CancerGuide: Clinical Trials and Experimental Treatments
The Classic Phase III Trial: Primary Objectives
Phase III trials compare two treatments for a particular kind of cancer. Typically an experimental treatment is compared to a standard treatment. The usual objective is to see if the new treatment produces better survival than the old one. In some cases the objective is to show that a treatment with lesser side-effects is at least as good as the standard treatment. In this case, the treatment with lesser side effects doesn't have to produce better survival than the standard treatment to be considered an advance. It only has to be as good. Finally, Phase III trials are used to compare treatments in common use where there is significant uncertainty or controversy over which is better (Uncertainty or controversy does not preclude the evidence favoring one treatment or the other, though this isn't always the case). Clear thinking about the precise objective of a Phase III trial will help you decide whether the trial fits with your goals and will point the way to what questions to ask and what to research.
Almost all modern Phase III trials also compare "Quality of Life" (QOL) with the different treatments. QOL is assessed using standardized questionnaires which ask the patient subjective questions about how they are feeling and functioning. QOL can be helpful in balancing the side effects of treatment with any benefits in objective measures such as survival.
The Classic Phase III Trial: Standard Design
The standard Phase III trial randomizes the patients between the treatments being tested in the trial.
The Importance of Randomization: The purpose of randomization is to eliminate any potential systematic difference between patients in the arms of the trial. If patients or their doctors choose their treatment, there might be some difference in the patients who entered one arm compared to those who selected the other. If one treatment is more toxic, patients who are in better shape might tend to choose that arm, and they might tend to do better because they are in better shape. Similarly, if results with a new treatment were compared with historical results (called a historical control), there are many sources of systematic differences between the historical group and the group getting the new treatment. Such a difference could cause differences in the results of treatment, which could then be mistaken for differences caused by the treatment itself. A systematic difference could also mask differences between treatments. For these reasons, the randomized trial is almost universally held to be the gold standard for reliable conclusions about whether a new treatment produces real benefits. Keep this in mind when you comb through the medical literature!
Myths of Randomized Trials
There are some amazingly pervasive myths about randomization which you need to avoid!
Other Aspects of the Standard Phase III Design
Phase III trials typically involve many more people than Phase I or II trials, often numbering in the hundreds. Phase III trials are therefore expensive to run. Because survival is the usual endpoint, they can take a long time to accrue the planned number of patients and follow them for long enough to get a meaningful result.
Almost all Phase III studies are multi-center studies, so you probably have choices as to where to get treated. Many are available at dozens of places across the country! One center may be more convenient for you, or may be in your health plan, or may be more experienced with the treatment. If you are offered a Phase III trial, and it is not at the location you would most prefer, you should ask where else it is being done.
Not all cancer treatments that have been through Phase II trials actually make it into Phase III. The most frequent reason is that the Phase II trial does not show enough promise to make Phase III trials worthwhile. On the other hand, if the treatment is a "home run" in Phase II it may be approved and become standard without there being a Phase III trial at all. This would be the case if there is no reasonable standard treatment for the cancer in question, and could also be the case if the results from the Phase II trial are so excellent that any reasonable person would conclude it is better than the standard treatment without the need of conducting a Phase III trial. Finally, Phase III trials are very expensive to conduct and lack of funding is certainly a major reason for not doing Phase III trials.
The Classic Phase III Trial: Key Eligibility Rules
Phase III trials require that you have a specific type and stage of cancer, but beyond that the requirements are highly variable depending on the trial. Unlike Phase I and II trials, Phase III trials are not just for people with advanced, difficult to treat cancer. There are Phase III trials comparing different first line treatments for people with localized and highly treatable or curable cancer. This means that if you have just been diagnosed with cancer, and are interested in trials, you should investigate your options before you embark on any standard treatment. Phase III trials for advanced cancer may not require measurable disease, since the primary endpoint is survival rather than response rate. Phase III trials with a secondary endpoint of response may still require measurable disease (For an explanation of measurable disease see Steve's Strategic Guide to Phase II Trials).
Phase III Trials: Variations
In the cross-over design, patients whose disease worsens are switched to the other arm's treatment. This is an advantage for the patient, particularly if you prefer one treatment, since it means that there is a chance to get both treatments, regardless of which arm you are randomized to first. If you prefer one treatment to the other, this is still not as good as getting the treatment you prefer first, since by definition the disease will have progressed, and may be more difficult to treat by the time you get to the second treatment. It is even possible that the disease will have progressed so much that you are no longer are well enough to qualify for treatment in the other arm. I have seen cross-over trials where fewer than half the patients were actually able to cross-over.
In randomized trials where one arm is placebo, or no treatment, then of course, only patients in the no treatment group get a chance to cross-over. Since in the cross-over design, patients may get both treatments, overall survival becomes difficult to analyze as an endpoint. Instead, the typical endpoint for cross-over trials is the time until the disease becomes measurably worse, known as the Time To Progression. This is a less definitive endpoint than survival (See my article on Endpoints for more on Time To Progression).
Some trials have more than just two arms. While this does not radically change the nature of the trial for a patient, it does reduce the chance of being randomized into any one of the arms, and increase the number of treatments you have to evaluate to understand the trial.
Phase III Trials: Special Note on Standard Treatments in Phase III Trials
In trials testing a new treatment against a standard treatment, the standard treatment is called the control treatment. Some Phase III trials test two well known equally plausible treatments, and so don't really have a control arm.
Since there are often several different treatments in standard use, there are often several different choices for the trial designers for the control treatment. You should not assume that the control treatment in a Phase III trial is the standard treatment. It may only be a standard treatment. In fact you may prefer a different standard treatment from the one chosen for the trial for any number of reasons. Therefore, it's critical to evaluate all of the treatments in a Phase III trial, including those described to you as "standard".
The following uncomfortable discussion was motivated by an observation I made in late 2004 about trials for kidney cancer (what I had). I noticed that many Phase III trials involving new drugs choose Interferon as the control treatment rather than Interleukin-2 based therapy. Interferon absolutely is a community standard for kidney cancer and is widely used, but I think IL-2 based treatment (also a standard and widely used) is more effective, though with worse short term side effects, and although it is more expensive. I have no way to know exactly why Interferon is being chosen so often, but what I saw got me to think about where the incentives are.
There are some interesting motivations in choosing a control you should be aware of. If a pharmaceutical company is running a trial of a new drug against a standard treatment, the drug will be approved by the FDA if it proves better than the standard treatment. The FDA may accept any of a variety of standard treatments which is in wide use by the community and for which there is evidence as the control. If one of these treatments is less expensive or easier to give, the drug company has a motivation to choose that one for the control. A large multi-site community trial run by a cooperative group might have the same motivation. Less expensive might not be your top priority. It gets worse.
If, though arguable, one of the treatments in standard use is probably less effective, the drug company actually has a motivation to choose the less effective treatment as the control. Why? Well simply because it's going to be easier to prove superiority to a less effective treatment than to a more effective treatment. Drug companies also have a powerful motivation to run a trial which will convince the world, so they also need to choose a treatment which is at least credible as a standard treatment (even assuming the FDA would accept something less credible). Still, there is sometimes room to choose what will probably be an easier hurdle to overcome, and this is not in your interest.
Just because there are incentives adverse to the interests of patients in a trial doesn't mean anyone necessarily gave in to them. Furthermore, it's hardly always true that this situation arises. Often there is no slightly less effective, but still acceptable standard treatment. My advice here is not to be paranoid, but only to keep your eyes open and take care to evaluate control treatments every bit as carefully as the experimental treatment when trying to judge a phase III trial.
Phase III Trials: Strategies and Ethics
Dealing with randomization and the fact that randomized trials involve evaluating not just one, but at least two different treatments, significantly increases the complexity of decision making! Randomization, while held to be the best way of getting reliable answers, becomes ethically controversial if participants, including you, the patient, believe that the treatments in a trial are in some important sense unequal. Because of the complexity and controversy, I present a somewhat extended ethical and strategic analysis in the hopes of clarifying (or at least stimulating) your thinking about randomized trials and maximizing the odds.
Phase III Trials: Ethics and Equipoise
Randomly assigning patients to treatment would be unethical if it were known in advance that one of the treatments was inferior to the other, simply because intentionally treating patients with less than the best is unethical.
In order to ethically conduct a randomized trial, there has to be a belief that both treatments are equal, or, more accurately, given that the trial is being conducted because it is not known which treatment is better, it has to be uncertain which treatment is better. If there is sufficient uncertainty about which is better, equipoise is said to exist and the trial is considered to be ethical. As far as I can tell, equipoise is a rather fuzzy concept, and there is no clarity about just how much uncertainty there has to be in order for there to be equipoise. Just exactly who has to have equipoise to make randomization ethically acceptable turns out to be a very interesting question! Often individual doctors have strong opinions as to which treatment is better, but don't agree. It has been argued that it is ethical for a doctor to offer a randomized trial to his patient even if he personally thinks one of the treatments is better, as long as it can be said that there is uncertainty in the medical community as a whole (perhaps demonstrated in part by disagreement over which treatment is best). This is called community equipoise.
One basic question is how much the evidence can favor one of the treatments before the trial is unethical. If based on the best available evidence there is no reason to believe the evidence favors either trial you have the ideal ethical situation for a randomized trial. But what if the best estimate of the chance that treatment A is better than treatment B is 51%? Would it still be ethical to randomize? What if the chance were 75%? Or even 95%? I suspect few patients would be happy about being assigned to a treatment which has a 75% chance of being inferior compared to the alternative, much less 95%! Yet science and medicine demand a high level of proof, roughly akin to 95% probability one treatment is better than another, before it is considered to be reliably established. This is in direct conflict with maximizing the odds of survival, which demands going with whatever has the best evidence now.
I don't think there is any clear answer to the question of how much the evidence can favor one arm in a trial before it becomes unethical to randomize, nor can I discern any clear practice in the conduct of trials. One thing I am absolutely certain of is that you can decide how much uncertainty there has to be before you are willing to be randomly assigned! In the end, when it comes to your decision about a randomized trial, the only equipoise which matters is your own. If you believe the evidence substantially favors one arm of the trial, then you are not in equipoise. This does not necessarily mean the trial could not be a choice for you (see more in the section on strategies below) but it does mean you are very likely to be uncomfortable with being randomized and risking assignment to a treatment which in your view is probably inferior.
The equality of treatments is not a simple matter of equality of estimated efficacy. Treatments vary not only in efficacy, but also in other very important qualities, such as the nature and degree of the side effects and risks. Two treatments which truly have equal evidence of efficacy may not be at all equal overall depending on your personal values and goals.
In his excellent book on the development of immunotherapy, The Transformed Cell, Stephen Rosenberg, MD Ph.D. describes what I consider to be an extreme example of a randomized trial of treatments with radically differing characteristics In this trial, patients with bone sarcoma were randomized to either initial chemotherapy followed by limited surgery to save the affected limb or to... amputation! It was controversial and uncertain whether the limb sparing treatment was as effective as amputation, which was the standard treatment. Obviously, whether it's worth it to risk your life to save your leg is an intensely personal decision. No doctor or amount of objective evidence can possibly decide this for you! Scientific equipoise is not personal equipoise! If you were truly indifferent to these two radically different options, you might be happy to contribute to scientific knowledge and help future patients by taking part in such a trial, but otherwise you would want to choose. In my view, it is not inherently unethical to offer a randomized trial in circumstances such as these, as long as both treatments are otherwise available for patients who choose to choose, and as long as patients are very clearly informed of this. In this case, both treatments were available, and patients were indeed informed that they could obtain either outside of the trial. The trial had a happy result: the dispute was ended and the limb sparing option was found to be as effective as chemotherapy so that patients with this disease can now avoid amputation in the knowledge that they are not risking their lives to do so.
Phase III Trials: Strategies
The thought of having major treatment decisions made by a roll of the dice may raise your emotional temperature. But I believe the best approach is a cold blooded calculation of whether or not a given randomized trial is on the whole your best option, after fully considering the objective evidence in light of your own hot blooded personal values. Keep in mind that in a randomized trial it is critical to examine the merits of all of the arms of the trial, including any standard treatment arms, since you don't know which treatment you would get if you signed up.
I advise considering any option relative to all other options. Thus if you are offered a randomized trial, the choice isn't between taking part in that particular trial or taking standard treatment. There may be other trials and other standard treatments to consider. If you find there is a treatment which you consider more promising than any arm of some randomized trial, and which is unconditionally available to you, then you needn't consider the randomized trial further. There is no strategic difficulty.
Therefore, the rest of this section assumes at least one of the arms in the randomized trial you are considering is a treatment you think is your best option.
If you are in equipoise about this trial:
If you are not in equipoise about this trial:
Then you prefer one arm of the trial to the other arm.
Phase III Trials: Special Questions for Blinded or Placebo Controlled Trials
While as I described above, most phase III trials are not blinded or placebo controlled, if you are considering a blinded or placebo controlled trial there are some special questions you should ask:
This CancerGuide Page By Steve Dunn. © Steve Dunn
Page Created: March 3, 2002, Last Updated: October 3, 2004