It is the best of times; it is the worst of times: Is there a way forward? A plethora of treatment options for rheumatoid arthritis, but critical trial design issues


We as rheumatologists are currently trying to come to grips with the truly remarkable times we live in. Clearly, these are the “best of times” to provide clinical care for patients with rheumatoid arthritis (RA), since we have a seemingly ever-increasing plethora of effective therapeutic options for our patients. Currently, at least 17 different individual medications that qualify as disease-modifying antirheumatic drugs are available to prescribe, and at least another dozen are in development. If we use these 17 drugs individually or in combinations of 2 to 4 drugs as we often do, we have a spectacular but dizzying array of options (6,683 if you are counting, excluding the possibility of multiple biologic combinations). Therein lies the paradox; we have little or no idea of how to approach an individual patient to best utilize this vast arsenal that we now command. At first glance this is not cause for great alarm—many of these options are terrific, and surely soon we will have studies to help us utilize these in an appropriate and evidence-based way for effective and economical care of our patients … Wrong! Unfortunately, if we continue to perform clinical trials using the current models, we will never get where we need to be. To quote Irwin Corey, “If we don't change direction soon we will end up where we are going.”

What, then, are the paramount issues with current clinical trial design in RA? Why are we not getting the answers we as clinicians need? And what can we do about it?

The vast majority of clinical trials, especially in the US, are pharmaceutical trials (1–9). Paradoxically, this has brought highly effective and badly needed treatments to our patients while teaching us much about RA, but at the same time it has created a huge and widening gap in our knowledge about how to optimize these options to provide the best care. Pharmaceutical trials are designed as they should be, with companies' interests in mind; anything less would clearly be unfair to the stockholders. The trial reported by van der Heijde et al in this issue of Arthritis & Rheumatism is an example of this paradox (1). While it is nice to know that etanercept continues to work for at least 2 years and is well tolerated by patients who experienced an initial response, what is the real relevance of a 2-year comparison? In practice, with all the options available, no clinician should let a patient who has active disease continue to receive any given therapy months to years past the time at which the patient should have had a maximal response.

Following is a partial list of the current problems in contemporary trial design:

  • 1Almost all trials compare active therapy with placebo.
  • 2Rather than comparing treatment strategies, trials most frequently randomize patients to therapy A versus therapy B in a rigid manner and ask the question, “Who is better after 1 year? After 2 years?”
  • 3There has been little or no organized attempt to look at parameters (clinical combined with biologic) that predict response.
  • 4Essentially no meaningful data on cost effectiveness have emerged.
  • 5Substantial fundamental ethical issues exist, including the use of placebos (for which there are many proven options), lengthy continuation of failed therapies, and the timing of providing knowledge about treatment assignment to the patient and treating physician.

During the past decade, the most common design of a clinical trial has compared a given therapy with placebo in RA patients with active disease despite treatment with methotrexate (2–8). This has served us well and has brought to market many therapies that are indicated for RA; paradoxically, however, because of the many proven effective therapies (12 if you count only those that are commercially available [2–11]) for this group of patients, clinicians need comparison trials to optimally use these therapies. How helpful is it to have a 13th or 14th option for this group of patients when at last count there were zero reported studies that had compared active therapies with each other? How, then, can a clinician make an appropriate therapeutic choice among these 12 options for this common patient?

Clinical trials, in general, compare 2 therapies with each other in a rigid manner, often with fixed doses of the medications for a fixed period of time, sometimes as long as a year or two (1–11). Current best clinical practice demands that we change and/or adjust therapies in all of our patients to achieve at least a low level of disease activity if not remission (12–15). Further, we should make these adjustments as soon as we have given therapies an opportunity to be maximally effective—with most of our current therapies this is certainly no longer than 3 months or, at the most, 6 months. Why, then, are studies that compare therapy A with therapy B (or worse, therapy A versus a placebo) for a year or two relevant to clinicians when therapies that have obviously failed in many of the studied patients have been continued for months after we knew they had failed and would have been changed long ago if the same patient was being followed up in the clinic?

A decade from now, hopefully, we will no longer be interested in what happens in clinical trials to large unselected groups of RA patients. Questions such as, “What is the best therapy for earlier RA? Or for RA patients with active disease despite treatment with methotrexate?” will be largely irrelevant. We will have progressed to the point at which we will be able to profile individual patients; we will be able to select among therapies that we know will have a very high success rate because of the patients' genetics, their cytokine profile, their autoantibody profile, the initial trigger of their disease, or a combination of these factors (and probably many others). Armed with this information and the literally thousands of options clinicians already have, we will be poised to take the next quantum leap forward in therapy for our RA patients. Unfortunately, many clinical trials have not attempted to address these questions because of the lack of biologic specimens (serum, DNA, RNA, synovial tissue, etc.) that are married to the clinical data. Sponsors are not interested in these answers because of the substantial expense involved or a myriad of other reasons. Until this happens, we will continue to spin the therapeutic option wheel.

Ultimately, we as clinicians will be asked to prescribe the most cost-effective therapy for all of our patients. This is already happening in many countries and in many health care delivery systems. We are particularly ill-equipped to address this concern—essentially no meaningful data on the cost effectiveness of our current therapies exist from clinical trials that in any way mirror clinical practice. Most clinical trials have compared active products with placebos, and any cost-effective analysis that compares a therapy with a placebo (if other effective treatments are available) is at best highly suspect. Unfortunately, if good cost effectiveness data are not available, many systems may force us to use the least expensive therapy, which could have grave consequences for our patients and ultimately for society as we bear the burden of the increased indirect costs that will certainly result from such an approach.

We should have grave ethical concerns about the current “accepted” clinical trial design (1–9, 11). Major concerns with the current “standard” include the use of placebos, the duration of continuation of failed therapies, and the timing of providing treatment information to patients and clinicians subsequent to study completion or withdrawal. Placebo controls are currently widely used in clinical trials in situations in which many proven therapies exist, and guidelines and trial data clearly show that patients would benefit if these therapies were introduced earlier. Many experts strongly advocate that aggressive therapy be used very early in the treatment of RA, while at the same time taking part in trials in which placebos are continued for months to years. With our current state of knowledge, how do we justify continuing a failed therapy for 6 months or more past the point at which it should have been maximally effective? Finally, patients and their treating clinicians are often blinded to the therapy they received in a clinical trial for months or even years after they have reached the primary end point of the trial. How do we justify withholding this vital information from patients and their treating physicians?

How can we change the direction in which we are going and truly allow our patients to reap the benefits of the remarkable therapeutic options we currently have? Pharmaceutical trials have given us effective treatment options, but they can be expected to take us only so far. Clinical trials funded by agencies (the National Institutes of Health, the Canadian Institutes of Health Research, the Veterans Administration, insurers, governments, etc.) that do not have a stake in the outcome are critically needed to supplement pharmaceutical trials. Trial design of the next decade needs to catch up with best clinical practice and be designed to more closely mirror clinical practice. Examples of this would be designing switches of therapies within a trial (in a blinded manner) for patients who have not achieved an optimal response in a reasonable period of time, or requiring patients who do not achieve improvement within a reasonable period of time to be removed from the trial and counted toward a failure of the therapy. If active therapies are compared in randomized blinded clinical trials and appropriate data are collected, we will for the first time have the opportunity to collect meaningful cost effectiveness data. When active therapies are compared, we will need to consider equivalence trials and come to grips with what clinical equivalence means, especially in the context of the cost differential of many of our therapies. Trials must be married to biologic repositories so that genetic and serologic parameters can be closely examined to allow us to ultimately profile patients so that future patients will get the most effective treatments in a timely manner.

We may not be dealing with something as historically significant as the French Revolution, as Dickens was with the title quote, but we badly need a revolution in trial design. If we are successful in getting trials funded that come to grips with some of the issues raised, we will truly be in “the best of times.” If we can not, and if we continue in the direction we are headed, we may indeed “end up where we are going” with more options for treatment than we have any idea what to do with and payers telling us we must use the cheapest until we can provide evidence to the contrary.