Bilateral, partially attached posterior vitreous detachment in patient 1. A, Scanned optical coherence tomographic images from the right eye (left panel) and left eye (right panel) showing superior and inferior vitreous attachments (arrows). B, Retinal nerve fiber layer (RNFL) thickness measurements showing abruptly thickened nerve fiber layers corresponding to the vitreous detachments. S indicates superior; T, temporal; N, nasal; I, inferior; Imax, inferior quadrant maximum; Smax, superior quadrant maximum; Tavg, temporal quadrant average; Navg, nasal quadrant average; Savg, superior quadrant average; Iavg, inferior quadrant average; solid line, right eye; and dashed line, left eye.
Unilateral, partially attached posterior vitreous detachment in patient 2 (right eye, with superior and inferior vitreous attachments). A, Scanned optical coherence tomographic images of the right eye (left panel) and left eye (right panel). Arrows indicate the partially attached posterior vitreous detachment. B, Retinal nerve fiber layer (RNFL) thickness measurements showing thickening of the nerve fiber layer corresponding to the vitreous attachment in the right eye as well as the difference in all parameters compared with the left eye. S indicates superior; T, temporal; N, nasal; I, inferior; Imax, inferior quadrant maximum; Smax, superior quadrant maximum; Tavg, temporal quadrant average; Navg, nasal quadrant average; Savg, superior quadrant average; Iavg, inferior quadrant average; solid line, right eye; and dashed line, left eye.
Batta P, Engel HM, Shrivastava A, Freeman K, Mian U. Effect of Partial Posterior Vitreous Detachment on Retinal Nerve Fiber Layer Thickness as Measured by Optical Coherence Tomography. Arch Ophthalmol. 2010;128(6):692-697. doi:10.1001/archophthalmol.2010.99
To evaluate the effect of partially attached posterior vitreous detachments (pPVDs) at the optic disc on retinal nerve fiber layer (RNFL) thickness as measured by optical coherence tomography.
A retrospective study was conducted using stored Stratus optical coherence tomography III scans of patients with suspected glaucoma from January 2003 to September 2006 at the Montefiore Medical Center, Bronx, New York. All scans were evaluated for vitreous attachments at the disc and were divided into control (without pPVD) and pPVD groups. The RNFL thickness was compared using the fast RNFL protocol. Patients were defined as glaucoma suspects based on clinical findings of either glaucomatous-appearing optic discs or elevated intraocular pressure. All study patients had normal Humphrey visual fields.
A total of 110 eyes from 110 patients were included; 59 were in the pPVD group and 51 were controls. Partial PVD was found in 40% of the glaucoma suspects. The mean RNFL thickness of eyes with pPVD was significantly broader than that of controls (101.6 μm vs 95.6 μm, respectively; P < .001). The average RNFL thickness of each quadrant was greater in the pPVD group than in the control group, with statistically significant differences in superior and inferior quadrants (P < .001 and P = .001, respectively).
More than one-third of this population of glaucoma suspects had a pPVD, indicating that this is a common phenomenon. The results suggest that RNFL thicknesses are greater in patients with pPVD than in controls. This may indicate a limitation of using RNFL thickness as a criterion for evaluating glaucomatous damage in patients with pPVD.
The diagnosis of glaucoma is generally established by detecting characteristic visual field defects in association with visible alterations in the surface contour of the optic nerve and elevation of the intraocular pressure. A substantial minority of retinal ganglion cells may be lost before a change in appearance of the optic disc can be discerned.1 High-resolution retinal images obtained by optical coherence tomography (OCT) can be used to quantitatively assess retinal nerve fiber layer (RNFL) thickness.2- 4 Because thinning of the RNFL precedes ophthalmoscopically visible alteration in the cup-disc ratio, OCT has the potential to support a diagnosis of glaucoma at an early stage.5- 8
High-resolution OCT provides reproducible visualization of the vitreoretinal interface. Partially attached posterior vitreous detachment (pPVD) appears as a distinct curvilinear signal extending from the retinal surface into the vitreous cavity.4 Our preliminary observations based on OCT studies in eyes with pPVD suggest that the RNFL is notably thicker at points of vitreous attachment, presumably due to traction exerted by the vitreous. Greater depth of the RNFL, associated with pPVD, may undermine the utility of OCT in diagnosing early glaucomatous change in patients with pPVDs. We performed a retrospective cross-sectional analysis to investigate the effect of pPVDs on RNFL measurements acquired by OCT.
This retrospective, observational study was approved by the Montefiore Medical Center Institutional Review Board and was compliant with the tenets of the Declaration of Helsinki (1964). The OCT scans were selected consecutively from an alphabetized list of stored patient images (from January 2003 to September 2006) at the Henkind Eye Institute, Montefiore Medical Center, Bronx, New York. Scans were eligible for inclusion in the study if the patient had undergone 2 or more fast RNFL scans for both eyes (on different days) and at least 1 of these scans was performed in the randomly chosen 6-month period of January 1, 2006, to July 1, 2006.
Scans were included in the pPVD group if at least 1 area of vitreous attachment was clearly visible at the peripapillary vitreoretinal interface. In the pPVD group, a pPVD was noted in either eye or both eyes. For the few patients with pPVD in both eyes, 1 eye was randomly selected for the study. The control group consisted of eyes with no discernible pPVD on any OCT image.
All scans included in the study were of patients having a diagnosis of suspected glaucoma based on either a suspicious optic nerve in the study eye (group 1) or ocular hypertension (group 2). Group 1 patients included those with a cup-disc ratio greater than 0.5, interocular asymmetry of the cup-disc ratio greater than 0.2, or thinning or notching of the neuroretinal rim. Group 2 patients had 2 consecutive intraocular pressure measurements greater than 25 mm Hg, measured via applanation tonometry (corrected for central corneal thickness), and also showed normal and symmetrical optic nerves.
All study patients had undergone a Humphrey visual field test within the specified 6-month period and were found to have either a normal visual field or only nonspecific and nonprogressive visual field defects in the study eye. Patients were excluded if the Humphrey visual field test results were unreliable (defined as >33% fixation loss, a false-positive response >33%, and a false-negative response >33%). Scans with an average RNFL thickness less than 80 μm were also excluded, even if visual fields were intact, to preclude inadvertent inclusion of cases with early glaucomatous damage. Patients with a history of laser treatment for retinal disease or vitrectomy were also ineligible. The control and pPVD groups were not age matched. We recorded age, ethnicity, and sex.
All OCT images were acquired with the Stratus OCT III (Carl Zeiss Meditec, Inc, Dublin, California) by a trained technician. Images were obtained using the standard technique for the fast RNFL protocol, which analyzes RNFL thickness based on 3 circular scans of the peripapillary RNFL. (The aiming circle used by this protocol was set at a default of 3.4 mm in diameter and was manually centered on the optic disc.) The RNFL thickness, calculated by averaging the 3 scans to determine means in the 4 quadrants as well as average thickness, was generated automatically by the Stratus OCT software.
Scans included in the study fulfilled the following criteria: the fundus image was clear with a visible scan circle centered on the optic disc, color distribution was even and dense throughout the scan, and the RNFL borders appeared to be accurately detected by the OCT algorithm.
Descriptive statistics are presented as means and standard deviations for continuous, normally distributed variables and as relative frequencies for rank order or categorical variables. Differences between pPVD and control groups with regard to continuous variables were tested for significance using t tests for independent samples, and those of pPVD/control groups and subgroups group 1 (suspicious disc)/group 2 (ocular hypertension) were tested for significance with analysis of variance. Associations with categorical variables were tested using χ2 or Fisher exact tests. Associations between continuous variables and age were derived using Pearson correlation coefficients, and multiple linear regression analysis was used to control for differences in groups due to age. One-way and 2-way analyses of variance were performed to identify differences among subsets. All analyses were conducted with SAS version 9.1 statistical software (SAS Institute, Inc, Cary, North Carolina) at α = .05 using 2-tailed tests.
A total of 1368 patients had OCT fast RNFL scans at the Montefiore Medical Center in the 6-month period from January 1, 2006, to July 1, 2006. Of these, 541 were found to have pPVD present on at least 1 scan for either eye, indicating a prevalence rate of 40% in this population. Generally, the vitreous was separated nasally and temporally (Figure 1 and Figure 2). In eyes harboring pPVD, attachments were noted superiorly in 66% and inferiorly in 75%.
A total of 110 eyes of 110 patients with suspected glaucoma were included in our analysis. Fifty-nine of these were in the pPVD group and 51 were controls (showing no pPVD). This latter group included patients with intact vitreous, complete vitreous detachment, or an indeterminate vitreoretinal interface. In the subgroup analysis, group 1 (suspicious discs) included 33 eyes with pPVD and 29 control eyes. Group 2 (ocular hypertension) comprised 26 eyes with pPVD and 22 control eyes.
The mean RNFL thickness of eyes with pPVD was significantly increased compared with that of controls (101.6 μm vs 95.6 μm, respectively; P < .001). The average RNFL thickness of each quadrant was greater in the pPVD group than in the control group, with statistically significant differences in the superior and inferior quadrants (P < .001 and P = .001, respectively). Maximum RNFL thicknesses for superior and inferior quadrants were significantly higher in the pPVD group than in controls (both P < .001). Maximum-minimum RNFL thickness was also higher in the pPVD group than in controls (P < .001). No significant differences among the pPVD group and controls were recognized in the temporal or nasal quadrants (Table 1).
Individuals in the control groups were significantly older (mean difference, 11.2 years; 95% confidence interval, 7.4-15.0; P < .001), but the age differences were insufficient to discount the variance in RNFL thickness when analyzed using linear regression models controlling for age (P > .10). Analyses with ethnicity and sex as covariates also failed to reduce the significance of the difference in RNFL thickness between eyes with pPVD and controls (P > .10) (Table 2).
Multiple linear regression evaluating both the influence of pPVD and the clinical status (suspicious discs vs ocular hypertension), controlling for age, confirmed the thickness difference between control eyes and pPVD eyes for all measurements (P ≤ .002) except those made in the nasal and temporal quadrants (Table 3). There was little or no evidence of a differential effect of pPVD on eyes with suspicious discs vs those with ocular hypertension (Table 3, tests of interaction). No significant differences were found in RNFL thickness in the temporal or nasal quadrants in either the collective analysis or the subgroup analysis.
In our review of more than 1300 optic nerve OCT scans performed to assess the RNFL, pPVD was recognized in 40% of eyes. In most cases of pPVD, the vitreous detached from the nasal and temporal quadrants but remained adherent in the superior and inferior quadrants. Eyes of glaucoma suspects harboring pPVD showed a statistically significantly thicker RNFL with respect to control eyes, and the RNFL thickening was most pronounced in the superior and inferior quadrants.
Our study was a retrospective analysis using stored OCT scans. We did not study normal eyes. The number of variables in the groups was reduced by limiting the prescreening diagnosis to suspected glaucoma with no defects on automated visual field testing. The sample sizes had adequate power to analyze variance between the control and pPVD groups. Although our groups were not matched for age, ethnicity, and sex, these factors were evaluated and were not responsible for the variance in study and control scans. We did not record axial length, which may play a role in RNFL thickness. High myopia is associated with a thinner RNFL.9 These data were not available in our retrospective study. It is unlikely that the disparity in RNFL thickness measurements could be attributed to other variables.
The average age was considerably younger (by 11.2 years) in those with pPVDs compared with the control group. This may be best explained by considering the natural history of PVD. As an individual ages, complete PVD is more likely to develop. Our controls included individuals with either no PVD or complete PVD in the peripapillary region. Normative data suggest that there is an average thinning in RNFL thickness of 0.2 to 0.3 μm per year. However, the difference in the RNFL thickness between eyes with pPVD and controls was considerably higher than could be explained by the age difference.7
Greater variability was seen in the subgroup analysis than in the overall statistical review. A 2-factor analysis of variance demonstrated that in general, the subgroups behaved similarly to the collective measurements. The control groups for group 1 (glaucoma suspects with enlarged or asymmetric cup-disc ratios) and group 2 (glaucoma suspects with normal nerves and ocular hypertension) were generally similar except for the superior quadrant average (Table 3). This comparability was maintained when the groups were controlled for age (except for the superior quadrant average). The explanation for the variation in the subgroups and especially in the superior quadrant is not obvious.
That pPVD is associated with a thicker RNFL is not unexpected. Examples of alterations of the vitreoretinal interface causing changes in retinal and optic nerve structure that may be registered by OCT have been noted in eyes undergoing vitrectomy for diabetic macular edema, vitreomacular traction syndrome, and vitreopapillary traction.10- 13 We found that pPVD is associated with increased RNFL thickness, especially in superior and inferior quadrants, and is most likely explained by traction from the adherent vitreous on these quadrants. (Two-thirds of eyes with pPVD showed attachments in the superior or inferior quadrants of the nerve.) Given our finding that more than one-third of patients undergoing optic nerve OCT scanning have pPVD, this effect may have important implications for OCT use in diagnosing glaucoma.
Correspondence: Umar Mian, MD, Department of Ophthalmology and Visual Sciences, Montefiore Medical Center, 111 E 210 St, Bronx, NY 10467 (firstname.lastname@example.org).
Submitted for Publication: October 8, 2007; final revision received September 1, 2009; accepted October 5, 2009.
Financial Disclosure: None reported.