Guidelines on the diagnosis and management of the thrombotic microangiopathic haemolytic anaemias.

Abstract

Thrombotic thrombocytopenic purpura (TTP) was first described by Moschowitz (1924). The classic pentad of diagnostic features has been recognized for many years. However, several other syndromes are also characterized by similar features. These include haemolytic uraemic syndrome (HUS), eclampsia and the HELLP syndrome (haemolysis, elevated liver enzymes, low platelets). The concept has arisen that they might represent an overlapping spectrum of disease, although with varying pathophysiological features (see Table I). The recent characterization of a novel von Willebrand factor (VWF)-cleaving metalloprotease activity (Furlan et al, 1996; Tsai, 1996) and its deficiency or inhibition in some forms of microangiopathic haemolysis (Furlan et al, 1997, 1998; Tsai & Lian, 1998) has led to speculation that a pathogenic mechanism for individual patients can be defined more readily and appropriate treatment introduced more rapidly. However, there is still considerable confusion, a lack of properly conducted randomized clinical trials and poor co-ordination of clinical data. This is, in part, because these patients present to a range of specialists including haematologists, obstetricians, nephrologists and infectious disease physicians. These guidelines attempt to define the various clinical subtypes, specify the recognized diagnostic features and look critically at management options. It is acknowledged that there is a lack of evidence from well-conducted studies on which to support some of the recommendations made. Thrombotic thrombocytopenic purpura (TTP) is rare. The reported incidence is 3·7 per million (Torok et al, 1995). However, its prompt recognition and treatment is vital, as delays in initiating treatment have been shown to adversely affect outcome (Pereira et al, 1995). TTP is a clinical diagnosis. It is characterized by the classic pentad of thrombocytopenia, microangiopathic haemolytic anaemia, fluctuating neurological signs, renal impairment and fever, often with insidious onset. Neurological impairment has multiple manifestations including headache, bizarre behaviour, transient sensorimotor deficits (TIAs), seizure and coma. Presence of coma at presentation is a poor prognostic indicator (Pereira et al, 1995; Sarode et al, 1997). Additional complications may be seen: gastrointestinal ischaemia (manifest as abdominal pain) and serous retinal detachment are recognized associations. However, up to 35% of TTP patients do not have neurological symptoms or signs at presentation (Rock et al, 1991). As the triad of acute renal insufficiency, MAHA and thrombocytopenia defines HUS, diagnostic uncertainty may arise. Moreover, fever and renal impairment are present in only a minority of patients (Rock et al, 1991, 1998). In practice, therefore, a diagnosis of TTP may be made in the presence of a microangiopathic haemolytic anaemia and thrombocytopenia in the absence of any other identifiable cause. A number of different clinical variants of TTP have been documented. Clinical subtype may influence management and those recognized are listed in Table II. The predominant histological abnormality found in TTP is the formation of platelet microvascular thrombi. The renal and cerebral circulations are primarily affected, thus accounting for the clinical features of the disease. Excessive platelet aggregation occurs when platelet-rich plasma (PRP) from patients with congenital TTP is exposed to shear stress (Moake et al, 1994). This is mediated by ultra-large VWF multimers (ULVWF) (Moake et al, 1994; Karpman et al, 1997). ULVWF are not a normal constituent of circulating plasma. Instead, VWF circulates as smaller multimeric forms resulting from proteolytic degradation of ULVWF. VWF fragments with mobility corresponding to 189, 176 and 140 kDa are consistently detected in normal plasma in addition to the predominant 225 kDa subunit (Zimmerman et al, 1986; Tsai et al, 1991). These originate as a consequence of cleavage of a single peptide bond between residues Tyr-842 and Met-843 of the mature subunit (Dent et al, 1991). Identical fragments may be generated in vivo by a novel metalloproteinase activity (Furlan et al, 1996; Tsai, 1996). The protease has recently been characterized as a new member of the ADAMTS (a disintegrin and metalloproteinase with thrombospondin type-I motif) family, ADAMTS13 (Fujikawa et al, 2001; Gerritsen et al, 2001; Levy et al, 2001). Deficiency of this VWF-cleaving protease (VWF-CP) activity has been associated with acquired and congenital TTP. While all cases of idiopathic TTP have, to date, been associated with severe protease deficiency, secondary TTP may occur in the context of normal protease activity (Veyradier et al, 2001). In a series of 111 patients with thrombotic microangiopathies of whom 66 manifested with TTP (25 idiopathic and 41 secondary) and 45 with HUS, protease deficiency had a sensitivity of 89% and specificity of 91% for TTP. Initial reports suggested that idiopathic TTP is secondary to an inhibitory auto-antibody of IgG subtype (Furlan et al, 1998; Tsai & Lian, 1998), but in the above series, a protease inhibitor was identified in only 14 patients (56%). While congenital TTP appears secondary to a constitutional deficiency (Furlan et al, 1997, 1998), presentation may be delayed until adulthood (Lämmle et al, 2001). Cirrhosis (Mannucci et al, 2001), uraemia (Mannucci et al, 2001), acute inflammation (Mannucci et al, 2001), disseminated intravascular coagulation (DIC) (Loof et al, 2001) and malignancy (Oleksowicz et al, 1999) have now also been associated with reduced VWF-CP activity. Thus, although sensitive, reduced VWF-CP activity is not specific for TTP. Moreover, this model fails to explain the anatomical distribution of thrombi. The endothelium is a heterogeneous organ and is subject to regulation by multiple factors, including cytokines (Drake et al, 1993), microenvironment (Aird et al, 1997) and shear stress (White & Fujiwara, 1986). Alterations in any one of these parameters could influence either VWF-CP activity per se or the susceptibility of VWF to proteolysis. TTP is often characterized by severe thrombocytopenia, which may be useful in its differentiation from HUS. However, in one series, although the mean platelet count was lower in TTP than HUS (18 × 109/l vs 36 × 109/l), there was a wide range and considerable overlap (Vesely et al, 2000). Severe thrombocytopenia at diagnosis (platelet count < 20 × 109/l) has been suggested to be a poor prognostic indicator, conferring increased mortality (Rock et al, 1998), although this is not a uniform observation (Sarode et al, 1997). Thrombocytopenia is typically accompanied by overt microangiopathic haemolysis. Thus, examination of the blood film usually shows striking red cell fragmentation and polychromasia. However, schistocytes may be absent from the peripheral blood film in the first 24–48 h following clinical presentation. Routine coagulation profiles are usually normal (Monteagudo et al, 1991; Sagripanti et al, 1996; Rock et al, 1998), although slight increases in D-dimer, fibrin degradation products and thrombin–anti-thrombin complex (TAT) may be seen (Monteagudo et al, 1991; Sagripanti et al, 1996; Wada et al, 1998). Secondary DIC may, however, arise from prolonged tissue ischaemia and is an ominous prognostic indicator. Evidence of endothelial perturbation is demonstrated by increased plasma levels of plasminogen activator inhibitor (PAI-1) (Anthony et al, 1998) and thrombomodulin. The latter has also been identified as a poor prognostic factor (Wada et al, 1998). Plasma VWF levels are often elevated acutely (Rock et al, 1998). Abnormalities of VWF multimers are also common and were identified in 86% of patients either at the onset of or during an acute episode of TTP (Moake & McPherson, 1989). These ranged from the presence of ULVWF multimers in 31% to a relative decrease in the largest plasma VWF forms in 36%. Acute changes in VWF multimeric distribution do not appear to correlate with clinical outcome. However, the finding of ULVWF multimers during periods of remission has been associated with intermittent TTP (Moake & McPherson, 1989). Renal function is normal in the majority of patients: only 18% patients presenting to the Canadian Apheresis Group had evidence of renal impairment (Shumak et al, 1995). Liver function tests often show not only hyperbilirubinaemia but also a raised transaminase level. This is thought to represent hepatic ischaemia. Evans' syndrome may be excluded by a negative direct anti-globulin test. Recommended diagnostic laboratory investigations to be performed at presentation are given in Table III. In those TTP patients in whom significant renal impairment is a feature, it may be impossible to confidently exclude HUS. Where there is doubt, a presumptive diagnosis of TTP should be made and plasmapheresis initiated. Occasionally renal biopsy performed after recovery of the platelet count may allow accurate retrospective diagnosis. In TTP, arteriolar and capillary thrombosis is prominent. Thrombi are largely composed of platelets and stain strongly for VWF. Only weak staining for fibrin and fibrinogen is seen in contrast to those thrombi formed in DIC (Asada et al, 1985). Aneurysmal dilatation of arterioles may lead to the formation of glomeruloid structures with relative sparing of the glomeruli (Katoh & Shigematsu, 1999). These features should be contrasted with those seen in HUS when the primary histological changes are glomerular and arteriolar fibrin thrombi and subendothelial widening of the glomerular capillary wall on electron microscopy (Remuzzi & Ruggenenti, 1995). If a diagnosis of TTP is made, consideration must be given to the presence of precipitants. These are drugs, autoimmune disease, malignancy and infection, particularly Escherichia coli 0157:H7 and human immunodeficiency virus (HIV). Although E. coli 0157:H7 is more closely linked with HUS, there have been cases with typical TTP features (Morrison et al, 1986; Kovacs et al, 1990). Of note VWF-CP deficiency was detected in one of 29 children with epidemic HUS, 25 of whom were positive for verotoxin (Hunt et al, 2001). In some series, up to 14% of TTP episodes have been associated with HIV infection (Ucar et al, 1994), although mortality data from the United States for 1988–1991 gives a figure of only 4·4% (Torok et al, 1995). Risk appears greatest at CD4 counts of less than 250 × 109/l (de Man et al, 1997). Serological testing for HIV should, therefore, be performed at diagnosis in all patients. As treatment (see below) results in multiple donor exposure, hepatitis B and C serology is also recommended in all patients at presentation. Recommendation. While there is no available diagnostic test for TTP, TTP may be diagnosed and treatment initiated if a patient presents with a microangiopathic haemolytic anaemia and thrombocytopenia in the absence of any other identifiable clinical cause. Routine investigations at presentation should include the following: full blood count, film, clotting screen, lactate dehydrogenase (LDH), direct anti-globulin test, urea and electrolytes, liver function tests, and urine dipstick for protein. An underlying precipitant should be considered. It is recommended that HIV and hepatitis serology tests are performed at diagnosis (Grade C, level IV). The mainstay of treatment of acute TTP is daily plasma exchange. Prior to its institution, mortality rates were in excess of 90% and have now fallen to 10–30%. Plasma exchange is superior to plasma infusion. A prospective randomized study performed by the Canadian Apheresis Group assigned a total of 102 patients to receive either plasma exchange or infusion with fresh-frozen plasma (FFP) on 7 of the first 9 d after entry to the trial. Plasma exchange resulted in significantly superior response rates at both the end of the first treatment cycle and at 6 months (response rates 47% and 78%vs 25% and 49% respectively). Mortality was also reduced at 22%vs 37% (Rock et al, 1991). Plasma exchange should be instituted within 24 h of presentation as delay in treatment initiation may increase treatment failure (Pereira et al, 1995). Moreover, it would seem appropriate to commence plasma exchange as soon as practicable if renal impairment, cardiac failure or coma is present. Reduced level of consciousness has been identified as a poor prognostic factor with an overall survival of 54% (Sarode et al, 1997). The duration of plasma exchange therapy required to achieve remission is highly variable. The average number of procedures required for remission in the above study was 15·8 (range 3–36). As the premature omission of a single plasma exchange may be associated with exacerbation, patients should be treated in centres able to provide a daily plasmapheresis service. Although undoubtedly efficacious, the optimal plasma exchange regimen has not been determined. In the Canadian apheresis trial, 1·5 × plasma volume exchange was performed on the first 3 d followed by 1·0 plasma volume exchange thereafter. Whether this intensity is superior to single plasma volume replacement from presentation is unclear. Currently, many centres initiate single-volume plasma exchange at presentation, reserving more intensive exchange for resistant cases. Similarly, the optimal duration of plasma exchange is unknown. It is empirically recommended that daily exchanges should continue for a minimum of 2 d after complete remission is obtained, defined as normal neurological status, platelet count and LDH with a rising haemoglobin. This is in agreement with the American Association of Blood Banks (AABB), which recommends daily plasma exchange until the platelet count is above 150 × 109/l for 2 to 3 d (AABB Extracorporeal Therapy Committee, 1992). Although it is accepted practice to taper the frequency of exchange procedures rather than stopping abruptly in an effort to minimize the risk of early relapse; this is not based on randomized clinical trials. The optimal replacement fluid administered also remains contentious. Possibly of note, cryosupernatant lacks the largest VWF multimers that are present in FFP and cryoprecipitate. Cryosupernatant is at least as efficacious as FFP. When used in previously untreated patients, the response rate was 75% after seven exchanges. Survival was 95% at 1 month, which compared favourably with historical control subjects treated with FFP (Rock et al, 1996). In contrast, a prospective randomized trial performed by the North American TTP Group failed to identify a significant difference in outcome between plasmapheresis with FFP or cryosupernatant from diagnosis. It should be noted, however, that only 27 patients were included in this trial (Zeigler et al, 2001). Larger trials are required to address this issue and a randomized controlled trial comparing plasma exchange with FFP versus cryosupernatant is currently being performed in Canada. Side-effects secondary to plasmapheresis are common. In one study, 9·7% of procedures were complicated by adverse reactions with anaphylactoid reactions occurring in 0·25% (Mokrzycki & Kaplan, 1994). Solvent/detergent-treated (S/D) plasma not only reduces viral risk but may be beneficial in reducing allergic reactions. This is because the process of plasma pooling results in extreme dilution of those antibodies responsible for immune-mediated reactions. S/D plasma has a similar favourable multimer profile to cryosupernatant and has been used as replacement fluid from presentation (Evans et al, 1999). However, numbers treated are small and there is no published comparative data with FFP or cryosupernatant. Further clinical experience is required to ascertain the role of S/D plasma in the primary treatment of TTP. Although plasma exchange remains the treatment of choice, plasma infusion (30 ml/kg/d) may still be indicated if there is to be an unavoidable delay in plasma exchange. It must be undertaken with care, however, as cardiac function may be compromised and cardiac failure can be precipitated. Recommendation. Single-volume daily plasma exchange should be commenced at presentation (Grade A, level Ib) and ideally within 24 h of presentation (Grade C, level IV). Plasma exchange using cryosupernatant may be more efficacious than that using FFP (Grade B, level III). Daily plasma exchange should continue for a minimum of 2 d after complete remission is obtained (Grade C, level IV). Corticosteroids. Steroids have been widely used in the treatment of TTP although there is scanty evidence documenting their efficacy. However, patients with TTP lacking central nervous system abnormalities other than headache have been shown to respond to steroid treatment alone, although of 54 such patients 44% required plasma exchange because of deterioration or failure to improve (Bell et al, 1991). Steroids have also been combined with plasma exchange in initial treatment of TTP. The addition of intravenous methylprednisolone 2 mg/kg/d to daily plasma exchange resulted in a complete remission rate of 76% (Perotti et al, 1996). As yet, no trial has addressed whether such a combination approach is superior to plasma exchange alone. Not surprisingly, there is no consensus regarding dose or mode of administration. Despite the lack of evidence, the addition of steroids to plasma exchange as standard therapy is attractive. Recent findings suggest that a functional deficiency of a novel VWF-CP activity secondary to a circulating inhibitory antibody of IgG subtype is of prime importance in the pathogenesis of TTP (Furlan et al, 1998; Tsai & Lian, 1998). It would, therefore, appear reasonable to institute steroids in all patients. Recommendation. All patients should receive adjuvant corticosteroid therapy (Grade B, level III). To achieve potent immunosuppression while minimizing long-term steroid side-effects, pulse methylprednisolone 1 g i.v. for 3 d is recommended (Grade C, level IV). Anti-platelet agents. The use of anti-platelet agents in TTP remains controversial. Ticlopidine and its analogue clopidogrel inhibit ADP–platelet interactions and interfere with shear-induced aggregation and might, therefore, be predicted to be of use in TTP. Certainly ticlopidine appeared to reduce the risk of subsequent relapse from 21·4% to 6·25% when used as maintenance therapy for 12 months after remission of TTP was achieved (Bobbio-Pallavicini et al, 1997). However, at publication, follow-up in this trial was incomplete. Subsequently ticlopidine (Bennet et al, 1998) and clopidogrel (Bennet et al, 2000) have been associated with TTP. The estimated incidence of ticlopidine-associated TTP is 1 per 1600–5000 patients treated while that following clopidogrel is 1·5–5·8 per million treated. As the latter is similar to the reported incidence of idiopathic TTP an association remains controversial. Nevertheless in both cases, antibody inhibitors to VWF-CP have been identified (Bennet et al, 2000; Tsai et al, 2000). In keeping with this, plasmapheresis is effective treatment with reported survival rates of 76–91% and is superior to plasma infusion (Bennet et al, 1998, 2000). For these reasons, ticlopidine and clopidogrel should be avoided in patients with a previous history of TTP. Aspirin and dipyridamole have both been used in the initial treatment of TTP. A 78% response rate at 6 months was achieved when aspirin and dipyridamole were administered in conjunction with plasma exchange (Rock et al, 1991). In a prospective randomized trial designed to address the effect of the addition of aspirin and dipyridamole to standard treatment (plasma exchange and steroids), a similar overall response was obtained in both groups. There was, however, a trend to reduced mortality at 15 d in those treated with anti-platelet agents. No excess haemorrhage was seen in the treatment group (Bobbio-Pallavicini et al, 1997), in contrast to the findings of a small retrospective study in which serious bleeding complications occurred in five of 14 patients receiving anti-platelet medication (Rosove et al, 1982). It should be noted, however, that large doses were administered in these last patients (0·9–2·7 g/d aspirin). Although not yet of proven benefit, there is nevertheless a rationale for instituting aspirin therapy when the platelet count rises over 50 × 109/l as anecdotally a rapid rise in platelet count during recovery has been associated with thromboembolic events. Recommendation. Low-dose aspirin (75 mg o.d.) should be commenced on platelet recovery (platelet count > 50 × 109/l) (Grade C, level IV). Supportive therapy. Red cell transfusion is an essential component of treatment. However, there is no single reliable parameter to guide the need for red cell transfusion. Identical survival rates resulted when a transfusion was given based on a haemoglobin threshold of 7 g/dl rather than 10 g/dl in euvolaemic critically ill patients (Hebert et al, 1999). As TTP may be complicated by rapid haemolysis and cardiac insufficiency secondary to myocardial microvascular thrombi, a patient should be transfused according to critical clinical evaluation after due consideration of the risks and benefits of transfusion. All patients should receive folic acid supplementation. Platelet transfusions are contraindicated unless there is life-threatening haemorrhage as they have been temporally associated with disease exacerbation (Harkness et al, 1981; Gordon et al, 1987). Some centres advocate the use of prophylactic phenytoin in an attempt to minimize secondary seizures. However, neurological features are completely absent in 35% of patients and seizures only occur in a minority of the remaining patients (Rock et al, 1991). Secondary rather than primary prophylaxis of seizures with phenytoin would, therefore, seem appropriate. Although fever is one of the defining features of TTP, an underlying infection should be sought actively; if untreated, occult infection may prevent response to plasma exchange or precipitate early relapse. Hepatitis B vaccination should be given to all patients and can be safely administered using a platelet threshold of 50 × 109/l. Recommendation. Red cell transfusion should be administered according to clinical need (Grade B, level III). Folate supplementation is required in all patients. Platelet transfusions are contraindicated in TTP unless there is life-threatening haemorrhage. Hepatitis B vaccination is recommended in all patients (Grade C, level IV). Despite the improvement in survival, there remains a subgroup with a slow or incomplete response to plasma exchange ± steroids. Refractory disease may be defined as persistent thrombocytopenia (platelet count < 150 × 109/l) or LDH elevation after a total of seven daily plasma exchange procedures. Treatment of this group of patients is difficult and a wide range of additional approaches has been used. Unfortunately randomized clinical trials have not been performed because of the rarity and heterogeneity of this condition. Manipulation of plasma exchange. There are reports documenting responses to plasma exchange in refractory cases of TTP following the substitution of either cryosupernatant (Molinari et al, 1993) or S/D plasma (Harrison et al, 1996) for FFP. Like cryosupernatant, S/D plasma lacks the largest plasma VWF multimers and this may be of importance. If, therefore, there is a suboptimal response to plasma exchange after 7 d or rapid clinical deterioration despite daily plasma exchange, an alternative replacement fluid should be substituted. While methylene blue (MB)-treated fresh-frozen plasma has been used in the management of TTP (Martinez et al, 2000), clinical experience is extremely limited at present and, unlike S/D plasma, VWF multimeric structure is not modified. Whether MB cryosupernatant has a role in the management of TTP will require further clinical experience. Intensification of plasma exchange has also been used in cases of refractory TTP with the introduction of either 12 h or double-volume plasma volume exchanges. At present, this approach is empirical. Recommendation. In the presence of refractory disease an alternative plasma product lacking high-molecular-weight VWF multimeric forms, cryosupernatant or S/D plasma should be used for plasma exchange (Grade C, level IV). Intensification of plasma exchange procedures should also be considered in life-threatening cases (Grade C, level IV). Vincristine. Although vincristine is often used in the treatment of refractory TTP, published literature supporting its efficacy comprises only case reports or small retrospective studies. Nevertheless, these suggest that the administration of vincristine in refractory TTP may be temporally associated with platelet recovery (Welborn et al, 1990; O'Connor et al, 1992; Bobbio-Pallavicini et al, 1994). A role for the early administration of vincristine (within 3 d of presentation) has also been advocated following a small retrospective study (Mazzei et al, 1998). However, such practice would carry a risk of inducing neuropathy without proof of clinical benefit. Until this finding can be corroborated by larger controlled trials, vincristine should be reserved for refractory cases. A number of dosage regimens have been employed with no clear advantage for any single one. A schedule of 1 mg repeated every 3 to 4 d for a total of four doses is popular as it may limit toxicity while retaining efficacy. Higher dose regimens have, however, been used apparently successfully. The mechanism of action of vincristine remains unclear. Recommendation. Vincristine 1 mg repeated every 3 to 4 d for a total of four doses is recommended in refractory TTP (Grade C, level IV). Cyclophosphamide. Cyclosphosphamide has also been advocated for the treatment of TTP, particularly those in patients who experience recurrent relapses (severe intermittent TTP) (Bird et al, 1990; Udvardy & Rak, 1990; Strutz et al, 1998). Both daily dosing and pulsed therapy have been used successfully, although reported numbers are extremely low. Cyclophosphamide is known to be a potent immunosuppressant and this is thought to underlie its efficacy. Cyclosporine. Although cyclosporine is associated with an increased risk of post-bone marrow transplant microangiopathy, there are reports of its successful application to the treatment of refractory (Hand et al, 1998), severe intermittent (Pasquale et al, 1998) and post-autologous bone marrow transplantation (BMT) TTP (Van Ojik et al, 1997). This is consistent with the current autoimmune model of TTP and cyclosporine's immunosuppressive action. While spontaneous resolution cannot be excluded in these patients, clinical and haematological response uniformly occurred 7 to 14 d after initiating treatment. Cyclosporine may, therefore, prove to be a useful therapeutic modality in these difficult patients. However, much remains unanswered. The optimal duration of treatment is unknown, with relapses documented after cessation of therapy (Hand et al, 1998; Pasquale et al, 1998). The optimal target therapeutic range is also unknown: trough serum levels of 200–300 µg/l have been used. The potential toxicity of this drug must also be considered. Recommendation. Intensive immunosuppression using either cyclophosphamide or cyclosporine is indicated in severe refractory or recurrent TTP (Grade C, level IV). The advent of VWF-CP assays is now beginning to confirm the clinical suspicion that TTP currently represents a heterogeneous group of conditions. It has long been recognized that plasma exchange is rarely effective in the treatment of BMT-associated TTP, suggesting that an alternative pathological process might be involved. This theory is supported by the recent finding that VWF-CP activity was normal in seven and only moderately reduced in one of eight patients with BMT-associated TTP (van der Plas et al, 1999). Effective treatment for this group of patients is, however, lacking. Resolution of autologous BMT-associated TTP has been reported following initiation of cyclosporine (Van Ojik et al, 1997), although paradoxically cyclosporine therapy is a recognized risk factor, along with total body irradiation, for allogeneic BMT-associated TTP. In the latter setting, cyclosporine should be stopped. Whether protein-A column immunoabsorption might be useful in such patients is unclear. Certainly this technique has been successfully employed in the treatment of malignancy-associated TTP in which plasma exchange is often found to be ineffective. In one small retrospective series, it was found to be of benefit in seven of 10 patients who had been unresponsive to plasma exchange (Gaddis et al, 1997). Recommendation. Malignancy and BMT-associated TTP are often refractory to plasma exchange. Protein-A column immunoabsorption may be considered (Grade C, level IV). Although remission is now attained in over 80% patients, subsequent relapse remains problematic. Data from the Canadian Apheresis Group estimate that over a 10-year follow-up period up to 36% of TTP patients relapse. Relapse has occurred up to 8 years after the index event (Shumak et al, 1995). All patients should be aware of the possibility of relapse and advised to report early if symptoms suggestive of relapse develop. At present, it is impossible to identify those patients at greatest risk, although the presence of ULVWF during periods of remission is associated with intermittent disease (Moake & McPherson, 1989). There is no consensus whether there is any effective intervention that might reduce this risk. Splenectomy has been advocated as a means of reducing the relapse rate. One small retrospective study including six patients showed encouraging results with relapse rates falling from 2·3 ± 2·0 to 0·1 ± 0·1 events per year when the operation was performed during haematological remission (Crowther et al, 1996). However, acute exacerbations of TTP have occurred post-operati