New Drugs—What's Important?
Abstract
Dr. Clifford A. Hudis is Chief, Breast Cancer Medicine Service, Memorial Sloan-Kettering Cancer, New York, NY. View profile.
The topic of what is important in new drugs has become especially heated and more controversial than ever before, given the sometimes modest improvements in conventionally recognized endpoints and the extent to which new drugs are expensive. Even while regulatory authorities have generally not been charged with evaluating cost effectiveness, as opposed to effectiveness, the issue of cost and what we are getting for the expenditure of resources is more acute than ever before.
With that as a background, I am going to address some selected aspects of new drug development, with the focus on what is most important for clinicians who are considering the incorporation of new drugs and agents into their practices.
I’m going to start by highlighting a list of potential concerns, which one should keep in mind: There is regulatory approval, meaning the availability of new drugs; there is cost; there is relative toxicity compared to alternative therapies; there is the impact, downstream, on subsequent treatment options to consider; and, then, there’s simple practicality—the day-to-day impact of one treatment option over another. A simple example is every–third week intravenous treatment versus oral daily therapy. Each requires a different kind of involvement and attention to detail on the part of the patients for a good outcome.
It is critical from the regulatory point of view, to remind everybody of the real role of the regulatory authorities. In 1962, there was a New Drug Application guidance provided via an amendment to the Food, Drug, and Cosmetic Act that required manufacturers to demonstrate the effectiveness of their products by conducting adequate and well-controlled studies. This, of course, immediately raises a values-based question about what constitutes sufficient evidence of effectiveness.
The types of evidence that we get can be considered on different axes, as well. For example, there is outright superiority defined by improved efficacy. Overall survival (OS) is a relatively obvious one, but without differences in OS, improvements in progression-free survival (PFS) may be in themselves important, even if they no longer are serving as accurate predictors, or surrogates, for OS.
Symptom-free survival is another potential mark of efficacy. But, here again, adequate assessment of symptoms and their duration and severity remain somewhat challenging for us. Overt measures of toxicity, like quality of life and standard clinical assessments of toxicity at the time of onset, all have their current weaknesses.
So, all of that said, once we see the differences between new treatment options and historical, or standard, ones we still have the problem of individualizing these results to our particular patients. I am not providing any answers here because I think these represent very challenging problems.
To highlight the somewhat unpredictable way in which we incorporate new clinical trial results, or perhaps, the very predictable way that we do it using subjective assessments, I would remind everybody of several recent examples of where incorporation of new trial results into the clinic has been somewhat nonstandard. On the one hand, when trastuzumab with chemotherapy was clearly shown to be better than chemotherapy alone in a single randomized phase III trial with a primary endpoint of time to progression, there was very rapid uptake of that agent by clinicians. A subsequent update showed that there was OS benefit, as well. But, nonetheless, this was quickly taken up. However, when doublets of chemotherapy (that is, two chemotherapy drugs versus one), were shown to offer similar kinds of improvements, they were not so quickly adopted. Very concrete examples are gemcitabine with paclitaxel and capecitabine with docetaxel. What this highlights is that, in a subjective way, clinicians are probably weighing the toxicities on the one hand and the appeal of the science on the other when they make decisions about using new drugs.
Another example is the drugs that have clearly improved PSF, but not yet OS. Bevacizumab is the obvious example, with multiple clinical trials demonstrating a fairly consistent, potentially clinically important, improvement in PFS; but, also, a fairly consistent lack of improvement in OS. This is something that has led, in some communities, to rapid uptake of routine use of bevacizumab, but not in others. Lapatinib plus capecitabine is a similar, albeit smaller scale, example of this phenomenon.
All of these examples taken together do not really convey, in my opinion, a single message. They highlight the fact that, despite similar kinds of results, clinicians will make very different choices in terms of the routine incorporation of novel treatment into practice. Some of the considerations that probably influence this include the linkage of the control group to currently existing standard therapies—how the new treatment fits into the existing treatment algorithms. The toxicity profile of a novel therapy and, again, the appeal of the clinical science are of some importance in this regard.
There are going to be any number of continued challenges to the existing order in terms of new systemic therapies, and our ability to rationally incorporate them into practice is going to be challenged. An upcoming example is the development of the Poly (ADP-ribose) polymerase (PARP) inhibitors. If the currently pending randomized phase III trial of the PARP inhibitor, BSI 201, combined with gemcitabine and carboplatin demonstrates superior PFS and OS over that of the control group, which is gemcitabine and carboplatin without the PARP inhibitor, some clinicians will quickly jump on this as a clear advance. On the other hand, there will be those clinicians who recognize and perhaps overemphasize that the gemcitabine/carboplatin arm is not a completely standard control group. While the addition of the PARP inhibitor may represent a significant advance, where it stacks up compared to other conventional options, like weekly paclitaxel, may be somewhat unknown.
I think, here again, final cost and toxicity will matter, as will the impact on OS. I do think that with improvements in OS, it becomes a little bit easier for a clinician to consider the incorporation of targeted drugs, especially when the toxicity is modest. I am emphasizing this because, as I noted earlier, we have had combinations of chemotherapy drugs that have clearly been shown to improve OS, but these have not really been rapidly accepted as routine standards in the clinic.
Overall, I think it is going to be very important for all of us to consider what appropriate endpoints we rely on when we design clinical trials, and it would also be useful for us to agree on standard endpoints across clinical trials. Until we do that, we will continue to have debates about the relative merits of new treatment options as they become available.
In that regard, I would highlight for you a much more conventional clinical trial result that will certainly be on everybody’s minds this year, and that is, of course, the results of the EMBRACE Trial comparing eribulin, a new—but in a sense old-fashioned— chemotherapy-type agent, with conventional standard therapies. In this case, using a novel trial design, a large number of such options were available. What provided a tremendous interest in this trial is the choice of endpoint, which was OS, and the fact that the researchers achieved a statistically significant improvement, even in this palliative setting with patients who had received extensive prior cytotoxic therapy.
The more nuanced point about this trial is that the investigators allowed the control group to vary with the clinicians’ choice, so that it reflected standard practice in the community, but it is not tied to any single control option. This kind of design, where OS is selected at the outset as the definitive endpoint of the trial, may be necessary, again, as we have continued discussion and debate about the meaning of various surrogates, like response rate and PFS. Moreover, it is much easier to quantify this kind of a benefit if it is real in the clinic. Therefore, patients who are confronting the alternative of more versus less toxic therapies, or more versus less expensive therapies, may have an easier time selecting a treatment option when OS is clearly different.
Looking forward, I would repeat myself by saying that one of the goals that we as a community might consider would be standardizing an approach to drug development such that certain endpoints are routinely accepted as evidence of superiority and, that the way in which the endpoint data are actually collected is done in a consistent fashion so that cross-study comparisons are somewhat easier facilitating clinical judgments regarding competing alternatives.
Click on any of these tags to subscribe to Topic Alerts. Once subscribed, you can get a single, daily email any time PracticeUpdate publishes content on the topics that interest you.
Visit your Preferences and Settings section to Manage All Topic Alerts