By: James R. O’Dell, M.D.
University of Nebraska Medical Center
Dr. O’Dell has no significant financial interest or relationships to disclose.
It really is the best of times to be a physician taking care of patients with rheumatoid arthritis (RA) and no one would question that. There are so many treatment options available: drug products and biological products, etc that were never dreamed of just a few years ago. However, there are some clouds on the horizon. There are gaps in our ability to take the available options and really apply them to patients in any kind of a scientific way. A good question to address is what do we need to do to get where we would like to be a decade from now.
Participants will be able to:
- State the importance of measurements.
- Restate the flaws in current trial design.
- Discuss the ethical concerns of current clinical trials.
Over the last 20+ years, thinking about RA and caring for patients with RA has changed drastically. We have a lot of tools now that we didn’t have twenty years ago.
Paradigm shifts of the last decade
Early Treatment is Critical. This wasn’t so obvious until the last 10 years or so how really important this is.
The Use of DMARD Combinations. In the mid-90’s nobody used combinations of disease modifying drugs to treat RA. A patient would be prescribed one drug; if it stopped working; lost its efficacy, a new drug would be prescribed. Rheumatologists are very timid. When we started our trials, we had a lot of trouble getting people to accept the concept of being more aggressive and we had to have INDs and the FDA made us use sulfasalazine at smaller doses than we would ever think about today because people were scared of combinations. Today combinations of DMARD are used in vast majorities of our patients.
Biological Therapies. Biological therapies have obviously been a huge change in the last decade
The Critical role of comorbidities. One of the big things also that has become hugely apparent over the last five years is the importance of comorbidities; chief among these cardiovascular: comorbidity of all our inflammatory diseases.
RA Treatment: A Remarkable Success Story?
Here we have a 56-year-old woman who has sero-positive, erosive RA. She has 17 tender and swollen joints. She gets treated with methotrexate (MTX) and they pushed the MTX dose. They switched it to parental when it wasn’t working completely, which is something many rheumatologists forget to do and should remember. She still has many active joints. But, she is an ACR70. So she is 70% better in a composite way overall and in any trial she would be labeled as one of the successes. This highlights a huge problem we have in clinical trials and that is, should we really consider this patient a success?
So what is wrong with that picture? Although a lot of patients would be happy that they only have five joints that are tender and swollen instead of 17, rheumatologists would not be happy with that. We are not happy when someone has five tender and swollen joints in the clinic. There is a big problem with what we measure as success and it begs the question of what the goals should be.
Where is the hemoglobin A1C; the LDL level; the blood pressure level that we should be targeting in our disease?
Lord Kelvin (left) had some thoughts on measuring things. “When you can measure what you are speaking about, and express it in numbers, you know something about it; but when you cannot measure it [and] express it in numbers, your knowledge is of a meager and unsatisfactory kind.”
He was not talking about measurements in rheumatoid arthritis but it certainly applies. He also said, “if you cannot measure it you cannot prove it”. That says a lot about the measures of success in rheumatoid arthritis. It lends itself to how we define success in the clinics, what we are shooting for.
What is the goal of therapy?
In clinical trials it may be an ACR 20, 50, or 70, or a disease activity score of DAS 28, or some radiographic outcome. These are very different measurements. In the clinic it is probably just a Gestalt for most of us. Particularly in this country, not too many people do DAS 28s in the clinics. There are other clinical measures–GAS, Rapid CDAI–that are evolving and need to continue to evolve if we are going to get to a uniform goal for patients.
ACR Core Set
- Physician global estimate
- Functional status – health assessment questionnaire (HAQ) multidimensional HAQ (MDHAQ)
ACR Core Set and Disease Activity Score (DAS)
- Tender joint count
- Swollen joint count
- Acute phase reactant – ESR, CRP
- Patient global estimate
Above are the 7 measures that are the core set that makes up the ACR 20, 50 and 70; and those used in the DAS. The problem with the ACR set is that it measures a percent improvement. DAS measures an absolute level of disease activity, but it has a lot of problems too. There is nothing in the DAS about functional status; it is much too sensitive to changes in sedimentation rates; painful joints are overemphasized compared with swollen joints.
DAS28 = 0.56 * square root (tender joint count 28) + 0.28 * square root (swollen joint count 28) + 0.70 * log e (ESR) +0.014 (patient assessment of global status or activity)
The DAS is a complicated mathematical formula but that is not the reason it is not useful to us in clinic. The reason that it is not useful to us in clinic is that most people don’t do formal 28 joint counts and even if they do, they don’t have sedimentation rates or CRP’s done at the time they are seeing the patient. There is also a CRP DAS that can be calculated.
RA treatment: questions for the next decade
The ACR put out an RFA on a new measure that would be able to be used in the clinic and that it would be measuring an absolute level of disease activity. The ACR hybrid is neither, but it is a better measure of outcomes in clinical trials and that is an important point. In fact it is good enough that if it’s used as an outcome in clinical trials, the number of patients used can be halved. It is an improvement in trials, but it is not going to be helpful in clinics because it can never be a goal for an individual patient because, again, it is based upon a percent improvement and that is a huge problem to get around.
To highlight some of the problems with the way we measure things, here is a study that looked at a combination methotrexate versus adalimumab versus methotrexate alone. In this case, one would have been much better off picking methotrexate compared with adalimumab. There is a statistically significant difference between methotrexate and adalimumab when measuring ACR 20. So for measuring signs and symptoms of RA, one is better off with methotrexate in terms of the ACR 20, 50 and 70. However if comparing xrays, one would be better off choosing adalimumab because the radiographic progression after two years is better in the adalimumab group compared with methotrexate alone. So what do we choose? Do we want the patient to be clinically better and have less tender and swollen joints? Or do we want them to have less radiographic damage? Again, we’re questioning what we are measuring in clinical trials and how we define success.
This is a patient who has very deforming arthritis. You may recognize this patient from the ACR collection. Here are the x-rays on that particular patient. Sharp score 0. There are no erosions there. Because of the subluxation that may be called some joint space narrowing but there is truly no joint space narrowing. So this patient has deforming arthritis but no Sharp score progression.
One of the problems with clinical trial interpretation is not knowing what to do with radiographic outcomes. Disability can and does occur without ever having any xray changes. So again, what should we really be measuring in clinical trials? This is a patient with SLE, but it could be a patient with RA who has significant deformities because of all the synovitis. This is much more debilitating for the patient than the progression of a few Sharp score points.
What is the goal of RA treatment? Unfortunately we can’t cure RA right now, but remission is a very realistic goal and should be our goal with most patients. If we treat patients early, then we should be able to achieve remission in a significant percentage of our people. What is significant? Significant depends on how we measure remission.
The TICORA trial was done in Glasgow and it asked the question: what happens if we have one set of patients that we treat to a goal (intensive care) and another set of patients that we treat with the best care (routine care)? The goal was the full DAS of less than 2.4. They didn’t use biological therapies in this trial; the therapy followed a progression:
Patients fared better with a goal.
5 Tender, 4 Swollen, ESR = 22
1 Tender, 0 Swollen, ESR = 7
What does this tell us? It tells us that if we had a goal for our patients and we made decisions to get to that goal—escalating therapy when we need too—all of our patients would do better. Unfortunately DAS is cumbersome, so we need something in clinic that is easier to follow.
Global Arthritis Score. The global arthritis score (GAS) is based upon measurements that the patient performs. This is a self-recorded tender joint count and is very easy to do. This may not be what we should be measuring in clinic, but we need to validate an outcome measure that is easy to use in clinic. The GAS has been shown to correlate very nicely with DAS changes in most cases.
What should the goal of our therapy be? Or how de we define success?
- absolute level of disease activity vs percent improvement
- clinical vs radiographical
- easy vs cumbersome
- real-time vs tomorrow or next week
Back to the 56-year-old woman: ACR 20 response but five tender and five swollen joints. Rheumatologists are not happy with the outcome, but there is no trial data that shows we can do better in this patients with only five tender and five swollen joints. All of our clinical trials have been based upon people with much more active disease. At minimum, to be in cilincal trials patients have six tender and six swollen joints. Most of the patients seen in clinics don’t qualify for clinical trials. The main reason they don’t qualify is that they have too little activity of the disease. That makes it difficult to extrapolate knowledge from clinical trial data and apply it to the care of an individual patient.
ATTRACT trial. The ATTRACT study was intended to study patients with aggressive disease who were inadequately responding to methotrexate. Patients had active RA, despite methotrexate treatment, defined as at least 6 swollen and tender joints and at least 2 of the following: morning stiffness of 45 minutes or more, erythrocyte sedimentation rate of a minimum of 28 mm/hr, or a C-reactive protein of at least 2 mg/dl. All patients had to have been treated with methotrexate for at least 3 months at a minimum stable dose of 12.5 mg/week for at least 4 weeks prior to screening and no other concomitant DMARDs were allowed during the trial. Patients were, however, permitted to receive stable low-dose corticosteroids at less than or equal to 10 mg/day and non-steroidal anti-inflammatory drugs.
The patient population in ATTRACT was reflective of a typical population of patients with active RA, having a median age of 54 years, female predominance and a majority of patients with positive rheumatoid factor. Patients had been on a median of 3 prior DMARDs, including methotrexate, with a range of 2 to 8. Patients were on therapeutic doses of methotrexate prior to study entry, with a median dose of 15 mg/week. Despite being on therapeutic doses of methotrexate at baseline, patients had active disease. They had a median of 20 swollen and 31 tender joints as well as a median CRP of 2.6 mg/dL. They also had substantial disability at study entry, indicated by a median HAQ score of 1.8.
Our patient began with 10 swollen and 10 tender joints, so she is going to qualify for all those other trials. She is still an ACR 40 and that is a success in most of our trials, but 10 tender and 10 swollen joints is bad. Clinical trials do not tell us how to treat this patient either; there are no trials that compare active therapies to each other in patients with active disease despite methotrexate. There are a lot of choices. All of the data that is available to us in the world literature about which of those therapies work better than the other one don’t indicate when to use them. In other words there is no data out there on the particular important question.
How soon is soon enough?
Early therapy is important. How early, we don’t know the answer to, but it is an important question. If a patient says, “Doctor, no matter what, I need to be able to go to work tomorrow”, which of our therapies are going to work quickly?
This shows the first 14 patients who were ever treated with steroids in the world (Mayo Clinic, 1949):
|ESR Start||ESR Finish||ESR Response|
They all had RA. RA was selected as the first disease to treat because it was thought that RA was a steroid deficiency disease. Some people still believe that and there is some evidence that might support that. Within a month, 100% of these patients had an ESR of 50. In other words, their sedimentation rate was about 50% better. Eighty percent of them had an ESR of 70. One therapy that works quickly is steroids.
What about more modern day data? The Cobra trial compared participants that received 60 mg of prednisolone each day for a week and then tapered fairly rapidly with a group that received sulfasalazine alone. The results showed that if the steroids were administered up front, the ESR got 75% better within two weeks. Sulfasalazine alone did the same thing but it took it six months. TNF inhibition works very quickly too. Those are the two options for a response within days or a week.
Are current trial designs relevant to current future or clinical practice?
What is the current standard trial design?
- Vast majority of trials especially US are pharmaceutical trials
- Most trials compare active therapy to placebo
- Randomize patients to A vs B in a rigid manner and ask question who is better after 1 year? 2 years? Or heaven forbid even longer!
- Little or no attempt to look at parameters (clinical or biological) that predict response
- Essentially no meaningful data on cost effectiveness
- Ethical issues: Placebos, duration of failed therapies, timing of knowledge about treatment assignment
First of all, the vast majority of trials, especially in the United States, are pharmaceutical trials. Because of pharmaceutical trials, we have so many options for patients that we didn’t have a decade ago. However, pharmaceutical trials are designed to ask a specific question by a group of people that have a vested interest in the outcome. They do not necessarily ask the questions that are most relevant for rheumatologists to take care of patients in clinic. Pharmaceutical trials are necessary and important, but there is a need for other trials that ask other important questions.
Most trials in patients with active disease despite methotrexate compare active therapy to placebos, which is ethically unconscionable. Why should we be doing yet another trial that shows us the 13th therapy works in that group of patients with suboptimal response to methotrexate when we know we have 12 ways to make that patient better? Most trials randomize patients to therapy A versus therapy B in a rigid manner and ask the question: who is better a year later or two years later? This is not how we take care of patients in clinic. If a patient is not doing better at three or six months they will switch to something else We need to do something different about trial design to get around this particular problem. There has been little or no attempt to look at parameters, clinical or biological, that predict responses.
In 2017 we are not going to be talking about RA the same way we talk about RA as a disease. We are going to be talking about subsets of RA and how those subsets are going to be defined is the very important question that we don’t know the answer to. Hopefully, we are going to have trial data that is married to biological repositories that will address the question of how can we predict who is going to have a 90% success rate on what therapy. There is essentially no meaningful data of cost effectiveness in our diseases. How can comparisons between doing something that is active versus something that is a placebo give accurate cost effectiveness data? We need trials that are comparing active therapies to each other. Until we have those, we essentially have no useful cost effectiveness data.
The use of placebos in trials is a huge problem, it helps answer some questions but it is unconscionable from the patient’s standpoint. The duration of failed therapies – we need trial design that lets patients come out of trials or lets them switch therapies at an appropriate clinical time. If we don’t do that, then the results of that trial aren’t very helpful to you we you try and take care of a patient. We need timing and knowledge of the treatment assignment. Another thing that is unconscionable is physicians and patients enrolled in trials not finding out the therapies administered until all participants are finished. That knowledge could have helped the physician understand how to treat that patient. If a patient comes out of the trial because of side effects, the code is not broken until everyone is out of that trial unless there are extreme situations. That is not the way we ought to be taking care of our patients. There are a lot of ethical issues involved in having patients participate in clinical trials.
Current Study Designs
ATTRACT Study Design and Methods. The pre-specified, primary and secondary endpoints at weeks 30, 54 and 102 in the ATTRACT study are shown here. The primary endpoint at week 30 was the proportion of patients achieving an ACR 20 level of response in each of the 5 treatment groups. The primary, week-54 endpoint was degree of inhibition of structural damage progression as measured by the change from baseline in the modified Sharp score. At 102 weeks, the primary endpoint was improvement in physical function as assessed by the change in HAQ over time. Secondary endpoints included change in modified Sharp score at weeks 30 and 102, ACR 20 response rates at weeks 54 and 102, and change in HAQ over time through week 54.There is no reason that we should be following someone for 54 weeks or 102 weeks if they haven’t done well on the therapy that we selected. They should be out of this trial as a treatment failure much, much earlier than that. We shouldn’t be putting patients at risk.
TEAR (Treatment of Early Aggressive Rheumatoid) trial. It is an investigator-initiated trial. It started out an NIH trial but didn’t end up that way. This is a trial where we are looking at patients with early disease and we are randomizing them to either the strategy of triple therapy or the strategy of methotrexate plus TNF inhibition, but only half of the patients are getting that therapy up front. The other half in both groups are just being started on methotrexate and then they are only stepped up to that if they have active disease, which is kind of the way we do things in clinic. The question then is what happens after two years.
RACAT (Rheumatoid arthritis comparison to active therapy) trial. This is a trial that is done in the VA with the collaboration of the Canadian Institute of Health, CIHR. It addresses the comparison of active therapies in methotrexate suboptimal responders by looking at a strategy. The question is based upon whether you started on methotrexate, all these patients are on methotrexate, sulfasalazine hydroxychloroquine or etanercept in the beginning. Based upon what one you started on what happens after a year, in terms of getting a DAS 28 below 3.2 (low level of disease activity). The therapy used isn’t the main point; it’s how well the patient is doing. These patients at 24 weeks will switch over to the other therapy if they haven’t had a clinically significant improvement, which is defined as a 1.2 on the DAS 28 in a blinded fashion. It is all done in a blinded fashion. At the end of the day some of these patients who originally got sulfasalazine hydroxychloroquine will actually be getting Etanercept. At the end of this day some of the patients might be on triple therapy.
Strengths of the RACAT trial:
- First trial to compare active therapies in methotrexate suboptimal responders
- Tests a strategy not fixed/rigid approach (single head-to-head comparison despite suboptimal results)
- No pharmaceutical funding, influence on design or data interpretation or control of the data
- Large concentration of men
- All study medications supplied by the VA
- Robust economic analysis
- Serum and DNA bank
How can we profile patients effectively?
This is a huge problem, and this is where we are going to make a lot of headway in the next decade. If I am making this decision for a new patient in my clinic, how do I treat patients with RA? I have 6,683 choices I can make. We have 17 disease-modifying drugs, six of which are biological. Consider that it is not unusual for us to use four disease-modifying drugs, so if you go through the math on that, in terms of using all those drugs individually, using two at a time, three at a time, four at a time that is 6,683 possible choices. I can’t spin a wheel and pick, I need a lot better tools to help me make that decision for our patients and when I have those tools I will be a lot smarter. How are we going to get there?
The shared epitope has been predicted in some models. CCP and CCP isotypes might help us, serum cytokines might help us, and other genetic markers might help us. There are many candidates out there, but not a lot of information right now. But that is one of the ways we are going to make quantum leaps forwards.
Ideal situation: 2017:
- Individual RA Treatment Profile
- Serum: IgA and IgG4 CCP positive, MMP3 –high, sTNFR – low
- DNA: DRB1 -0401/X heterozygote, MTHFR –C/T
- RNA: Not required unless shared epitope negative
- Synovial Bx: Not required unless active disease at 4 months
- LDL and HDL: 138 and 37
- Answer: Two weeks of anti-TNF X397 (dose of 374 mg q week) and maintenance therapy with methotrexate 18 mg q 8 days, Statin 242 45mg/d
Can we treat rheumatoid before it happens?
Yes, it is going to happen and it is going to happen in most of your lifetimes. CCP antibodies and rheumatoid factors are present long before disease happens, so immunologically you get RA five years before you show up in the clinic. There are studies now that are tracking those people and watching them develop RA and defining the earliest changes in terms of autoantibody production. Those autoantibodies are allowing us to predict who is actually going to develop RA with high degrees of specificity.
Those same studies are looking at cytokine changes or other things that may be even earlier changes in RA.
What about critical co-morbid conditions?
That is a huge issue because it is why patients die. They die because they have coronary artery disease and strokes, because of the inflammation. That may be one of the important variables in answering how soon is soon enough. Sherine Gabriel has elegant data that shows the heart disease of RA starts before the clinical RA. It makes sense that they have this inflammatory process going on years before they develop RA. I think we are doing a much better job of this, we are really paying attention to the risk factors that our patients have for those conditions. We are using statins, which may help their rheumatoid a bit as well, but we need to do an even better job.
How will we pay for all of these therapies that our patients are going to need?
Unless we have the data that shows which therapies which patients need and whether they are cost effective or not, insurance companies will shy away from payment. But even with that information it would be very reasonable for them to come to us and say “you have to try x, y and z before that patient can get on TNF therapy”. The data comparing x, y and z to TNF therapy needs to exist to fight the insurance companies.
Some of the problems with where we are right now may indicate some of the worst of times. However, it is still the best time to be a rheumatologist and take care of patients with RA. We did not have 6,683 choices for initial therapy a few years ago and that is exciting. It is dramatically different in clinic now, compared to what it was like a decade or 20 years ago with regard to the number of wheelchairs, the number of people that had C1, C2 problems, the number of people that had vasculitis and chronic nonhealing leg ulcers, etc. We don’t see a lot of that anymore. It is dramatically different, we have made a lot of progress.