Thursday, August 16, 2007

Warfarin's Labeling Fiasco

As practicing cardiac electrophysiologist, I have great respect for the blood thinning medication, warfarin (Coumadin®). You see, a good portion of my patients with atrial fibrillation take this drug to reduce their risk of stroke.

Warfarin has a complicated pharmacology: the dose you take today isn’t likely to affect you blood thinning level for about three days; and it has a complicated, yet very specific effect, blocking the production of Vitamin K in one’s liver. It blocks Vitamin K by “competitive inhibition:”
Warfarin inhibits epoxide reductase (specifically the VKORC1 subunit), thereby diminishing available vitamin K and vitamin K hydroquinone in the tissues, which inhibits the carboxylation activity of the glutamyl carboxylase. When this occurs, the coagulation factors are no longer carboxylated at certain glutamic acid residues, and are incapable of binding to the endothelial surface of blood vessels, and are thus biologically inactive. As the body stores of previously-produced active factors degrade (over several days) and are replaced by inactive factors, the anticoagulation effect becomes apparent. The coagulation factors are produced, but have decreased functionality due to undercarboxylation; they are collectively referred to as PIVKAs (proteins induced [by] vitamin K absence/antagonism). Hence, the effect of warfarin is to diminish blood clotting in the patient.
But Vitamin K is not just made in the liver. Vitamin K is also made by bacteria that reside in the gut.

This is why blood thinning levels sky-rocket when people take antibiotics – not because they necessarily affect the production of Vitamin K in the liver, but rather that the antibiotic kills the gut bacteria that provide an important amount of Vitamin K to the blood stream. Genetic testing tests humans, not the bacteria that reside in human’s gut. And the number of other confounding drug interactions reads like a Who’s Who of pharmacology.

So it is interesting that today’s Wall Street Journal discusses that the FDA is going to add a labeling change to warfarin that says that the initial lower warfarin dose “should be considered for patients with certain genetic variations.”

Never mind that the testing is not covered by all insurers.

Never mind that this testing adds significant costs.

Never mind that the testing takes about 10 days to return, is performed at only one lab in North Carolina, and returns only AFTER the initial dosing is performed in the first place.

Never mind that there have been no prospective, multi-center, randomized real-world clinical trials demonstrating that the application of these tests effects outcomes or reduces the incidence of bleeding in patients – especially patients carefully monitored in specialized “coumadin clinics.”

Never mind that it exposes doctors to significant claims of negligence if the tests are not used.

Never mind the conflict of interest that exists within the FDA’s fee-for-review structure that makes this labeling change sound like a boon for the genetic testing companies and a kick-back to the FDA.

And what if a patient refuses warfarin because they were found to be positive for a genetic defect, citing their risk of bleeding might be too high without really knowing the risks?

Will the FDA want to talk to my patient after their stroke?


18:24 - Addendum: Links fixed.
19:30 - The FDA's News Release and the Package Insert (pdf) - Genetic recommendations are on page 25.


rlbates said...

Very nice post. Good questions, but I have no answers.

Unknown said...

I'll bet the Lovenox people are pretty pleased about all of this :)

Eric, AKA The Pragmatic Caregiver said...

There's a point-of-care test in late-stage testing that returns VKORC1 and 2C9 in about an hour, which is reasonable in this setting - give them a midrange dose based on age, height and gender (AKA, the Sconce algorithm), and in an hour, refine the dose based on the genetics information. Even if the POC assay is run in a lab setting, you're looking at a day or two out of range, not weeks, and that's going to impact morbidity and mortality.

There are other providers of 2C9 and VKORC1 genetic testing - they just don't have a specific FDA approval for the indication of setting warfarin levels.

As an aside, $500 for the genetics testing isn't bad - each test gets billed to CMMS as a nurse visit and the assay, and we were schlepping the Old Guy in weekly until we got him stabilized. We would have paid $500 out of pocket to not blow two hours for a five minute test.


DrWes said...


"...and we were schlepping the Old Guy in weekly until we got him stabilized."

Thanks for your insights. But I guess I don't see how this genetic testing will change practice, except to raise costs.

For instance, I don't start 45kg 85 year-old ladies on Amiodarone or antibiotics at 5 mg of coumadin a day (I make a mental adjustment and start them on much lower dosing like 1 mg daily). Do I really need a genetic test to tell me this? If you're being vigilant to test INR's religiously, as you mentioned (and you will still need to, even with genetic testing) then why spend the $500?

The Sconce algorithm is based retrospective data on 1000 patients, and there is not general agreement that this algorithm has been validated in a prospective way. Remember, we're talking 30 million people on warfarin, here. Further, it's website states that for new patients:

"... the initial estimate of therapeutic dose explains 53% of the variability in a warfarin dose."

In other words, I might as well just keep checking the INR's frequently until the patient is stable since the algorithm is less reliable at first. I'd save a considerable amount of money this way and avoid complications.

"$500 for the genetics testing isn't bad..."

$500 IS bad if it's not picked up by your insurer and you're left paying the bill.