Recent Advances in Thrombotic Thrombocytopenic Purpura

Joel L. Moake, MD^*

Thrombotic thrombocytopenia purpura (TTP) is a severe microvascular occlusive “thrombotic microangiopathy” characterized by systemic platelet aggregation, organ ischemia, profound thrombocytopenia (with increased marrow megakaryocytes), and fragmentation of erythrocytes.¹ The RBC fragmentation occurs, presumably, as blood flows through turbulent areas of the microcirculation partially occluded by platelet aggregates. Schistocytes, or “split” red cells, appear on the peripheral blood smear (> 1% of total RBCs) as an indication of “microangiopathic hemolytic anemia.” Serum levels of lactate dehydrogenase (LDH) are extremely elevated as a consequence of hemolysis and the leakage of LDH from ischemic or necrotic tissue cells.

The systemic platelet clumping in TTP is often associated with platelets below 20,000/μL. Occlusive ischemia of the brain or the gastrointestinal tract is common, and renal dysfunction may occur. A “pentad” of signs and symptoms was long associated with TTP: thrombocytopenia; microangiopathic hemolytic anemia; neurologic abnormalities; renal failure; and fever. In current clinical practice, thrombocytopenia, schistocytosis, and an impressively elevated serum LDH value are sufficient to suggest the diagnosis.¹ Clotting studies are usually normal.

A clinically similar thrombotic microangiopathy can occur weeks to months after exposure to mitomycin C; inhibitors of the Ca²⁺-activated phosphatase, calcineurin (cyclosporine or tacrolimus [FK 506]); quinine; combinations of chemotherapeutic agents; total-body irradiation; or allogeneic bone marrow, kidney, liver, heart, or lung transplant.¹ The microvascular thrombi may be either predominantly renal or systemic in these heterogenous entities. Other thrombotic microangiopathies include the hemolytic-uremic syndrome (HUS), which can be acquired by ingestion of enterohemorrhagic bacteria that produce Shiga toxin (e.g., Escherichia coli of serotype 0157:H7).¹ Familial types of HUS are caused by a regulatory protein defect in the alternative complement pathway (plasma factor H or membrane-cofactor protein [CD46]) or defective intracellular cobalamin reduction/cofactor function. The discussion to follow is limited to TTP associated with deficient plasma von Willebrand factor (VWF)–cleaving metalloprotease (ADAMTS13) (Table 1).

Familial TTP is rare, usually (but not always) appears initially in infancy or childhood, and may recur as “chronic relapsing TTP” episodes at about 3-week intervals.^1–,3 Acquired idiopathic (“out-of-the-blue”) TTP has become a commonly recognized disorder that occurs in adults and older children^1,2 and, following successful treatment, recurs at irregular intervals in 11%–36% of patients. A small fraction of patients treated for arterial thrombosis with the platelet adenosine diphosphate receptor-inhibiting thienopyridine drugs, ticlopidine (Ticlid) or clopidogrel (Plavix) develop TTP within a few weeks after the initiation or therapy.¹ TTP occurs occasionally late in pregnancy or immediately after delivery.

In 1982, “unusually large” (UL) VWF multimers found in plasma samples taken repeatedly from 4 patients with chronic relapsing TTP were proposed as the systemic agglutinating agents.² The patients described were believed to have defective ULVWF multimeric “processing.”² Convincing evidence for ULVWF multimers as the cause of platelet clumping in TTP has accumulated in the subsequent 22 years.¹ 

Brief comments on VWF biochemistry and physiology are included here for orientation. Monomers of VWF (280,000 daltons) are linked by disulfide bonds into multimers with varying molecular masses that range into the millions of daltons. Multimers of VWF are constructed within megakaryocytes and endothelial cells, and stored within platelet α-granules and endothelial cell Weibel-Palade bodies. Most plasma VWF multimers are derived from endothelial cells. Both endothelial cells and platelets produce VWF multimers larger than the multimers in normal plasma. These ULVWF multimers bind more efficiently than the largest plasma VWF multimers to the glycoprotein (GP) Ibα components of platelet GPIb-IX-V receptors.^4,5 The initial attachment of ULVWF multimers to GPIbα receptors, and subsequently to activated platelet GPIIb-IIIa complexes, induces platelet adhesion and aggregation in vitro in the presence of elevated levels of fluid shear stress.⁴ After retrograde secretion by endothelial cells, ULVWF multimers become entangled in subendothelial collagen, thereby maximizing the VWF-mediated adhesion of blood platelets to any subendothelium exposed by vascular damage and endothelial cell desquamation. An efficient “processing activity”^2,6 in normal plasma prevents the highly adhesive ULVWF multimers that are also secreted antegrade into the vessel lumen from persisting in the bloodstream.

The ULVWF “processing activity” is now known to be a specific VWF-cleaving metalloprotease present in normal plasma.^3,7,8 The enzyme degrades ULVWF multimers by cleaving 842Tyr-843Met peptide bonds in susceptible A2 domains of VWF monomeric subunits. The VWF-cleaving metalloprotease is number 13 in a family of 19 distinct ADAMTS-type enzymes identified to date. The ADAMTS enzymes are numbered 1–20, but ADAMTS5 and ADAMTS11 have been found to be identical. ADAMTS13 is a disintegrin and metalloprotease with eight thrombospondin-1-like domains composed (Figure 1) of an amino-terminal reprolysin-type metalloprotease domain followed by a disintegrin domain; a thrombospondin-1-like domain; a cysteine-rich domain containing an arginine-glycine-aspartate (RGD) sequence and an adjacent spacer portion; 7 additional thrombospondin-1-like domains; and 2 similar CUB domains at the carboxyl-terminal end of the molecule. CUB domains, found only in ADAMTS13 among the ADAMTS enzymes, contain peptide sequences present in Complement subcomponents C1r/C1s; embryonic sea Urchin protein egf; and Bone morphogenic protein-1. ADAMTS13 is a Zn²⁺- and Ca²⁺-requiring 190,000 dalton glycosylated protein that is encoded on chromosome 9q34 and produced predominantly in the liver. ADAMTS13 activity is inhibited in vitro by ethylenediaminetetraacetic acid (EDTA) and, therefore, functional assays of the enzyme are usually performed using plasma anticoagulated with citrate.^3,7– 11 Plasma anticoagulated with heparin, D-phenylalanylprolylarginyl chloromethyl ketone (PPACK), hirudin and other direct thrombin inhibitors, or even serum, may also be satisfactory for testing.

ULVWF multimers are cleaved by ADAMTS13 as they are secreted in long “strings” from stimulated endothelial cells^12,13,(Figure 2A). The ULVWF multimeric strings may be anchored in the endothelial cell membrane to P-selectin molecules that are secreted concurrently with the ULVWF multimers from Weibel-Palade bodies.¹⁴ Endothelial cells are stimulated to secrete ULVWF multimers by histamine, Shiga toxin, tumor necrosis factor-alpha (TNFα), interleukin (IL)-8 and IL-6 in complex with IL-6 receptor.¹⁵ CUB domains, as well as one or more of the thrombospondin-1-like domains, may be involved in binding the ADAMTS13 to ULVWF multimers as they are secreted by the endothelial cells. Specifically, ADAMTS13 enzymes may attach under flowing conditions to accessible A3 domains in the monomeric subunits of ULVWF multimers,¹³ and then cleave Tyr 842–843 Met peptide bonds in adjacent A2 domains (Figure 2B ). Partial unfolding of emerging ULVWF multimers by fluid shear stress may increase the efficiency of ADAMTS13 attachment to, or cleavage of, ULVWF multimers.¹² 

Failure to degrade ULVWF multimers has long been suspected to cause the familial and acquired idiopathic types of TTP or predispose an individual to these disorders^1,2 (Figure 2C ). Critical experiments verifying this concept were reported in 1997–1998. In 1997, Furlan et al³ described 4 patients with chronic relapsing TTP who had a deficiency of VWF-cleaving protease activity in plasma. Because no inhibitor of the enzyme was detected, the deficiency was ascribed to an abnormality in the production, survival or function of the protease. The following year, elegant papers by Furlan et al⁷ and Tsai and Lian⁸ elucidated the pathogenesis of the more common acquired idiopathic type of TTP. The acquired idiopathic patients had little if any plasma VWF-cleaving protease activity during acute episodes; however, the activity returned to normal upon recovery. An IgG autoantibody produced transiently against the enzyme accounted for the lack of protease activity during the acquired idiopathic TTP episodes.⁸ 

Patients with familial, chronic relapsing TTP almost always have ULVWF multimers in their plasma^1,2 because they chronically have less than about 5% of normal plasma ADAMTS13—unless they have recently received plasma infusions. ULVWF multimers are also detectable in some patient plasma samples during acute episodes of acquired idiopathic TTP, but not after recovery,¹ due to transient inhibition of ADAMTS13 to less than about 5% of normal only during acute TTP episodes.^1,3,7,8,16 Severe deficiency of ADAMTS13 activity in TTP patient plasma often correlates with a failure to cleave ULVWF multimers as they emerge in long strings from the surface of stimulated endothelial cells¹² (Figure 2C).

ULVWF multimeric strings are anchored to the endothelial cells^12,13 via P-selectin molecules, which have transmembrane domains and are secreted along with ULVWF multimers from Weibel-Palade bodies.¹⁴ P-selectin molecules are predominantly retained in the endothelial cell membrane. Passing platelets adhere via their GPIbα receptors to the long ULVWF multimeric strings anchored to P-selectin.¹² (Platelets do not adhere to the smaller VWF forms that circulate after cleavage of ULVWF multimers, perhaps because the binding site for platelet GPIbα in VWF A1 domains is not exposed on smaller VWF forms.⁵) Many additional platelets subsequently aggregate under flowing conditions, probably via their activated IIb-IIIa complexes, onto the ULVWF multimeric strings to form large, potentially occlusive, platelet thrombi.

ULVWF multimeric strings are capable of detaching from endothelial cells in the absence of ADAMTS-13 activity, the presence of fluid shear stress, and the increasing torque generated by ULVWF-platelet adherence.^12,13 The detached ULVWF-platelet strings may “embolize” to microvessels downstream and contribute to organ ischemia. The formation of ULVWF-platelet thrombi and emboli may account for the detection of ULVWF multimers in the plasma of familial TTP patients chronically and of acquired idiopathic TTP patients during acute episodes.^1,2,6 Increased VWF antigen has been found by flow cytometry on platelets during episodes of familial or acquired TTP, and abundant VWF antigen (but not fibrinogen) has been observed by immunohistochemistry on platelet occlusive lesions in TTP.¹ 

The absent or severely reduced plasma ADAMTS13 activity in familial TTP^1,3,7 is usually a consequence of homozygous (or double heterozygous) mutations in both of the ADAMTS13 alleles located at chromosome 9q34.¹⁰ Mutations in familial TTP have been detected all along the gene, in regions encoding different domains.¹⁰ In severe familial deficiency of ADAMTS13 activity, episodes of TTP usually commence in infancy or childhood (Table 1). In others, however, overt TTP episodes do not develop for years (e.g., during a first pregnancy), if ever. These latter observations suggest that (1) in vivo ADAMTS13 activity on ULVWF multimers emerging from stimulated endothelial cells may exceed the estimates of plasma enzyme activity using in vitro non-physiologic assays, and/or (2) that accentuated secretion of ULVWF multimers by endothelial cells induced by estrogen or pro-inflammatory cytokines¹⁵ is required to provoke TTP episodes in some patients with severe plasma ADAMTS13 deficiency.

Many patients with acquired idiopathic TTP also usually have absent or severely reduced plasma ADAMTS13 activity only during an initial episode or later recurrence^1,7,8,(Table 1). IgG autoantibodies that inhibit plasma ADAMTS13 activity during acquired episodes can be detected using the nonphysiologic techniques currently available in 44%–94% of patients,^1,7,8 suggesting the presence of a transient, or intermittently recurrent, defect of immune regulation. Antibodies that inhibit plasma ADAMTS13 have also been demonstrated in a few patients with ticlopidine- or clopidogrel-associated TTP.¹ It is not known if there is a transient, severe defect of metalloprotease production or survival in patients with acquired idiopathic TTP who do not have detectable autoantibodies against ADAMTS13, or if the results are explained by the limited test sensitivity.

In one recent study¹⁷ of ADAMTS-13 autoantibodies in 25 acquired TTP patients, the epitope targets always included the cysteine-rich/spacer domain (100% of autoantibodies), and were exclusively directed against the cysteine-rich/spacer domain in 3 of the 25 autoantibodies. The other 22 autoantibodies reacted with the cysteine-rich/spacer domain plus either the CUB domains (64%), the metalloprotease/disintegrin-like/1st thrombospondin-1-like domain combination (56%), or the 2nd–8th thrombospondin-1-like domain combination (28%). The propeptide region was also identified by 20% of autoantibodies,¹⁷ indicating that propeptide removal is not required for secretion of active ADAMTS13. The recent demonstration of acquired idiopathic TTP caused by IgG autoantibodies against ADAMTS13 in identical twin sisters emphasizes that production of these autoantibodies is genetically determined.¹⁸ 

Plasma ADAMTS13 activity in healthy adults ranges from about 50% to 178%. Activity is often reduced below normal in liver disease, disseminated malignancies, chronic metabolic and inflammatory conditions, pregnancy, and newborns.¹⁹ With the exception of those peri-partum women who develop overt TTP, the ADAMTS13 activity in these conditions is not reduced to the extremely low values (< 5% of normal) found in most patients with familial or acquired idiopathic TTP.

Neither bone marrow transplantation-associated thrombotic microangiopathy nor “classical” HUS is usually associated with absent or severely reduced plasma ADAMTS13 activity.^1,7 The explanation for VWF abnormalities in the plasma of a few chemotherapy/transplant-associated thrombotic microangiopathy patients is not known.

In 1977, Byrnes and Khurana²⁰ reported that relapses of TTP could be prevented or reversed by the infusion of only a few units of fresh-frozen plasma or its cryoprecipitate-poor fraction (cryosupernatant), without concurrent plasmapheresis. It was shown in 1985 that the processing of ULVWF multimers was restored in patients with familial, chronic relapsing TTP by transfusing fresh-frozen plasma or cryosupernatant.⁶ Tsai and Lian⁸ and Furlan et al⁷ demonstrated in 1998 that these plasma products, as well as solvent/detergent-treated plasma,¹ contain active ADAMTS13.

The infusion about every 3 weeks of normal ADAMTS13 into familial TTP patients producing defective ADAMTS13 molecules^3,10 is sufficient to prevent TTP episodes. The plasma t_1/2 of infused ADAMTS13 activity is about 2 days, so perhaps ADAMTS13 molecules dock and cleave one secreted ULVWF multimeric string after another over a longer time period.^1,12,13 

The sequence of ADAMTS13 has been determined, and it has been partially purified^1,9 and produced in recombinant active form²¹ for ultimate therapeutic use. Because plasma ADAMTS13 of only about 5% is often sufficient to prevent or truncate TTP episodes,^1,16 gene therapy may eventually extend remissions in familial, chronic relapsing TTP.

Adults and older children with acquired idiopathic TTP episodes associated with ADAMTS13 deficiency require daily plasma exchange. This combines plasmapheresis (which may remove circulating ULVWF multimer-platelet strings, agents that stimulate endothelial cells to secrete ULVWF multimers, and autoantibodies against ADAMTS13) and the infusion of fresh-frozen plasma or cryosupernatant (containing uninhibited ADAMTS13). Both solvent-detergent–treated plasma^1,8 and methylene blue/light-treated plasma (for inactivation of lipid envelope viruses) also contain active ADAMTS13; however, protein S activity is below normal in solvent-detergent plasma.

Plasma exchange allows about 80%–90% of acquired TTP patients to survive an episode, usually without persistent overt organ damage. Lower titers of ADAMTS13 autoantibodies are associated with better responses to plasma exchange procedures. Production of ADAMTS13 autoantibodies may be suppressed by high-dose glucocorticoids, 4–8 weekly doses of rituximab (monoclonal antibody against CD20 on B-lymphocytes) or removal of autoantibody-producing cells by splenectomy.

Toshiyuki Miyata, PhD,^* and Koichi Kokame, PhD

Thrombotic thrombocytopenic purpura is typically characterized by thrombocytopenia and microangiopathic hemolytic anemia, with variable degrees of renal failure, neurological dysfunction, and fever. In most cases the underlying mechanism of idiopathic TTP is explained by the accumulation of ULVWF multimers in the plasma, leading to the formation of platelet thrombi in microvessels. In turn, the accumulation of ULVWF multimers is caused by congenital or acquired deficiency of a metalloprotease, ADAMTS13. By 2001, this enzyme had been purified and cloned by several groups.¹ Since then, our understanding of TTP and ADAMTS13 has greatly increased.

VWF is synthesized primarily by vascular endothelial cells. In the endoplasmic reticulum, VWF dimer formation occurs through disulfide bond formation near the C-termini. The pro-VWF dimers transit to the Golgi apparatus, where the pro-sequence is cleaved and multimers form by N-terminal disulfide bond formation. ULVWF multimers, disulfide-bonded at both N- and C-terminal domains, are formed in their mature form and stored in Weibel-Palade bodies or secreted into the plasma. ULVWF multimers are highly active in collagen binding and platelet aggregation. Under normal physiological conditions, ULVWF multimers depolymerize into smaller multimers ranging in size from 500 to 20,000 kDa. Depolymerization is catalyzed by a plasma metalloprotease, ADAMTS13, which specifically cleaves the peptide bond between Y1605 and M1606 in the A2 domain of VWF. Functional deficiency of ADAMTS13, caused by genetic mutation, inhibitory autoantibodies, or other etiologies, leads to the accumulation of ULVWF multimers in plasma. These multimers promote microvascular thrombi of platelets that result in platelet consumption and hemolysis. Once microvascular thrombi are formed in the brain or kidney, patients may suffer neurological dysfunction or renal failure.

ADAMTS13 has a multidomain structure like other ADAMTS proteases and its characteristics have been described in Section I. The ADAMTS13 precursor consists of 1427 amino acids and contains a signal peptide, a short propeptide, a reprolysin-like metalloprotease domain, a disintegrin-like domain, a thrombospondin type I repeat (TSP1), a Cys-rich domain, a spacer domain, 7 additional TSP1 repeats, and 2 CUB domains.¹ The metalloprotease domain of ADAMTS13 is necessary but not sufficient for VWF cleavage, which appears to require other domains as well. A mutant with a common single nucleotide polymorphism leading to the P475S substitution in the Cys-rich domain, found in Japanese populations with an allelic frequency of about 5%, showed reduced VWF cleavage, indicating a significant role for the Cys-rich domain.² The analysis of sequentially truncated C-terminal mutants suggests that both the Cys-rich and spacer domains are indispensable for VWF-cleaving activity.^3,4 In addition, epitope analysis of ADAMTS13 autoantibodies in acquired TTP has shown that the Cys-rich and spacer domains usually are involved in antibody reactivity.^4,5 Taken together, the Cys-rich and spacer domains seem to be essential for the VWF-cleaving activity of ADAMTS13. Although removal of the propeptide sequence in ADAM family members is usually essential for their enzymatic activity, pro-ADAMTS13 exhibited normal proteolytic activity toward VWF, indicating that propeptide removal is not required for the activity of ADAMTS13.⁶ 

As of 2002, five main assays for ADAMTS13 activity were in use to study ADAMTS13 activity in thrombotic microangiopathies and other pathophysiological conditions. However, more accurate and simpler methods were needed. This year, we developed a more convenient ADAMTS13 substrate based on the amino acid sequence of the A2 domain of VWF. In this section, I will describe conventional assay methods using multimeric VWF and then introduce our novel method, which uses a recombinant monomeric protein as the substrate. The assay methods for ADAMTS13 are summarized in Table 2 .

The first assay method was developed by Furlan et al.⁷ The substrate for this method was protease-free plasma VWF. The plasma samples are first diluted, then activated by barium chloride, mixed with substrate and dialyzed overnight against low ionic strength buffer, pH 8.0, containing 1.5 M urea. The reaction products were separated by sodium dodecyl sulfate (SDS)–1.4% agarose gel electrophoresis followed by immunoblotting. The resolution of the ladders of the degraded VWF multimers was excellent and reproducible, but the method required several days to complete.

The second method was reported by Tsai.⁸ Plasma samples were incubated with guanidine HCl-treated protease-free VWF for one hour. Next, the products were separated by SDS–polyacrylamide gel electrophoresis, followed by immunoblotting for dimers of C-terminal 176-kDa fragments of VWF. An advantage of this method is that dimer formation directly reflects the cleavage of the scissile bonds.

Obert et al reported an elegant and high-throughput method,⁹ in which the plasma samples were incubated with recombinant VWF overnight and the degraded VWF fragments were detected by two-site immunoradiometric assay in microtiter plates. This test can be performed in the setting of a hospital laboratory and the cumbersome immunoblot technique is not required. In 102 healthy volunteers, the range (± 2 SD) of ADAMTS13 activity was 50%–178%, where the mean was 114%.

Gerritsen et al reported a functional assay based on the preferential binding of high molecular weight forms of VWF to collagen.¹⁰ Plasma that was treated with EDTA and dialyzed against the buffer was used as substrate. Since EDTA is a chelator of divalent cations, EDTA treatment of plasma abolishes the VWF-cleaving activity. Depolymerized forms of VWF showed impaired binding to microtiter plates coated with collagen type III, and collagen-bound VWF was quantified using a specific anti-VWF antibody. In 177 healthy volunteers, ADAMTS13 activity was widely distributed, ranging from 40% to 170%.

Böhm et al used the ristocetin-induced platelet aggregation activity to quantitate the residual VWF multimers for their ADAMTS13 assay.¹¹ This assay is based on the positive correlation between VWF multimer size and ristocetin cofactor activity of platelet aggregation. The method requires no special laboratory equipment or expertise. In 80 healthy volunteers, the ADAMTS13 activity ranged from 52% to 134%.

To evaluate the various assays for ADAMTS13 activity, a pilot multicenter comparison of the four different assays—the quantitative immunoblotting of VWF (Furlan’s method), the two-site immunoradiometric assay (Obert’s method), the residual collagen binding assay (Gerritsen’s method), and the residual ristocetin cofactor assay (Böhm’s method)—has been performed.¹² The test consisted of 30 plasma samples from patients with hereditary and acquired TTP and other conditions associated with ADAMTS13 levels ranging from < 3% to > 100%. All assays identified plasma with severe ADAMTS13 deficiency (< 5%), notwithstanding some exceptions observed using the collagen-binding assay, suggesting that all were useful to screen for suspected TTP. For samples with normal to moderately reduced activity, results were less concordant.

Coagulation proteases such as thrombin or factor Xa that are classified as “serine proteases” have a restricted substrate specificity recognizing short amino acid sequences on the N-terminal side of the scissile bonds. Based on these sequences, peptidyl substrates specific for many of these proteases were developed and utilized for activity assays. ADAMTS13 is a metalloprotease rather than a serine protease, and it cleaves a peptide bond between Y1605 and M1606 in the A2 domain of VWF. The development of peptidyl substrates for ADAMTS13 has been attempted but not achieved, suggesting that cleavage depends not only on specific residues in the close vicinity of the scissile bond, but also on more remote sequences in VWF.

We utilized a recombinant protein approach to develop an ADAMTS13 substrate, using the amino acid sequence of the A2 domain of VWF that contains the scissile bond.¹³ To identify the minimal region of VWF required for cleavage by ADAMTS13, we prepared 5 recombinant proteins, each containing a region of the A2 domain of VWF (Figure 3A ). Two criteria were set: (1) the region must contain the cleavage site by ADAMTS13, and (2) it must not contain Cys residues that often interfere with the proper folding of recombinant proteins. The longest region that meets the criteria was D1459 to R1668 of VWF. For purification and detection, the protein was flanked with N-terminal glutathione-S-transferase (GST) and C-terminal 6xHis (H) tags. This was designated GST-D1459R1668-H (substrate A). We also prepared 4 shorter recombinant proteins (substrates B-E), as shown in Figure 3A .

Five recombinant proteins were expressed in E coli and purified by affinity chromatography. The purified proteins were incubated with plasma for 1 hour at 37°C and the reaction mixture was electrophoresed on an SDS-polyacrylamide gel. The cleaved products were then detected by Western blot using anti-GST antibodies. Four out of 5 recombinant proteins were successfully cleaved with plasma ADAMTS13; however, substrate E (the shortest) was not (Figure 3B ). Therefore, the minimum substrate for ADAMTS13 was substrate D, which contains the region from D1596 to R1668 (73 amino acid residues). We designated substrates D and E as GST-VWF73-H and GST-VWF64-H, respectively. The amino acid sequence of substrate D is shown in Figure 3C . As described, substrate E was resistant to cleavage. Therefore, the region of nine amino acid residues (EAPDLVLQR) underlined in Figure 3C  was likely essential for cleavage by ADAMTS13. Mass spectrometry analysis of both substrates, VWF73-H and VWF64-H, showed that their estimated molecular weights were exact, indicating no modification. To understand the substrate specificity of ADAMTS13, we determined the solution structure of ¹H and ¹⁵N double-labeled substrates VWF73-H and VWF64-H by nuclear magnetic resonance (NMR). The results indicated an extended structure for both peptides, suggesting an induced-fit substrate recognition mechanism (unpublished observations).

It was previously reported that ADAMTS13 cleaves VWF in vitro in the presence of urea and low ionic strength conditions. In hypotonic buffer containing 5 mM Tris-HCl and 10 mM BaCl₂, GST-VWF73-H was cleaved by plasma in a time-dependent manner (Figure 3D ). The supplementation of 1.5 M urea or 150 mM NaCl reduced the cleavage, indicating low ADAMTS13 activity under these conditions. GST-VWF73-H was cleaved in a dose-dependent manner with high sensitivity. This assay was able to detect as little as 3% activity (Figure 3E ).¹³ GST-VWF73-H was not cleaved by either plasma from patients with congenital deficiency of ADAMTS13 activity due to ADAMTS13 mutations (designated as Upshaw-Schulman syndrome [USS]) or plasma from patients with acquired TTP associated with ticlopidine (Tc-TTP). ADAMTS13 activity was observed in 4 patients with idiopathic TTP, but no activity was found in 5 other patients. All patients with hemolytic uremic syndrome (HUS) exhibited activity. Most patients with acquired TTP have autoantibodies that inhibit ADAMTS13 activity in their plasma. GST-VWF73-H was also useful for the detection of autoantibody when normal plasma was incubated with plasma from the patients with acquired TTP.¹³ 

Another attempt at developing recombinant substrates has been reported in which the E coli–expressed A2 domain was used and proved to be a substrate for ADAMTS13 (Table 2 ).¹⁴ 

There are several advantages to our method. First, ours is a direct assay for measuring ADAMTS13 product generation, which should provide a more accurate analysis than indirect methods that measure substrate depletion. Second, the bacteria-expressed protein avoids the need to make protease-free VWF. The recombinant protein is simple to prepare: we can produce 1 mg of pure GST-VWF73-H, sufficient for more than 2000 assays, from 100 mL of E coli culture in 2 days. Third, the bacterial expression system is easily modified. For example, we have expressed a mouse version of GST-VWF73-H to measure the mouse ADAMTS13 activity.¹⁵ Fourth, having the substrate tagged with two different molecules makes it easy to modify the detection of product. For example, Zhou and Tsai used GST-VWF73-H to develop an ELISA system to measure ADAMTS13 activity in a 96-well format (Table 2 ).¹⁶ Finally, VWF73 will be used as the lead compound to develop more convenient and rational substrates for ADAMTS13 activity assays.

One potential disadvantage of this method is that GST-VWF73-H is not a natural substrate. A recent report showed that ADAMTS13 may recognize the A3 domain of VWF.¹⁷ Therefore, if the defects of the enzyme in patients with TTP affect the ADAMTS13 binding site for the A3 domain, cleavage of GST-VWF73-H will not reflect these defects.

TTP is a life-threatening disease. If untreated, mortality may exceed 90%, but plasma exchange therapy significantly reduces mortality. Therefore, measurement of ADAMTS13 activity could be useful for early diagnosis, which is essential for successful treatment of an acute episode. For example, in July 2003, a Japanese pharmaceutical company began ADAMTS13 monitoring, using GST-VWF73-H, for the early diagnosis of patients with suspected TTP associated with ticlopidine. This approach may be effective in preventing the morbidity caused by drug-associated TTP. In addition, periodical ADAMTS13 measurements may be important for monitoring the therapeutic efficacy of plasma exchange or plasma infusion.

Low levels of ADAMTS13 activity have been observed occasionally in several conditions, including heparin-induced thrombocytopenia, severe sepsis, and HUS.¹⁸ But ADAMTS13 activity in these conditions is consistently detectable, and severe ADAMTS13 deficiency appears to be specific for TTP.^19,20 Consequently, ADAMTS13 assays may help in the differential diagnosis of thrombocytopenia, especially when accompanied by microangiopathic changes. In particular, the discrimination of TTP from HUS can be an urgent issue for treatment. Therefore, an assay for ADAMTS13 activity with a narrow normal range is needed. We are currently developing more rapid and quantitative methods for determining ADAMTS13 activity. One of our goals is to develop a bed-side assay for ADAMTS13 activity, for the diagnosis of congenital or autoimmune TTP.

James N. George, MD^*

The evaluation of patients with suspected thrombotic thrombocytopenic purpura is difficult because the diagnostic criteria—(1) thrombocytopenia, (2) microangiopathic hemolytic anemia, and (3) no other clinically apparent etiology—are not specific.¹ The value of ADAMTS13 measurements for establishing the diagnosis of TTP and determining the indication for plasma exchange treatment remains uncertain. Although a severe deficiency of ADAMTS13 activity (< 5%) may be an abnormality that is specific for TTP,² the presenting symptoms of patients with severe ADAMTS13 deficiency are often not severe and not distinguishable from many other common illnesses. Patients can have the characteristic presenting features and clinical course of TTP without severe ADAMTS13 deficiency or even with normal ADAMTS13 activity (> 50%),³ and also patients can have severe ADAMTS deficiency for many years with no illness.⁴ 

Patients with TTP and severe ADAMTS13 deficiency are heterogeneous, with remarkably variable presenting features (Table 3 ).^3,5 Nine of the 22 patients with severe ADAMTS13 deficiency (< 5%) in the Oklahoma TTP-HUS Registry (Table 3 ) had no neurologic abnormalities, not even mild symptoms such as headache and confusion. Nonspecific symptoms, such as weakness, abdominal pain, nausea, vomiting, and diarrhea, were common presenting complaints. Some patients had had symptoms for 3 weeks before their diagnosis; the median duration of symptoms prior to diagnosis was 6 days. The presenting symptoms of these patients were similar to patients diagnosed with idiopathic TTP-HUS but who did not have severe ADAMTS13 deficiency.³ 

Patients with severe illnesses that can mimic the presenting clinical features of TTP, such as sepsis and preeclampsia, may have low but detectable levels of ADAMTS13.^3,5,6 The clinical importance of less severe ADAMTS13 deficiencies is unknown. Furthermore, because of the long half-life of ADAMTS13 in plasma, estimated to be 2.6 days,⁷ transfusion of red cells and platelets before the diagnosis of TTP is considered can increase plasma ADAMTS13 activity.

Initial studies reported that a severe deficiency of ADAMTS13 distinguished syndromes described as TTP from the related syndromes described as HUS.⁸ Although this distinction is valid in many patients, overlap of these syndromes can occur. Acute renal failure, the characteristic clinical feature that is often used to designate syndromes as HUS, may rarely occur in adults with acquired severe ADAMTS13 deficiency.³ Children with congenital TTP caused by mutations of the ADAMTS13 gene may have severe renal failure.^9,10 Also children with typical HUS caused by E coli 0157:H7 infection may rarely have severe ADAMTS13 deficiency.¹¹ 

Although the value of ADAMTS13 measurements for establishing the diagnosis of TTP is uncertain, demonstration of severe ADAMTS13 deficiency and documentation of the inhibitor titer may help to anticipate the clinical course of patients with acquired TTP. Four case series^3,12–,14 have reported lower mortality among patients diagnosed with TTP who had severe ADAMTS13 deficiency (Table 4 ). In the two studies^3,13 that distinguished patients who had idiopathic TTP from patients who had potential etiologies and associated conditions that could have contributed to their illness, the mortality rates were less in the patients with idiopathic TTP. These reports suggested that the high rates of mortality among patients who did not have severe ADAMTS13 deficiency may have been caused by other serious conditions, such as systemic infections. Among the four case series reporting mortality in patients with severe ADAMTS13 deficiency,^3,12–,14 overall mortality was 13% (Table 4 ). Among the 3 case series reporting relapses in patients with severe ADAMTS13 deficiency, the frequencies of relapse were 10 of 44 (23%),¹⁵ 6 of 16 (38%),¹³ and 7 of 16 patients (44%)³ (Table 4 ).

Zheng et al¹³ have reported a further distinction among patients with severe ADAMTS13 deficiency, between patients who had no demonstrable inhibitor at presentation and those patients who had a high titer of inhibitory activity. In patients with no demonstrable inhibitor, plasma exchange was effective in 8 of 9 patients; none of the 8 patients who recovered have subsequently relapsed. In contrast, 4 patients with severe ADAMTS13 deficiency and a high titer inhibitor had more prolonged courses with more frequent serious complications; all suffered relapses and 2 have died, in spite of additional intensive immunosuppressive treatment.

The demonstration of severe ADAMTS13 deficiency with a high titer inhibitor may be an appropriate indication for intensive immunosuppressive treatment. Current clinical practice is inconsistent regarding the use of glucocorticoids and other immunosuppressive agents.¹ Part of this inconsistency is related to the uncertain initial diagnosis in many patients, and the presumed lack of efficacy of immunosuppressive agents for patients with non-idiopathic TTP. However, even among patients with acquired idiopathic TTP and severe ADAMTS13 deficiency, some respond promptly and completely to short courses of plasma exchange, without additional glucocorticoid treatment.^3,13 Patients who respond quickly and completely with only plasma exchange treatment may have only minimal inhibitor activity or an inhibitor that is not detectable by current methods. Patients with high-titer inhibitors may require not only glucocorticoids but also more intensive immunosuppressive treatment with agents such as rituximab or cyclophosphamide.

However, the correlation between the clinical course and ADAMTS13 levels is not consistent. Several patients have been described who had prolonged stable hematologic remissions in spite of persistent undetectable ADAMTS13 activity.^13,16 These observations, similar to observations of patients with congenitally absent ADAMTS13 activity who may live many years or even a lifetime without evidence of TTP,⁴ demonstrate that severe ADAMTS13 deficiency alone is not sufficient to cause the clinical syndrome of TTP. The suggestions of clinical correlation are also tempered by observations that some assays for ADAMTS13 activity may not be consistent across different laboratories, although a multicenter study documented excellent agreement among most assay techniques.¹⁷ Therefore management decisions should remain based upon the clinical course. Patients who respond promptly and completely to plasma exchange may require no additional treatment. Patients whose platelet count does not increase within several days, or in whom recurrent thrombocytopenia recurs when plasma exchange treatments are diminished or discontinued, will likely benefit from glucocorticoid treatment. Patients who have a more severe course, with more severe neurologic abnormalities, and who either do not respond or exacerbate in spite of continuing plasma exchange and glucocorticoid treatment may benefit from more intensive immunosuppressive treatment.

The major concern of patients who achieve remission from TTP is the risk for relapse. Current data suggest that the risk for relapse is almost totally restricted to patients who have severe ADAMTS13 deficiency, and severe ADAMTS13 deficiency is usually, but not always, restricted to patients who have idiopathic TTP. Data from the Oklahoma TTP-HUS Registry are presented in Table 5 . Relapse has not occurred among survivors who had TTP following stem cell transplantation or that was associated with dose-dependent drug toxicity, or who had a prodrome of bloody diarrhea. Two patients with an immune-mediated drug association have relapsed, but these recurrent episodes were caused by repeated ingestion of quinine. Three women whose initial episode of TTP was associated with pregnancy have relapsed. In 1 woman who had recurrent mid-trimester fetal losses, the diagnoses of both the initial and recurrent episode of TTP were uncertain. One woman had 2 episodes of TTP during the first trimester of each of her first 2 pregnancies, and then a third episode when she was not pregnant; ADAMTS13 activity was not measured. The third woman, whose initial episode of TTP was diagnosed at the time of delivery of her first child, had severe ADAMTS13 deficiency with a low titer inhibitor (Figure 4 , Patient 7); she had 4 relapses during the following year when she was not pregnant; cholecystectomy and splenectomy were performed following her fourth relapse; she has been asymptomatic since that time and has had 2 subsequent uncomplicated pregnancies.

Two patients who had additional disorders that may have contributed to the development of TTP have had multiple relapses. One woman with scleroderma had 3 relapses; her ADAMTS13 activity was not measured initially but was 40% at the time of her first relapse. One man with characteristic features of TTP was incidentally diagnosed with HIV infection at the time of his initial episode when his ADAMTS13 activity was 60%. He subsequently has had 4 relapses and his ADAMTS13 activity has decreased with each episode; ADAMTS13 activity was undetectable at the time of his fourth relapse.

Among patients with idiopathic TTP, 19 have relapsed (Table 5 ). Twelve of these patients have had ADAMTS13 activity measurements; 11 patients had undetectable activity either at the time of their initial diagnosis or at the time of a subsequent relapse. The exception is a woman with 3 episodes of TTP who had ADAMTS13 measured only at the time of her third episode; ADAMTS13 activity was 100%. Three of these 12 patients had ADAMTS13 activities of 5%–25% at the time of their initial episode and then undetectable ADAMTS13 activity at the time of a relapse. Their initial higher levels of ADAMTS13 activity may have been related to multiple red cell and platelet transfusions given prior to consideration of the initial diagnosis of TTP-HUS.

Figure 4  illustrates the course following recovery from the initial episode of TTP in 18 patients who initially had severe ADAMTS13 deficiency. Four additional patients had severe ADAMTS13 activity but died during their initial episode, 3 from unresponsive disease and 1 from acute hemorrhage related to central venous catheter insertion. Among the 18 survivors, 8 (44%) have had 1 or more relapses; the occurrence of relapse did not appear to be related to the strength of inhibitor activity. Among the patients who have relapsed, 1 (patient 5) had no demonstrable inhibitor, 3 (patients 6, 7, and 8) had trace or mild inhibitory activity, 3 (patients 1, 3, and 16) had moderate inhibitory activity, and 1 patient (patient 2) had strong inhibitory activity.³ At the time of his relapse, patient 2 was diagnosed with systemic lupus erythematosus and this has been his continuing major problem. Of the 6 patients who had strong inhibitory activity (patients 2, 4, 9, 11, 12, 15), only 1 has relapsed. Although the follow-up of these patients is limited, with a median duration of 3.3 years, these observations suggest that in most patients who will relapse, the initial relapse will occur within the first year (6 of 8 patients). These observations also suggest that in patients who relapse, most will have only 1 relapse (6 of 8 patients). Therefore these observations demonstrate the difficulty of assessing the benefit of any treatment to prevent future relapses. For example, reports of success of splenectomy for preventing future relapses may only be observing the natural history of TTP, with diminishing risk for relapses over time from the initial diagnosis.^18,19 

The observations that essentially all patients who have recurrent episodes of TTP have severe ADAMTS13 deficiency, and that nearly all of these patients have demonstrable inhibitory activity, suggests that it is appropriate to begin immunosuppressive therapy together with plasma exchange for management of patients with a recurrent episode.

Although almost all recurrent episodes of TTP are immediately recognizable, recent observations suggest the potential for diagnostic difficulty. Three women have been reported who had stroke symptoms following recovery from TTP without accompanying thrombocytopenia; recurrent TTP was considered and symptoms improved with plasma exchange treatment. Subsequently the diagnosis of TTP was supported by finding absent ADAMTS13 activity.^20,21 

Since TTP primarily affects women, and since pregnancy appears to be associated with TTP,²² the risks of a future pregnancy are a common concern among women who have recovered from TTP. Case reports suggest that the risk for recurrent TTP with a subsequent pregnancy is high; among 49 women who have been reported to have 70 subsequent pregnancies, 36 (73%) had recurrent TTP with a pregnancy.²³ However, these reports may be biased by descriptions of unusual patients with complex outcomes.²³ Our experience with 19 women who have had 30 pregnancies following recovery from TTP is more encouraging. Only 5 (26%) women have been diagnosed with recurrent TTP during a subsequent pregnancy, each only during 1 subsequent pregnancy.²³ In 3 of these 5 women, the diagnosis of recurrent TTP was uncertain because of coexisting complications: severe preeclampsia, intrauterine fetal death, and extreme hypertension in a woman with chronic renal failure. This experience emphasizes the difficulty of diagnosing TTP in the presence of other pregnancy-related complications, especially when anticipation is high because of a previous diagnosis of TTP. However, this experience, with complete documentation of all uncomplicated pregnancies, also suggests that recurrent TTP with a subsequent pregnancy is uncommon.

Although the risk for relapse is the major concern of patients who have recovered from TTP, it is not the only concern. Many patients describe persistent cognitive abnormalities for many years following recovery that can be documented by tests of new learning and recent memory.²⁴ The etiology of these abnormalities is unclear. Perhaps successful treatment with plasma exchange and immunosuppressive therapy will reveal additional long-term sequelae that require further study and perhaps additional supportive care.