- Email: [email protected]
Simulating an Anti–Vascular Endothelial Growth Factor Switch in Neovascular Age-Related Macular Degeneration
Simulating an AntieVascular Endothelial Growth Factor Switch in Neovascular Age-Related Macular Degeneration A HARBOR Subanalysis Marco Zarbin, MD, PhD,1 Min Tsuboi, PharmD,2 Lauren F. Hill, MS,2 Ivaylo Stoilov, MD2 Purpose: A simulated switching study assessed the effects of continuing the same antievascular endothelial growth factor (VEGF) treatment among patients who typically are considered for a therapy switch. Post hoc analysis of data from HARBOR was undertaken. Patients with neovascular age-related macular degeneration who demonstrated a suboptimal response after 3 or 6 months of ranibizumab treatment were identified as switching candidates. Rather than switching, however, patients continued on ranibizumab treatment, and visual and anatomic outcomes from the point of the hypothetical switch were examined. Design: Post hoc analysis of the phase 3 HARBOR clinical trial. Participants: Patients were included in 3- and 6-month switcher analyses if they received 3 of 3 initial monthly ranibizumab doses and 5 of 6 initial monthly ranibizumab doses, respectively, and met all the following: 5-letter or fewer gain from baseline, best-corrected visual acuity (BCVA) 20/40 or worse, and intraretinal or subretinal fluid with central foveal thickness (CFT) equal to or greater than central subfield thickness. Methods: Patient data were examined at months 3 and 6 to identify those who met predetermined switching criteria. Best-corrected visual acuity and CFT were examined from the point at which switching criteria were met through months 6, 12, 18, and 24 of HARBOR and compared with those who did not meet the criteria. Main Outcome Measures: Outcome measures included mean BCVA and CFT change over time from the point (month 3 or 6) at which switching criteria were met. Results: By months 3 and 6, only 44 of 1059 patients (4.2%) and 37 of 769 patients (4.8%), respectively, met the inclusion criteria for hypothetical switching. Patients who met switching criteria at month 3 gained, on average, 5.3 letters from months 3 to 12 and 2.7 letters from months 3 to 24. Month 6 switchers gained, on average, 1.6 letters from months 6 to 12 and 1.8 letters from months 6 to 24. Both groups experienced significant CFT reductions over 24 months. Conclusions: Month 3 hypothetical switchers achieved vision and anatomic improvement while continuing their original ranibizumab treatment. Month 6 switcher outcomes replicated those commonly reported in published anti-VEGF switching studies: stable vision or nominal improvements in vision with continued substantial anatomic improvement. Ophthalmology 2019;-:1e7 ª 2019 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/).
Age-related macular degeneration (AMD) is one of the leading causes of blindness in the United States among persons 50 years of age or older.1,2 Vision loss in AMD can occur as the result of vascular endothelial growth factor (VEGF)-mediated growth of new choroidal vessels and vascular leakage affecting the macula.2 The current standard of care for treating neovascular AMD (nAMD) involves blocking VEGF signaling using anti-VEGF agents.3 Several anti-VEGF agents currently are used for the treatment of nAMD in the United States, including ranibizumab (Lucentis; Genentech, Inc., South San Francisco, CA, and Novartis Pharma, Basel, Switzerland), aflibercept (Eylea; Regeneron Pharmaceuticals, Inc., Tarrytown, NY),4e7 and off-label bevacizumab (Avastin; Genentech, Inc., South San Francisco, CA).8,9 ª 2019 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.
Most patients with nAMD achieve clinically meaningful visual benefits after anti-VEGF therapy, as demonstrated by a marked reduction in the prevalence of blindness resulting from nAMD since the approval of the first anti-VEGF treatment more than 10 years ago.10 However, a small proportion of patients do not show an early or satisfactory response to intravitreal anti-VEGF therapy. This suboptimal response may be the result of a number of factors, including misdiagnosis, inadequate treatment, or development of resistance to the therapeutic method. When patients with nAMD fail to show significant clinical improvement within the first 3 to 6 months of initiating anti-VEGF treatment, they often are switched to another anti-VEGF agent.11,12 However, the benefits of such a switch are not well established in randomized controlled trials. https://doi.org/10.1016/j.ophtha.2019.01.003 ISSN 0161-6420/19
1
Ophthalmology Volume -, Number -, Month 2019 Meta-analyses examining numerous anti-VEGF switching studies have shown that most patients with nAMD who switch treatment from either ranibizumab or off-label bevacizumab to aflibercept demonstrate continued anatomic improvement with vision stabilization or limited visual improvement.9,13 Although these benefits generally are attributed to the new anti-VEGF agent, limitations in study design preclude definitive conclusions regarding the cause of the improvement. Most of these studies do not include a comparator arm, and as a result, it is unclear if the benefits achieved are the result of the new anti-VEGF drug, the result of continuing anti-VEGF treatment for a longer duration, or something else. Another issue to consider when assessing the results of switching studies is the potential impact of confounding factors. Uncontrolled studies of the treatment of diseases that have a variable natural course, such as nAMD, are susceptible to the confounding effects of regression toward the mean and sampling bias. Regression toward the mean describes the phenomenon in which, for example, if an individual central subfield thickness (CST) value is extreme when measured on a particular occasion, subsequent CST measurements will tend to be closer to the average CST value as measured for that patient on previous and subsequent occasions. Sampling bias occurs when the patients enrolled in a study are not representative of the entire population targeted for treatment. Figure 1 illustrates 2 hypothetical patients who are receiving treatment for nAMD. Although the mean CSTs of the 2 patients are similar, patient 1 exhibits a more variable response to monthly injections than patient 2. As a result, assessment of patient 1 at 6 months when the CST is relatively high (indicated by the arrow in the figure) may result in a decision to switch treatment to a different anti-VEGF agent. However, as illustrated in Figure 1, even without switching, the subsequent CST measurement at month 9 is lower because of regression toward the mean. If patient 1 had switched treatment at month 6, the subsequent improvement in CST probably would have been attributed to the switch. Thus, regression toward the mean can lead to an incorrect interpretation of the consequences of a therapeutic intervention.14 Patients similar to patient 1 are more likely to be enrolled in switching studies than those similar to patient 2, which may contribute to sampling bias in uncontrolled switching studies, because these patients may not be representative of the larger population. These considerations highlight the need for a control group in switching studies if the effect of switching is to be assessed accurately. Without a control group, interpretation of the results of uncontrolled switching studies can be compromised by sampling bias and overestimation of the magnitude of a treatment effect induced by the switch. The present post hoc analysis of patients from the 24-month HARBOR AMD trial simulates a switching study by identifying patients who met a set of predefined switching criteria, hypothetically switching them to a different anti-VEGF agent, and assessing the changes in vision and anatomic features after the fictitious switch date. Because patients were not actually switched, but remained treated with ranibizumab throughout
2
the study, this analysis examines the effect of continuing the original anti-VEGF treatment in patients who typically would have been switched to a different agent.
Methods Patient Selection The present analysis examined the 1097 treatment-naive patients with nAMD and active subfoveal choroidal neovascularization who were randomized in the 24-month phase 3 HARBOR trial (clinicaltrials.gov identifier, NCT00891735). Patients received ranibizumab 0.5-mg or 2.0-mg intravitreal injections either monthly or as needed (pro re nata [PRN]) after 3 consecutive monthly loading doses. The HARBOR clinical trial was conducted in compliance with the tenets of the Declaration of Helsinki. The independent ethics committee or institutional review boards at each site approved the HARBOR study protocol, and all patients provided written informed consent before enrollment, which extended to the use of individual patient data for further analyses.
Simulated Switching Criteria Patient data were examined at months 3 and 6 to determine whether they met all the hypothetical switching criteria. To be eligible for a switch at month 3, patients must have received all 3 monthly ranibizumab loading doses. To meet the switching criteria at month 6, patients must have received at least 5 monthly doses in the first 6 months of the study. At either time point, switching criteria included gaining 5 letters or fewer from baseline, having a best-corrected visual acuity (BCVA) of 20/40 or worse, and presence of either intraretinal or subretinal fluid, with central foveal thickness (CFT) greater than or equal to CST. It is important to note that despite meeting all the predefined switching criteria, none of these patients were switched from their original treatment with ranibizumab.
Outcome Measures Outcome measures were quantified as change from the point of the hypothetical switch (i.e., month 3 or month 6) to months 6, 12, 18, and 24 for month 3 switchers and to months 12, 18, and 24 for month 6 switchers. The primary visual outcome measure of this study was mean change in BCVA over time. The primary anatomic outcome measure was mean change in CFT over time. Patients in the ranibizumab 0.5-mg or 2.0-mg monthly and PRN treatment arms were pooled for all analyses.
Statistical Analysis Descriptive statistics using observed data were used to evaluate changes in BCVA and CFT outcomes over time for each hypothetical switching group. All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC).
Results Of the 1097 patients who were randomized in HARBOR, 1059 patients received 3 of 3 loading doses by month 3. Only 44 of these 1059 patients (4.2%) met the additional visual and anatomic hypothetical switching criteria at month 3. Similarly, 769 of 1097 patients had received at least 5 of 6 initial monthly injections by month 6, and 37 of 769 patients (4.8%) met the additional visual and anatomic criteria at month 6 for inclusion in this analysis (Fig 2).
350
HARBOR Simulated Switching
10
More likely to select patient 1 for switching (sampling bias), and patient 1 more likely to show improvement without a switch (regression to the mean)
300
Mean = 249 μm
CST, μm
250
200 Mean = 244 μm 150
100
Mean Change in BCVA from Month 3, ETDRS Letters
Zarbin et al
5.3
4.6 2.7
0
–5 50
5.2
5
3
6
12
18
24
40
41
Month n = 44
0 0
3
6
9
12
15
18
Month
Figure 1. Graph showing that patient 1 is most likely to be selected for switching because of sampling bias. However, patient 1 can show improvement without a switch (because of regression toward the mean), which may result in attributing the change in clinical status to a change in treatment, when the change in status would have occurred without the change in treatment. CST ¼ central subfield thickness.
Vision Outcomes On average, patients who met the switching criteria at month 3 (n ¼ 44) continued to demonstrate vision gains through the remaining 21 months of the study, despite continuing their original treatment regimen with ranibizumab (Fig 3). Patients gained, on average, 5.2, 5.3, 4.6, and 2.7 Early Treatment Diabetic Retinopathy Study (ETDRS) letters from month 3 to months 6, 12, 18, and 24, respectively. Patients who were fictitiously switched at month 6 (n ¼ 37) demonstrated minimal improvement in vision over the remainder of HARBOR (Fig 4), on average gaining 1.6, 0.2, and 1.8 ETDRS letters from month 6 to months 12, 18, and 24, respectively.
Anatomic Outcomes Patients who met the hypothetical switching criteria at month 3 (n ¼ 44) demonstrated continued reduction in CFT over the remainder of the trial. From month 3, mean reductions in CFT of 77.5 mm,
Initial monthly treatment: • Month 3 switch: 3/3 initial monthly doses by month 3 • Month 6 switch: ≥5/6 initial monthly doses by month 6 Hypothetical switch criteria: • ≤5-letter gain from baseline • Visual acuity 20/40 or worse • Intraretinal or subretinal fluid with CFT ≥ CST Outcome measures: • Visual: Mean change in BCVA • Anatomic: Mean change in CFT • All changes were from the month 3 or month 6 visit when criteria for switching were met
44
42
Figure 3. Graph showing best-corrected visual acuity (BCVA) changes for month 3 switching candidates. Patients meeting the month 3 switching criteria demonstrated modest BCVA gains from month 3. These vision gains were maintained through completion of the trial at month 24. Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
92.5 mm, 93.5 mm, and 111.7 mm were observed at months 6, 12, 18, and 24, respectively (Fig 5). A similar trend was observed in patients who met the switching criteria at month 6. From the point at which the criteria were met, patients demonstrated a mean reduction in CFT of 99.6 mm, 112.3 mm, and 136.2 mm at months 12, 18, and 24, respectively (Fig 6).
Sensitivity Analysis An additional analysis examined these patients under a less stringent set of switching criteria, in which patients were required to meet only the following criteria: (1) received all 3 monthly ranibizumab injections by month 3 (3-month switchers) or at least 5 monthly injections by month 6 (6-month switchers), (2) demonstrated either intraretinal or subretinal fluid, with CFT greater than or equal to CST, and (3) gained 5 letters or fewer from baseline or demonstrated BCVA of 20/40 or worse. The difference in this set of criteria compared with those used in the primary analysis was that rather than having to meet both the criteria for 5 letters or fewer gained and for BCVA of 20/40 or worse, patients were eligible for switching if they
HARBOR: treatment-naive patients with neovascular AMD and active subfoveal CNV; ranibizumab monthly or PRN for 2 years (N = 1097)
Month 3 switch: 1059 eligible patients
Month 6 switch: 769 eligible patients
Month 3: 44 (4.2%) patients met switch criteria
Month 6: 37 (4.8%) patients met switch criteria
Patients identified did not switch treatments, but remained on their initial treatment
Figure 2. Flow diagram showing study design and patient selection. AMD ¼ age-related macular degeneration; BCVA ¼ best-corrected visual acuity; CFT ¼ central foveal thickness; CNV ¼ choroidal neovascularization; CST ¼ central subfield thickness; PRN ¼ pro re nata.
3
Ophthalmology Volume -, Number -, Month 2019 25
5 1.6
1.8 0.2
0
Mean Change in CFT from Month 6, μm
Mean Change in BCVA from Month 6, ETDRS Letters
10
–75 –99.6
–100
–112.3
–125
–136.2
–150
–200
6
12
18
24
34
33
Figure 4. Graph showing best-corrected visual acuity (BCVA) changes for month 6 switching candidates. Patients meeting the month 6 switching criteria demonstrated BCVA stability from month 6 through the conclusion of the trial at month 24. Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
met only 1 of these in addition to those previously outlined in the primary analysis. At month 3, 85 patients (8.0%) met these less stringent criteria, whereas 66 patients (8.6%) met them at month 6. In this sensitivity analysis, month 3 switchers demonstrated minimal vision improvement from the time the switching criteria were met through the remainder of the trial, with a mean gain of 1.2 letters from months 3 to 24 (Fig 7A). A similar trend was observed in the month 6 switchers, with a mean gain from months 6 to 24 of 0.8 letters (Fig 7B). Both groups continued to experience steady improvements in CFT over the remainder of the 24-month trial after the switch, with CFT reductions of 110.3 mm (Fig 8A) and 130.9 mm (Fig 8B) for the 3- and 6-month switchers, respectively.
Discussion HARBOR reported the 24-month outcomes of the 1097 patients randomized to receive ranibizumab 0.5-mg or 2.0-mg injections monthly or PRN after 3 consecutive loading doses. At month 24, patients in the ranibizumab 0.5-mg PRN, 0.5-mg 25 0 –25 –50 –77.5
–75
–92.5
–100
–93.5 –111.7
–125
–175 –200
24
37
34
34
33
A
10
5
0
–5
3
6
12
18
24
18
24
Month
B
25 0 –25 –50 –75 –100 –125 –150 –175 3
6
12 Month
3
6
12
44
44
42
18
24
40
41
Month n=
18
monthly, 2.0-mg PRN, and 2.0-mg monthly treatment arms gained a mean of 7.9, 9.1, 7.6, and 8.0 letters from baseline, respectively.15 Although on average, these patients achieved clinically meaningful vision gains in the clinical trial setting, it is important to note that in practice, individual patients may exhibit a less robust response to treatment. This is
–200
–150
12
Figure 6. Graph showing central foveal thickness (CFT) changes for month 6 switching candidates. Patients meeting the month 6 switching criteria demonstrated steady CFT improvement from month 6 through the completion of the trial at month 24. Error bars represent 95% confidence intervals.
Mean Change in BCVA from Month 3, ETDRS Letters
35
n=
Mean Change in CFT from Month 3, μm
n = 37
6
Month
Month
Mean Change in CFT from Month 3, μm
–50
–175
–5
Figure 5. Graph showing central foveal thickness (CFT) changes for month 3 switching candidates. Patients meeting the month 3 switching criteria demonstrated steady CFT improvement from month 3 through the completion of the trial at month 24. Error bars represent 95% confidence intervals.
4
0 –25
Figure 7. A, Graph showing mean best-corrected visual acuity (BCVA) change from month 3 using less-stringent switching criteria. B, Graph showing mean central foveal thickness (CFT) change from month 3 using less stringent switching criteria. Patients who met the less stringent switching criteria at month 3 demonstrated BCVA and CFT improvements. These gains were comparable with those of patients who met the complete switching criteria used in the standard analysis (Figs 3 and 5, respectively). Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
Zarbin et al
A
HARBOR Simulated Switching
Mean Change in BCVA from Month 6, ETDRS Letters
10
5
0
–5
6
12
18
24
18
24
Month
Mean Change in CFT from Month 6, μm
B
25 0 –25 –50 –75
–100 –125 –150 –175 –200
6
12 Month
Figure 8. A, Graph showing mean best-corrected visual acuity (BCVA) change from month 6 using less stringent switching criteria. B, Graph showing mean central foveal thickness (CFT) change from month 6 using less stringent switching criteria. Patients who met the less stringent switching criteria at month 6 demonstrated BCVA and CFT improvements. These gains were comparable with those of patients who met the complete switching criteria used in the standard analysis (Figs 4 and 6, respectively). Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
reflected by the 2018 American Society of Retina Specialists Preferences and Trends Survey report, which showed that under the current treatment paradigm, nearly 80% of United States retina specialists consider a change in anti-VEGF treatment after only 3 to 6 injections.12 Despite the common practice of switching drugs early in the course of treatment, the data in the literature that support changing treatment within 3 to 6 months of initiation (versus continuing the same treatment) are limited. In a meta-analysis of 28 switching studies, patients gained a mean of only 1.11 and 0.63 letters 6 and 12 months after their change in treatment, respectively.10 However, these patients exhibited reductions in central retinal thickness of 61.9 and 50.0 mm at 6 and 12 months after changing anti-VEGF therapies, respectively. It is important to note that many of these switching studies contain 1 critical design flaw: they lack an adequate control group. This control group should represent patients who, despite meeting the typical criteria used for switching anti-VEGF therapies in nAMD, continue their original therapy for the remainder of the study. Admittedly, given the lack of available data on the matter, the inclusion of such a control group carries the risk of denying patients the best possible treatment. The present analysis attempts to address this concern without compromising patient outcomes by retrospectively identifying and following up patients from HARBOR who commonly would be considered for a switch in anti-VEGF treatment in the clinical practice
setting, but who instead continued their original treatment. Ferris et al3 performed a pseudoswitching analysis of Comparison of AMD Treatments Trials (CATT) data and found that patients demonstrated continued visual and OCT improvement, despite meeting switching criteria and continuing the initial treatment assignment (ranibizumab or bevacizumab). This report prompted us to use the HARBOR data to test the hypothesis that the result generally was valid and not simply an anomaly of the CATT clinical trial data. For this reason, we used switching criteria that closely mimicked those used by Ferris et al.3 The results of the present analysis indicate that only a minority of patients (approximately 5%) in HARBOR met the hypothetical anatomic and visual switching criteria detailed above. This result is similar to the proportion who met switching criteria in the study by Ferris et al.3 Even in the present sensitivity analysis, which involved reduced stringency for switching, only approximately 8% of patients met the criteria for switching. These results indicate that the vast majority of patients show a good response to antiVEGF therapy (in this case, ranibizumab) during the first 2 years of treatment and that switching early and often may not be needed for most patients. After meeting the switching criteria in this HARBOR analysis, pseudoswitchers demonstrated either modest vision gains or vision stability, with continued reductions in CFT through month 24. Those who met the switching criteria at month 3 continued to demonstrate anatomic and visual improvements, despite continuing their original therapy, suggesting that even those patients who did not demonstrate a robust response to initial treatment were capable of achieving meaningful vision gains or stability as well as a significant anatomic response without a change in anti-VEGF agent. Those eligible for a switch at month 6 demonstrated minimal vision gains coupled with significant anatomic improvements by the conclusion of the trial. In a secondary analysis using less stringent switching criteria, these results were replicated. These findings mirror those reported by Ferris et al3 using the CATT database. The findings reported here are remarkably similar to the overall results of uncontrolled switching trials summarized in 2 meta-analyses.9,13 Spooner et al13 reported mean 1.11- and 0.63-letter gains at months 6 and 12, respectively, after switching from ranibizumab or bevacizumab to aflibercept. Neither change was significantly different from a 0-letter gain. In contrast, there was a statistically significant 62- and 50-mm reduction in mean central retinal thickness after switching treatments at months 6 and 12, respectively. The HARBOR data presented here and the CATT data reported by Ferris et al3 indicate that without an adequate control group, it is difficult to ascertain whether the visual and anatomic improvements reported in these true switching studies are directly the result of the new anti-VEGF treatment or simply the result of continued, possibly more aggressive, anti-VEGF therapy. To assess definitively the benefits of changing anti-VEGF treatments for unresponsive patients with nAMD versus continuing a single, consistent or more intensive treatment regimen, there should be, at a minimum, a contemporaneous randomly assigned comparator group that does not change treatment and is followed up for a comparable period. One also may consider
5
Ophthalmology Volume -, Number -, Month 2019 reassignment to the original treatment regimen after a switch to demonstrate regression in clinical response elicited by the second agent after having resumed treatment with the original anti-VEGF agent. The present study is not without its own limitations. With only 5% of patients from HARBOR meeting the hypothetical switching criteria, it was necessary to pool data from the monthly and PRN treatment arms for analysis. Thus, we cannot ascertain whether patients treated PRN would have demonstrated the same results as those treated with monthly injections after meeting the criteria for switching. However, our analysis of these small subgroups did not reveal any meaningful differences in outcomes (data not shown). Furthermore, there is no comparator group (i.e., true switchers), so we cannot assess what the outcome would have been if these patients demonstrating an initial suboptimal response actually were switched to a different anti-VEGF agent. The present study does not prove that switching treatment methods is ineffective in any patient at any time. There may be occasions when patients with nAMD exhibit a more robust response to one anti-VEGF agent versus another. Physicians may consider alternatives to switching, such as increasing injection frequency, when managing a patient who seems unresponsive to therapy. In this regard, it may be helpful to have a patient return for evaluation 1 to 2 weeks after injection with repeat testing to assess whether there has been a reduction in subretinal fluid or retinal edema shortly after the antiVEGF injection.16 If there has been a response, then one may conclude that the problem is not one of drug resistance, but rather one of the durability of the treatment response, and may consider increasing the injection frequency accordingly. However, the fact that only a small (5%) proportion of patients met the switching criteria in HARBORdand that among this subgroup, maintenance of initial therapy resulted in visual and anatomic improvementdsuggests that physicians should expect that continuation of initial therapy in many cases will result in gradual improvement or stabilization of the eye.
3.
4.
5. 6. 7. 8. 9.
10.
11.
12.
13.
14. 15.
References 1. Friedman DS, O’Colmain BJ, Muñoz B, et al. Prevalence of age-related macular degeneration in the United States. JAMA Ophthalmol. 2004;122:564e572. 2. Wong T, Chakravarthy U, Klein R, et al. The natural history and prognosis of neovascular age-related macular
16.
degeneration: a systematic review of the literature and metaanalysis. Ophthalmology. 2008;115:116e126. Ferris 3rd FL, Maguire MG, Glassman AR, et al. Evaluating effects of switching anti-vascular endothelial growth factor drugs for age-related macular degeneration and diabetic macular edema. JAMA Ophthalmol. 2017;135:145e149. Busbee BG, Ho AC, Brown DM, et al. Twelve-month efficacy and safety of 0.5 mg or 2.0 mg ranibizumab in patients with subfoveal neovascular age-related macular degeneration. Ophthalmology. 2013;120:1046e1056. Kasap M, Sazci A. The comparison of VEGFR-1-binding domain of VEGF-A with modelled VEGF-C sheds light on receptor specificity. J Theor Biol. 2008;253:446e551. Regeneron Pharmaceuticals, Inc. Eyelea [package insert]. Tarrytown, NY: Regeneron Pharmaceuticals, Inc.; 2018. Genentech, Inc. Lucentis [package insert]. South San Francisco, CA: Genentech, Inc.; 2018. Genentech, Inc. Avastin [package insert]. South San Francisco, CA: Genentech, Inc.; 2018. Seguin-Greenstein S, Lightman S, Tomkins-Netzer O. A metaanalysis of studies evaluating visual and anatomical outcomes in patients with treatment resistant neovascular age-related macular degeneration following switching to treatment with aflibercept. J Ophthalmol. 2016;2016:4095852. Mehta H, Tufail A, Daien V, et al. Real-world outcomes in patients with neovascular age-related macular degeneration treated with intravitreal vascular endothelial growth factor inhibitors. Prog Retin Eye Res. 2018;65:127e146. Kitchens JW, Kassem N, Wood W, et al. A pharmacogenetics study to predict outcome in patients receiving anti-VEGF therapy in age related macular degeneration. Clin Ophthalmol. 2013;7:1987e1993. Singh RP, Stone TW, eds. 2018 Global Trends in Retina Survey. https://www.asrs.org/content/documents/2018-global-trendsin-retina-survey-highlights-website.pdf; 2018. Accessed 7.12. 18. Spooner K, Hong T, Wijeyakumar W, Chang AA. Switching to aflibercept among patients with treatment-resistant neovascular age-related macular degeneration: a systematic review with meta-analysis. Clin Ophthalmol. 2017;11:161e177. Morton V, Torgerson DJ. Effect of regression to the mean on decision making in health care. BMJ. 2003;326: 1083e1084. Ho AC, Busbee BG, Regillo CD, et al. Twenty-four-month efficacy and safety of 0.5 mg or 2.0 mg ranibizumab in patients with subfoveal neovascular age-related macular degeneration. Ophthalmology. 2014;121:2181e2192. Stewart M, Rosenfeld P, Penha F, et al. Pharmacokinetic rationale for dosing every 2 weeks versus 4 weeks with intravitreal ranibizumab, bevacizumab, and aflibercept (vascular endothelial growth factor trap-eye). Retina. 2012;32: 434e457.
Footnotes and Financial Disclosures Originally received: October 31, 2018. Final revision: December 18, 2018. Accepted: January 7, 2019. Available online: ---.
2017, New Orleans, Louisiana; and 41st Annual Macula Society Meeting, March 2018, Beverly Hills, California. Manuscript no. 2018-2461.
1
Institute of Ophthalmology and Visual Science, New Jersey Medical School, Rutgers University, Newark, New Jersey.
2
Genentech, Inc., South San Francisco, California.
Presented in part at: 2nd International MaculArt Meeting, July 2017, Paris, France; American Academy of Ophthalmology Annual Meeting, November
6
Financial Disclosure(s): The author(s) have made the following disclosure(s): M.Z.: Consultant e Boehringer Ingelheim, Cell Cure, Chengdu Kanghong Biotechnology Group Co., Coherus BioSciences, Daiichi Sankyo, Foundation Fighting Blindness, Frequency Therapeutics, Genentech, Inc./Roche, Healios K.K., IRIDEX, Isarna Therapeutics, Makindus, Novartis Pharma AG, Ophthotech, Percept Corporation; Financial support, Patent e Rutgers University.
Zarbin et al
HARBOR Simulated Switching
M.T.: Employee e Genentech, Inc. (South San Francisco, CA). L.F.H.: Consultant e Aerpio; Employee e Genentech, Inc. (South San Francisco, CA). I.S.: Employee e Genentech, Inc. (South San Francisco, CA). Supported by Genentech, Inc., South San Francisco, California; Funding was provided by Genentech, Inc., a member of the Roche Group, for the study and third-party writing assistance, which was provided by Jack W. Pike, PhD, of Envision Pharma Group. Qualified researchers may request access to individual patient-level data through the clinical study data request platform (www.clinicalstudydatarequest.com). Further details on Roche’s criteria for eligible studies are available here (https://clinicalstudydatarequest.com/Study-Sponsors/Study-SponsorsRoche.aspx). For further details on Roche’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see here (https://www.roche.com/research_and_development/who_we_are_how_we_work/clinical_trials/ our_commitment_to_data_sharing.htm). HUMAN SUBJECTS: Human subjects were included in this study. The present analysis is a post hoc analysis of the HARBOR Primary Manuscript (Ophthalmology 2013;120:1046-1056). The original HARBOR study protocol was approved by respective institutional review boards prior to the start of the study, and all participants provided
written, informed consent. All research adhered to the tenets of the Declaration of Helsinki. No animal subjects were included in this study. Author Contributions: Conception and design: Zarbin, Tsuboi, Hill, Stoilov Analysis and interpretation: Zarbin, Tsuboi, Hill, Stoilov Data collection: Zarbin, Tsuboi, Hill, Stoilov Obtained funding: Zarbin, Tsuboi, Hill, Stoilov Overall responsibility: Zarbin, Tsuboi, Hill, Stoilov Abbreviations and Acronyms: AMD ¼ age-related macular degeneration; BCVA ¼ best-corrected visual acuity; CATT ¼ Comparison of AMD Treatments Trials; CFT ¼ central foveal thickness; CST ¼ central subfield thickness; nAMD ¼ neovascular age-related macular degeneration; PRN ¼ pro re nata; VEGF ¼ vascular endothelial growth factor. Correspondence: Marco Zarbin, MD, PhD, Institute of Ophthalmology and Visual Science, New Jersey Medical School, Rutgers University, Doctors Office Center, Room 6155, 90 Bergen Street, Newark, NJ 07103. E-mail: zarbin@ earthlink.net.
7
Age-related macular degeneration (AMD) is one of the leading causes of blindness in the United States among persons 50 years of age or older.1,2 Vision loss in AMD can occur as the result of vascular endothelial growth factor (VEGF)-mediated growth of new choroidal vessels and vascular leakage affecting the macula.2 The current standard of care for treating neovascular AMD (nAMD) involves blocking VEGF signaling using anti-VEGF agents.3 Several anti-VEGF agents currently are used for the treatment of nAMD in the United States, including ranibizumab (Lucentis; Genentech, Inc., South San Francisco, CA, and Novartis Pharma, Basel, Switzerland), aflibercept (Eylea; Regeneron Pharmaceuticals, Inc., Tarrytown, NY),4e7 and off-label bevacizumab (Avastin; Genentech, Inc., South San Francisco, CA).8,9 ª 2019 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.
Most patients with nAMD achieve clinically meaningful visual benefits after anti-VEGF therapy, as demonstrated by a marked reduction in the prevalence of blindness resulting from nAMD since the approval of the first anti-VEGF treatment more than 10 years ago.10 However, a small proportion of patients do not show an early or satisfactory response to intravitreal anti-VEGF therapy. This suboptimal response may be the result of a number of factors, including misdiagnosis, inadequate treatment, or development of resistance to the therapeutic method. When patients with nAMD fail to show significant clinical improvement within the first 3 to 6 months of initiating anti-VEGF treatment, they often are switched to another anti-VEGF agent.11,12 However, the benefits of such a switch are not well established in randomized controlled trials. https://doi.org/10.1016/j.ophtha.2019.01.003 ISSN 0161-6420/19
1
Ophthalmology Volume -, Number -, Month 2019 Meta-analyses examining numerous anti-VEGF switching studies have shown that most patients with nAMD who switch treatment from either ranibizumab or off-label bevacizumab to aflibercept demonstrate continued anatomic improvement with vision stabilization or limited visual improvement.9,13 Although these benefits generally are attributed to the new anti-VEGF agent, limitations in study design preclude definitive conclusions regarding the cause of the improvement. Most of these studies do not include a comparator arm, and as a result, it is unclear if the benefits achieved are the result of the new anti-VEGF drug, the result of continuing anti-VEGF treatment for a longer duration, or something else. Another issue to consider when assessing the results of switching studies is the potential impact of confounding factors. Uncontrolled studies of the treatment of diseases that have a variable natural course, such as nAMD, are susceptible to the confounding effects of regression toward the mean and sampling bias. Regression toward the mean describes the phenomenon in which, for example, if an individual central subfield thickness (CST) value is extreme when measured on a particular occasion, subsequent CST measurements will tend to be closer to the average CST value as measured for that patient on previous and subsequent occasions. Sampling bias occurs when the patients enrolled in a study are not representative of the entire population targeted for treatment. Figure 1 illustrates 2 hypothetical patients who are receiving treatment for nAMD. Although the mean CSTs of the 2 patients are similar, patient 1 exhibits a more variable response to monthly injections than patient 2. As a result, assessment of patient 1 at 6 months when the CST is relatively high (indicated by the arrow in the figure) may result in a decision to switch treatment to a different anti-VEGF agent. However, as illustrated in Figure 1, even without switching, the subsequent CST measurement at month 9 is lower because of regression toward the mean. If patient 1 had switched treatment at month 6, the subsequent improvement in CST probably would have been attributed to the switch. Thus, regression toward the mean can lead to an incorrect interpretation of the consequences of a therapeutic intervention.14 Patients similar to patient 1 are more likely to be enrolled in switching studies than those similar to patient 2, which may contribute to sampling bias in uncontrolled switching studies, because these patients may not be representative of the larger population. These considerations highlight the need for a control group in switching studies if the effect of switching is to be assessed accurately. Without a control group, interpretation of the results of uncontrolled switching studies can be compromised by sampling bias and overestimation of the magnitude of a treatment effect induced by the switch. The present post hoc analysis of patients from the 24-month HARBOR AMD trial simulates a switching study by identifying patients who met a set of predefined switching criteria, hypothetically switching them to a different anti-VEGF agent, and assessing the changes in vision and anatomic features after the fictitious switch date. Because patients were not actually switched, but remained treated with ranibizumab throughout
2
the study, this analysis examines the effect of continuing the original anti-VEGF treatment in patients who typically would have been switched to a different agent.
Methods Patient Selection The present analysis examined the 1097 treatment-naive patients with nAMD and active subfoveal choroidal neovascularization who were randomized in the 24-month phase 3 HARBOR trial (clinicaltrials.gov identifier, NCT00891735). Patients received ranibizumab 0.5-mg or 2.0-mg intravitreal injections either monthly or as needed (pro re nata [PRN]) after 3 consecutive monthly loading doses. The HARBOR clinical trial was conducted in compliance with the tenets of the Declaration of Helsinki. The independent ethics committee or institutional review boards at each site approved the HARBOR study protocol, and all patients provided written informed consent before enrollment, which extended to the use of individual patient data for further analyses.
Simulated Switching Criteria Patient data were examined at months 3 and 6 to determine whether they met all the hypothetical switching criteria. To be eligible for a switch at month 3, patients must have received all 3 monthly ranibizumab loading doses. To meet the switching criteria at month 6, patients must have received at least 5 monthly doses in the first 6 months of the study. At either time point, switching criteria included gaining 5 letters or fewer from baseline, having a best-corrected visual acuity (BCVA) of 20/40 or worse, and presence of either intraretinal or subretinal fluid, with central foveal thickness (CFT) greater than or equal to CST. It is important to note that despite meeting all the predefined switching criteria, none of these patients were switched from their original treatment with ranibizumab.
Outcome Measures Outcome measures were quantified as change from the point of the hypothetical switch (i.e., month 3 or month 6) to months 6, 12, 18, and 24 for month 3 switchers and to months 12, 18, and 24 for month 6 switchers. The primary visual outcome measure of this study was mean change in BCVA over time. The primary anatomic outcome measure was mean change in CFT over time. Patients in the ranibizumab 0.5-mg or 2.0-mg monthly and PRN treatment arms were pooled for all analyses.
Statistical Analysis Descriptive statistics using observed data were used to evaluate changes in BCVA and CFT outcomes over time for each hypothetical switching group. All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC).
Results Of the 1097 patients who were randomized in HARBOR, 1059 patients received 3 of 3 loading doses by month 3. Only 44 of these 1059 patients (4.2%) met the additional visual and anatomic hypothetical switching criteria at month 3. Similarly, 769 of 1097 patients had received at least 5 of 6 initial monthly injections by month 6, and 37 of 769 patients (4.8%) met the additional visual and anatomic criteria at month 6 for inclusion in this analysis (Fig 2).
350
HARBOR Simulated Switching
10
More likely to select patient 1 for switching (sampling bias), and patient 1 more likely to show improvement without a switch (regression to the mean)
300
Mean = 249 μm
CST, μm
250
200 Mean = 244 μm 150
100
Mean Change in BCVA from Month 3, ETDRS Letters
Zarbin et al
5.3
4.6 2.7
0
–5 50
5.2
5
3
6
12
18
24
40
41
Month n = 44
0 0
3
6
9
12
15
18
Month
Figure 1. Graph showing that patient 1 is most likely to be selected for switching because of sampling bias. However, patient 1 can show improvement without a switch (because of regression toward the mean), which may result in attributing the change in clinical status to a change in treatment, when the change in status would have occurred without the change in treatment. CST ¼ central subfield thickness.
Vision Outcomes On average, patients who met the switching criteria at month 3 (n ¼ 44) continued to demonstrate vision gains through the remaining 21 months of the study, despite continuing their original treatment regimen with ranibizumab (Fig 3). Patients gained, on average, 5.2, 5.3, 4.6, and 2.7 Early Treatment Diabetic Retinopathy Study (ETDRS) letters from month 3 to months 6, 12, 18, and 24, respectively. Patients who were fictitiously switched at month 6 (n ¼ 37) demonstrated minimal improvement in vision over the remainder of HARBOR (Fig 4), on average gaining 1.6, 0.2, and 1.8 ETDRS letters from month 6 to months 12, 18, and 24, respectively.
Anatomic Outcomes Patients who met the hypothetical switching criteria at month 3 (n ¼ 44) demonstrated continued reduction in CFT over the remainder of the trial. From month 3, mean reductions in CFT of 77.5 mm,
Initial monthly treatment: • Month 3 switch: 3/3 initial monthly doses by month 3 • Month 6 switch: ≥5/6 initial monthly doses by month 6 Hypothetical switch criteria: • ≤5-letter gain from baseline • Visual acuity 20/40 or worse • Intraretinal or subretinal fluid with CFT ≥ CST Outcome measures: • Visual: Mean change in BCVA • Anatomic: Mean change in CFT • All changes were from the month 3 or month 6 visit when criteria for switching were met
44
42
Figure 3. Graph showing best-corrected visual acuity (BCVA) changes for month 3 switching candidates. Patients meeting the month 3 switching criteria demonstrated modest BCVA gains from month 3. These vision gains were maintained through completion of the trial at month 24. Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
92.5 mm, 93.5 mm, and 111.7 mm were observed at months 6, 12, 18, and 24, respectively (Fig 5). A similar trend was observed in patients who met the switching criteria at month 6. From the point at which the criteria were met, patients demonstrated a mean reduction in CFT of 99.6 mm, 112.3 mm, and 136.2 mm at months 12, 18, and 24, respectively (Fig 6).
Sensitivity Analysis An additional analysis examined these patients under a less stringent set of switching criteria, in which patients were required to meet only the following criteria: (1) received all 3 monthly ranibizumab injections by month 3 (3-month switchers) or at least 5 monthly injections by month 6 (6-month switchers), (2) demonstrated either intraretinal or subretinal fluid, with CFT greater than or equal to CST, and (3) gained 5 letters or fewer from baseline or demonstrated BCVA of 20/40 or worse. The difference in this set of criteria compared with those used in the primary analysis was that rather than having to meet both the criteria for 5 letters or fewer gained and for BCVA of 20/40 or worse, patients were eligible for switching if they
HARBOR: treatment-naive patients with neovascular AMD and active subfoveal CNV; ranibizumab monthly or PRN for 2 years (N = 1097)
Month 3 switch: 1059 eligible patients
Month 6 switch: 769 eligible patients
Month 3: 44 (4.2%) patients met switch criteria
Month 6: 37 (4.8%) patients met switch criteria
Patients identified did not switch treatments, but remained on their initial treatment
Figure 2. Flow diagram showing study design and patient selection. AMD ¼ age-related macular degeneration; BCVA ¼ best-corrected visual acuity; CFT ¼ central foveal thickness; CNV ¼ choroidal neovascularization; CST ¼ central subfield thickness; PRN ¼ pro re nata.
3
Ophthalmology Volume -, Number -, Month 2019 25
5 1.6
1.8 0.2
0
Mean Change in CFT from Month 6, μm
Mean Change in BCVA from Month 6, ETDRS Letters
10
–75 –99.6
–100
–112.3
–125
–136.2
–150
–200
6
12
18
24
34
33
Figure 4. Graph showing best-corrected visual acuity (BCVA) changes for month 6 switching candidates. Patients meeting the month 6 switching criteria demonstrated BCVA stability from month 6 through the conclusion of the trial at month 24. Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
met only 1 of these in addition to those previously outlined in the primary analysis. At month 3, 85 patients (8.0%) met these less stringent criteria, whereas 66 patients (8.6%) met them at month 6. In this sensitivity analysis, month 3 switchers demonstrated minimal vision improvement from the time the switching criteria were met through the remainder of the trial, with a mean gain of 1.2 letters from months 3 to 24 (Fig 7A). A similar trend was observed in the month 6 switchers, with a mean gain from months 6 to 24 of 0.8 letters (Fig 7B). Both groups continued to experience steady improvements in CFT over the remainder of the 24-month trial after the switch, with CFT reductions of 110.3 mm (Fig 8A) and 130.9 mm (Fig 8B) for the 3- and 6-month switchers, respectively.
Discussion HARBOR reported the 24-month outcomes of the 1097 patients randomized to receive ranibizumab 0.5-mg or 2.0-mg injections monthly or PRN after 3 consecutive loading doses. At month 24, patients in the ranibizumab 0.5-mg PRN, 0.5-mg 25 0 –25 –50 –77.5
–75
–92.5
–100
–93.5 –111.7
–125
–175 –200
24
37
34
34
33
A
10
5
0
–5
3
6
12
18
24
18
24
Month
B
25 0 –25 –50 –75 –100 –125 –150 –175 3
6
12 Month
3
6
12
44
44
42
18
24
40
41
Month n=
18
monthly, 2.0-mg PRN, and 2.0-mg monthly treatment arms gained a mean of 7.9, 9.1, 7.6, and 8.0 letters from baseline, respectively.15 Although on average, these patients achieved clinically meaningful vision gains in the clinical trial setting, it is important to note that in practice, individual patients may exhibit a less robust response to treatment. This is
–200
–150
12
Figure 6. Graph showing central foveal thickness (CFT) changes for month 6 switching candidates. Patients meeting the month 6 switching criteria demonstrated steady CFT improvement from month 6 through the completion of the trial at month 24. Error bars represent 95% confidence intervals.
Mean Change in BCVA from Month 3, ETDRS Letters
35
n=
Mean Change in CFT from Month 3, μm
n = 37
6
Month
Month
Mean Change in CFT from Month 3, μm
–50
–175
–5
Figure 5. Graph showing central foveal thickness (CFT) changes for month 3 switching candidates. Patients meeting the month 3 switching criteria demonstrated steady CFT improvement from month 3 through the completion of the trial at month 24. Error bars represent 95% confidence intervals.
4
0 –25
Figure 7. A, Graph showing mean best-corrected visual acuity (BCVA) change from month 3 using less-stringent switching criteria. B, Graph showing mean central foveal thickness (CFT) change from month 3 using less stringent switching criteria. Patients who met the less stringent switching criteria at month 3 demonstrated BCVA and CFT improvements. These gains were comparable with those of patients who met the complete switching criteria used in the standard analysis (Figs 3 and 5, respectively). Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
Zarbin et al
A
HARBOR Simulated Switching
Mean Change in BCVA from Month 6, ETDRS Letters
10
5
0
–5
6
12
18
24
18
24
Month
Mean Change in CFT from Month 6, μm
B
25 0 –25 –50 –75
–100 –125 –150 –175 –200
6
12 Month
Figure 8. A, Graph showing mean best-corrected visual acuity (BCVA) change from month 6 using less stringent switching criteria. B, Graph showing mean central foveal thickness (CFT) change from month 6 using less stringent switching criteria. Patients who met the less stringent switching criteria at month 6 demonstrated BCVA and CFT improvements. These gains were comparable with those of patients who met the complete switching criteria used in the standard analysis (Figs 4 and 6, respectively). Error bars represent 95% confidence intervals. ETDRS ¼ Early Treatment Diabetic Retinopathy Study.
reflected by the 2018 American Society of Retina Specialists Preferences and Trends Survey report, which showed that under the current treatment paradigm, nearly 80% of United States retina specialists consider a change in anti-VEGF treatment after only 3 to 6 injections.12 Despite the common practice of switching drugs early in the course of treatment, the data in the literature that support changing treatment within 3 to 6 months of initiation (versus continuing the same treatment) are limited. In a meta-analysis of 28 switching studies, patients gained a mean of only 1.11 and 0.63 letters 6 and 12 months after their change in treatment, respectively.10 However, these patients exhibited reductions in central retinal thickness of 61.9 and 50.0 mm at 6 and 12 months after changing anti-VEGF therapies, respectively. It is important to note that many of these switching studies contain 1 critical design flaw: they lack an adequate control group. This control group should represent patients who, despite meeting the typical criteria used for switching anti-VEGF therapies in nAMD, continue their original therapy for the remainder of the study. Admittedly, given the lack of available data on the matter, the inclusion of such a control group carries the risk of denying patients the best possible treatment. The present analysis attempts to address this concern without compromising patient outcomes by retrospectively identifying and following up patients from HARBOR who commonly would be considered for a switch in anti-VEGF treatment in the clinical practice
setting, but who instead continued their original treatment. Ferris et al3 performed a pseudoswitching analysis of Comparison of AMD Treatments Trials (CATT) data and found that patients demonstrated continued visual and OCT improvement, despite meeting switching criteria and continuing the initial treatment assignment (ranibizumab or bevacizumab). This report prompted us to use the HARBOR data to test the hypothesis that the result generally was valid and not simply an anomaly of the CATT clinical trial data. For this reason, we used switching criteria that closely mimicked those used by Ferris et al.3 The results of the present analysis indicate that only a minority of patients (approximately 5%) in HARBOR met the hypothetical anatomic and visual switching criteria detailed above. This result is similar to the proportion who met switching criteria in the study by Ferris et al.3 Even in the present sensitivity analysis, which involved reduced stringency for switching, only approximately 8% of patients met the criteria for switching. These results indicate that the vast majority of patients show a good response to antiVEGF therapy (in this case, ranibizumab) during the first 2 years of treatment and that switching early and often may not be needed for most patients. After meeting the switching criteria in this HARBOR analysis, pseudoswitchers demonstrated either modest vision gains or vision stability, with continued reductions in CFT through month 24. Those who met the switching criteria at month 3 continued to demonstrate anatomic and visual improvements, despite continuing their original therapy, suggesting that even those patients who did not demonstrate a robust response to initial treatment were capable of achieving meaningful vision gains or stability as well as a significant anatomic response without a change in anti-VEGF agent. Those eligible for a switch at month 6 demonstrated minimal vision gains coupled with significant anatomic improvements by the conclusion of the trial. In a secondary analysis using less stringent switching criteria, these results were replicated. These findings mirror those reported by Ferris et al3 using the CATT database. The findings reported here are remarkably similar to the overall results of uncontrolled switching trials summarized in 2 meta-analyses.9,13 Spooner et al13 reported mean 1.11- and 0.63-letter gains at months 6 and 12, respectively, after switching from ranibizumab or bevacizumab to aflibercept. Neither change was significantly different from a 0-letter gain. In contrast, there was a statistically significant 62- and 50-mm reduction in mean central retinal thickness after switching treatments at months 6 and 12, respectively. The HARBOR data presented here and the CATT data reported by Ferris et al3 indicate that without an adequate control group, it is difficult to ascertain whether the visual and anatomic improvements reported in these true switching studies are directly the result of the new anti-VEGF treatment or simply the result of continued, possibly more aggressive, anti-VEGF therapy. To assess definitively the benefits of changing anti-VEGF treatments for unresponsive patients with nAMD versus continuing a single, consistent or more intensive treatment regimen, there should be, at a minimum, a contemporaneous randomly assigned comparator group that does not change treatment and is followed up for a comparable period. One also may consider
5
Ophthalmology Volume -, Number -, Month 2019 reassignment to the original treatment regimen after a switch to demonstrate regression in clinical response elicited by the second agent after having resumed treatment with the original anti-VEGF agent. The present study is not without its own limitations. With only 5% of patients from HARBOR meeting the hypothetical switching criteria, it was necessary to pool data from the monthly and PRN treatment arms for analysis. Thus, we cannot ascertain whether patients treated PRN would have demonstrated the same results as those treated with monthly injections after meeting the criteria for switching. However, our analysis of these small subgroups did not reveal any meaningful differences in outcomes (data not shown). Furthermore, there is no comparator group (i.e., true switchers), so we cannot assess what the outcome would have been if these patients demonstrating an initial suboptimal response actually were switched to a different anti-VEGF agent. The present study does not prove that switching treatment methods is ineffective in any patient at any time. There may be occasions when patients with nAMD exhibit a more robust response to one anti-VEGF agent versus another. Physicians may consider alternatives to switching, such as increasing injection frequency, when managing a patient who seems unresponsive to therapy. In this regard, it may be helpful to have a patient return for evaluation 1 to 2 weeks after injection with repeat testing to assess whether there has been a reduction in subretinal fluid or retinal edema shortly after the antiVEGF injection.16 If there has been a response, then one may conclude that the problem is not one of drug resistance, but rather one of the durability of the treatment response, and may consider increasing the injection frequency accordingly. However, the fact that only a small (5%) proportion of patients met the switching criteria in HARBORdand that among this subgroup, maintenance of initial therapy resulted in visual and anatomic improvementdsuggests that physicians should expect that continuation of initial therapy in many cases will result in gradual improvement or stabilization of the eye.
3.
4.
5. 6. 7. 8. 9.
10.
11.
12.
13.
14. 15.
References 1. Friedman DS, O’Colmain BJ, Muñoz B, et al. Prevalence of age-related macular degeneration in the United States. JAMA Ophthalmol. 2004;122:564e572. 2. Wong T, Chakravarthy U, Klein R, et al. The natural history and prognosis of neovascular age-related macular
16.
degeneration: a systematic review of the literature and metaanalysis. Ophthalmology. 2008;115:116e126. Ferris 3rd FL, Maguire MG, Glassman AR, et al. Evaluating effects of switching anti-vascular endothelial growth factor drugs for age-related macular degeneration and diabetic macular edema. JAMA Ophthalmol. 2017;135:145e149. Busbee BG, Ho AC, Brown DM, et al. Twelve-month efficacy and safety of 0.5 mg or 2.0 mg ranibizumab in patients with subfoveal neovascular age-related macular degeneration. Ophthalmology. 2013;120:1046e1056. Kasap M, Sazci A. The comparison of VEGFR-1-binding domain of VEGF-A with modelled VEGF-C sheds light on receptor specificity. J Theor Biol. 2008;253:446e551. Regeneron Pharmaceuticals, Inc. Eyelea [package insert]. Tarrytown, NY: Regeneron Pharmaceuticals, Inc.; 2018. Genentech, Inc. Lucentis [package insert]. South San Francisco, CA: Genentech, Inc.; 2018. Genentech, Inc. Avastin [package insert]. South San Francisco, CA: Genentech, Inc.; 2018. Seguin-Greenstein S, Lightman S, Tomkins-Netzer O. A metaanalysis of studies evaluating visual and anatomical outcomes in patients with treatment resistant neovascular age-related macular degeneration following switching to treatment with aflibercept. J Ophthalmol. 2016;2016:4095852. Mehta H, Tufail A, Daien V, et al. Real-world outcomes in patients with neovascular age-related macular degeneration treated with intravitreal vascular endothelial growth factor inhibitors. Prog Retin Eye Res. 2018;65:127e146. Kitchens JW, Kassem N, Wood W, et al. A pharmacogenetics study to predict outcome in patients receiving anti-VEGF therapy in age related macular degeneration. Clin Ophthalmol. 2013;7:1987e1993. Singh RP, Stone TW, eds. 2018 Global Trends in Retina Survey. https://www.asrs.org/content/documents/2018-global-trendsin-retina-survey-highlights-website.pdf; 2018. Accessed 7.12. 18. Spooner K, Hong T, Wijeyakumar W, Chang AA. Switching to aflibercept among patients with treatment-resistant neovascular age-related macular degeneration: a systematic review with meta-analysis. Clin Ophthalmol. 2017;11:161e177. Morton V, Torgerson DJ. Effect of regression to the mean on decision making in health care. BMJ. 2003;326: 1083e1084. Ho AC, Busbee BG, Regillo CD, et al. Twenty-four-month efficacy and safety of 0.5 mg or 2.0 mg ranibizumab in patients with subfoveal neovascular age-related macular degeneration. Ophthalmology. 2014;121:2181e2192. Stewart M, Rosenfeld P, Penha F, et al. Pharmacokinetic rationale for dosing every 2 weeks versus 4 weeks with intravitreal ranibizumab, bevacizumab, and aflibercept (vascular endothelial growth factor trap-eye). Retina. 2012;32: 434e457.
Footnotes and Financial Disclosures Originally received: October 31, 2018. Final revision: December 18, 2018. Accepted: January 7, 2019. Available online: ---.
2017, New Orleans, Louisiana; and 41st Annual Macula Society Meeting, March 2018, Beverly Hills, California. Manuscript no. 2018-2461.
1
Institute of Ophthalmology and Visual Science, New Jersey Medical School, Rutgers University, Newark, New Jersey.
2
Genentech, Inc., South San Francisco, California.
Presented in part at: 2nd International MaculArt Meeting, July 2017, Paris, France; American Academy of Ophthalmology Annual Meeting, November
6
Financial Disclosure(s): The author(s) have made the following disclosure(s): M.Z.: Consultant e Boehringer Ingelheim, Cell Cure, Chengdu Kanghong Biotechnology Group Co., Coherus BioSciences, Daiichi Sankyo, Foundation Fighting Blindness, Frequency Therapeutics, Genentech, Inc./Roche, Healios K.K., IRIDEX, Isarna Therapeutics, Makindus, Novartis Pharma AG, Ophthotech, Percept Corporation; Financial support, Patent e Rutgers University.
Zarbin et al
HARBOR Simulated Switching
M.T.: Employee e Genentech, Inc. (South San Francisco, CA). L.F.H.: Consultant e Aerpio; Employee e Genentech, Inc. (South San Francisco, CA). I.S.: Employee e Genentech, Inc. (South San Francisco, CA). Supported by Genentech, Inc., South San Francisco, California; Funding was provided by Genentech, Inc., a member of the Roche Group, for the study and third-party writing assistance, which was provided by Jack W. Pike, PhD, of Envision Pharma Group. Qualified researchers may request access to individual patient-level data through the clinical study data request platform (www.clinicalstudydatarequest.com). Further details on Roche’s criteria for eligible studies are available here (https://clinicalstudydatarequest.com/Study-Sponsors/Study-SponsorsRoche.aspx). For further details on Roche’s Global Policy on the Sharing of Clinical Information and how to request access to related clinical study documents, see here (https://www.roche.com/research_and_development/who_we_are_how_we_work/clinical_trials/ our_commitment_to_data_sharing.htm). HUMAN SUBJECTS: Human subjects were included in this study. The present analysis is a post hoc analysis of the HARBOR Primary Manuscript (Ophthalmology 2013;120:1046-1056). The original HARBOR study protocol was approved by respective institutional review boards prior to the start of the study, and all participants provided
written, informed consent. All research adhered to the tenets of the Declaration of Helsinki. No animal subjects were included in this study. Author Contributions: Conception and design: Zarbin, Tsuboi, Hill, Stoilov Analysis and interpretation: Zarbin, Tsuboi, Hill, Stoilov Data collection: Zarbin, Tsuboi, Hill, Stoilov Obtained funding: Zarbin, Tsuboi, Hill, Stoilov Overall responsibility: Zarbin, Tsuboi, Hill, Stoilov Abbreviations and Acronyms: AMD ¼ age-related macular degeneration; BCVA ¼ best-corrected visual acuity; CATT ¼ Comparison of AMD Treatments Trials; CFT ¼ central foveal thickness; CST ¼ central subfield thickness; nAMD ¼ neovascular age-related macular degeneration; PRN ¼ pro re nata; VEGF ¼ vascular endothelial growth factor. Correspondence: Marco Zarbin, MD, PhD, Institute of Ophthalmology and Visual Science, New Jersey Medical School, Rutgers University, Doctors Office Center, Room 6155, 90 Bergen Street, Newark, NJ 07103. E-mail: zarbin@ earthlink.net.
7