Differences in the characteristics of subjects achieving complete, partial, or no resolution of macular edema in the READ-3 study

Abstract

Purpose

To identify baseline characteristics of subjects enrolled in the READ-3 study that would predict the response of macular edema to ranibizumab (RBZ) therapy at year 1.

Methods

In this post hoc analysis of the READ-3 randomized, multicenter phase 2 clinical trial, subjects with diabetic macular edema (DME) were randomized to receive monthly intravitreal injections of RBZ (0.5 or 2.0 mg) for 6 consecutive injections followed by as-needed treatments based on pre-defined retreatment criteria. In this sub-study, subjects were divided into three groups (persistent, rebound, and resolved) based on edema status at month 12 (M12). Multi-logistic regression was utilized to assess the probability of edema outcomes M12, based on the baseline characteristics.

Results

One hundred twenty-three out of 152 subjects were analyzed for this sub-study. A significant difference was observed in the baseline (BL) central subfield thickness (CST) among the study groups (p < 0.05). BL CST was a significant predictor for edema outcome at M12 with > 80% probability of the subject having persistent edema if BL CST was > 570 μm (p < 0.05). This association persisted when controlled for the dose of RBZ (relative risk (RR), 1.007; p < 0.05). BL CST was also a significant predictor for having persistent edema at M12 in subjects without vitreomacular adhesion (VMA) (> 80% probability of edema persistence at CST > 570 μm [RR, 1.006; p < 0.05]). However, in the presence of VMA, BL CST was no longer a significant predictor of having persistent edema at month 12 (RR, 1.005; p > 0.05).

Conclusions

Subjects with high CST (> 570 μm) at baseline may not benefit from repeated intravitreal injections of anti-VEGF for resolution of edema.

Introduction

Diabetic macular edema (DME) is a leading cause of diabetes-related vision loss and affects up to 21 million out of 93 million patients affected with diabetic retinopathy worldwide [1–3]. Anti-vascular endothelial growth factor (anti-VEGF) therapy, while the mainstay for diabetic macular edema, does not always result in complete resolution of the edema, despite multiple, frequent injections [4–7].

Nearly 50% of the subjects in the RESTORE study, who were treated with anti-VEGF, had a central subfield thickness (CST) of > 275 μm at month 12, on time domain optical coherence tomography (TD-OCT), demonstrating that a significant number of ranibizumab (RBZ)–treated subjects still had residual intra-retinal fluid [7]. Following two years of treatment of DME with RBZ, in the READ-2 study, about half of the subjects still had residual edema [8]. Similarly, approximately 41% of the subjects in the READ-3 study had a CST of > 250 μm at month 24 [9]. All three studies reported CST values measured by TD-OCT.

Long-term follow-up studies of DRCR.net demonstrated that approximately 40% of the subjects treated with IVT ranibizumab had persistent DME at 3 years [10]. In the study comparing various IVT anti-VEGF agents, subjects demonstrated the rates of 44.2%, 68.2%, and 54.5% with aflibercept, bevacizumab, and ranibizumab, respectively, at 2 years for persistent DME [11]. Visual outcomes of subjects with persistent DME were poor compared to those without DME.

There is a high cost associated with repeated anti-VEGF agents for subjects with DME. The incremental cost-effectiveness ratio of aflibercept and ranibizumab compared with bevacizumab has been reported to be $1,110,000 per quality-adjusted life-year (QALY) and $1,730,000 per QALY, respectively [12–14]. However, a significant number of subjects with DME show no or partial resolution of edema after anti-VEGF therapy. Therefore, there is a need to identify biomarkers that could assist in establishing the likelihood of response, thereby allowing treating physicians to convey a more precise message to the patient regarding their expectations.

In this analysis, we assessed the differences in outcomes of subjects enrolled in the READ-3 study and evaluated the utility of various baseline parameters as predictors of outcomes at month 12.

Methods

Data from the Ranibizumab for Edema of the Macula in Diabetes: Protocol 3 with High Doses (READ-3) clinical trial were used for this analysis. The READ-3 study was a phase 2, randomized, multi-centered clinical trial designed to evaluate the effects of 2 doses of ranibizumab (0.5 and 2.0 mg) in eyes with DME. Subjects in both groups received monthly intravitreal injection of ranibizumab from baseline to month 6 and on an as-needed basis, thereafter. Patients in both groups received an injection of RBZ at baseline and months 1, 2, 3, 4, and 5. Month 6 was the primary end point of the study. After month 6, patients in both groups were evaluated every month and were deemed eligible to receive additional RBZ if the CSF thickness was ≥ 250 μm or if there was evidence of any macular fluid on OCT. OCT was performed during each visit.

The study is registered at www.clinicaltrials.gov under the identifier NCT01077401 and was conducted in compliance with the Declaration of Helsinki, the United States Code of Federal Regulations Title 21, and the Harmonized Tripartite Guidelines for Good Clinical Practice (1996). The study was approved by Western institutional review board and local institutional review boards for selected sites. Written informed consent was obtained from all participants of the study.

Inclusion and exclusion criteria

Subjects were included in the study if they met the following criteria: (1) participation in the READ-3 trial and completion of the month 12 visit and (2) availability of spectral-domain optical coherence tomography (SD-OCT) images of sufficient quality for grading. Inclusion and exclusion criteria of the READ-3 study have been published previously [15].

Edema status assessment

In this sub-study analysis, data of all the subjects in READ-3 study was combined and categorized into three groups (persistent, rebound, and resolved) based on presence or absence of edema as visualized on SD-OCT scans on each monthly visit from baseline to month 12. In READ-3 study protocol, presence of edema was defined as having a central subfield thickness of ≥ 250 μm on TD-OCT or if there was evidence of fluid in the macular region on SD-OCT. Only eyes with SD-OCT were included in this analysis.

Eyes which showed no resolution of edema at any time point during the study were classified into the persistent edema group. Eyes in which the edema disappeared after baseline but reappeared before or at month 12 were classified into the rebound group. Eyes in which the edema resolved at any time point after baseline and did not reappear till month 12 were labeled as the resolved group.

Interleukin-6 level assessment

Subjects enrolled in the READ-3 study underwent anterior chamber (AC) fluid (50 μl) tap quarterly from baseline to month 12. The levels of interleukin 6 (IL-6) were quantified using an MSD^® MULTI-SPOT Cytokine Assay System (Meso Scale Diagnostics, LLC; Rockville, MD).

Outcome measures

Outcome measures assessed in the study included (1) differences in the baseline characteristics between the three study groups; patient demographics, prior anti-VEGF therapy, presence of vitreomacular adhesion (VMA), HbA1c, CST, and best-corrected visual acuity (BCVA); (2) probability of edema outcomes based on the baseline CST, VMA status, and prior anti-VEGF therapy; (3) mean change in CST as determined by SD-OCT from baseline to month 12; and (4) mean change in BCVA from baseline to month 12 as measured by the number of Early Treatment Diabetic Retinopathy Study (ETDRS) letters read at 4 meters.

Statistical analysis

Statistical analysis was performed using Stata v14.1 (Stata Corp, TX). Frequencies were compared using the chi-square test and Fisher exact test. Chi² and ANOVA were used to assess the differences among the study groups. Multi-logistic regression modeling was utilized to assess the probability of edema outcomes based on the baseline CST, VMA status, and prior anti-VEGF therapy.

Results

One hundred fifty-two (152) subjects were randomized in the READ-3 study. Of 152 subjects, 123 were eligible for this sub-study based on the pre-specified inclusion and exclusion criteria. Twenty-nine subjects were excluded from this sub-study due to non-availability of SD-OCT images at all study visits. Among the eligible subjects, 82 (66.7%) subjects were classified into the persistent edema group, 21 (17.1%) subjects were assigned to the rebound group, and 20 (16.3%) subjects had resolved edema.

Table 1 describes the baseline characteristics of the study groups. No significant difference was observed in the baseline age, gender, race, body mass index (BMI), smoking status, HbA1c, and BCVA between the study groups (p > 0.05). However, a significant difference was observed in the baseline CST among the study groups at baseline (p = 0.0002). The majority (56.1%) of the subjects in the persistent edema group were receiving a low dose (0.5 mg) of RBZ while the majority of the subjects in the resolved group received a high dose (2.0 mg) of RBZ (55.0% and 71.4%, respectively) (p, 0.07).

Table 1.

Clinical features of the study groups

	Persistent N = 82	Rebound N = 20	Resolved N = 21	p value
Mean age (year ± SD)	65.0 ± 9.0	65.5 ± 8.5	60.9 ± 12.0	0.18
Females [% (n)]	40.2 (33)	65.0 (13)	47.6 (10)	0.13
Mean BMI (kg/m2 ± SD)	31.3 (8.4)	31.4 (6.1)	30.3 (5.8)	0.85
Race [% (n)]
Caucasian	66.7 (46)	81.3 (13)	57.1 (10)	0.52
African American	18.8 (13)	12.5 (2)	7.1 (1)
Asian	7.3 (5)	6.3 (1)	14.3 (2)
Native Hawaiian/Pacific Islander	4.4 (3)	0 (0)	14.3 (2)
Other	2.9 (2)	0 (0)	7.1 (1)
Smoking status [% (n)]	29.3 (24)	40.0 (8)	38.1 (8)	0.55
Prior anti-VEGF therapy [% (n)]	9.1 (6)	6.3 (1)	16.7 (3)	0.55
VMA at baseline	19.1 (13)	25.0 (4)	23.8 (5)	0.82
Mean baseline BCVA (ETDRS letters)	28.1 ± 10.0	26.7 ± 14.1	28.4 ± 12.1	0.85
Mean baseline CST (μm)	536.7 ± 127.1	424.6 ± 82.8	456.5 ± 108.7	0.0002*
Mean baseline HbA1c (%)	7.8 ± 1.5	7.9 ± 2.0	7.3 ± 1.6	0.47
Dose of ranibizumab [%(n)]
2.0 mg	43.9 (36)	55.0 (11)	71.4 (15)	0.07
0.5 mg	56.1 (46)	45.0 (9)	28.6 (6)
Mean number of intravitreal injections (number ± SD)	11.1 ± 1.34	8.8 ± 1.4	8.5 ± 1.9	< 0.0001*
Median Interleukin-6 (pg/mL)	13.1	6.15	23.5	0.38

BCVA best-corrected visual acuity, CST central subfield thickness, ETDRS Early Treatment Diabetic Retinopathy Study, SD standard deviation

Prediction of macular edema outcome at month 12

Multi-logistic regression analysis identified baseline CST as a significant predictor for edema status at month 12, with 80% probability of a patient having persistent edema at month 12 if their baseline CST was > 570 μm (relative risk (RR), 1.006; p, 0.01) (Fig. 1). This association persisted even when the model was controlled for the dose of RBZ (RR, 1.007; p, 0.008).

Fig. 1.

Probability of edema outcome based on baseline central subfield thickness among the subjects in study groups

Baseline CST was also a significant predictor for having persistent edema at M12 in subjects with no VMA at baseline with > 80% probability of a patient having persistent edema at month 12 if their baseline CST was > 570 μm (RR, 1.006; p, 0.04) (Fig. 2). However, in the presence of VMA, baseline CST was no longer a significant predictor and the probability of having persistent edema was found to be 80% at a CST of 730 μm (RR, 1.005; p, 0.29).

Fig. 2.

Probability of persistent edema in the presence of vitreomacular adhesion at baseline among the subjects in study groups

Subjects with prior anti-VEGF therapy had an 80% probability of having persistent edema if their baseline CST was > 510 μm; however, this regression model was insignificant (RR, 1.02; p, 0.22).

None of the other baseline characteristics (age, gender, race, BMI, smoking status, HbA1c, and BCVA) could be identified as predictors of the edema status.

Change in best-corrected visual acuity and central subfield thickness

The mean change in BCVA in the persistent edema group was 8.6 ± 9.5 letters at month 12; in the rebound edema group, 5.6 ± 8.2 letters; and 11.0 ± 8.4 letters in the resolved edema group. The mean change in BCVA at month 12 was not statistically different among the three edema status groups (p, 0.18). Figure 3 illustrates the mean change in BCVA from baseline for all the groups.

Fig. 3.

Mean change from baseline in best-corrected visual acuity (ETDRS letters) among the subjects in study groups

At month 12, the mean change in CST was − 184.9 ± 142.1 μm for the persistent edema group, − 139.1 ± 105.0 μm for the rebound edema group, and − 210.6 ± 116.7 μm for the resolved edema group. The difference in the mean reduction in CST, at month 12, among the three groups was not statistically significant (p, 0.22). Figure 4 illustrates the mean change from baseline in CST in the study groups.

Fig. 4.

Mean change from baseline in central subfield thickness among the subjects in study groups

At month 12, 7/82 (8.5%) eyes had increased CRT than the baseline in the persistent group, 2/20 (10%) had increased CRT than the baseline in the rebound group, and none of the eyes had increased CRT at month 12 when compared to the baseline in the resolved edema group (p value = 0.36).

Interleukin-6

At baseline, the median IL-6 level of patients whose CST was ≥ 570 μm was 34.15 pg/mL (range, 2–349) while the median IL-6 in patients with CST < 570 μm was 11.4 pg/mL (range, 2–676). This difference was statistically significant with a p value of 0.03. The median IL-6 level at baseline was the highest (median, 23.5; range, 2–677 pg/mL) in the resolved group.

Discussion

Intravitreal ranibizumab (IVT RBZ) is a well-established therapy for achieving visual improvement, reducing the risk of diabetic retinopathy progression, and resolution of macular edema in patients with DME [7, 16–18]. In this study, logistic regression analysis identified baseline CST as a significant predictor of status of macular edema at month 12. No significant relationship was observed between other baseline characteristics (age, gender, race, BMI, smoking status, HbA1c, and BCVA).

Despite the favorable outcomes with IVT RBZ therapy, there are subjects who demonstrate a poor response to IVT RBZ in clinical practice [8, 19]. Subjects with chronic persistent DME have demonstrated worse visual outcomes compared to subjects who had DME which resolved in the long-term follow-up studies [10, 11]. Hence, it is important to identify markers which can help clinicians and the patients to identify the response to the treatment they can expect to achieve. Only a few studies have explored the prognostic factors for anatomical outcomes and visual improvement after IVT RBZ for DME [8, 19, 20]. There is a dearth of literature regarding the prediction of macular edema outcome after IVT RBZ therapy based on the baseline characteristics of subjects.

Our study demonstrates that subjects with high (> 570 μm on SD-OCT) baseline CST may not resolve edema with monthly IVT RBZ. We believe that the strong relationship found between the high baseline CST and the outcome of macular edema in the subjects participating in the READ-3 study is representative of the routine clinical use of RBZ. The retreatment criteria used in the READ-3 study are similar to those encountered in clinical practice, in which therapy is given at the clinician’s discretion based on the findings from clinical examination and imaging. Therefore, baseline CST can be used to predict response in terms of resolution of macular edema.

Various potential reasons can be cited for why high baseline CST may result in persistent macular edema. Since high CST is representative of more fluid in the macular region, it is possible that eyes with high CST require monthly injections for an extended period of time to dry out and stabilize the macula [21, 22].

Additionally, patients with high baseline CST had higher median IL-6 levels which indicates an active inflammatory pathway which compounds to the macular edema in such cases. However, higher aqueous concentrations of IL-6, although having been shown to correlate with ocular inflammatory status, may not always reflect the treatment response. In the current study, no significant difference at the median levels of IL-6 was found in between the 3 groups. The IL-6 levels were higher in the resolved and persistent groups. Treatment response in DME may not be defined solely by IL-6 as the mechanism of macular edema is rather complex and multifactorial. IL-6 itself also is a pleiotropic cytokine with a broad range of biological functions, and its effects are highly dependent on signaling receptor protein gp130, soluble IL-6 receptors, and other various factors. Regardless of the underlying reasons for the findings, the ability to identify the subjects that will achieve resolution of macular edema has important implications for the risk and benefit profile of anti-VEGF therapy. The benefits to be gained from IVT anti-VEGF therapy in eyes with high baseline CST should be discussed with the patients in great detail to cover the expectations and to justify the high cost associated with anti-VEGF therapy which potentially adds burden on the healthcare system [12, 13].

Our study also shows that presence of VMA has a beneficial effect and decreases the probability of having persistent macular edema. These findings are in accordance with the literature which has shown a similar beneficial effect of VMA in subjects with DME who are treated with IVT anti-VEGF therapy [9, 23].

Potential limitations of our study include a 12-month duration of follow-up of the study subjects. It may be of interest to observe the response in terms of edema resolution in a real-life situation with longer follow-up. Furthermore, more detailed anatomical characteristics of the retina were not taken into account for the prediction of macular edema which can further help the clinicians to make an informed decision.

Conclusion

In summary, this analysis assessed the predictive value of baseline characteristics to resolve macular edema in subjects with DME who are treated with intravitreal ranibizumab in the first year of a clinical trial. A high baseline CST (> 570 μm) may not benefit from repeated intravitreal injections of anti-VEGF for resolution of macular edema. However, the presence of VMA reduces the probability of having persistent edema at month 12. These findings may become important measures and parameters in the clinical decision-making for patients with DME.

Key messages.

• Almost half of the eyes treated with intravitreal anti-VEGF injections for diabetic macular edema have persistent diabetic macular edema (DME).On the other hand; there is still a lack of a clear predictor of response of DME to ranibizumab therapy.

• Higher baseline central subfoveal thickness was a significant predictor for the persistence of macular edema during intravitreal Ranibizumab (RBZ) treatment

• Vitreomacular adhesion status may have an effect on diabetic macular edema persistence during RBZ injections.