Management Of Thrombotic Thrombocytopenic Purpura


Dr Gail A. Rock, Chief, Division of Hematology and Transfusion Medicine, Ottawa Hospital, c/o Civic Campus, 1053 Carling Avenue, Ottawa ON K1Y 4E9. E-mail:

The syndrome

Thrombotic thrombocytopenic purpura (TTP) is not a common disorder, but the young age, acute onset, fulminant course and sometimes fatal nature of the disease make it remarkable. Although first described by Dr Eli Moschkowitz in 1924, the pathogenesis of the disorder has only recently begun to be truly understood ( Moschkowitz, 1924). Moschkowitz (1924) reported the presentation of a 16-year-old girl with fever, anaemia, central nervous system impairment, renal dysfunction and cardiac failure. The patient died after 2 weeks, with the autopsy showing hyaline thrombi in the terminal arterioles of the majority of organs – a finding considered to be characteristic of the disorder.

Over the years, a classical textbook description of a pentad of symptoms for TTP, consisting of microangiopathy, haemolytic anaemia, thrombocytopenia, fluctuating central nervous system abnormalities, fever and renal impairment, has developed. However, it now appears that the classical pentad is infrequently present in the early stages of disease. Only after there is widespread formation of microthrombi and a resultant impact on various organ systems does the full pentad express. In a series of 135 patients that we have recently reported, all patients had schistocytic haemolytic anaemia and thrombocytopenia ( Rock et al, 1998 ). However, only 30 had fever and 86 had neurological abnormalities. Renal impairment was present in 18% of patients. These findings are supported by the literature; a review of published cases by Ridolfi & Bell (1981) reported that 98% of patients had microangiopathic haemolytic anaemia (MHA), 83% thrombocytopenic purpura, 84% neurological symptoms and 76% had renal disease.

We have therefore proposed that TTP should be redefined as a syndrome of Coomb’s negative microangiopathic haemolytic anaemia and thrombocytopenia in the absence of other possible causes of these manifestations. In our experience, TTP occurs, for the most part, in previously healthy, relatively young individuals who suffer the sudden onset of a thrombotic disorder in which platelet microaggregates deposit in the arterial microvasculature. In our largest series of patients, 85 were women and 50 were men, with a mean age of 41·7 years (range 18–72) ( Rock et al, 1998 ).

Occlusion of small arterioles by platelet plugs containing variable quantities of von Willebrand factor (VWF) characterizes the disease. Electron microscopy has shown the thrombi to be composed of degranulated and altered platelets with little fibrinogen or fibrin ( Asada et al, 1985 ).

Incidence of TTP

Although TTP is uncommon, more than 200 cases have been reported each year for the last 5 years to the Canadian Apheresis Group, which captures the data on most TTP patients in the country. This is somewhat higher than the data from a recent US epidemiological study, which indicated an incidence of 3·7 cases per 1 000 000 residents ( Torok et al, 1995 ), but may be reflective of an enhanced awareness of the disorder in Canada because of our several TTP studies and the fact that the incidence of TTP appears to be rising recently. It is possible that the true mortality of the disease is underestimated as the majority of deaths occur within 48 h after presentation ( Brailey et al, 1999 ).

Presentation of TTP

Clinical Multiple forms of this syndrome are recognized, ranging from those in which the causative factor has been identified, such as in Shiga toxin-induced TTP/haemolytic uraemic syndrome (HUS), to the more obscure idiopathic forms. Most commonly seen is the acute single-incidence episode which, with appropriate therapy, generally resolves within weeks. Chronic relapsing forms of the disease are also seen. In four such patients, there has been documentation of the presence of unusually large factor VIII multimers in the intervals between relapse ( Moake et al, 1982 ). A juvenile chronic relapsing form has also been seen in which these patients appear to respond to infusion of plasma administered on a periodic schedule established to prevent relapses. Four of these patients have been shown to have a congenital deficiency of a protease which reduces high molecular weight forms of VWF ( Furlan et al, 1998 ).

Unfortunately, the initiating agent(s) of most forms of the syndrome is not yet known.

Laboratory Generally, the earliest initial presentation is purpura owing to thrombocytopenia. We found that the initial platelet count correlated with mortality: 32% of patients with a platelet count of ≤ 20 × 109/l died compared with 18% with a higher platelet count ( Rock et al, 1998 ). However, others have not found this association ( Brailey et al, 1999 ). Although it is well known that platelet microaggregates are formed, the evidence is variable regarding the activation of platelets in TTP. Several studies have shown an increase in CD62, an activation marker, but other studies have found this not to be the case ( Ahn et al, 1996 ).

Inevitably, there is a Coombs’-negative schistocytic anaemia and usually an increased serum lactate dehydrogenase (LDH). It is notable that, in 1998, Cohen et al (1998) characterized the elevated serum LDH and the isoenzyme profile in 10 consecutive patients with classic acute TTP, reporting that the elevated LDH was due to release of LDH5 from a variety of tissues damaged as a result of systemic ischaemia and not the direct result of an increase in isoenzymes LDH1 and 2 attributable to erythrocytes.

It is important to recognize that this is not a primary disorder of coagulation. Despite the formation of platelet microthrombi, disseminated intravascular coagulation (DIC) or other overt clinical or laboratory abnormalities of the coagulation system are rarely seen and the prothrombin time (PT), partial thromboplastin time (PTT) and fibrinogen levels are generally not altered. In our 135 patients, the PT and PTT were normal at entry in 128 patients ( Rock et al, 1998 ).

Although the patients presented at variable points in their disease, the majority had a quantitative elevation of von Willebrand Factor. This points to the generally accepted concept that TTP is ultimately related to some form of endothelial cell damage. This fact is strongly supported by both histological evidence and altered endothelial function, including decreased prostacyclin production, impaired fibrinolytic activities, release of thrombomodulin and, most recently, the finding of induction of apoptosis in microvasculature endothelial cells by TTP plasma ( Dang et al, 1999 ).

Recently, a considerable amount of evidence has accumulated for an immunological basis for at least some forms of TTP. The Italian Co-operative TTP group ( Porta et al, 1999 ) has just reviewed these data, which include evidence showing the presence of anti-C36 antibodies as well as an anti-von Willebrand factor cleaving metalloproteinase. Although the initiating factor(s) of the syndrome is not yet certain, it is apparent that there is a final common pathway of disease manifestation, involving the presence of increased amounts of von Willebrand factor both in the plasma and on the surface of the platelets and the formation and deposition of platelet aggregates in the microvasculature.


A number of possible pathophysiologies have been considered to account for this, as follows.

The presence of platelet aggregating factors In 1988, Siddiqui & Lian (1988) isolated a 37-kDa protein from plasma of some acute TTP patients. This protein agglutinates platelets through interaction with CD36, a platelet membrane glycoprotein ( Lian et al, 1991 ). Another group has demonstrated a proteolytic enzyme, calpain, in TTP plasma during active phases of the disease ( Murphy et al, 1987 ). Calpain is an intercellular enzyme that is found normally in many tissues, including platelets. It is thought to have a role in proteolysing platelet membrane GP1b and activating GPIIb/IIIa, the membrane binding site for VWF. Calpain also proteolyses VWF, producing characteristic multimer patterns on electrophoresis. This altered VWF is highly reactive with activated platelets and binds to GPIIb/IIIa, and has been shown to cause formation of platelet aggregates ( Moore et al, 1990 ).

von Willebrand Factor VWF is normally present in plasma at ≈ 1 U/ml and is involved in platelet adhesion to subendothelium and aggregation. A series of VWF multimers can be demonstrated on electrophoresis. Unusually large molecular weight (ULHMW) multimers are normally found in the Wiebel-Palade bodies of endothelial cells and in platelet alpha granules. These multimers, but not normal plasma VWF forms, can induce aggregation of platelets at high shear stress ( Moore et al, 1990 ).

Both quantitative and qualitative abnormalities of plasma VWF have been reported for patients with TTP. These involve the appearance in plasma of very high molecular weight multimers during remission of a chronic relapsing form of TTP ( Moake et al, 1982 ) and loss of the larger forms of normal multimers during acute or relapsing TTP. VWF is clearly important in the pathogenesis of the intravascular platelet aggregation and thrombus formation, as supported by the finding of VWF in the thrombi formed in TTP.

Mannucci et al (1989) found that VWF proteolysis was enhanced in acute TTP, but did not lead to loss of the larger multimers. Members of the Italian TTP Registry also found that in familial and sporadic TTP cases there was enhanced fragmentation of VWF during acute disease as demonstrated by a decrease in the native 225-kDa VWF subunits, suggesting that abnormal cleavage of VWF might occur ( Galbusera et al, 1999 ).

In our study of 135 patients with acute TTP ( Rock et al, 1991 ), we found VWF to be elevated but with a variable presentation of VWF multimers. Few patients had ULVWF forms, most had a normal distribution of multimers or lacked the larger normal forms. VWF multimers did not correlate with outcome, treatment or any other variable. Certainly, proteolytic enzymes including plasmin, calpain and elastase could all be responsible for the loss of larger multimers which are highly susceptible to proteolytic cleavage. At this time, the mechanisms governing the appearance and interaction of the VWF multimers are unclear, with a variety of reports giving different information. As various patients with chronic relapsing disease are seen to have ULVWF multimers in their plasma in quiescent times, some secondary or quantitative effect may be necessary to initiate the acute disease.

Antibodies to platelets and/or endothelial cells An aetiological role for antiplatelet antibodies was first suggested by elevated platelet-associated IgG (PAIgG) which resolved as the patient recovered ( Morrison & McMillan, 1977). It was suggested that this represented immune complexes possibly resulting from the response to bacterial or viral infection which interacted with platelets and caused them to aggregate and release. Subsequently, PAIgG has been reported to be elevated in many TTP patients ( Morrison & McMillan, 1977; Neame, 1980), but, as is the situation with most other markers in TTP, it has been reported as normal in others ( Ansell et al, 1978 ). However, the concept that certain antiplatelet antibodies lead to autothrombotic activity is now well appreciated and exemplified by heparin-induced thrombocytopenia (HIT) and the lupus anticoagulant with thrombosis. Antibody binding to specific platelet epitopes can cause activation, leading to episodic thrombotic events via FcγRII receptors and complement pathway activation.

We have found antiplatelet antibodies in the sera of 80 out of 102 patients studied ( Tandon et al, 1994 ). Protein blotting of patients’ sera demonstrated a significant number of antibodies directed against CD36. Of these, 23/27 (85%) reacted by immunoprecipitation and 17/28 (60%) by dot blots.

Human CD36 is a single-chain integral membrane polypeptide (also known as GPIV and IIIb). It is expressed in platelets as well as many other cells, organs and tissues. Many functions have been described for CD36, including a role as a cell-surface receptor interacting with a large number of ligands and implicated in intercellular signalling transcription ( Huang et al, 1991 ) . Interestingly, CD-36 is characteristically restricted to capillary endothelial cells and is not seen in the endothelium of large vessels ( Sverlick et al, 1992 ), thereby corresponding to the pattern of microthrombi deposition in TTP. We found that these anti-CD36 antibodies activated platelets in most cases and that this reaction was enhanced in the presence of purified VWF. The antibodies appear to be a part of a spectrum of autoantibodies arising from immune system hyperactivity or in response to platelet membrane changes in these TTP patients.

Recently, Schultz et al (1998) reported their study in which they also found anti-CD36 antibodies in 8/11 TTP patients using a PAIgG assay and 10/14 by immunoblot. Of note, they report two different forms of CD36 with both the classic 88-kDa form and an 85-kDa (less glycosylated) form. Patient sera reacted more strongly to the latter; monoclonal antibodies to the former. This may help to explain the variable results seen in different studies.

We have also found anti-CD36 autoantibodies in patients with the lupus anticoagulant and thrombotic complications, but not those with antiphospholipid antibodies without thrombosis ( Rock et al, 1994 ).

Antiendothelial cell antibodies have also been reported in TTP ( Leung et al, 1988 ). Burns & Zucker-Franklin (1982) showed that plasma from three patients with TTP caused time-dependent immune destruction of cultured endothelial cells and spontaneous aggregation of normal platelets in vitro. In our experience, the majority of TTP sera show reactivity by protein blotting against microvascular endothelial cells, again demonstrating interaction with CD36 ( Rock et al, 1998 ). Immune injury to vascular endothelial cells could expose thrombogenic subendothelial surfaces and release VWF from intracellular stores which could then potentiate platelet agglutination through some secondary interaction, such as high shear rate or platelet antibodies.

Changes in platelet function Early on it was suggested that patients with TTP have either a deficiency of platelet prostacyclin (PGI2) or of a precursor that promotes the formation of this prostaglandin which normally inhibits platelet adherence, aggregation and release ( Remuzzi et al, 1978 ). In our study ( Rock et al, 1998 ), the level of 6-keto prostaglandin F3α (PGFlα) was measured for the first 77 patients entered into the study. However, the results were highly variable and it was not considered useful to continue this assay.

Two other specific platelet abnormalities have been described: Murphy et al (1987) and Rock et al (1988) indicated that the quantity of platelet membrane GPIb was decreased in some patients with TTP. Murphy et al (1987) attributed this to the activation of the enzyme calpain. It should be noted that if platelets are treated in vitro with proteolytic enzymes to remove the glycocalicin (GPIb peptide) these platelets subsequently become hypo- rather than hyperaggregable. However, our finding that absorption of normal VWF multimers from plasma was impaired in three out of three patients with TTP gives support to an alteration in a surface membrane receptor which is either quantitatively or functionally diminished ( Rock et al, 1988 ). This suggestion that VWF is bound to the platelets in TTP has recently been confirmed by others using flow cytometry ( Chow et al, 1998 ).

Chow et al (1998) also found that although both single episode and recurrent adult TTP patients had platelet aggregates in their blood and increased VWF on single platelets the platelet α-granule protein P-selectin was not increased in most TTP blood samples. This suggests that the VWF bound to the platelets arises from plasma rather than from the α-granules. As the platelets themselves are not activated, this would appear to indicate that the VWF is coming from the damaged endothelium and binds to the platelets externally without specific activation.

This binding of VWF to platelets during disease manifestation may account for some of the variability seen in the plasma pattern for VWF multimers.

Antibodies against VWF cleaving metalloproteinase VWF is secreted by endothelial cells as a very large polymer of polypeptides joined by disulphide bonds ( Tsai et al, 1989 ). Subsequently, it is cleaved in the circulation between tyrosine at position 842 and methionine at position 843 by a 200-kDa metalloproteinase into smaller functionally low-adhesive dimers of 176 kDa and 140 kDa respectively ( Dent et al, 1990 ). Furlan et al (1997) reported that two brothers with a relapsing form of TTP had a constitutional deficiency of this VWF cleaving protease without any inhibitor. After plasma exchange, both patients had normalized platelet counts and LDH. The biological half life of the VWF cleaving protease was determined to be 3·3 and 2·1 d in these patients.

Recently, Tsai & Lian (1998) in New York and Furlan et al (1998) in Switzerland have independently found that the level of plasma VWF cleaving metalloproteinase activity is greatly reduced or absent during acute TTP episodes, with a return to baseline values after recovery. They have detected an autoantibody directed against the enzyme which appears to account for the lack of metalloproteinase activity during the acute disease. Furlan et al (1999) further reported that although TTP patients had impaired metalloproteinase activity this was not the case for familial or acquired HUS, and suggested that this may provide a useful tool to distinguish between the two diseases. However, another report ( Galbusera et al, 1999 ) of enhanced fragmentation of VWF in many TTP patients suggests factors other than ULVWF forms as being important in TTP.

Nevertheless, whatever the form, it is clear that the presence of excess VWF, and its binding to platelets, perhaps as a result of alterations in VWF structure, play a crucial role in the evolution of TTP.

Related events (secondary associations)

In most cases, TTP has an acute onset with no apparent initiating factors. However, whereas the majority of cases appear to be idiopathic, secondary associations with other disorders are well documented. These include Escherichia coli 0157:H7 infection ( Chart et al, 1991 ), pregnancy ( Caggiano et al, 1983 ), hormone contraceptive therapy ( Kwaan, 1987), bone marrow transplantation, chemotherapeutic agents including cisplatin and mitomycin ( Murgo, 1987) and human immunodeficiency virus (HIV) infection ( Ucar et al, 1994 ).

Haemolytic uraemic syndrome (HUS) is a disorder overlapping with TTP in that its final manifestation is the formation of platelet microthrombi with a haemolytic schistocytic anaemia, but differing in that the predominant affect is seen in the kidneys with deposition of platelet microthrombi in the glomeruli. Many consider TTP and HUS to be essentially the same disorder, perhaps with different initiating events but both having the same result of formation and deposition of platelet aggregates. In childhood HUS, a specific association with verotoxin producing E. coli has been determined ( Karmali et al, 1983 ). This verotoxin is known to have several direct effects, including the release of high molecular weight VWF from endothelial cells. It acts as an endothelial cell cytotoxin and inhibits protein synthesis. Overall, there has been relatively little evidence of an infectious complication in TTP, although some cases of TTP have been documented to follow haemorrhagic colitis due to E. coli 0147:117 ( Kovacs et al, 1990 ) and, in association with Bartonella infection ( Tarantolo et al, 1997 ), TTP has also been seen with occult infection and peridontal abscess.

The association between TTP/HUS syndrome and cancer is well established ( Gordon & Kwaan, 1999). In addition to the microangiopathic haemolytic anaemia, severe thrombocytopenia and renal failure which are always present, pulmonary oedema is commonly seen. The non-cardiogenic pulmonary oedema is said to be characteristic of this variety of TTP/HUS, making it distinct from the other types. TTP is most commonly observed in gastric adenocarcinoma, followed by carcinoma of the breast. Immune complexes are present in the plasma, whereas the Coombs’ test is negative. In a recent review, Gordon & Kwaan (1999) stated that the presence of immune complexes in ≈ 90% of cases was an unusual feature of cancer-associated TTP. However, in our early series of 102 patients with TTP, we found immune complexes in all patients, although it was necessary to use three different methods to determine this fact, suggesting some variability in the type of complexes which are formed. This may not be surprising if it is considered that the primary insult and therefore the stimulatory antigen may be different in different cases of TTP, but that there is a final common pathway of immune complex formation and endothelial injury with platelet deposition. An important observation made by Gordon & Kwaan (1999) is that DIC and TTP/HUS may co-exist in cancer patients and may confuse the diagnosis.

Patients receiving chemotherapy also appear to have a particular susceptibility to acquiring TTP ( Murphy et al, 1992 ), although the relative importance of chemotherapeutic agents such as mitomycin C and the underlying neoplasm itself is difficult to assess. The endothelial damage caused by certain of the chemotherapeutic agents is well known, as is the platelet-aggregating effect of certain neoplasms.

Ticlopidine, a thienopyridine compound, has also been associated with cases of TTP. First marketed in the USA in 1991, the drug alters platelet function by inhibiting the binding of adenosine 5′-diphosphate to its adenylyl cyclase-coupled receptor site. Given orally, the effect persists up to 7–10 d after drug discontinuation. Ticlopidine is used to prevent stroke and clot formation after cardiac stent placement ( Arcan et al, 1988 ). Recently, Bennett et al (1998) reported on 60 patients who developed TTP during ticlopidine treatment. These patients were mostly men aged over 60 years who had received ticlopidine for less than 1 month. The mechanism by which ticlopidine induces this syndrome is not known. These investigators have pointed out that ticlopidine-associated TTP has been markedly under-reported during its first years on the market. This is an important consideration in that clopidogrel (an agent that is chemically related to ticlopidine) has now captured 55% of the antiplatelet market in the USA. The presence of the skin rash in many cases of ticlopidine- or clopidogrel-associated TTP and the relative infrequency of the syndrome in patients receiving these drugs suggests that an autoimmune phenomenon may be involved.

Aids-related TTP has been reported to be common in some series ( Torok et al, 1995 ). In our Canadian experience, this is not frequently seen.

There have been reports of a relationship between progestogen-only contraceptives and TTP ( Fraser et al, 1996 ), and physicians have been cautioned to consider this development in users of the Norplant system. TTP may also develop during pregnancy, where it is more commonly seen in the third trimester. Ezra et al (1996) reported that women who are either pregnant or in the postpartum period make up 10–25% of TTP patients, and stated that once the disease occurs during the pregnancy it tends to recur in subsequent pregnancies. In our original series of 102 patients randomized to our trial, seven patients were pregnant or had just delivered. Overlap with pre-eclampsia and HELLP syndrome may complicate diagnosis.

TTP has also been described in association with systemic lupus erythematosus (SLE). However, this association is rare and the diagnosis may be challenging. Autoimmune mechanisms including platelet antibodies may be shared with SLE ( Musio et al, 1998 ). In one series of patients with the lupus anticoagulant and thrombus, we have demonstrated the presence of antibodies to CD36. These antibodies were not present in other patients that had the lupus anticoagulant but no thrombosis ( Rock et al, 1994 ).

Therapy of TTP

For many years, TTP remained an almost universally fatal disorder. More than a quarter of a century after the disease was first described, it was discovered that plasma infusion was able to reverse the course of disease, a fact that was attributed to the presence of a substance in plasma that inhibited the factor responsible for causing platelet aggregation ( Byrnes & Khurana, 1977). Early success with plasma infusion led to trials of plasma exchange based on the theory that there might be benefit to the simultaneous removal of any toxic factors. The development of cell separation devices, which revolutionized the approach to therapeutic plasma exchange, made it possible to achieve large volume plasma removal so that a 1–1·5× plasma volume exchange could be achieved within a matter of hours with replacement of the patient’s plasma with appropriate other fluids. A review ( Amorosi & Ultmann, 1966) showed that before widespread plasma exchange the mortality rate was very high, exceeding 90%.

In 1991, the Canadian Apheresis Group reported on the results of a trial in which 102 patients were randomized to receive either plasma exchange with fresh frozen plasma (FFP) or plasma infusion on 7 of the first 9 d of the trial ( Rock et al, 1991 ). The plasma exchange (PE) procedure required 1·5× plasma volume exchange for the first three procedures followed by 1·0× plasma volume replacement thereafter. The infusion patients received 30 ml FFP/kg over 24 h then 15 ml/kg each day after.

We found that plasma exchange was preferable to plasma infusion in the treatment of TTP, with 78% survival at 1 month. In so reporting, the authors recognized that the volume of plasma administered in the exchange arm was threefold greater than that which was infused. Therefore, if therapy is dependent upon the delivery of a large quantity of a putative factor, plasma exchange may have been more successful simply by delivering a larger volume of this factor. Our study was designed to assess the difference between two therapies; clearly, a limiting factor in the use of plasma infusion is the volume which can be tolerated by the patient. Thus, although not directly resolving the question of the relative benefit of removal of toxic component vs. replacement of missing component, the study was definitive in defining the relative benefits of two forms of therapy: plasma exchange vs. plasma infusion (PI). The fact that some benefit was seen with PI only (compared with historical results) argues for a missing or altered factor in the circulation.

Bell et al (1991) have also reported that vigorous plasma exchange with FFP has been most effective primary therapy for TTP. Limited data from several studies using 5% albumin for the first half of the exchange followed by FFP suggests that this approach is equivalent or better than using FFP alone.

In our next study, we evaluated the use of cryosupernatant plasma (CSP), which is the supernatant plasma obtained after cryoprecipitate production and removal. This therapy was first suggested by Byrnes & Khurana (1977) and is based on the hypothesis that high molecular weight VWF multimers are involved in the pathophysiology of the disease. It is reasoned that as cryoprecipitate supernatant is relatively deficient in the higher molecular weight multimers of factor VIII this material could provide the putative missing factor, while not adding to the dose of ULVWF forms.

A comparison of the VWF multimer patterns in plasma, cryoprecipitate supernatant and the cryoprecipitate made from that plasma using standard techniques is shown in Fig 1. The quantity of VWF is decreased in CSP compared with FFP; following individual units of fresh plasma through the cryoprecipitation step, we have found an average of 1·01 units/ml of VWF in FFP and 0·18 units/ml in CSP. This varies with blood group, with CSP from AB group patients having the lowest level of VWF ( Gerhard et al, 1998 ). Recent unpublished work that we have carried out with Dr Tsai (Montefiore Medical Center, New York, USA) indicates that the metalloproteinase that he has described is present in CSP in the same concentration as in plasma – it is not removed in the cryoprecipitate.

Figure 1.

VWF multimer patterns of plasma, cryoprecipitate and cryosupernatant plasma. Two different preparations were sampled from left to right: PC, plasma control; P, plasma; C, cryoprecipitate; S, supernatant plasma; Plt C, platelet lysate control. Plasma was obtained from normal healthy blood donors and cryoprecipitate was made according to the standard operating procedures of the Canadian Blood Services with the cryosupernatant retained for analysis. VWF multimers were run using a 0.75% agarose stacking gel and 1.5% running gel. Rabbit anti-VWF was obtained from Diagnostica Stago. Protein blots were developed with NBT BCIP and alkaline phosphatase with goat anti-rabbit conjugate (Biorad).

As an initial study, patients with TTP who had not responded to plasma exchange with FFP replacement were exchanged with cryosupernatant plasma following the protocols and volumes used in our first study. Eighteen patients who failed the first course (average 7·7 plasma exchanges with FFP) received a further seven exchanges with CSP. Eleven patients responded (defined as an increase in platelet count to greater than 150 × 109/l and no neurological events). After seven exchanges, we saw a 60% response (platelet count < 150 000 with no new neurological events) and 83% of patients survived at 1 month.

Next, 40 previously untreated patients received CSP. The response rate was 75% at the end of the first cycle and 95% of the patients were alive after 1 month. These values were significantly better than those reported in our earlier study comparing plasma exchange and plasma infusion and better than those seen in other TTP patients treated concurrently at the same centres but receiving FFP replacement because of the limited availability of CSP. These data are statistically significant. We have now initiated a randomized prospective trial of plasma exchange comparing cryosupernatant plasma with fresh frozen plasma ( Rock et al, 1996 ).

In a series of 15 of these patients, we have followed the levels of von Willebrand factor and the multimer patterns on days 0, 3 and 5 of plasma exchange. As shown in Fig 2, the levels of von Willebrand factor were lower throughout the course of therapy in those patients who received cryosupernatant plasma. The multimer patterns showed some decrease in quantity and in the presence of larger forms ( Fig 3) with CSP after therapy. A retrospective analysis by Owens et al (1995) also compared patient survival after the use of either FFP or CSP. The survival was 72% in the group that received CSP and 47% in the group that received FFP. Both of these figures were considerably lower than those that we obtained in our series but the relative outcomes were the same, showing the superiority of CSP.

Figure 2.

Concentration of VWF in plasma during PE therapy. Data are from two individual patients.

Figure 3.

VWF multimer patterns before and after plasma exchange with FFP or CSP. PC, plasma control; 1, plasma before day 1; 2, plasma after day 1; 3, before day 3; 4, after day 3; 5, before day 5; 6, after day 5; Plt C, platelet lysate control. (A) Patient treated with FFP. (B) Patient treated with CSP.

In our experience, patients with TTP/HUS, i.e. with evidence of renal dysfunction, have responded to PE in the same way as those with TTP alone ( Rock et al, 1992 ).

Schedule for plasma exchange

There are no scientific studies that have precisely determined the optimal plasma exchange (PE) treatment schedule. The American Association of Blood Banks recommends daily PE until the platelet count is above 150 × 109/l for 2–3 d ( AABB Extracorporeal Therapy Committee, 1992). The American Society for Apheresis recommends daily PE until the platelet count is above 100 × 109/l and continues to rise after cessation of treatment and the LDL level is near normal ( Gilcher et al, 1993 ). However, there is wide variability of therapeutic targets, as shown by a recent survey of 20 institutions in the USA ( Bandarenko & Brecher, 1998). Sixty per cent of the institutions established a platelet count of at least 150 × 109/l as a requirement before discontinuing daily PE, 35% have established a platelet count of 100 × 109/l and one institution continued daily PE until a platelet count of 200 × 109/l was reached. Additionally, some institutions used LDH as a therapeutic target, with 50% of the institutions requiring the serum LDH to be within the normal reference range before discontinuing daily PE. Some centres did not follow serum LDH in their management. This study determined that 60% of the centres did not routinely utilize a tapering regimen for management of initial episodes of PE.

We have used a tapering schedule involving five PE treatments over 2 weeks for responders and five PE per week in partial responders. The response to PE therapy or indeed any other therapy is rarely immediate. In our studies, incomplete responses were relatively common at the end of the first cycle of therapy; for 51 patients, the treatment period averaged 15·8 d (range 3–36 d), another (non-study) patient received 91 plasma exchanges.

Use of solvent detergent treated plasma

One of the drawbacks to the use of fresh frozen plasma or cryosupernatant plasma is the exposure of patients to the plasma from very large numbers of donors, even when apheresis plasma (500–800 ml per donor bag) is used. A relatively new plasma product, which is treated with a solvent and detergent (SDP) to inactivate lipid-enveloped viruses, has been developed. SDP may have an added advantage for use in TTP because, for unknown reasons, it has been reported to lack the high molecular weight forms of VWF ( Moake et al, 1994 ) and may therefore be beneficial as the primary plasma replacement. We have recently examined the multimer patterns from SDP of various blood groups, as shown in Fig 4 (material kindly provided by Dr F. Darr, American Red Cross) and found that the higher molecular weight forms of VWF found normally in plasma are indeed decreased in all samples.

Figure 4.

VWF multimer patterns in solvent detergent-treated plasma. PC 1 and 2, two different normal plasma controls; SDP A, blood group A solvent detergent-treated plasma; SDP O, blood group O; SDP B, blood group B; SDP AB, blood group AB; Plt C, platelet lysate control.

A disadvantage of this material is that the process does not inactivate non-lipid viruses. This includes such infectious agents as parvovirus and hepatitis A. A recent recall of SDP in the USA was based on parvovirus transmission. Therefore, steps are being taken to include nucleic acid testing of plasma for parvovirus. Further, SDP is derived from large plasma pools and therefore any new non-lipid infectious agent would be disseminated throughout the pool, thereby potentially increasing donor exposure.

Data are available showing successful treatment of TTP patients with SDP plasma ( Moake et al, 1994 ). Although the metalloproteinase is reportedly present in SDP, large-scale comparative trials are necessary to prove that SDP is at least equivalent to FFP.

Protein A columns

Selective removal of immunoglobulins through on line immunoabsorption using columns of protein A has proven to be therapeutic in disorders such as haemophilia A with inhibitors in which the antibody responsible for the pathophysiology of disease is well known. The presence of immune complexes, antiplatelet and antiendothelial cell antibodies in TTP is now well documented. Therefore, it is reasonable to consider that their selective removal or removal of a protease inhibitor would be of benefit. Protein A immunoabsorption has been reported by Gaddis et al (1997) to be effective in the therapy of 10 TTP patients who had previously been refractory to plasma exchange. Snyder et al (1993) have reported responses in 25/55 cases of cancer-associated TTP/HUS when immunoabsorption columns were used to remove immune complexes from the circulation.

Recent advances

Recent advances by both Tsai & Lian (1998) and Furlan et al (1998) have helped to shed light on the pathophysiology of the disease and may eventually have significant impact on primary therapy. As described earlier, both of these investigators have shown the presence in the patient’s plasma of an inhibitor to a metalloproteinase which is normally present in the plasma and is responsible for the cleavage of the unusually large molecular weight VWF which can induce platelet aggregation. Therefore, removal of antibody to the metalloproteinase and replacement of the inactive (blocked) enzyme with the protease from fresh frozen plasma may explain the reason for the success of plasma exchange. If so, this would uniquely define a therapy which was empirically decided upon – because it was most often successful – without an understanding of the mechanism by which it worked. Thus, the recent advances in elucidating the possible pathophysiological basis may have significant impact in TTP. This will be particularly so if it is possible to purify the enzyme from plasma and use it to treat specifically these patients. This would require infusion of sufficient amounts of the enzyme to overcome the inhibitor, in much the same way as factor VIII is administered to haemophiliac patients with inhibitors. Additionally, various adjuvant therapies might be useful to effect immune modulation and antibody suppression.

Adjuvant therapy in TTP

Historically, a variety of adjunctive treatments have been used in combination with plasma exchange for the treatment of TTP. Prominent among these are a number of antiplatelet drugs, particularly aspirin, dipyramidole and sulphinpyrazone. Ticlopidine has also been used with some benefit to treat TTP ( Ishii et al, 1984 ). In our first Canadian study, patients in both arms of the trial received combined therapy with aspirin (325 mg/d) and dipyrimadole (400 mg/d). We continue to use aspirin in combination with plasma exchange. The rationale for this is based on the formation of platelet aggregates and the consideration that inhibition and down-regulation of platelet responses would be of benefit, although, as stated earlier, the literature is variable in regard to demonstrating platelet activation in TTP. Bobbio-Pallavicini et al (1997) randomized 72 TTP patients to receive PE and steroids with or without aspirin and dipyrimadole and evaluated efficacy after 15 d. They concluded that the results showed the usefulness of antiplatelet agents in the treatment of acute phase TTP and that 1 year of triclopidine maintenance therapy appeared to be beneficial in preventing relapses.

As with other therapies in TTP, the results of splenectomy, either carried out early in the disease or to prevent recurrent relapse, have been variable. In 1996, Crowther et al (1996) followed six consecutive patients who had one or more relapses of TTP over a 10-year period. They found that splenectomies carried out during immunological remission reduced the frequency of acute relapse and the resulting need for medical therapy. It should be kept in mind that splenectomy is not a new therapy for TTP. Therefore, although there is mounting evidence of an immune aetiology, past history has not shown an overwhelming response to this therapy and, thus, splenectomy alone or even in combination with PE is unlikely to provide the final answer to treatment.

Based on the concept of an immunological basis for the disease, a number of immune modulatory approaches have been taken. Corticosteroids have been widely used. Moake et al (1994) suggested a dose of methylprednisolone of 75 mg/kg i.v. every 12 h or prednisone orally at about 1·0 mg/kg. An appropriate randomized study to assess the role of steroids has not been carried out. Consequently, it is difficult to determine for certain the benefits of steroid administration. However, as evidence supporting an immune-mediated basis for the disorder continues to accumulate, steroid therapy increases in acceptance.

Vincristine sulphate is a strong immunosuppressant and has also been used to treat different types of thrombocytopenia, including ITP and secondary thrombocytopenias ( Ahn et al, 1974 ). Again, based on the acceptance of an immunological disorder in TTP, there is considerable rationale for the use of vincristine, which is thought to cause an immunomodulatory effect on the endothelium whose disruption seems key in TTP. Burns & Zucker-Franklin (1982) reported that endothelial cells exposed to TTP plasma were damaged by IgG antibodies and that the use of vincristine prevents this interaction. Further, vincristine is known to alter platelet membrane glycoprotein receptors which would prevent up-regulation and activation.

In one report, survival increased to 88% when vincristine was administered within 3 d of beginning PE ( Mazzel et al, 1998 ). In a retrospective study by Bobbio-Pallavicini et al (1994) , patients receiving vincristine after failing to respond to PE had a survival rate of 50%. These investigators also found benefit in concurrent therapy ( Bobbio-Pallavicini et al, 1994 ). Whether the use of vincristine adjuvant therapy after failure for PE will compromise outcome is not known.

IVIgG has also been used in TTP, both as a primary (rarely) and as an adjunctive therapy. When used in combination with PE, it is important, as with other drugs, to time the administration to avoid depletion during the exchange procedure.

The recent finding of a metalloproteinase inhibitor added further weight to the concept of an immunological cause for TTP. Consequently, there has been an increased interest in the use of immune modulatory approaches, including the use of IVIgG and splenectomy. With IVIgG, there is the historical example of the success seen with IVIgG in both childhood and adult ITP, a thrombocytopenia which is known to be mediated by platelet antibodies. This enthusiasm, however, is tempered by the fact that both IVIgG and splenectomy have been used in TTP in the past with variable and unpredictable results. Therefore, in spite of the fact that more data are accumulating to indicate that TTP is an immunological disease, the historical data do not support the use of IVIgG.

In fact, IVIgG has not seen much acceptance in the treatment of TTP, although it has been available and sporadically used for years. It may be that standard IVIgG preparations do not contain appropriate anti-idiotype antibodies which are inhibitory to the metalloproteinase or antiplatelet antibodies. Until more definitive information on the specific role of antibodies in TTP is available, it will be difficult to convince any group of the benefit of using IVIgG to treat TTP in a randomized prospective trial. Without such a trial, IVIgG will probably remain an adjuvant therapy used particularly in patients who do not respond to plasma exchange.

One concept that has clearly evolved from the literature is that there appears to be general agreement to avoid platelet transfusions in these thrombocytopenic patients based on the understanding that the patient is not bleeding but rather that platelets are part of the pathophysiological event and that provision of additional platelets would only contribute to formation of more aggregates. Additionally, there have been anecdotal reports of fatalities after platelet transfusion ( Harkness et al, 1981 ; Gordon et al, 1987 ). Therefore, platelets should be used to treat TTP only in patients experiencing life-threatening haemorrhage.


A 10-year retrospective analysis of our original group of 102 patients showed that 36% of patients have relapsed, with this occurring at variable numbers of years after initial recurrence ranging from 8 months to 9 years ( Shumak et al, 1995 ).

Pasquale et al (1998) reported a single patient with 18 episodes of relapsing TTP in which the last 11 episodes were treated with regimes containing cyclosporin. The initial episode of TTP had failed to respond to PE and steroids, remitting only after splenectomy. Subsequent episodes (2–7) responded to FFP plus prednisone, whereas episode 8 failed to respond to steroids plus FFP but remitted with cyclosporin plus prednisone.

The role of splenectomy in preventing relapses has yet to be defined. Although splenectomy is infrequently used at first presentation, it has been common practice in some areas to carry out splenectomy after remission in order to prevent further recurrence. Several small studies have described patients who have apparently benefited from such an approach ( Winslow & Nelson, 1995).

We and others ( Bandarenko & Brecher, 1998) have found no predictable laboratory marker for relapse, including either final platelet count or the LDH level. Should the presence of a protease inhibitor be determined to be the cause of acute TTP then a diagnostic test may soon exist for predicting relapse and also assist in the determination of the number of therapeutic procedures required.

The high rate of relapse emphasizes the importance of both physician and patient awareness of this possibility. In our experience, early intervention in these relapsing cases appears to be most successful and is often triggered by a patient who recognizes the significance of new purpuric lesions.

Refractory TTP

In spite of aggressive PE and adjuvant therapy, some patients do not respond. This is most often seen with secondary causes, such as in bone marrow transplant-related and mitomycin-induced TTP in which the disease does not respond particularly well to PE therapy.

Refractory cases of TTP have been treated by the use of intravenous immune globulin, protein A columns, splenectomy, cyclophosphamide and vincristine, sometimes with success. The reports of TTP associated with various microbial agents suggests a potential role for empirical antimicrobial coverage in the management of TTP. Cryosupernatant plasma has also been reported to be effective in the treatment of otherwise refractory TTP ( Byrnes et al, 1990; Rock et al, 1998 ).

Because of the implication of immune mechanisms in TTP and the demonstration of antibodies to both platelets and a metalloproteinase, splenectomy should be considered in refractory patients. The data from Crowther et al (1996) supports this approach.

Nonetheless, the treatment of refractory TTP remains a continuing problem. Until more definite information on pathophysiology is available, it is likely that concerned physicians will continue to attempt treatment of refractory patients using any or all of a mixture of available therapies in the hope of resolving the disease.

Future research

Three areas of research activity are immediately apparent, as follows:

Purification of the metalloproteinase At this time, there is considerable excitement about the role of an inhibitor to the metalloprotease in TTP. Isolation and purification of this protease is currently a matter of extreme interest and is certain to be a focus of research in the next months. One of the early difficulties in pursuing this isolation was the relatively complicated assay necessary to demonstrate protease activity through cleavage of von Willebrand factor. However, several laboratories now have functional assays which are easier to carry out. This should allow purification of the protease with a view to ultimately using it to replace specifically the inhibited protein – in much the same way as factor VIII concentrates are currently used to treat factor VIII patients with inhibitors.

Development of a clinical assay for protease deficiency A simplified assay that can be easily performed in a routine clinical laboratory would facilitate the early diagnosis and treatment of these patients. This assay would need to be able to quantify either the metalloproteinase itself or the inhibitors to this protease.

Importance of a reduced level of von Willebrand factor in the solution used for therapy must be defined It is of immediate concern that a definitive clinical trial comparing cryosupernatant plasma with FFP be completed in order to determine whether the absence of the higher molecular weight forms of von Willebrand factor are of benefit in therapy. That is to say, even if a protease inhibitor is the primary cause of TTP, the question remains ‘do the high molecular weight forms of VWF “fuel the fire”?’ It should also be resolved as to whether the effort involved in preparing cryosupernatant plasma is justified.

In the same way, it will be important to assess in a large-scale randomized clinical trial whether solvent detergent-treated plasma is as good as or better than cryosupernatant plasma and whether the alterations in von Willebrand factor in the solvent detergent-treated plasma have any impact on the disease.


The first step in patient management is confirmation of the diagnosis. The clinical presentation must be considered with emphasis on the fact that the expression of a full pentad of symptoms is not required for diagnosis. A full blood count and blood smear should be reviewed and LDH and results of coagulation tests including PT, PTT and fibrinogen considered.

If TTP is diagnosed or suspected, PE therapy should be commenced immediately. The 1·5× plasma volume PE with FFP or CSP should be started with daily exchange for at least the first 3–4 d after presentation. Therapy should be maintained until the platelet count has risen to > 150 000 × 109/l and has stabilized for at least 2 d. Long-term (months) PE should not be ruled out. Adjuvant therapy should also be used as considered warranted. There are no data indicating that antiplatelet agents have contributed to bleeding problems in these patients and they do appear to have benefit; the side-effects of steroid therapy are well known.

In refractory patients, there appears to be no predictable pattern of response, therefore any of the adjuvant or alternative therapies including protein A immunoabsorption, vincristine and IVIgG should be considered.

Until a definite aetiology and diagnostic test(s) are determined, practitioners will have to contend with the fact that although remarkable progress has been made in the treatment of TTP the treatment is still empirical and not yet predictably effective. Only by defining the true pathophysiology of this disorder will definitive therapy be possible. Certainly, much has been accomplished, but patients continue to die from this acute onset disorder involving platelet aggregate formation.


The author wishes to acknowledge and thank the members of the Canadian Apheresis Group and the Canadian Association of Apheresis Nurses who have contributed patients and data to our TTP studies. Thanks also to Elizabeth Kipp for processing this manuscript, Dr Graham Jamieson for review and discussion and Marisa Freedman for technical assistance. This work has been supported in part through a grant from the Bayer Blood Partnership Fund.