Differential Response to Anti-VEGF
`Regimens in Age-Related Macular
`Degeneration Patients with Early Persistent
`Retinal Fluid
`
`Glenn J. Jaffe, MD,1 Peter K. Kaiser, MD,2 Desmond Thompson, PhD,3 Andrea Gibson, PhD,3
`Namrata Saroj, OD,3 Robert Vitti, MD,3 Alyson J. Berliner, MD, PhD,3 Jeffrey S. Heier, MD4
`
`Purpose: To compare the effect of intravitreal aflibercept or ranibizumab drug type and frequency on visual
`acuity outcomes in eyes with neovascular age-related macular degeneration (NVAMD) and early persistent retinal
`fluid after 3 initial monthly injections.
`Design: A post hoc analysis of eyes enrolled in VIEW 1 and VIEW 2, 2 similarly designed, randomized, phase
`3 trials.
`Participants: A total of 1815 eyes with NVAMD from VIEW 1 and VIEW 2.
`Methods: Analyses included patients with known fluid status at baseline and weeks 4, 8, and 12 in 3
`treatment groups: ranibizumab 0.5 mg every 4 weeks (Rq4) (n ¼ 595), intravitreal aflibercept injection (IAI) 2 mg
`every 4 weeks (2q4) (n ¼ 613), and IAI 2 mg every 8 weeks (2q8) after 3 monthly injections (n ¼ 607).
`Main Outcome Measures: Mean best-corrected visual acuity (BCVA) change from baseline over weeks 16
`to 52 and the proportion of eyes that gained 15 letters or lost 5 letters were evaluated in eyes with and without
`persistent fluid (cystic intraretinal or subretinal fluid at all 4 initial visits). Visual outcomes also were assessed in
`eyes with persistent fluid by fluid type (intraretinal and subretinal fluid).
`Results: The proportions of eyes with persistent fluid were 29.4%, 18.8%, and 20.3% in the Rq4, 2q4, and
`2q8 groups, respectively. In these eyes, mean BCVA gain from baseline to week 52 was greater with 2q4
`compared with Rq4 (P < 0.01) and 2q8 (P < 0.05), whereas it was similar with Rq4 and 2q8 (P ¼ 0.294). At week
`52, similar proportions of eyes gained 15 letters (31.5%e35.2%), whereas fewer eyes lost 5 letters with 2q4
`compared with Rq4 and 2q8 (6.5% vs. 16.6% and 16.2%). The pattern of visual outcomes was similar regardless
`of fluid type. In eyes without persistent fluid, BCVA changes were similar across treatment groups.
`Conclusions:
`In patients with early persistent fluid, 2q4 may provide additional clinical benefit over 2q8 or
`Rq4. Ophthalmology 2016;123:1856-1864 ª 2016 by the American Academy of Ophthalmology. This is an open
`access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
`
`Aflibercept is a soluble human fusion protein. It acts as a
`decoy receptor
`for vascular endothelial growth factor
`(VEGF) and placental growth factor, and has potent anti-
`VEGF activity.1 In 2 pivotal phase 3 trials, VIEW 1 and
`VIEW 2, in patients with neovascular age-related macular
`degeneration (NVAMD), outcomes were similar for the
`primary and secondary visual acuity end points when 2 mg
`of intravitreal aflibercept injection (IAI) was given every 4
`weeks (2q4) or every 8 weeks (2q8) after 3 initial monthly
`injections, or when 0.5 mg ranibizumab was given every 4
`weeks (Rq4) at week 52.2 These data led to an approval by
`the United States Food and Drug Administration (FDA) in
`2011 of
`IAI
`to treat NVAMD in the United States
`monthly or every 2 months after 3 initial doses.
`In VIEW 1 and VIEW 2, as well as the Comparison of
`Age-related Macular Degeneration Treatment Trials
`(CATT), there was a rapid reduction of macular fluid, as
`measured by optical coherence tomography (OCT), in the
`
`majority of treated eyes during the first several weeks of
`therapy.2,3 However, a subgroup of eyes had residual fluid
`on OCT despite injections with anti-VEGF drugs.2,3 In the
`CATT, at each study visit, eyes with residual intraretinal
`fluid had worse visual acuity than those without intraretinal
`fluid, whereas eyes with subretinal fluid had better visual
`acuity than eyes without fluid in that location. These find-
`ings were particularly prominent when the fluid affected the
`fovea and were independent of drug or treatment regimen.4
`In multiple randomized trials, on average, eyes that are
`given fixed-dosing anti-VEGF treatments had better visual
`acuity at 1 and 2 years than those given less frequent
`treatments.2,3,5e9 Furthermore, when patients were given
`monthly treatment for 1 year and then switched to a pro re nata
`(PRN) regimen, the visual acuity decreased to a level similar to
`that of patients given PRN treatment all along.10 Despite these
`findings, in the United States, the majority of patients with
`NVAMD are given fewer
`than monthly intravitreal
`
`1856
`
`Ó 2016 by the American Academy of Ophthalmology
`This is an open access article under the CC BY-NC-ND license
`(http://creativecommons.org/licenses/by-nc-nd/4.0/). Published by Elsevier Inc.
`
`http://dx.doi.org/10.1016/j.ophtha.2016.05.016
`ISSN 0161-6420/16
`
`Regeneron Exhibit 2018
`Page 01 of 09
`
`

`

`Jaffe et al
`
` Visual Acuity in NVAMD with Persistent Fluid
`
`anti-VEGF injections and are treated using according to a PRN
`or “treat-and-extend” strategy.11 Treatment decisions often are
`driven by the presence of fluid seen on OCT.11 However, the
`factors that determine the needed anti-VEGF treatment fre-
`quency in any individual patient are not well understood.
`Because most patients with NVAMD receive fewer than
`monthly injections based on PRN and treat-and-extend
`dosing approaches,5 it would be desirable to identify during
`initiation of therapy those treatment-naïve patients who will
`benefit most
`if they are given monthly anti-VEGF in-
`jections. However, there are few data available to guide the
`clinician in this regard, even from the primary data from the
`VIEW studies. The design of the VIEW studies afforded us
`the opportunity to study outcomes based on fluid status. We
`began by evaluating the fluid status over time and treatment
`effect on fluid status. This analysis subsequently led us to
`focus on the fluid status in the initial phase of therapy. Data
`from this on-treatment analysis have the potential to provide
`valuable information to the clinician. Specifically, from
`VIEW 1 and VIEW 2, we determined (1) the degree of retinal
`fluid fluctuation, (2) the time to absence of retinal fluid, and
`(3) whether the presence of persistent retinal fluid observed
`after 3 initial monthly intravitreal injections, as seen on time-
`domain OCT, predicted longer-term outcomes and visual
`acuity, and whether these visual acuity changes depended on
`the treatment regimen. We also assessed whether the visual
`acuity outcomes observed throughout the study would differ
`on the basis of fluid type (subretinal or intraretinal). If so, this
`information could help to inform the clinician about injection
`frequency strategy in eyes with NVAMD.
`
`Methods
`
`VIEW 1 and VIEW 2 Study Designs
`
`The VIEW 1 and VIEW 2 studies were 2 similarly designed,
`randomized, double-masked, active-controlled, multicenter, 96-
`week, phase 3 trials comparing the efficacy and safety of IAI
`and ranibizumab in treatment-naïve eyes with NVAMD. The
`design of the VIEW studies has been described.2,12 Each clinical
`site’s
`respective institutional
`review board/ethics committee
`approved the study. All patients provided written informed con-
`sent. Participants were randomized in a 1:1:1:1 ratio to 1 of the
`following 4 regimens for the first 52 weeks: (1) 0.5 mg intravitreal
`ranibizumab every 4 weeks (Rq4), (2) 2 mg IAI every 4 weeks
`(2q4), (3) 0.5 mg IAI every 4 weeks, and (4) 2 mg IAI every 8
`weeks (2q8) after 3 initial monthly injections. From week 52 to
`week 96, patients were treated with the same dose as the first 52
`weeks at
`least every 12 weeks, with monthly evaluations for
`injections.12 Analyses presented in this article were
`additional
`limited to data through week 52 because the dosing regimen in
`all treatment arms changed to PRN from week 52 to week 96,
`and thus this period is not included in our analysis.
`
`Outcome Measures
`
`Independent masked readers at 2 central reading centers, Duke
`Reading Center (VIEW 1) and Vienna Reading Center (VIEW 2),
`determined the presence (termed “wet”) or absence (denoted “dry”)
`of retinal fluid. The OCT image grading was harmonized between
`the Duke and Vienna OCT reading centers. This process included
`conference calls and joint grading exercises on representative
`
`image samples to ensure that the same grading procedure was
`followed at the 2 institutions.
`Retinal fluid was defined as the presence of intraretinal (cystic)
`fluid or subretinal fluid on time-domain OCT images at baseline
`and then monthly through week 52. The time to a single absence
`of retinal fluid and sustained absence of retinal fluid (fluid absent
`on 2 consecutive visits) was determined. Eyes that were not dry
`by week 12 were then further evaluated. Because retinal fluid can
`fluctuate depending on anti-VEGF treatment, we defined an “early
`persistent fluid” group as one that had retinal fluid as defined
`earlier on all initial 4 visits (baseline and weeks 4, 8, and 12). This
`period included all 3 visits during the initial loading dose period
`when all eyes were treated every 4 weeks, as well as 4 weeks after
`the third treatment. Specific baseline characteristics were evaluated
`to determine the influence on persistent fluid status. The mean
`change from baseline best-corrected visual acuity (BCVA), as
`measured by Early Treatment Diabetic Retinopathy Study letters
`over week 52, and the proportion of eyes gaining 15 letters and
`losing 5 letters in eyes with early persistent fluid were compared
`with those without early persistent fluid. In addition, we deter-
`mined whether the fluid type, intraretinal or subretinal, influenced
`visual outcomes for those eyes with and those without early
`persistent fluid.
`
`Statistical Analysis
`
`Of the 2457 eyes randomized into the VIEW studies, 2412 pa-
`tients received study medication and had a baseline and at least 1
`postbaseline BCVA assessment (full analysis set). The 597 eyes
`treated with 0.5 mg IAI every 4 weeks were not included in the
`analysis because IAI 2 mg is the Food and Drug Admin-
`istrationeapproved dose, and the only dose currently available to
`clinicians in the United States. Thus, a total of 1815 eyes (Rq4,
`n ¼ 595; 2q4, n ¼ 613; and 2q8, n ¼ 607) were included in these
`analyses. There were 91 eyes (34 eyes in the Rq4 group, 29 eyes
`in the 2q4 group, and 28 eyes in the 2q8 group) with missing
`observations in at least one visit during baseline to week 12. These
`eyes were classified as wet if the preceding and the following
`visits were classified as wet.
`Time to absence of
`retinal fluid was evaluated by the
`KaplaneMeier method and a proportional hazard analysis. These
`approaches provided a means to estimate the cumulative incidence
`and accounts for eyes that did not have complete follow-up. The
`log-rank test was used to test the difference between the cumula-
`tive incidence curves of the treatment groups. The relative risks (or
`relative hazards) comparing the various IAI treatment groups with
`the Rq4 group were estimated by the proportional hazards anal-
`ysis. Time at risk for each patient is the minimum from the time at
`randomization to the first of any of the following: (1) the date of
`discontinuation, (2) the first episode of dryness, or (3) the end of
`the study. Time at risk is expressed as person-years at risk, and the
`rate is the number of events/person-years at risk. The relative risk
`(or hazard) is determined as the ratio of the rate in the ranibizumab
`group to the rate in the IAI group. The relative risk analyses were
`stratified by study (VIEW 1 vs. VIEW 2). Only within-stratum
`analyses contributed to the overall relative risks. Logistic regres-
`sion was used to determine baseline factors that were useful to
`predict eyes that had persistent retinal fluid. Repeated-measures
`methodology was used to evaluate the differences in the means
`of groups defined by on-treatment variables. On-treatment vari-
`ables are those that are evaluated after randomization. Analysis of
`covariance was used to assess differences in BCVA between the
`treatment groups. The covariates used in the model are baseline
`characteristics listed in Table 1. Least-squares means are reported
`and implicitly corrected for sample size imbalances among the
`treatment groups.
`
`1857
`
`Regeneron Exhibit 2018
`Page 02 of 09
`
`

`

`Ophthalmology Volume 123, Number 9, September 2016
`
`Table 1. Baseline Characteristics of Patients with and without Persistent Retinal Fluid* by Treatment Groups
`
`Patients (n)
`BCVA (ETDRS letters), mean  SD
`CST (mm), mean  SD
`CNV lesion size, n (%)
`10.16 mm2
`>10.16 mm2
`Missing
`CNV lesion type, n (%)
`Occult
`MC
`PC
`Missing
`
`Rq4 (N [ 595)
`Not Persistent
`Persistent
`
`IAI 2q4 (N [ 613)
`Not Persistent
`Persistent
`
`IAI 2q8 (N [ 607)
`Not Persistent
`Persistent
`
`420
`53.6 (13.4)
`289.8 (122.1)
`
`175
`54.5 (13.5)
`311.0 (123.5)
`
`498
`54.1 (13.8)
`292.1 (127.3)
`
`115
`53.3 (12.7)
`325.9 (118.9)
`
`484
`53.9 (13.4)
`292.8 (126.4)
`
`123
`52.4 (13.7)
`357.2 (149.3)
`
`324 (77.1)
`91 (21.7)
`5 (1.2)
`
`163 (38.8)
`145 (34.5)
`106 (25.2)
`6 (1.4)
`
`123 (70.3)
`51 (29.1)
`1 (0.6)
`
`68 (38.9)
`60 (34.3)
`46 (26.3)
`1 (0.6)
`
`373 (74.9)
`124 (24.9)
`1 (0.2)
`
`194 (39.0)
`187 (37.6)
`115 (23.1)
`2 (0.4)
`
`82 (71.3)
`31 (27.0)
`2 (1.7)
`
`39 (33.9)
`30 (26.1)
`44 (38.3)
`2 (1.7)
`
`349 (72.1)
`134 (27.7)
`1 (0.2)
`
`191 (39.5)
`180 (37.2)
`111 (22.9)
`2 (0.4)
`
`99 (80.5)
`23 (18.7)
`1 (0.8)
`
`38 (30.9)
`36 (29.3)
`48 (39.0)
`1 (0.8)
`
`Full analysis set.
`BCVA ¼ best-corrected visual acuity; CST ¼ central subfield thickness; ETDRS ¼ Early Treatment Diabetic Retinopathy Study; IAI ¼ intravitreal
`aflibercept injection; MC ¼ minimally classic; PC ¼ predominantly classic; Rq4 ¼ 0.5 mg intravitreal ranibizumab every 4 weeks; SD ¼ standard deviation;
`2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2 mg every 8 weeks.
`*Persistent fluid was defined as the presence of intraretinal or subretinal fluid on all initial 4 visits (baseline and weeks 4, 8, and 12).
`
`Results
`
`Retinal Fluid Status over Time
`
`Retinal fluid status based on the presence or absence of retinal fluid
`was first determined at each study visit for each treatment group.
`During the 52 weeks of follow-up, the retinal fluid status as defined
`earlier fluctuated in all treatment groups, although greater fluctu-
`ation over time was noted in the 2q8 group. Specifically, the
`average proportions of visits without retinal fluid were 57.4%,
`75.1%, and 57.9% in the Rq4, 2q4, and 2q8 groups, respectively.
`Figure 1 shows the proportion of eyes without retinal fluid by visit
`and treatment assignment.
`
`Time to Absence of Retinal Fluid
`
`Time to retinal fluid absence (“dryness”) was compared among the
`3 treatment groups (Fig 2A). By week 52, there were 512, 571, and
`548 study eyes with retinal fluid absent on at least 1 study visit,
`
`with resultant cumulative incidences of 86.9%, 93.9%, and
`91.9% in the Rq4, 2q4, and 2q8 groups, respectively. The IAI
`groups were more likely than the Rq4 group to have an episode
`of
`absent
`retinal fluid: 1.5 (2q4)
`and 1.4 (2q8)
`times,
`respectively. The overall
`rates (calculated as the number of
`events per 100 person-years at risk of dryness) between the 2q4
`and 2q8 groups were similar. The separation between the IAI and
`Rq4 groups was seen early and was maintained through 52 weeks.
`As expected, the 2q4 and 2q8 cumulative incidence curves were
`almost identical until week 16. Thereafter, there was a slight sep-
`aration between the 2 IAI groups. In both the 2q4 and 2q8 groups,
`more than 50% of the study eyes had the first episode of retinal
`fluid absence by week 4. At week 12, more than 75% of the eyes
`treated with 2q4 or 2q8 were dry on at least 1 visit. The 75%
`cumulative incidence of a dry retina on at least 1 visit for eyes
`treated with Rq4 was reached at week 20.
`The proportion of eyes with sustained dryness, defined as absent
`fluid on at least 2 consecutive visits, and the time to sustained dryness
`were then assessed for each of the 3 treatment groups (Fig 2B). There
`
`Figure 1. Proportion of patients without retinal fluid over 52 weeks of study. IAI ¼ intravitreal aflibercept injection; Rq4 ¼ 0.5 mg intravitreal ranibizumab
`every 4 weeks; 2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2 mg every 8 weeks after 3 initial monthly injections.
`
`1858
`
`Regeneron Exhibit 2018
`Page 03 of 09
`
`

`

`Jaffe et al
`
` Visual Acuity in NVAMD with Persistent Fluid
`
`Figure 2. Cumulative incidence of (A) single and (B) sustained absence of retinal fluid. Sustained absence refers to absent retinal fluid on 2 or more
`consecutive visits. Full analysis set. CI ¼ confidence interval; IAI ¼ intravitreal aflibercept injection; PYR ¼ person-years; R 0.5q4 ¼ 0.5 mg intravitreal
`ranibizumab every 4 weeks; 2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2 mg every 8 weeks.
`
`were 430, 512, and 436 study eyes with retinal fluid absent on 2 or
`more consecutive study visits by week 52 that
`resulted in
`cumulative incidences of 73.7%, 84.8%, and 73.8% in the Rq4,
`2q4, and 2q8 groups, respectively. The rates of the events are
`provided in the table within Fig 2B. On the basis of the relative
`hazard ratio for
`sustained dryness, eyes
`treated 2q4 were
`approximately 1.5 times more likely than those treated Rq4 and
`2q8 to achieve sustained dryness. During the first 12 weeks, the
`cumulative incidence of sustained dryness in the 2q4 and 2q8
`groups was similar, but from week 16 onward, the cumulative
`incidence of sustained dryness in the 2q8 group was lower than
`that observed in the 2q4 group. The separation for the cumulative
`incidence curves between 2q4 and Rq4 was seen early, at
`approximately week 4. The 50th percentile for the cumulative
`incidence of sustained dryness was achieved at week 8 in both of
`the IAI groups. The 75th percentile was achieved at week 24 for
`the 2q4 group but was never achieved for the 2q8 or Rq4 groups.
`
`Evaluation of Early Persistent Fluid
`
`We next determined how frequently retinal fluid persisted over
`each of the first 4 study visits (including baseline) through week
`
`12 (termed “early persistent fluid”) and whether the specific anti-
`VEGF treatment affected fluid persistence. For this analysis, eyes
`were determined to have early persistent fluid if the presence of
`fluid was confirmed at all initial 4 visits: baseline, week 4, week 8,
`and week 12. Those that did not meet this criterion were deter-
`mined not to have early persistent fluid. Most eyes (1402 [77.2%])
`did not have early persistent fluid over the first 4 visits through
`week 12: 420 eyes (70.6%), 498 eyes (81.2%), and 484 eyes
`(79.7%) in the Rq4, 2q4, and 2q8 treatment groups, respectively.
`Overall, 413 eyes (22.8%) had early persistent retinal fluid during
`the first 4 visits. The proportion of eyes with early persistent fluid
`was similar in the IAI arms (2q4:18.8%, 115/613; 2q8: 20.3%,
`123/607; combined 19.5%, 238/1220), as expected since both
`were dosed monthly during the loading phase, while eyes treated
`with Rq4 were 51% more likely (95% confidence interval [CI]:
`27%, 78%) to have early persistent fluid (Rq4: 29.4%; 175/595).
`Eyes with and without early persistent fluid through week 12 had
`similar baseline visual acuity, lesion size, and lesion type across
`treatment groups (Table 1). However, eyes with early persistent
`fluid had greater baseline central subfield thickness on OCT
`across treatment groups compared with eyes without early
`persistent fluid.
`
`1859
`
`Regeneron Exhibit 2018
`Page 04 of 09
`
`

`

`Ophthalmology Volume 123, Number 9, September 2016
`
`Figure 3. Least square mean change in best-corrected visual acuity (BCVA) from baseline through week 52 in eyes (A) without and (B) with persistent
`retinal fluid through week 12. ETDRS ¼ Early Treatment Diabetic Retinopathy Study; IAI ¼ intravitreal aflibercept injection; Rq4 ¼ 0.5 mg intravitreal
`ranibizumab every 4 weeks; 2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2 mg every 8 weeks.
`
`Relationships between Early Persistent Fluid
`and Visual Acuity
`
`For eyes without early persistent retinal fluid (from baseline
`through week 12), there were no differences among the treatment
`groups in mean BCVA change from baseline over the interval
`beginning at week 16 and spanning through week 52 (Fig 3A). An
`examination of the means and error bars at each visit suggests no
`differences between the treatment groups at any visit.
`In contrast, in eyes with early persistent fluid from baseline
`through week 12, the mean BCVA gains from baseline over the
`interval spanning weeks 16 to 52 were observed as early as week
`16 for the 2q4 group and were consistently greater than those for
`the 2q8 and Rq4 groups. An analysis of covariance using cova-
`riates from Table 1 confirms and quantifies these suggestions at
`week 52. A test of
`the equality of
`the means among the
`treatment groups was rejected (P ¼ 0.006). Pairwise treatment
`group comparisons yield the following: 2q4 versus 2q8 and 2q4
`versus Rq4 are P ¼ 0.006 and P ¼ 0.049, respectively, and the
`P value for Rq4 versus 2Q8 is 0.294. The adjusted mean
`
`changes from baseline at week 52 were 11.7, 8.5, and 7.5 letters
`for the 2q4, Rq4, and 2q8 groups, respectively (Fig 4). These
`mean changes correspond to difference of 3.2 letters for 2q4
`versus Rq4, 4.2 letters for 2q4 versus 2q8 and 1.0 letters for Rq4
`versus 2q8.
`At week 52, the percentages of study eyes that gained 15
`letters were similar among the 3 study groups: 33.7% (95% CI,
`26.5e41.0), 35.2% (95% CI, 26.2e44.2), and 31.5% (95% CI,
`22.9e40.2)
`for Rq4, 2q4, and 2q8,
`respectively (Fig 5A).
`However, a lower percentage of eyes in the 2q4 group lost 5
`letters compared with eyes in the Rq4 and 2q8 groups (6.5%
`[95% CI, 1.8e11.1] vs. 16.6% [95% CI, 10.9e22.3] and 16.2%
`[95% CI, 9.4e23.1]) (Fig 5B).
`The described analysis in eyes with early persistent fluid
`through week 12 was repeated separately for persistent fluid sub-
`types: 52 eyes (8.7%), 43 eyes (7%), and 45 eyes (7.4%) with early
`persistent intraretinal fluid, and 113 eyes (19.0%), 57 eyes (9.3%),
`and 76 eyes (12.5%) with early persistent subretinal fluid in the
`Rq4, 2q4, and 2q8 treatment groups, respectively. However, the
`numbers of eyes available for these analyses were small. In
`
`1860
`
`Regeneron Exhibit 2018
`Page 05 of 09
`
`

`

`Jaffe et al
`
` Visual Acuity in NVAMD with Persistent Fluid
`
`Figure 4. Adjusted mean changes in best-corrected visual acuity (BCVA)
`from baseline at week 52 in eyes with persistent retinal, intraretinal, or
`subretinal fluid through week 12. Rq4 ¼ 0.5 mg intravitreal ranibizumab
`every 4 weeks; 2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2 mg every 8 weeks.
`
`general, the pattern of visual acuity changes from week 16 to week
`52 was similar regardless of whether the early persistent fluid was
`intraretinal or subretinal.
`At week 52, the percentages of study eyes with early persistent
`intraretinal fluid that gained 15 letters were 30.0% (95% CI,
`17.3e42.7), 31.0% (95% CI, 17.0e44.9), and 29.3% (95% CI,
`15.3e43.2) in the Rq4, 2q4, and 2q8 treatment groups, respec-
`tively (Fig 5A), whereas the percentages of eyes with early
`persistent intraretinal fluid that lost 5 letters were 16.0% (95%
`CI, 5.8e26.2), 4.8% (95% CI, 0.0e11.2), and 22% (95% CI,
`9.3e34.6), respectively (Fig 5B). In these eyes, at week 52, the
`mean changes in BCVA from baseline were 6.4, 11.8, and 7.3
`letters for Rq4, 2q4, and 2q8, respectively (Fig 6A).
`More eyes with persistent subretinal fluid treated with 2q4 gained
`15 letters, and fewer lost5 letters of visual acuity than eyes treated
`with Rq4 or 2q8 (Figs 5A and B). The proportions of eyes that gained
`15 letters were 34.9% (95% CI, 25.8e44.0), 40.7% (95% CI,
`27.6e53.9), and 25.8% (95% CI, 15.2e36.3), whereas
`the
`proportions of eyes that lost 5 letters were 20.8% (95% CI,
`13.0e28.5), 9.3% (95% CI, 1.5e17.0), and 15.2% (95% CI,
`6.5e23.8) for Rq4, 2q4, and 2q8, respectively (Fig 5A and B). In
`this group, the mean BCVA changes from baseline at week 52 were
`8.3, 11.9, and 5.5 letters for Rq4, 2q4, and 2q8, respectively (Fig 6B).
`
`Discussion
`
`To our knowledge, this report is the first to evaluate the
`proportion of eyes with early persistent fluid after 3 anti-
`VEGF “loading” injections and the effect of this fluid on
`subsequent visual acuity. In the pivotal phase 3 VIEW 1 and
`VIEW 2 trials, overall, 2 mg IAI monthly and every 2
`months (after 3 initial injections) were clinically equivalent
`to ranibizumab monthly for the primary end point of mean
`change in visual acuity from baseline at week 52 in patients
`with NVAMD.2 However, fluid status, particularly early
`persistent fluid, and its effect on visual outcomes were not
`explored in the primary analyses. Likewise, other large
`randomized anti-VEGF treatment
`trials, such as CATT
`and Inhibition of VEGF in Age-related choroidal Neo-
`vasularisation (IVAN), showed that ranibizumab and bev-
`acizumab had similar visual acuity outcomes when the
`totality of the data were considered, but they did not report
`the proportion of eyes with persistent fluid or the effect of
`persistent fluid on subsequent visual acuity.3,6 Our data
`
`Figure 5. Proportion (95% confidence interval) of eyes with persistent
`retinal, intraretinal, or subretinal fluid through week 12 that (A) gained
`15 letters or (B) lost 5 letters from baseline at week 52. Rq4 ¼ 0.5 mg
`intravitreal ranibizumab every 4 weeks; 2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2
`mg every 8 weeks.
`
`show that although the proportion of eyes with retinal or
`subretinal fluid fluctuated during the first year of therapy,
`most eyes did not have early persistent fluid, and for this
`group of eyes, IAI given every 4 or 8 weeks, or ranibizumab
`every 4 weeks, had a similar effect on visual acuity,
`consistent with the overall results reported in the VIEW 1
`and VIEW 2 studies.2 However, when early persistent fluid
`was present after the initial 3 injections (a finding present in
`approximately 20% of eyes initially treated with IAI and in
`30% of eyes with Rq4), there may be a benefit to monthly
`IAI compared with the other regimens as demonstrated by
`a higher proportion of dry retinas, a greater visual acuity
`improvement, and a smaller proportion with visual acuity
`loss in eyes treated with 2q4 compared with eyes treated
`with Rq4 or 2q8 at week 52.
`The reason for better visual acuity outcomes in the IAI
`2q4 treatment group among eyes with persistent fluid is
`unknown. In the CATT, continuous monthly injections of
`bevacizumab and ranibizumab resulted in slightly better
`BCVA than PRN injections at years 1 and 2.3,10 We
`hypothesize that treatment-naïve eyes, in which the fluid
`persists after initial treatment, may be less sensitive to the
`
`1861
`
`Regeneron Exhibit 2018
`Page 06 of 09
`
`

`

`Ophthalmology Volume 123, Number 9, September 2016
`
`Figure 6. Mean change in best-corrected visual acuity (BCVA) from baseline through week 52 in eyes with (A) persistent intraretinal fluid or (B) persistent
`subretinal fluid through week 12. IAI ¼ intravitreal aflibercept injection; ETDRS ¼ Early Treatment Diabetic Retinopathy Study; Rq4 ¼ 0.5 mg intravitreal
`ranibizumab every 4 weeks; 2q4 ¼ 2 mg every 4 weeks; 2q8 ¼ 2 mg every 8 weeks.
`
`anti-VEGF antipermeability effects and require continuous
`anti-VEGF treatment to avoid the adverse effects of fluid on
`photoreceptors and other neurosensory retinal structures.
`We were unable to assess the effect of early persistent
`fluid on retinal microanatomy because lower-resolution
`time-domain OCT was used in this study. However, in
`ongoing or future anti-VEGF NVAMD treatment trials in
`which spectral-domain OCT is used, it will be of interest to
`evaluate the effect of persistent fluid on inner and outer
`retinal structures, such as the ganglion cell complex, inner
`nuclear layer, external limiting membrane, ellipsoid zone,
`and photoreceptor outer segments.
`In previous anti-VEGF NVAMD interventional studies,
`the specific type of fluid had differing effects on visual
`acuity. Indeed, in the CATT, eyes with intraretinal fluid had
`worse visual acuity, whereas eyes with subretinal fluid had
`better visual acuity than those without these types of fluid.3
`These data led us to speculate that the effects of anti-VEGF
`
`therapy in eyes with early persistent fluid after initial
`loading doses might differ depending on whether the fluid
`was intraretinal fluid or subretinal fluid. Of note, we found
`that the type of fluid was not important in this regard; eyes
`with early persistent intraretinal fluid benefited from anti-
`VEGF therapy to the same degree as those with subretinal
`fluid, and eyes with early persistent fluid, regardless of fluid
`type, benefited more from IAI 2q4 compared with IAI 2q8
`or monthly ranibizumab. In a different analysis of VIEW 1
`and VIEW 2 data, eyes with baseline intraretinal fluid had
`worse visual acuity at week 52 than those without intra-
`retinal fluid.13 The CATT data and those reported by
`Waldstein et al13 are not inconsistent with the findings in
`the present report because the study methodology and
`main end points were different. In the CATT, the subtype
`of fluid was correlated with visual acuity at each study
`visit, and in the study by Waldstein et al,13 the analysis
`was conducted to determine how the baseline morphology
`
`1862
`
`Regeneron Exhibit 2018
`Page 07 of 09
`
`

`

`Jaffe et al
`
` Visual Acuity in NVAMD with Persistent Fluid
`
`characteristics (that included intraretinal fluid) affected the
`week 52 visual acuity. In contrast, in this report, the goal
`was to determine the effect of drug type and treatment
`frequency in eyes with or without persistent fluid over the
`first 4 visits on subsequent visual acuity outcomes.
`We found that persistent subretinal fluid did not prevent
`15 letter gains in any of the 3 treatment groups, but was
`associated with a 5 letter loss in the Rq4 and 2q 8 cohorts.
`The reason for this finding is unclear. Regardless, the overlap
`within the wide CIs across treatment groups with such a small
`sample size in this subgroup makes it difficult to definitively
`determine whether these results are clinically meaningful.
`When there was early persistent fluid, eyes treated with
`monthly aflibercept had better visual acuity at week 52 than
`those treated with aflibercept less frequently or those treated
`with ranibizumab. The differences in visual acuity among the
`treatment groups were seen early, at week 16, the first visit after
`the defined period of persistent fluid, and were maintained
`through week 52. This observation is consistent with the
`hypothesis that visual acuity differences among treatment
`groups at week 52 reflect fluid status during the initial early
`treatment period.
`It might be tempting to speculate that elimination of
`subretinal fluid after this initial early period would have a
`beneficial effect on visual acuity. However, this study was
`not designed to determine the effect of subretinal fluid res-
`olution on visual acuity after the first 12 weeks. Further-
`more, although the CATT, VIEW 1, and VIEW 2 have
`recently shown that subretinal fluid did not adversely affect
`visual acuity at individual study visits,4,14 those studies also
`were not designed to study the effect of subretinal fluid
`resolution on visual acuity. Future studies designed to
`determine how aggressively to treat subretinal fluid when it
`is present after the first 12 weeks are warranted and would
`help to answer this question. Overall, eyes treated with both
`monthly and every 2 months IAI were less likely to have
`persistent fluid through week 12 and achieved a dry retina
`earlier and in a higher proportion of patients compared with
`ranibizumab. Furthermore, in eyes treated with monthly IAI,
`a dry retina was more likely to be sustained than in those
`treated with IAI every 2 months or ranibizumab every
`month. Together, these data indicate that IAI, at the doses
`used in this study, has a more pronounced antipermeability
`effect than ranibizumab in eyes with NVAMD. This anti-
`permeability effect has potential clinical
`implications
`because eyes treated with IAI were less likely to have
`persistent fluid through week 12; therefore, theoretically, in
`these patients without persistent fluid, a higher percentage of
`IAI-treated eyes could be managed with therapy every 2
`months when compared with those treated with ranibizu-
`mab. As described next, this hypothesis would need to be
`tested in a prospective trial.
`This study has several strengths. Data were obtained from 2
`large, randomized, masked trials. Accordingly, there was
`adequate power to determine the effects of persistent fluid on
`visual acuity. Furthermore, bias was minimized during data
`collection because trained readers masked to treatment group
`analyzed images at 2 independent reading centers, and masked
`visual
`acuity examiners performed the visual
`acuity
`measurements.
`
`Study Limitations
`
`Our analysis has limitations. The data were analyzed post
`hoc, and the original studies were not designed to determine,
`in a prospective manner, whether eyes with early persistent
`fluid after IAI treatment should be maintained on monthly
`IAI or whether eyes with persistent fluid after ranibizumab
`should be switched to monthly IAI. Adjustments for mul-
`tiplicity were not made because these analyses were not
`prespecified. Thus, there is potential for type I and II errors.
`Furtherm