Non-responders to ranibizumab Anti-VEGF treatment are actually short-term responders: A prospective SD-OCT study

Abstract

Purpose

To investigate the inter-individual variability in duration of anti-VEGF (ranibizumab) treatment effect in neovascular age-related macular degeneration (nvAMD).

Methods

Prospective observational study of nvAMD patients treated with ranibizumab and followed weekly for four weeks with spectral-domain optical coherence tomography (SD-OCT) assessing the time to maximal reduction of central retinal thickness (CRT) as well as the presence of intra- and subretinal fluid.

Results

Forty-eight eyes of 48 patients (74.79 ± 8.25 years, 60% female, 52% treatment naive, 35% pseudophakic) were assessed. Two-thirds reached maximal CRT reduction earlier than the standard 4-week interval, 6.25% one-week post-injection, 22.92% two weeks post-injection, and 35.42% three weeks post-injection. Only 35.42% of patients had maximal CRT reduction at 4 weeks. Twenty-one percent of treatment naive and 44% of non-naive patients had a week-4 CRT that was >35 microns thicker than the CRT at the earlier CRT nadir time point. The time to maximal CRT reduction was not related to axial length, age, lens status or prior history of injections.

Conclusions

Optimal dosing interval for maximal CRT reduction may be less than 4 weeks for a significant proportion of patients. Most patients will be classified as complete responders if intervals shorter than 4 weeks are used to assess anti-VEGF treatment response. Disease load rather than eye size appears to be the driver of treatment effect and dosing interval, which needs to be optimized in the cohort of short-term responders.

INTRODUCTION

Anti-vascular endothelial growth factor (VEGF) therapy has had a significant impact on the reduction of visual loss due to vascular leakage and is considered the mainstay of treatment for neovascular age-related macular degeneration (nvAMD)[1,2]. Monthly dosing with the anti-VEGF inhibitor ranibizumab results in a significant visual acuity improvement in over 30% of patients, as supported by the pivotal clinical trials ANCHOR [3,4] and MARINA [5]. Monthly dosing regimen was based mostly on extrapolation from small animal and human pharmacokinetics studies [6–8]. Although pro re nata (PRN) and treat and extent treatment options [9,11] have been used in practice to decrease the burden of monthly injections, most evidence point that monthly treatments are likely the approach to achieve best overall results [11,12]. However, even with monthly administrations there is a substantial proportion of patients with suboptimal or inadequate response, estimated to range anywhere from 10 to 50% depending on the study and type of criteria used [13–18].

A patient can be classified as having an inadequate response to anti-VEGF for nvAMD based on a combination of anatomical and functional criteria. Despite the lack of a global consensus[19] for incomplete response to therapy, persistent retinal edema may imply a reduction of <10% in central retinal thickness (CRT) from baseline, or persistent CRT of more than 300μm [20]. These patients are classified as poor responders or non-responders in a monthly dosing schedule after a 12-month treatment period as some patients may gradually improve during this time interval. The most effective management strategy of these patients and optimal interval dosing is being debated. In the Phase 2 Tolerability and Efficacy study of dose escalation of Ranibizumab, 2- and 4-week intervals were studied (with increasing doses from 0.3mg to 2.0mg from first to last injection) [21]. Two-week interval was found to be tolerated and showed trends for superior outcomes compared to 4-week intervals but for convenience, monthly dosing was eventually tested in Phase 3[22].In contrast, an OCT guided ranibizumab treatment trial of 40 patients in 2007 reported that visual acuity (VA) was the same when examined 2 weeks or 4 weeks after the initial injection and that the presence of intra- or subretinal fluid was not better at week 2 versus week 4 [22]. However, Stewart et al. performed theoretical calculations based on pharmacokinetic data and suggested that optimal dosing interval is less than 4 weeks and likely to be 2–3 weeks [23]. More recently Mimouni et al reported on a retrospective study of 27 patients treated with bi-weekly bevacizumab (not ranibizumab) for refractory nvAMD cases and reported that 20% of the patients would improve in both central macular thickness and total macular volume [24].

In this study, we wanted to assess the time course of ranibizumab treatment effects on retinal fluid by conducting weekly SD-OCT assessments in patients with nvAMD after one 0.5mg of ranibizumab injection in order to determine the time to maximal response post-injection and assess if patients who exhibit poor or no response at week 4, have actually been short-term responders if evaluated at an earlier time point.

METHODS

■ Study Design

This is a prospective observational study. The research protocol was conducted in accordance with Health Insurance Portability and Accountability Act requirements and the tenets of the Declaration of Helsinki. The Institutional Review Boards of the Eye Clinic of the University Hospital of Heraklion, Greece, and the Korgialenio Benakio General hospital, Athens, Greece, approved this study.

■ Study Population and Study Protocol

Subjects from three centers were considered and identified.The purpose of the study was explained to the patients and written informed consent was obtained. For all considered subjects, inclusion criteria were: Age of at least 55 years old with neovascular AMD, adequately clear ocular media for SD-OCT imaging, intra- and/ or sub-retinal fluid on SD-OCT, absence of vitreomacular traction, as well as epiretinal membrane (ERM), and retinal atrophy. Patients were excluded if they had previously been treated with corticosteroids in either eye. Patients with a history of intraocular infections, retinal detachment, uncontrolled intraocular pressure (>21mmHg) or vitrectomized eyes were also excluded. None of the patients had active diabetic retinopathy or active ocular and periocular infection or inflammation. Both treatment naive (52.1%, n=25) as well as patients who had previously received treatment with ranibizumab (47.9%, n=23) for more than one month prior to their enrolment were also recruited. Patients who were previously treated with other anti-VEGF agents, but had not received any injections within 2 months before entry into the study were included as well.

Intravitreal injections were performed using a standard protocol with a dosage of (0.5mg/0.05ml) ranibizumab at the baseline visit. Weekly SD-OCT imaging was obtained using Spectralis SD-OCT (Heidelberg Engineering, Carlsbad, California, USA) for a total period of 4 weeks. (an example of SD-OCT of a patient with nvAMD as documented at baseline and follow up visits at 1–4 weeks post injection is presented in Figure 1)

FIGURE 1:

SD-OCT of two patients with nvAMD. Intraretinal edema with subretinal and intraretinal fluid is observed at baseline. Short-term treatment response is shown at week 1–3. Relapse of the disease is shown at week 4.

Basic demographics, past systemic and ocular history were recorded. Visual acuity (VA) was measured at baseline and at the last follow-up visit. Lens status was determined with the LOCSIII scale and axial length (AL) was measured using optical biometry (IOL Master; Carl Zeiss Meditec).

■ Imaging analysis and data collection

We reviewed weekly OCT images and thickness maps for each subject and recorded CRT and total retinal volume values. Furthermore, grading for qualitatively evaluation of the OCT images for intra- and subretinal fluid was performed by two experienced and independent observers (G.B., L.I.), masked to patients identity and to the stages of intervention. A third grader was recruited in cases of different opinions (3/48) patients.

■ Statistical analysis

Statistical analysis was performed using SPSS for Windows, software version 21 (SPSS, Chicago, IL, USA). The Shapiro-Wilk test was used to examine normal distributions for all variables. Correlations were examined by calculating the Spearman’s correlation coefficient. Distributions of quantitative variables are described as means (±SD). Qualitative variables were summarized by count and percentage. Graphical displays were illustrated using GraphPad Prism (Graphpad Software Inc, San Diego, CA).

RESULTS

There were 48 eyes of 48 patients (19 men and 29 women) with a mean age of 74.79 ± 8.25 years. Of the studied eyes, 31 were phakic and 17 pseudophakic; and mean AL was 23.61 ± 1.15 mm. For the 23 patients that had prior treatments the mean number of prior injections was 8.48 ± 5.46 [range 3 – 25]

Quantitative Results (based on weekly CRT value)

After administration of ranibizumab, there was a rapid decline in CRT values in the first week with the smallest CRT value (“nadir point”) reached on week 3 for treatment naive patients and week 2 for non-naive previously treated patients (Fig. 2A) This was not statistically different. Three patients had the smallest CRT value on the first week (6.25%), eleven patients (22.91%) on the second week, seventeen (35.42%) on the third week and seventeen (35.42%) on the fourth week post-injection (Fig 2B). Overall, a quarter of all patients (22.91%) presented with maximal CRT reduction on SD-OCT within 2-weeks post-injection and almost half of all patients (20/48) had a CRT value at week 4 that was higher when compared to week 3 or week 2(Fig 2B). If we consider changes of 10% of normal thickness as clinical meaningful differences in retinal fluid, 21% of treatment naive and 44% of non-naive patients had a week-4 CRT that was >35 microns thicker than the earlier CRT nadir time point (Table 1). The differences between naïve and prior treated group are not statistically significant.

FIGURE 2:

(A) Left: Line graphs illustrate weekly time course of mean CRT values of nvAMD after anti-VEGF injection. (B) Right: Frequency histogram displaying number of patients reaching CRT_nadir at each week post-injection within a 4-week interval.

Table 1.

Percent of patients with clinically significant retinal fluid on week 4 vs. time point of maximal CRT reduction (= nadir)

	>35 microns CRT4 weeks - CRTnadir	<10 microns CRT4 weeks - CRTnadir
All Patients	13/48 27.1%	22/48 45.8%
Naive Patients	5/25 20.0%	13/25 52.0%
Non-Naive Patients	8/23 34.8%	9/23 39.1%

The time point at which patients presented with the smallest CRT value was quite variable (1–4 weeks) but there were no significant correlations with axial length, baseline CRT, age, or lens status (Table 2). Furthermore, the difference between week 4 CRT values and CRT nadir values (CRT_wk4-CRT_nadir=ΔCRT) did not correlate with axial length, age, lens status, or week ofCRT nadir (Table 2). It must be noted that due to the size of the study, we are underpowered to determine small effects of these variables, nevertheless, these data suggest that disease load and/or clearance of the drug rather than eye volume and intraocular drug concentration are more likely responsible for anti-VEGF effect duration and time to maximal structural response.

Table 2.

Correlation coefficients (r) between patient characteristics and CRT_nadir, ΔCRT_{week 4 CRT -nadir CRT} CRT_baseline

	CRTnadir		ΔCRT		CRTbaseline
	r*	p	r*	p	r*	p
CRTbaseline	−0.126	0.863	0.314	0.03	-	-
Patient Age	0.166	0.266	0.159	0.292	0.328	0.023
Axial Length	0.025	0.869	−0.094	0.534	−0.037	0.801
IOL	0.081	0.589	0.051	0.737	0.094	0.525

^*Spearman’s rank correlation coefficient

Qualitative Results (based on weekly SRF and IRF presence)

Qualitative evaluation of OCT fluid characteristics (IRF, SRF, PED) at weekly time points from baseline until week 4 after an injection of Ranibizumab revealed that the least amount of IRF and SRF was observed earlier than week four, with IRF being least at week 2, while SRF and PED were least at week 3 (Table 3).

Table 3.

Distribution of nvAMD OCT lesion characteristics at weekly time points from baseline until week 4 post-injection with Ranibizumab.

OCT lesion characteristics (n=48)	Baseline	Week 1	Week 2	Week 3	Week 4
Retinal Cysts	32 (66.67%)	15 (31.25%)	10 (20.83%)	12 (25%)	16 (33.33%)
Subretinal Fluid	26 (54.17%)	16 (33.3%)	13 (27.08%)	11 (22.92%)	16 (33.33%)
Pigment Epithelial Detachment	27 (56.25%)	18 (37.5%)	13 (27.08%)	12 (25%)	13 (27.08%)

To assess the proportion of patients that are complete or partial responders at the standard monthly dosing interval vs. the CRT nadir time point, two observers (GB and LI) examined all OCT images and determined the presence of Grade 1 (no fluid), Grade2 (small cystic spaces / subtle subretinal fluid ), Grade 3 (moderate intraretinal and/or subretinal fluid), or Grade 4 (amount of fluid equivalent to baseline or worse) at the corresponding time points. As seen in Figure 3, while 11 patients present with large amounts of retina fluid (grade 3 and 4) on week 4, only six patients demonstrated similar amounts of fluid at the nadir time point.

FIGURE 3:

Frequency histogram of macular edema severity assessed at the time point of maximal edema reduction (nadir point) vs. week 4 after intravitreal ranibizumab injection.

These data taken together suggest that the proportion of poor or non-responders after anti-VEGF treatment is significantly less if optimal timing for assessment is used.

DISCUSSION

Our study suggests that optimal dosing interval for maximal CRT reduction may be less than 4 weeks for a significant proportion of our patients (Table 1 and Fig 2). It also suggests that most patients who are labelled as poor- or non-responders may be responders if assessed at an earlier time point (Figure 1 and 3). Furthermore, these results imply that with current dosing regimen of 4 weeks or more, there is a partial escape of VEGF suppression for many patients which may have implications for progression of neovascular and non-neovascular components of the disease.

Optimal dosing interval for ranibizumab (and other anti-VEGF agents) has not been well studied to date. Pharmacokinetics studies from animals [6, 25] and humans [26,27] suggest that the half-life of ranibizumab is 3 days (with a variable range of 2–4 days). Dosing every 28 days equates to dosing every 7–14 half-lives of this drug. This dosing interval is likely suboptimal unless a very high loading dose is used. The dose escalation Phase 2 study of ranibizumab included 2- and 4-week intervals (with increasing doses from 0.3mg to 2.0mg from first to last injection) and showed a signal for better efficacy at bi-weekly dosing [21]. However, for convenience, monthly dosing was eventually tested in Phase 3[21].Based on available half-life information, Stewart et al. using a mathematical model concluded that the optimal dosing of ranibizumab should be bi-weekly, and provided one clinical case as proof of concept[23].The results of our study are in agreement with the idea that more frequent dosing may lead to improved anatomical treatment results for a large subset of patients. Of course, as Stewart et al. described[23], more frequent injections carry logistical and other problems. Furthermore, based on clinical experience, most ophthalmologists agree that the response to treatment for many of our patients varies over time; likely both due to the evolution of the disease as well as due to the effects of the chronic pharmaceutical management.

It is known that nvAMD is quite variable among individuals and VEGF concentrations are quite variable among individual patients and have been reported to range from 74 pg per milliliter of vitreous humor to 521 pg per milliliter in one recent study[28]. Yet, major studies are performed with the same pharmaceutical dose and dosing interval for all patients. This “one-size fits-all” design of studies is likely due to practical reasons despite the fact that employing a fixed amount of anti-VEGF on a same time-schedule for all will be unable to neutralize equally and to the same extent of time the variable VEGF load in each patient (even not accounting of the variability of drug clearance among patients). Multiple factors are thought to contribute to individual variability [29], however, conclusive reproducible evidence lacks for many of them. Axial length is associated with a larger vitreous volume, resulting in higher distribution volume for intravitreal drugs. Surprisingly however, Krohne et al. found that intraocular drug concentrations are not influenced by the ocular volume[26] and our study failed to see a correlation as well. Lens status has also been investigated and it has been hypothesized that in pseudophakic eyes, diffusion of intravitreal drugs into the anterior chamber is facilitated and clearance can be accelerated from elimination through the Schlemm’s canal. However, Krohne et al. found no relationship of lens status to intraocular drug concentration, and our study also failed to find any association. However, a meta-analysis of the ANCHOR and MARINA studies reported that pseudophakic eyes receiving ranibizumab are at greater risk of visual loss compared to phakic eyes [31].

Although anti-VEGF therapy has had a significant benefit for most of the patients, inadequate response to treatment and treatment failures to treatment have been reported and several theories have been proposed to explain treatment failure. First, VEGF concentrations might exceed the binding capability of the anti-VEGF agents during the 1-month interval between two injections [32]. While it may appear reasonable that the drug can be overwhelmed by the increasing VEGF levels after its clearance from the vitreous, higher doses of ranibizumab (2mg) that will last for a longer duration have been tested and did not result in better final VA outcomes [33]. Other factors that have been hypothesized for affecting anti-VEGF effectiveness and treatment variability may include: inadequate drug diffusion due to variable vitreous liquification; or drug reflux through the scleral injection site; as well as the contribution to vascular leakage from other cytokines. A rarer phenomenon to explain the decrease in the biological response and tolerance to anti-VEGF drugs is tachyphylaxis [34]. It has been suggested that tachyphylaxis may occur as early as after 2 anti-VEGF injections or after 10 or more injections-[35,36].Tachyphylaxis to anti-VEGF therapy has been described at a rate of 10% for bevacizumab [37] and at of 2%–7.7% for ranibizumab [38,39]. Findings imply that suboptimal response to anti-VEGF agents, might be explained due to a short-term response that stays undetected at monthly follow-up visits.

Theoretically, an increase in the injection frequency from every 28 days to every 14 days would result in a significant increase of VEGF binding capacity [23] and our study seems to support this for a subgroup of patients. A recent study [24] which used bevacizumab instead of ranibizumab biweekly reported that this approach was effective in nearly one-quarter of nvAMD “non-responders”. Although individualized dosage and injection frequency to achieve optimal dosing could increase the burden to a subset of patients, it will decrease it for others. The current monthly dosing allows for a see-saw, non-continuous VEGF suppression which may be desired if one is concerned about completely blocking the neurotrophic action of VEGF [40,41]. A recent report by Campochiaro et al[43] may further alleviate concern about chronic VEGF suppression.

The required frequency of injection and interval dosing may also change over time for any individual patient as the disease evolves. In the CATT protocol [42], after the first year of treatment, ranibizumab dosed as needed gave equivalent results compared to monthly injections. Indeed, visual acuity was maintained in the setting of stable, chronic fluid on OCT, and a completely dry macula was not necessarily associated with better visual outcomes. In our study we did not see significant vision changes among short-term responders versus non-short-term responders (short term responders gaining 1.7 less letters than non-short-term responders; statistically not significant; Suppl Table 2). Of-course our study is limited in duration and longer duration studies may uncover differences.

The strengths of our study were that it was prospective and used the same anti-VEGF agent for all patients. Its limitations include small sample size and thus a larger study is needed to confirm our results. Furthermore, our study was unable to examine the relationship between duration of disease and the effective duration of the treatment. Although studying a single drug is a strength, it is also a limitation, and additional studies are needed to compare other available anti-VEGF agents.

In conclusion, we found that the time to maximal effect of ranibizumab in reducing macular thickness in nvAMD is variable and occurs at 2–3 weeks post injection for 59% of patients in our study. Current 4-week interval dosing allows for an anti-VEGF see-saw suppression in a significant proportion of patients. Most patients that appear to not respond at 4 weeks actually have a significant drug response if examined at an earlier time point. Disease load rather than eye size appears to be the driver of dosing interval and effect. Larger and longer prospective studies are needed to determine the inter-individual variability in in dose response and to develop an algorithm for dosing interval for each patient at the onset of treatment and over the course of the disease.

Highlights.

A prospective, weekly SD-OCT study shows that non-responders to ranibizumab anti-VEGF treatment are actually short-term responders. Optimal dosing interval may be less than 4 weeks for a significant proportion of patients.