Warfarin and "Warfarinlike" Anticoagulants
Oral anticoagulants can be divided into two groups: (1) hydroxycoumarins, including warfarin (commonly called by its trade name Coumadin), difenacoum, panwarfarin, warficide, coumachlor, coumafuryl, fumasol, prolin, ethyl biscoumacetate (Tromexan), phenprocoumon, dicumarol bishydroxycoumarin, and acenocoumarin (Sintrom); and (2) indanediones, including chlorophacinone, pindone, pivalyn, diphacinone, diphenadione, phenindione, and anisindione. Regardless of the classification, their mechanism of action involves inhibition of the vitamin K cycle. Vitamin K is a cofactor in the postribosomal synthesis of clotting factors II, VII, IX, and X (Fig. 59–2). The vitamin K—sensitive enzymatic step that occurs in the liver involves the γ-carboxylation of 10 or more glutamic acid residues at the amino terminal end of the precursor proteins, to form a unique amino acid γ-carboxyglutamate.53,174,179,180 These amino acids chelate calcium in vivo, which allows the binding of the four vitamin K—dependent clotting factors to phospholipid membranes during activation of the coagulation cascade.195
Vitamin K is inactive until it is reduced from its quinone form to a quinol (or hydroquinone) form in hepatic microsomes. This reduction of vitamin K must precede the carboxylation of the precursor factors. The carboxylation activity is coupled to an epoxidase activity for vitamin K, whereby vitamin K is oxidized simultaneously to vitamin K 2,3-epoxide (see Fig. 59–2).179,195 This inactive form of the vitamin is converted back to the active form by two successive reductions.53,116,146 In the first step, an epoxide reductase (known as vitamin K 2,3-epoxide reductase) uses reduced nicotinamide adenine dinucleotide (NADH) as a cofactor to convert vitamin K 2,3-epoxide to a quinone form.139,179 Subsequently, the quinone is reduced to the active vitamin K quinol form (see Antidotes in Depth A16: Vitamin K1).
Warfarin is a racemic mixture of R warfarin and S warfarin enantiomers. In rodents, S warfarin is three to six times more potent than R warfarin at producing hypoprothrombinemia.29 In humans, S warfarin may only be about 1.5 times as potent as R warfarin.30 Warfarin and all warfarinlike compounds inhibit the activity of vitamin K 2,3-epoxide reductase, as can be demonstrated by the observation of elevated concentrations of vitamin K 2,3-epoxide in orally anticoagulated subjects.39,198 Additional evidence suggests that another enzyme, vitamin K quinone reductase, is also inhibited by warfarin and its related compounds (Fig. 59–2).53,57 This reduction in the cyclic activation of vitamin K subsequently inhibits the formation of activated clotting factors.
Orally ingested warfarin is virtually completely absorbed, and peak serum concentrations occur approximately 3 hours after administration.178 Because only the free warfarin is therapeutically active, concurrent administration of xenobiotics that alter the concentration of free warfarin, either by competing for binding to albumin or by inhibiting warfarin metabolism, may markedly influence the anticoagulant effect.12,61,178 The pharmacologic response to warfarin is a polygenic trait with approximately 30 genes contributing to its therapeutic effects.101 Table 59–1 lists the xenobiotics that interfere with or potentiate warfarin's effects. Although vitamin K regeneration is altered almost immediately, the anticoagulant effect of warfarin, as well as other oral anticoagulants, is delayed until the existing stores of vitamin K are depleted and the active coagulation factors are removed from circulation. Because vitamin K turnover is rapid, this effect is largely dependent on factor half-life (t1/2), with factor VII (t1/2 ~5 hours) depleted most rapidly.61 For a prolongation of the INR to occur, factor concentrations must fall to approximately 25% of normal values. Assuming complete inhibition of the vitamin K cycle, this suggests that in most patients who are not originally anticoagulated, at least 15 hours (three factor VII half-lives) are required before warfarin's effect is evident.59 In fact, complete inhibition does not occur, and hence the onset of coagulation is even further delayed.
Table 59–1. Common Xenobiotic Interactions With Warfarin Anticoagulation |Favorite Table|Download (.pdf)
Table 59–1. Common Xenobiotic Interactions With Warfarin Anticoagulation
|Anabolic steroids||Nonsteroidal antiinflammatory drugs||Cholestyramine|
|Fluconazole fluoroquinolone antibiotics||Vitamin E|
|HMG-CoA reductase Inhibitors|
Because the half-life of warfarin in humans is 35 hours, its duration of action may be as long as 5 days.29,178 On average, it takes approximately 6 days of warfarin administration to reach a steady-state anticoagulant effect.
R warfarin is metabolized by isozymes CYP1A2 and CYP3A4, and S warfarin is metabolized by CYP2C9 of the hepatic microsomal P450 enzyme system. R warfarin is metabolized by side-chain reduction to secondary alcohols that are subsequently excreted by the kidney, whereas S warfarin is metabolized by hydroxylation to 7-hydroxy warfarin, which is excreted into the bile.178 The elimination of S warfarin is more rapid than that of R warfarin.30
The therapeutic dose of warfarin is established for both adults and children. Typical adult recommendations are to give a starting dose of 5 mg/d with subsequent doses based on nomograms, computer programs, and/or clinical experience.64 Previous recommendations of initiating with a "loading" dose appear to be unnecessary.5 Wide variability of maintenance dosing also exists, depending on, for example, individual responsiveness, comorbid health conditions, and age. For children, the suggested starting dose of warfarin is 0.2 mg/kg, followed by continued loading over 3 days, followed by a daily maintenance dose to maintain the INR between 2 and 3.127,152 For patients with mechanical heart valves, depending on the type of valve, an increased intensity of anticoagulation (INR 3.0 to 4.0) may be recommended.54
Dosing of warfarin and other vitamin K antagonists is potentially problematic in certain individuals. In one study, genetic polymorphisms of the vitamin K epoxide reductase complex 1 (VKORC1) and CYP2C9 genes appear to be the strongest predictors of interindividual variability in the anticoagulant effect of warfarin.101 Furthermore, pharmacogenomic research with complex xenobiotics, such as warfarin, may improve the treatment of patients and predict or prevent interactions with other xenobiotics. In fact, the US Food and Drug Administration (FDA) has recently approved a commercially available test to identify variants within these genes.97
Pharmacology of Long-Acting Anticoagulants
Within the coumarin group are two 4-hydroxycoumarin derivatives—difenacoum and brodifacoum differ from warfarin by their longer, higher-molecular-weight polycyclic hydrocarbon side chains (Fig. 59–3). Together with chlorophacinone, an indandione derivative, they are known as "superwarfarins," or long-acting anticoagulants.
Structural comparison of prototypical short-acting (warfarin) and long-acting (brodifacoum) anticoagulants.
Long-acting anticoagulants were designed to be effective rodenticides in warfarin-resistant rodents.113 Their mechanism of action is identical to that of the traditional warfarinlike anticoagulants, as demonstrated by the measurement of increased concentrations of vitamin K 2,3-epoxide after long-acting anticoagulant administration.28,31,32,107,145 The ability of these xenobiotics to perform as superior rodenticides is attributed to their high lipid solubility and concentration in the liver.107,113,145 They also may saturate hepatic enzymes at very low concentrations, as demonstrated by zero-order elimination following overdose.32 These factors make them about 100 times more potent than warfarin on a molar basis.107,113,145 In addition, they have a longer duration of action than the traditional warfarins.107,113,145 For example, to obtain 100% lethality in a mouse, more than 21 days of feeding with a warfarin-containing rodenticide (0.025% anticoagulant by weight of bait) is required.113 Similar efficacy can be achieved with a single ingestion of brodifacoum (0.005% anticoagulant by weight of bait).113
Many animals have been poisoned with long-acting anticoagulants, either secondary to the unintentional ingestion of rodenticides or intentionally for scientific investigation. In rats, the half-life of brodifacoum is reported to be 156 hours.9 The half-life in dogs is reported to be between 6 and 120 days.200 Horses intentionally poisoned with brodifacoum had a half-life of 1.22 days.23 The veterinary literature is replete with reports of fatalities and of animals that remained anticoagulated in excess of 1 month.130,176
Likewise, many cases of intentional overdose of long-acting anticoagulants in humans are also described in the literature. These patients' clinical courses are characterized by a severe coagulopathy that may last weeks to months, often accompanied by consequential blood loss. The most common sites of bleeding are the gastrointestinal and genitourinary tracts. Although initial parenteral vitamin K1 doses as high as 400 mg have been required for reversal,33 daily oral vitamin K1 requirements may be in the range of 50 to 100 mg. Recent experience in both animals and humans suggests that parenteral vitamin K1 therapy might not be required (see Antidotes in Depth A16: Vitamin K1).32,200 It should also be noted that although ingestions of these xenobiotics are the most common route of exposure and subsequent cause of toxicity, dermal absorption can occur also resulting in coagulopathy.172
Patients with unintentional ingestions must be distinguished from those with intentional ingestions, because the former individuals demonstrate a low likelihood of producing coagulation abnormalities and have only rare morbidity or mortality. Prolongation of the INR is unlikely with a single small ingestion of a superwarfarin rodenticide. Clinically significant anticoagulation is even rarer. In a combined pediatric case series, prolongation of the INR occurred in only eight of 142 children (5.6%) reported with single small ingestions of long-acting anticoagulants.15,94,95,171 Only one child in this group was reported to have "abnormal prolonged bleeding," but this required no medical attention.171 In a single case report, a 36-month-old child developed a coagulopathy manifested by epistaxis and hematuria, with anticoagulation persisting for more than 100 days after a presumed, but unwitnessed, single unintentional ingestion of brodifacoum.182 Clinically significant coagulopathy can result, however, following small repeated ingestions. Two children reportedly became poisoned by repeated ingestions of a long-acting anticoagulant. One child presented with a neck hematoma that compromised his airway, and the other with a hemarthrosis.69 Similarly, a 7-year-old girl required multiple hospitalizations over a 20-month period following repeated nonsuicidal ingestions of brodifacoum.192 Finally, a 24-month-old child who presented with unexplained bruising and a PT greater than 125 seconds was the victim of brodifacoum poisoning because of a Munchausen syndrome by proxy.8
Most patients (usually children) are entirely asymptomatic and have a normal coagulation profile following an acute unintentional exposure. Knowing that the risk of coagulopathy is low and that it takes days to develop, most authors recommend supportive care only.93,171 Despite the fact that significant toxicity from superwarfarins is rare, it should be recognized that the reported benign courses of pediatric exposures may be misleading. Multiple retrospective studies suggest that children with unintentional acute exposures do not require any follow-up coagulation studies.128,132,147,166 However, this conclusion and approach to management may be an unjustified attempt to decrease the cost of "unnecessary" coagulation studies. There are clearly insufficient data to justify this conclusion, as many of these "exposed" children were never documented to have ingested long-acting anticoagulants (see Chap. 135). We recommend that clinicians continue to manage these children as possible significant exposures, and that all children be followed up with at least a single INR at least 48 hours after the exposure. A baseline INR is usually unnecessary but may be performed if there is a suspicion of chronic ingestion.
Typical warfarin rodenticides contain only small concentrations of anticoagulant, 0.025% (or 25 mg of warfarin per 100 g of product). Using the data previously listed, a 10-kg child would require an initial dose of 2.5 mg of warfarin (or 10 g of rodenticide). These quantities are far greater than those that occur in typical "tastes." Thus, single small unintentional ingestions of warfarin-containing rodenticides pose a minimal threat to normal patients.93 In contrast, intentional and large unintentional ingestions of pharmaceutical-grade anticoagulants have the potential to produce a coagulopathy and bleeding. In one study describing 12 patients with surreptitious ingestion of oral anticoagulants, nine were healthcare professionals.140 These patients presented with bruising, hematuria, hematochezia, and menorrhagia, the typical manifestations of impaired coagulation. Hemorrhage into the neck with resultant airway compromise is a rare but life-threatening complication that has occurred.24
Although intentional ingestions of warfarin-containing products are uncommon, adverse drug events resulting in excessive anticoagulation and bleeding frequently occur. The risk of hemorrhage during oral anticoagulant therapy depends on a myriad of factors, including the intensity of anticoagulation, patient characteristics, and comorbid conditions such as hypertension, renal insufficiency, hepatic dysfunction, malignancy, length of anticoagulant therapy, and indications for anticoagulation—cerebrovascular disease, prosthetic heart valves, atrial fibrillation, ischemic heart disease, and venous and arterial thromboembolism. Although the significance of each of these clinical conditions varies among different reports, most studies demonstrate that there is a greater incidence of bleeding complications with increasing INR,41 increasing intensity (or variation) of coagulation, advanced age, a history of previous bleeding episodes while on therapeutic warfarin, drug interactions, impaired liver function, and dietary changes.59,61,71,76,150,197 Clearly, the most serious complication of excessive anticoagulation is intracranial hemorrhage, which is reported to occur in as many as 2% of patients on long-term therapy.61 This complication is associated with a fatality rate as high as 77%.121 A recent study of patients with intracranial hemorrhage found that decreased level of consciousness and increased size of hemorrhage were predictors of poor prognosis.203 Somewhat surprisingly, the degree of INR elevation was not associated with worse outcome.203
An Outpatient Bleeding Risk Index was created and shown to be more accurate than physician judgment in classifying patients according to the risk of major bleeding.18 The index was based on independent risk factors: age 65 years or older; history of cerebrovascular accident; history of gastrointestinal bleeding; and history of recent myocardial infarction, hematocrit less than 30%, serum creatinine greater than 1.5 mg/dL, or diabetes mellitus. The sum of the number of risk factors successfully predicted major bleeding at 48 months to be 3% in low-risk (zero risk factors), 12% in intermediate risk (one to two risk factors), and 53% in high-risk (three to four risk factors) patients. Because physicians are often unable to accurately estimate the probability of bleeding, use of the Outpatient Bleeding Risk Index seems appropriate to improve awareness and treatment of these high-risk patients and was validated in at least one subsequent study.193
In a study of 32 patients who developed life-threatening hemorrhage while on warfarin therapy, most patients had multiple risk factors including excessive anticoagulation.197 The gastrointestinal tract was identified as the source of bleeding in 67% of the patients.197 Sixty-six percent of patients were given vitamin K1, 50% were given fresh-frozen plasma (FFP), and 7% were given both therapies.197
Established screening tests are helpful for diagnosis. Four studies—PT (INR), PTT, thrombin time, and fibrinogen concentration—are usually adequate. Prothrombin time is calculated by adding standardized thromboplastin reagent (phospholipid and tissue factor) to a sample of the patient's citrated plasma (the citrate removes calcium to prevent clotting). Calcium is then introduced and the time to clotting measured. With the exception of factor X, the PT is unaffected by the presence or absence of factors VIII to XIII, platelets, prekallikrein, and HMWK. An individual's PT was formerly expressed as a ratio (PT observed to PT control). Because this ratio is directly affected by both laboratory methodology and the source of the thromboplastin reagent used, the generated results suffered from significant variability. Thus, a standard, the INR, was developed in an attempt to limit interlaboratory variability.78,136 The INR is derived by raising the PT ratio to a power value known as the International Sensitivity Index (ISI): (PT ratio)ISI. The ISI is a measure of responsiveness of the particular thromboplastin to warfarin. Although the use of the INR does not completely eliminate variability,80,135 it does improve the potential for standardized interpretation and limits interinstitutional variations. It should be noted that in patients taking oral anticoagulants, specifically warfarin, the INR is extremely effective at monitoring the extent of anticoagulation. However, use of the INR measurement in the setting of fulminant hepatic failure, as in the recently developed Model for End-Stage Liver Disease (MELD) score, a tool to predict the need for liver transplantation, is unwarranted.35 In these patients the INR is extremely variable and inaccurate as a consequence of the variability in thromboplastins.154 This problem may be mitigated in the United States because of the use of recombinant human preparations of thromboplastin, which results in greater consistency.36
The partial thromboplastin time is measured by adding kaolin or celite to citrated plasma in order to activate the "contact" components of the intrinsic system. This mixture is then recalcified and the time to clotting observed. Some tests use phospholipids in the reagent to activate the remaining coagulation factors, thereby giving rise to the term activated partial thromboplastin time (aPTT). Because the PTT and aPTT are essentially interchangeable, the term PTT is used hereafter to represent the concept. The PTT is not affected by alterations in factors VII, XIII, or platelets.
The thrombin time, determined by adding exogenous thrombin to citrated plasma, evaluates the ability to convert fibrinogen to fibrin, and is thus unaffected by abnormalities of factors II, V, VII to XIII, platelets, prekallikrein, or HMWK. Finally, either a fibrinogen concentration or a determination of fibrin degradation products will help distinguish between problems with clot formation and consumptive coagulopathy (disseminated intravascular coagulation). An evaluation of the combination of normal and abnormal results of these tests usually determines a patient's clotting abnormality (Table 59–2).
Table 59–2. Evaluation of Abnormal Coagulation Times |Favorite Table|Download (.pdf)
Table 59–2. Evaluation of Abnormal Coagulation Times
|PT normal, PTT prolonged, bleeding|
|Deficiencies of factors VIII, IX, XI|
|von Willebrand disease|
|PT normal, PTT prolonged, no bleeding|
|Deficiencies of factor XII, prekallikrein, high-molecular-weight kininogen inhibitor syndrome|
|PT prolonged, PTT normal|
|Deficiency of factor VII|
|Warfarin therapy (early)|
|Vitamin K deficiency (mild)|
|Liver disease (mild)|
|PT and PTT prolonged, thrombin time normal, fibrinogen normal|
|Deficiencies of factors II, V, IX; vitamin K deficiency (severe)|
|Warfarin therapy (late)|
|PT and PTT prolonged, thrombin time abnormal, fibrinogen normal|
|PT and PTT prolonged, thrombin time abnormal, fibrinogen abnormal|
|Disseminated intravascular coagulation|
Inhibitors can be diagnosed by "mixing studies," because only a small percentage of the coagulation factors present in normal plasma are necessary to have normal clotting studies. If the patient with an abnormal PT or PTT suffers from even a severe factor deficiency, restoration of that factor activity to 50% of normal will completely normalize the PT or PTT. Thus the presence of an abnormal PT or PTT that will not correct by incubation of the patient's plasma with an equal volume of normal plasma is diagnostic of an inhibitor of coagulation. Heparin-induced anticoagulation results in an elevated PTT that corrects when mixing studies are performed. More sophisticated studies can be used to identify specific coagulation-factor deficiencies. The reader is referred to one of several standard references for a more detailed discussion of the approach to patients with abnormal coagulation studies.156
Although warfarin concentrations may be useful to confirm the diagnosis in unknown cases and to study drug kinetics,72,138 the routine use of simple and inexpensive measures such as INR determination seems more appropriate.
Laboratory Evaluation of Long-Acting Anticoagulants
For patients who have ingested long-acting anticoagulants and who are considered likely to develop a coagulopathy, baseline coagulation studies are not usually helpful, but they may provide information about chronic exposures. If the history is reliable and the patient is healthy, baseline studies can be avoided. A single INR at 48 hours should identify all patients at risk of coagulopathy.171 Depending on the social situation, these studies can be obtained while the patient remains in the home setting.
In contrast, all patients with intentional ingestion of long-acting anticoagulants should be presumed to be at risk for a severe coagulopathy. In fact, most patients do not seek medical care until bruising or bleeding is evident.11,32,33,38,56,82,95,100,133,181 These events often occur many days after ingestion, which obviates the need for gastric decontamination unless there is a suggestion of repetitive ingestion. These patients should be managed as described below.
For patients who have suspected long-acting anticoagulant overdose, daily or twice-daily INR evaluations for 2 days should be adequate to identify most patients at risk for coagulopathy. Early detection through coagulation factor analysis may be preferred,72,82 however, and concentrations of long-acting anticoagulants can now be measured.102,138