Bernhard Lämmle, Department of Hematology and Central Hematology Laboratory, Inselspital, University Hospital, CH 3010 Bern, Switzerland. Tel.: 41-31-632 33 02; e-mail: email@example.com
Summary. This overview summarizes the history of thrombotic thrombocytopenic purpura (TTP) from its initial recognition in 1924 as a most often fatal disease to the discovery in 1997 of ADAMTS-13 deficiency as a major risk factor for acute disease manifestation. The cloning of the metalloprotease, ADAMTS-13, an essential regulator of the extremely adhesive unusually large von Willebrand factor (VWF) multimers secreted by endothelial cells, as well as ADAMTS-13 structure and function are reviewed. The complex, initially devised assays for ADAMTS-13 activity and the possible limitations of static in vitro assays are described. A new, simple assay using a recombinant 73-amino acid VWF peptide as substrate will hopefully be useful. Hereditary TTP caused by homozygous or double heterozygous ADAMTS-13 mutations and the nature of the mutations so far identified are discussed. Recognition of this condition by clinicians is of utmost importance, because it can be easily treated and – if untreated – frequently results in death. Acquired TTP is often but not always associated with severe, autoantibody-mediated ADAMTS-13 deficiency. The pathogenesis of cases without severe deficiency of the VWF-cleaving protease remains unknown, affected patients cannot be distinguished clinically from those with severely decreased ADAMTS-13 activity. Survivors of acute TTP, especially those with autoantibody-induced ADAMTS-13 deficiency, are at a high risk for relapse, as are patients with hereditary TTP. Patients with thrombotic microangiopathies (TMA) associated with hematopoietic stem cell transplantation, neo-plasia and several drugs, usually have normal or only moderately reduced ADAMTS-13 activity, with the exception of ticlopidine-induced TMA. Diarrhea-positive-hemolytic uremic syndrome (D+ HUS), mainly occurring in children is due to enterohemorrhagic Escherichia coli infection, and cases with atypical, D− HUS may be associated with factor H abnormalities. Treatment of acquired idiopathic TTP involves plasma exchange with fresh frozen plasma (FFP), and probably immunosuppression with corticosteroids is indicated. We believe that, at present, patients without severe acquired ADAMTS-13 deficiency should be treated with plasma exchange as well, until better strategies become available. Constitutional TTP can be treated by simple FFP infusion that rapidly reverses acute disease and – given prophylactically every 2–3 weeks – prevents relapses. There remains a large research agenda to improve diagnosis of TMA, gain further insight into the pathophysiology of the various TMA and to improve and possibly tailor the management of affected patients.
Historical aspects of thrombotic thrombocytopenic purpura (1924–1998)
In 1924, Dr Eli Moschcowitz described a 16-year-old girl who died within 2 weeks after the abrupt onset and progression of petechial bleeding, pallor, fever, paralysis, hematuria and coma . Disseminated microvascular ‘hyaline’ thrombi were detected at autopsy, and these widespread thrombi in arterioles and capillaries, later found to be largely composed of platelets, remain the pathologic hallmark of Moschcowitz’ disease or thrombotic thrombocytopenic purpura (TTP) today [2,3]. Moschcowitz suspected that a powerful agglutinative and hemolytic poison was responsible for this disease .
In a landmark paper of 1966, Amorosi and Ultmann  reviewed some 250 reported patients with TTP, added 16 new cases and established a pentad of clinical and laboratory features still considered to be the key diagnostic criteria: microangiopathic hemolytic anemia with fragmented erythrocytes (schistocytes) in the peripheral blood smear (Fig. 1), thrombocytopenia, (often fluctuating) neurologic signs and symptoms, renal dysfunction and fever. Nowadays, it is generally believed that intravascular platelet clumping under high shear stress in the microcirculation results in thrombocytopenia, ischemic neurologic, renal and other organ dysfunction and intravascular fragmentation of red blood cells in the partially occluded arterioles and capillaries.
Hemolytic uremic syndrome (HUS), reported in 1955 by Gasser et al.  in five children, is a disease clinically very similar to TTP. In routine clinical practice, TTP was often diagnosed in adult patients with predominant neurologic symptoms whereas a diagnosis of HUS was often made in children with predominant renal failure. Nevertheless, this distinction was not universally accepted and some authors adopted the term ‘TTP/HUS’ presuming a similar pathomechanism with variable organ tropism. Whereas many cases of TTP occur in previously healthy persons, childhood HUS is often associated with preceding hemorrhagic colitis caused by verocytotoxin-producing Escherichia coli O157:H7 infection, and this illness is nowadays labelled typical (diarrhea-positive or D+) HUS .
In addition, thrombotic microangiopathies (TMA) associated with pregnancy, HELLP syndrome (hemolysis, elevated liver enzymes, low platelets), disseminated cancer, anticancer agents such as mitomycin C, hematopoietic stem cell transplantation, various drugs such as cyclosporine A, ticlopidine, clopidogrel, quinine and others, and human immunodeficiency virus infection have been observed and variably referred to as TTP, HUS, TTP-HUS, TTP-like disease or secondary TTP (for review see Refs [8,9]).
Numerous hypotheses on the etiology and pathogenesis of idiopathic TTP have been put forward over the years (for reviews, see Refs [8,10–12]). Among others, endothelial injury, e.g. by oxidative stress, decreased prostacyclin production, reduced fibrinolytic capacity of the vessel wall, anti-endothelial cell autoantibodies and specifically antibodies toward glycoprotein IV (CD36) [13,14] that is located on microvascular endothelial cells and platelets, and the capacity of TTP plasma to induce apoptosis of microvascular endothelial cells  have been proposed as pathogenetic factors. Moreover, a 37-kDa protein , and a 59-kDa protein or a calcium-dependent cysteine protease (calpain) [17,18] were identified in serum or plasma from patients with acute TTP and suggested to be responsible for in vivo platelet aggregation. In 1982, Moake et al.  reported the presence of unusually large von Willebrand factor (ULVWF) multimers in plasma of four patients with a chronic relapsing course of TTP during remission. They suspected that these highly polymeric VWF multimers, similar in size to those found in endothelial cell-culture supernatant, were responsible for in vivo platelet clumping in the microvasculature. Moake et al.  hypothesized that the lack of a ‘depolymerase’ was responsible for the persistence of these ULVWF multimers in their patients.
In 1996, Furlan et al.  and Tsai  simultaneously isolated a hitherto unknown plasma protease that specifically cleaved VWF multimers at the peptide bond Tyr842–Met843 of the mature VWF subunit (Tyr1605–Met1606 in amino acid numbering including the VWF propeptide), the peptide bond previously shown to be cleaved during physiologic processing of VWF in vivo . One year later, four patients, including two brothers, with a chronic relapsing TTP and showing ULVWF multimers in their plasma during remission, were found to completely lack any VWF-cleaving protease (VWF-cp) activity . In 1998, we observed another patient with a severe course of TTP lacking any VWF-cp activity whose plasma contained an IgG autoantibody inhibiting VWF-cp activity in normal plasma . The inhibitor disappeared transiently after plasma exchange and replacement of fresh frozen plasma (FFP), corticosteroid and vincristine treatment and this was paralleled by normalization of VWF-cp and clinical remission. Reappearance of the IgG inhibitor with disappearance of protease activity preceded the first clinical relapse and only splenectomy performed 1 year after disease onset led to persistent clinical remission, absence of inhibitor and normal VWF-cp activity . Two separate retrospective studies on large cohorts of patients with TTP and HUS appearing in the same issue of the New England Journal of Medicine [25,26] demonstrated that the majority of patients with acute sporadic TTP had a severe deficiency of VWF-cp, most of them with inhibiting autoantibodies that disappeared in all  or some  patients in remission. Six familial cases (three pairs of siblings) had a complete protease deficiency without inhibitors  and 23 patients with a diagnosis of HUS had normal or subnormal VWF-cleaving protease activity .
Acute TTP was mostly fatal until the empirical introduction of plasma therapy in the 1970s . In a prospective randomized study, the Canadian Apheresis Study Group  showed the superiority of plasma exchange and FFP replacement over FFP infusion. Using plasmapheresis and FFP replacement, some 80% of the patients survive the acute TTP episode . The number of plasma exchange procedures and hence the treatment duration varies greatly and many patients relapse during follow-up [29–31]. Often, additional treatment, such as corticosteroids , vincristine, other immunosuppressive medication and/or splenectomy is given, especially in refractory or relapsing cases (for reviews, see Refs [8–10,33]). These largely empirical treatments seemed to be pathophysiologically supported by the discovery that many patients with acute TTP had an autoantibody-mediated deficiency of the specific VWF-cp: plasma exchange and corticosteroids would probably remove circulating autoantibodies and suppress formation of VWF-cp inhibitors, respectively, and FFP replacement would supply the lacking protease.
The very careful clinical observation during long-term follow-up of a patient with frequently recurring severe thrombocytopenia and microangiopathic hemolytic anemia since childhood, repeatedly showing a prompt response within a few hours to simple plasma infusion, led Upshaw  to conclude that his and Schulman's similar patient  were congenitally deficient in a plasma factor protecting from hemolysis and thrombocytopenia.
For reviews giving personal accounts of the discovery of ULVWF in TTP, VWF-cleaving protease and its deficiency in TTP the reader is referred to three interesting recent historical sketches by Furlan , Tsai  and Moake .
Cloning, structure and function of the von Willebrand factor-cleaving protease (ADAMTS-13)
von Willebrand factor-cleaving protease was purified to homogeneity and subjected to N-terminal amino acid sequence analysis [39–41]. This allowed to identify VWF-cp as a new member of the ADAMTS (a disintegrin and metalloprotease with thrombospondin type 1 motifs) family of metalloproteases, denoted as ADAMTS-13  and to locate the gene to chromosome 9q34.
Simultaneously, Levy et al. , performing a genome-wide linkage analysis in patients with hereditary TTP displaying severe VWF-cp deficiency and their family members detected the same gene, ADAMTS-13. They identified several different mutations as being presumably responsible for the severely deficient protease activity and hereditary TTP in homozygous or double heterozygous carriers of mutated alleles, whereas family members with a heterozygous mutation had about 50% of protease activity and were clinically asymptomatic .
The ADAMTS-13 gene spans approximately 37 kb, contains 29 exons and encodes a precursor polypeptide composed of 1427 amino acid residues [42,43] (Fig. 2). The polypeptide consists of a signal peptide, a short propeptide with a C-terminal furin cleavage site, a catalytic domain with a typical reprolysin-type active site sequence (HEXGHXXGXXHD) coordinating a Zn2+ ion and a Ca2+ binding site motif (E83, D173, C281, D284), a disintegrin domain, a thrombospondin type 1 domain, a cysteine-rich domain, a spacer domain, seven additional thrombosponding type 1 motifs and two CUB domains (Fig. 2).
The calculated molecular mass is 145 kDa, the protein isolated from human plasma has an apparent mass of 180 kDa  and is heavily glycosylated . Northern blotting of various tissues revealed a 4.7-kb mRNA transcript in liver tissue [41,42] and in situ hybridization showed that ADAMTS-13 was mainly expressed in the perisinusoidal cells of the liver . Low expression of ADAMTS-13 mRNA was found in several other organs  and mRNA was recently detected in platelets . In addition, a shorter 2.4-kb mRNA transcript was isolated from placenta, skeletal muscle and tumor cell lines .
Recombinant ADAMTS-13 has been transiently expressed in mammalian cell lines [46,48–50]. Secreted ADAMTS-13 has the propeptide cleaved off, is functionally active  and when added to congenitally ADAMTS-13-deficient plasma, it dose-dependently restored its VWF-cleaving protease activity .
ADAMTS-13 purified from normal plasma showed a series of bands with apparent Mr of 180, 170, 160 and 120 kDa , all having an identical N-terminal sequence, thus being obviously truncated at various distances from the C-terminus. Similar C-terminally truncated species have been found in supernatants but not in lysates of transiently transfected mammalian cell lines  suggesting that they resulted from proteolysis upon secretion.
Recombinant ADAMTS-13 recovered from conditioned medium of transfected cells is an active enzyme that does not need further activation. Regulation of VWF cleavage may be provided by the substrate VWF itself. In vivo, shear stress as observed in the microcirculation may be needed to make the cleavage site in the A2 domain of the VWF subunit accessible to ADAMTS-13 [21,52], whereas in vitro partial unfolding of the VWF substrate by guanidine-HCl  or urea and low ionic strength  similarly renders VWF accessible and cleavable by ADAMTS-13. rADAMTS-13 obtained from lysed transfected cells shows only minimal VWF-cleaving activity . Whether this is due to the cell-lysing procedure, to incomplete glycosylation or to intracellular inhibitor(s) is not known. Recombinant as well as plasma derived ADAMTS-13 specifically cleave recombinant or plasma-derived VWF at Tyr1605–Met1606 [44,46], and rADAMTS-13 or plasma-derived ADAMTS-13 incubated with rVWF results in the generation of the typical triplet pattern as seen in plasma VWF multimers [44,53]. Besides VWF, no other protein substrate of ADAMTS-13 has been identified [20,50]. Metal ion chelators, such as ethylenediaminetetraacetic acid (EDTA), inhibit ADAMTS-13 activity [20,21] and very recently, free hemoglobin was shown to inhibit plasma ADAMTS-13 activity .
Zheng et al.  and Soejima et al.  constructed rADAMTS-13 deletion mutants by progressively truncating the C-terminal region. Truncation of the CUB domains and the thrombospondin type 1 repeats 2–8 led to the slight loss of VWF-cp activity in a static in vitro system. Activity was (almost) completely lost by further truncation of the spacer domain. On the contrary, it was shown that rADAMTS-13 synthesized in a furin-deficient cell line and secreted with the propeptide attached had normal VWF-cp activity . In vivo, the thrombospondin type 1 repeats 2–8 and the CUB domains may still be important for proper interaction with VWF .
Dong et al.  showed that cultured endothelial cells, upon stimulation with histamine, secreted extremely long VWF multimer strands that remained attached to the endothelial cells, at least partly in a P-selectin-dependent manner . If these stimulated endothelial cells were perfused with platelets in buffer, platelets attached to the ultra-large VWF and formed long ‘beads-on-a-string’ structures. Further perfusion with normal plasma or partially purified ADAMTS-13 rapidly detached the adhering platelets by cleaving endothelium-anchored ULVWF. Perfusion with plasma from patients with severe constitutional or acquired ADAMTS-13 deficiency, however, was unable to cleave VWF and to detach the platelets . Further studies suggested that ADAMTS-13 binds to the A1 and A3 domains of VWF  enabling the protease to cleave the peptide bond in the near A2 domain becoming accessible under tensile forces. These findings suggest that ADAMTS-13 regulates the size of VWF multimers by its capacity to cleave VWF upon its secretion in the form of ULVWF on the endothelial surface .
Several assays for assessing ADAMTS-13 activity have been developed (for reviews, see Refs [60,61]). They are based on the degradation of purified, plasma-derived or recombinant VWF multimers by patient plasma and (i) measuring the disappearance of larger VWF multimers using sodium dodecyl sulfate (SDS)-agarose gel electrophoresis and immunoblotting [20,25], (ii) assaying the generation of disulfide-linked homodimers (Mr 350 kDa) of C-terminal VWF proteolytic fragments [21,26], (iii) quantitating the residual collagen-binding activity  or (iv) quantitating the ristocetin cofactor activity  of degraded VWF. A two-site immunoradiometric assay using monoclonal antibodies against C- and N-terminal VWF epitopes has also been reported . In all these assays, the cleavage site at Tyr1605–Met1606 in VWF multimers is made accessible to ADAMTS-13 by partial VWF unfolding using either guanidine-HCl  or 1.5 m urea and low ionic strength ; and barium  or calcium ions  are used to ‘activate’ ADAMTS-13.
Because of major vigorous debates about the suitability of these various methods, we initiated a multicenter comparison involving several laboratories performing different assay techniques on identical aliquots of 30 plasma samples with varying ADAMTS-13 activity levels . This evaluation showed a generally good agreement between the five participating laboratories concerning detection of severe ADAMTS-13 deficiency (<5% of the activity in pooled normal plasma), whereby the convenient but very delicate collagen-binding assay yielded some erroneous results (two false-positive diagnoses of severe deficiency and one failure to detect a severe deficiency) and the interlaboratory agreement on samples with slightly reduced or normal activity values was less good . A larger comparative exercise involving 10 expert laboratories and 11 methods  essentially confirmed and extended these results.
Inhibitors of ADAMTS-13 are measured with all reported methods by pre-incubating (heat-inactivated ) patient plasma with pooled normal plasma for 2 h and measuring the residual VWF-cleaving activity of normal plasma [65,67]. Our multicenter study showed good interlaboratory agreement for the detection of strong inhibitors (more than 1 BU mL−1), whereas there was considerable disagreement for low-titer inhibitors . It is evident that this mixing technique allows only the detection of free IgG inhibitors in plasma that have not been bound to ADAMTS-13 and is – therefore – not very sensitive.
A probably more physiologic method for measuring ADAMTS-13 activity was proposed by Dong et al. [57,68]. Based on the earlier observations by Tsai et al.  that shear stress enhances the proteolysis of VWF, they assay the ability of plasma to detach platelets adhering to ULVWF strings on stimulated endothelial cell cultures using a parallel plate perfusion system . This method has the conceptual advantage of testing not only the proteolytic activity of ADAMTS-13 but also its ability to attach to VWF and/or the endothelial cells under flow conditions which mirrors the physiologic function of ADAMTS-13 better than the mere cleavage of partially denatured VWF multimers under static conditions. Nevertheless, this assay is very complex and in the above-mentioned multicenter study  the reproducibility of this assay was not perfect: only seven of 10 replicate plasma samples with severe ADAMTS-13 deficiency were correctly identified, whereas three of 10 replicate samples with 40% ADAMTS-13 activity were judged to show a severe deficiency .
To improve static assays of ADAMTS-13 activity, recombinant VWF A1-A2-A3  or rVWF A2 domain  have been proposed as substrates. Kokame et al.  expressed a series of recombinant VWF peptides containing the cleavage site 1605Tyr–1606Met and identified a peptide containing 73 amino acids from 1596Asp–1668Arg as the minimal substrate required for efficient cleavage by ADAMTS-13. Adding N- and C-terminal glutathione S-transferase and histidine tags, respectively, substrate purification and detection of substrate as well as cleavage product by SDS-polyacrylamide gel electrophoresis (PAGE) is greatly facilitated . This GST-VWF73-H peptide allows sensitive measurement of plasma ADAMTS-13 activity of only 3% of normal plasma  and by introducing fluorescent probes on both sites of the cleavage site, a convenient fluorescence resonance energy transfer (FRET) assay has been established that will greatly facilitate and accelerate ADAMTS-13 activity measurement and provide an assay suitable for the routine laboratory .
In addition, methods for measuring ADAMTS-13 antigen (F. Scheiflinger unpubl. obs.) and for detecting autoantibodies directed against ADAMTS-13  using enzyme-linked immunosorbent assays (ELISAs) have been developed. It is noteworthy that in at least one reported patient with acquired TTP and severe ADAMTS-13 deficiency, a high-titer IgG autoantibody toward ADAMTS-13 was found by ELISA, whereas functional assays did not reveal an inhibitor of VWF-cleaving protease in this plasma . This suggests that in some cases non-inhibiting autoantibodies may result in severe ADAMTS-13 deficiency, e.g. by accelerating the clearance of the enzyme.
In 1978, Upshaw  concluded that his patient, a young woman with recurring episodes of severe microangiopathic hemolytic anemia and thrombocytopenia was congenitally deficient in a plasma factor that protected from intravascular hemolysis and thrombocytopenia. This plasma factor was identified in 1997 as VWF-cleaving protease in two brothers with chronic relapsing TTP . Since then, several cases of constitutional TTP because of severely deficient ADAMTS-13 activity have been reported [67,74–83]. Reviewing the clinical courses in 23 patients from 16 families with severe constitutional VWF-cleaving protease deficiency identified until 2001 in our laboratory revealed a striking age-dependent clustering of disease onset : about half of the patients had their first bout of TTP between the neonatal period and 5 years of age, whereas the other half had disease onset in adulthood, between 20 and 41 years of age, and two male subjects diagnosed in the frame of a family investigation were still asymptomatic at age 37 and 44 years, respectively. Frequent relapses were noted in patients with early or late onset, once a first bout of TTP had occurred. The diagnosis of TTP was often greatly delayed and several siblings of affected patients had died ; we have recently reported a fatal outcome in an 8-year-old boy where the diagnosis of TTP was made only at autopsy . Schneppenheim et al.  made similar clinical observations, several of their pediatric patients had been misdiagnosed as having atypical, Coombs-negative Evans’ syndrome or immune thrombocytopenic purpura. Some patients with severe constitutional VWF-cleaving protease deficiency had been clinically diagnosed as having HUS because of prominent renal failure [69,84]. In fact, the index patient (brother A1), completely lacking VWF-cleaving protease activity also showed transient severe renal failure at his first attack . Other patients with severe hereditary protease deficiency suffer mainly from cerebrovascular ischemia , and others have mainly recurring hemolytic anemia and thrombocytopenia. It should also be noted that in earlier reports, published at a time when ADAMTS-13 was still unknown, TTP and HUS were described to coexist in siblings [85,86]. This stresses the fact that, on clinical grounds, TTP and HUS cannot be clearly separated .
In a landmark study, Levy et al.  identified 12 different mutations in the ADAMTS-13 gene accounting for 14 of 15 disease alleles in their families with hereditary TTP. This study convincingly related hereditary TTP to the ADAMTS-13 gene. As of January 2005, a total of 75 candidate mutations in patients with constitutional TTP have been described [43,48,51,53,54,87–101; J. A. Kremer Hovinga, B. Lamoule unpubl. data] (Fig. 2). Affected patients are double heterozygous or homozygous carriers of mutated alleles, the heterozygous parents being consistently asymptomatic. Only about one-third of the reported mutations have been expressed in mammalian cells [48,92–94,102]. Most of the mutated proteins are not secreted [48,92–94,102], others are secreted but show deficient VWF-cleaving activity [92,94].
So far, no correlation between genotype and phenotype, i.e. disease severity, organ tropism and age at disease onset is evident. Other disease-modifying genetic or environmental triggering factors may be needed to result in clinical disease. In this context, we became aware of two pairs of sisters from two families who suffered their first bout of hereditary TTP during pregnancy, whereas their protease-deficient brothers were asymptomatic at age >35 years . It may be hypothesized that increased VWF synthesis during pregnancy, a known risk factor for TTP [103,104], may be such a triggering factor.
It is generally assumed that hereditary TTP is extremely rare. Nevertheless, we believe that this disease may have been greatly underestimated and misdiagnosed. A single nucleotide transition in the ADAMTS-13 gene leading to a Pro475Ser substitution has an allele frequency of approximately 5% in the Japanese population . Expression of this mutant showed normal secretion of an obviously dysfunctional protease, the activity in a static assay being about 5% as compared with wild-type ADAMTS-13 . Whether homozygous carriers of this mutation are at risk for TTP or whether the presumably quite low ADAMTS-13 activity level is completely asymptomatic has not yet been reported. A disease-associated mutation, however, a single nucleotide insertion in exon 29, 4143insA, was identified in four unrelated patients from Germany , in two brothers from Sweden  and one patient from Australia . Another seven patients have been recently detected and preliminary haplotype analysis suggests a common founder mutation, possibly arising in an ancestor near the Baltic sea region . Because several apparently not consanguineous patients with a homozygous 4143insA mutation have been identified, it is likely that many still undiagnosed homozygous carriers, either patients with undiagnosed constitutional TTP or high risk candidates to suffer from TTP, may exist.
It is of utmost importance that pediatricians, internists, nephrologists and hematologists be aware of hereditary TTP because efficient treatment and prophylactic measures are available (see below).
Patients with acute acquired TTP, like those with an acute bout of hereditary TTP, are usually severely ill, most of them are not suffering from a pre-existing underlying disease. Because of the high fatality rate in untreated patients [5,9], diagnosis is urgent, but may be difficult if the patient does not present the complete pentad of clinical and laboratory findings (see above). Schistocytic hemolysis and thrombocytopenia, not explained by another condition, may allow a tentative diagnosis even though these are not highly specific diagnostic criteria .
Following the reports in 1998, that 20 of 24  and all of 37  patients with acute acquired TTP had severe VWF-cleaving protease deficiency, most often associated with a circulating IgG inhibitor, autoantibody-mediated VWF-cleaving protease deficiency became a candidate laboratory criterion for the diagnosis of idiopathic TTP. Early critical reports suggested that VWF-cleaving protease deficiency was not specific for TTP, but was also found in disseminated intravascular coagulation (DIC) , in thrombocytopenic disorders different from TTP , in various acute inflammatory conditions, liver cirrhosis, uremia, during later stages of pregnancy, in newborns , and even in some healthy controls . However, in these studies, VWF-cp activity was either not rigorously quantitated or was only moderately decreased. In a study on 68 patients with thrombocytopenia from various causes except TTP or HUS, 12 had values lower than 30%, but none <10%, in clear distinction to patients with acute TTP . This suggests that a severely deficient activity of VWF-cp (<5% of normal) is specific for TTP. Nevertheless, subsequent cohort studies on patients diagnosed with acute (idiopathic) TTP showed that only about 33–100% had a severe protease deficiency [25,26,30,96,110–114] (Table 1). Obviously, not all patients clinically diagnosed with idiopathic TTP have a severe ADAMTS-13 deficiency as measured with current static protease activity assays.
Table 1. Proportion of patients with severely deficient ADAMTS-13 activity (defined as <5% of normal in all studies, except <10% in ) in reported TTP case series
Design of study
Denominators refer to: *patients classified as having acute TTP; †patients classified as having acute idiopathic TTP; ‡patients with first attack or relapse of acute idiopathic TTP.
An ADAMTS-13 epitope mapping of autoantibodies from 25 patients with acute TTP, severe or borderline to severe ADAMTS-13 deficiency and protease inhibiting autoantibodies showed that all 25 patients had antibodies reacting with the Cys-rich/spacer domain, 16 of 25 plasmas reacted with the CUB1+2 domains, 14 of 25 with the first thrombospondin type 1 domain, 14 of 25 with the catalytic/disintegrin/thrombospondin type 1/1 domains, seven of 25 with the thrombospondin type 1/2-8 domains, and five of 25 recognized the propeptide . This shows that autoantibodies toward different antigenic regions are present in acute TTP. It seems likely that antibodies directed toward the Cys-rich/spacer domain account for the inhibition of ADAMTS-13 activity in static in vitro assays, but it is conceivable that antibodies toward other antigenic sites may impair the ADAMTS-13 interaction with endothelial cell-anchored ULVWF in vivo and that such antibodies may lead to clinical manifestions of TTP. Alternatively, other pathophysiologic mechanisms not involving ADAMTS-13-VWF interaction may lead to a syndrome clinically indistinguishable from that associated with severe acquired ADAMTS-13 deficiency.
In the Oklahoma TTP-HUS registry, 142 consecutive patients referred for plasma exchange treatment for suspected TTP-HUS between November 1996 and December 2001 had ADAMTS-13 assays performed on their stored admission serum samples . Of 48 patients with idiopathic TTP, 16 had severe ADAMTS-13 deficiency, whereas 32 had moderately decreased, subnormal or normal protease values. The presenting clinical and laboratory findings were not different among the two groups, with the exception of acute renal failure being more frequent in those without severe deficiency (11/32) than in severely deficient patients (1/16). Clinical outcome, i.e. response to plasma exchange therapy, exacerbation, number of necessary plasma exchanges and death rate was not different among patients with and without lacking protease. Relapse rate in surviving patients, however, was significantly higher in patients with severe acquired ADAMTS-13 deficiency than in those without (6/14 vs. 2/25) . Raife et al.  reported higher median creatinine levels, less severe thrombocytopenia, and lower relapse rate in non-severely ADAMTS-13 deficient when compared with severely deficient patients with acute TMA.
The relapse rate in survivors of TTP is high (up to 36%), recurrences may occur many years after a first attack , but are most frequent during the first year after disease onset . Patients with severe acquired ADAMTS-13 deficiency usually lose their circulating inhibitors and normalize their ADAMTS-13 activity after obtaining a clinical remission [24,26], but some patients achieve remission despite remaining severely deficient [25,114,117,118]. Zheng et al.  found that patients with high titer inhibitors at presentation usually remained severely deficient in ADAMTS-13 activity in early remission and had a high relapse risk. On the contrary, we followed up a patient who was splenectomized because of plasma-refractory TTP due to autoantibody-mediated severe ADAMTS-13 deficiency . Three years after splenectomy, a high titer autoantibody leading to complete disappearance of protease activity reappeared, the patient nevertheless remaining asymptomatic for more than 32 months.
Patients with TMA after hematopoietic stem cell transplantation consistently had measurable ADAMTS-13 activity [30,112,120–122] as did most patients with neoplasia- or anticancer treatment-associated TMA [112,123]. TMA induced by ticlopidine  and possibly by clopidogrel , on the contrary, was reported to be associated with autoantibody-mediated severe ADAMTS-13 deficiency, as were several – though not all – cases of postpartum- or pregnancy-associated TMA [11,30,114]. D+ HUS was generally not associated with severe ADAMTS-13 deficiency [126,127] whereby one of 29 children in the latter study had a transiently lacking protease activity caused by an autoantibody . None of 120 plasma samples from patients having been diagnosed with HUS, referred to our laboratory between January 2001 and July 2003, had a severe ADAMTS-13 deficiency .
The annual incidence of TTP has been estimated to be 3.7 per million population . The Oklahoma TTP-HUS registry has recently calculated age-, sex-, and race-standardized incidence rates for all patients with a TTP-HUS syndrome referred for plasma exchange therapy (thus excluding childhood D+ HUS), for idiopathic TTP, and for those with severe ADAMTS-13 deficiency, the figures being 11.3, 4.5 and 1.7/106 per annum, respectively .
Treatment of TTP
Treatment of TTP has been reviewed by several authors [9,12,33,130] and guidelines have been proposed by the British Committee for the Standardization in Haematology  even though levels of evidence on which recommendations are based are mostly weak, i.e. level IV. Still, the dramatic improvement of survival from approximately 10% to 80% by the empirical introduction of plasma exchange and FFP replacement in the 1970s  certainly forbids to undertake a randomized study with a control group not receiving plasma therapy. Moreover, the Canadian Apheresis Study Group , in a prospective controlled trial on 102 patients demonstrated that plasma exchange of 1.5 plasma volumes daily for 3 days, followed by exchanges of 1 volume daily, and FFP replacement was superior to simple FFP infusion, response rates after the first treatment cycle and at 6 months being 47% and 78% vs. 25% and 49%, respectively, and mortality being 22% vs. 37%. Corticosteroids have been frequently used in addition, and sometimes alone in milder cases  and were often found to be useful. In light of the recent findings that many patients with acute TTP have severe autoantibody-induced ADAMTS-13 deficiency, a pathophysiological basis for these approaches seems to be at hand: plasma exchange may remove autoantibodies, FFP replaces the lacking protease and corticosteroids may suppress autoantibody formation. Cryosupernatant, lacking the larger plasma VWF multimers, has been used as replacement fluid instead of plasma and seemed to be more efficacious if compared with a historical control group in which FFP was used . However, this has not been substantiated in subsequent randomized studies [133,134].
Plasma-refractory TTP patients or relapsing patients are often treated with more intensive plasma exchange regimens, e.g. twice daily. Splenectomy, best performed in remission after relapse, seemed to be often effective [135–137], its beneficial effect may mainly relate to elimination of autoantibody-producing B cells [24,119], even though other mechanisms may underlie its beneficial effects . Recently, several patients with relapsing TTP caused by autoimmune-mediated ADAMTS-13 deficiency have been treated with rituximab, a monoclonal anti-CD20 antibody, and successful short-term outcomes have been reported [138–143]. Whether there is any long-term advantage over corticosteroids, other immunosuppressive treatment or splenectomy, needs to be tested in prospective trials.
There has been controversy as to whether plasma exchange therapy would be indicated for those patients with clinically diagnosed TTP not having autoantibody-mediated severe ADAMTS-13 deficiency. A small retrospective study from Japan  showed that 10 of 12 TTP patients with severe acquired ADAMTS-13 deficiency survived, whereas four of six without severe deficiency died despite plasma exchange treatment. In contrast, among the 48 patients with idiopathic TTP reported by Vesely et al. , three of 16 patients with and seven of 32 without severe acquired ADAMTS-13 deficiency died of their disease, thus plasma exchange seemed to be effective also in the latter patients. Zheng et al. , based on their experience with 37 patients, also questioned the indication for plasma exchange in patients without severe ADAMTS-13 deficiency. However, our interpretation of their data is that outcome was rather related to the clinical diagnosis than to ADAMTS-13 levels: 10 of 17 patients with secondary TMA (many associated with hematopoietic stem cell transplantation, neoplasia or anticancer agents), all having normal or mildly reduced ADAMTS-13 activity, died whereas only three of 20 patients with idiopathic TTP succumbed, notably three of 16 with severe ADAMTS-13 deficiency but none of four without.
Thus, based on present knowledge, plasma exchange therapy with FFP replacement clearly remains mandatory, at least for all patients with idiopathic TTP, also in the absence of severe acquired ADAMTS-13 deficiency.
Constitutional TTP caused by homozygous or double heterozygous ADAMTS-13 gene defects may be treated by simple FFP infusion [11,34] and relapses may be prevented by regular FFP infusion every 2–3 weeks, in some patients with frequent recurrences over many years . The ADAMTS-13 has an in vivo half-life in plasma of about 2–4 days  and ADAMTS-13 activity after infusion of 1–2 U of FFP may rise to only about 10–20% of normal. It is not quite clear why the effect on platelet count will last for up to 3 weeks, because the plasma level will fall below 5% within some 3–8 days. One may speculate that ADAMTS-13 remains available, e.g. on the microvascular endothelial surface, for longer than anticipated from its plasma levels.
Diarrhea-positive HUS in children is generally treated by supportive therapy using renal dialysis as required, and plasma infusion or exchange has not been found to improve outcome . In a cohort of elderly nursing home residents with D+ HUS, plasma exchange with FFP replacement was performed in 16 of 22 patients without evidence of clear benefit . Activation of coagulation with high prothrombin fragment F1 + 2 concentrations has been reported in children with D+ HUS  and in an animal model, lepirudin prevented death from HUS in two of three dogs challenged with shiga-like toxin . Thus, thrombin inhibition may be a therapeutic strategy to be further explored in D+ HUS.
Effective treatment for hematopoietic stem cell transplantation- or neoplasia-associated TMA is unknown, plasma exchange is often used but in severe cases the mortality rate is very high with death occurring either from the underlying neoplasia or the TMA itself.
Conclusion and research agenda
Despite exciting pathophysiologic advances achieved in recent years, many questions concerning diagnosis, pathogenesis and treatment of the thrombotic microangiopathies remain to be answered.
1An important point is clarification of terminology. TTP, HUS, TTP-HUS, idiopathic TTP, secondary TTP, TTP-like disease, typical HUS (D+ HUS) and atypical HUS (D− HUS) are too many terms for partly overlapping, partly distinct, and pathophysiologically only partially clarified disease states. We propose to use the comprehensive term TMA for all these disease states with schistocytic hemolysis and consumptive thrombocytopenia and to list: (i) idiopathic TTP with hereditary severe ADAMTS-13 deficiency, (ii) idiopathic TTP with and (iii) without severe acquired ADAMTS-13 deficiency as distinct entities (the latter as provisional category, until its pathophysiology is better characterized). D+ HUS caused by enterohemorrhagic E. coli infection should be considered a separate entity as well, and the role of factor H deficiency or mutations [148–152] as causal for atypical HUS should be further investigated. It is probably justified to provisionally list other disease states as TMA associated with the underlying condition, such as neoplasia, anticancer agents, hematopoietic stem cell transplantation, until their pathogenesis will be unravelled.
2Quantitation of red cell fragmentation  should be more widely introduced in the clinical laboratory. Some schistocytes are seen in many conditions such as DIC, sepsis and others, usually in much lower numbers than in clear-cut TTP or HUS.
3It would be extremely useful to reinvestigate the histological patterns of microvascular thrombi in several forms of TMA. These microthrombi are said to consist mainly of platelets and VWF in TTP [2,3], but rather of fibrin in D+ HUS [3,126]. If confirmed, the presumed difference in pathogenesis will be further substantiated, and additional histopathologic studies may further our insight into the pathophysiology of other TMA, e.g. idiopathic TTP without severe ADAMTS-13 deficiency.
4Similarly, it will be interesting to evaluate whether exaggerated thrombin generation is specific for D+ HUS  and whether activation markers of plasmatic coagulation and fibrinolysis can help to distinguish other forms of TMA. It should be recalled, that elevated coagulation activation markers have also been reported in TTP .
5An important topic will be to find out whether idiopathic TTP without severe ADAMTS-13 deficiency as measured by static assays is still related to an altered ADAMTS-13-VWF interaction in vivo. Further structure-function studies of ADAMTS-13 and search for possible autoantibodies not inhibiting ADAMTS-13 protease activity in traditional assays but precluding its binding to endothelium-anchored ULVWF are needed.
6Cofactors for triggering disease onset in severe hereditary or acquired ADAMTS-13 deficiency should be investigated to clarify why some patients with severely deficient protease activity consistently relapse every few weeks , whereas others remain asymptomatic into their forties.
7The true frequency of hereditary TTP needs to be re-evaluated, this condition having been largely ignored or misdiagnosed. A registry of such patients will certainly be helpful to sensitize physicians and improve the health of affected patients.
8Having identified twin sisters who developed autoantibody-mediated ADAMTS-13 deficiency and TTP 1 year apart  the question arises as to whether a genetic predisposition to ADAMTS-13 autoimmunity can be identified.
9The role of persisting or recurring acquired ADAMTS-13 deficiency in survivors of acute TTP as a prognostic marker needs to be addressed, because it may allow to prevent relapses at an asymptomatic state using immunosuppressive treatment.
10Finally, the role of ADAMTS-13 in platelets  has to be investigated, both in terms of its physiological role and with respect as to whether a platelet reservoir of ADAMTS-13 may be protective even in the absence of any plasma ADAMTS-13 activity.
11TTP and the other TMA have attracted and continue to attract many researchers. Many controversies have been raised and debates are likely to go on. It is hoped that continuing research into the pathophysiology and treatment of the TMA will further benefit affected patients who are still at high risk of premature and often preventable death.
We dedicate this paper to our friend, Professor Miha Furlan, who until his retirement in 2000 guided the research project on von Willebrand factor-cleaving protease at our institute, made most important contributions to the field, and still follows with great interest and helpful advice our continuing research. Irmela Sulzer is acknowledged for technical assistance.
Studies from our laboratory are supported by the Swiss National Science Foundation (grant no. 32-66756.01 and 32-65337.01), the Fondation pour la Recherche sur l’ Arteriosclérose et la Thrombose, and by the Mach-Gaensslen Foundation Switzerland.