Summarizing the Key Anti-VEGF Study Results
Several comparison studies should help guide clinicians through the decision-making process.
Dr. Kaiser: The results of several studies — CATT, IVAN, MANTA, HARBOR and VIEW — comparing different anti-VEGF treatments (and their dosing regimens) have recently been reported. What are your thoughts on these trial results?
CATT, IVAN and MANTA Trials
Dr. Boyer: CATT essentially showed that monthly injections of ranibizumab or bevacizumab resulted in approximately the same visual outcomes at the end of 1 and 2 years. In the PRN group, with as-needed treatment beginning after the first injection, the bevacizumab group did not show non-inferiority. It actually was inferior to the monthly injection groups, where ranibizumab given on a PRN basis provided a better result. Total drying on OCT was seen more in the ranibizumab-treated arm that was receiving monthly treatment than in the monthly bevacizumab-treated arm. A discrepancy in systemic safety was found in the first year. There were more serious adverse events (SAEs) in the first year in the patients receiving bevacizumab. The events did not seem to be related to any known anti-VEGF side effects and the discrepancy was dismissed. But in the second year, a higher SAE rate was again reported in the bevacizumab group. Safety of all the anti-VEGF agents is really a question, although I cannot find a biological reason for some of the events, and that is a problem. We do not know why it is biologically occurring, but safety issues with bevacizumab occurred at a higher rate in both years of CATT.
Dr. Brown: The CATT also showed that bevacizumab does not dry out the retina as well as ranibizumab. The treatment burden is increased with bevacizumab. I have a hard time extrapolating the bevacizumab CATT data to my clinic because the study used an FDA-monitored fill and finish formulary where there is almost no chance of infection, or a very limited chance, anyway. There is a much higher chance of infection when using a compounding pharmacy.
“We do not know why it is biologically occurring, but safety issues with bevacizumab occurred at a higher rate in both years of CATT .”
— David Boyer, MD
Dr. Kaiser: We recently heard the interim results of the IVAN study. What are some of the design differences between CATT and IVAN that clinicians should understand?
Dr. Brown: In IVAN, there was a three-letter margin of inferiority; CATT had five letters. The PRN treatment group in IVAN received three injections before moving to as-needed treatments, but if there was any leakage, the patient received another three monthly doses. So, a patient with two recurrences could have had nine or 10 injections, making the overall number of injections higher in IVAN than in CATT. IVAN aggregated data between the monthly doses, and the PRN data were also aggregated. So, we do not have data on monthly ranibizumab; without the four individual groups it is difficult to interpret the data. In the primary outcome, ranibizumab did not meet the non-inferiority margin, but by a very small amount. The P value is 0.056, so literally two patients may have skewed the results. Had five patients gone the other way, bevacizumab would have been shown to be inferior to ranibizumab. Not “non-inferior,” inferior period. Unlike CATT, IVAN showed systemic anti-VEGF suppression with a serum sample, not a plasma sample. In CATT, the monthly curves are almost on top of each other and look truly equivalent, but in IVAN, it is clear that bevacizumab was not equal.
Dr. Kaiser: I agree. If you look at the confidence intervals around the mean difference between the drugs, the interim outcome was almost inferior. We will see what happens in the second year when the primary outcome is reported.
Dr. Csaky: Whenever fluid was detected, the protocol necessitated three injections. We typically do not mandate three injections in day to day clinical practice. The PRN regimen was much more aggressive in the trials and that is one reason why there were more injections and better visual outcomes in IVAN. That is why, I think, there was no change between the discontinuous and continuous dosing.
Dr. Heier: The one consistent finding between CATT and IVAN is the ability to dry out the retina; both studies found ranibizumab was better than bevacizumab in that respect. We do not have hard data yet to confirm if it makes a visual difference, but I believe that visual outcomes will be affected by persistent fluid.
Dr. Kaiser: The MANTA study also compared ranibizumab and bevacizumab using a PRN dosing regimen after three monthly loading doses with the primary outcome at 1 year in 317 patients at 10 study centers in Austria. Like the bevacizumab used in CATT, the bevacizumab was compounded at central pharmacies. MANTA had a non-inferiority margin of seven letters, so once again it is different from CATT and IVAN. At 1 year, there was no difference in the mean change in vision from baseline, or in the five or 15 letter gainers, but more patients numerically lost 15 letters in the ranibizumab group, although this was not statistically significant. The reasons are not completely clear as to why this occurred and further analysis is being done. The systemic safety was the same between the two groups.
VIEW I and VIEW II
Dr. Heier: VIEW I and VIEW II included more than 2,400 patients, so there were 600 patients in each arm: 2.0 mg aflibercept monthly and 0.5 mg aflibercept monthly; 2.0 mg aflibercept with a loading phase and then every 8 weeks; and the control arm, ranibizumab, 0.5 mg monthly. The data showed that essentially all four of these groups were comparable. In particular, the finding from VIEW Year 1 that was most interesting to us was not simply that the 2.0 mg aflibercept and the 0.5 mg ranibizumab were comparable, but that the 2.0 mg dosed every 8 weeks after the loading phase was comparable. In VIEW Year 2, patients were treated in a capped PRN and followed monthly, but there was a mandatory quarterly dosing so patients would not go more than 3 months without a treatment. Again, the four arms largely maintained the gains they had, but there was some falloff ranging from 0.8 to 1.7 letters. What was interesting is that the number of injections was essentially the same, except in patients who received the most injections. Patients in the ranibizumab group were more likely than those in the aflibercept group to need more than six injections.
Dr. Kaiser: Some would argue that the number of injections of ranibizumab and aflibercept were essentially the same in the second year of VIEW and that aflibercept did not appear to maintain biologic activity longer than other drugs, though this has been postulated.
Dr. Brown: In VIEW Year 2, we did not want a true PRN where we knew vision would be lost. The capped PRN gave everyone a floor they could not fall below. The average number of injections in the second year was 4.1 in the aflibercept arm and 4.2 in the ranibizumab arm. If you look at those numbers, they do not seem to be much different. However, when you compare patients who actually required 9, 10 or 11 injections in the second year, there were more of those patients in the ranibizumab arms than in the aflibecept arms, which implies more of a difference in patients who require more anti-VEGF therapy. The anatomic OCT data at 1 year also suggest aflibercept dries out more retinas. The odds ratio of being dry was much better for patients who were randomized to the every 2 months (q8w) aflibercept arm than the monthly ranibizumab arm based on OCTs at 1 year. I think that is the strongest data of all that aflibercept q8w is keeping patients seemingly drier than ranibizumab monthly.
“Patients who have been treated with anti-VEGF agents are slowly losing vision in the 5- to 7-year framework.”
— Peter Kaiser, MD
HARBOR and SAVE
Dr. Kaiser: Another study was the HARBOR trial, which compared 0.5 mg vs. 2.0 mg ranibizumab monthly as well as PRN after a three monthly loading dose. What were some highlights of this comparison study?
Dr. Boyer: HARBOR compared 0.5 mg and 2.0 mg ranibizumab monthly and PRN after three initial injections. At 1 year, no increased incidence of systemic adverse events was found and the visual outcomes were approximately the same between the 0.5 mg and the 2.0 mg dose in the two different dosing regimens. There was no significant difference found between the doses.
Dr. Kaiser: One concern raised about increased dosing of anti-VEGF agents was an increase in SAEs, and in particular systemic SAEs, and that did not appear to be the case in HARBOR. How was HARBOR different from the SAVE study?
Dr. Brown: The PRN was not shown to be non-inferior. That is only at 1 year. I think the 2-year data will be very interesting. The SAVE study showed that in recalcitrant patients, the higher dose did dry out more patients and achieved a statistically significant improvement in visual acuity. HARBOR found that in a treatment-naïve patient, dose levels do not make a difference. Maybe these patients over time develop resistance, or tachyphylaxis, or some other mechanism wherein they need a higher dose to be effective. But in my mind, tachyphylaxis has never really been demonstrated in anti-VEGF treatments. It would, however, be one explanation as to why these recalcitrant, hard-to-treat patients might need more drugs.
Dr. Csaky: We did not see significant differences in the CNV change with 2.0 mg or 0.5 mg, so this suggests that we are at the top end of the curve in the majority of naïve patients in the amount of anti-VEGF treatments needed. Even with 0.5 mg ranibizumab, we are starting at the top dose and I do not think, in the majority of patients, we can achieve a better response in naïve patients with 2.0 mg.
Dr. Heier: We had a series of patients we considered sub-optimal and we doubled the dose to 1.0 mg. A significant number did have an anatomic response, but a minimal visual response. It’s clear some patients will benefit from the higher doses, but the number is relatively small when you look at the clinical trials. Yet it is relatively high when you look at patients who occupy our exam rooms.
Dr. Csaky: This gets back to the issue of patient variability. Each patient demonstrates to some degree a unique mode of presentation. That is why we can to some degree generalize about clinical trial results but at the same time for an individual patient, I do not think we can generalize and say that 2.0 mg does not work or that 0.5 mg is always the best for every patient.
“The anatomic OCT data at 1 year also suggest aflibercept dries out more retinas. The odds ratio of being dry was much better for patients who were randomized to the every 2 months aflibercept arm than the monthly ranibizumab arm based on OCTs at 1 year.”
— David Brown, MD
Are We Undertreating?
Dr. Kaiser: The second year of the CATT study clearly showed that monthly dosing was better than PRN dosing. So are we undertreating our patients, or are we discounting the study conclusions?
Dr. Brown: We are slowly realizing we are undertreating. After MARINA and ANCHOR, we thought 24 shots would be enough, and people would not need more. But there are people who need monthly shots at year 7, 8 and 9. Every time we try to extend, I think we run the risk of losing some of those vision gains we saw in the curve. The risk of endophthalmitis is about 1 in 3,000, but PRN dosing eliminates the gains you have achieved. We hope we are somewhere in the middle with treat-and-extend.
Dr. Kaiser: In the SEVEN-UP study (7-year follow-up of patients in ANCHOR and MARINA) when patients were converted to PRN after 24 monthly injections and followed over a long period of time, one-third of the patients were 20/200 or worse at 7 years, whereas in the first 2 years of monthly dosing, fewer than 10% were 20/200 or worse. I think we are running into that challenge with the anti-VEGF agents. How do we maintain good vision at years 6, 7, or longer after initiating therapy? Patients who have been treated with anti-VEGF agents are slowly losing vision in the 5- to 7-year framework. We have to be careful with some of our early decisions because we do not know the repercussions of these decisions down the road.
Dr. Boyer: The other surprising finding in the SEVEN-UP study was that 50% of the patients still had active neovascularization 7 years after beginning anti-VEGF therapy. So, we are not really curing anything, we are just controlling the leakage for an unspecified amount of time while we continue to treat. I think we may need to use auto-fluorescence or fluorescein angiography periodically to check for disease progression.
Dr. Heier: We have a few patients who had been enrolled in the original multi-injection ranibizumab study (circa 2001) and one patient in particular developed AMD in the second eye during the study, so we could not treat the fellow eye. That second eye is 20/400 but the one enrolled in the study is still 20/30, and the patient is treated regularly. I think a more pressing issue is non-exudative disease that is also ongoing; it can be difficult to differentiate between the two.
Dr. Kaiser: The CATT 2-year data found patients receiving monthly ranibizumab had larger amounts of geographic atrophy (GA) and a greater enlargement of GA area than the other groups. What are your thoughts on this finding — is this real?
Dr. Brown: I do not think it is likely to be real. We have not seen it in the diabetic macular edema (DME) trials, we have not seen it in the retinal vein occlusion (RVO) studies, and we do not think we have seen it much in patients we have been treating for 10 years. I would like to see that replicated in other studies — HARBOR, VIEW I and II — before I put a lot of credence in it.
“We have to be careful with some of our early decisions because we do not know the repercussions of these decisions down the road.”
— Peter Kaiser, MD
Dr. Csaky: One hypothesis that has been proposed is that patients who have a thin choroid with CNV may be more susceptible to the development of some GA with anti-VEGFs. Do you think measuring choroidal thickness at baseline has a role in this development of GA? Do you think this is a place where we should concentrate more efforts?
Dr. Brown: There are patients with GA with really robust, healthy choroids and patients with almost no choroid where their retinal pigment looks solid, so there is certainly not a linear correlation. If we kill off a patient’s choroid and get secondary RPE atrophy, it is usually from PDT rather than anti-VEGF. I have not seen any data to demonstrate that we are killing off choroid with anti-VEGF therapy.
Dr. Heier: Looking at the anatomic effects of the treatments leads me to believe the persistence of fluid is more likely to have a long-term effect than the development of GA. In most of the studies and in our own experience, under-treatment seems to cost more vision than over-treatment.
Dr. Brown: We all have patients 10 years out seeing well. If we’re hurting the RPE, we need to figure out does one agent do it more than others. Until we have other evidence, it does not concern me that there were small amounts of non-foveal GA seemingly increased in one cohort of the CATT trial.
Dr. Kaiser: I am not certain that the GA seen in CATT was truly due to monthly ranibizumab versus a spurious statistical finding. Were there baseline differences in GA, for instance? For example, it is difficult to see GA in wet maculas, and the monthly ranibizumab subjects had the driest maculas so the GA was easy to see. Maybe the other groups had similar levels of GA, but their wet retinas prevented the GA from being seen. We have to look at other studies where monthly ranibizumab was delivered to see if they also showed similar GA events. Did we see increased GA in HARBOR or VIEW? In HARBOR where they quadrupled the dose of ranibizumab, I would expect to see more atrophy in those patients if the CATT findings are real.
Dr. Brown: We do not have the 2-year long-term data of GA sizes yet. There was autofluorescence, but it was not uniformly used.
Dr. Heier: It is similar with VIEW. It is very hard to assess when you were not looking for it at baseline.
Dr. Brown: It is hard to assess on OCT, too. We looked at it in the SAVE trial where we had a population where all these patients had monthly OCTs. We were disappointed with it because it was so hard to get reproducible signals to grade if there was overlying choroidal neovascular membrane or fibrosis.
Dr. Csaky: The caveat to that is the main reason people go to 20/200 or worse in both ANCHOR and MARINA was the development of GA. So, is it just the natural history of AMD progression, or is it an anti-VEGF issue? I think it needs more study.
Dr. Kaiser: I agree. The GA issue is one we need to watch closely, and in future clinical trials we have to prospectively follow it. I am not convinced yet that the anti-VEGF drugs produce GA. ✦