3; FL, 0%; FP, 1%; FN, 0%. BF, average blinking same was also true of gender. However this does not deny a frequency; FL, fixation loss; FN, false negative; FP, false positive; GT, gaze relationship between age and GT parameters, and a different outcome may be observed if a larger population is analysed. tracking; move3-5, average frequency of eye movement per stimulus Nonetheless, it was suggested there is a significant relationship between 3° and 5°; move≥6, average frequency of eye movement per stimulus more than 6°; SPK, superficial punctate keratopathy; TBUT, tear between dry eye and GT parameters in the current study. break-up time; TFF, average tracking failure frequency per stimulus; TMV, In the current study, FLs, FPs and FNs were not related to 27 tear meniscus volume. ocular surface measurements. On the other hand, Yenice et al

Arai T, et al. Br J Ophthalmol 2017;0:1–6. doi:10.1136/bjophthalmol-2017-310309 5 Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com Clinical science reported that the use of artificial tears improved these indices 6 Bengtsson B, Heijl A. False-negative responses in glaucoma perimetry: indicators of in patients with glaucoma and dry eye. These somewhat contra- patient performance or test reliability? Am J Ophthalmol Vis Sci 2000;130:689–2204. 7 Ishiyama Y, Murata H, Mayama C, et al. An objective evaluation of gaze tracking in dictory results could be ascribed to differences in the enrolled Humphrey perimetry and the relation with the reproducibility of visual fields: a pilot patients. In our study, the presence of dry eye was not required. study in glaucoma. Invest Ophthalmol Vis Sci 2014;55:8149–52. As a result, very few patients suffered from severe dry eye, as 8 Ishiyama Y, Murata H, Asaoka R. The usefulness of Gaze Tracking as an index of Visual suggested by the mild SPK scores. Thus, a different outcome may Field reliability in Glaucoma patients. Invest Ophthalmol Vis Sci 2015;56:6233–6. have been observed if the current investigation had been carried 9 Ishiyama Y, Murata H, Hirasawa H, et al. Estimating the usefulness of Humphrey Perimetry Gaze Tracking for evaluating Structure-Function relationship in Glaucoma. out in patients with severe dry eye. It should also be noted that Invest Ophthalmol Vis Sci 2015;56:7801–5. the number of studied eyes was based on a sample size calcu- 10 Humphrey Field Analyzer series 700 serveice guide section 1, 1994. lation to analyse the relationship between dry eye status and 11 Choplin NT, Traverso CE. Atlas of glaucoma third edition. Boca Raton, FL: CRC Press, GT parameters, using linear mixed modelling. The comparison 2014. 12 Asaoka R, Murata H, Iwase A, et al. Detecting preperimetric Glaucoma with between SPK+ and SPK− groups, carried out as a subanalysis, Standard Automated Perimetry using a Deep Learning Classifier. Ophthalmology may require a different sample size; hence the results of the 2016;123:1974–80. comparison between these two groups could be different with a 13 Asaoka R, Iwase A, Hirasawa K, et al. Identifying "preperimetric" glaucoma different sample size. in standard automated perimetry visual fields. Invest Ophthalmol Vis Sci In conclusion, the presence of SPK is associated with increased 2014;55:7814–20. 14 Anderson DR, Vm P. Automated static perimetry. 2nd ed. Louis: StMosby, 1999. rates of move3-5 and move≥6. In addition, high TMV is associated 15 Miyata K, Amano S, Sawa M, et al. A novel grading method for superficial punctate with an increased rate of TFF. Careful consideration is needed keratopathy magnitude and its correlation with corneal epithelial permeability. Arch when assessing GT results in patients with these characteristics. Ophthalmol 2003;121:1537–9. 16 Ohtomo K, Ueta T, Fukuda R, et al. Tear meniscus volume changes in Contributors Conception or design of the work: RA, HM, TU. Acquisition, dacryocystorhinostomy evaluated with quantitative measurement using anterior analysis or interpretation of data: RA, HM, TU, TA, MM. Drafting the work or segment optical coherence tomography. Invest Ophthalmol Vis Sci 2014;55:2057–61. revising: RA, HM, TU, TA, MM. Final approval of the version published: RA, HM, TU, 17 Akiyama R, Usui T, Yamagami S. Diagnosis of Dry Eye by Tear Meniscus Measurements TA, MM. using Anterior Segment Swept Source Optical Coherence Tomography. Cornea Funding This work was supported by Grant 17K11418 from the Ministry of 2015;34(Suppl 11):S115–S120. Education, Culture, Sports, Science, and Technology of Japan and Japan Science and 18 Akiyama-Fukuda R, Usui T, Yoshida T, et al. Evaluation of Tear Meniscus Dynamics Technology Agency (JST) CREST JPMJCR1304. using Anterior Segment Swept-Source Optical Coherence Tomography after Topical solution instillation for Dry Eye. Cornea 2016;35:654–8. Competing interests None declared. 19 Leung EW, Medeiros FA, Weinreb RN. Prevalence of ocular surface disease in Patient consent Obtained. Glaucoma patients. J Glaucoma 2008;17:350–5. 20 Jin SW, Min JS. Clinical evaluation of the effect of diquafosol ophthalmic solution in Ethics approval Ethics Board of Tokyo University. Glaucoma patients with dry eye syndrome. Jpn J Ophthalmol 2016;60:150–5. Provenance and peer review Not commissioned; externally peer reviewed. 21 Hoste A, Salmon J, Thygeson J, et al. Tips archive. A dry eye can hamper visual field © Article author(s) (or their employer(s) unless otherwise stated in the text of the test interpretation. 2012 http://www.eugs.org/ eng/tips_ showtip.asp? id=2110 article) 2017. All rights reserved. No commercial use is permitted unless otherwise 22 Kocabeyoglu S, Mocan MC, Bozkurt B, et al. Effect of artificial tears on automated expressly granted. visual field testing in patients with Glaucoma and dry eye. Can J Ophthalmol 2013;48:110–4. 23 Van Went C, Alalwani H, Brasnu E, et al. Corneal sensitivity in patients treated References medically for glaucoma or ocular hypertension. J Fr Ophtalmol 2011;34:684–90. 1 Vingrys AJ, Demirel S. The effect of fixational loss on perimetric thresholds and 24 Wang Y, Toor SS, Gautam R, et al. Blink frequency and duration during perimetry reliability. Amsterdam: Kugler Publications, 1992. Perimetry Update 1992/93. and their relationship to test-retest threshold variability. Invest Ophthalmol Vis Sci 2 Frankhauser F, Spahr J, Bebie H. Some aspects of the automation of perimetry. Surv 2011;52:4546–50. Ophthalmol 1977;22:131–41. 25 Fukuda R, Usui T, Miyai T, et al. Tear meniscus evaluation by anterior segment swept- 3 Katz J, Sommer A. Screening for glaucomatous visual field loss. the effect of patient source optical coherence tomography. Am J Ophthalmol 2013;155:620–4. 624.e1-2. reliability. Ophthalmology 1990;97:1032–7. 26 Wang J, Palakuru JR, Aquavella JV. Correlations among upper and lower tear 4 Sanabria O, Feuer WJ, Anderson DR. Pseudo-loss of fixation in automated perimetry. menisci, noninvasive tear break-up time, and the Schirmer test. Am J Ophthalmol Ophthalmology 1991;98:76–8. 2008;145:795–800. 5 Demirel S, Vingrys AJ. Eye Movements during Perimetry and the effect that Fixational 27 Yenice O, Temel A, Orüm O. The effect of artificial tear administration on visual field instability has on Perimetric Outcomes. J Glaucoma 1994;3:28–35. testing in patients with glaucoma and dry eye. Eye 2007;21:214–7.

6 Arai T, et al. Br J Ophthalmol 2017;0:1–6. doi:10.1136/bjophthalmol-2017-310309 Downloaded from http://bjo.bmj.com/ on August 28, 2017 - Published by group.bmj.com

The association between ocular surface measurements with visual field reliability indices and gaze tracking results in preperimetric glaucoma Takahiro Arai, Hiroshi Murata, Masato Matsuura, Tomohiko Usui and Ryo Asaoka

Br J Ophthalmol published online August 1, 2017

Updated information and services can be found at: http://bjo.bmj.com/content/early/2017/07/31/bjophthalmol-2017-31030 9

These include: References This article cites 22 articles, 4 of which you can access for free at: http://bjo.bmj.com/content/early/2017/07/31/bjophthalmol-2017-31030 9#BIBL Email alerting Receive free email alerts when new articles cite this article. Sign up in the service box at the top right corner of the online article.

Notes

To request permissions go to: http://group.bmj.com/group/rights-licensing/permissions

To order reprints go to: http://journals.bmj.com/cgi/reprintform

To subscribe to BMJ go to: http://group.bmj.com/subscribe/