August 30th, 2011
Is Warfarin Still the First Choice in Atrial Fibrillation?
Stephen Fleet, MD
There’s an old maxim in medicine that one shouldn’t be the first to prescribe a new drug, nor the last.
A fascinating debate between Michael Ezekowitz from the U.S. and Felicita Andreotti from Italy highlighted the differences between warfarin and the newer oral anticoagulants (NOACs) apixaban, rivaroxaban, and dabigatran.
Even Dr. Ezekowitz, the warfarin protagonist for purposes of the debate, had to concede the superiority of the NOACs, now supported as safer and more effective in three clinical trials: RE-LY, ROCKET-AF, and ARISTOTLE. In general, the factor Xa inhibitors prevent ischemic strokes, reduce mortality, and limit the incidence of dreaded intracranial hemorrhage, with less bleeding as well.
How, then, could a cardiologist not rush to the electronic prescription pad and immediately take most patients off warfarin and prescribe NOACs?
Well, for one thing, there’s the cost. True, INR monitoring comes with societal and health care costs, plus the cost of paying for excess strokes. The patient, however, is more concerned with the copay: “Do I have to pay $10 per month for a generic drug or hundreds for a newer medication?”
There’s also the question of patient subgroups. Until we have more information, subgroups such as patients with prosthetic metal valves will need to remain on warfarin. The current lack of a specific drug antidote to the NOACs concerns many observers as well.
I also find intriguing the way statistical data are presented. For example, in the ARISTOTLE trial, death occurred in 3.52% of the apixaban group and 3.94% of the warfarin group. You can declare, then, that by using apixaban instead of warfarin 8 deaths are prevented for every 1000 patients treated. On the other hand, you could say that 96.48% of the apixaban group and 96.06% of the warfarin group didn’t die. Is that really a compelling difference for the individual patient sitting before you in the exam room?
For now, I think I’m going to continue prescribing warfarin for patients who are already well controlled on that medication. For those with poorly controlled INRs or new patients, perhaps the time for NOACs is now.
20 Responses to “Is Warfarin Still the First Choice in Atrial Fibrillation?”
NEJM Journal Watch — Recent Cardiology Articles- Anticoagulation in Patients with Cancer-Associated Pulmonary Embolism: How Long is Long Enough?
- Spironolactone Use After Percutaneous Intervention for Acute Myocardial Infarction
- CRISPR/Cas9 Gene Editing for Transthyretin Cardiomyopathy
- Does Colchicine Reduce Cardiac Events After Myocardial Infarction?
- First-Line Therapy for Ventricular Tachycardia After MI: What's the Best Approach?
I would not make mortality predictions, as mortality was not a primary endpoint in any othe trials. The only outcome argument to switch warfarin patients with High TTRs is a lower intrracranial bleed rate, which appears not to interact witj TTR (at least with dabigatran).
Competing interests pertaining specifically to this post, comment, or both:
BI speaker in past
I suppose I would prefer a rate of NOT having an ICH of 99.76% a year to 99.53%, but does it really matter in the individual patient?
Impute placebo into the comparison of apixaban and warfarin. We know that warfarin is dramatically better than placebo from the old atrial fibrillation trials, and dramatically better than aspirin. Apixaban would positively shine against placebo and the mortality reduction would be dramatically impressive — the only NNT the patient cares about is whether they get the drug (either warfarin or apixaban) or they do not (placebo). Incremental NNT’s are misleading.
Agree that the use of some anticoagulant is more important than which anticoagulant, but most clinicians would prefer, in theory, to use the better product.
Consider the competing choices:
-drug A interacts with about 30 different classes of medications; drug B does not
-drug A requires titration, nomograms, blood tests, physician visits, trips to the pharmacy, dose changes; drug B does not
-drug A requires dietary restrictions; drug B does not
-drug A requires an entire specialty and clinic set up to “optimize” care; drug B anyone can prescribe
-drug A levels may vary when the patient is sick, gets prescribed an antibiotic, changes what they eat, and the consequences of this variation may be quite severe; drug B does not
-drug A must be stopped at least 5-7 days prior to surgery; drug B just hours for wash-out
-drug A carries the very rare consequence of induced skin necrosis; drug B does not
-drug A has higher mortality, more total hemorrhage, more intracerebral hemorrhage (the worst type of stroke, with highest rates of functional dependence, mortality and nursing home disposition); drug B has less mortality, less total hemorrhage, less intracerebral hemorrhage
-drug B prevents ischemic stroke and systemic embolic events better than drug A
-drug A efficacy depends to a great extent on “black box” metabolizer polymorphisms in a given patient as well as hepatic function, general state of nutrition and so forth; drug B does not.
Given these parameters, which drug would you, as the physician, want to start today for your atrial fibrillation?
Dan,
Familiarity (over 50 years of experience) and cost (inexpensive by several-fold) of drug A are important considerations that deserve a seat on the table amongst the otherwise very comprehensive accounting of attributes you listed.
The choice of NOAC’s over warfarin should be obvious when you read Dan’s thorough accounting; yet the low copayment, familiarity, and easy management of many patients tend toward sticking with warfarin in those already on it and doing well. The warfarin clinics are not suddenly folding.
Dear Sanjay and Stephen,
I agree with all your comments. Economic issues aside, it would seem the NOACs are superior to warfarin. My gosh — how many ischemic strokes and hemorrhages have I seen in patients who are either under- or over-anticoagulated with warfarin (respectively)!! This has been amply documented in the literature as a systemic problem with warfarin use in the ‘real world’ (outside of trials). Why do only 50% of eligible AF patients in the community get warfarin if it’s so easy to use and physicians are so familiar with it? The reason is not fear of bleeding but the true difficulty of stickhandling this most narrow of therapeutic index drugs.
How long it has been available is possibly irrelevant – digoxin has been available for over 250 years (a la Withering) and has seen its footing substantially lost in the management of heart failure and atrial fibrillation with the advent of better drugs, devices and procedures.
The reason the warfarin clinics are not suddenly folding is purely for economic reasons (related to poor update of NOACs) and has nothing to do with efficacy or even cost-effectiveness (when you consider INR checks, drug interactions, phone calls, bleeds, pharmacy scripts, strokes in the under-anticoagulated, etc). Our provincial payer still will not reimburse prescriptions for this agent — again, it’s a purely economic phenomenon.
As physicians are we charged with providing the best care for our patients or is our priority to reduce costs for insurers, even if that means providing a clearly inferior product? Granted adherence is going to be a problem if patients have to pay out of pocket for these agents and the importance of this therapy, despite its cost, is not carefully explained to them.
Dan,
It is important to keep in mind that dabigatran was not granted a claim of superiority over warfarin. FDA is loathe to allowing superiority claims on the basis of noninferiority trials which rest on a very unstable foundation, and dabigatran did not beat warfarin in patients with optimal target INR. The take home message is that in patients with well-controlled INRs, there is really no advantage in switching to an expensive agent.
Rivaroxaban is not approved yet, and won’t be surprised if it has the same set of regulatory and marketing challenges as dabigatran.
As far as apixaban is concerned, I will reserve my judgment until I have seen the FDA’s primary analysis and review. I have seen enough disparities between the published original report and the latter to warrant prudent skepticism.
Lets wait and see how clear or fuzzy the NOAC landscape is going to look like in the near future!
Just for the purpose of clarification, the article made reference to Dabigatran being a Factor Xa. That is incorrect. It is a DTI (Direct Thrombin Inhibitor) Different MOA.
The recent editorial accompanying rapid online publication of ARISTOLE pointed out that in all three of the NOAC vs warfarin trials, intracerebral hemorrhage was reduced in the NOAC arms (significantly so), with mortality either statistically significantly reduced or nearly so.
It would be interesting to imagine a scenario in which all three NOACS were approved therapies and warfarin was being trialed as a new experimental agent (in RE-LY, ARISTOLE and ROCKET-AF). This is easy to envision as we just take the reciprocal of all the point estimates. Across all three trials, warfarin increased intracerebral hemorrhage and in ARISTOLE, mortality was significantly increased for warfarin. Would warfarin be approved under these conditions? Particularly given its many drug interactions and inter- and intra-individual variability in drug response?
What I am getting at is that there is a bit of a double standard at play here. Just because we have a cheap, established agent with lots of clinical experience and many well-controlled asymptomatic patients on it for years without problems does not mean warfarin is a good therapy (by the way, I hope these well-controlled patients do not go on vacation and get an antibiotic for a URTI, cellulitis or UTI from a physician who does not know them well – the results can be catastrophic). Presumably, the many irritants and drawbacks of warfarin use are what has driven the development of so many new oral anticoagulants in the past 5-10 years.
Had there been a placebo arm in any of these three large clinical trials there would be no question of superiority (admittedly, we can’t do that – both for ethical reasons and because the regulator requires incremental improvement on the current standard of care, namely warfarin). Yet the NNTs and ARRs against placebo would be striking.
Again, the main thing holding up use of these agents (pending approval) is economics. I myself have seen a number of patients (even retirees) who are fully willing to pay out of pocket for these drugs. I myself would have no qualms in taking one of these agents, having reviewed the assembled literature.
Competing interests pertaining specifically to this post, comment, or both:
None.
Regarding the economic issue I would expect a reaction from an official body of physicians regarding the pricing of these agents. In the midst of all the consternation about the rising cost of Healthcare, the impending insolvency of Medicare, reduced reimbursement for imaging, drug companies break new barriers with novel cancer treatments that exceed $100,000 and these agents at $200 a month. Anticoagulants are not niche agents. They treat common conditions. Do the companies need to recoup their significant investment in a year or two?
Competing interests pertaining specifically to this post, comment, or both:
None
I’ll stick with warfarin unless there is difficulty with INR control or other compelling evidence not to use it. It has stood the test of time well. I need to see more post marketing surveillence for the newer agents. I also know how to reverse warfarin and it is QD or less.
Competing interests pertaining specifically to this post, comment, or both:
None
Dr. DeMedio’s points are well taken. Dronedarone is a good example of what can happen if we get too excited about prescribing new medications. The art of medicine is knowing when to throw your hat in the ring.
As a neurologist at a major stroke center, I see a new, afib-related infarction almost every day in our stroke unit. Today it was a complete left middle cerebral territory infarct. Her INR was 1.07. She has a 50% chance of dying in the next month and a 100% chance of life long debility. She will likely require a craniectomy, a fairly radical procedure and controversial since this is a language- dominant hemisphere, if she is to survive.
Every week I see a primary brain hemorrhage on coumadin. INRs often times have not been assessed in a month, sometimes more. Sometimes a drug has been added causing a change to a previously stable INR. Yes, this is poor care. But this is the real world. I am not managing these patient’s coumadin, but I get to deal with their devastating events.
I cannot understand how cardiologists, who see the afib, but do not often see the most devastatng consequences of afib cannot concede that coumadin is inferior to the new agents in most every way.
A drug like dabigatran that shows in sub-analysis a 25% lower risk of ischemic stroke, the primary reason from my point-of-view to take coumadin, is a revolution. The primary end point is equally important. I was very sorry to see Ximelagatran not come to market years ago despite the hepatic problems. I’d probably be taking care of a smaller number of devastated stroke patients if it had been available the last five years.
Coumadin is a drug who’s time has come and gone. It has a terribly narrow therapeutic index, and other problems as noted in the posts above. In many ways it is like phenytoin, narrow T.I., many drug interactions, cheap, requires frequent lab tests, major adverse effects. Neurologists have moved away from phenytoin for chronic outpatient use in all but the patient’s that can only afford $4 list medications.
Another point I’d like to make is that study patients are a different group of people than your average clinic patient, or worse, the patient you pick up at the hospital that you may never see again. Study patients are motivated and more compliant, and tolerant of adverse effects. In the real world, I believe the new anticoagulants will be much safer and more effective than RE-LY, ROCKET AF, and ARISTOTLE data suggest.
Also, given the fact than patent life is short, competition is plentiful, and liability is very high for an anticoagulant due to the potential for devastating adverse effects, I think the pricing strategy for Pradaxa is better than I expected. No, it is not an option for those without insurance, unless they can get on assistance. But, most folks without insurance don’t get great INR regulation either. Many end up on aspirin, which has only about one-fourth the efficacy of coumadin for stroke prevention in afib.
I think prescribing coumadin will shortly become a liability. A 60% lower risk of intracranial hemorrhage is dramatic. How many of you have recently taken care of an ICH? There are no proven therapies. And by the way, there is no good data that reversing anticoagulation with coumadin in the setting of an ICH alters outcomes.
Although the market penetration here is not high yet for Pradaxa, we have not yet seen a Pradaxa-related ICH in 10 months or so on the market, but I continue to see a coumadin ICH weekly.
As an investigator for the apixiban vs. ASA study in patients deemed by their primary doctor not to be a coumadin candidate (AVERROES; stopped early for overwhelming efficacy with similar safety) as well as the trial vs coumadin (ARISTOTLE), the difference between the two studies was dramatic. Every day I had to sign off on adverse events in the ARISTOTLE study, typically massive bleeds. I had almost nothing to sign off for AVERROES.
Competing interests pertaining specifically to this post, comment, or both:
ARISTOTLE and AVERROES investigator, passionate Pradaxa speaker, stroke prevention advocate.
devils advocate here – but the novel agents have been tested in non-inferiority trial against warfarin. They are not inferior. Superior efficacy is not definitively demonstrated by the nature of the trial design.
As a neurologist – you are certainly only seeing the bad side – completely unrepresentative of the population at large.
Experience with an agent should be valued – there are millions of patient years of experience with warfarin as opposed to a few thousand with the newer agents. Skeptics will remain that it took several years for the downsides for COX II inhibitors or rosiglitazone to be recognised.
Warfarin is cheap and effective in the majority of patients – some are not controlled well and I am fairly sure that those patients will be switched rapidly.
It’s also probably that in 5 years time, Warfarin use will have dramatically fallen, similar to that of unfractionated heparin a few years after low molecular weight heparins were introduced.
Competing interests pertaining specifically to this post, comment, or both:
none
Dr. Graff’s thoughtful comments show the perspective from a major stroke center.
What you see depends on where you’re sitting. I follow hundreds of patients with atrial fibrillation and almost never see a stroke. One stroke I can recall occurred when someone was taken off warfarin years ago by a physician because of apparent return to normal sinus rhythm and clinical stability.
One issue also is that if someone is poorly compliant to warfarin, he/she will likely be poorly compliant to the NOAC’s as well.
“I suppose I would prefer a rate of NOT having an ICH of 99.76% a year to 99.53%, but does it really matter in the individual patient?”
Stephen, there are several reasons I am skeptical about the above comment. First, by focusing only on ICH we are missing other important events like mortality, ischemic stroke and other bleeds.
Second, applied on a population level, wherein we will see a 50% increase in a.fib. by 2020 from 2000, this will translate into many fewer ICH (not to mention the other endpoints discussed above).
Third, we’ve tied the ARR to the short duration of this trial. Patients on warfarin or NOAC will likely receive it for the remainder of their lives (except if they have an ICH, but then they are less likely to, on NOAC). The ARR over a score of years will be considerably greater, as the baseline risk will incrementally grow over years. Even more importantly, the ischemic stroke ARR will be considerably greater over years and years.
Fourth, clinical trial patients are often far less sick than patients in clinic, whose baseline risk is far higher and the consequent NNT/ARR gains much more substantial. The number therefore to take out of the trial is not the NNT/ARR but rather the RRR, to be applied to the patient in front of you (e.g. with CHADS2 or CHA2DS2-VASc), in calculating an NNT/ARR.
Fifth, compliance will be better on drugs that have fewer events, both ischemic and hemorrhagic, as well as drugs with fewer drug/diet interactions and need for monitoring. This will eventuate in further gains over time.
One has to compliment big pharma in this case for coming up with a solution to the undertreatment problem in atrial fibrillation. Now if we can only get the price right! (and the drugs approved, of course 🙂
Larry Husten’s article on ARISTOTLE provided the following information:
The investigators calculated that for every 1000 patients treated with apixaban instead of warfarin for 1.8 years,
•stroke would be avoided in 6 patients,
•major bleeding would be avoided in 15 patients, and
•death would be avoided in 8 patients.
Thus, in a moderate sized practice I calculate that for every 100 patients treated with apixaban instead of warfarin for approx. 1 year
stroke would be avoided in 0.3 patients
major bleeding would be avoided in 0.75 patients, and
death would be avoided in 0.4 patients.
If the medication costs approx. US $200/month, it would cost approximately a quarter of a million dollars to prevent the above in one year. If my math is correct, 1000 patients for two years would be about 5 million dollars. (And this is just one medication within the whole breadth of health care!)
It has become clear to me from the excellent discussions above that we–particularly clinicians on the front lines–need to face the daunting task of studying and reviewing statistics in further depth so that we can make reasonable decisions as these new, often costly, medications and interventions arrive.
Not only do we need to study statistics per se, but we also need to figure out what the statistics really mean to us. For example, if a cheaper drug results in a given bad outcome 0.5% of the time, should I be thinking of the few patients in the office who will likely never experience that outcome or the thousands or millions of people treated worldwide where a tiny percentage may mean quite a few bad outcomes overall?
As society grapples with deficits and constrained budgets, these considerations will become very important. Where the dollars go will require a lot of statistical, medical, economic, and philosophical thought and discussion.