Pharmacogenetic Algorithm Accurately Predicts Warfarin Dosing

Reading time: 4 – 7 minutes

This article was written by Noelle K. LoConte, M.D.

Warfarin (brand name Coumadin) is one of the most commonly used anticoagulants (meaning a medication that thins the blood). It is used in a variety of medical situations, including atrial fibrillation, blood clots and when there is an increased risk of blood clotting due to genetic predisposition. When a patient is on warfarin, they need frequent blood draws to measure blood thinness and frequent dose adjustments until they have reached a stable level of blood thinning.


However, clinicians realize that each patient’s warfarin dose can be highly variable for any number of reasons, including genes, diet and concurrent medical illnesses. Sometimes it can take many weeks or months to get a patient to the right warfarin dose, and during that time the patient’s blood can either be too thin, putting them at risk of bleeding, or too thick, placing them at risk for developing blood clots. A more refined way of dosing warfarin is needed that takes into account individual patient factors.

An article published recently in the New England Journal of Medicine by the International Warfarin Pharmacogenetics Consortium seeks to address these concerns [1]. The authors include 21 research groups from around the world, representing a broad group of patients and investigators. Initially, they looked at 4,043 patients who were at their target warfarin dose, and looked back at a variety of clinical and genetic characteristics, including:

Genotype variants called single nucleotide polymorphisms (SNPs) of CYP2C9 and VKORC1 are strongly associated with warfarin responsiveness [2]. The researchers were not able to include information on diet or smoking status. From this information the scientists created a mathematical model which predicted the optimal warfarin dose. They called this the pharmacogenetic algorithm. They then applied the algorithm to 1,009 patients to see if the model predicted the final dose of warfarin used. They compared the result to a model that used only clinical (non-genetic) data and to a second model that had every patient taking an initial dose of 5 mg of warfarin daily (which, in clinical practice, is a common place to start warfarin dosing).

The pharmacogenetic algorithm predicted a lower dose of warfarin (8.3 mg/day) then the clinical algorithm (10 mg/day) or the flat dosing approach (13.3 mg/day), and was closer to the ultimate maintenance dose of warfarin achieved. The largest differences in doses were noted among patients who ended up requiring either very low or very high doses (in essence, those patients at greatest risk of bleeding or clotting when a physician is starting them on warfarin). This represented 46% of the patients in the study — a substantial number.

What is unknown is whether using this algorithm translates to better outcomes for patients, such as less bleeding or clotting, or even fewer blood draws. Although one assumes that will be the result, a study should be performed to confirm this. Also, this investigation was done retrospectively and should be confirmed with a prospective study (meaning in real time), which is a study design that is less subject to potential bias. In that setting, the impact of diet and smoking on warfarin therapy should also be assessed.

This study is one of a growing number demonstrating how pharmacogenetics can be used to individualize and optimize drug therapy. This is a promising application of “personalized medicine”, whereby drugs and drug doses will be safer and more effective for patients because they will be selected based on an individual’s genetic makeup.

Additional resources can be found in the Personalized Medicine category of the Highlight HEALTH Web Directory.

About the author: Noelle K. LoConte, M.D. is an assistant professor of medicine at the University of Wisconsin. Dr. LoConte’s clinical interests lie in gastrointestinal cancer and geriatrics. She has authored numerous studies on gastrointestinal cancer and has been interviewed by several media outlets about her expertise in this area. She blogs about current research, trials, treatments and issues in the hematology/oncolgy field at HemOnc Today.


  1. International Warfarin Pharmacogenetics Consortium. Estimation of the warfarin dose with clinical and pharmacogenetic data. N Engl J Med. 2009 Feb 19;360(8):753-64. DOI: 10.1056/NEJMoa0809329
    View abstract
  2. Cooper et al. A genome-wide scan for common genetic variants with a large influence on warfarin maintenance dose. Blood. 2008 Jun 5. [Epub ahead of print]
    View abstract
About the Author

From time to time, we publish contributed articles. If you have something that you think might be of interest to our readers, feel free to email us your article for consideration.


  1. Nice article and will be nice when we can finally start testing and getting reimbursed for the test! I’ve done a few posts on the subject and there are those debating on whether or not if is cost effective from the insurance side of things. Risk management can sure kill almost anything, including us these days!

    Also, (grin) nice to see someone besides me use the word “algorithm” since I take such a beating over it as I am attempting to get everyone who reads familiar with that word, the code controls so much in life everywhere with decision making processes.

    Back on subject, hopefully this will go forward with the warfarin testing soon!

    • Hi Barbara! Very interesting – I’ll have to read up on the research evaluating the cost-effectiveness of genotype testing. Thanks for your comment!