Portal de Periódicos da CAPES

Acquired Haemophilia: Review and Meta‐Analysis Focused on Therapy and Prognostic Factors

2003; Wiley; Volume: 121; Issue: 1 Linguagem: Inglês

10.1046/j.1365-2141.2003.04162.x

1365-2141

Julio Delgado, Víctor Jiménez‐Yuste, F. Hernández‐Navarro, Ana Villar,

Coagulation, Bradykinin, Polyphosphates, and Angioedema

Acquired haemophilia (AH) is a rare disease that occurs at a rate of approximately 1 person per million each year. Anti-factor VIII is the most commonly recognized autoantibody directed against a clotting factor, and is associated with bleeding complications that can be life threatening. These bleeding episodes, however, can be controlled when the correct diagnosis is made quickly and appropriate therapy is applied. Although the aetiology of this disorder remains obscure, about 40–50% of cases are associated with other conditions, mainly the post-partum period, underlying malignancies, drug administration or autoimmune diseases such as rheumatoid arthritis or systemic lupus erythematosus. This article provides a succinct review of the clinical features, laboratory diagnosis, prognostic factors and therapeutic management of patients with AH. The bleeding pattern of AH patients is quite different from that observed in congenital haemophilia. The most common complaints are bleeding into the skin or muscles, haematuria, haematemesis or melaena and prolonged post-partum or post-operative bleeding (Green & Lechner, 1981; Morrison et al, 1993). Iatrogenic bleeding is also common, after attempts to insert intravenous lines or catheterize the bladder. Often the first indication of the presence of an inhibitor is persistent bleeding after a surgical procedure that defies all attempts to control blood loss (Boggio & Green, 2001). In contrast to patients with congenital haemophilia, haemarthroses are relatively uncommon. The reason for this variant bleeding pattern, however, is unclear (Hay, 1998). Severe or life-threatening bleeding occurs in more than 80% of patients, and some 20% die, either directly or indirectly related to the presence of the autoantibody. Deaths are more frequent during the first few weeks after its appearance (Green & Lechner, 1981). Patients are rarely diagnosed in the presymptomatic state via the discovery of a prolonged activated partial thromboplastin time (APTT). The APTT assay is a reliable screening test for factor VIII (FVIII) inhibitor detection as it is typically prolonged when FVIII activity decreases to 45% of the mean normal level or less. Furthermore, mixing studies with patient plasma and normal plasma will not normalize the APTT, and FVIII activity will be reduced. Weak autoantibodies, however, may not prolong the APTT unless the mixture is incubated for at least 1 or 2 h at 37°C (Boggio & Green, 2001). Of note, when high-titre FVIII inhibitors are present, activity of factors XII, XI and IX may be artificially depressed but, if the assays are repeated in increasing dilutions of patient plasma, the true level of the above-mentioned factors will emerge while the FVIII activity will remain at the lower levels (Pruthi & Nichols, 1999). The inhibitor can be roughly quantified using the Bethesda assay (Kasper et al, 1975), which measures residual FVIII after incubation of patient plasma with normal plasma for 2 h at 37°C. The Nijmegen modification of the Bethesda assay uses a buffered normal plasma and FVIII-deficient plasma, instead of buffer, to dilute normal and patient plasmas, thereby stabilizing FVIII activity (Verbruggen et al, 1995). This modification results in fewer false-positive results, although it is more expensive. Both methods, however, tend to underestimate the inhibitor potency because of its non-linear complex reaction kinetics, which often results in the persistence of low levels of FVIII after prolonged incubation (Hay, 1998). Indeed, the recovery and half-life of exogenous FVIII may be markedly reduced even in patients with low inhibitor titres (Morrison & Ludlam, 1995). Finally, the Bethesda assay can be modified to measure the degree to which an inhibitor inactivates porcine FVIII, in order to determine whether this product could be a viable therapeutic alternative (Kasper, 1991). In most patients, FVIII autoantibodies are idiopathic. However, the disorder is associated with other conditions in about 40–50% of cases, which mainly occur in relation to the post-partum period, autoimmune diseases, underlying malignancies and drug administration. The association between pregnancy and AH has long been recognized. The bleeding disorder usually follows pregnancy, most commonly 1–4 months after delivery, although cases appearing more than 1 year post partum have been described (Michiels, 2000). In addition, the inhibitor appears during pregnancy in 2·5–14% of patients (Solymoss, 1998; Michiels, 2000). The potency of the inhibitor is rather low in the majority of cases, with a median titre of about 20 Bethesda units (BU) (Hauser et al, 1995; Solymoss, 1998). This may explain why the natural history of pregnancy-associated AH is characterized in the majority of cases by a spontaneous disappearance of the inhibitor after a mean period of 30 months (Michiels et al, 1997; Michiels, 2000). Mortality rates also tend to be lower in this subgroup of patients, ranging between 0% and 6% (Hauser et al, 1995; Solymoss, 1998). It is advisable to evaluate all these patients for lupus or rheumatoid arthritis because a certain subset of patients might have one of these disorders, which would require a change in the therapeutic approach (Shobeiri et al, 2000). An important clinical issue regarding pregnancy-related inhibitors is its possible recurrence in subsequent pregnancies, which makes counselling of women with this profile essential. In this regard, results are conflicting and, although most series have found that inhibitors tend not to recur in patients who attain a complete remission (Coller et al, 1981; Michiels, 2000; Yee et al, 2000), Solymoss (1998) reported on three patients who had six subsequent pregnancies; four of these pregnancies (66%) had an anamnestic response of the inhibitor. FVIII inhibitors may be associated with additional derangement of the immune system such as rheumatoid arthritis, systemic lupus erythematosus, Sjögren's syndrome, dermatomyositis and graft-versus-host disease after allogeneic bone marrow transplantation (Green & Lechner, 1981; Kessler, 2000). More rarely, AH is associated with organ-specific autoimmune diseases such as myasthenia gravis, multiple sclerosis, Graves' disease and autoimmune haemolytic anaemia (Hoyle & Ludlam, 1987; Sievert et al, 1996; Van den Brink et al, 2000; Meiklejohn & Watson, 2001). In contrast to post-partum inhibitors, rheumatoid arthritis patients usually have high-titre inhibitors that do not decline without therapy or with prednisone alone. The addition of an alkylating agent, however, results in the disappearance of the inhibitor in the majority of cases (Green & Lechner, 1981). Approximately 10% of patients with AH have an underlying malignancy, either solid or haematologic (Green & Lechner, 1981; Rizza et al, 2001). The association between AH and haematologic malignancies, especially lymphoproliferative disorders, is in line with the broad range of autoimmune phenomena that frequently complicate these conditions (Sallah et al, 2000). Nevertheless, the association between AH and solid tumours is not so clear. In fact, some authors consider that the appearance of FVIII antibodies in patients with solid tumours may well be a non-causal association as these neoplasms are common in elderly patients in whom AH usually appears (Hultin, 1991; Hay, 1998). To clarify this issue, Hauser & Lechner (1999) reviewed all 27 well-documented cases found in the literature in which FVIII antibodies were associated with solid tumours, and found that immunosuppressive treatment led to the disappearance of the antibody in the majority of cases, whereas removal of the tumour by surgery was successful in only one case. In keeping with this, they concluded that immunosuppression should be regarded as the front-line therapy in patients with cancer-associated FVIII inhibitors. Indeed, they found it unlikely that tumour treatment could effectively remove the antibody. Sallah et al (1998), nevertheless, reported five patients with AH and solid tumours who underwent tumour treatment with very different results. Three patients with early-stage neoplasms had treatment directed towards the inhibitor, which did not completely eliminate it, whereas surgical resection of the tumour or chemotherapy resulted in its complete eradication. Conversely, in the remaining two patients with metastatic carcinoma, the inhibitor persisted despite long-term immunotherapy. These results prompted the authors to perform a new review of the literature, and they found 41 cases of patients with cancer-associated FVIII inhibitors, including patients with solid and haematologic malignancies (Sallah & Wan, 2001a). In 22% of patients, the treatment of cancer led to the disappearance of the inhibitor, whereas the majority of patients who failed treatment had an advanced or metastatic malignancy. The authors thus concluded that: (1) it seems reasonable to treat the primary malignancy in patients with cancer-associated FVIII inhibitors because it is easier to eradicate the antibody when the tumour is controlled; and (2) the presence of an underlying malignancy in a patient with a FVIII inhibitor should not be considered as a constraint against the use of immunotherapy to suppress the production of the antibody, even in those cases that fail to respond to treatment of the primary tumour. The decision to administer immunosuppressive agents in these patients should, nevertheless, take other factors, such as age, type of malignancy and severity of bleeding, into consideration. Despite the fact that it is difficult to relate any illness specifically to a drug in elderly patients, many of whom are on a range of medications, certain agents including antibiotics (penicillins, sulphonamides, chloramphenicol) and anticonvulsants (diphenylhydantoin) have well-established associations with AH (Green & Lechner, 1981; Morrison et al, 1993; Bossi et al, 1998). Frequently, drug-induced anti-FVIII arises after hypersensitivity reactions and remits shortly after withdrawing the offending drug. However, the pathophysiology of this phenomenon remains unknown. On the other hand, the strong immune properties of both interferon (IFN) alpha and fludarabine may explain the appearance of autoantibodies against FVIII and other immune phenomena reported with their use (Tiplady et al, 2000; Sallah & Wan, 2001b). The two main goals of AH treatment are to arrest bleeding and to eradicate the inhibitor. The first objective is necessary because bleeding episodes are usually relentless and can be life threatening. The second objective is required to restore normal haemostasis and may be accomplished using some type of immunotherapy. The management of acute bleeding episodes in patients with AH is complicated by the lack of prospective studies evaluating the different therapeutic modalities and the variable response to these approaches. One important principle to consider is the prevention of risk situations for bleeding. Severe bleeding can occur from trivial injuries, intramuscular injections, intra-arterial blood sampling and any invasive procedure, all of which should be avoided. Drugs that interfere with platelet function, such as aspirin or non-steroidal anti-inflammatory agents, should also be avoided. The choice of therapeutic agents depends on the severity of bleeding and the inhibitor titre. Ideally, the most effective treatment is one that will increase the plasma FVIII level sufficiently to control bleeding. Several strategies, such as desmopressin, human FVIII (hFVIII) or porcine FVIII (pFVIII), may raise the FVIII level. If the inhibitor titre is high (> 5 BU), or bleeding continues despite infusions of hFVIII or pFVIII concentrates, FVIII bypassing agents, such as activated prothrombin complex concentrates (APCC) or recombinant factor VIIa (rFVIIa), may be used. During the last two decades, desmopressin (1-deamino-8-δ-vasopressin, DDAVP) has been demonstrated to be effective in both mild haemophilia A and von Willebrand's disease (VWD) (Mannucci et al, 1977). The intravenous infusion of 0·3 µg/kg to a normal person produces a two- to threefold increase in FVIII and von Willebrand factor (VWF) plasma levels. This release of endogenous stored FVIII protein could overcome or neutralize the inhibitor in patients with low-titre inhibitors, usually < 3 BU (De la Fuente et al, 1985; Vicente et al, 1985; Chistolini et al, 1987; Mudad & Kane, 1993). Another possible action mechanism for this agent is a desmopressin-induced increase in VWF, which may block the FVIII autoantibody by inhibiting the binding of FVIII to VWF (Boggio & Green, 2001). Desmopressin, however, is only effective in a few patients with low inhibitor titres (< 3 BU), and a trial of desmopressin in patients with high inhibitor titres or with severe bleeding only delays the use of more effective approaches. Large doses of plasma-derived or recombinant hFVIII may be useful in patients with inhibitor titres under 5 BU (Morrison & Ludlam, 1995; Cohen & Kessler, 1996). The dose of hFVIII needed to neutralize a circulating inhibitor and provide FVIII haemostatic levels [0·3–0·5 international units (IU)/ml] can only be predicted roughly from the inhibitor titre because of the variable kinetics of these autoantibodies. The recommended dose is 20 IU/kg for each BU of inhibitor plus 40 additional IU. The plasma FVIII level should be determined 10–15 min after the initial bolus and, if it is not adequate, another bolus should be administered (Kasper, 1995). Another approach is to administer an initial bolus of 200–300 IU/kg followed by continuous infusion of about 4–14 IU/kg/h (Blatt et al, 1977). Some authors double or triple the dose of hFVIII that should be given to a haemophiliac of the same weight without inhibitor and often find beneficial clinical responses despite poor post-infusion assay results (Ekert et al, 1979; Rizza & Matthews, 1982). As expected, the half-life of hFVIII administered to patients with AH cannot be predicted. Autoantibodies are extremely difficult to saturate by the addition of antigen (FVIII); therefore, hFVIII therapy must be closely monitored and the dosage adjusted or discontinued depending on plasma FVIII levels. Porcine FVIII concentrates (Hyate:C®) have been used extensively for the treatment of bleeding episodes in patients with both auto- and alloantibodies (Morrison et al, 1993; Hay et al, 1996). The degree of homology between hFVIII and pFVIII means that pFVIII can function haemostatically in humans, whereas the critical epitopes recognized by human autoantibodies are often sufficiently different to achieve high levels of circulating FVIII. These facts provide the rationale for the therapeutic use of pFVIII concentrates (Hay, 1998, 2002; Pruthi & Nichols, 1999). An anamnestic rise in the antihFVIII inhibitor titre has been reported to follow for 25–35% of patients with congenital haemophilia treated with pFVIII (Gatti & Mannucci, 1984; Kernoff, 1984; Hay et al, 1996). Nevertheless, in a multicentre survey, only 15% of patients with AH increased their antihFVIII inhibitor titre after treatment with pFVIII (Morrison et al, 1993). The pharmacokinetic profile of pFVIII is unpredictable in patients with AH. Close laboratory monitoring is therefore recommended to adjust the dose according to plasma FVIII recovery and clinical response. The initial dose of pFVIII can be calculated using this formula: plasma volume (ml) × antiporcine antibody titre (U/ml) (neutralizing dose) + desired FVIII coagulant activity (FVIII:C) increment × body weight (kg)/1·5 (augmenting dose) (Gatti & Mannucci, 1984). If the antipFVIII inhibitor titre is unknown, 50–100 IU/kg pFVIII concentrate should be given to patients whose antihFVIII antibody titre is < 50 BU. In patients with inhibitors titres between 50 and 100 BU, pFVIII can be administered at a dose of 100–200 IU/kg (Kernoff, 1984). In the largest reported series of patients with AH treated with pFVIII, a good or excellent response was observed in 78% of patients, whereas the response was partial and poor in 11% and 9% of patients respectively (Morrison et al, 1993). The average dose was 90 IU/kg repeated every 12 h. It is generally believed that pFVIII recovery in vivo, monitored by ex vivo assays of plasma FVIII activity, provides a guide to response and cost-effective dosing, usually in the range of 1 U/ml. Over time, pFVIII levels may decline as antibodies develop against the porcine protein, rendering the patient resistant to this agent. Interestingly, the kinetics of pFVIII alloantibodies are similar to those appearing in haemophiliacs treated with hFVIII (type 1 pattern) (Green et al, 1999). The main adverse events associated with pFVIII administration are allergic reactions and thrombocytopenia. The incidence of transfusion reactions is low (about 3–7% of infusions) and mainly dose related (Hay et al, 1996). The risk of reactions may be reduced if the product is given by continuous infusion (O'Gorman et al, 2001; DiMichele et al, 2002; Hay, 2002). The fall in platelet count, which has been attributed to porcine VWF contaminating the concentrate (Altieri et al, 1986), is usually mild and of no clinical significance unless large doses of pFVIII are administered several times a day (Gringeri et al, 1991; Hay et al, 1996; Hay, 2002). The incidence of thrombocytopenia depends on the monitoring frequency and dose of pFVIII. This complication may also be reduced by the administration of pFVIII via continuous infusion, which suggests that this might be the preferred way of administering this product (O'Gorman et al, 2001; DiMichele et al, 2002; Hay, 2002). Activated prothrombin complex concentrates (APCC) have been used extensively in the treatment of bleeding episodes in patients with FVIII inhibitors (Kurczynski & Penner, 1974). The products in most widespread use are FEIBA® and Autoplex®. The composition of Autoplex® has been examined recently (Lundblad et al, 1998), and the major components are activated factor IX (FIXa) and FVIIa with small amounts of FXa, FXIa and thrombin. This suggests that, after its infusion, FXa and thrombin are rapidly inactivated by antithrombin, whereas FXIa and FIXa generate additional FVIIa, which may be the main active component (Turecek et al, 1999). The composition of FEIBA® has also been examined, and similar FVIIa, but lower FIXa, concentrations were found compared with Autoplex® (Roberts, 1999). Although there are some reports about the use of APCC in AH (Söhngen et al, 1997; Ji et al, 1998), most of the experience has been gained in patients with congenital haemophilia and inhibitors (Lusher et al, 1980; Hilgartner & Knatterud, 1983; Negrier et al, 1997; Penner, 1999). In a French study, FEIBA® was administered to 60 patients with FVIII inhibitors, six of whom had autoantibodies (Negrier et al, 1997). The product was considered effective in 81% of bleeding episodes. Doses averaged 70 IU/kg every 8–12 h and generally did not exceed 200 IU/kg/d (Negrier et al, 1997). Furthermore, excellent results were obtained in 13 of 17 patients receiving Autoplex® at doses > 50 IU/kg, although smaller doses were generally ineffective (Penner, 1999). The recommended dose for both products is 50–200 IU/kg/d in divided doses. Larger doses (> 200 IU/kg/d) have been associated with adverse thrombotic events (Gruppo et al, 1983; Lusher, 1994; Stenbjerg & Jorgensen, 1977). Indeed, in a survey of adverse thrombotic events with FEIBA®, three patients with AH developed disseminated intravascular coagulation (DIC), myocardial infarction and thrombosis respectively (Ehrlich et al, 2002). Efficacy has to be judged clinically and the dose adjusted to the clinical response because there are no laboratory assays that correlate with response (Hay, 1998). Recombinant FVIIa (rFVIIa; NovoSeven®) has demonstrated clinical effectiveness in patients with congenital haemophilia and inhibitors (Hedner et al, 1988; Lusher, 1996; Key et al, 1998; Hedner, 1999; Scharrer, 1999; Negrier & Lienhart, 2000). The physiological haemostatic mechanism is initiated by the formation of a complex between tissue factor (TF) and FVIIa, which, after several reactions, results in the generation of thrombin (Mann, 1999; Hoffman & Monroe, 2001). Moreover, rFVIIa might work through TF-independent activation of FX on the surface of activated platelets. In fact, full thrombin generation could be induced on the activated platelet surface by the addition of supraphysiological doses of rFVIIa, even in the absence of FVIII or IX (Monroe et al, 1997, 1998; Hoffman et al, 1998; Ewenstein, 2001). In an alternative action mechanism dependent on TF, supraphysiological concentrations of rFVIIa may overcome the inhibition produced by zymogen FVII over thrombin generation in the absence of FVIII (van't Veer et al, 2000). The optimal dose of rFVIIa has not been established. Although patients reported by Hay et al (1997) received a wide range of doses (median 90 µg/kg every second hour, range 45–181 µg/kg), there was no difference in the intensity of treatment between responders and non-responders. A regimen of boluses of 90–120 µg/kg, given every other hour is the current recommended dose. The short half-life of this product (mean half-life in adults 2·7 h) necessitates such frequent infusions (Lindley et al, 1994). Minor episodes are usually treated with two or three doses, but the treatment of major bleeds may continue for several days. Continuous infusion is a new method of administration (Schulman et al, 1998; Santagostino et al, 2001; Smith et al, 2001), with a substantial individual variation in the clearance of the product and the haemostatic response (Ewenstein, 2001). The most extensive experience with the use of rFVIIa in patients with AH comes from Hay et al (1997). The authors treated 74 bleeding episodes with rFVIIa. The average dose of rFVIIa was 90 µg/kg every 2–6 h for a median of 3·9 d, and the median number of doses was 28. When rFVIIa was used as salvage therapy, 75% and 17% of patients had good and partial responses respectively; when the agent was given as first-line therapy, the response rate was 100%. Those patients who did not respond within 24 h were unlikely to respond if rFVIIa treatment was subsequently continued (Hay et al, 1997). rFVIIa has good tolerability and few side-effects. However, there have been some reports of myocardial infarction, clinical thrombosis and DIC in patients receiving rFVIIa (Peerlinck & Vermylen, 1999). Recently, the experience of patients receiving more than 180 000 standard doses of rFVIIa has been published (Roberts, 2001). Thrombotic events occurred rarely, and most could be attributed to improvements in the clotting mechanism rather than a direct effect of the rFVIIa itself (Roberts, 2001). Nevertheless, the close temporal association of thrombosis with administration of the drug makes close patient monitoring necessary when using this agent. Other disadvantages of rFVIIa are its high financial cost and the lack of monitoring possibilities. The determination of post-infusion prothrombin time and FVII:C levels has been suggested for monitoring the effect of rFVIIa, although adequate haemostatic levels have not yet been defined (Hedner, 1996; Johannessen et al, 2000). Immunomodulatory therapy of FVIII inhibitors is aimed at the eradication of the autoantibody or the suppression of the cell clone responsible for its synthesis. Despite the fact that up to 36% of patients who do not receive immunosuppressants experience a spontaneous resolution of their autoantibodies (Lottenberg et al, 1987), this occurrence is unpredictable, and the patient remains at great risk as long as the antibody persists (Green et al, 1993). As such, attempts to accelerate the disappearance of these inhibitors are warranted in the best interests of the patient. Many drugs have been used to inhibit antibody production even though the, often anecdotal, nature of the reports and the near absence of controlled clinical trials makes interpretation of data difficult. Furthermore, the choice of immunosuppressive agents should take into consideration the age of the patient, the presenting bleeding complications and the titre of the antibody (Pruthi & Nichols, 1999). Prednisone, given at doses of 1 mg/kg/d, abolishes the inhibitor in ≈ 30% of patients (Green & Lechner, 1981; Spero et al, 1981; Green et al, 1993), but the addition of cyclophosphamide (1–2 mg/kg/d) may increase this response rate to 60–100% (Green et al, 1993; Shaffer & Phillips, 1997; Bayer et al, 1999). Based on the earlier studies, the combination of prednisone and cyclophosphamide has been the mainstay of treatment for many years. However, other combinations, such as prednisone with azathioprine (Söhngen et al, 1997) or prednisone with cyclophosphamide and vincristine (Lian et al, 1989, 2002), have also proved effective. In the event of failure, cyclosporin A (200–300 mg/d) has been used successfully, either alone or in combination with prednisone, as a salvage therapy (Schulman et al, 1996; Brox et al, 1998; Petrovic et al, 2000; Saxena et al, 2000). The successful use of intravenous immunoglobulin (IVIg) in patients with AH was first reported by Sultan et al (1984). Its effect is attributed to the presence of anti-idiotypic antibodies in the IVIg, although in some patients, the inhibitor level may diminish permanently, suggesting that there is also suppression of antibody synthesis. In a prospective multicentre study on the efficacy of high-dose IVIg in patients with AH, the overall response rate was 50% (Schwartz et al, 1995), although some of these patients also received other immunosuppressants. In contrast, a subsequent literature review of patients treated by IVIg with no other concomitant immunosuppressive therapy was more disappointing in that the complete response (CR) rate was only 12% (Crenier et al, 1996). Therefore, the actual benefit of IVIg is questionable, and current clinical results are insufficient to make IVIg a first choice for the suppression of FVIII autoantibodies. Extracorporeal removal of the antibody may be indicated when a rapid reduction is needed, especially in cases of severe bleeding. Several methods have been used, such as plasmapheresis (Slocombe et al, 1981; Narukawa et al, 1999; Sunagawa et al, 1999) and immunoadsorption to staphylococcal protein A (Guillet et al, 2001) or polyclonal sheep antibodies against human immunoglobulins (Jansen et al, 2001). All these procedures have proved effective in this clinical setting but, with few exceptions (Toepfer et al, 1998; Ogata et al, 1999), they are always combined with immunosuppressive drugs, and their specific contribution is very difficult to assess. In fact, as these procedures do not appear to shorten the time to CR, they should be used in association with immunosuppressive agents in order to obtain long-term remissions and prevent inhibitor recurrence. Immune tolerance induction (ITI) regimens with hFVIII concentrates are used primarily in young patients with congenital haemophilia and FVIII alloantibodies (Colowick et al, 2000). In contrast, ITI protocols are rarely, if ever, implemented in adults with AH. A few authors have administered a combination of hFVIII concentrate and immunosuppressive agents in order to eradicate the inhibitor (Green, 1971; Lian et al, 1989; Nemes & Pitlik, 2000). The rationale behind this approach is that intravenous hFVIII infusion will theoretically stimulate the abnormal lymphoid clone responsible for anti-FVIII antibody synthesis and thus facilitate more specific and effective elimination of these cells by cytotoxic agents. Lian et al (1989) treated 12 patients with cyclophosphamide, vincristine and prednisone (CVP) after priming with hFVIII concentrate, and a CR was obtained in 11 out of 12 patients. Nevertheless, the same authors have recently reported very similar results in a series of six patients treated with CVP without FVIII infusion, concluding that CVP without FVIII might be as good as CVP with FVIII (Lian et al, 2002). On the other hand, Nemes & Pitlik (2000) administered hFVIII, cyclophosphamide and methyl-prednisolone to 14 patients with AH and compared their results with six historical controls treated with steroids ± cyclophosphamide. Although statistical analyses were not performed, a shorter time to CR and a lower bleeding-related mortality were observed in the ITI group. Finally, hFVIII was administered orally to three patients with long-lasting FVIII autoantibodies who had failed to respond to different immunosuppressive agents (Lindgren et al, 2000). In one patient, the FVIII inhibitor disappeared, and the FVIII level increased from < 0·01 to 0·17 IU/ml 6 months after cessation of FVIII administration. However, these results are preliminary, and further studies will be needed to establish the true efficacy of these ITI protocols. Many authors have stated that AH is associated with a high mortality rate, which ranges between 7·9% and 22% (Green & Lechner, 1981; Hay et al, 1997). However, few investigators have emphasized that most of these patients are very old and have other comorbidities, which may or may not be related to the FVIII inhibitor (Godreuil et al, 2001). Moreover, the majority of these patients receive immunosuppressive agents, which are associated with multiple side-effects in this age group (Rochon & Gurwitz, 1995). Because of this, we feel that AH should be considered not only as a potentially lethal condition but also as a disorder with an increased morbidity, which is mainly therapy related. Unfortunately, few studies have systematically addressed therapy-related side-effects. In their classic paper, Lottenberg et al (1987) treated three patients with cyclophosphamide, and two of them developed bacterial sepsis from neutropenia. Two years later, Lian et al (1989) treated 12 patie