Hematologic diseases

John W Eikelboom,* Graeme J Hankey,† Wai Khoon Ho,‡ Cindy H Lee§ * Haematologist, Thrombosis Service, McMaster University, HHS General Divison, 237 Barton Street East, Hamilton, ON L8L2X2, Canada; † Neurologist, ‡ Fellow in Haematology, § Senior Registrar in Haematology, Royal Perth Hospital, Perth, WA. eikelbj@mcmaster.ca In reply: Pulmonary embolism (PE) remains a complex diagnosis despite the availability of validated prediction models and D-dimer testing to direct the need for diagnostic imaging. We agree with Bailey that the ability to exclude the diagnosis of PE on clinical grounds in patients with a low pretest probability is highly desirable. Unfortunately, clinical features lack sensitivity and specificity for the diagnosis of PE, and clinical prediction models, laboratory investigations, and diagnostic imaging are likely to remain an integral part of the clinical work-up. As suggested by Bragg, it may be possible to simplify the diagnostic approach by using a highly sensitive D-dimer assay, and simplified pretest probability models have been proposed. However, this may come at a cost of reduced specificity,1 which leads to unnecessary diagnostic imaging studies and thus limits the clinical utility of these approaches. Further improvements in the diagnostic approach to PE are clearly needed. There are emerging data demonstrating the accuracy of computed tomography pulmonary angiography (CTPA) for the diagnosis of PE. However, CTPA has limitations (a large contrast load, high radiation dose, and lack of sensitivity of first generation scanners for small thrombi2), some of which are evident in the recently published validation study referred to by Bailey:3 25% of screened patients with suspected PE were not eligible for this study because of renal impairment, a contraindication to CT, or other reasons. The diagnostic algorithm that we provided in our review suggests that either ventilation perfusion (V/Q) scanning or CTPA can be used for patients with suspected PE who require diagnostic imaging,2 with the choice determined by patient factors and availability. The diagnosis of PE during pregnancy is challenging because of concerns about radiation exposure and uncertainty about whether CTPA or V/Q delivers more radiation to the fetus.4 Furthermore, clinical decision rules have not been validated in pregnancy. However, recommendations from experts and professional bodies suggest that V/Q scanning can be used in combination with compression ultrasound to establish or exclude the diagnosis of PE during pregnancy in most cases with minimal fetal radiation exposure.5,6 The comments by Bailey and Bragg concerning the interpretation of high probability V/Q scan results highlight the pitfalls of performing diagnostic imaging without considering the patient’s pretest probability of PE. Although a high probability V/Q scan is diagnostic in patients with a moderate or high pretest probability of PE (prevalence of disease ≥ 90%), the prevalence of disease is only about 50% in those with a low pretest probability.7,8 Therefore, V/Q scanning should not be performed in patients with a low pretest probability unless the D-dimer test is positive. In this situation the algorithm for moderate or high pretest probability should be followed,2 and a high probability scan reliably establishes the diagnosis.

John W Eikelboom · Graeme J Hankey · Wai Khoon Ho · Cindy H Lee

Richard F O’Reilly,* Ian A Burgess,† Bernard Zicat‡ * Physician, † Radiologist, ‡ Orthopaedic Surgeon, Mater Misericordiae Hospital, Rocklands Road, North Sydney, NSW 2060. roreillyATbigpond.net.au To the Editor: In a recent editorial, Gallus estimates the cost of doing ultrasonography in all patients after unilateral hip or knee replacement, with further testing in the 9% or 26% of patients, respectively, found to have deep vein thrombosis (DVT), to be about $200 000 per 1000 patients.1 We agree. He then states, “Many would argue that extended prophylaxis is likely to be the simplest, cheapest and perhaps safest solution”. However, prophylaxis is also expensive. Subcutaneous enoxaparin 40 mg administered daily for 30 days costs $170, or $170 000 per 1000 patients.2 In our study, we found DVTs in 1086 of 5999 patients (18.1%) before discharge,3 so that extended prophylaxis would involve 81.9% of patients receiving prophylactic doses of anticoagulants, with the risk of unwanted bleeding, despite the absence of DVT on ultrasound at Day 7 postoperatively. In addition, if an ultrasound scan was not done before discharge, the 18.1% of patients with a DVT would receive only prophylactic (not therapeutic) doses of anticoagulant for their DVT. We plan a further study to check the prevalence of post-discharge DVT by repeating ultrasonography at 90 days postoperatively in patients without DVT on ultrasound at Day 7. We suspect the prevalence is lower than suggested in the literature, as the data on late presentation of DVTs have been obtained by retrospective study of the number of patients re-admitted to hospital with DVT. Finally, on the question of whether performing ultrasonography on all patients has clinical benefit, we concur with Gallus when he writes that “Logic suggests it should . . .”.

Richard F O’Reilly · Ian A Burgess · Bernard Zicat

In reply: O’Reilly and colleagues belatedly address the need to consider bleeding risk and costs when choosing between management routines designed to prevent venous thrombosis and pulmonary embolism. Their otherwise valuable article1 failed to record bleeding rates when patients (almost 17%) with subclinical calf-vein thrombosis were exposed to therapeutic (not prophylactic) anticoagulant dosages. Nor did they evaluate the dollar and manpower costs of their complex management routines. Present evidence-based international guidelines from the Seventh ACCP (American College of Chest Physicians) Conference on Antithrombotic and Thrombolytic Therapy recommend effective prophylaxis for at least 10 days in all patients having hip or knee replacement, extending to 28–35 days after hip replacement.2 The ACCP guidelines also recommend against routine use of ultrasound screening because it is “neither clinically effective nor cost effective”.2 This is a Grade 1A recommendation from the ACCP (“Grade 1” implies certainty “that the benefits do, or do not, outweigh the risks, burdens, and costs”; “Grade A” refers to recommendations based on “randomized clinical trials with consistent results [that] provide evidence with a low likelihood of bias”).3 To reverse this recommendation would require randomised comparisons between routine prophylaxis alone or routine prophylaxis supplemented by screening ultrasonography — powered to permit meaningful measures, in both groups, of thromboembolism rates, bleeding rates and costs. Routinely screening for subclinical thrombosis after major joint surgery should not be done outside suitably designed clinical trials until such evidence is available. The role of logic in medicine is to generate hypotheses, which must then be tested by clinical trial. Unfortunately, evidence derived from uncontrolled cohort studies remains limited to Grade C (based on “observational studies or [on] generalization from one group of patients included in randomized trials to a different, but somewhat similar, group of patients”).3

Alexander S Gallus

Pulmonary embolism (PE) affects 0.5–1 per 1000 people in the general population each year, and is one of the most common preventable causes of death among hospitalised patients. The clinical diagnosis of PE is unreliable and must be confirmed objectively with ventilation perfusion scanning or computed tomography pulmonary angiography. The diagnosis of PE can be reliably excluded, without the need for diagnostic imaging, if the clinical pretest probability for PE is low and the D-dimer assay result is negative. The initial treatment of PE is low-molecular-weight heparin or unfractionated heparin for at least 5 days, followed by warfarin (target international normalised ratio [INR], 2.0–3.0) for at least 3–6 months. Patients with a high clinical pretest probability of PE should commence treatment immediately while awaiting the results of the diagnostic work-up. Thrombolysis is indicated for patients with objectively confirmed PE who are haemodynamically unstable. Percutaneous transcatheter or surgical embolectomy may be life-saving in patients ineligible for, or unresponsive to, thrombolytic therapy. Unresolved issues in the management of venous thromboembolism include the roles of thrombophilia testing, thrombolysis for the treatment of stable PE patients who present with right ventricular dysfunction, and new anticoagulants; and the duration of anticoagulation for first unprovoked venous thromboembolism.

Cindy H Lee MB BS · Graeme J Hankey MD, FRACP, FRCP · Wai Khoon Ho MB ChB, FRACP, FRCPA · John W Eikelboom MB BS, FRACP, FRCPA

An antipodean’s history of haematology. Robert E Sage. Norwood, SA: Peacock Publications, 2004 (528 pp) ISBN 1 876087 96 X. Thomas Carlyle, 19th century Scottish historian and philosopher, once observed that “History is the essence of innumerable biographies.” And this is the style of An antipodean’s history of haematology by retired Australian haematologist, R Ed Sage. The book falls into two sections. In the first six chapters, Sage attempts to cover the history of haematology from antiquity to the 1960s, but sadly, clarity is drowned in the overwhelming detail. It is the following 17 chapters that make the book “a bloody good story”. Sage traces the history of antipodean haematology according to Carlyle’s tenet; he explores the professional lives of 99 Australasian haematologists. Indeed, this is the essence of the book’s uniqueness — the alignment of the professional lives of these haematologists with the growth of Australasian haematology, the foundation and growth of the Haematology Society of Australia (1961) and the Australasian Society of Blood Transfusion (1964), and events beyond. Sage presents insights into personalities and politics, including the revelations of the intransigence between transfusionists in Melbourne and Sydney over such monumental matters as the transfusion bottle! There were also the philosophical differences over what a haematologist is — a rancorous debate, which reached its peak in the 1970s, and embroiled the Austral(as)ian Colleges of Physicians and Pathologists, along with academics and mainstream haematologists. The bad blood was to ooze for years. Other stories include: the reasons for Melbourne’s erstwhile position as the power house of Australian haematology, a supremacy sustained by attracting the cream of Sage unearths some interesting facts. William Osler is called an English physician (he was Canadian); John Hunter is claimed to have been publishing and describing syndromes in 1817 (he died in 1793); the University of St Louis is said to be located in Mississippi (St Louis University is in Missouri); and Wollongong Hospital is said to be located in Sydney (the denizens of Wollongong, some 100 km south of Sydney, would be surprised to learn this). Nonetheless, the book is a significant history, adorned by innumerable mug shots of people in their prime or decline, and precious photos of local historical events. At its price, it is a bargain and should be on the shelves of all Australasian haematologists and haematology departments. I thoroughly recommend it. Martin B Van Der WeydenFormer Haematologist, and currently Editor, Medical Journal of Australia, Sydney, NSW

Objective: To evaluate the frequency and management of anaemia in Australian adults with solid and haematological malignancies.Design: 6-month observational, prospective, multicentre study.Participants: 694 patients recruited from outpatient oncology clinics in 24 hospitals in five Australian states between 9 April 2001 and 31 July 2001.Main outcome measures: Frequency of anaemia (haemoglobin [Hb] level < 120 g/L) at enrolment and over ensuing 6 months, by tumour type, disease status and cancer treatment; anaemia treatment and “trigger” Hb level for this treatment.Results: Participants had median age 60 years, and 61% were women. Prevalence of anaemia at enrolment was 35% (199/562), with 78% of these 199 having mild anaemia (Hb, 100–119 g/L). Frequency of anaemia (either present at enrolment or developing during the study) was 57% overall (323/566), and varied with tumour type, from 49% (lymphoma/myeloma) to 85% (urogenital cancer). Patients who received radiotherapy either in combination or concomitant with chemotherapy were more likely to have anaemia (73%) than those receiving chemotherapy alone (58%) (P = 0.004). Of all chemotherapy patients not anaemic at enrolment, 23% developed anaemia by the second monthly follow-up. Independent predictors for anaemia in chemotherapy patients were low baseline Hb level (odds ratio [OR], 5.4; 95% CI, 2.7–10.9) and use of platinum chemotherapeutic agents (OR, 4.8; 95% CI, 2.1–11.4) (P < 0.001). Anaemia was treated in 41% of patients with anaemia at enrolment — by transfusion (36%), iron (5%) and erythropoietic agents (2%). Frequency of anaemia treatment varied between tumour types, from 19% (breast cancer) to 60% (leukaemia). The mean “trigger Hb” for initiating transfusion was 95 g/L.Conclusions: Anaemia is prevalent among Australian patients with cancer managed in hospital oncology units. Its management varies between tumour types. Many patients do not receive treatment for their anaemia.

Tara Seshadri MB BS · H Miles Prince FRACP, MD · David R Bell FRACP · Paul B Coughlin FRACP, PhD · Philip P B James DM, FRACP · Gary E Richardson FRACP · Boris Chern FRACP, FAChPM · Peter Briggs FRACP · John Norman FRACP · Ian N Olver MD, PhD, FRACP · Chris Karapetis FRACP, MMedSc · John Stewart FRACP

Venous thromboembolism (VTE) affects 1–2 per 1000 people in the general population each year. Clinical diagnosis of deep venous thrombosis (DVT) is unreliable, and must be confirmed by compression ultrasonography or venography. A low clinical pretest probability of DVT and negative D-dimer result reliably exclude the diagnosis, with no need for diagnostic imaging. Initial treatment of DVT is with low-molecular-weight heparin or unfractionated heparin for at least 5 days, followed by warfarin (target INR, 2.0–3.0) for at least 3 months. A vena cava filter is indicated in patients who are ineligible for anticoagulant therapy or who experience embolism despite therapeutic anticoagulation. Thrombolysis or surgical embolectomy may be used as a limb-saving measure in patients with extensive proximal DVT and circulatory compromise that threatens the viability of the leg. Decisions regarding the optimal duration of anticoagulation to prevent recurrent VTE should be individualised and balance the risk of recurrence if warfarin is stopped against the risk of major bleeding and inconvenience of continuing treatment. The risk of recurrence is highest in people with recurrent unprovoked DVT or chronic predisposing factors (eg, cancer) who require indefinite anticoagulant treatment.

Wai Khoon Ho MB ChB, FRACP, FRCPA · Graeme J Hankey MD, FRACP, FRCP · Cindy H Lee MB BS · John W Eikelboom MB BS, FRACP, FRCPA

To the Editor: The recent position statement by the Warfarin Reversal Consensus Group provides clear and concise guidelines for a number of clinical scenarios related to the use of warfarin. 1 Unfortunately, it makes the general statement about the periprocedural management of warfarin in patients with atrial fibrillation (AF), “clinical experience suggests that bridging therapy is not required” [page 496]. Clinicians caring for patients with large ischaemic stroke in these circumstances may beg to differ. Although studies of bridging therapy in patients with AF in the periprocedural period are lacking, there are data which suggest that there is a considerably higher risk of thromboembolism during this period than would be expected by simply calculating the risk for several days off anticoagulation therapy.2-4 My own study of such patients undergoing endoscopy found a stroke risk of up to 3% in those at high risk.2 Many of these strokes were severe. The prothrombotic periprocedural environment may be a factor here, although advanced age and vascular risk factors may also contribute. The outstanding risk factor, however, is a previous history of stroke,2 and this is also a major risk factor for perioperative stroke in patients without AF.5 I would suggest careful, individualised assessment of all patients, and judicious bridging therapy where possible for patients with AF who have a past history of stroke.

David J Blacker

To the Editor: The recent position statement from the Warfarin Reversal Consensus Group provides a comprehensive, coherent and practical approach to warfarin reversal management. 1 In reviewing the position statement, and with particular reference to the paragraph about modifiers of warfarin response, we noted that the contribution of cytochrome P450 2C9 (CYP2C9) genotype to the response to warfarin was not addressed. There is debate in the current literature about the clinical utility of evaluating CYP 2C9 genotype in patients already taking or about to start warfarin therapy. Nonetheless, a significant body of evidence supports the contribution of CYP2C9 genetic variants as modifiers of response to warfarin therapy. CYP2C9 is the enzyme principally involved in metabolising warfarin.2 Several studies have identified the presence of single nucleotide polymorphisms in the CYP2C9 gene resulting in the expression of two allelic variants of CYP2C9 (CYP2C9*2 and CYP2C9*3) that are associated with reduced enzymatic activity, impaired metabolism of and increased sensitivity to standard warfarin doses.2,3 The allelic frequency of the mutant genotypes is in excess of 21% in the white population2 (they occur at reduced frequencies in African American populations and are rare in Asian populations4). The presence of allelic variants (CYP2C9*2 and CYP2C9*3) with reduced enzymatic activity is closely correlated with increased bleeding complications.2,3 Thus, there is potential for a considerable clinical impact given the large number of patients taking warfarin. We have identified an allelic frequency of CYP2C9*2 and CYP2C9*3 genotypes in an Australian population of patients attending an anticoagulant clinic comparable to that reported in the literature.2,3 We also identified international normalised ratios in excess of the target range in patients with the CYP2C9*2 or CYP2C9*3 genotype undergoing induction warfarin therapy with standard dosing regimens, relative to those who did not have these genotypes (personal, unpublished data, presented as: Cytochrome P450 CYP2C9 genotyping and warfarin induction therapy, presented at the 2004 Annual Scientific Meeting of the Haematology Society of Australia and New Zealand [Oct 17–20, Melbourne, Australia]). Recent reports suggest that CYP2C9 genotyping before inducing warfarin therapy may avert bleeding complications.5 However, CYP2C9 genotyping is currently only available within research institutes and larger corporations with research and development facilities and does not attract a Medicare rebate. While simple and inexpensive, genetic CYP2C9 screening has yet to be proven cost effective. However, genotyping may be of benefit in averting over-anticoagulation in certain clinical scenarios. These include commencing warfarin therapy in “high risk” elderly patients; those in whom low-dose, long-term, low-testing-frequency warfarin regimens are being contemplated; and in other “high risk” patients, such as those with conditions affecting warfarin metabolism, including liver disease, and in those taking medications known to interact with the hepatic metabolism of warfarin.

David J Blacker · Faye Gray · Keith Byron

To the Editor: The article by Baker et al was a timely review of managing anticoagulation therapy and balancing the risks of thrombosis and bleeding.1 However, in managing anticoagulation therapy before non-cardiac surgery in patients with mechanical cardiac valve prostheses, the suggested 5-day cessation of warfarin therapy, with only subcutaneous heparin cover, is not appropriate. I have had three patients with mechanical bileaflet mitral prostheses develop valve thrombosis while under this protocol, two with a fatal outcome. I have also had one patient with a mechanical bileaflet aortic valve develop a popliteal arterial embolus requiring thrombectomy, despite being treated according to the protocol. The consequences of valve thrombosis and thromboembolism far outweigh the lesser complications of increased bruising or bleeding associated with non-cardiac surgery. To avoid the potentially devastating complications of valve thromboembolism associated with the routine cessation of warfarin therapy 5 days before surgery, warfarin ought to be continued to maintain an INR (international normalised ratio) of around 2.0, supplemented with subcutaneous heparin. Alternatively, full intravenous heparinisation can be used while ceasing warfarin treatment, and continued postoperatively until the INR is restored to the therapeutic level. Warfarin should never be reversed with vitamin K, except in cases of life-threatening haemorrhage. Apropos of the therapeutic INR ranges generally recommended for mechanical cardiac valve replacements, the current generation of prostheses does not require the anticoagulation intensity of the older style prostheses.2,3 Lower intensity anticoagulation is sufficient to prevent thromboembolism at decreased risk of haemorrhagic complications.4 My personal practice for patients with bileaflet mechanical prostheses is to maintain an INR of 2.0–2.5 for aortic valves, and 2.5–3.0 for mitral valves. The higher intensity for mitral prostheses relates to potential increased thrombogenicity because of lower leaflet opening pressures, as well as the common association of left atrial dilatation and atrial fibrillation.

Serge Lubicz

To the Editor: A middle-aged woman with atrial fibrillation had her warfarin therapy stopped for 2 days before dental extraction. She had a catastrophic stroke and is now a plaintiff. I was asked if her medical management accorded with common practice. At the October 2004 Annual Conference of the Royal Australian College of General Practitioners, I conducted a straw poll of 20 experienced GPs, of whom 18 said they would stop warfarin for between 2 and 4 days before a dental extraction. Some of these GPs regarded a dental extraction as elective surgery and pointed me to authoritative (but slightly ambiguous) sources to back up their view. 1,2 However, a review of the medical and dental literature shows that this is an example of common practice lagging behind clinical evidence. The first controlled trial of dental extraction in patients on warfarin therapy was conducted in 1983.3 It showed that it was not necessary to cease warfarin prophylaxis for patients whose international normalised ratio (INR) was within the normal therapeutic range. Since then, two major literature reviews have confirmed these conclusions.4,5 A recent Australian review on warfarin reversal expresses a similar point of view.6 The incidence of a serious embolic complication from stopping therapy with warfarin is 1%, and this is three times more likely to occur than bleeding complications in patients whose warfarin therapy was continued.4 Furthermore, a stroke is a catastrophic event, while a bleeding tooth socket is simply messy and usually easily controlled. An authoritative review and position statement on warfarin therapy and dental procedures from the Australasian Society of Thrombosis and Haemostasis may be the catalyst required to align common practice with clinical evidence.

Max Kamien

In reply: We thank Blacker for his constructive and helpful comments. Our recommendations on bridging therapy in patients with atrial fibrillation were for patients with chronic atrial fibrillation who had not previously had a thromboembolic event.1 We do agree with Blacker that extreme care needs to be exercised in patients with atrial fibrillation and a previous thromboembolism. These patients should be managed along the same lines as patients who are at relatively high risk of recurrent thromboembolism. We also wish to emphasise that it is extremely important to assess each individual patient carefully, and to use the consensus guidelines as guiding principles, and not apply them blindly. Dear and his colleagues correctly point out that there are several published studies that have confirmed increased warfarin sensitivity in allelic variants of the cytochrome P450 2C9 (CYP2C9) enzyme. Polymorphisms associated with reduced enzymatic activity have been reported to be associated with increased warfarin sensitivity. They suggest that determining the genotype of individuals before commencing warfarin therapy may be of benefit in reducing the incidence of over-anticoagulation in a select group of patients. We do not believe that this approach is currently practical or possible. From a practical point of view we recognise several reasons why patients become over-anticoagulated when treated with warfarin. In our article we discussed several important modifiers that contribute to an individual’s sensitivity to warfarin. While we agree that polymorphisms of the CYP2C9 gene on its own have been linked with increased sensitivity to warfarin, we are not aware of any studies showing a synergistic interaction of the polymorphism with other clinically recognised causes of increased warfarin sensitivity. Furthermore, we are not aware of any properly conducted studies that have attempted to address the clinical or economic viability of screening for CYP2C9 polymorphisms in patients for whom warfarin therapy is planned. Finally, the time required to obtain the results of this investigation would preclude its application in the routine management of patients who require warfarin therapy. The letter by Lubicz highlights the difficulties encountered in bridging anticoagulant therapy in patients with prosthetic valves. As pointed out in our article, the management of these patients is controversial and mostly anecdotal.1 We believe that the recommendations in our article are useful for most patients, but would like to emphasise the need to consult with the relevant experts in order to avoid bleeding or thrombosis. We would not recommend routine full therapeutic anticoagulation therapy with heparin immediately after surgery, or the combined use of warfarin at any international normalised ratio (INR) with subcutaneous heparin before surgery. Such approaches are more likely to cause confusion and predispose the patients to either bleeding or the risk of thrombosis. Patients with prosthetic valves require careful handling, and involving experts in their management is critical. Kamien’s comments are important and illustrate the difficulties in changing entrenched practices. We hope that our recommendations will go some way to improving the way we manage patients on warfarin therapy who are about to undergo surgery.

on behalf of the Warfarin Reversal Consensus Group

Objective: To determine the prevalence of venous thromboembolism (VTE) after total hip replacement (THR), total knee replacement (TKR) or bilateral TKR in a large sample of patients in a major hospital orthopaedic unit.Design, setting and patients: The Mater Misericordiae Hospital, North Sydney, NSW, a 195-bed private hospital. All patients who had THR, TKR or bilateral TKR at the hospital between 1 April 1995 and 31 December 2001 had physical prophylaxis (graduated compression elastic stockings or intermittent pneumatic compression, or both) and chemical prophylaxis (anticoagulant) against VTE. All underwent ultrasonography of both legs before discharge, with a small, symptomatic group also undergoing a ventilation/perfusion lung scan (V/Q scan) and computed tomographic pulmonary angiography.Main outcome measures: Prevalence of deep-vein thrombosis (DVT) and symptomatic pulmonary embolism (PE) before discharge.Results: Among a total of 5999 patients, the pre-discharge prevalence of DVT after THR, TKR or bilateral TKR was 8.9%, 25.6% and 36.9%, respectively. The prevalence of symptomatic non-fatal in-hospital PE was 1.9%, while the prevalence of fatal in-hospital PE was 0.05%.Conclusions: Despite short-term chemical and physical thromboprophylaxis, the prevalence of DVT after lower-limb joint replacement, measured by pre-discharge ultrasonography, was high. The rate of symptomatic non-fatal in-hospital PE was moderate, but fatal in-hospital PE was rare.

Richard F O’Reilly MB BS, FRACP · Ian A Burgess MB BS, FRANZCR · Bernard Zicat MD, FRCSC, FRACS

A 28-year-old woman from Sudan who had lived for 9 years in Victoria, Australia, was diagnosed with falciparum malaria 2 months after splenectomy for massive splenomegaly of unknown cause. Chronic falciparum malaria can occasionally present years after leaving endemic areas in partially immune patients. It should be considered in such patients with presentations possibly related to malaria, including splenomegaly, anaemia, or a long history of intermittent fevers and chills. Infection with Plasmodium falciparum is well known as a cause of acute malaria among travellers from endemic areas, such as Africa and South-East Asia. However, chronic infection persisting for months may occur in endemic areas among those with a degree of partial immunity, and cases have been reported in people who left an endemic area up to 5 years previously.1,2 We report a case of falciparum malaria recurring 9 years after the patient migrated from Sudan to Victoria, Australia. The anopheles mosquito vector does not occur in this region of Australia and, if imported, is unlikely to survive for long. The patient initially presented with fever and massive splenomegaly, and the diagnosis of malaria was made after a splenectomy had been performed. Clinical recordA 28-year-old woman from Africa who had been living in Australia for 9 years presented to the emergency department with a one-week history of fever, rigors, abdominal pain, nausea and vomiting. She had no abnormalities on physical examination, apart from mild dehydration, and was discharged with a presumptive diagnosis of viral gastroenteritis. The next day, she presented to the gastroenterology outpatient clinic. Malaria was considered in the differential diagnosis, and she was admitted for further investigation and treatment. Past history: The patient was born in Eritrea and lived there for 6 years, followed by 12 years in Sudan, before migrating to Australia. She had had multiple episodes of malaria while in Africa, but could not recall exactly the drug therapy she received. In Australia, she experienced multiple episodes of nausea, abdominal pain and fever every 3 to 6 months, almost identical to her previous episodes of malaria. Although thick and thin malaria blood films were performed during each of these episodes in Australia, a diagnosis of malaria could not be confirmed. Ten months before current presentation: The patient’s general practitioner noted hepatosplenomegaly and, given her history of probable schistosomiasis exposure through freshwater irrigation canals in East Africa, tested her for schistosomiasis. Stool samples were positive for eggs of Schistosoma mansonii, and schistosoma serological tests were also positive (indirect haemagglutination titre, 256 [positive, > 32]; enzyme immunoassay IgM ratio, 1.5 [positive, > 1.2]). Computed tomography and ultrasound examination of the abdomen demonstrated massive splenomegaly (17 cm) and hepatomegaly (15 cm), with no radiological evidence of portal hypertension. Laboratory studies demonstrated anaemia and neutropenia consistent with hypersplenism. The platelet count could not be measured because of clumping (Box 1). She was treated with praziquantel, but continued to have episodes of abdominal pain and anorexia. Two months before current presentation: Because of these continuing episodes, as well as haematological evidence of hypersplenism, splenectomy was performed after appropriate vaccinations. Repeat thick and thin films before the splenectomy were again negative for malaria, although an immunochromatography card test (ICT) for malarial antigens was not performed. Initial histological examination of the spleen revealed only congestion and some mononuclear-cell infiltration. Current presentation: The patient had not travelled to a malaria-endemic area since arriving in Australia 9 years previously, nor had she been near an airport in the preceding 6 months. Results of haematological and biochemical tests are shown in Box 1. A rapid ICT for malaria antigen was performed (NOW ICT Malaria P.f/P.v. Test, Binax Inc, Portland, USA) and was positive for falciparum malaria (Box 2). The patient was admitted to hospital, and treatment begun with intravenous quinine (600 mg three times daily), as recommended for this form of malaria. The following day, thick and thin blood films were reported as showing 0.5% malaria parasitaemia. Parasite morphology, along with the prolonged period between the last possible exposure to malaria and illness, was considered consistent with Plasmodium malariae infection. Treatment was changed to oral chloroquine (620 mg initially, followed by 310 mg 6 hours later and on Days 2 and 3). Because of the conflicting results of the blood film and ICT test, polymerase chain reaction (PCR) tests for malarial antigens were performed at the Victorian Infectious Diseases Reference Laboratory and at the Institute of Clinical Pathology and Medical Research, Westmead Hospital, Sydney, NSW. Results of both tests a month later confirmed P. falciparum as the causative parasite. Further expert review of the blood film showed features consistent with P. falciparum (Box 3). Progress and follow-up: The fever, vomiting and abdominal pain resolved rapidly, and the patient was discharged after 3 days in hospital. After the positive PCR result for P. falciparum, she was treated again, as an outpatient, with atovaquone plus proguanil (1000 mg and 400 mg, respectively, daily for 3 days). Two weeks after completion of therapy, thick and thin films were found to be negative for malaria parasites. Serological tests for human immunodeficiency virus, hepatitis B and hepatitis C were all negative. A stored blood sample that had been taken 7 months before the current presentation was tested and found to be positive for malaria antibodies (immunofluorescent antibody titre, 160 [positive, > 20]). Further review of the computed tomography scan performed before splenectomy showed no features of chronic liver disease, and detailed histological review of the spleen again demonstrated congestion and a lymphocytic infiltrate (compatible with hyperreactive malarial splenomegaly), but no malarial parasites or pigment. At 6-month review, the patient reported no further symptoms. Screening of her children and husband for malaria (by thick and thin blood films and ICT) and schistosomiasis (by indirect haemagglutination assay) gave negative results. DiscussionThis case is notable because of the prolonged period (9 years) between the last possible exposure to malaria and the diagnosis of falciparum malaria, which was confirmed by PCR. Although it is well recognised that P. malariae infection can persist for many years,3 to our knowledge the longest previously reported delay between exposure and subsequent diagnosis of falciparum malaria is 5 years. In that case, the patient donated blood 5 years after leaving a malarious area, and was found to have falciparum malaria on testing after the recipient of the blood transfusion developed malaria; details in the report are limited.2 Recently, mathematical modelling was used to estimate the duration of P. falciparum infection after interruption of transmission.4 The authors estimated that the maximum duration of infection was about 4 years. Chronic falciparum malaria may occur in people who have lived in endemic areas and have developed partial immunity to the malaria parasite, resulting in low-grade parasitaemia.5 Antimalarial antibodies have been detected in high titres in such patients.6 Massive splenomegaly, now termed “hyperreactive malarial splenomegaly syndrome”,7 is a manifestation of chronic malaria. The demonstration of malarial parasites after splenectomy in patients not recently exposed to malaria raises the possibility of this syndrome.8 Major diagnostic criteria include: massive splenomegaly (> 10 cm) when no other cause can be found; immunity to malaria (ie, demonstration of antimalarial antibodies); and a clinical and immunological response (fall in antibody levels) to antimalarial therapy,9 or a significant reduction in spleen size and improvement in haematological parameters with antimalarial therapy.10 Our case highlights the fact that malaria may still present a major diagnostic challenge. In hindsight, it was likely that our patient had hyperreactive malarial splenomegaly syndrome. Use of other diagnostic tests, including PCR, before splenectomy might have enabled a trial of antimalarial treatment and possibly averted the need for surgery.9 Although thick and thin blood film examination using Field or Giemsa–Wright stain is the established “gold” standard for malaria diagnosis,11 repeated appropriate blood films in our patient before splenectomy were negative. An experienced laboratory can achieve sensitivity of 50 parasites/μL blood (0.001% red blood cells infected), but a survey of UK laboratories found that most achieved sensitivity of 500 parasites/μL blood when compared with a reference laboratory.12 As this case demonstrates, the new, more sensitive ICT card tests can be valuable in difficult-to-diagnose cases of P. falciparum infection. These tests detect circulating P. falciparum histidine-rich protein 2 (HRP-2) in whole blood and provide an immediate result, with sensitivity of 77%–98% and specificity exceeding 95% for falciparum malaria, correlating with counts of 100–300 parasites/μL blood.13 Other ICT kits that detect different antigens are available and can detect all four Plasmodium species.13 Although the sensitivity of these rapid antigen tests is good, a negative result does not exclude malaria. PCR techniques are even more sensitive, detecting levels as low as 5 parasites/μL blood,14 and are available on special request at reference laboratories around Australia. PCR can detect the specific plasmodial species and is therefore useful when morphological diagnosis is difficult, or when clinical suspicion warrants further attempts at diagnosis despite negative results from blood films and ICT. A recent study in Sudan showed that P. falciparum can survive for months in human hosts during the 9-month dry season, when no transmission occurs.5 Many people had ongoing PCR positivity for falciparum malaria, despite having levels of parasitaemia below the threshold for detection on thick and thin blood films. This phenomenon was demonstrated by our patient, whose blood films were repeatedly negative over years. However, splenectomy may unmask underlying chronic P. falciparum infection sufficiently to allow detection of parasitaemia on blood films.15 Chronic falciparum malaria should be considered in the differential diagnosis in patients from endemic areas presenting with symptoms possibly related to malaria, even years after their last possible exposure. Although thick and thin blood films remain the standard laboratory investigation, the relatively inexpensive and more sensitive malaria ICT card test should be a routine adjunct to blood films to detect P. falciparum. PCR testing may be warranted before excluding chronic malaria as the diagnosis. 1 Results of laboratory investigations Test Reference range 7 months before 2 months before* Current admission Haemoglobin (g/L) 115–165 105 104 97 White cell count (× 109/L)† Total 4.0–11.0 2.7 3.1 8.2 Neutrophils 2.0–7.5 0.91 0.95 1.64 Lymphocytes 1.0–4.0 1.39 1.71 5.58 Monocytes 0.1–0.8 0.38 0.38 0.9 Eosinophils < 0.4 0.02 0.07 0.08 Mean cell volume (fL) 82–95 90 85 84 Mean cell Hb concentration (g/L) 320–360 334 331 321 Albumin (g/L) 36–48 39 40 Bilirubin (μmol/L) < 18 23 26 ALP (U/L) 35–104 45 206 ALT (U/L) < 55 16 153 GGT (U/L) < 45 9 43 Urea (mmol/L) 2.5–7.7 4.3 1.9 Creatinine (mmol/L) 0.03–0.11 0.046 0.047 Thick and thin malaria blood films Negative Negative Positive‡ Hb = haemoglobin. ALP = alkaline phosphatase. ALT = alanine aminotransferase. GGT = γ-glutamyltransferase. * Pre-splenectomy. †Platelet count was not recordable because of clumping. ‡Films showed 0.5% parasitaemia; rapid immunochromatography card test and polymerase chain reaction tests were also positive for Plasmodium falciparum. 2 Rapid immunochromatography card test (ICT) for malaria Positive result for Plasmodium falciparum on immunochromatography card test (ICT) in our patient. The card test is performed on whole blood and gives a result within 10 minutes. 3 Thin blood film from the patient The ring form of Plasmodium falciparum (arrow) is apparent in a normal-sized red blood cell.

Benjamin P Howden FRACP, FRCPA · Gautam Vaddadi MB BS · M Lindsay Grayson MD, FRACP, FAFPHM · Joseph Manitta BAppSci(MLS), AIMS, MASM

In most cases, childhood leukaemia has a fetal origin, but multiple molecular events are required after birth for pre-leukaemic cells to progress to leukaemia. Cure rates for acute lymphoblastic leukaemia (ALL) now approach 80%. A high level of minimal residual disease detected by polymerase chain reaction in patients with ALL in remission has profound prognostic importance and is the focus of a major Australian study attempting to prevent relapse in these children. Greater awareness of the late effects of chemotherapy has led to changes in the treatment protocols for ALL, with improvement in neurocognitive outcomes and reduced rates of second malignancies. Pharmacogenetics is a new field of research that aims to enhance treatment efficacy by assessing the individual’s metabolism of and response to chemotherapeutic agents. Targeted therapies currently being developed show some promise of being able to further improve cure rates. Adolescents with ALL have a better prognosis if treated with paediatric rather than adult protocols.

David S Ziegler MB BS · Glenn M Marshall MB BS, FRACP · Luciano Dalla Pozza MB BS, FRACP · Keith D Waters MB BS, FRACP

Re: “Warfarin reversal: consensus guidelines, on behalf of the Australasian Society of Haemostasis and Thrombosis”, by Ross I Baker, Paul B Coughlin, Alex S Gallus, et al on behalf of the Warfarin Reversal Consensus Group (Med J Aust 2004; 181: 492-497). In Box 5, an asterisk in the Box title should have attributed Prothombinex-HT to CSL Limited, the manufacturer of all plasma-derived therapeutic products fractionated from Australian plasma. The error only appeared in the printed edition. The html and pdf versions of the article published in the eMJA were correct on publication.

Ross I Baker PhD, FRACP · Paul B Coughlin PhD, FRACP, FRCPA · Hatem H Salem MD, FRACP, FRCPA · Alex S Gallus FRACP, FRCPA, FRCP · Paul L Harper MD, FRACP, FRCPA · Erica M Wood FRACP, FRCPA

This bone marrow smear was taken from a child after chemotherapy for acute lymphoblastic leukaemia. The marrow showed no residual leukaemia, and provided this photographic evidence of her steps towards remission.

Annmarie A Bosco · Giselle Kidson-Gerber

For most warfarin indications, the target maintenance international normalised ratio (INR) is 2–3. Risk factors for bleeding complications with warfarin use include age, history of past bleeding and specific comorbid conditions. To reverse the effects of warfarin, vitamin K1 can be given. Immediate reversal is achieved with a prothrombin complex concentrate (PCC) and fresh frozen plasma (FFP). Vitamin K1 is essential for sustaining the reversal achieved by PCC and FFP. When oral vitamin K1 is used for warfarin reversal, the injectable formulation of vitamin K1 is preferable to tablets because of its flexible dosing; this formulation can be given orally or injected. To temporarily reverse the effect of warfarin when there is a need to continue warfarin therapy, vitamin K1 should be given in a dose that will quickly lower the INR to a safe, but not subtherapeutic, range and will not cause resistance once warfarin is reinstated. Prothrombinex-HT is the only PCC approved in Australia and New Zealand for warfarin reversal. It contains factors II, IX and X, and low levels of factor VII. FFP should be added to Prothrombinex-HT as a source of factor VII when used for warfarin reversal. Simple dental or dermatological procedures may not require interruption to warfarin therapy. If necessary, warfarin therapy can be withheld 5 days before elective surgery, when the INR usually falls to below 1.5 and surgery can be conducted safely. Bridging anticoagulation therapy for patients at high risk for thromboembolism should be undertaken in consultation with the relevant experts.

Ross I Baker PhD, FRACP · Paul B Coughlin PhD, FRACP, FRCPA · Hatem H Salem MD, FRACP, FRCPA · Alex S Gallus FRACP, FRCPA, FRCP · Paul L Harper MD, FRACP, FRCPA · Erica M Wood FRACP, FRCPA

Melagatran is a synthetic, small-peptide direct thrombin inhibitor with anticoagulant activity. Ximelagatran, an oral prodrug, undergoes rapid enzymatic conversion to melagatran. Melagatran has rapid onset of action, fixed twice-daily dosing, stable absorption, apparent low potential for medication interactions, and no requirement for monitoring drug levels or dose adjustment. There is no specific antidote, but the drug has a short plasma elimination half-life (about 4 hours). In clinical studies, melagatran/ximelagatran is not inferior to warfarin for stroke prevention in patients with non-valvular atrial fibrillation, to heparin–warfarin for acute treatment and extended secondary prevention of deep vein thrombosis, and superior to warfarin for prevention of venous thromboembolism after major orthopaedic surgery. Major bleeding with melagatran/ximelagatran occurred at rates similar to those in patients treated with warfarin. 6%–12% of patients taking ximelagatran develop asymptomatic elevated liver enzyme levels (predominantly alanine aminotransferase) after 1–6 months of therapy; this usually resolves with cessation of therapy. Less than 1% of patients develop abnormal liver function while taking ximelagatran; this rarely persists or develops into clinical illness.

Timothy A Brighton FRACP, FRCPA, MD

High-quality primary evidence from an Australian and New Zealand study provides a definitive answer The 1998 report of a meta-analysis by the Cochrane Injuries Group Albumin Reviewers presented an important public health issue and questioned the practices of many doctors in Australian intensive care units (ICUs).1 Using data from 24 studies involving 1419 patients, the meta-analysis reported that the administration of albumin-containing fluids to critically ill patients increased the absolute risk of death by 6%, suggesting one extra death for every 17 patients given albumin. The authors recommended that albumin should not be administered to critically ill patients outside the context of rigorously conducted, randomised trials. A subsequent meta-analysis did not resolve the issue of albumin’s safety in the critically ill.2,3 Due to its ready availability, albumin has been widely used in Australian ICUs. Even after the publication of the Cochrane review, half of all patients in surveyed ICUs in Australia received albumin during their ICU stay (R Bellomo, S Finfer, unpublished data). Extrapolating these results would mean that 50 000 patients received albumin in Australian ICUs each year and, if the Cochrane meta-analysis was correct, this would result in an additional 3000 deaths annually. Thus, the issue of albumin’s safety was of particular importance in Australia. The recent publication of the SAFE (Saline versus Albumin Fluid Evaluation) study in the New England Journal of Medicine4 not only brings certainty to the issue of albumin’s safety in a heterogeneous population of adult ICU patients, it also marks the coming of age of clinical research in Australian and New Zealand ICUs. The SAFE Study, a collaboration of the Australian and New Zealand Intensive Care Society Clinical Trials Group, the Australian Red Cross Blood Service, and The George Institute for International Health, was a double-blind, randomised controlled trial of albumin versus saline for fluid resuscitation involving 6997 patients. Conducted in 14 ICUs in Australia and two in New Zealand, it was funded by the National Health and Medical Research Council and the Health Research Council of New Zealand, and by direct grants from the Australian federal, state and territory governments, the two New Zealand hospitals and CSL Ltd. Internationally, the SAFE Study is the largest randomised controlled trial conducted in intensive care or transfusion medicine to date. An accompanying editorial acknowledged that the SAFE Study heralded a new era in critical care marked by the large, simple randomised trial.5 The design of the study reflected the SAFE collaborators’ desire to conduct a definitive trial that would answer an important clinical question and provide results that would be widely applicable in ICUs around the world. As a result, the trial sought to include as many adult patients admitted to participating ICUs as possible by using broad, simple inclusion criteria and few exclusion criteria. The main inclusion criterion was that the treating clinician believed that fluid administration was indicated for the treatment of intravascular volume depletion. The primary outcome measure was all-cause mortality 28 days after randomisation. The only broad patient groups excluded were those admitted to the ICU after liver transplantation, burns or cardiac surgery. The study enrolled 6997 patients in 69 weeks, an average recruitment rate of 101 patients per week. The rapid recruitment rate was made possible by the commitment of the clinical staff in the participating ICUs, the provision for delayed consent, and web-based randomisation and fluid distribution. The blinded study design was possible as both study fluids were manufactured by CSL Ltd and packaged in specially designed blinding materials6 before distribution by the Australian Red Cross Blood Service, New Zealand Blood Service and participating centres’ blood banks. What did the study find? The primary outcome was available for 99.1% of the 6997 patients randomised; of these, 726 assigned albumin (20.9%) and 729 assigned saline (21.1%) had died by Day 28. The relative risk of death for patients assigned albumin compared with patients assigned saline was 0.99 (95% CI, 0.91–1.09; P = 0.87). There were no differences in secondary outcomes, with patients assigned albumin and saline having similar incidences of new organ failure, similar durations of mechanical ventilation and renal replacement therapies, and similar ICU and hospital lengths of stay. Key findings of the study are shown in the Box. The study identified six predefined subgroups: patients with and without trauma, with and without severe sepsis, and with and without acute respiratory distress syndrome (ARDS). As a previous meta-analysis had suggested that trauma patients resuscitated with colloid solutions had higher mortality than those resuscitated with crystalloid solutions,7 trauma was a stratification variable in the study. Patients with severe sepsis and ARDS were identified at baseline to determine whether the increased capillary permeability to albumin seen in those conditions8,9 resulted in a treatment effect that differed from that seen in the study patients without those conditions. Within the predefined subgroups there was limited evidence of a treatment effect favouring saline in patients with trauma, and favouring albumin in patients with severe sepsis. The possibly detrimental effect of albumin in patients with trauma was limited to patients with evidence of traumatic brain injury, namely those patients admitted to the ICU as a result of trauma who had a documented unsedated Glasgow Coma Scale score less than 14 and evidence of brain injury on cerebral computed tomography. The investigators cautioned readers that such subgroup differences frequently occur by chance, and the accompanying editorial advised cautious interpretation of the subgroup findings.5 Thus, the study demonstrated that, in this heterogeneous population of adult ICU patients, albumin can be considered safe, without demonstrating any clear efficacy advantage over saline. The SAFE Study achieved its goal of providing a definitive answer to an important clinical question. The result is widely applicable in those ICUs around the world where purified albumin is available. In addition, the study has demonstrated that investigators in Australian and New Zealand ICUs are capable of conducting large-scale collaborative trials on modest budgets and in a realistic timeframe. The SAFE Study has been described as a landmark study that heralds a new era in critical care.5 We hope that it will be only the first of many such studies. Key findings of the SAFE Study4 6997 patients were randomised to receive either albumin (3497) or saline (3500). The primary outcome (alive or dead at 28 days) was available for 6933 patients (99.1%). No significant difference was seen between the albumin and saline groups in: 28-day all-cause mortality (20.9% v 21.1%; P = 0.87) days in the intensive care unit (6.5 [SD, 6.6] v 6.2 [SD, 6.2]; P = 0.44) days in hospital (15.3 [SD, 9.6] v 15.6 [SD, 9.6]; P = 0.30) days of mechanical ventilation (4.5 [SD, 6.1] v 4.3 [SD, 5.7]; P = 0.74) days of renal replacement therapy (0.5 [SD, 2.3] v 0.4 [SD, 2.0; P = 0.41)

Simon R Finfer FRCA, FRCP, FJFICM · Neil W Boyce FRACP, PhD · Robyn N Norton PhD, MPH

Randomised trials suggest that ximelagatran is “non-inferior” to warfarin for preventing stroke in patients with non-valvular atrial fibrillation, but important questions remain Atrial fibrillation is a strong and independent risk factor for stroke because it predisposes to thrombus formation in the left atrial appendage, and subsequent embolism to the brain.1 Each year, at least 6000 cardioembolic ischaemic strokes occur among an estimated 150 000 Australians with atrial fibrillation,2,3 and these numbers are expected to rise substantially with the ageing of the Australian population and associated increase in the prevalence of atrial fibrillation. The only two treatments proven to reduce the risk of stroke among patients with atrial fibrillation are aspirin and adjusted-dose warfarin.4 However, both have limitations. Aspirin is only modestly effective, reducing the risk of stroke by about a fifth compared with placebo (absolute risk reduction [ARR], 1.7% per year; number of patients needed to treat for one year to prevent one stroke [NNT], 59). Warfarin reduces the risk of stroke by about two-thirds compared with placebo (ARR, 3.1% per year; NNT, 32) and by about a third compared with aspirin (ARR, 0.8% per year; NNT, 125), but causes at least twice as many intracranial and extracranial bleeds as aspirin, particularly in patients at increased risk of bleeding (eg, those aged over 75 years, those with a history of bleeding; see Box 1).4 Warfarin is also inconvenient to use because it has a narrow therapeutic index, interacts with numerous drugs and food, and requires close laboratory monitoring (Box 2).5 Consequently, only a third to a half of patients with atrial fibrillation who are appropriate candidates for warfarin therapy actually receive it.6 Reducing the intensity of warfarin therapy to an international normalised ratio (INR) of less than 2.0 lowers the risk of bleeding, but is associated with an increased incidence of ischaemic stroke and worse stroke outcomes compared with standard-intensity warfarin therapy (INR ≥ 2.0).7 Direct thrombin inhibitors are a new class of anticoagulant drugs that bind directly to thrombin and block its interaction with substrates, thus inhibiting fibrin formation, thrombin-mediated activation of coagulation, and thrombin-induced platelet aggregation. Hirudin is the only direct thrombin inhibitor currently available for use in Australia, but must be given parenterally and is approved only for the treatment of heparin-induced thrombocytopenia. Ximelagatran, a pro-drug of melagatran, is an orally administered direct thrombin inhibitor and the newest drug in this class. It is rapidly absorbed from the gut and converted to its active form, melagatran. Melagatran is not metabolised or bound to plasma proteins. It is cleared predominantly (about 80%) by the kidneys, and has a half-life of 4–5 hours, which means ximelagatran needs to be administered twice daily (Box 2). Two large phase III randomised trials have recently evaluated ximelagatran as a replacement for warfarin to prevent thrombotic complications in patients with non-valvular atrial fibrillation.8,9 The primary objective of the Stroke Prevention using the ORal direct Thrombin Inhibitor ximelagatran in patients with non-valvular atrial Fibrillation (SPORTIF) III and V trials was to determine whether ximelagatran given in a fixed dose of 36 mg twice daily without laboratory monitoring was non-inferior to adjusted-dose warfarin (INR, 2.0–3.0) for the prevention of stroke or systemic embolism in patients with non-valvular atrial fibrillation and at least one additional major risk factor for stroke. The prespecified criterion for non-inferiority required that the lower confidence interval for the difference in the rate of stroke or systemic embolism between ximelagatran and warfarin did not exceed the prespecified threshold of 2% per year.10 Establishment of non-inferiority would imply that ximelagatran has either equivalent or superior effectiveness to warfarin and would allow clinicians to select ximelagatran over warfarin for convenience or safety. The design of the two SPORTIF trials was identical, except that SPORTIF III (3407 patients) was conducted in Europe, Asia, Australia and New Zealand and treatment allocation was open label, while SPORTIF V (3922 patients) was conducted in North America and treatment allocation was double blinded. The pooled results of the SPORTIF III and V trials (which had mean follow-up periods of 17 or 20 months, respectively) showed no significant difference in the risk of stroke or systemic embolism between ximelagatran (2.5%) and warfarin (2.5%; Box 1). In both trials, findings for ximelagatran fulfilled the criterion for non-inferiority. However, the pooled results of the SPORTIF trials also showed that ximelagatran significantly reduced the risk of major bleeding compared with warfarin (2.5% for ximelagatran; 3.4% for warfarin; estimated annualised ARR, 0.6%; NNT for 1 year to avoid one major bleed, 167) and increased the risk of transiently elevated levels of liver alanine aminotransferase (ALT) enzymes (6.1% for ximelagatran; 0.8% for warfarin; absolute risk increase [ARI], 5.3%; number of patients needed to treat with ximelagatran to harm [NNH] with increased ALT, 19). Raised ALT levels typically occurred 2–6 months after initiation of ximelagatran therapy, but produced no symptoms, were transient (returning to baseline spontaneously or after cessation of treatment), and without sequelae in all cases reported in the SPORTIF trials. These results suggest the beginning of the end of warfarin, because ximelagatran is not only associated with less major bleeding than warfarin, but it also has a predictable pharmacokinetic profile (uninfluenced by the patient’s age, sex, weight, ethnicity or diet). Therefore, it is not necessary to monitor anticoagulation activity or adjust the dose of ximelagatran (except in patients with renal dysfunction, in whom a decrease in dose or longer dosing interval is likely to be required). Furthermore, ximelagatran has a wider therapeutic margin than warfarin, and a low potential for drug interactions (Box 2). Although the cost of ximelagatran is likely to be substantially higher than the cost of warfarin, it may prove to be more cost effective because of its lower risk of bleeding and superior convenience (eg, no laboratory monitoring). Yet, important questions remain. First, there was significant heterogeneity between the two SPORTIF trials (P = 0.02). In the SPORTIF III trial, random allocation to open-label ximelagatran was associated with an absolute reduction in stroke or systemic embolism of 0.7% per year compared with warfarin, whereas in the SPORTIF V trial allocation to double-blinded ximelagatran was associated with an absolute increase in stroke or systemic embolism of 0.4% per year compared with warfarin. The cause of this heterogeneity remains uncertain, but might, at least in part, be accounted for by diagnostic suspicion or reporting bias in the open-label SPORTIF III trial. Second, the 2% per year threshold that was chosen as the criterion for non-inferiority does not reliably exclude even a near doubling of risk of stroke or systemic embolism with ximelagatran compared with warfarin. Third, unexpected hepatic side-effects of ximelagatran are an important concern given their high incidence in the short-term (6%), the large population potentially eligible for ximelagatran, and the likely long-term exposures to ximelagatran (and possibility of other long-term adverse effects). Monitoring of liver function is likely to be required during the first 6 months of treatment, and additional long term outcome data are required. The SPORTIF data signal the emergence of ximelagatran as an effective, safe and more convenient long-term alternative to warfarin for preventing stroke in patients with non-valvular atrial fibrillation. Safety concerns and cost issues are likely to delay its approval and eventual uptake by clinicians in Australia. In the meantime a range of other new antithrombotic drugs are also being evaluated for this indication. Both idraparinux (a selective clotting factor Xa inhibitor administered by once-weekly subcutaneous injection) and the combination of aspirin and clopidogrel are being tested in clinical trials, and novel oral preparations of direct-thrombin inhibitor and factor Xa inhibitors are in clinical development. This is heartening news for patients with atrial fibrillation, their doctors, and also public health professionals and governments faced with a looming epidemic of morbidity caused by atrial fibrillation in the ageing Australian community. 1: Estimated benefits and risks of treating a typical cohort of 1000 patients with non-valvular atrial fibrillation with aspirin, adjusted-dose warfarin, or ximelagatran* Aspirin (v placebo) Warfarin (v placebo) Warfarin (v aspirin) Ximelagatran (v warfarin)† Stroke‡ ARR ↓ 17 NNT 59 ARR ↓ 31 NNT 32 ARR ↓ 8 NNT 125 ARR§ 0 NNT§ — Major extracranial bleeds¶ ARI ↑ 1 NNH 1000 ARI ↑ 3 NNH 333 ARI ↑ 2 NNH 500 ARR ↓ 6 NNT 167 ALT ≥ 3 times upper limit of normal — — — — — — ARI ↑ 53 NNH 19 ARR =absolute risk reduction. ARI = absolute risk increase. NNT = number of patients needed to treat for one year to prevent or avoid one event. NNH = number of patients needed to treat for one year to harm by causing one event. ALT = alanine aminotransferase. * Data for aspirin v placebo, warfarin v placebo, and warfarin v aspirin are adapted from Hart et al.4 † ARR and NNT in the SPORTIF trials were calculated by dividing the pooled event rate by the mean duration of follow-up in years (approximately 1.5 years). ‡ Includes haemorrhagic stroke. § Includes stroke and systemic embolism. ¶ Event rates are likely to be substantially higher outside clinical trial settings, in the elderly, and in those with major comorbid conditions. 2: Comparison of the pharmacology and costs of aspirin, adjusted-dose warfarin and ximelagatran to prevent stroke in patients with non-valvular atrial fibrillation Aspirin Warfarin Ximelagatran Route Oral Oral Oral Dose 150–325 mg Variable* 36 mg Frequency Once daily Once daily Twice daily Half-life 20 minutes 40 hours 4–5 hours Clearance Systemic Hepatic Renal† Laboratory monitoring Not required INR Liver function tests‡ Antidote No Yes — Vitamin K No Reversal of antithrombotic effect Platelet transfusion Vitamin K Fresh frozen plasma Prothrombinex Discontinue ximelagatran Maintain diuresis Haemodialysis Food interactions Nil Multiple Nil known Drug interactions Uncommon Multiple Nil known Major side-effects Gastrointestinal bleeding Bleeding Bleeding Abnormal liver function test results Precautions and contraindications Bleeding diathesis Peptic ulcer Allergy Bleeding diathesis Alcoholism Dementia Impaired liver function Bleeding diathesis Impaired renal function Impaired liver function Approximate costs $2 per month§ $10 per month§ Unknown¶ INR = international normalised ratio. * Dose adjusted according to the results of the INR. † Trials of ximelagatran in atrial fibrillation have been restricted to patients with a creatinine clearance rate of ≥ 30 mL/min. ‡ Monitoring of liver function is likely to be required for the first 6 months. § Pharmaceutical Benefits Scheme November 2003: aspirin, $6.13 for 112 100 mg enteric-coated tablets; warfarin, $8.40 for 50 5 mg tablets (does not include the cost of laboratory monitoring). ¶ Cost of ximelagatran is not known but is likely to be at least $100 per month for a private prescription.

John W Eikelboom MSc, FRACP, FRCPA · Graeme J Hankey MD, FRCP, FRACP

Unravelling the uncertainties The antiphospholipid antibody syndrome is an autoimmune disease that commonly presents with either venous thromboembolism (VTE) (deep vein thrombosis, pulmonary embolism), arterial thrombosis (ischaemic stroke, coronary thrombosis, peripheral arterial occlusion) or unexplained fetal loss. A high index of suspicion for the diagnosis is raised in patients with unusual thrombosis, those without obvious risk factors, or those who experience recurrent events. The diagnosis is based on presentation with one of the above clinical criteria and detection in the laboratory of persistent antiphospholipid antibodies. Clinically relevant antiphospholipid antibodies are identified by either functional coagulation assays (lupus anticoagulant) or immunoassays (anticardiolipin antibody, β2-glycoprotein-1 antibody).1,2 When considering the diagnosis of the antiphospholipid antibody syndrome, it is important to perform both clotting and serological tests, as they are concurrently positive in 50% of unequivocal cases.3 The detection of a lupus anticoagulant and/or medium- to high-titre IgG antiphospholipid antibodies is associated with an approximately 10-fold increase in risk of VTE.4,5 Unlike the treatment for other autoimmune disorders, anticoagulation rather than immunosuppression is the mainstay of treatment.1-3 About 40 000 tests for thrombophilia are performed in Australia each year . . . Why are we performing these tests? Significantly elevated levels of antiphospholipid antibodies are detected in up to 20% of patients with VTE, compared with only 2% of healthy adults.6 This means that clinicians will be frequently asked to make decisions based on positive antibody test results. With about 40 000 tests for thrombophilia performed in Australia each year,7 the question arises as to the impact of testing for antiphospholipid antibodies in clinical practice. Why are we performing these tests? Up to now the reason has been twofold. Firstly, there is a perception that there is a high risk of recurrent thrombotic episodes, and even death, when anticoagulation treatment is stopped in people with antiphospholipid antibodies. Several retrospective studies have suggested that treatment with warfarin should be continued for at least 12 months, if not longer or lifelong, as up to 70% of patients will have recurrent events without ongoing treatment.8,9 For patients with arterial occlusion, particularly those without other atherosclerotic risk factors (eg, smoking, diabetes, hyperlipidaemia, hypertension), recurrence of thrombosis from significant levels of antiphospholipid antibodies would be catastrophic. For such patients, anticoagulation is usually continued. For patients with VTE and no obvious risk factors, testing of antibody levels may help to resolve the current uncertainty about the optimal duration of anticoagulation treatment and may even help to determine whether anticoagulation should be stopped at all. In a multicentre study, 412 patients were prospectively followed after a first episode of VTE: within 4 years after ceasing warfarin, 29% of patients with IgG anticardiolipin antibodies had experienced further events, compared with 14% without antibodies.6 However, maintaining people on long-term warfarin therapy carries the inevitable risk of serious haemorrhage (1.1 events/100 patient-years) and anticoagulant-related death (0.25 deaths/100 patient years).10 If we had further data on the optimal duration of anticoagulation treatment, we would be in a better position to discuss with patients the risk versus benefit of ongoing treatment. The second reason for determining the presence of antiphospholipid antibodies is that it assists in deciding the optimal intensity of anticoagulation treatment. There is evidence from retrospective case series that recurrence of thrombotic episodes can occur even with moderate-intensity warfarin therapy (target international normalised ratio [INR], 2.0–3.0) when compared with higher-intensity therapy (target INR, 3.0–4.5).8,9 However, this information comes from specialised expert clinics for high-risk patients, and to generalise the findings to a large number of VTE patients without better evidence is problematic. The cost of having a higher INR (3.0–4.4 rather than 2.0–2.9) is the doubling of the risk of any bleeding event.10 A recent prospective randomised controlled trial on the efficacy and safety of high-intensity warfarin treatment (target INR, 3.0–4.0) versus standard-intensity treatment (target INR, 2.0–3.0) in 114 patients with arterial thrombosis and VTE with antiphospholipid antibodies attempted to address this uncertainty.11 The results of the trial showed that there was no difference in the rates of recurrent thrombosis and bleeding between the two groups. Although the confidence intervals were wide because the number of events was small, the study does give some guidance regarding treatment decisions for the majority of patients with thrombosis and antiphospholipid antibodies. However, there are several caveats: Patients were included after already receiving warfarin for variable periods of time, and only a third of patients were allocated to one of the trial groups within 6 months of thromboembolism. This means that patients who were at highest risk of recurrence or bleeding while on warfarin were excluded; Patients who already had recurrent thrombosis while receiving warfarin with a target INR > 2.0 were specifically excluded; Control of INR range was supervised by expert clinic staff, who kept INRs within range 70% of the time for the standard-intensity arm and 40% of the time for the high-intensity arm. Most of the patients out of the INR target range in the high-intensity group were below the therapeutic range. Whether this degree of control for a given target INR is achieved in everyday practice is uncertain and needs further validation. It highlights the difficulty of maintaining a narrow therapeutic target INR range with warfarin despite the best intentions. Given the limitations of the published studies, the clinician is still faced with uncertainty in dealing with patients with thrombosis and antiphospholipid antibodies. For most people with thrombosis and antiphospholipid antibodies, long-term standard-intensity warfarin with a target range INR 2.0–3.0 may be appropriate. However, there will still be patients with extensive or unusual thrombosis for whom higher-intensity anticoagulation treatment could be considered. This situation applies particularly to people in whom recurrence may be threatening to life, limb or organ, particularly when no other recognised risk factor for either VTE or atherothrombosis can be removed or modified. To reliably answer the question of how to manage patients with thrombosis and a significant level of antiphospholipid antibodies would require an inception cohort study, with patients randomised at diagnosis to different intensities of anticoagulation treatment.

Ross I Baker MB BS, FRACP, FRCPA

The recently released PREVENT trial provides some answers Venous thromboembolism (VTE) affects 1–2 people per 1000 in the general population each year.1 It most commonly manifests as deep vein thrombosis of the leg, or as pulmonary embolism. There are many acute provoking factors or triggers (eg, major trauma, recent surgery), and many chronic predisposing factors, both genetic (eg, factor V Leiden) and acquired (eg, cancer). Most patients with provoked VTE have a low risk of recurrence (0–4% per year without anticoagulation), presumably because most have no major predisposing factors for VTE.2 Treatment for provoked VTE is short term and consists of giving intravenous unfractionated heparin or subcutaneous low-molecular-weight heparin (LMWH) for at least 5 days, followed by warfarin (target international normalised ratio [INR], 2.0–3.0) for 3 months.3 Further antithrombotic therapy is usually not required unless patients are re-exposed to known triggers for VTE. Standard-intensity therapy with warfarin remains the treatment of choice for the long-term prevention of recurrent VTE in patients who are at highest risk of recurrence Most patients with unprovoked VTE, however, have a higher risk of recurrence (≥ 5% per year without anticoagulation) over many years.4,5 This is because they are chronically exposed to one or more underlying genetic or acquired predisposing factors for VTE, which may be identifiable from the clinical history or through laboratory testing. Furthermore, the absence of a provoking factor or trigger for VTE is the most important predictor of recurrence in these patients. They require longer-term or indefinite treatment, which consists of giving intravenous unfractionated heparin or subcutaneous LMWH for at least 5 days, followed by warfarin (target INR 2.0–3.0).3 This is standard-intensity anticoagulation therapy, and is highly effective in preventing recurrent episodes of VTE for as long as it is continued. In trials in which patients were treated for a median of 4–6 months, it reduced the absolute risk by 7.6%, which is equivalent to preventing one event for every 13 patients treated (odds ratio [OR], 0.15; 95% CI, 0.10–0.23).6 In patients considered at highest risk of recurrent unprovoked VTE (eg, > 10% per year; see Box), warfarin therapy is continued indefinitely, whereas in most patients, it is discontinued after 6–12 months.3 This is because long-term anticoagulation is associated with a cumulative risk of bleeding, which is perceived to outweigh its benefits in preventing recurrent VTE. Standard-intensity therapy with warfarin causes minor "nuisance" bleeding in 5%–15%, major bleeding in 2%–3%, and fatal bleeding in 0.2%–0.6% of patients each year.7 A hitherto burning question for patients with unprovoked VTE is whether there are other anticoagulant treatment regimens with a more acceptable benefit-to-harm ratio, such as lower-intensity oral anticoagulation therapy. The recently reported Prevention of Recurrent Venous Thromboembolism (PREVENT) trial was initiated in July 1998 to test the hypothesis that long-term, low-intensity warfarin therapy (target INR, 1.5–2.0) might provide a safe and effective method of reducing the risk of recurrent VTE among patients who had a previous idiopathic (unprovoked) venous thrombosis.8 After completing at least 3 months of standard-intensity warfarin therapy (target INR, 2.0–3.0), 508 patients were randomly allocated to receive low-intensity warfarin therapy or placebo in a double-blinded fashion. The trial was terminated after a mean follow-up duration of 2.1 years because there was strong evidence of efficacy of warfarin. Of 253 patients assigned to placebo, 37 had recurrent venous thromboembolism (7.2 per 100 person-years), compared with 14 of 255 patients assigned to low-intensity warfarin therapy (2.6 per 100 person-years). This represents a relative risk reduction of 64% (hazard ratio [HR], 0.36; 95% CI, 0.19–0.67; P < 0.001), and an absolute risk reduction of 4.6%, equivalent to one event prevented for every 22 patients treated for 1 year. Bleeding episodes necessitating hospitalisation occurred in two patients in the placebo group (0.4 per 100 person-years), and five patients in the warfarin group (0.9 per 100 person-years); this difference was non-significant (P = 0.25).8 Although the PREVENT trial showed no significant excess of major bleeding with low-intensity warfarin therapy compared with placebo, event rates were low (5 v 2), and the 95% confidence intervals do not reliably exclude even a 13-fold increase in risk of major bleeding (HR, 2.53; 95% CI, 0.49–13.03). Yet, there is no doubt that low-intensity warfarin causes bleeding. In the PREVENT trial, "minor" bleeding was significantly increased in the warfarin group compared with the placebo group (12.8% v 6.7%; HR, 1.92; 95% CI, 1.26–2.93), with an increase in absolute risk of 6.1%, equivalent to one minor bleed caused for every 16 patients treated for 1 year. The results of the PREVENT trial indicate that low-intensity warfarin therapy is effective for long-term prevention of recurrent VTE. However, it was not shown to be sufficiently superior to placebo for low-intensity warfarin to be adopted for this indication. Standard-intensity warfarin is also superior to placebo when continued for up to 4 years after an initial thrombotic event.6,9-11 Indeed, it almost eliminates the risk of recurrent VTE in patients who continue the therapy, but is not routinely used because of the bleeding risks. Mini-dose warfarin therapy (fixed-dose, 1–2 mg daily) has never been shown to be effective for this indication, while low-intensity warfarin therapy is unlikely to offer any advantages over standard-intensity therapy in terms of convenience, and would only be a viable alternative if it were significantly safer. Indirect comparisons of the relative effectiveness and safety of low-intensity and standard-intensity therapy with warfarin, compared with placebo, are unreliable.8-11 For example, the apparently lower rates of bleeding in the PREVENT trial when indirectly compared with previous trials of warfarin might simply be explained by differences in patient selection. The PREVENT trial randomly allocated patients to treatment or placebo after they had completed a median of 6.5 months of warfarin treatment, and also included a 28-day run-in phase. It is thus likely that patients at increased risk of bleeding were excluded from the long-term phase of the study. By contrast, in most previous trials of long-term standard-intensity therapy with warfarin, patients were randomly allocated after no more than 3 months of treatment. This is as unreliable as comparing two sporting teams by their respective performances against another team rather than having them oppose each other directly. Indeed, the results of a recent direct head-to-head randomised comparison showed that low-intensity warfarin therapy was not only less effective than standard-intensity therapy for preventing recurrent VTE (absolute risk increase of 1.3% per patient year, equivalent to one event caused for every 77 patients treated for 1 year), but provided no advantage in terms of major bleeding (1.0% v 0.9% per patient-year; HR, 1.0; 95% CI, 0.4–2.7) or minor bleeding (4.9% v 3.6% per patient-year; HR, 1.3; 95% CI, 0.8–2.1).12 Taken together, these results indicate that standard-intensity therapy with warfarin is more effective for preventing recurrent VTE than low-intensity warfarin therapy, which, in turn, is more effective than placebo. However, because low-intensity warfarin therapy does not appear to be any safer in terms of bleeding and still requires close laboratory monitoring, it is difficult to justify this approach as an alternative to standard-intensity therapy for the long-term prevention of VTE, irrespective of a patient's baseline risk of recurrence or bleeding. The implications of these results for clinicians are that standard-intensity therapy with warfarin (target INR, 2.0–3.0) remains the treatment of choice for the long-term prevention of recurrent VTE in patients who are at highest risk of recurrence (eg, history of recurrent unprovoked VTE, major predisposing factor such as cancer; see Box) or with an initial life-threatening event (eg, major pulmonary embolism), and low risk of haemorrhagic complications. For patients with a first episode of unprovoked VTE or at increased risk of haemorrhagic complications, to decide about long-term treatment, doctors need to weigh the absolute risks of recurrent VTE and bleeding complications with and without warfarin treatment in each patient. In most cases, this is likely to result in the discontinuation of treatment after 6–12 months. The implications of these results for researchers are that more data are required to improve the reliability of clinical and laboratory predictors of recurrent VTE and haemorrhagic complications in individual patients, and that randomised controlled trials are required to evaluate the effectiveness and safety of alternative long-term antithrombotic therapies (eg, ximelagatran,13 antiplatelet agents) that are likely to be more convenient or have a more favourable benefit-to-risk profile than either standard-intensity or low-intensity warfarin therapy. Major determinants of the risk of recurrent venous thromboembolism Low risk (0–4% per year) Provoked event* Isolated distal deep vein thrombosis Intermediate risk (5%–10% per year) First unprovoked event Major predisposing factor(s)† Highest risk (> 10% per year) More than one unprovoked event First unprovoked event plus major predisposing factor(s)† Active cancer * Provoking factors include, in the last 3 months: hospitalisation, major surgery, trauma, leg fracture, plaster cast, puerperium. †Major predisposing factors include: prolonged immobility, neurological disease with paresis, homozygosity for factor V Leiden, combined (multiple) thrombophilic abnormalities, antiphospholipid antibody syndrome, inferior vena caval filter. Cancer is also a major predisposing factor but is mentioned separately because it is such a strong predisposing factor in its own right.

John W Eikelboom FRACP, FRCPA · Graeme J Hankey MD, FRACP

Antiplatelet drugs protect against myocardial infarction, stroke, cardiovascular death and other serious vascular events in patients with a history of previous vascular events or known risk factors for cardiovascular disease. Aspirin reduces the risk of serious vascular events in patients at high risk of such an event by about a quarter and is recommended as the first-line antiplatelet drug. Clopidogrel reduces the risk ...

Graeme J Hankey MD, FRACP · John W Eikelboom MB BS, FRACP

To the Editor: In a valuable review of low-molecular-weight heparins (LMWH), Eikelboom and Hankey1 stray off the beaten path into the unwelcoming area of obstetric therapeutics — a notoriously hostile environment replete with traps and hazards. Their statement that "low-molecular-weight heparins are being used increasingly in pregnant women with prosthetic heart valves and for the prevention and treatment of venous thromboembolism" is contentious and requires considerable qualification. The Journal has already published a position statement concerning the use of these heparins in pregnancy.2 It clearly stated that the initial treatment for pulmonary embolism in pregnancy remains intravenous unfractionated heparin, because so far there are no trials of LMWH in pulmonary embolism in pregnancy. The guidelines of the American College of Chest Physicians do endorse the use of LMWH for this indication,3 but base that view on data in non-pregnant patients. We believe that, as yet, there is insufficient evidence to recommend LMWH for the initial management of pulmonary embolism in pregnancy, although, on theoretical grounds, the treatment seems attractive. In anticoagulation therapy for artificial heart valves in pregnancy, there are serious problems. Unfortunately, the conscientious adviser must be very circumspect in counselling women with these prostheses. Pregnancy for these women presents significant risks. None of the heparins, unfractionated or low molecular weight, has been shown to protect reliably against embolism from, or thrombosis of, these valves in pregnancy. Whether LMWH is better than unfractionated heparin has not been established and awaits appropriate trials. Warfarin, which crosses the placenta, remains a valid, but worrying, choice in pregnancy for antico-agulation in patients with prosthetic heart valves. This drug provides optimal protection from valve thrombosis, but with the potential for teratogenicity in the first trimester and fetal (and maternal) haemorrhage later in pregnancy. Many experts use heparin and warfarin sequentially in this situation.3 Thus, anticoagulation therapy for pregnant women with serious medical problems remains, as always, perplexing, difficult and dangerous. While LMWH offer considerable promise and have undoubted utility in several areas, there are very compelling caveats about their current use for pulmonary embolism and prosthetic heart valves in pregnant women. For these reasons and others, women with prosthetic heart valves planning pregnancy, as well as those already pregnant, should be counselled about these problems by a physician experienced in managing medical problems in pregnancy.

Barry NJ Walters · Dorothy Graham