Prevalence and Factors Associated with Optic Disc Tilt in the Primary Open-Angle African American Glaucoma Genetics (POAAGG) Study

Abstract

Purpose:

To investigate the prevalence and factors associated with optic disc tilt in the eyes of African Americans with glaucoma.

Design:

Cross-sectional.

Participants:

Subjects with glaucoma participating in the Primary Open-Angle African American Glaucoma Genetics (POAAGG) study.

Methods:

Stereo pairs of optic disc images were assessed independently by POAAGG-certified non-physician graders for quantitative features including maximum and minimum linear disc diameters, and qualitative features including gradeability of images, shape of the cup, rim plane position, beta-peripapillary atrophy, sloping region adjacent to the outer disc margin, and rim pallor. Discrepancies were adjudicated by an ophthalmologist. Descriptive statistics and p-values were generated for associations of tilt with demographic and ocular characteristics. Stepwise multivariable analysis was performed with logistic regression using Generalized Estimating Equations (GEE) to account for inter-eye correlation within subjects.

Main Outcome Measures:

Tilt defined by Ovality Index (TOI) of >1.3 and Stereoscopically Identified optic disc Tilt (SIT).

Results:

Among 1251 subjects with data on both eyes, 104 (8.3%) had TOI. Subjects with TOI were less likely to be male (Adjusted Odds Ratio [aOR] 0.46, 95% CI 0.29–0.74, p<.001), Eyes with TOI were less likely to have large Cup Disc ratios (aOR 0.18, 95% CI 0.06–0.53, p<0.001) and less likely to have cylinder shaped cups compared to conical shaped cups (aOR 0.31, 95% CI 0.19–0.49, p<0.001). Among 1007 subjects with data on both eyes, 254 (25.2%) had SIT. Subjects with SIT were younger (aOR 0.95, 95% CI 0.93–0.96, p<0.001) and eyes with SIT were more likely to have oval shaped discs compared to round discs (aOR 1.82, 95% CI 1.32–2.52, p<0.001), more likely to have a sloping region adjacent to the outer disc margin instead of being flat (aOR 3.26, 95% CI 2.32–4.59, p<0.001) and less likely to have cylinder shaped cups compared to conical shaped cups (aOR 0.59, 95% CI 0.41–0.85, p<0.001). Both TOI and SIT were not associated with myopia.

Conclusions:

There are substantial numbers of tilted optic discs in glaucoma patients with African American ancestry. They occur more frequently in females and younger subjects and are associated with several ocular features but not with myopia.

A titled disc is an optic nerve that enters the eye at an oblique angle with rotation along the anterior-posterior axis.¹ Tilted discs have a rotation of the long axis of the optic nerve head (ONH) from a vertical reference point. Disc torsion occurs when the longest axis of the ONH falls beyond 15 degrees of the vertical axis.² Both tilted and torted discs have a wide phenotypic spectrum but most reports in the literature do not differentiate between tilt and tort when characterizing features of the ONH.

The prevalence of tilted discs in the general population is low, ranging from 0.09% to 3.5%.^3–6 Since tilted discs have largely been associated with myopia most investigations on tilted discs have focused on myopia or myopia occurring with concomitant glaucoma. In select populations with myopia, the prevalence of disc tilt is high, ranging from 18% to 57%.^4–9 Among Chinese Americans with myopia, 28% had tilted discs.⁶

Tilted discs have been reported to be associated with several features of primary open angle glaucoma (POAG). These features include lamina cribrosa defects^7,8; retinal nerve fiber layer (RNFL) thinning^9,10; faster rates of visual field progression^11,12; protection against glaucoma progression (various combinations of disc morphology changes, VF changes and RNFL thinning)^13,15; and a likelihood of overestimation of glaucoma due to incorrect assessment of tilt.¹⁶

POAG affects 2.22 million people in the United States.¹⁷ African Americans have 3 to 4 times higher age-adjusted prevalence of glaucoma than European Americans.¹⁸ The prevalence of titled discs among African Americans is not known. The Primary Open-Angle African American Glaucoma Genetics (POAAGG) study seeks to identify genetic and other risk factors that underlie POAG in African Americans and to understand the basis for the increased disease burden in this population.¹⁹ In this cross-sectional study, we investigated the prevalence of tilted discs among subjects with glaucoma in the POAAGG cohort and the possible associations with baseline demographic, ocular, and genetic characteristics.

Methods:

Subjects

The POAAGG study population includes self-identified African ancestry subjects (Black, Afro-Caribbean, or African American), aged 35 years or older, recruited from Philadelphia, Pennsylvania, USA. Subjects were identified from comprehensive and subspecialty ophthalmology clinics at the University of Pennsylvania and satellites, as well as two neighboring ophthalmology clinics in Philadelphia, Pennsylvania (Windell Murphy, MD; Temple University). Each subject received a clinical examination, including an onsite interview and ophthalmic examination. Examination data were recorded on case report forms, which were entered into the REDCap (Research Electronic Data Capture) database. Additionally, retrospective ophthalmic and systemic health data were extracted from the UPenn EPIC and MERGE databases.

A fellowship-trained glaucoma specialist classified each subject as a glaucoma case or control. Details on eligibility criteria, phenotyping, and baseline characteristics have been described previously.²⁰ All subjects signed an informed consent form and provided a genomic DNA sample. The University of Pennsylvania Institutional Review Board approved the study and the informed consent process, and the research adhered to the tenets of the Declaration of Helsinki.

Grading of Color Images

In this study, 30-degree color stereo disc photos of POAAGG subjects which were taken using the Topcon TRC 50EX retinal camera (Topcon Corp. of America, Paramus, New Jersey, USA) were analyzed. The stereo color images were uploaded to the Ophthalmology Reading Center at the University of Pennsylvania using a secure server. Images were taken between 01/13/2004 and 06/25/2019, uploaded to the Reading Center server between 01/22/2016 and 04/20/2021, and graded at the Reading Center between 06/06/2016 to 05/10/2021.

Three non-physician graders were trained by two glaucoma specialists to grade digital stereo color images of the optic disc using a stereo viewer (Screen-Vu stereoscope, Portland, OR). The details of this training have previously been described.²¹ For each patient visit, two graders independently performed quantitative and qualitative assessments on each stereo color image pair.

We selected an enriched sample of 30 eyes (10 eyes with tilt and 20 eyes without tilt based on the original adjudicated grading) for regrading independently by two graders with adjudication by the director for discrepancies. We assessed the agreement between original adjudicated grading vs. adjudicated regrading for presence of SIT and the shape of the cup using percent of agreement and Kappa.

Quantitative Grading of the ONH

Each optic nerve image was analyzed by two of the three trained graders, who were masked to all patient details. A detailed description of quantitative grading and reliability has been previously reported.²¹ In brief, the graders outlined seven structures on the ONH and adjacent peripapillary atrophy areas using the Image J/Fiji software (available at http://rsbweb.nih.gov/ij/; Rasband WS, Image J, US National Institutes of Health, Bethesda, MD, 1997e2012), while viewing the stereo images through a stereo viewer. The outlined structures included: the optic cup using the paler color cues of the cup, the optic cup contour using a combination of the color and vascular cues, the optic disc outer rim margin inner to the scleral ring, the outer border of the scleral ring, the outer border of beta peripapillary atrophy if present, the outer border of alpha peripapillary atrophy if present, and the site of the exiting central vessel trunk.

The height and width of these measurements were then calculated using the Image J/Fiji software. Region of interest (ROI) files that contained the coordinates of these outlines were saved for each image. The dice’s coefficients for both the disc and the cup contour ([2*area overlap of graders 1 and 2] / [area of grader 1 + area of grader 2]) were calculated. The difference between the two graders on the vertical cup-to-disc ratio (VCDR) of each image was also compared. When the dice’s coefficient was below 0.7, or the difference between graders of VCDR was greater than 0.2, the image was sent to adjudication by the Reading Center Director, an ophthalmologist, who reviewed the stereoscopic images to determine if a new drawing was required or if one of the graders drawings could be used when it was close to his evaluation. New drawings, if required, were produced by the Reading Center Director.

Qualitative Grading of the ONH

A standardized grading form was completed independently by two graders for each stereo image pair. Graded features included: quality of the stereo pair of images; disc shape (round, oval, other); shape of the cup (reader makes an assessment whether the walls of the cup slope towards the center [conical], abruptly drop vertically down [cylindrical], or are excavated [partial and complete bean-pot] (Figure 1 available at http://www.aaojournal.org), with the predominant shape chosen in instances where the configuration is mixed); cup depth (shallow, moderate, deep); rim plane position (reader makes an assessment to determine if the rim was at the same level in its entirety and if it was not, records the area where it was depressed); presence of beta and alpha peripapillary atrophy (PPA) at each clock hour surrounding the disc; sloping towards the outer rim of the disc; presence and location of hemorrhages; baring of lamina cribrosa and circumlinear vessels; bayonetting; cilioretinal vessels; gray crescents; and pallor of the neural rim. Discrepancies between the two graders were adjudicated by the Reading Center Director who reviewed the stereoscopic images used by the graders and determined if one of the values could be used or if a new value needed to be generated.

Measurement of Tilt

The tilt of the optic disc was evaluated using both quantitative and qualitative and methods:

1. Quantitative: The Tilt Ovality Index (TOI). Using a semi-automated approach developed in MATLAB 2018a (The MathWorks Inc., Natick, Massachusetts, USA) for each eye, the grader delineation of the optic disc was used to find the length of the maximum (L_Max) and minimum (L_Min) axis of the optic disc. The center of mass of the delineated disc region was used as the center point for both axes. The TOI was calculated as the ratio between the length of the maximum disc axis and the length of the minimum disc axis.^5,14,23–25 Eyes with a TOI ≥1.30 were defined as having a tilted optic disc, as in previous epidemiologic studies.^5,25

2. Qualitative: Stereoscopically Identified optic disc Tilt (SIT). Optic disc tilting was considered present when one margin of the optic disc was depressed below the opposite margin, as seen from stereoscopic photographs using a stereoscopic viewer.^5,22

Demographic Characteristics

During the onsite enrollment interview, Clinical Research Coordinators collected demographic, behavioral, and systemic disease information from each POAAGG subject. Demographic information included age at enrollment, gender, and self-described race. Behavioral information, including tobacco and alcohol use, was collected using a questionnaire.²¹ Information on history of systemic diseases (diabetes, hypertension), anthropometric measures (BMI), and family history of glaucoma were also obtained at the enrollment interview and from electronic medical records, as detailed in a previous publication.²⁷

Ocular Characteristics

All ocular characteristics were obtained during the onsite examination and were stored in REDCap. For this study, we specifically examined refractive error, intraocular pressure (IOP), central corneal thickness (CCT), and cup-to-disc ratio (CDR). Previous glaucoma surgery, previous cataract surgery, and presence/absence of myopia were also recorded. Qualitative features of the cup and disc, described above, were recorded on a standardized pre-formatted grading form and were also included in this analysis.

Genotype

We previously reported that the genome-wide association study (GWAS) identified several SNPs implicated in POAG and in quantitative traits associated with this disease in the POAAGG cohort.²⁸ We selected 2 SNPs from the POAAGG GWAS and tested their association with tilt in this study. Genotypes for the SNP rs112369934 (TT wildtype and TC variant) implicated with case-control status near the TRIM66 gene on chr11 and the SNP rs187699205 (GG wildtype and GC variant) SNP in the intron of the LMX1B gene, associated with the mean deviation of visual field loss were tested.²⁸ The frequency of the homozygous genotypes CC (rs112369934) and CC (rs187699205) is very low and so subjects with these genotypes were removed from further analysis for the corresponding SNPs.

Statistical Analysis

Both TOI and SIT outcomes were analyzed separately at the eye level. The laterality of each outcome was described by cross-tabulations comparing left and right eyes. Descriptive statistics and p-values from logistic regression analyses for presence of tilt were used to assess the association of tilt with various characteristics in univariate analyses. Continuous, numeric characteristics were analyzed both as a continuous variable and as a categorical variable defined by quartiles. Significant (p<0.05) characteristics were then included in a forward stepwise multivariable analysis starting with the variable with the lowest p-value on univariate analysis. P-values for both the univariate and multivariate analyses were derived from logistic regression using Generalized Estimating Equations (GEE) to account for inter-eye correlation within subjects. All analyses were performed using R Version 4.03.

Results

Tilt Identified by Tilt Ovality Index (TOI) ≥1.3

Among 2563 eyes of 1308 subjects, 124 (5%) eyes had an ovality index of ≥1.3 (i.e., TOI discs). Out of the 1308 subjects, 1251 had non-missing information on both right and left eyes (Figure 2 available at http://www.aaojournal.org). Among these 1251 subjects 104 (8.3%) had 1 or 2 tilted discs. Of the 104 subjects with TOI, 88 (84.6%) were unilateral and 16 (15.4%) were bilateral.

The univariate analysis results of participant associations with TOI are shown in Table 1. Subjects who had a TOI of ≥1.30 (tilted discs) were more likely to be female (76% vs 59%, p<0.001) and have a higher body mass index (BMI) (mean (SD): (31 (7) vs 30(7), p=0.045). The univariate analysis results of ocular associations with TOI is shown in Table 2. Eyes with tilt had more oval discs (99% vs 49%, p<0.001), more excavated PPA (34% vs 24%, p=0.03), more conical shaped cups (64% vs 36%, p<0.001), less baring of the lamina cribrosa (53% vs 70%, p=0.002), and less beta PPA (54% vs 67% p=0.007). Additionally, the mean CDR was lower among subjects with tilt (mean (SD): (0.65 (0.17) vs 0.71 (0.17), p=0.001). There was no association of TOI to either genotype (TRIM66, LMX1B) examined in this study (Table 1). The relationship between TOI and the spherical equivalent refractive error is shown in Figure 2.

Table 1:

Univariate Associations of Subject Characteristics with Tilt Ovality Index (>1.30)

Subject Characteristics	No Tilt (N=2439)	Tilt (N=124)	P value*
Age (years)			0.70
Mean (SD)	69.33 (11.28)	68.91 (10.62)
Gender			0.001
Female	1445 (59.2%)	94 (75.8%)
Male	994 (40.8%)	30 (24.2%)
Basal Metabolic Index			0.66
Mean (SD)	29.97 (6.66)	29.77 (6.38)
Diabetes			0.36
No	1321 (59.8%)	71 (64.5%)
Yes	889 (40.2%)	39 (35.5%)
Unknown	229	14
Hypertension			0.23
No	467 (21.1%)	29 (26.4%)
Yes	1743 (78.9%)	81 (73.6%)
Unknown	229	14
Family History of Glaucoma			0.09
No	922 (41.7%)	36 (32.7%)
Yes	1288 (58.3%)	74 (67.3%)
Unknown	229	14
Previous Glaucoma Surgery			0.948
No	1534 (69.4%)	76 (69.1%)
Yes	676 (30.6%)	34 (30.9%)
Unknown	229	14
LMX1B Genotype			0.78
GC	48 (2.0%)	3 (2.4%)
GG	2391 (98.0%)	121 (97.6%)
TRIM66 Genotype			0.26
TC	281 (11.5%)	19 (15.3%)
TT	2154 (88.5%)	105 (84.7%)
Unknown	4	0

^*GEE logistic regression

Table 2:

Associations of Ocular Characteristics with Tilt Ovality Index (>1.30)

Ocular Characteristics	No Tilt (N=2439)	Tilt (N=124)	p value*
Disc Shape			< 0.001
Round	1050 (50.7%)	1 (0.9%)
Oval	1020 (49.3%)	106 (99.1%)
Unknown	369	17
Shape of Cup			< 0.001
Conical	711 (35.7%)	63 (64.3%)
Cylindrical	1011 (50.8%)	26 (26.5%)
Unknown	448	26
Sloping Margin Adjacent to Outer Disc Rim			0.03
No	1522 (75.6%)	67 (65.7%)
Yes	490 (24.4%)	35 (34.3%)
Unknown	427	22
Baring of Lamina Cribrosa			0.002
No	628 (30.5%)	49 (46.7%)
Yes	1432 (69.5%)	56 (53.3%)
Unknown	379	19
Cilio Retinal Vessels			0.50
No	1618 (78.1%)	85 (81.0%)
Yes	453 (21.9%)	20 (19.0%)
Unknown	368	19
Gray Crescent			0.43
No	1925 (92.6%)	96 (90.6%)
Yes	154 (7.4%)	10 (9.4%)
Unknown	360	18
Pallor of the Disc			0.19
No	2005 (96.5%)	105 (99.1%)
Yes	72 (3.5%)	1(0.9%)
Unknown	362	18
Beta PPA			0.007
No	683 (33.0%)	48 (46.2%)
Yes	1388 (67.0%)	56 (53.8%)
Unknown	368	20
Cup to Disc Ratio			0.001
Mean (SD)	0.71 (0.17)	0.65 (0.17)
Unknown	27	2
Central Corneal Thickness			0.08
Mean (SD)	533.01 (39.70)	539.41 (37.98)
Unknown	50	0
Clinical Eye Characteristics
Refractive Error (Diopters)			0.68
Mean (SD)	−0.20 (1.87)	−0.30 (2.18)
Unknown	916	46
Myopia			0.98
No	725 (47.6%)	37 (47.4%)
Yes	798 (52.4%)	41 (52.6%)
Unknown	916	46
Intraocular Pressure (mmHg)			0.55
Mean (SD)	16.47 (5.21)	16.22 (4.14)
Unknown	8	0
Corrected Visual Acuity (Log MAR)			0.32
Mean (SD)	0.32 (0.71)	0.27 (0.50)
Unknown	164	6

^*GEE logistic regression

Multivariable analysis of TOI >1.30 showed that subjects with TOI were less likely to be male (adjusted Odds Ratio (aOR) 0.46, 95% CI 0.29–0.74, p<.001). Eyes with TOI were also less likely to have large CDRs (aOR 0.18, 95% CI 0.06–0.53, p<0.001) and less likely to have cylinder shaped cups compared to conical shaped cups (aOR 0.31, 95% CI 0.19–0.49, p<0.001) (Table 3).

Table 3.

Multivariable analyses for Tilt Ovality Index (TOI)

Patient Characteristics	Adjusted Odds Ratio	95% Confidence Limits	P value
Male	0.46	(0.29, 0.74)	0.001
Ocular Characteristics	Adjusted Odds Ratio	95% Confidence Limits	P value
Cylindrical Shape of cup	0.31	(0.19, 0.49)	<.001
Cup to Disc Ratio	0.18	(0.06, 0.53)	0.002

Stereoscopically Identified Tilt (SIT)

Among 2135 eyes of 1128 subjects with glaucoma, 330 (15.5%) eyes had SIT. Out of the 1128 subjects, 1007 had non-missing information on both right and left eyes (Figure 1). In these subjects, 254 (25.2%) had tilt in at least one eye. Among these 254 people, 199 (78.3%) were unilateral and 55 (21.7%) were bilateral.

The results of the univariate analysis of subjects with SIT are shown in Table 4. When compared with subjects without tilted discs, subjects with SIT were younger, with a mean age (SD) of 65.15 (11.08) vs 70.36 (10.93) years (p<0.001) and were more likely to be female (67% vs 60%, p=0.02). A higher proportion of subjects with tilt had a family history of glaucoma (66% vs 57%, p=0.02) and had had previous glaucoma surgery (33% vs 24%, p=.005). The results of the univariate analysis of ocular features with SIT are shown in Table 5. Eyes with tilt had a higher proportion of oval shaped optic discs (67% vs 50%, p<0.001), conical shaped cups (rather than cylinder shaped cups) (50% vs 35%, p<.001), excavated peripapillary atrophy (44% vs 21%, p<0.001) and grey crescents (13% vs 7%, p<0.001). Eyes with SIT had a lower proportion with pallor of the disc (0.3% vs 3.8%, p=0.01). Eyes with tilt also had more myopic refractive errors (mean (SD): (−0.5 (2) vs −0.1 (2), p=0.03). SIT was not different in subjects carrying the altered genotype for both the variants tested, as shown in Table 4. The percentages of subjects with tilt compared to subjects without tilt carrying the TC genotype near the TRIM66 gene were 12.7% vs 11.4% (p=0.54). In the LMX1B variant associated MD, it was 1.2% vs 1.9% (p=0.48) in GC genotype. Figure 3 shows the relationship between tilt and the spherical equivalent refractive errors.

Table 4:

Univariate Associations of Subject Characteristics with Stereoscopic Identified Tilt (SIT)

Subject Characteristics	No Tilt (N=1805)	Tilt (N=330)	P value*
Age (years)			< 0.001
Mean (SD)	70.36 (10.93)	65.15 (11.08)
Gender			0.02
Female	1082 (59.9%)	222 (67.3%)
Male	723 (40.1%)	108 (32.7%)
Basel Metabolic Index			0.045
Mean (SD)	29.84 (6.54)	31.16(6.63)
Diabetes			0.41
No	1008 (60.8%)	185 (63.6%)
Yes	651 (39.2%)	106 (36.4%)
Unknown	146	39
Hypertension			0.15
No	338 (20.4%)	72 (24.7%)
Yes	1321 (79.6%)	219 (75.3%)
Unknown	146	39
Family History of Glaucoma			0.02
No	709 (42.7%)	100 (34.4%)
Yes	950 (57.3%)	191 (65.6%)
Unknown	146	39
Previous Glaucoma Surgery			0.005
No	222 (76.3%)	1111 (67%)
Yes	69 (23.7%)	548 (33%)
Unknown	39	146
LMX1B Genotype			0.48
GC	34 (1.9%)	4 (1.2%)
GG	1771 (98.1%)	326 (98.8%)
TRIM66 Genotype			0.54
TC	205 (11.4%)	42 (12.7%)
TT	1598 (88.6%)	288 (87.3%)

^*GEE logistic regression

Table 5:

Univariate Associations of Ocular Characteristics with Stereoscopic Identified Tilt

Ocular Characteristics	No Tilt (N=1805)	Tilt (N=330)	P value*
Disc Shape			< 0.001
Round	907 (50.5%)	108 (32.9%)
Oval	888 (49.5%)	220 (67.1%)
Unknown	10	2
Shape of Cup			< 0.001
Conical	626 (34.8%)	158 (50.0%)
Cylindrical	933 (51.9%)	113 (35.8%)
Bean pot shaped	238 (13.2%)	45 (14.2%)
Unknown	8	14
Sloping Margin Adjacent to Outer Disc Rim			< 0.001
No	1412 (78.6%)	185 (56.2%)
Yes	385 (21.4%)	144 (43.8%)
Unknown	8	1
Baring of Lamina Cribrosa			0.61
No	555 (31.0%)	107 (32.5%)
Yes	1234 (69.0%)	222 (67.5%)
Unknown	16	1
Cilio-Retinal Vessels			0.36
No	1415 (78.7%)	252 (76.4%)
Yes	383 (21.3%)	78 (23.6%)
Unknown	7	0
Gray Crescent			< 0.001
No	1684 (93.3%)	287 (87.0%)
Yes	120 (6.7%)	43 (13.0%)
Unknown	1	0
Beta PPA			0.45
No	606 (33.8%)	119 (36.1%)
Yes	1189 (66.2%)	211 (63.9%)
Unknown	10	0
Cup Disc Ratio			< 0.001
Mean (SD)	0.71 (0.17)	0.67 (0.16)
Unknown	12	1
Pallor of the Disc			0.01
No	329 (99.7%)	1736 (96.2.%)
Yes	68(3.8%)	1 (0.3%)
Unknown	0	1
Clinical Eye Characteristics
Refractive Error (Diopters)			0.03
Mean (SD)	−0.10 (1.82)	−0.50 (2.22)
Unknown	686	135
Myopic			0.31
No	547 (48.9%)	87 (44.6%)
Yes	572 (51.1%)	108 (55.4%)
Unknown	686	135
Intraocular Pressure (mmHg)			0.14
Mean (SD)	16.48 (5.16)	16.06 (4.07)
Unknown	1	0
Corrected Visual Acuity (Log MAR)			0.02
Median (Q1, Q3)	0.10 (0.00, 0.30)	0.09 (0.00, 0.24)
Unknown	94	24

^*GEE logistic regression

Figure 3:

Scatter plot showing the unadjusted relationship between ovality index (TOI) and refractive error

The results of the multivariable analyses are given in Table 6. Subjects with SIT were younger (aOR 0.95, 95% CI 0.93–0.96, p<0.001). Every decade increase in age decreased the odds of having a tilted disc by 50% (p<0.001). Eyes with SIT were more likely to have oval shaped discs compared to round discs (aOR 1.82, 95% CI 1.32–2.52, p<0.001), more likely to have a sloping region adjacent to the outer disc margin instead of being flat (aOR 3.26, 95% CI 2.32–4.59, p<0.001) and less likely to have cylinder shaped cups compared to conical shaped cups (aOR 0.59, 95% CI 0.41–0.85, p<0.001). Neither TOI nor SIT was associated with myopia.

Table 6:

Multivariable analyses Stereoscopically Identified Tilt (SIT)

Patient Characteristics	Adjusted Odds Ratio	95% Confidence Limits	P value
Age	0.95	(0.93, 0.96)	<.001
Ocular Characteristics	Adjusted Odds Ratio	95% Confidence Limits	P value
Cylinder Shaped Cup	0.59	(0.41, 0.85)	0.004
Sloping Region Adjacent to Disc Margin	3.26	(2.32, 4.59)	<.001
Oval Shaped Disc	1.82	(1.32, 2.52)	<.001

The grade - re-grade percent of agreement for the shape of the cup was 87% with Kappa of 0.75 (95% CI: 0.52–0.98) and the grade – regrade percent agreement for SIT was 83% with Kappa of 0.62 (95% CI: 0.31–0.92). The results demonstrate substantial agreement in the grade / regrading of SIT and shape of cup.²⁹

Stereoscopically Identified Tilt and Ovality Index of ≥1.30

There were 1972 eyes from 1054 subjects that had non-missing values for both SIT and TOI (≥1.30). Agreement between SIT and TOI is shown in Table 7. In 1607 eyes (82%), both methods of identifying tilt agreed there was no tilt; in 268 eyes (14%), SIT identified tilt but TOI did not; in 62 eyes (3%) TOI identified tilt but SIT did not; and in 35 eyes (2%) both methods agreed to the presence of tilt. Of the 1316 total subjects who had their eyes evaluated for TOI and SIT, 339 (25.8%) had tilt by one or both methods of classification. Of the 1270 subjects with data on both eyes, 255 (20.1%) were unilateral and 77 (6.1%) were bilateral.

Table 7.

Agreement (Per Eye) of Tilt Ovality Index (TOI >1.3) and Stereoscopic Identified Tilt (SIT)

Stereoscopic Identified Tilt	Tilt Ovality Index
	Absent	Present	Total
Absent	81.49% (1607)	3.14% (62)	84.63% (1669)
Present	13.59% (268)	1.77% (35)	15.37% (303)
Total	95.08% (1875)	4.92% (97)	100.00% (1972)

Note: 1972 eyes from 1054 subjects have both TOI and SIT measurements

Discussion:

The prevalence of tilt measured by tilt ovality index of ≥1.30 (TOI) was 8.3% (95% C I 6.9%, 10.0%). The prevalence was higher at 25.2% (95% C I 22.6%, 28.0%) when tilt was identified by stereoscopy (SIT). When compared with several studies that have reported on the laterality of tilted optic discs, POAAGG subjects with glaucoma had a relatively high proportion of unilateral tilt.⁵ Overall, eyes with optic disc tilt were associated with several optic cup and disc parameters that could be indicators of less severe glaucoma.

Most of the investigations on tilted discs have been in populations where myopia is highly prevalent. It is well-established that populations with tilted discs have a higher prevalence of myopia and myopic open-angle glaucoma, ranging from 57% to 81%.^30–34 POAAGG glaucoma subjects with tilt did not show an association with myopic refractive error. Enrollment in the POAAGG study excluded glaucoma subjects with a concomitant myopia of > −8.00 D and therefore we could not fully ascertain whether higher myopia was associated with tilt in this population. African Americans have a higher prevalence of POAG and a lower prevalence of refractive errors, suggesting that the higher risk of developing POAG in this population is independent of refractive error and if an association does exist between refractive error and risk of glaucoma, it is likely weak.^35–37 Although TOI was not associated with myopia there was an association with SIT in the univariate analysis that was not sustained after multivariable analysis. This finding is similar to a study of healthy young individuals where SIT was more closely related with myopia than TOI.³⁸ However, the lack of association with myopia in our study may be due, in part, to the exclusion of subjects with refractive error higher than −8 diopters.

Around 22% of glaucoma subjects in the POAAGG cohort had bilateral tilted disc by one or both of the classification methods. The bilateral frequency reported in this study is much lower than in other population-based studies where the frequency of bilateral discs ranged from 37.5% to 80%.^5,39 Different definitions used for tilted discs could explain some of this variation in frequency.

Our study showed that eyes with TOI had a cone shaped configuration of the cup, while those without tilted discs had a greater frequency of cylindrical shaped cups. It is possible that the gradually sloping contour of the conical cup implies a less severe form of cupping when compared to cylindrical shaped cups which have an abruptly descending contour. SIT discs were also found to be strongly associated with sloping of the region adjacent to the outer border of the disc rim instead of maintaining a horizontal plane. In studying the optic disc margins in myopic eyes with glaucoma using spectral-domain OCT, the border tissue configuration has been described as externally oblique in the temporal region and internally oblique in the nasal region of the optic disc between the anterior scleral opening and the Bruch’s membrane opening. At 12 and 6 o’clock, in the transition region between the internally and externally oblique, a non-oblique configuration was commonly observed.^40,41 Our study did not record the clockwise distribution of the sloping region, but these OCT features observed at the margin of the disc may be related to the sloping at the disc margin recorded in our study. However, the significance of this finding is not clear. Although the shape of the cup and the sloping of the peripapillary atrophy is subjective and dependent on the level of stereo in the digital pair of images evaluated, these two features warrant further investigation with recently evolving and more accurate imaging modalities.

Beta PPA is commonly seen in myopic or myopic glaucoma eyes with tilted discs. The frequency of β-PPA was similar in both eyes with and without tilt in the POAAGG cohort, suggesting that myopia may not play a significant role. Tilted discs have been reported to be associated with a lower rate of diabetic retinopathy.⁴² In the POAAGG study a substantial number of subjects had diabetes but having the disease was not associated with tilted discs. The association of two SNPs in the LMX1B and TRIM66 genes, with the prevalence of tilted disc was also investigated in this study. Although we did not find an association with the variants examined, we aim to further explore the association of tilted discs with a polygenic risk score and meta-analysis of the POAAGG data.

We investigated tilt using both quantitative (TOI) and qualitative (SIT) measures in this study. Our results show that although TOI has a strong correlation with SIT, it does not capture all the SIT, specifically in eyes with less oval or round shaped discs (Figure 4). Three decades ago, a quantitative measurement of tilt, termed papillary index, was introduced, which divides the larger diameter of the optic disc by the shorter one.⁴³ Index of tilt and TOI were terms that were later used to describe the same quantitative measurement of tilt. TOI has been used in several papers as a surrogate for identifying tilt. However, evaluation of optic disc stereophotograph image pairs using a stereo-viewer allows a three-dimensional observation, albeit subjective, and is still considered a reliable measure of identifying tilted optic disc. The Tanjong Pagar study found moderate agreement between a quantitative measure of index of tilt (ratio of minimum to maximum optic disc diameter less than 0.75) and a qualitative assessment (one edge of the optic disc appeared more anteriorly located than its edge 180° away).⁴ This study reported that tilt identified by stereoscopic photographs was the gold standard; the sensitivity of the ovality index was 56.7% and the specificity was 94.0%.⁴ In our study the sensitivity for TOI when SIT was used as a gold standard was much lower at 10%. Another study which investigated optic disc vertical tilt angle in 112 subjects using spectral-domain optical coherence tomography found that the ovality index was only moderately correlated with the tilt angle, concluding that ovality index was not a good proxy measure for the vertical disc tilt.⁴⁴ The TOI rates in our subjects (8.3%) were slightly higher than those found in the adult population of several countries (0.09% to 3.5%).

Figure 4:

Box plot showing the unadjusted relationship between Stereoscopically Identified Tilt (SIT) and refractive error

A noteworthy result from this report is the substantial presence of tilted discs among African Americans with POAG. It is well known that individuals of African ancestry are four to five times more likely to have POAG than individuals of European ancestry.⁴⁵ It has been shown that when compared to glaucoma specialists, ophthalmology trainees and comprehensive ophthalmologists overestimated the likelihood of glaucoma in large discs when incorrectly assessing disc tilt, both attributes common among African Americans.¹⁶ If tilted discs are not taken into account in African Americans with POAG, misclassifications and erroneous treatments are possible in this vulnerable group of glaucoma subjects.

There are several limitations to our study. This was a cross-sectional study involving subjects with a wide variability in glaucoma characteristics, severity, and treatment history. Therefore, associations have been confined to demographic features and the Reading Center evaluation of the optic disc features that have been identified by both qualitative and quantitative methods described earlier. Medical history taken at enrollment could have been subject to misclassification due to recall bias. We also could not use the control group for comparison because imaging in this group was restricted to only those who had a medical reason for imaging the retina.

In summary, we show for the first time that in glaucoma patients with African Americans ancestry, there are a substantial number of tilted optic discs. Glaucoma subjects with tilted optic discs were more likely to be female and younger. Myopia did not appear to be independently associated with tilted discs, as has been commonly reported in literature. Eyes with tilted discs when compared to eyes with non-tilted discs had predominantly oval shaped discs, sloping regions adjacent to the outer disc rim, and cone shaped cups. Exploring the role of tilted discs in the pathology of glaucoma, its role and prevalence in subjects with African ancestry will help further in understanding the pathophysiology of this disease in this vulnerable population. We believe that primary open angle glaucoma represents many different diseases defined by differing phenotypes and different underlying genotypes. This study carefully addresses the phenotype of the optic nerve through precisely defined characteristics found in this understudied African Ancestry population. As we continue to discover variants of interest, we will correlate these with differing phenotypes described here.

Figure 5:

Examples of fundus color images of optic discs with and without tilt

A1: Left eye optic disc with no tilt. This eye had an Ovality Index (OI) of >1.30 (1.13) and did not have a Stereoscopically Identified Tilt (SIT). The Cup Disc Ratio (CDR) was 0.85. A2: Left eye optic disc with tilt identified by both OI (1.38) and SIT with a CDR of 0.70. A3: Right eye optic disc with tilt identified by SIT but not by OI (1.09) with a CDR of 0.70. A4: Left eye optic disc with tilt identified by OI (1.42) but not by SIT with a CDR of 0.80.

Optic disc tilt is common among glaucoma patients with African American ancestry and occurs more frequently in younger women. They are associated with several other ocular features but not with myopia.