- Email: [email protected]
Glaucoma Specialist Optic Disc Margin, Rim Margin, and Rim Width Discordance in Glaucoma and Glaucoma Suspect Eyes
Accepted Manuscript Glaucoma Specialist Optic Disc Margin, Rim Margin and Rim Width Discordance in Glaucoma and Glaucoma Suspect Eyes. Seung Woo Hong, Helen Koenigsman, Ruojin Ren, Hongli Yang, Stuart K. Gardiner, Juan Reynaud, Robert M. Kinast, Steven L. Mansberger, Brad Fortune, Shaban Demirel, Claude F. Burgoyne PII:
S0002-9394(18)30203-4
DOI:
10.1016/j.ajo.2018.04.022
Reference:
AJOPHT 10504
To appear in:
American Journal of Ophthalmology
Received Date: 12 December 2017 Revised Date:
20 April 2018
Accepted Date: 20 April 2018
Please cite this article as: Hong SW, Koenigsman H, Ren R, Yang H, Gardiner SK, Reynaud J, Kinast RM, Mansberger SL, Fortune B, Demirel S, Burgoyne CF, Glaucoma Specialist Optic Disc Margin, Rim Margin and Rim Width Discordance in Glaucoma and Glaucoma Suspect Eyes., American Journal of Ophthalmology (2018), doi: 10.1016/j.ajo.2018.04.022. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Abstract (250 words) Purpose: To quantify the variability of 5 glaucoma specialists’ optic disc margin (DM), rim margin (RM) and rim width (RW) estimates.
RI PT
Design: Inter-Observer Reliability analysis. Methods: Clinicians viewed stereo-photos from 214 subjects with glaucoma or ocular hypertension and digitally marked the DM and RM. For each photo, the
SC
centroid of each clinician’s DM was calculated, and an averaged DMcentroid was determined. The axis between the DMcentroid and the fovea was used to establish
M AN U
twelve 30° sectors. Measurements from the DM centroid to each clinician’s DM (DMradius) and RM (RMradius) were used to generate a RW (DMradius – RMradius) and cup disc ratio (CDR) (RMradius/DMradius) by sector. Parameter means, standard deviations and coefficient of variations (COVs) were calculated across all clinicians for each eye.
TE D
Parameter means for each clinician, and intra-class correlation coefficients (ICC), were calculated across all eyes by sector.
Results: Among all eyes, the median COV by sector ranged from 3-5% for DMradius,
EP
20-25% for RMradius, and 26-30% for RW. Sectoral ICCs for CDR ranged from 0.566 to 0.668. Sectors suspicious for rim thinning by one clinician were frequently
AC C
overlooked by others. Among 1724 sectors in which at least one clinician was suspicious for rim thinning, (CDR ≥ 0.7), all 5 clinicians’ CDRs were ≥ 0.7 in only 499 (29%) and 2 of the 5 clinicians failed to detect rim thinning (CDR < 0.7) in 442 (26%). Conclusion: In this study, glaucoma specialist RM, DM and RW discordance was frequent and substantial even in sectors that were suspicious for rim thinning.
ACCEPTED MANUSCRIPT Glaucoma Specialist Optic Disc Margin, Rim Margin and Rim Width Discordance in Glaucoma and Glaucoma Suspect Eyes.
RI PT
Authors: Hong, Seung Woo1, 2; Koenigsman, Helen1; Ren, Ruojin1; Yang, Hongli1; Gardiner, Stuart K.3; Reynaud, Juan1,3; Kinast, Robert M.4; Mansberger, Steven L.3,4; Fortune, Brad3; Demirel, Shaban3; Burgoyne, Claude F.1,3,4 1
M AN U
TE D
Corresponding Author: Claude F. Burgoyne Optic Nerve Head Research Laboratory Devers Eye Institute Legacy Research Institute 1225 NE 2nd Ave Portland, OR 97208-3950 Tel: +1 503 413 4739 Email: [email protected]
SC
Devers Eye Institute, Optic Nerve Head Research Laboratory, Legacy Research Institute, 1225 NE 2nd Ave, Portland, OR 97208-3950 2 Department of Ophthalmology, Medical College, the Catholic University of Korea, 222 Banpo- daero, Seocho-gu, Seoul, Republic of Korea 06591 3 Devers Eye Institute, Discoveries in Sight Research Laboratories, Legacy Research Institute, 1225 NE 2nd Ave, Portland, OR 97208-3950 4 Devers Eye Institute Glaucoma Service, 1040 NW 22nd Ave, Portland, OR 97210
Presented in part at the American Glaucoma Society Annual Meeting, March 2012, New York, New York and at the Association for Research in Vision and Ophthalmology Annual Meeting, May 2012, Fort Lauderdale, Florida.
EP
Short title: Glaucoma Specialist Optic Disc Rim Width Discordance
AC C
Address for reprints: Claude F. Burgoyne, MD Optic Nerve Head Research Laboratory Devers Eye Institute, Legacy Research Institute 1225 NE 2nd Ave, Portland, OR 97208-3950
ACCEPTED MANUSCRIPT Introduction Since Von Helmholtz’s description of the direct ophthalmoscope in 1851, clinical examination of the optic disc has been a cornerstone of ophthalmic practice.1-3 The optic disc is the clinical term
RI PT
for the clinically visible surface of the neural and connective tissues of the optic nerve head (ONH). By convention, the clinical disc examination requires an evaluation of both the amount and health of the neuroretinal rim tissue.1-3 An examination begins by identifying the outer and inner borders of the neuroretinal rim, which are known, respectively, as the optic disc margin, and the optic disc rim
contained within the rim margin is referred to as the “cup”.
SC
margin (hereafter referred to as the disc and rim margins, respectively). The central depression
M AN U
Based in large part on the concept of “cup disc ratio” (CDR) articulated by Armaly in the 1960s,4 the amount of rim tissue is estimated within the plane of the DM (as estimated by the examiner) as the ratio of the size of the “cup” to the size of the “disc”. Most commonly a single CDR is assigned without regard for regional variations in the amount of rim tissue. These concepts are
TE D
applied whether the examination is performed with direct ophthalmoscopy, slit-lamp biomicroscopy, or stereo optic disc photography. The examiner then characterizes the color and health of the rim regionally, or by clock hour, around the optic disc by checking for the presence of pallor, bowing,
EP
swelling, notching, hemorrhages and peripapillary retinal nerve fiber layer RNFL defects.1-3 At present, the clinical disc examination remains essential to the care of all patients because it informs
AC C
the decision to obtain additional testing, because it remains the first step in progression detection when comparison to baseline stereo-photos are employed, and because “pallor” of the rim and the presence of disc hemorrhages cannot be dependably detected by OCT. Until the recent descriptions of optical coherence tomography (OCT) minimum rim width (MRW) measurements made from the OCT-detected anatomic end of Bruch’s membrane (Bruch’s Membrane opening (BMO)),5-7 all forms of automated ONH rim assessment have mimicked the clinical examination of the rim tissues by employing similar concepts and parameters.8-10 Thus, instrument-based measurements of CDR, cup volume and rim area were made in a “transverse” or
ACCEPTED MANUSCRIPT “horizontal” orientation (i.e. in a plane parallel to the retina), after the instrument had best identified the clinical disc and rim margins, often requiring the operator to identify the clinical disc margin as the first step of the analysis. These approaches allowed quantification of CDR, cup area, rim area and cup volume, but lacking a consistent anatomic foundation, failed to provide stable reference planes,8consistent measurements between instruments,13-15 or disease discrimination that exceeded the
sensitivity and specificity of experienced clinician examiners.16-18
RI PT
12
While the use of the disc and rim margins to estimate the amount of neuroretinal rim tissue
19, 20
SC
remains a core component of the clinical disc examination, a review of current teaching materials2, 3, reveals that specific instructions on how to mark the disc and rim margin are either absent,
M AN U
inconsistent or lack anatomic explanation.6, 7, 21 Clinician variability in all aspects of the clinical disc examination is well documented,22-27 but, to our knowledge, there have been no studies documenting glaucoma specialists’ disc and rim margin variability (separate from their variability in assigning a cup disc ratio) so as to assess their separate contributions to rim width variability. We believe that such a
TE D
study, if linked to OCT ONH anatomy in the same eyes, (see below) may provide an anatomic foundation for teaching clinical rim width examination that will improve its accuracy and variability. In the present study, five fellowship trained glaucoma specialists digitally delineated the disc
EP
margin and rim margin within optic disc stereo-photographs of one eye of 214 high risk ocular hypertensive and glaucoma subjects from the Portland Progression Project that had undergone OCT
AC C
ONH imaging within 3 months of photos. Each clinician independently performed their disc margin and rim margin delineations (Figure 1) without prior instruction or awareness of the other clinician’s marks. To characterize the regional variability of the 5 clinician’s marks in each eye, (Figure 2), each stereo-photo was colocalized to the infrared fundus reflectance image obtained as part of the OCT scan acquisition. This approach enabled 30° optic d isc sectoral analysis to be performed on the basis of fixed anatomical landmarks (i.e. the position of the fovea relative to the center of the ONH, Figure 2) rather than on the basis of the acquired image frame.28
ACCEPTED MANUSCRIPT Disc margin, rim margin, rim width and cup disc ratio (CDR) discordance among the five clinicians was assessed by sector, for each eye and across all 214 eyes. In a second analysis, agreement and discordance among the sectors which were suspicious for rim thinning by two criteria (CDR ≥ 0.6 and CDR ≥ 0.7), was separately assessed. The focus of the present study was to quantify
RI PT
clinician rim width discordance without regard for the accuracy of their estimates. In a separate study to follow, the clinician rim width estimates from a subset of the study eyes of the present report will be compared to colocalized OCT MRW measurements,29 so as to assess each clinician’s ability to detect
SC
sectors with thin rim tissue identified by OCT criteria. Methods
M AN U
Background and demographics. This study was performed at the Devers Eye Institute, by five clinicians, included 3 full time attending glaucoma specialists (CFB, RMK, SLM), one visiting clinician-scientist specializing in glaucoma (RR), and one glaucoma fellow (HK). The 214 study subjects were recruited from the Portland Progression Project,28, 30 an NIH-funded, longitudinal study
TE D
of progression in participants with high risk ocular hypertension and glaucoma, that is based at the Devers Eye Institute. The protocol was approved and monitored by the Legacy Health Institutional Review Board. The study adheres to the tenets of the Declaration of Helsinki and complies with the
EP
Health Insurance Portability and Accountability Act of 1996. All participants provided written informed consent, after having the risks and benefits of participation explained to them.
AC C
Study participants had best corrected vision ≥ 20/40 and one or more of the following potential risk factors for glaucoma progression (as perceived at the time of Portland Progression Project study initiation, in 1997): age >70, systemic hypertension, migraine, diet-controlled diabetes, peripheral vasospasm, African ancestry, or family history of glaucoma. Participants with ocular, neurologic or systemic diseases, medications that can affect the visual field (VF) or previous ocular trauma including ocular surgery (except for uncomplicated cataract surgery) were excluded. Each subject underwent a variety of structural and functional tests including pachymetry, stereo-photography, OCT ONH and RNFL imaging and visual field testing.
ACCEPTED MANUSCRIPT Imaging. Stereoscopic ONH photography and OCT imaging were performed within 3 months of each other, most commonly (87%) on the same day. ONH stereo-photographs were obtained using a simultaneous stereoscopic camera (3-Dx; Nidek Co Ltd, Gamagori, Japan) after maximum pupil dilation. The images were acquired on 35-mm slide film, developed and processed into color slides.
RI PT
The slides were digitized with a slide scanner (Nikon LS-5000 ED, Nikon Corporation, Tokyo, Japan) at a resolution of 4800 dpi. OCT imaging included a 30-degree fundus image of the posterior pole including the ONH and fovea (using the integrated confocal scanning laser ophthalmoscope), a 48
SC
radial B-scan data set, (each B-scan was 15-degrees wide, contained 768 A-scans and was the average of 9 repetitions) and a single, standard circumpapillary B-scan used to measure RNFL
M AN U
thickness (Spectralis, Heidelberg Engineering GmbH, Heidelberg, Germany). One eye of each subject was randomly selected for analysis. For each study eye, the illumination and focus of the left and right images of the stereoscopic image-pair were qualitatively compared and the best was chosen for simultaneous digital delineation of the disc and rim margins, as outlined below.
TE D
Clinician disc margin and rim margin assessment within each stereo-photograph (Figure 1). Five glaucoma specialists independently viewed the stereo-photo-pairs of each study eye using a stereoscopic viewer (Screen Vu™, PS Manufacturing, Portland, OR) on a high resolution
EP
computer monitor without prior instruction as to how to mark the disc margin and rim margin points or awareness of the other clinicians’ marks. The disc margin and rim margin points of each examiner
AC C
were digitally “burned” onto a separate copy of the better-focused image of the stereo-photo-pair, using a second, adjacent monitor and custom software. Clinician-specific disc margin, rim margin, rim width and CDR equivalents for each stereo-photo (Figure 2). For each disc photo, each clinician’s disc margin centroid (DMcentroid) was determined and a pooled DMcentroid (defined to be the geometric average of the 5 individual clinician DMcentroids) was separately determined. The photograph containing the pooled DMcentroid was then registered to the infrared fundus reflectance image from the matched OCT data set using a previously described technique,31 performed within Image J (version 1.43u, TurboReg plug-in, National Institutes
ACCEPTED MANUSCRIPT of Health, Bethesda, MD). The line connecting the pooled DMcentroid to the center of the fovea observed in the OCT infrared fundus image was used to define the Foveal-DM axis, which in turn defined the boundaries of each 30-degree (“clock-hour”) sector. Thus, a common set of retinal and ONH anatomical landmarks were employed to consistently define the position of the 30⁰ ONH
RI PT
sectors in each study eye,28 rather than the orientation of the fundus relative to the acquired image frame, which can vary from eye to eye.
Prior to further data analyses, all left eye data were converted into right eye orientation.30 For
SC
each clinician, their disc margin and rim margin points, were connected by straight lines and a single sectoral measurement of DMradius, RMradius and rim width (DMradius – RMradius), were made from the
M AN U
pooled DMcentroid along the centerline of each 30⁰ sector. CDR was not a primary outcome parameter of this study, however sectoral CDRs were calculated as: sectoral RMradius / DMradius and rounded to the nearest single decimal place, (CDR equivalent, Figure 2) to more closely model clinical practice. Thus, for each eye a DMradius, RMradius, rim width and CDR equivalent, (hereafter referred to as CDR)
TE D
for each clinician was calculated for each 30⁰ sector.
Data Analysis. The mean DMradius, RMradius, rim width and CDR for each clinician was calculated across all eyes for each sector. The differences in DMradius, RMradius and rim width among
EP
the 5 clinicians and 12 sectors were assessed by one-way analysis of variance. The coefficient of variation (COV) among the 5 clinicians for each parameter was calculated for each sector for each
AC C
eye. The distribution of these data across all eyes was summarized for each parameter and sector by box plot. To compare the sectoral variation in clinician discordance, the inter-clinician standard deviation for each parameter was defined to be the standard deviation of the 5 clinician’s measures of each parameter in each sector of each study eye. The inter-clinician standard deviation for each parameter was then averaged across all eyes for each parameter by sector. For each parameter the differences in these sectoral averages were assessed by Wilcoxon signed rank test.
ACCEPTED MANUSCRIPT Intra-class correlation coefficients were calculated for each parameter by sector. While systems for grading have been proposed,32 we report these data without qualitative interpretation. To estimate how often clinician rim estimates were discordant in sectors that were suspicious for rim tissue thinning, we defined suspicious sectors in two ways. First, we assessed all sectors in which at
RI PT
least one clinician’s rim width estimate was equivalent to a CDR ≥ 0.6. Second, we assessed all sectors in which at least one clinician’s rim width estimate was equivalent to a CDR ≥ 0.7. For each definition of rim thinning, we assessed clinician agreement by reporting the number of clinicians
SC
whose CDR equivalent was ≥ 0.60, or ≥ 0.7, respectively. We choose these definitions because a series of glaucoma screening studies have chosen a CDR ≥ 0.6 or CDR ≥ 0.7 to be criterion for
M AN U
suspicion that required further evaluation.33-40 Results
This study included 214 eyes of 214 subjects ranging in age from 33 to 90 years old (mean ± standard deviation: 64 ±11 years). Most of the participants were Caucasian (200/214, 93%),
TE D
approximately half (111, 52%) were female, approximately half of the eyes were right eyes 111 (52%), and the mean visual field mean deviation was -0.63 ± 2.86 decibels (standard deviation), (range -16.43 to +3.29). The clinical training and years of experience as a glaucoma specialist for the
EP
5 clinicians ranged from 0 to 19 years.
Mean DMradius, RMradius, rim width and CDR values (averaged across all eyes) are reported by
AC C
sector and by clinician in Figure 3. Interestingly, the clinical disc margin and rim margin determination tendencies of each clinician were generally consistent across the 214 study eyes as evidenced by the fact that the rank-order of clinician means for DMradius, RMradius, rim width and CDR across all eyes were consistent across all 12 sectors. For example, in the case of rim width estimation, clinician 1 manifests the least generous and clinician 5 the most generous estimation in each sector. For rim width, except for the fact that the measurements by clinicians 2 and 3 were not significantly different from one another in 8 of the 12 sectors (P ≥ 0.098), the average measurements across all study eyes for each of the 5 clinicians were different from one another in all 12 sectors (P < 0.001, ANOVA).
ACCEPTED MANUSCRIPT Figure 4 depicts the distribution of the 214 COVs (one for each eye) for DMradius, RMradius and rim width by sector. The median COV for DMradius, RMradius and rim width ranged from 3-5%, 20-25%, and 26-30% by sector, respectively. However, for as many as 10% of the eyes, the sectoral COV for DMradius, RMradius and rim width ranged from 8% to 32%, 25% to 290%, and 38% to 110%,
RI PT
respectively. Representative eyes demonstrating high rim width discordance are shown in Figure 5. High disc and rim margin discordance are shown in Supplemental Figures 1 and 2, respectively. Sectors suspicious for rim thinning by one clinician were frequently overlooked by others.
SC
Figure 6 reports the frequency of clinician CDR agreement in study eye sectors in which at least one clinician’s CDR suggested rim thinning using two criteria: (CDR ≥ 0.6 and CDR ≥ 0.7). Among the
M AN U
2137 sectors in which at least one clinician was suspicious using a criteria of CDR ≥ 0.6, all 5 clinicians’ CDRs were ≥ 0.6 in only 986 (46%) and 2 of the 5 clinicians failed to detect rim thinning (CDR < 0.6) in 592 (28%). Among the 1724 sectors in which at least one clinician was suspicious using a criteria of CDR ≥ 0.7, all 5 clinicians’ CDRs were ≥ 0.7 in only 499 (29%) and 2 of the 5
TE D
clinicians failed to detect rim thinning (CDR < 0.7) in 442 (26%). Among the 214 eyes in which at least one clinician was suspicious for rim thinning in at least one sector by CDR ≥ 0.6 criteria, in 41 (19%) of those eyes, at least 1 of the remaining clinicians was not suspicious in any sector. Among the 211
EP
eyes in which at least one clinician was suspicious in at least one sector by CDR ≥ 0.7 criteria, in 99 (47%) of those eyes, at least 1 of the remaining clinicians was not suspicious in any sector.
AC C
Study subject DIS362, (Figure 5, second row from top), presents two examples of sectors in which at least one clinician’s rim width estimates suggest rim thinning, (clinician 1 superior temporally and Clinicians 1 and 4 inferior temporally), that are not obvious in the rim width estimates of clinicians 2 and 5. In the same figure, the sectoral CDR equivalents for study subject DIS432 (bottom row) are ≥ 0.9 for clinicians 1-4, inferiorly, but clinician 5’s CDR equivalent is < 0.60 in the same sectors. Finally, the inferior temporal sectoral rim width estimates of clinician 1 for subject DIS 409, (third row from the top) produce a CDR equivalent of ≥ 0.8, whereas the CDR equivalents of clinicians 2 and 5 are ≤ 0.6 within the same sectors.
ACCEPTED MANUSCRIPT To assess the sectoral distribution of clinician discordance for DMradius, RMradius, and rim width, a polar plot of the inter-clinician standard deviation for each parameter among the five clinicians for each eye, assessed by sector, and averaged across all 214 eyes is presented in Supplemental Figure 3. The average inter-clinician standard deviation across all eyes for all three parameters was greatest
RI PT
nasally and least temporally, with the variability of inferior and superior sectors being similar in magnitude to each other and midway between that of the nasal and temporal sectors. To assess inter-clinician agreement for each parameter, we report ICCs for DMradius, RMradius, rim width and CDR
SC
equivalent by sector in Supplementary Table 1. Sectoral ICCs for rim width and CDR were greatest in the inferior sector (0.633 and 0.684, respectively) and smallest within the temporal-superior sector
M AN U
(0.481 and 0.566, respectively). Discussion
This study documents that DM and RM demarcations, as well as the corresponding sectoral rim width estimates, varied substantially among 5 glaucoma specialists when they were assessed in a
TE D
masked fashion within stereo-photographs of 214 glaucoma patient and glaucoma suspect eyes. The fact that we did this assessment within 30⁰ sectors that were consistently applied to each study eye (i.e. using consistent retinal and ONH landmarks), not only provides a more robust anatomic
EP
foundation for their regionalization among clinicians and study eyes, but will also allow us to colocalize the rim width estimates of each clinician to OCT MRW measurements, regionalized in the
AC C
same manner, in a follow up study.
To our knowledge, every previous study that has reported the variability of the clinical disc examination has evaluated the reproducibility of global or regional CDR estimates.22-25, 41-47 In these studies, the degree of agreement in vertical CDR ranged from 0.19 to 0.90 using a weighted kappa value,22, 41-46 and 0.44 to 0.855 using ICC.47-49 Among glaucoma specialists, the reported weighted kappa values range between 0.62 and 0.9,44-46 and the ICCs range between 0.85 to 0.861.48, 49 While CDR was not a primary outcome measure of this study, when our glaucoma specialist rim width estimates are expressed in CDR equivalents, the sectoral ICCs ranged from 0.566 to 0.684.
ACCEPTED MANUSCRIPT When the superior and inferior sectoral rim estimates were combined to create a single vertical CDR estimate (data not shown), the ICC value was 0.720, which is substantially lower than most previous reports. As the vertical CDR equivalent of our study is different from the vertical CDR of the clinical disc examination (largest vertical cup diameter / largest disc margin diameter), our results
RI PT
cannot be directly compared to previous studies. However, the fact that our ICC value is lower than previous studies is noteworthy, because we used the FoBMO axis to generate anatomically consistent ONH sectors, which better reflects ‘inter-clinician discordance’ by removing inconsistent
SC
assignment of the “vertical” or “horizontal” CDR measurement location. Yet, even in a study setting that removed these “non-clinician” causes of discordance, ICC values for sectoral rim width ranged
M AN U
from 0.481 to 0.633, and median COV values by sector ranged from 24.3% to 31.9%. Among the factors that contribute to clinician disc margin and rim margin discordance is a lack of agreement on the anatomic landmarks underlying DM and RM. A second factor is that in the absence of anatomic foundations, there is little consistency in the way clinicians are taught how to
TE D
identify the disc and rim margins. The underlying anatomy of the clinical DM has traditionally been thought to be the “scleral lip” or “scleral ring”.2, 3, 19 However a series of recent OCT studies7, 31 have shown that the anatomy underlying the clinical disc margin varies among ONH sectors, but that it is
EP
most commonly defined by the BMO, unpigmented Bruch’s Membrane, and the Border Tissues of Elschnig, rather than the edge of sclera. Less commonly it involves unpigmented peripapillary sclera
AC C
and/or the anterior scleral canal opening. In this study, while disc margin discordance was present and substantial in a subset of eyes, it was small in frequency and magnitude compared to rim margin discordance. However, there is a separate discordance between the anatomic endpoint of BMO, where an accurate measurement of the rim can be most consistently made within OCT anatomy6, 7 and the clinically visible disc margin (Figure 7). The discordance between OCT-identified BMO and the clinical disc margin will affect the accuracy of clinicians’ rim width estimates, regardless of clinician disc margin discordance. This issue will be addressed further in the OCT versus clinician study to follow.
ACCEPTED MANUSCRIPT While the rim margin is commonly defined as the point where the optic disc surface dips below the level of the peripapillary retina,50 the point where the slope first develops an inward deflection and the point where the slope is steepest are also widely used.51 However, the three-dimensional perception of the ONH that can be achieved through stereo-photographic assessment or slit lamp
RI PT
stereo-biomicroscopy is inconsistent and limited compared to OCT. In this study we deliberately did not instruct the clinicians on disc margin and rim margin anatomy because our purpose was to document their current practice rather than to maximize their accuracy and reproducibility. To address
SC
these issues, studies that project clinicians’ rim margin estimates into OCT anatomy are required to determine if OCT identification of a more accurate and reproducible definition for the clinical rim
M AN U
margin can be taught.
CDR equivalent discordance was present in a substantial proportion of the eyes with sectors in which at least one clinician raised suspicion for glaucoma. While CDR rim width estimation is only one portion of the clinical disc examination, given that the clinical disc examination remains the most
TE D
important screening test for detecting the optic neuropathy of glaucoma, our findings indicate that it will be important to develop clinical teaching libraries focused on the sensitive and specific detection of rim thinning that is suspicious or abnormal by OCT criteria.
EP
Although the clinicians showed significant differences from one another, the rank order of clinicians’ average DMradius, RMradius, and rim width estimates were consistent across almost every
AC C
sector (e.g., whereby clinician 1’s estimates were consistently smallest and clinician 5’s consistently largest for rim width). The fact that this finding occurred in a cohort of eyes that encompass a wide range of physiologic and pathophysiologic differences in rim anatomy, suggests that the glaucoma specialists’ tendencies were internally consistent in terms of judging disc and rim margin across this wide range of eyes. This study does not address the accuracy of each clinician examiner, but our follow up study will do so by comparing each clinician’s rim width to OCT MRW by sector. In the study eye of subject DIS362, (Figure 5), clinician 1’s rim width estimates suggest rim thinning in the superior and superior-
ACCEPTED MANUSCRIPT temporal disc sectors that appears to be discordant with the other examiners. Figure 7 depicts colocalized, sectoral OCT MRW measurements for the same eye. The panels of this figure illustrate several points. First, in this eye, most clinicians overestimated the amount of rim tissue temporally, in a region of invisible BMO, which likely contributed to the failure to detect tissues that were thin by
RI PT
OCT criteria.6, 7 Second, while the rim width estimates of clinician 1 (purple with CDR equivalent in (7E)) superiorly and those of Clinicians 1 and 4, inferior temporally, appear to be outliers, in both instances, their estimates reflect areas of OCT suspicion and the discordance generated in these
SC
sectors reflects rim width over-estimation by the other clinicians. By CDR ≥ 0.7 criteria, only 1 of the 5 clinicians detected rim thinning in the temporal inferior quadrant and only 2 of the 5 clinicians
M AN U
detected it temporally. While there was substantial rim width discordance superior temporally, 4 of the 5 clinicians CDR equivalents suggested rim thinning by CDR 0.7 criteria. Whether this discordance and/or these cup disc ratios translate to an accurate or missed diagnosis, requires future studies in which the clinician is asked to identify suspicious sectors.
TE D
To this point specifically, our study is limited by the relatively small number of clinician examiners and the fact that it only addresses clinical rim width estimates within stereo-photographs and does not include the other aspects of the clinical examination such as pallor, nerve fiber layer
EP
thinning and nerve fiber layer hemorrhages that together may have made the glaucoma specialists more concordant in terms of a final diagnosis. We focused on DM, RM and rim width estimation
AC C
because these components of the clinical examination could be quantified and compared to OCT rim anatomy in the studies that will follow. Had we simply asked the clinicians if the disc was suspicious for glaucoma, clinician agreement may have been better even in the face of the rim width discordance we report. While the number of examiners is small, they are equal to or exceed the number of examiners in most previous studies22-25, 41-44, 46-49 and they span a range of training and experience that is likely to be present in most tertiary care glaucoma practices.
ACCEPTED MANUSCRIPT Finally, it is also worth noting that while one clinician was in glaucoma fellowship training during the study and a second was a young visiting clinician-scientist from China, neither of these clinicians, (clinicians 2 and 4, respectively) were outliers when compared to clinicians 1 and 5. Our study was not designed to demonstrate the effect of practice duration on clinician discordance. Such
RI PT
a study would require multiple clinicians at each level of experience over the full duration of an average career and include a non-linear assessment of the post-fellowship learning curve. These kinds of analyses are beyond the scope of the present report.
SC
In summary, our findings show that fellowship trained glaucoma specialists, exhibit substantial discordance in their estimates of the ONH rim tissue present in glaucomatous or at-risk eyes.
M AN U
Discordance occurred commonly in disc sectors that were judged by at least one clinician to be suspicious for demonstrating rim thinning. In a follow-up study, we will compare each clinician’s sectoral rim width estimates to OCT MRW measurements in the same eyes (as shown in Figure 7 of the present report). The goal of this work will be to identify and articulate anatomic explanations for
TE D
clinical DM and RM estimation that can provide guidance for improving the accuracy and consistency
AC C
EP
of clinical rim width estimation by all practioners.
ACCEPTED MANUSCRIPT Acknowledgments/ Disclosure: a. Funding/ Support: NIH/NEI R01-EY-021281; Legacy Good Samaritan Foundation The above listed sponsors/funding organizations had no role in the design, conduct, analysis or reporting of this research.
RI PT
b. Financial Disclosures: SW. Hong: None. R. Ren: None.
SC
H. Yang: None. S. Gardiner: None.
M AN U
C. Hardin: None.
S. Demirel: NIH/NEI R01-EY-019674; Legacy Good Samaritan Foundation; Carl Zeiss Meditec. C. Burgoyne: NIH/NEI R01-EY021281; Legacy Good Samaritan Foundation; Heidelberg Engineering, GmbH, Heidelberg, Germany - Dr. Burgoyne is a consultant to Heidelberg Engineering. In this role,
TE D
he receives unrestricted research support, instruments, software, occasional travel support, but no honorarium or personal income.
R. Kinast: Research: NEI (RO1 EY025181-01), Allergan, Ocular Therapeutix, American Glaucoma
EP
Society, Oregon Health & Sciences University Biomedical Innovation Program. S. Mansberger: Research: NEI (RO1 EY025181-01), Allergan, Ocular Therapeutix, American
AC C
Glaucoma Society; Consultant: Envisia, Santen, Gore,Aerie. B. Fortune: Inotek Pharmaceuticals; Glaucoma Research Foundation; BrightFocus Foundation; Legacy Good Samaritan Foundation. c. Acknowledgements: The authors wish to thank Cindy Albert, and Michael Whitworth for their clinical imaging of the Portland Progression Project participants and Dr. Jonathan He for his assistance with the preliminary statistical analysis and graphing of our data. We would also like to thank Julia Monaghan for assistance with manuscript preparation and submission.
ACCEPTED MANUSCRIPT References
1.
Prum BE, Jr., Rosenberg LF, Gedde SJ, et al. Primary Open-Angle Glaucoma Preferred
Practice Pattern Guidelines. Ophthalmology 2016;123(1):P41-P111. 2.
Nicolela MT. Optic Nerve: Clinical Examination. In: Giaconi JA, Law SK, Coleman AL, Caprioli
RI PT
J, editors. Pearls of Glaucoma Management. Berlin, Heidelberg: Springer Berlin Heidelberg, 2010:1521. 3.
Cantor LB, Rapuano CJ, Cioffi GA. Chapter 3: Clinical Evaluation. In: Basic and Clinical
Science Course (BCSC) 2017-2018: Glaucoma Section 10. San Francisco, CA: American Academy of Ophthalmology, 2017-2018:48.
Armaly MF. Genetic determination of cup/disc ratio of the optic nerve. Arch Ophthalmol
SC
4.
1967;78(1):35-43.
Povazay B, Hofer B, Hermann B, et al. Minimum distance mapping using three-dimensional
M AN U
5.
optical coherence tomography for glaucoma diagnosis. J Biomed Opt 2007;12(4):041204. 6.
Reis AS, O'Leary N, Yang H, et al. Influence of clinically invisible, but optical coherence
tomography detected, optic disc margin anatomy on neuroretinal rim evaluation. Invest Ophthalmol Vis Sci 2012;53(4):1852-60. 7.
Reis AS, Sharpe GP, Yang H, Nicolela MT, Burgoyne CF, Chauhan BC. Optic disc margin
TE D
anatomy in patients with glaucoma and normal controls with spectral domain optical coherence tomography. Ophthalmology 2012;119(4):738-47. 8.
Burk RO, Vihanninjoki K, Bartke T, et al. Development of the standard reference plane for the
Heidelberg retina tomograph. Graefes Arch Clin Exp Ophthalmol 2000;238(5):375-84. Poli A, Strouthidis NG, Ho T, Garway-Heath DF. Analysis of HRT images: comparison of
EP
9.
reference planes. Invest Ophthalmol Vis Sci 2008;49(9):3970-3975. Mwanza JC, Chang RT, Budenz DL, et al. Reproducibility of peripapillary retinal nerve fiber
AC C
10.
layer thickness and optic nerve head parameters measured with cirrus HD-OCT in glaucomatous eyes. Invest Ophthalmol Vis Sci 2010;51(11):5724-30. 11.
Leung CK, Chan WM, Hui YL, et al. Analysis of retinal nerve fiber layer and optic nerve head in
glaucoma with different reference plane offsets, using optical coherence tomography. Invest Ophthalmol Vis Sci 2005;46(3):891-9. 12.
Tan JC, Hitchings RA. Reference plane definition and reproducibility in optic nerve head
images. Invest Ophthalmol Vis Sci 2003;44(3):1132-7. 13.
Mistlberger A, Liebmann JM, Greenfield DS, et al. Heidelberg retina tomography and optical
coherence tomography in normal, ocular-hypertensive, and glaucomatous eyes. Ophthalmology 1999;106(10):2027-32.
ACCEPTED MANUSCRIPT 14.
Shin HY, Park HY, Jung KI, Park CK. Glaucoma diagnosis optic disc analysis comparing
Cirrus spectral domain optical coherence tomography and Heidelberg retina tomograph II. Jpn J Ophthalmol 2013;57(1):41-6. 15.
Yang B, Ye C, Yu M, Liu S, Lam DS, Leung CK. Optic disc imaging with spectral-domain
optical coherence tomography: variability and agreement study with Heidelberg retinal tomograph.
16.
RI PT
Ophthalmology 2012;119(9):1852-7. Girkin CA, McGwin G, Jr., Long C, DeLeon-Ortega J, Graf CM, Everett AW. Subjective and
objective optic nerve assessment in African Americans and whites. Invest Ophthalmol Vis Sci 2004;45(7):2272-8. 17.
Deleon-Ortega JE, Arthur SN, McGwin G, Jr., Xie A, Monheit BE, Girkin CA. Discrimination
SC
between glaucomatous and nonglaucomatous eyes using quantitative imaging devices and subjective optic nerve head assessment. Invest Ophthalmol Vis Sci 2006;47(8):3374-80. Vessani RM, Moritz R, Batis L, Zagui RB, Bernardoni S, Susanna R. Comparison of
M AN U
18.
quantitative imaging devices and subjective optic nerve head assessment by general ophthalmologists to differentiate normal from glaucomatous eyes. J Glaucoma 2009;18(3):253-61. 19.
Fingeret M, Medeiros FA, Susanna R, Jr., Weinreb RN. Five rules to evaluate the optic disc
and retinal nerve fiber layer for glaucoma. Optometry 2005;76(11):661-8. 20.
Susanna R, Jr., Vessani RM. New findings in the evaluation of the optic disc in glaucoma
21.
TE D
diagnosis. Curr Opin Ophthalmol 2007;18(2):122-8.
Chauhan BC, O'Leary N, Almobarak FA, et al. Enhanced detection of open-angle glaucoma
with an anatomically accurate optical coherence tomography-derived neuroretinal rim parameter. Ophthalmology 2013;120(3):535-43.
Varma R, Steinmann WC, Scott IU. Expert agreement in evaluating the optic disc for
EP
22.
glaucoma. Ophthalmology 1992;99(2):215-21. Klein BE, Magli YL, Richie KA, Moss SE, Meuer SM, Klein R. Quantitation of optic disc
AC C
23.
cupping. Ophthalmology 1985;92(12):1654-6. 24.
Lichter PR. Variability of expert observers in evaluating the optic disc. Trans Am Ophthalmol
Soc 1976;74:532-72. 25.
Varma R, Spaeth GL, Steinmann WC, Katz LJ. Agreement between clinicians and an image
analyzer in estimating cup-to-disc ratios. Arch Ophthalmol 1989;107(4):526-9. 26.
Jampel HD, Friedman D, Quigley H, et al. Agreement among glaucoma specialists in
assessing progressive disc changes from photographs in open-angle glaucoma patients. Am J Ophthalmol 2009;147(1):39-44 e1. 27.
Lloyd MJ, Mansberger SL, Fortune BA, et al. Features of optic disc progression in patients with
ocular hypertension and early glaucoma. J Glaucoma 2013;22(5):343-8.
ACCEPTED MANUSCRIPT 28.
He L, Ren R, Yang H, et al. Anatomic vs. acquired image frame discordance in spectral
domain optical coherence tomography minimum rim measurements. PLoS One 2014;9(3):e92225. 29.
Chauhan BC, Danthurebandara VM, Sharpe GP, et al. Bruch's Membrane Opening Minimum
Rim Width and Retinal Nerve Fiber Layer Thickness in a Normal White Population: A Multicenter Study. Ophthalmology 2015;122(9):1786-94. Ren R, Yang H, Gardiner SK, et al. Anterior lamina cribrosa surface depth, age, and visual
RI PT
30.
field sensitivity in the Portland Progression Project. Invest Ophthalmol Vis Sci 2014;55(3):1531-9. 31.
Strouthidis NG, Yang H, Reynaud JF, et al. Comparison of clinical and spectral domain optical
coherence tomography optic disc margin anatomy. Invest Ophthalmol Vis Sci 2009;50(10):4709-18. 32.
Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized
33.
SC
assessment instruments in psychology. Psychological Assessment 1994;6(4):284-290.
Iwase A, Suzuki Y, Araie M, et al. The prevalence of primary open-angle glaucoma in
34.
M AN U
Japanese: the Tajimi Study. Ophthalmology 2004;111(9):1641-8.
Kyari F, Entekume G, Rabiu M, et al. A Population-based survey of the prevalence and types
of glaucoma in Nigeria: results from the Nigeria National Blindness and Visual Impairment Survey. BMC Ophthalmol 2015;15:176. 35.
Gupta P, Zhao D, Guallar E, Ko F, Boland MV, Friedman DS. Prevalence of Glaucoma in the
Vis Sci 2016;57(6):2905-2913. 36.
TE D
United States: The 2005-2008 National Health and Nutrition Examination Survey. Invest Ophthalmol
Bokman CL, Pasquale LR, Parrish RK, 2nd, Lee RK. Glaucoma screening in the Haitian Afro-
Caribbean population of South Florida. PLoS One 2014;9(12):e115942. 37.
Kim CS, Seong GJ, Lee NH, Song KC, Namil Study Group KGS. Prevalence of primary open-
38.
EP
angle glaucoma in central South Korea the Namil study. Ophthalmology 2011;118(6):1024-30. Iwase A, Sawaguchi S, Sakai H, Tanaka K, Tsutsumi T, Araie M. Optic disc, rim and
AC C
peripapillary chorioretinal atrophy in normal Japanese eyes: the Kumejima Study. Jpn J Ophthalmol 2017;61(3):223-229. 39.
Pan CW, Zhao CH, Yu MB, et al. Prevalence, types and awareness of glaucoma in a multi-
ethnic population in rural China: the Yunnan Minority Eye Study. Ophthalmic Physiol Opt 2016;36(6):664-670. 40.
Ashaye A, Ashaolu O, Komolafe O, et al. Prevalence and Types of Glaucoma Among an
Indigenous African Population in Southwestern Nigeria. Investigative Ophthalmology & Visual Science 2013;54(12):7410-7416. 41.
Tielsch JM, Katz J, Quigley HA, Miller NR, Sommer A. Intraobserver and interobserver
agreement in measurement of optic disc characteristics. Ophthalmology 1988;95(3):350-6.
ACCEPTED MANUSCRIPT 42.
Abrams LS, Scott IU, Spaeth GL, Quigley HA, Varma R. Agreement among optometrists,
ophthalmologists, and residents in evaluating the optic disc for glaucoma. Ophthalmology 1994;101(10):1662-7. 43.
Harper R, Reeves B, Smith G. Observer variability in optic disc assessment: implications for
glaucoma shared care. Ophthalmic Physiol Opt 2000;20(4):265-73. Kong YX, Coote MA, O'Neill EC, et al. Glaucomatous optic neuropathy evaluation project: a
RI PT
44.
standardized internet system for assessing skills in optic disc examination. Clin Exp Ophthalmol 2011;39(4):308-17. 45.
Chan HH, Ong DN, Kong YX, et al. Glaucomatous optic neuropathy evaluation (GONE)
project: the effect of monoscopic versus stereoscopic viewing conditions on optic nerve evaluation.
46.
SC
Am J Ophthalmol 2014;157(5):936-44.
Bastawrous A, Giardini ME, Bolster NM, et al. Clinical Validation of a Smartphone-Based
47.
M AN U
Adapter for Optic Disc Imaging in Kenya. JAMA Ophthalmol 2016;134(2):151-8. Spry PG, Spencer IC, Sparrow JM, et al. The Bristol Shared Care Glaucoma Study: reliability
of community optometric and hospital eye service test measures. Br J Ophthalmol 1999;83(6):707-12. 48.
Arthur SN, Aldridge AJ, De Leon-Ortega J, McGwin G, Xie A, Girkin CA. Agreement in
assessing cup-to-disc ratio measurement among stereoscopic optic nerve head photographs, HRT II, and Stratus OCT. J Glaucoma 2006;15(3):183-9.
Tanito M, Sagara T, Takamatsu M, et al. Intraobserver and interobserver agreement of
TE D
49.
computer software-assisted optic nerve head photoplanimetry. Jpn J Ophthalmol 2014;58(1):56-61. 50.
Portney GL. Photogrammetric analysis of the three-dimensional geometry of normal and
46. 51.
EP
glaucomatous optic cups. Trans Sect Ophthalmol Am Acad Ophthalmol Otolaryngol 1976;81(2):239-
Sommer A, Pollack I, Maumenee AE. Optic disc parameters and onset of glaucomatous field
AC C
loss. I. Methods and progressive changes in disc morphology. Arch Ophthalmol 1979;97(8):1444-8.
ACCEPTED MANUSCRIPT Figure Legends
Figure 1. A stereo-photo (upper left and middle) of each ONH was individually viewed by each clinician while simultaneously marking the Disc Margin (green dots) and Rim Margin (black
RI PT
dots) within one image from the stereo-pair viewed within custom marking software on an adjacent monitor (upper right). Without prior instruction, for each study eye, each clinician used a stereoscopic viewer to examine its stereo disc photograph (left and middle) on a high resolution
SC
computer monitor. Simultaneous marking of the disc margin and rim margin on the better focused of the two stereo-photo images (left or right) occurred within custom marking software on a second,
M AN U
adjacent monitor. For this eye the right image of the stereo-pair (right panel) was preselected to be used to mark the disc margin and rim margin points. (Lower) Disc margin and rim margin points along with the resultant rim width (purple shading) variability amongst the 5 glaucoma specialists for
TE D
this study eye.
Figure 2. Establishing a consistent set of twelve 30
(clock hour) sectors for the clinician disc
margin, rim margin and rim width measurements of each eye. (A) The geometric centroid (center
EP
of gravity) for each clinician's disc margin delineation (DM centroid) was calculated. (B) A pooled DM centroid (black dot) was calculated by averaging the x,y coordinates of the 5 clinicians’ DM centroids.
AC C
(C) The photograph containing the pooled DM centroid was colocalized to the infrared fundus reflectance image from an optical coherence tomography (OCT) optic nerve head data set acquired within 3 months of the photograph. A line connecting the pooled DM centroid (black dot) and the foveal center (blue dot) was drawn to establish the Foveal-DM axis, which was used as the horizontal axis for twelve 30⁰ (clock hour) sectors (white borders). Note, that because the Foveal-DM axis was used as the nasal-temporal axis for each eye, rather than the horizontal axis of the acquired image, the regions are consistent for all 5 clinicians among all 214 study eyes. (D) The disc margin and rim margin points for clinician 3, (both green). (E) Clinician 3’s disc margin (still in green) and rim margin
ACCEPTED MANUSCRIPT points (now black) connected by straight lines. (F) For each sector, a single measurement of DM radius, RM radius and rim width (RW – in purple) was made along the midline of the sector. While not a primary outcome parameter, a cup disc ratio (CDR) was calculated for each sector as (sectoral RW radius / sectoral DM radius) which was then rounded to the nearest single decimal place (CDR
RI PT
equivalent) to more closely model clinical practice.
Figure 3. Polar plots of the mean values for disc margin radius and rim margin radius, (left),
SC
rim width (center), and cup disc ratio equivalent, (right), assessed by sector and averaged for each clinician across all 214 eyes. Data are in microns. Interestingly, the delineation tendencies of
M AN U
each clinician were generally consistent across the 214 study eyes as evidenced by the fact that the rank-order of clinician means for disc margin radius, rim margin radius, rim width and cup disc ratio equivalent were consistent across all sectors.
TE D
Figure 4. Distribution box-plots of the coefficient of variation (COV) among the 5 clinicians for each study eye, for each parameter, by sector. The central box of each plot defines the median with the upper and lower quartiles values. The thick and thin lines define the upper and lower 10thth
EP
and 5th percentiles, (respectively) of the distribution. The median COV for DM radius, RM radius and rim width ranged from 3 to 5%, 20 to 25%, and 26 to 30% by sector, respectively. However, for as
AC C
many as 10% of the eyes, the sectoral COV for DM radius, RM radius and rim width ranged from 8% to 32%, 25% to 290%, and 38% to 110%, respectively. Representative study eyes demonstrating high rim width discordance are shown in Figure 5. High disc margin and rim margin discordant eyes can be seen in Supplemental Figures 1 and 2, respectively.
Figure 5. Four representative high rim width discordance eyes.
ACCEPTED MANUSCRIPT Figure 6. Frequency of clinician agreement in study eye sectors in which at least one clinician’s CDR equivalent suggests rim thinning using two CDR criteria, (CDR ≥ 0.6 and CDR ≥ 0.7, respectively). (Left Column – Suspicion and Agreement Criteria CDR ≥ 0.6) For each sector in which at least one clinician’s CDR was ≥ 0.6, (n=2137 sectors), the level of agreement is
RI PT
reported as the number of clinicians whose CDR equivalent was ≥ 0.6. (Right Columns – Suspicion and Agreement Criteria CDR ≥ 0.7) For each sector in which at least one clinician’s CDR was ≥ 0.7, (n = 1724 sectors), the level of Agreement is reported as the number of clinicians whose CDR
SC
equivalent was ≥ 0.7. For all columns the darkest color denotes the number of sectors in which all 5 clinicians were suspicious for rim thinning using the stated criteria. The lightest color denotes the
were in agreement, by the stated criteria.
M AN U
number of sectors and eyes in which 1 clinician was suspicious and none of the other four clinicians
Figure 7. Clinician Disc Margin, Rim Margin and Rim Width estimates co-localized to sectoral
TE D
OCT minimum rim width (MRW) anatomy in a study eye (DIS362 from Figure 5, above) in which clinician discordance occurs in sectors that are structurally suspicious by OCT criteria. (A) Fundus photograph. (B) Bruch’s membrane opening (BMO), the centroid of BMO, the Foveal – BMO
EP
axis (FoBMO axis, red line) and the 12 FoBMO 30⁰ (clock hour) sectors. (C). BMO relative to the disc margin points of all four clinicians. Note that BMO is well inside the clinical disc margin points of
AC C
all 5 clinicians temporally, (black arrows), suggesting that BMO is clinically invisible in this region, as has been previously described6, 7 (D, Center Panel).. BMO, and each clinician’s disc margin and rim margin points have been superimposed on the funds photo in the center, and OCT rim anatomy along with its quantification relative to a Caucasian Normative Data Base29 are shown for each FoBMO sector. Note that the superior temporal, temporal and temporal inferior sector OCT anatomy boxes have been outlined in yellow because the OCT MRW anatomy (yellow arrow) is in the lowest 5th percentile of the NDB. Notice also that while they are considered “normal” the MRW values for the adjacent temporal superior and superior sectors are in the 8th and 9th percentiles, respectively. In a
ACCEPTED MANUSCRIPT white circle at the top of each OCT anatomy box, the number of clinicians that were suspicious for rim thinning by CDR ≥ 0.7 criteria are noted (E-I, bottom row). Each clinician’s rim width, (purple) and their CDR equivalent for the three sectors that are suspicious by OCT criteria,(in yellow) are shown.
RI PT
Supplemental Figure 1. Four representative high disc margin discordance eyes.
SC
Supplemental Figure 2. Four representative high rim margin discordance eyes.
Supplemental Figure 3. Polar plot of disc margin, rim margin and rim width discordance
M AN U
among the five clinicians for each eye, assessed by sector, and averaged across all 214 eyes. Data for each sector are the inter-clinician standard deviation for each parameter for each eye, averaged across all study eyes. On average the sectoral distribution of clinican discordance was similar for all three parameters and was greatest nasally, similarly moderate superiorly and inferiorly
AC C
EP
TE D
and least temporally.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
S0002-9394(18)30203-4
DOI:
10.1016/j.ajo.2018.04.022
Reference:
AJOPHT 10504
To appear in:
American Journal of Ophthalmology
Received Date: 12 December 2017 Revised Date:
20 April 2018
Accepted Date: 20 April 2018
Please cite this article as: Hong SW, Koenigsman H, Ren R, Yang H, Gardiner SK, Reynaud J, Kinast RM, Mansberger SL, Fortune B, Demirel S, Burgoyne CF, Glaucoma Specialist Optic Disc Margin, Rim Margin and Rim Width Discordance in Glaucoma and Glaucoma Suspect Eyes., American Journal of Ophthalmology (2018), doi: 10.1016/j.ajo.2018.04.022. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Abstract (250 words) Purpose: To quantify the variability of 5 glaucoma specialists’ optic disc margin (DM), rim margin (RM) and rim width (RW) estimates.
RI PT
Design: Inter-Observer Reliability analysis. Methods: Clinicians viewed stereo-photos from 214 subjects with glaucoma or ocular hypertension and digitally marked the DM and RM. For each photo, the
SC
centroid of each clinician’s DM was calculated, and an averaged DMcentroid was determined. The axis between the DMcentroid and the fovea was used to establish
M AN U
twelve 30° sectors. Measurements from the DM centroid to each clinician’s DM (DMradius) and RM (RMradius) were used to generate a RW (DMradius – RMradius) and cup disc ratio (CDR) (RMradius/DMradius) by sector. Parameter means, standard deviations and coefficient of variations (COVs) were calculated across all clinicians for each eye.
TE D
Parameter means for each clinician, and intra-class correlation coefficients (ICC), were calculated across all eyes by sector.
Results: Among all eyes, the median COV by sector ranged from 3-5% for DMradius,
EP
20-25% for RMradius, and 26-30% for RW. Sectoral ICCs for CDR ranged from 0.566 to 0.668. Sectors suspicious for rim thinning by one clinician were frequently
AC C
overlooked by others. Among 1724 sectors in which at least one clinician was suspicious for rim thinning, (CDR ≥ 0.7), all 5 clinicians’ CDRs were ≥ 0.7 in only 499 (29%) and 2 of the 5 clinicians failed to detect rim thinning (CDR < 0.7) in 442 (26%). Conclusion: In this study, glaucoma specialist RM, DM and RW discordance was frequent and substantial even in sectors that were suspicious for rim thinning.
ACCEPTED MANUSCRIPT Glaucoma Specialist Optic Disc Margin, Rim Margin and Rim Width Discordance in Glaucoma and Glaucoma Suspect Eyes.
RI PT
Authors: Hong, Seung Woo1, 2; Koenigsman, Helen1; Ren, Ruojin1; Yang, Hongli1; Gardiner, Stuart K.3; Reynaud, Juan1,3; Kinast, Robert M.4; Mansberger, Steven L.3,4; Fortune, Brad3; Demirel, Shaban3; Burgoyne, Claude F.1,3,4 1
M AN U
TE D
Corresponding Author: Claude F. Burgoyne Optic Nerve Head Research Laboratory Devers Eye Institute Legacy Research Institute 1225 NE 2nd Ave Portland, OR 97208-3950 Tel: +1 503 413 4739 Email: [email protected]
SC
Devers Eye Institute, Optic Nerve Head Research Laboratory, Legacy Research Institute, 1225 NE 2nd Ave, Portland, OR 97208-3950 2 Department of Ophthalmology, Medical College, the Catholic University of Korea, 222 Banpo- daero, Seocho-gu, Seoul, Republic of Korea 06591 3 Devers Eye Institute, Discoveries in Sight Research Laboratories, Legacy Research Institute, 1225 NE 2nd Ave, Portland, OR 97208-3950 4 Devers Eye Institute Glaucoma Service, 1040 NW 22nd Ave, Portland, OR 97210
Presented in part at the American Glaucoma Society Annual Meeting, March 2012, New York, New York and at the Association for Research in Vision and Ophthalmology Annual Meeting, May 2012, Fort Lauderdale, Florida.
EP
Short title: Glaucoma Specialist Optic Disc Rim Width Discordance
AC C
Address for reprints: Claude F. Burgoyne, MD Optic Nerve Head Research Laboratory Devers Eye Institute, Legacy Research Institute 1225 NE 2nd Ave, Portland, OR 97208-3950
ACCEPTED MANUSCRIPT Introduction Since Von Helmholtz’s description of the direct ophthalmoscope in 1851, clinical examination of the optic disc has been a cornerstone of ophthalmic practice.1-3 The optic disc is the clinical term
RI PT
for the clinically visible surface of the neural and connective tissues of the optic nerve head (ONH). By convention, the clinical disc examination requires an evaluation of both the amount and health of the neuroretinal rim tissue.1-3 An examination begins by identifying the outer and inner borders of the neuroretinal rim, which are known, respectively, as the optic disc margin, and the optic disc rim
contained within the rim margin is referred to as the “cup”.
SC
margin (hereafter referred to as the disc and rim margins, respectively). The central depression
M AN U
Based in large part on the concept of “cup disc ratio” (CDR) articulated by Armaly in the 1960s,4 the amount of rim tissue is estimated within the plane of the DM (as estimated by the examiner) as the ratio of the size of the “cup” to the size of the “disc”. Most commonly a single CDR is assigned without regard for regional variations in the amount of rim tissue. These concepts are
TE D
applied whether the examination is performed with direct ophthalmoscopy, slit-lamp biomicroscopy, or stereo optic disc photography. The examiner then characterizes the color and health of the rim regionally, or by clock hour, around the optic disc by checking for the presence of pallor, bowing,
EP
swelling, notching, hemorrhages and peripapillary retinal nerve fiber layer RNFL defects.1-3 At present, the clinical disc examination remains essential to the care of all patients because it informs
AC C
the decision to obtain additional testing, because it remains the first step in progression detection when comparison to baseline stereo-photos are employed, and because “pallor” of the rim and the presence of disc hemorrhages cannot be dependably detected by OCT. Until the recent descriptions of optical coherence tomography (OCT) minimum rim width (MRW) measurements made from the OCT-detected anatomic end of Bruch’s membrane (Bruch’s Membrane opening (BMO)),5-7 all forms of automated ONH rim assessment have mimicked the clinical examination of the rim tissues by employing similar concepts and parameters.8-10 Thus, instrument-based measurements of CDR, cup volume and rim area were made in a “transverse” or
ACCEPTED MANUSCRIPT “horizontal” orientation (i.e. in a plane parallel to the retina), after the instrument had best identified the clinical disc and rim margins, often requiring the operator to identify the clinical disc margin as the first step of the analysis. These approaches allowed quantification of CDR, cup area, rim area and cup volume, but lacking a consistent anatomic foundation, failed to provide stable reference planes,8consistent measurements between instruments,13-15 or disease discrimination that exceeded the
sensitivity and specificity of experienced clinician examiners.16-18
RI PT
12
While the use of the disc and rim margins to estimate the amount of neuroretinal rim tissue
19, 20
SC
remains a core component of the clinical disc examination, a review of current teaching materials2, 3, reveals that specific instructions on how to mark the disc and rim margin are either absent,
M AN U
inconsistent or lack anatomic explanation.6, 7, 21 Clinician variability in all aspects of the clinical disc examination is well documented,22-27 but, to our knowledge, there have been no studies documenting glaucoma specialists’ disc and rim margin variability (separate from their variability in assigning a cup disc ratio) so as to assess their separate contributions to rim width variability. We believe that such a
TE D
study, if linked to OCT ONH anatomy in the same eyes, (see below) may provide an anatomic foundation for teaching clinical rim width examination that will improve its accuracy and variability. In the present study, five fellowship trained glaucoma specialists digitally delineated the disc
EP
margin and rim margin within optic disc stereo-photographs of one eye of 214 high risk ocular hypertensive and glaucoma subjects from the Portland Progression Project that had undergone OCT
AC C
ONH imaging within 3 months of photos. Each clinician independently performed their disc margin and rim margin delineations (Figure 1) without prior instruction or awareness of the other clinician’s marks. To characterize the regional variability of the 5 clinician’s marks in each eye, (Figure 2), each stereo-photo was colocalized to the infrared fundus reflectance image obtained as part of the OCT scan acquisition. This approach enabled 30° optic d isc sectoral analysis to be performed on the basis of fixed anatomical landmarks (i.e. the position of the fovea relative to the center of the ONH, Figure 2) rather than on the basis of the acquired image frame.28
ACCEPTED MANUSCRIPT Disc margin, rim margin, rim width and cup disc ratio (CDR) discordance among the five clinicians was assessed by sector, for each eye and across all 214 eyes. In a second analysis, agreement and discordance among the sectors which were suspicious for rim thinning by two criteria (CDR ≥ 0.6 and CDR ≥ 0.7), was separately assessed. The focus of the present study was to quantify
RI PT
clinician rim width discordance without regard for the accuracy of their estimates. In a separate study to follow, the clinician rim width estimates from a subset of the study eyes of the present report will be compared to colocalized OCT MRW measurements,29 so as to assess each clinician’s ability to detect
SC
sectors with thin rim tissue identified by OCT criteria. Methods
M AN U
Background and demographics. This study was performed at the Devers Eye Institute, by five clinicians, included 3 full time attending glaucoma specialists (CFB, RMK, SLM), one visiting clinician-scientist specializing in glaucoma (RR), and one glaucoma fellow (HK). The 214 study subjects were recruited from the Portland Progression Project,28, 30 an NIH-funded, longitudinal study
TE D
of progression in participants with high risk ocular hypertension and glaucoma, that is based at the Devers Eye Institute. The protocol was approved and monitored by the Legacy Health Institutional Review Board. The study adheres to the tenets of the Declaration of Helsinki and complies with the
EP
Health Insurance Portability and Accountability Act of 1996. All participants provided written informed consent, after having the risks and benefits of participation explained to them.
AC C
Study participants had best corrected vision ≥ 20/40 and one or more of the following potential risk factors for glaucoma progression (as perceived at the time of Portland Progression Project study initiation, in 1997): age >70, systemic hypertension, migraine, diet-controlled diabetes, peripheral vasospasm, African ancestry, or family history of glaucoma. Participants with ocular, neurologic or systemic diseases, medications that can affect the visual field (VF) or previous ocular trauma including ocular surgery (except for uncomplicated cataract surgery) were excluded. Each subject underwent a variety of structural and functional tests including pachymetry, stereo-photography, OCT ONH and RNFL imaging and visual field testing.
ACCEPTED MANUSCRIPT Imaging. Stereoscopic ONH photography and OCT imaging were performed within 3 months of each other, most commonly (87%) on the same day. ONH stereo-photographs were obtained using a simultaneous stereoscopic camera (3-Dx; Nidek Co Ltd, Gamagori, Japan) after maximum pupil dilation. The images were acquired on 35-mm slide film, developed and processed into color slides.
RI PT
The slides were digitized with a slide scanner (Nikon LS-5000 ED, Nikon Corporation, Tokyo, Japan) at a resolution of 4800 dpi. OCT imaging included a 30-degree fundus image of the posterior pole including the ONH and fovea (using the integrated confocal scanning laser ophthalmoscope), a 48
SC
radial B-scan data set, (each B-scan was 15-degrees wide, contained 768 A-scans and was the average of 9 repetitions) and a single, standard circumpapillary B-scan used to measure RNFL
M AN U
thickness (Spectralis, Heidelberg Engineering GmbH, Heidelberg, Germany). One eye of each subject was randomly selected for analysis. For each study eye, the illumination and focus of the left and right images of the stereoscopic image-pair were qualitatively compared and the best was chosen for simultaneous digital delineation of the disc and rim margins, as outlined below.
TE D
Clinician disc margin and rim margin assessment within each stereo-photograph (Figure 1). Five glaucoma specialists independently viewed the stereo-photo-pairs of each study eye using a stereoscopic viewer (Screen Vu™, PS Manufacturing, Portland, OR) on a high resolution
EP
computer monitor without prior instruction as to how to mark the disc margin and rim margin points or awareness of the other clinicians’ marks. The disc margin and rim margin points of each examiner
AC C
were digitally “burned” onto a separate copy of the better-focused image of the stereo-photo-pair, using a second, adjacent monitor and custom software. Clinician-specific disc margin, rim margin, rim width and CDR equivalents for each stereo-photo (Figure 2). For each disc photo, each clinician’s disc margin centroid (DMcentroid) was determined and a pooled DMcentroid (defined to be the geometric average of the 5 individual clinician DMcentroids) was separately determined. The photograph containing the pooled DMcentroid was then registered to the infrared fundus reflectance image from the matched OCT data set using a previously described technique,31 performed within Image J (version 1.43u, TurboReg plug-in, National Institutes
ACCEPTED MANUSCRIPT of Health, Bethesda, MD). The line connecting the pooled DMcentroid to the center of the fovea observed in the OCT infrared fundus image was used to define the Foveal-DM axis, which in turn defined the boundaries of each 30-degree (“clock-hour”) sector. Thus, a common set of retinal and ONH anatomical landmarks were employed to consistently define the position of the 30⁰ ONH
RI PT
sectors in each study eye,28 rather than the orientation of the fundus relative to the acquired image frame, which can vary from eye to eye.
Prior to further data analyses, all left eye data were converted into right eye orientation.30 For
SC
each clinician, their disc margin and rim margin points, were connected by straight lines and a single sectoral measurement of DMradius, RMradius and rim width (DMradius – RMradius), were made from the
M AN U
pooled DMcentroid along the centerline of each 30⁰ sector. CDR was not a primary outcome parameter of this study, however sectoral CDRs were calculated as: sectoral RMradius / DMradius and rounded to the nearest single decimal place, (CDR equivalent, Figure 2) to more closely model clinical practice. Thus, for each eye a DMradius, RMradius, rim width and CDR equivalent, (hereafter referred to as CDR)
TE D
for each clinician was calculated for each 30⁰ sector.
Data Analysis. The mean DMradius, RMradius, rim width and CDR for each clinician was calculated across all eyes for each sector. The differences in DMradius, RMradius and rim width among
EP
the 5 clinicians and 12 sectors were assessed by one-way analysis of variance. The coefficient of variation (COV) among the 5 clinicians for each parameter was calculated for each sector for each
AC C
eye. The distribution of these data across all eyes was summarized for each parameter and sector by box plot. To compare the sectoral variation in clinician discordance, the inter-clinician standard deviation for each parameter was defined to be the standard deviation of the 5 clinician’s measures of each parameter in each sector of each study eye. The inter-clinician standard deviation for each parameter was then averaged across all eyes for each parameter by sector. For each parameter the differences in these sectoral averages were assessed by Wilcoxon signed rank test.
ACCEPTED MANUSCRIPT Intra-class correlation coefficients were calculated for each parameter by sector. While systems for grading have been proposed,32 we report these data without qualitative interpretation. To estimate how often clinician rim estimates were discordant in sectors that were suspicious for rim tissue thinning, we defined suspicious sectors in two ways. First, we assessed all sectors in which at
RI PT
least one clinician’s rim width estimate was equivalent to a CDR ≥ 0.6. Second, we assessed all sectors in which at least one clinician’s rim width estimate was equivalent to a CDR ≥ 0.7. For each definition of rim thinning, we assessed clinician agreement by reporting the number of clinicians
SC
whose CDR equivalent was ≥ 0.60, or ≥ 0.7, respectively. We choose these definitions because a series of glaucoma screening studies have chosen a CDR ≥ 0.6 or CDR ≥ 0.7 to be criterion for
M AN U
suspicion that required further evaluation.33-40 Results
This study included 214 eyes of 214 subjects ranging in age from 33 to 90 years old (mean ± standard deviation: 64 ±11 years). Most of the participants were Caucasian (200/214, 93%),
TE D
approximately half (111, 52%) were female, approximately half of the eyes were right eyes 111 (52%), and the mean visual field mean deviation was -0.63 ± 2.86 decibels (standard deviation), (range -16.43 to +3.29). The clinical training and years of experience as a glaucoma specialist for the
EP
5 clinicians ranged from 0 to 19 years.
Mean DMradius, RMradius, rim width and CDR values (averaged across all eyes) are reported by
AC C
sector and by clinician in Figure 3. Interestingly, the clinical disc margin and rim margin determination tendencies of each clinician were generally consistent across the 214 study eyes as evidenced by the fact that the rank-order of clinician means for DMradius, RMradius, rim width and CDR across all eyes were consistent across all 12 sectors. For example, in the case of rim width estimation, clinician 1 manifests the least generous and clinician 5 the most generous estimation in each sector. For rim width, except for the fact that the measurements by clinicians 2 and 3 were not significantly different from one another in 8 of the 12 sectors (P ≥ 0.098), the average measurements across all study eyes for each of the 5 clinicians were different from one another in all 12 sectors (P < 0.001, ANOVA).
ACCEPTED MANUSCRIPT Figure 4 depicts the distribution of the 214 COVs (one for each eye) for DMradius, RMradius and rim width by sector. The median COV for DMradius, RMradius and rim width ranged from 3-5%, 20-25%, and 26-30% by sector, respectively. However, for as many as 10% of the eyes, the sectoral COV for DMradius, RMradius and rim width ranged from 8% to 32%, 25% to 290%, and 38% to 110%,
RI PT
respectively. Representative eyes demonstrating high rim width discordance are shown in Figure 5. High disc and rim margin discordance are shown in Supplemental Figures 1 and 2, respectively. Sectors suspicious for rim thinning by one clinician were frequently overlooked by others.
SC
Figure 6 reports the frequency of clinician CDR agreement in study eye sectors in which at least one clinician’s CDR suggested rim thinning using two criteria: (CDR ≥ 0.6 and CDR ≥ 0.7). Among the
M AN U
2137 sectors in which at least one clinician was suspicious using a criteria of CDR ≥ 0.6, all 5 clinicians’ CDRs were ≥ 0.6 in only 986 (46%) and 2 of the 5 clinicians failed to detect rim thinning (CDR < 0.6) in 592 (28%). Among the 1724 sectors in which at least one clinician was suspicious using a criteria of CDR ≥ 0.7, all 5 clinicians’ CDRs were ≥ 0.7 in only 499 (29%) and 2 of the 5
TE D
clinicians failed to detect rim thinning (CDR < 0.7) in 442 (26%). Among the 214 eyes in which at least one clinician was suspicious for rim thinning in at least one sector by CDR ≥ 0.6 criteria, in 41 (19%) of those eyes, at least 1 of the remaining clinicians was not suspicious in any sector. Among the 211
EP
eyes in which at least one clinician was suspicious in at least one sector by CDR ≥ 0.7 criteria, in 99 (47%) of those eyes, at least 1 of the remaining clinicians was not suspicious in any sector.
AC C
Study subject DIS362, (Figure 5, second row from top), presents two examples of sectors in which at least one clinician’s rim width estimates suggest rim thinning, (clinician 1 superior temporally and Clinicians 1 and 4 inferior temporally), that are not obvious in the rim width estimates of clinicians 2 and 5. In the same figure, the sectoral CDR equivalents for study subject DIS432 (bottom row) are ≥ 0.9 for clinicians 1-4, inferiorly, but clinician 5’s CDR equivalent is < 0.60 in the same sectors. Finally, the inferior temporal sectoral rim width estimates of clinician 1 for subject DIS 409, (third row from the top) produce a CDR equivalent of ≥ 0.8, whereas the CDR equivalents of clinicians 2 and 5 are ≤ 0.6 within the same sectors.
ACCEPTED MANUSCRIPT To assess the sectoral distribution of clinician discordance for DMradius, RMradius, and rim width, a polar plot of the inter-clinician standard deviation for each parameter among the five clinicians for each eye, assessed by sector, and averaged across all 214 eyes is presented in Supplemental Figure 3. The average inter-clinician standard deviation across all eyes for all three parameters was greatest
RI PT
nasally and least temporally, with the variability of inferior and superior sectors being similar in magnitude to each other and midway between that of the nasal and temporal sectors. To assess inter-clinician agreement for each parameter, we report ICCs for DMradius, RMradius, rim width and CDR
SC
equivalent by sector in Supplementary Table 1. Sectoral ICCs for rim width and CDR were greatest in the inferior sector (0.633 and 0.684, respectively) and smallest within the temporal-superior sector
M AN U
(0.481 and 0.566, respectively). Discussion
This study documents that DM and RM demarcations, as well as the corresponding sectoral rim width estimates, varied substantially among 5 glaucoma specialists when they were assessed in a
TE D
masked fashion within stereo-photographs of 214 glaucoma patient and glaucoma suspect eyes. The fact that we did this assessment within 30⁰ sectors that were consistently applied to each study eye (i.e. using consistent retinal and ONH landmarks), not only provides a more robust anatomic
EP
foundation for their regionalization among clinicians and study eyes, but will also allow us to colocalize the rim width estimates of each clinician to OCT MRW measurements, regionalized in the
AC C
same manner, in a follow up study.
To our knowledge, every previous study that has reported the variability of the clinical disc examination has evaluated the reproducibility of global or regional CDR estimates.22-25, 41-47 In these studies, the degree of agreement in vertical CDR ranged from 0.19 to 0.90 using a weighted kappa value,22, 41-46 and 0.44 to 0.855 using ICC.47-49 Among glaucoma specialists, the reported weighted kappa values range between 0.62 and 0.9,44-46 and the ICCs range between 0.85 to 0.861.48, 49 While CDR was not a primary outcome measure of this study, when our glaucoma specialist rim width estimates are expressed in CDR equivalents, the sectoral ICCs ranged from 0.566 to 0.684.
ACCEPTED MANUSCRIPT When the superior and inferior sectoral rim estimates were combined to create a single vertical CDR estimate (data not shown), the ICC value was 0.720, which is substantially lower than most previous reports. As the vertical CDR equivalent of our study is different from the vertical CDR of the clinical disc examination (largest vertical cup diameter / largest disc margin diameter), our results
RI PT
cannot be directly compared to previous studies. However, the fact that our ICC value is lower than previous studies is noteworthy, because we used the FoBMO axis to generate anatomically consistent ONH sectors, which better reflects ‘inter-clinician discordance’ by removing inconsistent
SC
assignment of the “vertical” or “horizontal” CDR measurement location. Yet, even in a study setting that removed these “non-clinician” causes of discordance, ICC values for sectoral rim width ranged
M AN U
from 0.481 to 0.633, and median COV values by sector ranged from 24.3% to 31.9%. Among the factors that contribute to clinician disc margin and rim margin discordance is a lack of agreement on the anatomic landmarks underlying DM and RM. A second factor is that in the absence of anatomic foundations, there is little consistency in the way clinicians are taught how to
TE D
identify the disc and rim margins. The underlying anatomy of the clinical DM has traditionally been thought to be the “scleral lip” or “scleral ring”.2, 3, 19 However a series of recent OCT studies7, 31 have shown that the anatomy underlying the clinical disc margin varies among ONH sectors, but that it is
EP
most commonly defined by the BMO, unpigmented Bruch’s Membrane, and the Border Tissues of Elschnig, rather than the edge of sclera. Less commonly it involves unpigmented peripapillary sclera
AC C
and/or the anterior scleral canal opening. In this study, while disc margin discordance was present and substantial in a subset of eyes, it was small in frequency and magnitude compared to rim margin discordance. However, there is a separate discordance between the anatomic endpoint of BMO, where an accurate measurement of the rim can be most consistently made within OCT anatomy6, 7 and the clinically visible disc margin (Figure 7). The discordance between OCT-identified BMO and the clinical disc margin will affect the accuracy of clinicians’ rim width estimates, regardless of clinician disc margin discordance. This issue will be addressed further in the OCT versus clinician study to follow.
ACCEPTED MANUSCRIPT While the rim margin is commonly defined as the point where the optic disc surface dips below the level of the peripapillary retina,50 the point where the slope first develops an inward deflection and the point where the slope is steepest are also widely used.51 However, the three-dimensional perception of the ONH that can be achieved through stereo-photographic assessment or slit lamp
RI PT
stereo-biomicroscopy is inconsistent and limited compared to OCT. In this study we deliberately did not instruct the clinicians on disc margin and rim margin anatomy because our purpose was to document their current practice rather than to maximize their accuracy and reproducibility. To address
SC
these issues, studies that project clinicians’ rim margin estimates into OCT anatomy are required to determine if OCT identification of a more accurate and reproducible definition for the clinical rim
M AN U
margin can be taught.
CDR equivalent discordance was present in a substantial proportion of the eyes with sectors in which at least one clinician raised suspicion for glaucoma. While CDR rim width estimation is only one portion of the clinical disc examination, given that the clinical disc examination remains the most
TE D
important screening test for detecting the optic neuropathy of glaucoma, our findings indicate that it will be important to develop clinical teaching libraries focused on the sensitive and specific detection of rim thinning that is suspicious or abnormal by OCT criteria.
EP
Although the clinicians showed significant differences from one another, the rank order of clinicians’ average DMradius, RMradius, and rim width estimates were consistent across almost every
AC C
sector (e.g., whereby clinician 1’s estimates were consistently smallest and clinician 5’s consistently largest for rim width). The fact that this finding occurred in a cohort of eyes that encompass a wide range of physiologic and pathophysiologic differences in rim anatomy, suggests that the glaucoma specialists’ tendencies were internally consistent in terms of judging disc and rim margin across this wide range of eyes. This study does not address the accuracy of each clinician examiner, but our follow up study will do so by comparing each clinician’s rim width to OCT MRW by sector. In the study eye of subject DIS362, (Figure 5), clinician 1’s rim width estimates suggest rim thinning in the superior and superior-
ACCEPTED MANUSCRIPT temporal disc sectors that appears to be discordant with the other examiners. Figure 7 depicts colocalized, sectoral OCT MRW measurements for the same eye. The panels of this figure illustrate several points. First, in this eye, most clinicians overestimated the amount of rim tissue temporally, in a region of invisible BMO, which likely contributed to the failure to detect tissues that were thin by
RI PT
OCT criteria.6, 7 Second, while the rim width estimates of clinician 1 (purple with CDR equivalent in (7E)) superiorly and those of Clinicians 1 and 4, inferior temporally, appear to be outliers, in both instances, their estimates reflect areas of OCT suspicion and the discordance generated in these
SC
sectors reflects rim width over-estimation by the other clinicians. By CDR ≥ 0.7 criteria, only 1 of the 5 clinicians detected rim thinning in the temporal inferior quadrant and only 2 of the 5 clinicians
M AN U
detected it temporally. While there was substantial rim width discordance superior temporally, 4 of the 5 clinicians CDR equivalents suggested rim thinning by CDR 0.7 criteria. Whether this discordance and/or these cup disc ratios translate to an accurate or missed diagnosis, requires future studies in which the clinician is asked to identify suspicious sectors.
TE D
To this point specifically, our study is limited by the relatively small number of clinician examiners and the fact that it only addresses clinical rim width estimates within stereo-photographs and does not include the other aspects of the clinical examination such as pallor, nerve fiber layer
EP
thinning and nerve fiber layer hemorrhages that together may have made the glaucoma specialists more concordant in terms of a final diagnosis. We focused on DM, RM and rim width estimation
AC C
because these components of the clinical examination could be quantified and compared to OCT rim anatomy in the studies that will follow. Had we simply asked the clinicians if the disc was suspicious for glaucoma, clinician agreement may have been better even in the face of the rim width discordance we report. While the number of examiners is small, they are equal to or exceed the number of examiners in most previous studies22-25, 41-44, 46-49 and they span a range of training and experience that is likely to be present in most tertiary care glaucoma practices.
ACCEPTED MANUSCRIPT Finally, it is also worth noting that while one clinician was in glaucoma fellowship training during the study and a second was a young visiting clinician-scientist from China, neither of these clinicians, (clinicians 2 and 4, respectively) were outliers when compared to clinicians 1 and 5. Our study was not designed to demonstrate the effect of practice duration on clinician discordance. Such
RI PT
a study would require multiple clinicians at each level of experience over the full duration of an average career and include a non-linear assessment of the post-fellowship learning curve. These kinds of analyses are beyond the scope of the present report.
SC
In summary, our findings show that fellowship trained glaucoma specialists, exhibit substantial discordance in their estimates of the ONH rim tissue present in glaucomatous or at-risk eyes.
M AN U
Discordance occurred commonly in disc sectors that were judged by at least one clinician to be suspicious for demonstrating rim thinning. In a follow-up study, we will compare each clinician’s sectoral rim width estimates to OCT MRW measurements in the same eyes (as shown in Figure 7 of the present report). The goal of this work will be to identify and articulate anatomic explanations for
TE D
clinical DM and RM estimation that can provide guidance for improving the accuracy and consistency
AC C
EP
of clinical rim width estimation by all practioners.
ACCEPTED MANUSCRIPT Acknowledgments/ Disclosure: a. Funding/ Support: NIH/NEI R01-EY-021281; Legacy Good Samaritan Foundation The above listed sponsors/funding organizations had no role in the design, conduct, analysis or reporting of this research.
RI PT
b. Financial Disclosures: SW. Hong: None. R. Ren: None.
SC
H. Yang: None. S. Gardiner: None.
M AN U
C. Hardin: None.
S. Demirel: NIH/NEI R01-EY-019674; Legacy Good Samaritan Foundation; Carl Zeiss Meditec. C. Burgoyne: NIH/NEI R01-EY021281; Legacy Good Samaritan Foundation; Heidelberg Engineering, GmbH, Heidelberg, Germany - Dr. Burgoyne is a consultant to Heidelberg Engineering. In this role,
TE D
he receives unrestricted research support, instruments, software, occasional travel support, but no honorarium or personal income.
R. Kinast: Research: NEI (RO1 EY025181-01), Allergan, Ocular Therapeutix, American Glaucoma
EP
Society, Oregon Health & Sciences University Biomedical Innovation Program. S. Mansberger: Research: NEI (RO1 EY025181-01), Allergan, Ocular Therapeutix, American
AC C
Glaucoma Society; Consultant: Envisia, Santen, Gore,Aerie. B. Fortune: Inotek Pharmaceuticals; Glaucoma Research Foundation; BrightFocus Foundation; Legacy Good Samaritan Foundation. c. Acknowledgements: The authors wish to thank Cindy Albert, and Michael Whitworth for their clinical imaging of the Portland Progression Project participants and Dr. Jonathan He for his assistance with the preliminary statistical analysis and graphing of our data. We would also like to thank Julia Monaghan for assistance with manuscript preparation and submission.
ACCEPTED MANUSCRIPT References
1.
Prum BE, Jr., Rosenberg LF, Gedde SJ, et al. Primary Open-Angle Glaucoma Preferred
Practice Pattern Guidelines. Ophthalmology 2016;123(1):P41-P111. 2.
Nicolela MT. Optic Nerve: Clinical Examination. In: Giaconi JA, Law SK, Coleman AL, Caprioli
RI PT
J, editors. Pearls of Glaucoma Management. Berlin, Heidelberg: Springer Berlin Heidelberg, 2010:1521. 3.
Cantor LB, Rapuano CJ, Cioffi GA. Chapter 3: Clinical Evaluation. In: Basic and Clinical
Science Course (BCSC) 2017-2018: Glaucoma Section 10. San Francisco, CA: American Academy of Ophthalmology, 2017-2018:48.
Armaly MF. Genetic determination of cup/disc ratio of the optic nerve. Arch Ophthalmol
SC
4.
1967;78(1):35-43.
Povazay B, Hofer B, Hermann B, et al. Minimum distance mapping using three-dimensional
M AN U
5.
optical coherence tomography for glaucoma diagnosis. J Biomed Opt 2007;12(4):041204. 6.
Reis AS, O'Leary N, Yang H, et al. Influence of clinically invisible, but optical coherence
tomography detected, optic disc margin anatomy on neuroretinal rim evaluation. Invest Ophthalmol Vis Sci 2012;53(4):1852-60. 7.
Reis AS, Sharpe GP, Yang H, Nicolela MT, Burgoyne CF, Chauhan BC. Optic disc margin
TE D
anatomy in patients with glaucoma and normal controls with spectral domain optical coherence tomography. Ophthalmology 2012;119(4):738-47. 8.
Burk RO, Vihanninjoki K, Bartke T, et al. Development of the standard reference plane for the
Heidelberg retina tomograph. Graefes Arch Clin Exp Ophthalmol 2000;238(5):375-84. Poli A, Strouthidis NG, Ho T, Garway-Heath DF. Analysis of HRT images: comparison of
EP
9.
reference planes. Invest Ophthalmol Vis Sci 2008;49(9):3970-3975. Mwanza JC, Chang RT, Budenz DL, et al. Reproducibility of peripapillary retinal nerve fiber
AC C
10.
layer thickness and optic nerve head parameters measured with cirrus HD-OCT in glaucomatous eyes. Invest Ophthalmol Vis Sci 2010;51(11):5724-30. 11.
Leung CK, Chan WM, Hui YL, et al. Analysis of retinal nerve fiber layer and optic nerve head in
glaucoma with different reference plane offsets, using optical coherence tomography. Invest Ophthalmol Vis Sci 2005;46(3):891-9. 12.
Tan JC, Hitchings RA. Reference plane definition and reproducibility in optic nerve head
images. Invest Ophthalmol Vis Sci 2003;44(3):1132-7. 13.
Mistlberger A, Liebmann JM, Greenfield DS, et al. Heidelberg retina tomography and optical
coherence tomography in normal, ocular-hypertensive, and glaucomatous eyes. Ophthalmology 1999;106(10):2027-32.
ACCEPTED MANUSCRIPT 14.
Shin HY, Park HY, Jung KI, Park CK. Glaucoma diagnosis optic disc analysis comparing
Cirrus spectral domain optical coherence tomography and Heidelberg retina tomograph II. Jpn J Ophthalmol 2013;57(1):41-6. 15.
Yang B, Ye C, Yu M, Liu S, Lam DS, Leung CK. Optic disc imaging with spectral-domain
optical coherence tomography: variability and agreement study with Heidelberg retinal tomograph.
16.
RI PT
Ophthalmology 2012;119(9):1852-7. Girkin CA, McGwin G, Jr., Long C, DeLeon-Ortega J, Graf CM, Everett AW. Subjective and
objective optic nerve assessment in African Americans and whites. Invest Ophthalmol Vis Sci 2004;45(7):2272-8. 17.
Deleon-Ortega JE, Arthur SN, McGwin G, Jr., Xie A, Monheit BE, Girkin CA. Discrimination
SC
between glaucomatous and nonglaucomatous eyes using quantitative imaging devices and subjective optic nerve head assessment. Invest Ophthalmol Vis Sci 2006;47(8):3374-80. Vessani RM, Moritz R, Batis L, Zagui RB, Bernardoni S, Susanna R. Comparison of
M AN U
18.
quantitative imaging devices and subjective optic nerve head assessment by general ophthalmologists to differentiate normal from glaucomatous eyes. J Glaucoma 2009;18(3):253-61. 19.
Fingeret M, Medeiros FA, Susanna R, Jr., Weinreb RN. Five rules to evaluate the optic disc
and retinal nerve fiber layer for glaucoma. Optometry 2005;76(11):661-8. 20.
Susanna R, Jr., Vessani RM. New findings in the evaluation of the optic disc in glaucoma
21.
TE D
diagnosis. Curr Opin Ophthalmol 2007;18(2):122-8.
Chauhan BC, O'Leary N, Almobarak FA, et al. Enhanced detection of open-angle glaucoma
with an anatomically accurate optical coherence tomography-derived neuroretinal rim parameter. Ophthalmology 2013;120(3):535-43.
Varma R, Steinmann WC, Scott IU. Expert agreement in evaluating the optic disc for
EP
22.
glaucoma. Ophthalmology 1992;99(2):215-21. Klein BE, Magli YL, Richie KA, Moss SE, Meuer SM, Klein R. Quantitation of optic disc
AC C
23.
cupping. Ophthalmology 1985;92(12):1654-6. 24.
Lichter PR. Variability of expert observers in evaluating the optic disc. Trans Am Ophthalmol
Soc 1976;74:532-72. 25.
Varma R, Spaeth GL, Steinmann WC, Katz LJ. Agreement between clinicians and an image
analyzer in estimating cup-to-disc ratios. Arch Ophthalmol 1989;107(4):526-9. 26.
Jampel HD, Friedman D, Quigley H, et al. Agreement among glaucoma specialists in
assessing progressive disc changes from photographs in open-angle glaucoma patients. Am J Ophthalmol 2009;147(1):39-44 e1. 27.
Lloyd MJ, Mansberger SL, Fortune BA, et al. Features of optic disc progression in patients with
ocular hypertension and early glaucoma. J Glaucoma 2013;22(5):343-8.
ACCEPTED MANUSCRIPT 28.
He L, Ren R, Yang H, et al. Anatomic vs. acquired image frame discordance in spectral
domain optical coherence tomography minimum rim measurements. PLoS One 2014;9(3):e92225. 29.
Chauhan BC, Danthurebandara VM, Sharpe GP, et al. Bruch's Membrane Opening Minimum
Rim Width and Retinal Nerve Fiber Layer Thickness in a Normal White Population: A Multicenter Study. Ophthalmology 2015;122(9):1786-94. Ren R, Yang H, Gardiner SK, et al. Anterior lamina cribrosa surface depth, age, and visual
RI PT
30.
field sensitivity in the Portland Progression Project. Invest Ophthalmol Vis Sci 2014;55(3):1531-9. 31.
Strouthidis NG, Yang H, Reynaud JF, et al. Comparison of clinical and spectral domain optical
coherence tomography optic disc margin anatomy. Invest Ophthalmol Vis Sci 2009;50(10):4709-18. 32.
Cicchetti DV. Guidelines, criteria, and rules of thumb for evaluating normed and standardized
33.
SC
assessment instruments in psychology. Psychological Assessment 1994;6(4):284-290.
Iwase A, Suzuki Y, Araie M, et al. The prevalence of primary open-angle glaucoma in
34.
M AN U
Japanese: the Tajimi Study. Ophthalmology 2004;111(9):1641-8.
Kyari F, Entekume G, Rabiu M, et al. A Population-based survey of the prevalence and types
of glaucoma in Nigeria: results from the Nigeria National Blindness and Visual Impairment Survey. BMC Ophthalmol 2015;15:176. 35.
Gupta P, Zhao D, Guallar E, Ko F, Boland MV, Friedman DS. Prevalence of Glaucoma in the
Vis Sci 2016;57(6):2905-2913. 36.
TE D
United States: The 2005-2008 National Health and Nutrition Examination Survey. Invest Ophthalmol
Bokman CL, Pasquale LR, Parrish RK, 2nd, Lee RK. Glaucoma screening in the Haitian Afro-
Caribbean population of South Florida. PLoS One 2014;9(12):e115942. 37.
Kim CS, Seong GJ, Lee NH, Song KC, Namil Study Group KGS. Prevalence of primary open-
38.
EP
angle glaucoma in central South Korea the Namil study. Ophthalmology 2011;118(6):1024-30. Iwase A, Sawaguchi S, Sakai H, Tanaka K, Tsutsumi T, Araie M. Optic disc, rim and
AC C
peripapillary chorioretinal atrophy in normal Japanese eyes: the Kumejima Study. Jpn J Ophthalmol 2017;61(3):223-229. 39.
Pan CW, Zhao CH, Yu MB, et al. Prevalence, types and awareness of glaucoma in a multi-
ethnic population in rural China: the Yunnan Minority Eye Study. Ophthalmic Physiol Opt 2016;36(6):664-670. 40.
Ashaye A, Ashaolu O, Komolafe O, et al. Prevalence and Types of Glaucoma Among an
Indigenous African Population in Southwestern Nigeria. Investigative Ophthalmology & Visual Science 2013;54(12):7410-7416. 41.
Tielsch JM, Katz J, Quigley HA, Miller NR, Sommer A. Intraobserver and interobserver
agreement in measurement of optic disc characteristics. Ophthalmology 1988;95(3):350-6.
ACCEPTED MANUSCRIPT 42.
Abrams LS, Scott IU, Spaeth GL, Quigley HA, Varma R. Agreement among optometrists,
ophthalmologists, and residents in evaluating the optic disc for glaucoma. Ophthalmology 1994;101(10):1662-7. 43.
Harper R, Reeves B, Smith G. Observer variability in optic disc assessment: implications for
glaucoma shared care. Ophthalmic Physiol Opt 2000;20(4):265-73. Kong YX, Coote MA, O'Neill EC, et al. Glaucomatous optic neuropathy evaluation project: a
RI PT
44.
standardized internet system for assessing skills in optic disc examination. Clin Exp Ophthalmol 2011;39(4):308-17. 45.
Chan HH, Ong DN, Kong YX, et al. Glaucomatous optic neuropathy evaluation (GONE)
project: the effect of monoscopic versus stereoscopic viewing conditions on optic nerve evaluation.
46.
SC
Am J Ophthalmol 2014;157(5):936-44.
Bastawrous A, Giardini ME, Bolster NM, et al. Clinical Validation of a Smartphone-Based
47.
M AN U
Adapter for Optic Disc Imaging in Kenya. JAMA Ophthalmol 2016;134(2):151-8. Spry PG, Spencer IC, Sparrow JM, et al. The Bristol Shared Care Glaucoma Study: reliability
of community optometric and hospital eye service test measures. Br J Ophthalmol 1999;83(6):707-12. 48.
Arthur SN, Aldridge AJ, De Leon-Ortega J, McGwin G, Xie A, Girkin CA. Agreement in
assessing cup-to-disc ratio measurement among stereoscopic optic nerve head photographs, HRT II, and Stratus OCT. J Glaucoma 2006;15(3):183-9.
Tanito M, Sagara T, Takamatsu M, et al. Intraobserver and interobserver agreement of
TE D
49.
computer software-assisted optic nerve head photoplanimetry. Jpn J Ophthalmol 2014;58(1):56-61. 50.
Portney GL. Photogrammetric analysis of the three-dimensional geometry of normal and
46. 51.
EP
glaucomatous optic cups. Trans Sect Ophthalmol Am Acad Ophthalmol Otolaryngol 1976;81(2):239-
Sommer A, Pollack I, Maumenee AE. Optic disc parameters and onset of glaucomatous field
AC C
loss. I. Methods and progressive changes in disc morphology. Arch Ophthalmol 1979;97(8):1444-8.
ACCEPTED MANUSCRIPT Figure Legends
Figure 1. A stereo-photo (upper left and middle) of each ONH was individually viewed by each clinician while simultaneously marking the Disc Margin (green dots) and Rim Margin (black
RI PT
dots) within one image from the stereo-pair viewed within custom marking software on an adjacent monitor (upper right). Without prior instruction, for each study eye, each clinician used a stereoscopic viewer to examine its stereo disc photograph (left and middle) on a high resolution
SC
computer monitor. Simultaneous marking of the disc margin and rim margin on the better focused of the two stereo-photo images (left or right) occurred within custom marking software on a second,
M AN U
adjacent monitor. For this eye the right image of the stereo-pair (right panel) was preselected to be used to mark the disc margin and rim margin points. (Lower) Disc margin and rim margin points along with the resultant rim width (purple shading) variability amongst the 5 glaucoma specialists for
TE D
this study eye.
Figure 2. Establishing a consistent set of twelve 30
(clock hour) sectors for the clinician disc
margin, rim margin and rim width measurements of each eye. (A) The geometric centroid (center
EP
of gravity) for each clinician's disc margin delineation (DM centroid) was calculated. (B) A pooled DM centroid (black dot) was calculated by averaging the x,y coordinates of the 5 clinicians’ DM centroids.
AC C
(C) The photograph containing the pooled DM centroid was colocalized to the infrared fundus reflectance image from an optical coherence tomography (OCT) optic nerve head data set acquired within 3 months of the photograph. A line connecting the pooled DM centroid (black dot) and the foveal center (blue dot) was drawn to establish the Foveal-DM axis, which was used as the horizontal axis for twelve 30⁰ (clock hour) sectors (white borders). Note, that because the Foveal-DM axis was used as the nasal-temporal axis for each eye, rather than the horizontal axis of the acquired image, the regions are consistent for all 5 clinicians among all 214 study eyes. (D) The disc margin and rim margin points for clinician 3, (both green). (E) Clinician 3’s disc margin (still in green) and rim margin
ACCEPTED MANUSCRIPT points (now black) connected by straight lines. (F) For each sector, a single measurement of DM radius, RM radius and rim width (RW – in purple) was made along the midline of the sector. While not a primary outcome parameter, a cup disc ratio (CDR) was calculated for each sector as (sectoral RW radius / sectoral DM radius) which was then rounded to the nearest single decimal place (CDR
RI PT
equivalent) to more closely model clinical practice.
Figure 3. Polar plots of the mean values for disc margin radius and rim margin radius, (left),
SC
rim width (center), and cup disc ratio equivalent, (right), assessed by sector and averaged for each clinician across all 214 eyes. Data are in microns. Interestingly, the delineation tendencies of
M AN U
each clinician were generally consistent across the 214 study eyes as evidenced by the fact that the rank-order of clinician means for disc margin radius, rim margin radius, rim width and cup disc ratio equivalent were consistent across all sectors.
TE D
Figure 4. Distribution box-plots of the coefficient of variation (COV) among the 5 clinicians for each study eye, for each parameter, by sector. The central box of each plot defines the median with the upper and lower quartiles values. The thick and thin lines define the upper and lower 10thth
EP
and 5th percentiles, (respectively) of the distribution. The median COV for DM radius, RM radius and rim width ranged from 3 to 5%, 20 to 25%, and 26 to 30% by sector, respectively. However, for as
AC C
many as 10% of the eyes, the sectoral COV for DM radius, RM radius and rim width ranged from 8% to 32%, 25% to 290%, and 38% to 110%, respectively. Representative study eyes demonstrating high rim width discordance are shown in Figure 5. High disc margin and rim margin discordant eyes can be seen in Supplemental Figures 1 and 2, respectively.
Figure 5. Four representative high rim width discordance eyes.
ACCEPTED MANUSCRIPT Figure 6. Frequency of clinician agreement in study eye sectors in which at least one clinician’s CDR equivalent suggests rim thinning using two CDR criteria, (CDR ≥ 0.6 and CDR ≥ 0.7, respectively). (Left Column – Suspicion and Agreement Criteria CDR ≥ 0.6) For each sector in which at least one clinician’s CDR was ≥ 0.6, (n=2137 sectors), the level of agreement is
RI PT
reported as the number of clinicians whose CDR equivalent was ≥ 0.6. (Right Columns – Suspicion and Agreement Criteria CDR ≥ 0.7) For each sector in which at least one clinician’s CDR was ≥ 0.7, (n = 1724 sectors), the level of Agreement is reported as the number of clinicians whose CDR
SC
equivalent was ≥ 0.7. For all columns the darkest color denotes the number of sectors in which all 5 clinicians were suspicious for rim thinning using the stated criteria. The lightest color denotes the
were in agreement, by the stated criteria.
M AN U
number of sectors and eyes in which 1 clinician was suspicious and none of the other four clinicians
Figure 7. Clinician Disc Margin, Rim Margin and Rim Width estimates co-localized to sectoral
TE D
OCT minimum rim width (MRW) anatomy in a study eye (DIS362 from Figure 5, above) in which clinician discordance occurs in sectors that are structurally suspicious by OCT criteria. (A) Fundus photograph. (B) Bruch’s membrane opening (BMO), the centroid of BMO, the Foveal – BMO
EP
axis (FoBMO axis, red line) and the 12 FoBMO 30⁰ (clock hour) sectors. (C). BMO relative to the disc margin points of all four clinicians. Note that BMO is well inside the clinical disc margin points of
AC C
all 5 clinicians temporally, (black arrows), suggesting that BMO is clinically invisible in this region, as has been previously described6, 7 (D, Center Panel).. BMO, and each clinician’s disc margin and rim margin points have been superimposed on the funds photo in the center, and OCT rim anatomy along with its quantification relative to a Caucasian Normative Data Base29 are shown for each FoBMO sector. Note that the superior temporal, temporal and temporal inferior sector OCT anatomy boxes have been outlined in yellow because the OCT MRW anatomy (yellow arrow) is in the lowest 5th percentile of the NDB. Notice also that while they are considered “normal” the MRW values for the adjacent temporal superior and superior sectors are in the 8th and 9th percentiles, respectively. In a
ACCEPTED MANUSCRIPT white circle at the top of each OCT anatomy box, the number of clinicians that were suspicious for rim thinning by CDR ≥ 0.7 criteria are noted (E-I, bottom row). Each clinician’s rim width, (purple) and their CDR equivalent for the three sectors that are suspicious by OCT criteria,(in yellow) are shown.
RI PT
Supplemental Figure 1. Four representative high disc margin discordance eyes.
SC
Supplemental Figure 2. Four representative high rim margin discordance eyes.
Supplemental Figure 3. Polar plot of disc margin, rim margin and rim width discordance
M AN U
among the five clinicians for each eye, assessed by sector, and averaged across all 214 eyes. Data for each sector are the inter-clinician standard deviation for each parameter for each eye, averaged across all study eyes. On average the sectoral distribution of clinican discordance was similar for all three parameters and was greatest nasally, similarly moderate superiorly and inferiorly
AC C
EP
TE D
and least temporally.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT