Interview by L Scott Zoeller. 2012 May 2, Kathy Miller, MD
Dr. Miller is Associate Professor and Sheila D. Ward Scholar, Department of Medicine Division of Hematology/Oncology, Indiana University School of Medicine.
OncologySTAT: Results of recent analyses have dampened some of the excitement over the potential of antiangiogenic therapy in breast cancer, particularly bevacizumab, which has become controversial. Would you give an historical perspective on this topic and bring us up to date regarding research on the safety and efficacy of various antiangiogenic agents in breast cancer?
Dr. Miller: I actually find it an interesting question, because there really haven’t been any recent or new studies. What there has been is a difference in interpretation of studies that were reported several years ago. The historical perspective really goes back to basic biology and pathology studies that found that angiogenesis, or the growth and development of an increased circulatory system, was important for bringing increased perfusion, increased oxygen and nutrient delivery, and disposal of the tumor waste products to allow a tumor to grow. That led to the fairly obvious hypothesis that, if this process is important for the growth and progression of tumors, inhibiting the process ought to be a useful component of therapy.
That led to trying to understand the controls of this process and, like many important biologic processes, the control is typically tightly regulated with multiple redundant systems, with a large number of factors that increase angiogenesis and an equally large number that try to inhibit or stop angiogenesis to maintain the system in balance. So, there are many ways that you could inhibit angiogenesis.
One of those might be by inhibiting signaling of the vascular endothelial growth factor (VEGF) and its receptor pathways. Just focusing on that single ligand shows the complexity of the system, because there are multiple VEGF isoforms that have different tissue distributions and slightly different functions. There are at least three VEGF receptors and two neuropilin coreceptors with a lot of redundancy and cross-talk between which ligands bind to which receptors in which tissues, which then leads to downstream signaling and a whole host of downstream effects, including increased endothelial migration, chemotaxis, immune effects with changes in antigen-presenting dendritic cells, and changes in tissue permeability, as well as resistance to hypoxia and apoptosis.
So, inhibiting VEGF signaling is not nearly so simple and there are many ways to do that in the clinic. The one that entered the clinic first and is the farthest advanced is bevacizumab, a monoclonal antibody that binds to the VEGF-A ligand. That compound had quite modest activity as monotherapy in patients with heavily pretreated breast cancer, with an objective response rate of about 7% to 9%, up to a total of about 15% or 16%, of patients who were either responding or stable for about 5 months or more. But all of the preclinical data suggested that you would have much greater activity in combination with other agents. There was never really a thought that the major use of this sort of agent would be as monotherapy.
That led to the ECOG-2100 study, which looked at patients with newly diagnosed metastatic disease, or, I should say, patients newly requiring chemotherapy.1 Those patients who were estrogen receptor (ER)-positive may have actually had metastatic disease that had been controlled on various hormone maneuvers for quite some time before they came to this trial. It was a simple comparison of chemotherapy with weekly paclitaxel alone vs weekly paclitaxel with bevacizumab. What we reported first back in 2005 were significant improvements in objective response rate and progression-free survival (PFS) with the combination. There was an absolute improvement in PFS of about 5.5 months, with no difference in overall survival, and some increase in toxicity with the combination. The most common toxicity was hypertension, which, for most patients, was quite easy to manage and control. However, about 15% to 16% of patients required a medication to control their hypertension. The rates of potentially serious or life-threatening toxicities, including serious thromboembolic events, gastrointestinal perforations, and pulmonary hemorrhage, were quite low and we didn’t identify any new safety signals that had not been seen in the colorectal and lung cancer trials that were going on concurrently.
At the same time, there were two other phase III efforts being run independently by the corporate sponsor using different chemotherapy backbones. The first was the AVADO trial, which used every-3-week docetaxel.2 It had similar randomization, but included an exploration of a lower dose of bevacizumab. The results also met the endpoints of the trial, which were improvement in response rates and improvement in PFS, the primary endpoint. No difference was found in overall survival. There was also no difference in the toxicity profiles from the bevacizumab, although, with every-3-week docetaxel, the chemotherapy-related toxicity was a bit more substantial. But the absolute difference in PFS was about 2 months instead of 5.5 months.
The RIBBON-1 trial had an even broader chemotherapy backbone strategy.3Patients could have been treated with an anthracycline-based regimen with single-agent taxane-based therapy or with capecitabine. The trial was designed to be analyzed in two components, so the capecitabine cohort was analyzed independently. All of the anthracycline- and taxane-based groups were then lumped together and analyzed as a separate cohort. Both cohorts showed fairly similar results: improvements in response, improvements in PFS on the order of about 2 to 2.5 months, and no difference in overall survival.
All of that data were first available at least 3 years ago. So, there have not been any new phase III studies with bevacizumab in metastatic breast cancer since that time. However, there has been an ongoing debate about whether improvement in PFS in the absence of improvement of overall survival is worthwhile at all, and if it might be worthwhile in some instances. Is this one of them? Or, at what point does the magnitude of the PFS benefit justify the potential toxicities and the potential cost that bevacizumab therapy brings?
So, I’m honestly always puzzled when people ask me about recent studies in metastatic disease because there really haven’t been recent studies. You would think that there were new data when it has been all old data that we’re talking about. In fact, the U.S. Food and Drug Administration (FDA) had seen the top-line results from the AVADO trial before they granted the original accelerated approval, so they were fully aware that the absolute improvement in PFS in that trial with a different chemotherapy agent was not of the same magnitude as the improvement in PFS in E2100 when they originally granted approval.
OncologySTAT: In your opinion, does antiangiogenic therapy have an important future role in breast cancer treatment, be it immunotherapeutic, or tyrosine kinase inhibition, or some other mechanism?
Dr. Miller: I think it will. I think we have some work to do to find a way to get to that point, and the work is on several fronts. However, I think this will continue to be an important area of research and, ultimately, an important area of treatment. It just seems a very unlikely paradox to me to say that this process of increasing the capacity of the circulatory system to support a growing tumor mass in some way is biologically absolutely crucial, and yet therapeutically irrelevant. That just seems such an unlikely paradox that I have a hard time imagining that that is the situation.
This is such a redundant system that it is really naive to think that inhibiting a single ligand, or a single receptor fairly late in the course of the disease, will have such a profound and long-lasting effect. This is very different than inhibiting something like a driving oncogene on the epithelial tumor cells. So, this is not the equivalent of
HER2, where it is directly responsible for driving the tumor cell proliferation and growth. This is a process that plays a supporting role, rather than the leading actor or actress role.
We will need to figure out how to more completely inhibit the process, rather than just a single ligand or single receptor. We’re going to need to tackle the problem of whether the entire process can be inhibited without bringing substantially greater toxicity to bear, and we need to begin to understand which patients are most likely to benefit from which types of available antiangiogenic agents. We frequently hear people talk about bevacizumab or other antiangiogenic agents as one of the new targeted therapies and there is no such thing. Bevacizumab was designed to target a particular ligand, but we don’t use it in that way in the clinic. We don’t select patients based on any molecular feature in the patient’s inheritance or in their tumor. We use bevacizumab as if it were a broad-based therapeutic that is equally applicable, equally likely to be of benefit or harm to every patient who receives it, and that’s clearly not the case. We simply have not been able yet to identify which patients are most likely to benefit. That ultimately is going to be the key to using these agents successfully.
OncologySTAT: Is there any information available about what the optimal duration of therapy could be?
Dr. Miller: We have virtually no clinical information about that in breast cancer. All of the trials in the metastatic setting incorporated the same strategy into the trial design in that the protocols specified continuation of chemotherapy and bevacizumab until either disease progression or toxicity of sufficient severity or frequency such that it is not safe and reasonable to continue. If the toxicity was from the chemotherapy, then the chemotherapy could be discontinued while the bevacizumab was then continued.
Trials were able to do that with varying degrees of success. So, the number of patients in the E2100 trial with weekly paclitaxel who discontinued therapy because of toxicity before disease progression was a fairly small minority. In the AVADO trial, that was the majority of the patients because it was with a chemotherapy regimen that incorporated, or brought with it, much more toxicity. While the trial specified that patients could stop the chemotherapy and continue the bevacizumab, that’s not what all of the patients actually did. Some took that option, but others stopped all of their treatments when they ran into chemotherapy- related toxicity and enjoyed, at that point, a time completely off therapy.
Those factors may have played a role in the different results that we saw from these different trials, but it means we don’t have any good studies really designed to look at the duration of bevacizumab therapy in this setting. There is no study in breast cancer that randomized patients after an initial period of combination chemotherapy and bevacizumab to either continue bevacizumab or stop. We came closest to that in a trial in ovarian cancer (the GOG trial) which looked at concurrent therapy with bevacizumab stopping when the chemotherapy was completed vs concurrent therapy followed by bevacizumab.4 But, even in that study, bevacizumab wasn’t continued until progression. It was merely continued for a longer period of time. In that one trial, in that metastatic setting, there was a suggestion that longer duration of therapy had greater activity, but most of our trials have simply not been designed to answer that question. In metastatic colorectal cancer, the phase III ML 18147 study randomized patients at progression to chemotherapy plus bevacizumab or chemotherapy alone. A recent press release cited improved survival with bevacizumab given beyond progression, so those data are anticipated.5
OncologySTAT: What do you we know about the effectiveness of angiogenic agents based on tumor biology?
Dr. Miller:Nothing. It’s a simple, short answer. We know nothing about the effectiveness of these agents based on tumor biology. We have looked in all of the breast cancer trials at various clinically based subsets. Whether you want to look more directly at tumor biology based on expression of ER or a triple-negative subset, or based on other measures of disease aggressiveness—whether that is number of sites of disease or previous disease-free interval—none of those different ways of looking at tumor biology have at all predicted the potential benefit from these sorts of therapies.
OncologySTAT:How much do we know about the mechanisms of resistance and, also, the potential for these agents to elicit tumor progression?
Dr. Miller: We know a lot about the potential mechanisms of resistance in laboratory systems and animal models. We know very little about mechanisms of resistance in patients. In the animal models, mechanisms of resistance that have been documented are simply that increased production of the ligand overwhelms the blockade and increases the reliance on other parallel proangiogenic factors with different signaling pathways that then still result in the same downstream effects. There can be cross-talk among different receptor families that bypasses the blockade. There can be co-opting of normal vasculature, rather than creation of a new vasculature. And there can be a phenomenon called vascular mimicry, in which a subset of the epithelial tumor cells can begin to express endothelial cell surface markers and can be incorporated into vascular channels with endothelial cells as a way of restoring perfusion. The control of vascular mimicry, though, is not well understood at all. None of those mechanisms of resistance have really been well-documented in patients.
There has been a lot of talk about this potential for antiangiogenic agents to elicit tumor progression. On that front, we have to separate what sort of agents we’re talking about. The suggestion that tumor progression occurs first came from animal models that used tyrosine kinase inhibitors (TKIs), which typically inhibited the tyrosine kinases associated with one or several of the VEGF receptors or the related platelet-derived growth factor receptors. Those TKIs typically have a fairly short half-life, and one of the biologic responses seen in almost all of the model systems to inhibiting the activity of the tyrosine kinase is increased production of the ligand—and the ligands tend to have fairly long circulating half-lives.
Imagine that you have essentially put on a short-acting break with the TKI—the response to that is much greater production of a long-acting ligand. And then you stop the TKI, which results in more rapid disease progression, because you now have all of this excess ligand in the system and you’ve removed the blockade. So, in animal systems with those short-acting TKIs, you see fairly rapid progression after the TKIs are stopped.
In contrast, bevacizumab is a monoclonal antibody that has a very long half-life, so there is not this quick on-and-off phenomenon. There also is not the same increase in unbound available ligand to stimulate the system. In animal models with bevacizumab, we’ve never documented this rapid progression after cessation of therapy. We have looked at those bevacizumab clinical trials to see if there is any evidence of more rapid progression after therapy is stopped and, in that way, we’ve been able to take advantage of those patients in the trials that I mentioned earlier who had toxicity from the chemotherapy that led to them discontinue treatment. In fact, some of those patients discontinued both components of their therapy rather than continuing either the bevacizumab or placebo until progression.
So, we can look at those patients to see if there is a difference in the rate of progression, the timing of progression, or the quality of progression in those who stopped bevacizumab or in those who stopped placebo. In fact, progression is faster in patients who stopped placebo than in patients who stopped bevacizumab. Not only is there no preclinical evidence of this increasing progression with bevacizumab, there is no clinical evidence of it either. This is another question that I’m frequently asked that just stymies me because there doesn’t seem to be any amount of data that we can generate, at least with long-acting monoclonal antibodies, that will make this question go away.
OncologySTAT: How far along are we in determining which patients will respond to antiangiogenic therapies? Are there promising biomarkers on the horizon?
Dr. Miller: I can tell you all the things that didn’t work. Our research in this area has been hampered in many ways; the most important of which is that a lot of the trials didn’t collect the tumors, and the trials that did attempt to collect tumors successfully collected tumors in only a subset of patients, and what was collected in the breast cancer trials was the tumor at the time of initial diagnosis. However, we weren’t treating patients at initial diagnosis, we were treating patients at the time of metastatic disease. It is not likely that these angiogenic factors remain static in their expression and in their relative importance over several years of disease progression and under the pressure of different therapies. So, we don’t have any studies that I’m aware of that actually collected a biopsy from the metastatic tumor that we are now actively helping to treat to see if there are biomarkers there that might be helpful.
We have been advocating for collection of metastatic tumor samples for a long time, but it has been much more difficult to do than getting tumor samples from initial diagnosis.
We’ve been focusing a lot of our efforts now on host factors rather than tumor factors, and have found that there are substantial differences in a patient’s inherited capacity to respond to various proangiogenic stimuli. Those differences, since they’re based on host inheritance rather than on tumor factors, are not likely to change. So, we are focusing more of our efforts on determining whether those factors might be predictive.
OncologySTAT:What insights can be gained from use of these agents in other cancers?
Dr. Miller:I think it tells us a lot about how we need to conduct research differently. It simply is not likely to be successful going forward to do broad-based clinical trials of unselected patients without collecting biologic samples of the disease actively being treated. We will need to develop the willingness and the clinical trial regulatory processes and infrastructures to allow companion biologic investigations to really answer those crucial questions of: who benefits? how do patients become resistant? and how could that lead to other new therapies?
I hope that it also results in a larger conversation about what really is an effective successful therapy. When you say that this is a therapy effective for patients with metastatic disease, what does that mean? What do patients want? Are the only successful therapies those that improve median overall survival? Well, if that’s the only marker of effectiveness that we’re willing to accept, then our clinical trials need to be designed differently, because none of the trials that we’ve talked about today were individually designed to assess overall survival. They were all very dramatically underpowered for an overall survival endpoint.
There are a lot of other markers of potential benefit, but we don’t have consensus and have really not had a larger conversation about what we want from our therapies. Without the answer to those questions, I don’t know how to design the next generation of trials, because it’s not clear to me what we’re shooting for.
OncologySTAT:It seems this is going to be an even more relevant question as clinical trials increasingly look at subsets of patients instead of the norm of conducting enormous phase III trials.
Dr. Miller:Absolutely.
OncologySTAT: Which trials currently underway appear promising in terms of new agents for combination therapy?
Dr. Miller: There are still trials with bevacizumab underway, and a new one is about to start in the metastatic setting. They are still not selected for any molecular phenotype, because we don’t yet have a way to do that. However, they do incorporate planned analysis of some potential biomarkers. Also, there are other ways of inhibiting angiogenesis, both in inhibiting signaling through VEGF and inhibiting angiogenesis that is not dependent on VEGF signaling. Several of those have now made the move into the clinic and still look very promising. Preclinical data that supports them are just as strong as they ever were, so those trials are absolutely still continuing.
References
1. Miller K, Wang M, Gralow J, et al. Paclitaxel plus Bevacizumab versus Paclitaxel Alone for Metastatic Breast Cancer. N Engl J Med. 2007;357(26):2666–2676.
2. Miles DW, Chan A, Dirix LY, et al. Phase III Study of Bevacizumab Plus Docetaxel Compared With Placebo Plus Docetaxel for the First-Line Treatment of Human Epidermal Growth Factor Receptor 2–Negative Metastatic Breast Cancer. J Clin Oncol. 2010;28(20):3239–3247.
3. Brufsky A, Bondarenko IN, Smirnov V, et al. RIBBON-2: A randomized, double-blind, placebo controlled, phase III trial evaluating the efficacy and safety of bevacizumab in combination with chemotherapy for second-line treatment of HER2-negative metastatic breast cancer. Cancer Res. 2009;69(abstr 42):495s.
4. Burger RA, Brady, MF, Bookman MA. Incorporation of Bevacizumab in the Primary Treatment of Ovarian Cancer. N Engl J Med. 2011;365(26):2473–2483.
5. Roche. Investor Update. Avastin-based regimen extends survival when continued beyond initial treatment in patients with metastatic colorectal cancer. Available at:
http://www.roche.com/investors/ir_up...2012-01-26.htm. Accessed January 26, 2012.
Hopeful
Linear Mode
