Is INR between 2.0 and 3.0 the optimal level for Chinese patients on warfarin therapy for moderate-intensity anticoagulation?

Abstract

Aim

To examine the optimal range of International Normalized Ratio (INR) for Chinese patients receiving warfarin for moderate-intensity anticoagulation.

Methods

This was a retrospective cohort study conducted at the ambulatory setting of a 1400-bed public teaching hospital in Hong Kong. The INR measurements and occurrence of serious or life-threatening haemorrhagic and thromboembolic events among patients newly started on warfarin from 1 January 1999 to 30 June 2001 for indications with target INR 2–3 were analysed. The INR-specific incidence of bleeding and thromboembolism were calculated.

Results

A total of 491 patients were included, contributing to 453 patient-years of observation period. Forty-seven of the 491 patients experienced 25 haemorrhagic events (5.5 per 100 patient-years) and 27 thromboembolic events (6.0 per 100 patient-years). The percentage of patient-time spent within therapeutic INR range (2–3), INR <2 and INR >3 were 50, 44 and 6%, respectively. The incidence of either haemorrhagic or thromboembolic events was lowest (≤4 events per 100 patient-years) at INR values between 1.8 and 2.4.

Conclusions

An INR of 1.8–2.4 appeared to be associated with the lowest incidence rate of major bleeding or thromboembolic events in a cohort of Hong Kong Chinese patients receiving warfarin therapy for moderate-intensity anticoagulation.

Introduction

Warfarin is the most commonly prescribed anticoagulant for prevention and treatment of thromboembolic events [1]. The anticoagulation effect of warfarin is subject to wide inter- and intraindividual variability that possibly leads to haemorrhagic or thromboembolic events despite careful dosage titration. The risks of thromboembolism and haemorrhage depend on the intensity of anticoagulation as measured by the International Normalized Ratio (INR) [2, 3]. Both the British Society for Haematology (BSH) and the American College of Chest Physicians (ACCP) recommend a target INR range of 2–3 (moderate-intensity anticoagulation) for most indications of warfarin, such as prophylaxis and treatment of venous thrombosis, treatment of pulmonary embolism and prevention of systemic embolism secondary to tissue heart valves, valvular heart disease or atrial fibrillation [1, 4]. The exceptions that require high-intensity anticoagulation (INR 2.5–3.5) are patients with mechanical heart-valve prostheses. Despite that both recommendations are mainly derived from clinical trials conducted in western populations, the same recommendations have been adopted for management of anticoagulation therapy in populations of different ethnic backgrounds.

Various studies reporting the association of racial background to warfarin dosage requirement report that Iranian and Asian subjects are more sensitive to warfarin than North American and European subjects [5–7]. A prospective, randomized trial conducted in Japan on patients with nonvalvular atrial fibrillation showed that low-intensity anticoagulation therapy (INR 1.5–2.1) was associated with fewer major haemorrhagic complications than the conventional-intensity treatment (INR 2.2–3.5) for secondary prevention of stroke [8]. It has been reported that Chinese patients require a lower dose of warfarin than Caucasians although the intensity of the anticoagulation is comparable [7, 9–11]. As Chinese patients have a lower warfarin requirement, there are concerns whether the same target INR range for western populations is also the optimal anticoagulation level for Chinese patients. The objective of the present study was therefore to examine the optimal INR range, in which the incidence of both major thromboembolic and bleeding events were lowest, for Chinese patients receiving warfarin for moderate-intensity anticoagulation.

Method

Study design

This was a retrospective cohort study conducted at the ambulatory setting of a 1400-bed public teaching hospital in Hong Kong. The study was approved by the Clinical Research Ethics Committee of The Chinese University of Hong Kong. The INR measurements and occurrence of warfarin-related major events among the patients receiving warfarin for indications with target INR 2–3 were analysed. Outpatients who were newly receiving warfarin from 1 January 1999 to 30 June 2001 were included in the study. Data were retrieved from medical records until the patients stopped receiving warfarin therapy or 30 June 2001, whichever occurred first. Patients receiving warfarin for indications with target INR 2.5–3.5 (prosthetic heart valves) and patients receiving warfarin for less than 3 months were excluded from the present analysis. Patient demographics, dates and results of all INR assessments, occurrence of warfarin-related major complications including thromboembolic events and bleeding were collected.

The definitions of major events were adopted from Fihn et al.[2]. Major bleeding included gastrointestinal bleeding, gross haematuria, haemoptysis, bleeding leading to cardiopulmonary arrest, to surgical or angiographic intervention, to irreversible sequelae such as myocardial infarction, neurological deficit or massive haemothorax, to systolic hypotension (<90 mmHg), to critical anaemia (haematocrit ≤0.20) or to death. Major thromboembolic events were transient ischaemic attacks, stroke, recurrent deep venous thrombosis, pulmonary embolism and systemic embolism. A major event was included in the present analysis if the INR was obtained at the time of the hospital admission or a measurement of INR was performed in the clinic in less than 7 days before the event.

The percentage of patient-time spent in each INR level was estimated using linear interpolation between measured INR values developed by Rosendaal et al. as described elsewhere [12]. Briefly, the INR value between two measurements over a known period of interval was assumed to vary linearly. Patient-time between two INR measurements was divided into days and allocated equally to all INR values with increment of 0.1 INR over the range of time interval.

The incidence rates of bleeding and thromboembolic events for each specific level of INR was calculated by the following equation [12]:

$$\left(\text{Number of major events occurred at a INR level}\right)/\left(\text{Number of total patient}-\text{years spent at the same INR level}\right)$$

Results

A total of 491 patients were included in the present study, contributing to 453 patient-years of observation period. The mean age of the patient was 65.8 ± 14.2 years and 237 (48%) patients were male. The main indications for warfarin were atrial fibrillation (72%), deep vein thrombosis (12%), pulmonary embolism (3%), cerebrovascular accident (3%) and rheumatic heart disease (3%).

Forty-seven of the 491 patients experienced 25 haemorrhagic events (5.5 per 100 patient-years) and 27 thromboembolic events (6.0 per 100 patient-years). Of these 52 major events, INR was documented at the time of the event in 44 cases and they were used for the analysis of INR-specific incidence rates. Twenty-two of the 44 events were bleedings (5 intracranial; 17 extracranial) and others were thromboembolic events (12 cerebral infarction; 10 peripheral embolism). Five fatal events were all caused by bleeding. There was no significance difference in gender (P = 0.075) and in age (P = 0.342) between patients with major events (67.9 ± 14.6 years) and patients without major events (65.6 ± 14.2 years). The distribution of the patient-years over the INR intervals was shown in Figure 1. The percentage of patient-time spent within therapeutic INR range (2–3), INR <2 and INR >3 were 50, 44 and 6%, respectively.

Figure 1.

Distribution of patient-years among INR intervals

The incidence rates of bleeding and thromboembolic events at specific INR level fell in a narrow U-shaped distribution (Figure 2). The incidence rate of bleeding events increased sharply from four events to 11.7 events per 100 patient-years as INR rose from 2.4 to 2.9, and it further increased to 40 events per 100 patient-years when INR reached 3.4. When the INR fell from 1.8 to 1.5, the thromboembolic event rate increased from 3.3 to 8.7 events per 100 patient years and it reached 71 events per 100 patient-years as INR decreased to 1. In Figure 2, it was estimated that the incidence rate of either bleeding or thromboembolic events was lowest (≤four events per 100 patient-years) at INR values between 1.8 and 2.4. The incidence rate of all events (bleeding and thromboembolism) was about eight events per 100 patients-years at these INR levels. The incidence rates of events in the INR of 1.8–2.4 were significantly lower than the rates in the INR above and below this range (P = 0.037).

Figure 2.

Incidence of major events at specific INR levels. Bleeding event (▴); thromboembolic event (▪).

Discussion

The optimal therapeutic range of anticoagulation therapy varies for different indications and for patients with various characteristics. Bleeding is the major complication of warfarin therapy and it is related to the intensity of anticoagulation [2, 13–15]. In the process of searching for an optimal INR range, studies have therefore focused on establishing the lowest effective therapeutic ranges [16–20]. Randomized trials comparing different target INR ranges are recommended as the most reliable method to establish an optimal anticoagulation intensity, when comparing with other study designs such as indirect comparisons of results from randomized trials, subgroup analyses of anticoagulation group from randomized trials and case-control studies [1]. Based upon clinical evidence predominantly generated in randomized clinical trials, The BSH and the ACCP recommended that a moderate-intensity INR of 2.0–3.0 is effective for most indications [1, 4].

A number of studies have shown lower warfarin requirements in Chinese patients (3 mg day⁻¹) compared with Caucasians (4–6 mg day⁻¹), suggesting that Chinese patients are more sensitive than Caucasians to the anticoagulation effect of warfarin [7, 10, 11, 21–23]. Age and the target intensity of anticoagulation therapy were identified as the two most important factors affecting warfarin dose requirement [11, 23]. Yet little is known about the optimal intensity of anticoagulation, i.e. the INR range in which the incidence of both the major thromboembolic and bleeding events are lowest, in the Chinese population.

In the present study, the INR-specific incidence of major events in a cohort of Chinese patients receiving moderate-intensity anticoagulation therapy was examined. The average age of patients in the present cohort was >65 years and there was no significant difference in age and in gender between the patients with and without major events. Our results showed that the event rate was lowest at an INR of 1.8–2.4 for the present cohort. In this range, the incidence rate of either major bleeding or thromboembolism was ≤ four events per 100 patient-years. The higher mortality rate related to major bleeding events (5 of 22 events were fatal), compared with thromboembolic events (0 fatal events), in this cohort of Chinese patients suggested that the target level of anticoagulation should aim for lower INR.

Cannegieter et al. examined the INR-specific incidence rates of thromboembolism and major bleeding in 1608 Dutch patients with mechanical heart valves and determined the optimal intensity of anticoagulation to be an INR of 2.5–4.9 [24]. The incidence rates of haemorrhagic and thromboembolic events formed a wide U-shaped distribution that the incidence rate at an INR of 2.5–4.9 was only about two events per 100 patient-years. The incidence rate of all events rose sharply when the INR fell below 2.5 or rose to 5.0 or above.

A subgroup analysis of the anticoagulation cohort of the European Atrial Fibrillation Trial, a secondary prevention trial in patients with nonrheumatic atrial fibrillation, was conducted to determine the optimal intensity of oral anticoagulation [25]. The INR-specific incidence rates for occurrence of ischaemic or haemorrhagic complications in 214 patients indicated that the rate was lowest at an INR of 2.0–3.9. The incidence rate of ischaemic events increased to 18 events per 100 patient-years when the INR fell below 2.0. The bleeding incidence rate increased to over 20 events per 100 patient-years when the INR rose to 4.0 or above.

The tolerance of the present Chinese cohort to the anticoagulation effect appeared to be lower than the European patients as the incidence rate of major bleeding increased steeply when the INR rose above 2.4. The incidence rate of all events at an INR of 1.8–2.4 (about eight events per 100 patient-years) in the present cohort were almost 4-fold the rate of all events in the Dutch patients at an INR of 2.5–4.9.

One of the possible explanations for the high bleeding incidence at a moderate level of anticoagulation (INR >2.4) is that the mean age of the present cohort was >65 years and older patients are at higher risk for warfarin-related bleeding [3, 26–32]. Use of herbal medicine is another possible cause for the high sensitivity to anticoagulation effect. The use of herbal medicine was common (26%) among Hong Kong Chinese patients receiving warfarin therapy [33]. A number herbal medications were identified to have antiplatelet and/or antithrombotic effects, and therefore potentially interact with warfarin and increase risk of bleeding [34–43].

Another potential explanation is the polymorphism of cytochrome P450 (CYP) 2C9. Warfarin is a racemic mixture of two enantiomers, S- and R-warfarin, and S-warfarin is 3–5 times more potent than the R-warfarin [44]. The metabolism of S-warfarin is primarily catalysed by CYP2C9 [44, 45]. The functional significance of various mutations on the coding region of the CYP2C9 gene were examined and it has shown that the allelic variants of CYP2C9 affect metabolism clearance of warfarin [46, 47]. CYP2C9 polymorphism was associated with lower warfarin dosage requirement and with increased risk of major bleeding events during the induction phase of warfarin therapy [48–52]. Nevertheless, the frequencies of both known functionally defective CYP2C9*2 and *3 were low in the Chinese population, yet Chinese patients required a 40–50% lower maintenance dose of warfarin than Caucasian patients [23, 53]. This difference is not explainable entirely by variations in age, body weight, sex, dietary vitamin K intake, clinical indications for warfarin use and target level of anticoagulation. It implies that other functionally defective polymorphisms in the coding and noncoding regions of CYP2C9, such as the 5′-flanking region, may also have a significant effect on the phenotype of CYP2C9. Further study is therefore required to determine the association of high warfarin sensitivity in Chinese patients with CYP2C9 polymorphisms.

A limitation of the present study was the retrospective, observational study design that limited the level of evidence presented. Our findings supported further investigation of a low-intensity of anticoagulation compared with the conventional-intensity (INR 2–3) for Chinese patients in randomized clinical trials. The present study was also limited by the nature of retrospective cohort studies that data on some factors affecting the risk of bleeding, such as blood pressure control, vitamin K dietary intake, use of herbal and over-the-counter medications, and concurrent use of antiplatelets or aspirin, could not be gathered. Also, INR values were not documented for eight (15%) of the 52 major events at the time of admission or within 7 days prior to the event, consequently affecting the distribution of incidence of major events over the INR range.

Direct thrombin inhibitors and selective factor Xa inhibitors are recently developed with potential clinical and safety advantages over warfarin for the prevention and management of thromboembolic events [54–56]. Until full-scale pharmacoeconomic evaluations of these new classes of antithrombotic agents are available, warfarin remains the mainstay of drug therapy for prevention and treatment of thrombotic disorders.

In conclusion, a cohort of Hong Kong Chinese patients receiving warfarin therapy for moderate-intensity anticoagulation showed high incidence of major bleeding when INR was >2.4 and an INR of 1.8–2.4 appeared to be associated with the lowest incidence rate (≤ four events per 100 patient-years) of major bleeding or thromboembolic events. The effectiveness and safety of a low-intensity of anticoagulation for Chinese patients should be further examined in randomized clinical trials.