Association Between First- and Third-Month Responses to Intravitreal Ranibizumab for Diabetic Macular Edema

Abstract

Purpose:

This work examines the relationship between first- and third-month anatomical and visual response with antivascular endothelial growth factor for diabetic macular edema.

Methods:

We prospectively evaluated 58 eyes with center-involved diabetic macular edema. Response was categorized upon the anatomical status after 3 monthly doses based on the reduction of central macular thickness (CMT) from baseline (≥20% or not). Correlation analysis between the anatomical response status, gained letters, optical coherence tomography morphological features, and other baseline characteristics were obtained.

Results:

Twenty-five eyes (43.1%) achieved an anatomical reduction of ≥20% at the third month. Those with a reduction of ≥20% of CMT had subretinal fluid (P < .01), lower hemoglobin A_1c values (P < .01), lower proportion of intraretinal cysts (P < .01), a greater anatomical reduction, and visual improvement at the first month of treatment. Multiple logistic regression analysis, showed that the change of CMT after the first injection was an independent predictor for the anatomical reduction of ≥20% after the loading phase (P < .05). Best corrected visual acuity gain after the first dose showed a significant association with an improvement of ≥10 letters after the loading phase (P < .05), but not for macular thickness reduction.

Conclusions:

First month anatomical reduction was associated with the anatomical response at 3 months (P = .042) after monthly ranibizumab therapy. Visual improvement at the first month was predictive only for the visual outcome after the 3 monthly doses (P = .032).

Introduction

The leading cause of vision loss in patients with diabetes is diabetic macular edema (DME), which has a high prevalence in the Mexican population, ¹ and several options are available for its management. ² Several molecular mediators are related to the abnormal vascular permeability in eyes with DME, such as vascular endothelial growth factor (VEGF) and other inflammatory cytokines. ³ Many clinical trials have confirmed the efficacy of anti-VEGF drugs, which are now the standard of care for the management of DME. ^{4 -14} Anti-VEGF intravitreal injections are effective, but a subgroup of patients is unresponsive to treatment and often requires other alternatives, including other anti-VEGF substances, laser, and/or intravitreal steroids. ¹⁵ Persistent DME (> 6 months) can produce permanent changes in vision that could halt the visual recovery even when the edema resolves. Reports from the RISE and RIDE (A Study of Ranibizumab Injection in Subjects With Clinically Significant Macular Edema With Center Involvement Secondary to Diabetes Mellitus) studies have suggested that prompt treatment of DME is important to increase visual outcomes and that the delay of effective treatment may adversely affect final visual outcome. ⁷ Another problem of anti-VEGF therapy is the cost, because it is a major financial burden for patients. ¹⁶

An average reduction of 40% in central macular thickness (CMT) is observed by optical coherence tomography (OCT) within the first 3 months after starting treatment. ⁷ However up to 39% of patients treated with ranibizumab do not respond adequately to anti-VEGF therapy and receive macular laser treatment. ⁷ Additionally, the Diabetic Retinopathy Clinical Research Network (DRCR) protocol I study has shown that almost 70% of patients fail to gain more than 10 letters. ⁵ The EARLY study has reported that the best-corrected visual acuity (BCVA) outcomes after 3 years of ranibizumab treatment could be predicted by visual outcomes at the third month, ¹⁷ which supports the decision of changing therapy when edema persists and vision does not increase at the third month. Other reports have evaluated anatomical reduction as a predictive factor, showing that a 20% or greater CMT reduction at 12 weeks is related to good visual response to anti-VEGF therapy in the long term. ^18,19 However, other findings have shown an inconsistent association between early anatomical response and visual improvement in the long term. ^20,21 Switching treatment needs to be considered when a patient has failed to respond to anti-VEGF therapy, and it would be helpful to predict whether a patient will respond in the early phase of treatment. We sought to find the association between early response (first month) and the visual and anatomic outcomes at month 3 in patients treated with intravitreal ranibizumab for DME.

Methods

Study Design

A prospective cohort study from a single center was carried out. It was performed in 46 consecutive patients (58 naive eyes) with a diagnosis of center-involved DME at the Retina Service at the Mexican Institute of Ophthalmology, Querétaro, Mexico, from March 1, 2018 through January 2019.

Eligibility and Exclusion Criteria

Inclusion criteria were patients with: (1) age 18 years and older, (2) type 1 or type 2 diabetes mellitus, (3) center-involved DME, defined as a CMT of greater than 250 µm, as measured by spectral-domain OCT (SD-OCT) RTVue (Optovue, Inc), and (4) no treatment with any therapy (macular laser or/and panretinal photocoagulation, steroid, or anti-VEGF) for the prior 6 months. Patients with end-stage renal failure or renal transplant (due to diabetic nephropathy), glycated hemoglobin (HbA_1c) levels greater than 12%, vitreomacular traction or epiretinal membrane, cataract or vitreoretinal surgery within 6 months, rubeosis iridis, severe glaucoma, high-risk proliferative diabetic retinopathy, increased foveal avascular zone, and any conditions (cataract or corneal opacities) that could interfere with OCT measurement or VA were excluded.

Participants, Follow-up, and Outcome Measures

After informed, signed consent was received, clinical and ophthalmic history was recorded and a complete ophthalmologic examination was performed, including BCVA assessment (using the Early Treatment Diabetic Retinopathy Study [ETDRS] VA charts undertaken by an optometrist or the investigator [D.V.C.]), applanation tonometry, and anterior segment with dilated slit-lamp biomicroscopic examination. SD-OCT imaging was performed at baseline with the RTVue (Optovue, Inc). CMT was defined by the mean thickness of the 1-mm area circle centered on the fovea determined by the 5-mm macular map protocol, and high-definition linear scans were analyzed for the evaluation of structural changes and used for statistical analysis. Qualitative SD-OCT analysis was performed at baseline and assessed the presence or absence of the following morphologic features: subretinal fluid (SRF), intraretinal fluid (cysts), disorganization of retinal inner layers (DRIL), disruption of the external limiting membrane (ELM) and ellipsoid zone (EZ), and the presence of hyperreflective foci. All patients had their blood pressure and HbA_1c level recorded at the first visit. All patients underwent an intravitreal injection of ranibizumab (0.3 mg in 0.05 mL) at their baseline visit (day 0).

Participants were subsequently reviewed monthly (at 4, 8, and 12 weeks). At each visit, anterior chamber evaluation, intraocular pressure measurement, and dilated fundoscopy were performed. CMT data and ETDRS VA were recorded by the clinical investigator at baseline and at 4 and 12 weeks. The response status was categorized based on the reduction of CMT at the third month (≥ 20% CMT reduction or not). Eyes that gained 10 or more ETDRS letters at 12 weeks were also analyzed and compared with baseline factors.

Objectives

The primary outcome measure of the study was the proportion of eyes that achieved a CMT reduction of 20% or more at the third month. Anatomic response status at 3 months was compared with the initial (first month) visual/anatomic change and visual improvement at 12 weeks. Other characteristics such as HbA_1c, baseline OCT characteristics, lens status, and history of arterial hypertension were analyzed to find an association with the anatomical response status at 3 months.

Injection Technique

Patients’ initial injection was at day 0, and a maximum of 3 injections were performed (3 monthly doses) in 3 months. In a designated intravitreal treatment room, under sterile conditions, topical 0.5% tetracaine hydrochloride (Ponti-Ofteno, Sophia Laboratories) drops were given; a lid speculum was inserted and the injection site was cleaned with a drop of 5% povidone-iodine into the conjunctival sac. A 30-gauge needle was then inserted through the pars plana (supra or infratemporal quadrant), and 0.05 mL (0.3 mg) of ranibizumab (Lucentis, Novartis) was injected. A drop of 0.3% tobramycin and 0.1% dexamethasone (Trazidex; Sophia Laboratories) was placed in the fornix at the end of the procedure. Fundus examination (patency of the central retinal artery) was conducted and intraocular pressure was measured within 20 to 40 minutes after the injection. After the procedure, topical 0.3% tobramycin and 0.1% dexamethasone (Trazidex) was instilled 4 times per day for 3 days. Intravitreal injections were paid for by the patient.

Statistics

Quantitative variables were described using means and SDs. Categorical variables were described using frequency measures and percentages. Differences were examined with the χ² test. Multivariable logistic regression analysis was carried out to determine if the reduction in macular thickness at the first month is a predictor of the response at 3 months to anti-VEGF therapy. The Stata statistical package version 15.1 (StataCorp 2015, Stata Statistical Software: release 15, StataCorp LP) was used.

Results

A total of 48 patients were included in this study, and 58 eyes were considered for analysis. Two patients discontinued the study, 1 voluntarily and 1 because of health complications unrelated to the therapy. Thirty-eight eyes (65%) were from female patients, with a mean age 59.8 ± 9.1 (range, 35-75 years) and a mean HbA_1c level of 8 ± 1 (6%-10%). Baseline characteristics (demographics, BCVA, CMT, and ETDRS letters) for the entire study population and treatment response are summarized in Tables 1 and 2. Regarding the anatomic reduction at the third month, 33 eyes (56.9%) did not achieve an anatomical reduction of 20% or more. Significant differences in baseline demographic and clinical characteristics were found according to the proportion of response at the third month (Table 2); for example, those with a 20% or greater CMT reduction had SRF (P < .01), lower HbA_1c values (P < .01), a lower proportion of intraretinal cysts (P < .01), as well as a greater CMT reduction (Figure 1) and visual improvement at the first month of treatment.

Table 1.

Baseline Characteristics (N = 58).

Variable	Mean ± SD	Range
Age, y	59.8 ± 9.1	35-75
Sex, n (%)
Female	38 (65)	NA
Hypertension, n (%)	37 (63.8)	NA
Uncontrolled n (%)	35 (94.5)	NA
HbA1c, %	8 ± 1	6-10
Lens status, n (%)
Phakic	18 (31)	NA
Cataract	25 (43)	NA
Pseudophakic	15 (26)	NA
OCT morphological features, n (%)
Diffuse retinal thickening	2 (3.5)	NA
Intraretinal cysts	39 (67.2)	NA
Subretinal fluid	17 (29.3)	NA
HRF, n (%)	53 (91.4)	NA
DRIL, n (%)	10 (17.2)	NA
EZ disruption, n (%)	14 (24.1)	NA
Baseline letter score, n	43.8 ± 17.8	NA
Central macular thickness, µm	432.7 ± 140	260-910
Previous PRP, n (%)	51 (87.9)	NA

Abbreviations: DRIL, disorganization of the retinal inner layers; EZ, ellipsoid zone; HbA_1c, glycated hemoglobin; HRF, hyperreflective foci; NA, not applicable; OCT, optical coherence tomography; PRP, panretinal photocoagulation.

Table 2.

Demographic and Clinical Characteristics According to Anatomical Response to Treatment After 3 Injections (N = 58).^a

Variable	< 20% CMT n = 33	≥ 20% CMT n = 25	P valueb
Age, mean ± SD, y	61.9 ± 7.9	57.1 ± 10.0	.21
Sex, n (%)
Female	23 (69.7)	15 (60)	.44
Hypertension, n (%)	22 (66.7)	15 (60)	.60
Uncontrolled, n (%)	21 (95.5)	14 (93.3)	.78
HbA1c, mean ± SD, %	8.4 ± 0.9	7.6 ± 0.9	.004
Pseudophakic, n (%)	6 (18.2)	9 (36)	.12
Baseline OCT morphological features
Intraretinal cysts, n (%)	29 (87.9)	10 (40)	< .001
Subretinal fluid, n (%)	2 (6)	15 (60)	< .001
HRF, n (%)	31 (93.9)	22 (88)	.43
DRIL, n (%)	6 (18.2)	4 (16)	.83
ELM/EZ disruption, n (%)	7 (21.2)	7 (28)	.55
CMT, mean ± SD, µm	387.8 ± 119.7	492 ± 144.7	.001
Baseline letter score, mean ± SD, n	46.1 ± 17.1	40.8 ± 18.6	.35
Previous PRP, n (%)	32 (96.9)	19 (76)	.02
Response at first and third month, mean ± SD
CMT change after 1 injection, µm	+9.6 ± 74.5	–155 ± 99.1	< .001
Letters gained after 1 injection, n	+1.2 ± 11.6	+9.7 ± 8.3	.003
Letters gained after 3 injections, n	+2.48 ± 15.5	+15.32 ± 9.86	< .001
≥ 10 Letters gained after 3 injections, n (%)	10 (38.5)	16 (62)	.01

Abbreviations: CMT, central macular thickness; DRIL, disorganization of the retinal inner layers; ELM, external limiting membrane; EZ, ellipsoid zone; HbA_1c, glycated hemoglobin; HRF, hyperreflective foci; OCT, optical coherence tomography; PRP, panretinal photocoagulation.

^a Bold values show results with P < .01.

^b T or Mann-Whitney U test for quantitative variables and chi-square test for qualitative variables, as appropriate.

Figure 1.

Correlation between central macular thickness (CMT) change after the first injection and after the loading phase.

Several baseline OCT morphological features such as presence of DRIL, SRF, hyperreflective foci, disruption of ELM/EZ, or cystoid intraretinal lesions, were analyzed to assess their association with the response status at the third month. No significant differences were found among treatment response groups regarding the presence of DRIL (P = .42) and ELM/EZ disruption area (P = .82) at baseline. However, a greater anatomical response (≥ 20%) was associated with a higher percentage of eyes with SRF (60%) and higher baseline CMT; these differences were statistically significant (P < .01) (see Table 2).

CMT reduction after the first injection showed a strong correlation with the anatomical response after the 3 loading doses as well as the association between BCVA change at the first month with visual outcome after the third injection (P < .001) (see Figures 1 and 2). The anatomic reduction after the first injection showed a moderate correlation with change in VA after the loading phase (Figure 3). The multiple logistic regression analysis, when adjusting the variables to different baseline parameters and changes after the first injection, showed that the CMT change after the first injection was an independent predictor for a 20% or greater CMT reduction after the loading phase (P < .05) (Table 3). On the other hand, visual improvement after the first dose showed a significant correlation with an improvement of 10 or more letters after the loading phase (P < .05) (Table 4) but not for CMT reduction.

Figure 2.

Correlation between letters change after the first injection and after the loading phase.

Figure 3.

Correlation between central macular thickness (CMT) change after the first injection and letters improvement after the loading phase.

Table 3.

Factors Associated With a Reduction of 20% or More of Central Macular Thickness After the Loading Phase.^a

Variable	Regression coefficient	95% CI	P valueb
Constant	–17.610	–46.02 to 10.80	.22
Age, y	0.144	–.144 to .432	.33
Female sex	–3.507	–8.41 to 1.40	.16
Baseline letter score	0.038	–.095 to .172	.57
Baseline CMT, µm	0.013	–.007 to .032	.21
CMT after first injection, µm	–0.127	–.248 to –.0048	.04
Letters gained after first injection, n	0.237	–.152 to .626	.23
Previous PRP, n (%)	–4.656	–9.86 to .554	.08

Abbreviations: CMT, central macular thickness; PRP, panretinal photocoagulation.

^a Bold values show results with P < .05.

^b Multiple logistic regression analysis.

Table 4.

Factors Associated With Improvement of 10 or More Letters After the Loading Phase.^a

Variable	Regression coefficient	95% CI	P valueb
Constant	4.989	–11.25 to 21.23	.55
HbA1c	–1.105	–3.18 to .97	.30
Baseline CMT, µm	0.012	–.0062 to .032	.19
Baseline letter score	–0.144	–.322 to .033	.11
CMT change after 1 injection, µm	0.008	–.0072 to .025	.28
Letters gained after 1 injection, n	0.926	.052 to 1.80	.04

Abbreviations: CMT, central macular thickness; HbA_1c, glycated hemoglobin.

^a Bold values show results with P  <  .05.

^b Multiple logistic regression analysis.

Conclusions

The concept of responders to anti-VEGF therapy for DME has been reported for a while, and multiple molecular pathways help to explain this phenomenon. ^22,23 Despite the well-known evidence of the efficacy of anti-VEGF therapy for the management of DME, clinical trials data have shown that 50% to 60% of patients with treatment will not have a BCVA improvement of > 20/25 and/or will have persistent or recurrent edema at 12 months ^18,24,25 ; this group may correspond to patients with suboptimal response. Because chronic and persistent DME can lead to permanent diminished VA even after therapy is initiated, the need to start treatment early and identify the response status quickly, is of tremendous importance. ²⁶

Although macular thickness is widely used for monitoring the management of DME, it has been shown to have a weak relationship with visual outcomes. ^{27 -29} In our study, we analyzed the relationship between first- and third-month anatomical and visual responses. We prospectively observed eyes with center-involved DME after initiating a loading dose (3 monthly intravitreal ranibizumab injections). A mean CMT decrease of 67 µm (–11.4%) was obtained immediately after the loading dose, with a significant recovery of BCVA (+8.01 letters; P < .001). These anatomic and visual outcomes are similar to other known clinical trials (RESTORE, RISE, and RIDE) with the same treatment regimen, which also show a significant visual and anatomic improvement at the first month. ^6,13

Our analyses indicated that, at the cohort level, proportional anatomical response (< 20% vs ≥ 20% CMT reduction) after the loading phase was significantly correlated with proportional CMT response at the first month (see Figure 1). Similar to our findings, one study has shown that a patient with a CMT reduction of more than 15% after the first injection would often go on to have a 25% anatomical reduction at 12 weeks with a 75% and 92% sensitivity and specificity, respectively. ³⁰ Another report has shown that a CMT reduction of 8.7% after the first injection was able to predict 73% of patients who gained more than 5 letters at 12 weeks, with a modest specificity, ³¹ and a post hoc analysis of the protocol I study data has suggested that early and sustained anatomical reduction during ranibizumab therapy is associated with better long-term visual outcomes. ^18,25 However, analysis from clinical trials data (RISE/RIDE and READ-2) has reported a dissociation between early anatomical and long-term visual outcomes, ^32,33 which is consistent with other reports from the DRCR.net studies that have revealed a weak-to-modest correlation between visual outcomes and macular thickness. ^27,28

Protocol I study data have demonstrated that eyes with an early anatomical response (at 12 weeks) maintained this initial improvement during long-term treatment (83% of eyes continued to show a ≥ 20% CMT reduction at 3 years) in comparison with eyes that showed little or no anatomical improvement after the loading dose, which did not maintain improvement (48% of eyes had a ≥ 20% CMT reduction) at 3 years despite continuous treatment; another relevant outcome was that there was no significant association between this early anatomical response (≥ 20% CMT reduction at week 12) and the visual improvement at either 1 or 3 years. ¹⁹ These data support the outcomes of our study because we observed that OCT findings after 1 injection were predictive for anatomical reductions at 3 months, and protocol I data have shown that the early anatomical response (at 3 months) was correlated with the anatomical status after 1 and 3 years. They also have demonstrated that the anatomical reduction was not associated with visual outcome, similar to our study (Figure 3).

Another post hoc analysis of the DRCR protocol I study data had evaluated the predictability of the early visual response (12 weeks) after ranibizumab therapy and has shown that the letters gained after the loading phase (3 injections) were correlated with visual outcomes at 1 and 3 years. ¹⁷ These findings support the outcomes of our study because we found that visual improvement at the first month was predictive for visual outcome at 3 months (see Figures 4 and 5), and data from protocol I have shown that vision at the third month was correlated with visual outcome at 1 and 3 years. Even if it is not possible to conclude that anatomical and visual outcomes after 1 injection will predict vision 3 years later, a trend could be suggested.

Figure 4.

An optical coherence tomography image from an eye showing an anatomical reduction greater than 20% after 3 injections. (A) A 64-year-old pseudophakic patient with a hemoglobin A_1c of 6% presented with a cystoid macular edema and subretinal fluid; the baseline vision score was 7 letters and central macular thickness (CMT) was 649 µm. (B) The patient had a CMT reduction of 33% and vision gain of 9 letters after the first injection. (C) After the loading phase, a CMT decrease greater than 20% with an improvement of more than 10 letters was found. N indicates nasal; T, temporal; S, superior; I, inferior.

Figure 5.

An optical coherence tomography image from an eye with a reduced anatomical response (< 20%) after 3 injections. (A) A 63-year-old phakic patient with a hemoglobin A_1c of 6.5% presented with cystoid macular edema and several hyperreflective foci; baseline vision score was 38 letters and central macular thickness was 339 µm. (B) Vision improved by 2 letters after the first injection with no anatomical response. (C) After the loading phase there were no changes in central macular thickness, but there was an improvement of 8 letters. N indicates nasal; T, temporal; S, superior; I, inferior.

In contrast, there are reports showing that there is a continuous long-term vision gain despite an initial limited response (gain of less than 5 letters) at the third month ^34,35 (late responders). They have observed that even after an inadequate initial response, late responders' VAs at 2 years were similar to those that had an adequate initial response. ^32,35 Some authors maintain that vision improvement in patients with DME is usually a progressive and gradual process and that they will achieve a peak of visual improvement around 6 to 9 months or more after the initiation of therapy. ³⁶ That is why some authors consider 12 months to be the average time point to evaluate response status, alleging that it is better to wait longer to identify true responders. ³⁷

Our finding of a categorized anatomical response (≥ 20% of CMT reduction) obtained at 3 months was associated with a higher baseline CMT, which is consistent with other reports that have shown the greatest anatomical reductions occur in eyes with the highest baseline macular thickness. ^8,13,18 Regarding intraretinal fluid, many studies have found that the location and the size of the cysts are thought to be correlated with anatomic and visual outcomes over the course of treatment with intravitreal ranibizumab therapy, because those located in the outer layers and those of greater size (> 200 µm) are more likely to cause a poor visual outcome. ^21,38 Our study evaluated the presence of intraretinal cysts and found they were was associated with a limited anatomical response. Because we did not analyze the characteristics (size and location) of the cysts, the interpretation of this outcome is limited.

Although the presence of SRF as a predictive factor is not fully elucidated yet, our findings are consistent with others who found an association of SRF with greater anatomical reduction and better visual outcomes, because patients with SRF were more likely to achieve a final BCVA of better than 20/40 at 2 years of follow-up. ^21,39 Others studies have also found that the presence of SRF was associated with better visual outcomes after ranibizumab therapy (+21 gained letters in the SRF group vs +7.8 letters in the non-SRF group after the same number of injections; P = .004). ⁴⁰

In contrast to our study, HbA_1c levels have not been found to be related to functional or anatomical results at 1 year in eyes treated with ranibizumab for DME. ^41,42 Limitations of the present analysis include the small sample size, short follow-up, and the absence of information on additional factors such as DME angiographic subtype (focal vs diffuse) that are known to influence anatomical and VA responses to ranibizumab therapy.

Our analysis tried to minimize the time before the clinician makes a decision to continue the same anti-VEGF therapy or switch to alternative therapies (eg, intravitreal steroids). The present analysis indicates that the early (first-month) anatomical response to ranibizumab is strongly associated with anatomical outcome after the loading dose (3 months), and the early visual response is significantly associated with letters gained at the third month in eyes with DME. Although an early anatomical response does not prevent long-term anatomical improvement with continuous treatment, we have to bear in mind that the consolidation of the anatomical response in these circumstances could be prolonged, and if it does not reach the therapeutic objective of achieving timely elimination of all the fluid, patients could end up with a worse visual outcome. Making a recommendation based on our results would not be appropriate because of the methodological issues of our study, but we believe that the treatment response (visual and anatomic) in the first month could be taken into account when evaluating the response status after the loading phase of the treatment.

Our study suggested that the first-month anatomical reduction was associated with anatomical response at 3 months (P = .042) after anti-VEGF therapy for DME. Visual improvement at the first month was predictive of visual outcome at the third month (P = .032). The first-month anatomical reduction showed a moderate correlation with visual outcome after the loading phase, as well as the relationship between CMT reduction and letters gained after the loading dose.