To determine the association between baseline subfoveal choroidal thickness and short-term response to intravitreal anti–vascular endothelial growth factor (anti-VEGF) therapy in diabetic macular edema (DME).
Retrospective, consecutive case series.
Fifty-three eyes from 42 patients diagnosed with treatment-naïve DME were treated with 3 monthly intravitreal injections of ranibizumab or bevacizumab. Serial enhanced depth imaging optical coherence tomography scans were used to measure subfoveal choroidal thickness and central macular thickness (CMT). Anatomic response (CMT decrease ≥50 μm) and functional response (best-corrected visual acuity gain ≥1 line) were assessed at 3 months follow-up using univariate and multivariate analyses.
After 3 monthly anti-VEGF treatments, subfoveal choroidal thickness decreased significantly (225 μm at baseline, 201 μm at 3 months, P < .0001). The anatomic responder group (32 eyes) had a greater baseline choroidal thickness (243 ± 15 μm) than the nonresponder group (21 eyes, 198 ± 13 μm, P = .03). Similarly, the functional responder group (28 eyes) tended to have a greater baseline subfoveal choroidal thickness (239 ± 12 μm) than the nonresponder group (25 eyes, 211 ± 16 μm, P = .08). Multivariate analyses revealed that a greater baseline subfoveal choroidal thickness was associated with a better anatomic (odds ratio = 1.12 for every 10 μm increase, P = .03) and functional response (odds ratio = 8.45 for >200 μm vs ≤200 μm, P = .008).
Baseline subfoveal choroidal thickness may help predict which patients with DME will respond more favorably in the short term to intravitreal anti-VEGF pharmacotherapy. In this study, eyes with a thicker baseline subfoveal choroidal thickness had better short-term anatomic and functional responses.
Diabetic macular edema (DME) is the leading cause of blindness in patients with diabetic retinopathy worldwide. While alterations in retinal vasculature resulting in compromise of the blood-retinal barrier have been demonstrated to play a critical role in the pathophysiology of the disease, changes in the underlying choroidal vasculature may also play a contributing role.
With the advent of enhanced depth imaging spectral-domain optical coherence tomography (EDI OCT) imaging, detailed visualization of the choroidal compartment and direct comparison between physiologic and pathologic states are now possible. Recent studies applying EDI OCT to evaluate choroidal thickness in eyes with DME and proliferative diabetic retinopathy (PDR) have demonstrated altered choroidal morphologic features in both processes. Querques and associates found an overall thinning of the choroid compared to healthy age- and sex-matched controls. They hypothesized that decreased choroidal thickness may lead to retinal hypoxia, higher levels of vascular endothelial growth factor (VEGF), and breakdown of the blood-retinal barrier, propagating DME. In a separate study exploring the effect of anti-VEGF treatment and laser photocoagulation on choroidal thickness in diabetic retinopathy, Lains and associates found that eyes receiving intravitreal anti-VEGF injections experienced a reduction in choroidal thickness. In contrast, those treated with macular laser showed no significant change in thickness.
Despite growing evidence demonstrating alterations to the choroidal vasculature in DME, the relationship between subfoveal choroidal thickness and treatment response to anti-VEGF therapy has not been elucidated. We investigated the association between initial subfoveal choroidal thickness in treatment-naïve eyes with DME and short-term treatment response to monthly anti-VEGF therapy at 3 months follow-up. We further delineated baseline characteristics that may predict a positive anatomic as well as functional therapeutic response.
Following approval from the Institutional Review Board at Wills Eye Hospital, a retrospective, consecutive chart review of patients diagnosed and treated with DME at the Retina Service of Wills Eye Hospital and the outpatient offices of Mid Atlantic Retina was performed. The billing records of all patients seen between December 1, 2010 and September 30, 2013 were reviewed for the corresponding ICD-9 362.07 code.
Inclusion criteria consisted of patients who were treatment naïve prior to their first anti-VEGF injection and were determined to have clinically significant macular edema by the examining physician as defined by the Early Treatment Diabetic Retinopathy Study. Center-involving DME with central macular thickness (CMT) ≥300 μm on OCT was also required. Only patients treated with 3 consecutive monthly anti-VEGF injections were eligible for study inclusion. Patients were excluded if they had any other treatments related to their diabetic retinopathy before the first intravitreal anti-VEGF injection or during the course of the initial 3 monthly intravitreal anti-VEGF injections with any of the following interventions: intravitreal or sub-Tenon injections of corticosteroids, corticosteroid implants, focal/grid macular laser photocoagulation, panretinal photocoagulation (PRP), or pars plana vitrectomy. Additionally, individuals were excluded if they had any of the following concomitant ocular diseases aside from nonproliferative diabetic retinopathy (NPDR) in the treated eye: age-related macular degeneration, central/branch retinal vein occlusion, choroidal neovascularization, history of ocular trauma, or any history of prior intraocular surgery (with the exception of cataract surgery).
Pertinent clinical data recorded included patient age, sex, hypertension, duration of diabetes mellitus, baseline and follow-up best-corrected Snellen visual acuity (BCVA), complete biomicroscopic examination findings, injection history, and length of follow-up.
Enhanced depth imaging optical coherence tomography (Spectralis; Heidelberg Engineering, Heidelberg, Germany) was performed for all patients. Automated CMT measurements were generated using a 25-line raster scan pattern protocol. Serial EDI OCT scans were analyzed and subfoveal choroidal thickness was manually measured beneath the fovea using the caliper tool in the Heidelberg Eye Explorer software, from the hyperreflective line of the Bruch membrane to the hyperreflective line of the chorioscleral interface. Eyes in which the chorioscleral interface was not easily discernible were excluded from the study. The Heidelberg’s tracking feature enabled sequential scans to be captured in identical locations, providing accurate assessments of any changes in retinal morphology. Two examiners (N.R., E.R.) who were experienced at analyzing EDI OCT scans performed these measurements and were masked to patient clinical data at the time of acquisition.
Paired 2-tailed t test was used to compare outcome measures taken at baseline and at 3 months of follow-up. The associations of baseline subfoveal choroidal thickness with anatomic response (ie, CMT change from baseline) and functional response (visual acuity change from baseline) at 3 months were first assessed using linear regression analysis by modeling CMT change and BCVA change as a continuous measure. To further evaluate whether baseline subfoveal choroidal thickness may predict clinically relevant response to anti-VEGF treatment at 3 months, we defined anatomic responders as patients having a decrease in CMT of ≥50 μm from baseline and functional responders as patients with BCVA gains of ≥1 line from baseline. The differences between responders and nonresponders were compared using unpaired, 2-tailed t test. To determine the baseline predictors for anatomic responders and functional responders, univariate analysis was performed, followed by multivariate logistic regression analysis that included predictors with P < .10 from the univariate analysis. Odds ratios (OR) and their corresponding 95% confidence intervals (95% CI) were calculated from the multivariate logistic regression. BCVA was converted to logarithm of the minimal angle of resolution (logMAR) values for statistical analysis. All statistical analyses were performed using SAS v9.3 (SAS Institute Inc, Cary, North Carolina, USA).
A total of 53 eyes from 42 consecutive patients with treatment-naïve DME meeting inclusion criteria for the study were evaluated. Baseline demographics and ocular characteristics are outlined in Table 1 . Overall (53 eyes), patients’ mean logMAR BCVA improved from 0.74 ± 0.43 (Snellen equivalent: 20/110) at baseline to 0.60 ± 0.43 (Snellen equivalent: 20/80, P < .001) at 3 months follow-up. Furthermore, mean initial CMT was 477 ± 124 μm and decreased to 386 ± 94 μm ( P < .001) following 3 monthly anti-VEGF injections. Similarly, mean overall subfoveal choroidal thickness decreased from 225 ± 70 μm at baseline to 201 ± 70 μm ( P < .001) at 3 months follow-up.
|Demographic Characteristics (N = 42 patients)|
|Mean (SD)||63.8 (12.6)|
|Median (range)||66 (19–84)|
|Duration of diabetes (y)|
|Ocular Characteristics (N = 53 eyes)|
|Intravitreal injection received|
|Bevacizumab only||22 (41.5%)|
|Ranibizumab only||17 (32.1%)|
|Baseline BCVA (logMAR)|
|Mean (SD)||0.74 (0.44)|
|Median (range)||0.54 (0.20–2.00)|
|Mean (SD)||477 (124)|
|Median (range)||445 (307–939)|
|Baseline SFCT (μm)|
|Mean (SD)||225 (70)|
|Median (range)||225 (109–424)|
|Baseline ellipsoid zone|
Comparison of Anatomic Responders and Nonresponders
Baseline and month 3 characteristics of the anatomic responders and nonresponders as defined based on CMT change from baseline are outlined in Table 2 . The mean ± standard error (SE) baseline CMT for anatomic responders was 504 ± 27 μm, compared to 436 ± 23 μm for nonresponders ( P = .07). Following 3 monthly intravitreal injections, anatomic responders’ mean CMT decreased to 350 ± 12 μm ( P < .0001). Conversely, the nonresponders’ mean CMT had increased to 439 ± 25 μm ( P = .71).
|Outcome Measures||Anatomic Responders (n = 32)||Anatomic Nonresponders (n = 21)||P Value b|
|Mean (SE)||Mean (SE)|
|Baseline||504 (27)||436 (23)||.07|
|3 months||350 (12)||439 (25)||.006|
|Change from baseline||−153 (24)||3.6 (9.8)||<.0001|
|P value c||<.0001||.71|
|Baseline||243 (15)||198 (13)||.03|
|3 months||213 (14)||183 (14)||.13|
|Change from baseline||−30 (6)||−15 (5)||.08|
|P value c||<.0001||.006|
|Baseline||0.74 (0.09)||0.73 (0.09)||.90|
|3 months||0.59 (0.08)||0.62 (0.09)||.81|
|Change from baseline||−0.15 (0.04)||−0.11 (0.04)||.45|
|P value c||<.0001||.01|
With respect to choroidal thickness, patients in the anatomic responder group had a significantly greater mean baseline subfoveal choroidal thickness of 243 ± 15 μm than the nonresponder group, which was 198 ± 13 μm ( P = .03). At 3 months follow-up, subfoveal choroidal thickness had decreased to 213 ± 14 μm (change 30 ± 6 μm; P < .0001) in the responder group and 183 ± 14 μm (change 15 ± 5 μm; P = .006) in the nonresponder group. The decrease in choroidal thickness from baseline to 3 months tended to be greater for anatomic responders compared to nonresponders ( P = .08).
Baseline mean logMAR BCVA was 0.74 ± 0.09 (Snellen equivalent: 20/110) in the anatomic responder group and 0.73 ± 0.08 (Snellen equivalent: 20/110) in the nonresponder group ( P = .90). At 3 months follow-up, patients in the anatomic responder group had experienced a significant improvement in vision to 0.59 ± 0.08 (Snellen equivalent: 20/80, P < .0001), and patients in the anatomic nonresponder group also had an improved mean visual acuity of 0.62 ± 0.09 (Snellen equivalent: 20/83, P = .01). The change in BCVA from baseline to month 3 was not statistically significant in the anatomic responders compared to the nonresponders ( P = .45).
Comparison of Functional Responders and Nonresponders
Baseline and month 3 characteristics of the functional responders and nonresponders are outlined in Table 3 . The functional responders had a mean baseline logMAR BCVA of 0.79 ± 0.08 (Snellen equivalent: 20/120), which was comparable to the vision in the nonresponder group of 0.67 ± 0.09 (Snellen equivalent: 20/95, P = .34). At month 3, mean logMAR BCVA in responders significantly improved to 0.50 ± 0.07 (Snellen equivalent: 20/60, P < .0001), while nonresponders significantly declined to 0.71 ± 0.09 (Snellen equivalent: 20/100, P = .04).
|Outcome Measures||Functional Responder (n = 28)||Functional Nonresponder (n = 25)||P Value b|
|Mean (SE)||Mean (SE)|
|Baseline||0.79 (0.08)||0.67 (0.09)||.34|
|3 months||0.50 (0.07)||0.71 (0.10)||.09|
|Change from baseline||−0.29 (0.03)||0.04 (0.02)||<.0001|
|P value c||<.0001||.04|
|Baseline||497 (24)||455 (22)||.13|
|3 months||387 (18)||384 (21)||.91|
|Change from baseline||−109 (26)||−71 (20)||.15|
|P value c||<.0001||.0005|
|Baseline||239 (12)||211 (16)||.08|
|3 months||218 (12)||183 (14)||.03|
|Change from baseline||−21 (5)||−28 (6)||.41|
|P value c||.0001||<.0001|
Functional responders had a mean baseline CMT of 497 ± 24 μm, which was greater than that of functional nonresponders (455 ± 22 μm), though this difference did not reach statistical significance ( P = .13). Following 3 monthly intravitreal injections, the mean CMT significantly decreased, to 387 ± 18 μm ( P < .0001) in the functional responder group and 384 ± 21 μm ( P = .0005) in the nonresponder group. The change in CMT from baseline to month 3 demonstrated a trend toward a greater decrease in the functional responders (109 ± 26 μm) compared to the nonresponders (71 ± 20 μm), yet the difference did not reach statistical significance ( P = .15).
Initial mean subfoveal choroidal thickness in the functional responder group was 239 ± 12 μm, which tended to be greater than in the nonresponders group (211 ± 16 μm; P = .08). After 3 months of follow-up, subfoveal choroidal thickness had decreased to 218 ± 12 μm (change 21 ± 5 μm; P = .0001) in the functional responders, compared to 183 ± 14 μm (change 28 ± 6 μm; P < .0001) in the nonresponders. The change in subfoveal choroidal thickness from baseline to 3 months was similar between the 2 groups ( P = .41).
Baseline Predictors of Anatomic Responders
Univariate analysis demonstrated that patients with a higher baseline subfoveal choroidal thickness were more likely to attain anatomic response (CMT decrease ≥50 μm from baseline) at 3 months (linear trend, P = .03). Similarly, a higher baseline CMT was significantly associated with a better anatomic response at 3 months follow-up (77.8% for baseline CMT >440 μm vs 42.3% for baseline CMT <440 μm, P = .01; Supplemental Table 1 , available at AJO.com ).
Multivariate analysis revealed that both a higher baseline subfoveal choroidal thickness and CMT were independent predictors for achieving anatomic response. The adjusted odds ratio anatomic response was 1.12 for every 10 μm increase in subfoveal choroidal thickness (95% CI: 1.01–1.23, P = .03), and 4.47 for CMT >440 μm vs CMT ≤440 μm (95% CI: 1.25–16.0, P = .02).
Baseline Predictors for Functional Responders
In univariate analysis, patients who had a higher baseline subfoveal choroidal thickness (>200 μm) were more likely to attain BCVA improvement of 1 line or more at 3 months follow-up than patients with a baseline subfoveal choroidal thickness ≤200 μm (65.7% vs 27.8%, P = .01; Supplemental Table 2 , available at AJO.com ). In addition, having a worse baseline BCVA (>0.50 logMAR) was significantly associated with gaining 1 or more lines in BCVA (63.6% vs 35.0%, P = .04). Eyes with a disrupted ellipsoid zone (n = 33) at baseline had a worse mean baseline BCVA (Snellen equivalent: 20/140) than eyes with an intact ellipsoid zone (n = 20, Snellen equivalent: 20/72, P = .02). However, the integrity of the ellipsoid zone at baseline was not found to be associated with functional response ( P = .41). The functional response rate was 57.6% in eyes with a disrupted ellipsoid zone, compared to 45% in eyes with an intact ellipsoid zone.
Multivariate analysis demonstrated that having a higher baseline subfoveal choroidal thickness or a worse baseline BCVA were independent predictors for achieving functional response. The adjusted odds ratio for functional response was 8.45 for subfoveal choroidal thickness >200 μm vs ≤200 μm (95% CI: 2.01–35.5, P = .003), and 5.88 for baseline BCVA >0.50 vs ≤0.50 logMAR (95% CI: 1.46–23.8, P = .008).
When functional response was evaluated using BCVA change from baseline as a continuous measure, we also found that a higher baseline subfoveal choroidal thickness and a worse baseline BCVA were significantly associated with greater BCVA improvement in multivariate linear regression analysis. For eyes with a baseline subfoveal choroidal thickness greater than 200 μm, mean BCVA improved by 0.15 logMAR at 3 months, whereas an increase of 0.04 logMAR was noted in eyes with a subfoveal choroidal thickness of 200 μm or less ( P = .050). For eyes with a baseline logMAR BCVA greater than 0.5, BCVA improved by 0.18 logMAR at 3 months, whereas an increase of 0.01 logMAR occurred in eyes with a baseline logMAR BCVA 0.5 or less ( P = .0008).