Optical coherence tomography (OCT) has been used to monitor disease progress and response to therapy in eyes within diabetic macular edema (DME).1,2 OCT also can be used to quantitatively analyze tissue optical properties, as the OCT signal depends on the total attenuation and backscattering coefficients.3,4 Treatment with anti-vascular endothelial growth factor (VEGF) agents such as bevacizumab (Avastin; Genentech, South San Francisco, CA), ranibizumab (Lucentis; Genentech, South San Francisco, CA), and aflibercept (Eylea; Regeneron, Tarrytown, NY) has been shown to be effective for DME.5–10 Anti-VEGF treatment for DME improves visual acuity (VA) and decreases central retinal thickness, but the effect is often short-lived and multiple injections are required.11 In addition, despite a decrease in retinal thickness, VA may not improve in some subjects.6,11
Bressler et al.12 from the Diabetic Retinopathy Clinical Research network trials demonstrated only a modest correlation between VA and retinal thickness in eyes with DME. Alasil et al.13 observed that the relationship could be significantly strengthened if the thickness of the outer retinal bands (photoreceptor inner and outer segments) was considered. In addition, Castro Lima et al.14 reported improved VA after anti-VEGF therapy in patients with normal outer retinal layer morphology at baseline. In contrast, patients with a disturbed outer retinal layer showed a reduction in retinal thickness but no significant visual improvement.
The challenge of relying on the integrity of the outer retinal bands to predict visual outcomes in patients with DME is that marked retinal thickening or poor image quality due to media opacity may impair visualization of the outer retinal bands and confound accurate assessments. In addition, assessing damage to the inner retina due to retinal ischemia may also contribute to visual compromise in these patients. In fact, Sun et al.15 identified disorganization of the retinal inner layers (DRIL) to be an important predictor. It is not always easy to determine that DRIL is present, however, when the retina is edematous, particularly with extensive cystoid edema. If one could estimate the actual tissue loss (inner and/or outer) present in the retina in the setting of overall thickening due to fluid, one may be able to better predict visual or anatomic outcomes following successful therapy.16
With this goal in mind, we proposed and studied a new metric, dry retinal volume (DRV), as an approach for calculating the expected volume (or thickness) of the retina if the fluid could be resolved. We compared this predicted DRV with the actual retinal volume as well as with visual acuity following anatomically successful anti-VEGF therapy for DME.
Patients and Methods
The medical records of all patients with DME who presented to the Medical Retina Clinics of the Doheny Eye Institute (single physician, SRS) between January 2013 and December 2014 were reviewed in this retrospective analysis. The study was approved by the institutional review boards of the University of California — Los Angeles and the research adhered to the tenets set forth in the Declaration of Helsinki. Eligibility criteria included: (1) foveal center-involved macular edema from diabetic retinopathy only (with no prior laser treatment), (2) baseline and follow-up volume OCT scans obtained with the same spectral-domain OCT (SD-OCT) (Cirrus HD-OCT; Carl Zeiss Meditec, Dublin, CA), (3) high-quality OCT images (signal strength 7 or higher), (4) receipt of intravitreal ranibizumab therapy in accordance with a modified DRCR Protocol I strategy (the modification being that treatment was suspended before the fourth injection if the foveal central subfield was deemed to be dry or the VA was 20/20), and (5) a minimum of 3 month's follow-up, including at least one follow-up visit in which the treating physician deemed the fovea to be maximally dry, prompting cessation of therapy. It should be noted that in some cases, maximally dry did not mean the complete absence of foveal cystoid edema, but that the physician determined no additional fluid reduction was observed despite a minimum of six consecutive injections. Because of these strict criteria, only 28 eyes of 26 subjects ultimately were deemed to be eligible and included in this analysis. Variability of follow-up was less of an issue in this study, as the primary goal was to compare the predicted dry retinal volume with an actual visit when the retina was dry or had achieved maximum dryness in the estimation of the treating physician (SRS). A retinal thickness cut-off was not used as the key indicator of dryness, since a damaged retina with tissue loss may have a normal thickness despite persistent cystoid edema.
Clinical and OCT data from baseline (just before beginning anti-VEGF therapy) and follow-up were collected for assessment. The follow-up visit chosen for analysis was the first visit in which the investigator determined that the maximal drying had been achieved and therapy was suspended. Raw SD-OCT volume scan (512 × 128 over a 6 mm × 6 mm square, centered on the fovea) data were exported from the Cirrus OCT for subsequent analysis. Non-OCT variables collected included gender, age, best-corrected VA (BCVA) using logMAR chart, and ophthalmoscopic and biomicroscopic findings. SD-OCT volume scan data were also collected from one eye each of 10 healthy normal volunteers aged between 20 years and 45 years with emmetropic, no ocular abnormalities, no systemic disease, and a normal exam by biomicroscopy and OCT.
Exported OCT data (including the manufacturer determined retinal boundaries) from both eyes with DME and controls were imported into 3D-DOCTOR grading software (Able Software Corp., Lexington, MA).17,18 A certified senior OCT reading center grader (MGN) inspected every B-scan of all volume scans and manually corrected any segmentation errors. The reproducibility of manual correction of segmentation errors by our group and this grader has been demonstrated in prior publications.19 In addition, the grader manually added one additional boundary corresponding to the outer border of the nerve fiber layer (NFL), allowing the NFL to be isolated from the volume scan for further analysis. The vitreous space was defined as all pixels internal to the internal limiting membrane. The brightness of the vitreous and NFL pixels was used to normalize signal strength as previously described.20
The vitreous and NFL21 were chosen for normalization because they were thought to be relatively unaffected by the disease process in this cohort. No subjects had evidence of vitreous hemorrhage or vitreous opacity on examination or by OCT. The vitreous appeared to be a reasonable reference standard for the darkest regions on the OCT. The visible pixels in the vitreous in most eyes were largely speckle noise. The NFL was used for normalization as it is the only retinal layer that tends not to manifest cystoid change in eyes with DME1 and could serve as a reasonable reference for the brightest structure on the OCT scan. To normalize the brightness of a given structure, one can subtract the mean vitreous brightness (ie, the “noise floor”) and divide by the mean NFL brightness, yielding a ratio of the normalized structure brightness.22 This operation is described by the equation: “Normalized structure intensity = [Retinal intensity − Vitreous intensity] / NFL intensity.”
Once the brightness of structures of interest is normalized, comparisons can be made between eyes (with the assumption that the brightness of the vitreous and NFL should be similar among eyes).
Dry Retinal Volume Calculation
To estimate the expected retinal volume if the edema was resolved (DRV), we assumed that fluid in the retina was dark or hyporeflective on OCT. Accumulation of fluid thickened the retina but also reduced its mean brightness or intensity. In addition, in our model, we hypothesized that in an edematous retina, although the mean brightness was lower than in a normal retina, assuming no tissue loss, the mean intensity of all pixels in the retina should be similar in edematous and normal retina. If the mean intensity of the edematous retina was less than the normal retina, we presumed this was a sign of tissue loss. The amount of tissue loss or thinning relative to normal could then be determined as a simple ratio, allowing for computation of DRV. This operation is described by the equation: DRV = [mean retinal reflectivity of case / mean retinal reflectivity of controls] * total retinal volumne (TRV) of case (at baseline). The dry foveal volume could similarly be calculated by restricting the sample to the foveal central subfield.
The main outcome measures for analysis were age, logMAR VA, retinal thickness, TRV, foveal center subfield (FCS) volume, and DRV. The correlations between DRV measurements and VA were assessed using bivariate Pearson correlations. The generalized estimating equation method was used to adjust for correlations between eyes. Paired t-test was used to compare pre- and posttherapy retinal thickness and volume. The difference between predicted DRV and actual retinal volume after anti-VEGF therapy was assessed by t-test. Relation between predicted DRV and actual retinal volume was analyzed using bivariate correlations. A P value of .05 or less was considered clinically significant. All statistical analysis was performed using SPSS 18 Statistical Software (SPSS, Chicago, IL).
DRV was calculated in 28 eyes of 26 subjects (mean age: 69 years ± 9.8 years; range: 45 years to 79 years; 28% female). The average post-anti-VEGF therapy follow-up period until a dry retina was achieved was 7.14 months ± 6.79 months (range: 3 months to 24 months). Mean number of anti-VEGF injections received was four (± standard deviation: two; range: one to seven injections). Mean logMAR BCVA was significantly (P = 0.01) improved from 0.56 ± 0.36 to 0.44 ± 0.32 (Snellen equivalent 20/60 to 20/50) at baseline and posttherapy.
Retinal thickness was significantly (P < .001) decreased from 367.98 μm ± 75.82 μm to 303.2 μm ± 50.49 μm, and retinal volume was significantly decreased (P < .001) from 13.25 mm3 ± 2.73 mm3 to 10.92 mm3 ± 1.42 mm3 from baseline to post-anti-VEGF therapy. Calculated DRV (10.79 mm3 ± 1.42 mm3) and retinal volume obtained from the OCT instrument after therapy were statistically similar (P = .73), and the correlation coefficient (r) was 0.48, P = .009 (Figure 2A), with a mean absolute difference of 1.18 mm3 ± 0.94 mm3 and a maximum difference of 3 mm3 (Figure 1). The DRV for the FCS of 0.21 mm3 ± 0.05 mm3 was significantly different from the posttreatment actual FCS volume of 0.25 mm3 ± 0.09 mm3 (P = .01) and the correlation coefficient r = 0.45, P = .02 (Figure 2B). Eighteen of 28 eyes (64%) were completely dry with no residual cysts at the post-therapy visit. Despite the correlation between predicted and actual DRV, these measurements did differ by more than 1 mm3 in 40% of our study eyes; however, when excluding eyes with residual cysts, only 18% of eyes showed this level of discrepancy.
Example study case (A) of diabetic macular edema in the left eye with wet retinal volume of 17.53 mm3. (B) Post-therapeutic reduced macular edema with original retinal volume of 10.19 mm3 and adjusted dry retinal volume of 10.45 mm3.
Scatter diagram showing relation between posttherapy total retinal volume and predicted dry retinal volume (A). Post-therapy foveal volume and predicted dry foveal volume (B).
After anti-VEGF therapy, the logMAR VA was improved by 1 line or more in 19 (68%) eyes, stable in five (18%) eyes, and decreased by 1 line or more in four (14%) eyes. Post-therapeutic improvement in logMAR VA was significantly correlated with TRV change (r = 0.50, P = .007) and FCS volume change (r = 0.53, P = .004). The logMAR BCVA was weakly correlated with DRV (r −0.19, P = .32) and with dry FCS volume (r = −0.15, P = .44), but the correlations were not statistically significant.
Several studies have reported that baseline parameters such as central retinal thickness,5,12 ischemia,5 and cyst size8 can be used to predict visual outcome. Along with these parameters, we suggest a novel OCT quantitative parameter, DRV, calculated by effectively mathematically subtracting the volume of cystic spaces and macular edema from the OCT measured retinal volume.
Retinal reflectivity has been described as an important OCT parameter to predict disease progression in pathologies such as diabetic retinopathy;3 however, OCT-based reflectivity has not been explored completely for its application in clinical practice. In this study, we used total retinal reflectivity to calculate DRV; the mean DRV was statistically similar to the actual post-therapeutic TRV. This measurement could be used to estimate the residual volume of tissue in macular edema.16
The DRV of the FCS was significantly lower than the actual post-therapy volume. This difference could be due to the presence of large cystoid spaces in the central macular region of the majority of the study cases, or it may be a result of damage to the outer retinal layers during the disease process, causing outer retinal layer atrophy. We propose further studies regarding this finding.
Despite the correlation between predicted and actual DRV, these measurements did differ by more than 1 mm3 in 40% of our study eyes. When we retrospectively reviewed these cases with the maximum discrepancy in prediction, two main confounding features were observed at baseline: marked lipid exudates or massive cystoid edema. Lipid exudates are a common finding in DME; they are associated with visual impairment23 and manifest as bright optical signals in OCT imaging. The presence of dense lipid exudates in the macular area can increase the total retinal intensity values and lead to an overestimation of DRV. In eyes with massive edema/retinal thickening, there is a loss of signal with tissue depth that can reduce the brightness of the outer retina and the overall total retinal intensity, leading to an underestimation of DRV. If cases with marked exudates or massive edema are excluded, the correlation between BCVA and DRV improves further to r = −0.30 (P = .29).
In this study, post-therapy improvement in BCVA was significantly and positively correlated with decrease in retinal thickness.6–9 BCVA was also moderately correlated with decreased retinal and FCS volume. But calculated DRV had no significant relationship with VA.
In spite of successful treatment effect on reducing retinal thickness (in macular edema), 14% of the cases in this cohort showed a decrease in VA of more than 1 line, potentially due to post-therapeutic outer retinal layer atrophy.8,16,23
Our study has several limitations that should be considered when assessing the significance of our findings. First, our cohort was relatively small, and it was even smaller when considering eyes that became completely dry as opposed to just maximally dry. Second, the persistence of some cysts may have confounded the DRV prediction. Third, we were unable to adjust for the effects of lipid exudates or massive edema on the computation of the total retinal intensity. Future studies excluding lipid pixels could potentially improve the prediction. Moreover, use of newer swept-source OCT devices, which feature less sensitivity loss with depth, may ameliorate some of the problems posed by massively thick retinas.
In conclusion, with further optimization, dry retinal volume may be a useful predictor of the morphology or thickness of the retina after anti-VEGF therapy for cystoid macular edema associated with diabetic retinopathy. VA, however, was not significantly associated with dry retinal thickness, suggesting that other parameters, such as the integrity of the outer retinal bands and the organization of the inner retina, may still need to be considered.
- Somfai GM, Tátrai E, Ferencz M, Puliafito CA, Debuc DC. Retinal layer thickness changes in eyes with preserved visual acuity and diffuse diabetic macular edema on optical coherence tomography. Ophthalmic Surg Lasers Imaging. 2010;41(6):593–597. doi:10.3928/15428877-20100830-04 [CrossRef]
- Lang GE. Optical coherence tomography findings in diabetic retinopathy. Dev Ophthalmol. 2007;39:31–47. doi:10.1159/000098498 [CrossRef]
- Gao W, Tátrai E, Ölvedy V, et al. Investigation of changes in thickness and reflectivity from layered retinal structures of healthy and diabetic eyes with optical coherence tomography. J Biomed Sci Eng. 2011;4:657–665. doi:10.4236/jbise.2011.410082 [CrossRef]
- Schmitt JM, Knüttel A, Bonner RF. Long-term effect of intravitreal bevacizumab (avastin) in patients with chronic diffuse diabetic macular edema. Retina. 2008;28(8):1053–1060. doi:10.1097/IAE.0b013e318176de48 [CrossRef]
- Arevalo JF, Fromow-Guerra J, Quiroz-Mercado H, et al. Primary intravitreal bevacizumab (Avastin) for diabetic macular edema: Results from the Pan-American Collaborative Retina Study Group at 6-month follow-up. Ophthalmology. 2007;114(4):743–750. doi:10.1016/j.ophtha.2006.12.028 [CrossRef]
- Arevalo JF, Sanchez JG, Fromow-Guerra J, et al. Comparison of two doses of primary intravitreal bevacizumab (Avastin) for diffuse diabetic macular edema: Results from the Pan-American Collaborative Retina Study Group (PACORES) at 12-month follow-up. Graefes Arch Clin Exp Ophthalmol. 2009;247(6):735–743. doi:10.1007/s00417-008-1034-x [CrossRef]
- Reznicek L, Cserhati S, Seidensticker F, et al. Functional and morphological changes in diabetic macular edema over the course of anti-vascular endothelial growth factor treatment. Acta Ophthalmol. 2013;91(7):e529–536. doi:10.1111/aos.12153 [CrossRef]
- Do DV, Nguyen QD, Vitti R, et al. Intravitreal aflibercept injection in diabetic macular edema patients with and without prior anti-vascular endothelial growth factor treatment: outcomes from the Phase 3 program. Ophthalmology. 2016;123(4):850–857. doi:10.1016/j.ophtha.2015.11.008 [CrossRef]
- Wells JA, Glassman AR, Diabetic Retinopathy Clinical Research Network et al. Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema. N Engl J Med. 2015;372(13):1193–1203. doi:10.1056/NEJMoa1414264 [CrossRef]
- Chun DW, Heier JS, Topping TM, Duker JS, Bankert JM. A pilot study of multiple intravitreal injections of ranibizumab in patients with center-involving clinically significant diabetic macular edema. Ophthalmology. 2006;113(10):1706–1712. doi:10.1016/j.ophtha.2006.04.033 [CrossRef]
- Browning DJ, Glassman AR, Diabetic Retinopathy Clinical Research Network et al. Relationship between optical coherence tomography-measured central retinal thickness and visual acuity in diabetic macular edema. Ophthalmology. 2007;14(3):525–536.
- Alasil T, Keane PA, Updike JF, et al. Relationship between optical coherence tomography retinal parameters and visual acuity in diabetic macular edema. Ophthalmology. 2010;117(12):2379–2386. doi:10.1016/j.ophtha.2010.03.051 [CrossRef]
- Castro Lima V, Rodrigues EB, Nunes RP, Sallum JF, Farah ME, Meyer CH. Simultaneous confocal scanning laser ophthalmoscopy combined with high-resolution spectral-domain optical coherence tomography: A review. J Ophthalmol. 2011;2011:743670. doi:10.1155/2011/743670 [CrossRef]
- Sun JK, Lin MM, Lammer J, et al. Disorganization of the retinal inner layers as a predictor of visual acuity in eyes with center-involved diabetic macular edema. JAMA Ophthalmol. 2014;132(11):1309–1316. doi:10.1001/jamaophthalmol.2014.2350 [CrossRef]
- Pelosini L, Hull CC, Boyce JF, McHugh D, Stanford MR, Marshall Jl. Optical coherence tomography may be used to predict visual acuity in patients with macular edema. Invest Ophthalmol Vis Sci. 2011;52(5):2741–2748. doi:10.1167/iovs.09-4493 [CrossRef]
- Sadda SR, Joeres S, Wu Z, et al. Error correction and quantitative subanalysis of optical coherence tomography data using computer-assisted grading. Invest Ophthalmol Vis Sci. 2007;48(2):839–848. doi:10.1167/iovs.06-0554 [CrossRef]
- Nittala MG, Konduru R, Ruiz-Garcia H, Sadda SR. Effect of OCT volume scan density on thickness measurements in diabetic macular edema. Eye (Lond). 2011;25(10):1347–1355. doi:10.1038/eye.2011.173 [CrossRef]
- Joeres A, Tsong JW, Updike PG, et al. Reproducibility of quantitative optical coherence tomography subanalysis in neovascular age-related macular degeneration. Invest Ophthalmol Vis Sci. 2007;48(9):4300–4307. doi:10.1167/iovs.07-0179 [CrossRef]
- Hu Z, Nittala MG, Sadda SR. Comparison of retinal layer intensity profiles from different OCT devices. Ophthalmic Surg Lasers Imaging Retina. 2013;44(6 Suppl):S5–10. doi:10.3928/23258160-20131101-02 [CrossRef]
- Otani T, Kishi S. Tomographic findings of foveal hard exudates in diabetic macular edema. Am J Ophthalmol. 2001;131(1):50–54. doi:10.1016/S0002-9394(00)00661-9 [CrossRef]
- Velaga SB, Nittala MG, Parinitha B, Sadda SR, Chhablani JK. Correlation between retinal sensitivity and cystoid space characteristics in diabetic macular edema. Indian J Ophthalmol. 2016;64(6):452–458. doi:10.4103/0301-4738.187675 [CrossRef]
- Charafeddin W, Nittala MG, Oregon A, Sadda SR. Relationship between subretinal hyperreflective material reflectivity and volume in patients with neovascular age-related macular degeneration following anti-vascular endothelial growth factor treatment. Ophthalmic Surg Lasers Imaging Retina. 2015;46(5):523–30. doi:10.3928/23258160-20150521-03 [CrossRef]
- Maheshwary AS, Oster SF, Yuson RM, Cheng L, Mojana F, Freeman WR. The association between percent disruption of the photoreceptor inner segment-outer segment junction and visual acuity in diabetic macular edema. Am J Ophthalmol. 2010;150(1):63–67. doi:10.1016/j.ajo.2010.01.039 [CrossRef]