Treating TTP/HUS with plasma exchange: a single centre’s 25-year experience


  • Brian R. Forzley,

    1. London Health Sciences Centre, Victoria Hospital
    2. Division of Nephrology, Department of Medicine, London Health Sciences Centre, University of Western Ontario
    Search for more papers by this author
  • Jessica M. Sontrop,

    1. Department of Epidemiology & Biostatistics, University of Western Ontario, London, ON, Canada
    Search for more papers by this author
  • Jennifer J. Macnab,

    1. Division of Nephrology, Department of Medicine, London Health Sciences Centre, University of Western Ontario
    2. Department of Epidemiology & Biostatistics, University of Western Ontario, London, ON, Canada
    Search for more papers by this author
  • Salina Chen,

    1. London Health Sciences Centre, Victoria Hospital
    Search for more papers by this author
  • William F. Clark

    1. London Health Sciences Centre, Victoria Hospital
    2. Division of Nephrology, Department of Medicine, London Health Sciences Centre, University of Western Ontario
    Search for more papers by this author

William F. Clark, 800 Victoria Hospital Room A2-343, Commissioners Road E., London N6A 4G5, ON, Canada. E-mail:


Thrombotic thrombocytopenic purpura/Haemolytic uremic syndrome (TTP/HUS) is a thrombotic microangiopathy with a 6-month mortality rate of 16–29%. The present study described the clinical features, treatment regime and 6-month all-cause mortality rate of TTP/HUS patients at the London Health Sciences Centre (LHSC), Canada. Data for this retrospective cohort study were obtained from inpatient and outpatient records for all patients referred for plasma exchange therapy at LHSC, Canada between 1981 and 2006. Patients (n = 110) were categorized as: idiopathic primary (38%) or relapsed (16%), and secondary responsive (30%) or non-responsive (16%). Mortality data were available for all but three patients. The all-cause 6-month mortality rate was 19% overall and was 12% and 26% among idiopathic and secondary TTP/HUS patients, respectively. No mortality events occurred among the 17 idiopathic patients who relapsed. Relapsed patients had the least severe presenting characteristics, the fastest response time, and experienced significant improvement in the severity of clinical features between the first and final presentation. These findings suggest an excellent outcome for relapsed TTP/HUS patients. Patient education, surveillance, and aggressive plasma exchange therapy are hypothesized to improve the likelihood of survival: these hypotheses should be tested in a randomized controlled trial.

Thrombotic thrombocytopenic purpura/Haemolytic uremic syndrome (TTP/HUS) is a relatively uncommon disorder characterized by unexplained haemolytic anaemia and thrombocytopenia. In the past it has been differentiated on clinical grounds with HUS classically affecting the kidney and TTP affecting the brain, kidney and other organs (Hosler et al, 2003). Work by both Tsai and Lian (1998) and Furlan et al (1998) showed promise for differentiating these disorders based on ADAMTS-13 (a disintegrin and metalloprotease with thrombospondin type 1 motifs) protease activity. However, the usefulness of this distinction for clinical purposes has been brought into question using larger cohort studies from less selected populations (Vesely et al, 2003; Peyvandi et al, 2004; Veyradier & Girma, 2004; Lammle et al, 2005). Vesely et al (2003) observed that only 33% of 142 idiopathic patients had severe protease deficiency and, more importantly, that protease level <5% did not predict treatment response. These data provide evidence that the clinical entity of TTP/HUS remains a heterogenous disorder not fully explainable by ADAMTS-13 deficiency. Alternative proteins, including Factor H, monocyte chemoattractant protein (MCP) and others, have more recently emerged as pathogenetically relevant to microangiopathies, and have been recently reviewed (Noris & Remuzzi, 2005). George et al (2004) recognized that, regardless of ADAMTS-13 level, patients with a clinical diagnosis of idiopathic TTP/HUS were similar in terms of treatment responsiveness and prognosis. This observation supplies evidence that the syndromic term acquired idiopathic TTP/HUS still applies in the era of molecular diagnosis, particularly for rapid decisions to initiate plasma exchange.

Although outcomes have improved with plasma exchange, modern treatment remains sub-optimal with mortality rates of 16–29% in published randomized clinical trials (Rock et al, 1991; Bobbio-Pallavicini et al, 1997; Zeigler et al, 2001; Mintz et al, 2006). Variability in survival could reasonably be ascribed to inadequate use of plasma exchange prescription; however, a recently published systematic review of randomized controlled trials involving patients with TTP/HUS did not identify any studies that examined the effect of plasma exchange dose on patient mortality (Brunskill et al, 2007). Across North America, prescriptions vary widely in terms of exchange volume and frequency, plasma product, use of a taper, and adjunctive therapy (Bandarenko & Brecher, 1998). Survival outcomes from different centres have been published, but comparison between studies is problematic due to differences in inclusion criteria, data collection period, health care systems, and differences in disease-specific care. This led us to examine retrospectively our own experience with delivery of a single centre TTP/HUS treatment protocol.

The objectives of the present study were to describe the presenting characteristics, treatment regime, and 6-month all-cause mortality of TTP/HUS patients treated at the plasmapheresis unit at the London Health Sciences Centre (LHSC) between 1981 and 2006.


Ethics approval was obtained from the University of Western Ontario Research Ethics Board to conduct a retrospective analysis of TTP/HUS over the last 25 years at LHSC in London, Canada. The inpatient and outpatient records of all patients referred for plasma exchange were identified through records from the plasmapheresis unit. A data extraction tool was constructed and tested by one of the authors in 10 consecutive medical records in order to confirm that the desired data was consistently available in the medical charts. Data was collected by two trained research assistants. Inclusion criteria required the detection of unexplained haemolytic anaemia (haemoglobin < 120 g/l) with schistocytes and thrombocytopenia (platelet count < 100 × 109/l). Exclusion criteria consisted of age < 15 years, International Normalized Ratio (INR) > 1·3, chart review detecting a cause for this syndrome that was not due to a microangiopathy (e.g. malaria, severe sepsis, valvular heart disease, endocarditis) or was due to a microangiopathy of another cause (e.g. haemolysis-elevated liver enzymes-low platelet (HELLP) syndrome, malignant hypertension, disseminated intravascular coagulopathy). The following baseline clinical data at presentation were obtained from the LHSC chart review: age (years), sex, weight (kg), haemoglobin (g/l), platelet count (×109/l), creatinine (μmol/l), dialysis at presentation, INR, lactate dehydrogenase (LDH; U/l), presence of fever (>38·0°C) and presence of neurological dysfunction (focal deficit, seizure, coma). The following treatment information was obtained: date of first exchange, frequency of exchanges during the first cycle of therapy (exchanges/day), and mean exchange volume during the first cycle of therapy (ml/kg per day). With regards to adjunctive therapy, it has not been our practice to use antiplatelet agents over the past 20 years. We have treated plasma reactions with antihistamines and methylprednisolone. In the past 10 years we used cytotoxic and Rituximab therapy in less than 4% of our patients that were deemed refractory to plasma exchange therapy. All-cause mortality at 6 months was corroborated by medical records.

Patients with TTP/HUS were grouped into four categories. Efforts to clearly categorize every case of TTP/HUS are inherently limited by the multifaceted nature of the disorder and its heterogeneous etiology; however, for the purpose of this research, the following definitions were used. Idiopathic TTP/HUS was defined as unexplained haemolytic anaemia and thrombocytopenia with no established trigger or associated disease. Idiopathic patients were sub-grouped into primary (no record of previous TTP/HUS presentations at study entry) or relapsed (patients with a history of TTP/HUS at study entry or idiopathic patients who relapsed over the course of the study). Secondary TTP/HUS was haemolytic anaemia and thrombocytopenia with an identified trigger (e.g. quinine) or associated disease (e.g. human immunodeficiency virus). In previous studies, some secondary causes demonstrated a similar prognosis to idiopathic TTP/HUS, while others have exhibited a worse prognosis (George, 2006). This lead our group to a priori distinguish secondary causes based on expected response to plasma exchange. Examples of responsive causes included non-chemotherapeutic drugs, autoimmune diseases, and viral infections. Examples of unresponsive causes included metastatic cancer, systemic chemotherapy and bone marrow transplantation. The first treatment cycle was defined as the period of time that began on the first day of plasma exchange and continued either to 9 d after the first treatment or until therapy was stopped due to patient death or response. The definition of response was restoration of the platelet count above 150 × 109 cells/l and LDH below 360 U/l leading to cessation of plasma exchange (George, 2006). We adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines in the reporting of our observational data (von Elm et al, 2007).


All-cause 6-month mortality rates of TTP/HUS sub-groups were compared using 95% confidence intervals (95% CI) (Hanley & Lippman-Hand, 1983; Newcombe & Altman, 2000). The number of days until response was compared across TTP/HUS subgroups using 95% CI; all survivors were included in this analysis, including patients defined as secondary non-responsive. Presenting clinical features were compared between idiopathic primary and relapsed groups using analysis of variance (anova) for continuous variables and Fisher’s exact test for categorical variables. Presenting clinical features and time to response among patients with two presentations during the study period were compared between first and final presentation using the paired t-test. A composite variable representing the severity of clinical features at presentation was created based on the presence of risk factors identified as such in previous research (Rose et al, 1993; Wyllie et al, 2006): hemoglobin < 90 g/l; age > 40 years; fever > 38°C; neurological dysfunction including focal motor deficits, seizure, or coma; creatinine > 132 μmol/l, platelet count < 20 × 109/l and LDH ≥ 1200 U/l. Associations between severity of clinical features at presentation and respectively, average first cycle exchange volume and mortality were evaluated using anova. For continuous data, means are presented with standard deviations in brackets. Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS), v 15.0 (SPSS Inc., Chicago, IL, USA).


We identified 128 patients treated with plasma exchange for suspected TTP/HUS over a period from 1981 to 2006. After eliminating patients who did not meet our inclusion criteria (n = 14) as well as patients with incomplete records (n = 4), a sample size of 110 patients was available for analysis. Data on 6-month mortality was unavailable for three of these patients.

Diagnostic categories and clinical features at first presentation are presented in Table I. The average number of days to response was fastest among relapsed patients with a mean of 5·6 d (95% CI, 2·8 to 6·2) (Fig 1). Patients received an average of 47·0 ml/kg per day (standard deviation [SD] 20·9) of plasma exchange during the first cycle.

Table I.   Characteristics of TTP/HUS patients: diagnostic categories and clinical features (n = 110).
 n (%)
  1. *Secondary causes expected to respond to treatment include: drug effect (eg. quinine, ticlodipine), bloody diarrhoea prodrome, human immunodeficiency virus, autoimmune disorders (eg. Lupus, anti-neutrophil cytoplasmic antibody vasculitis).

  2. †Secondary causes not expected to respond to treatment include: chemotherapy, systemic malignancy, bone marrow transplantation.

  3. LDH, lactate dehydrogenase.

Diagnostic category
  Primary42 (38)
  Relapsed17 (16)
  Responsive*33 (30)
  Unresponsive†18 (16)
Clinical Features
 Female76 (69)
 Age > 40 years74 (67)
 Fever > 38°C21 (21)
 Coma25 (24
 Focal motor deficit20 (19)
 Seizures17 (16)
 Creatinine ≥ 132 μmol/l66 (61)
 LDH ≥ 1200 U/l44 (41)
 Haemoglobin < 90 g/l61 (56)
 Platelet count < 20 × 109/l48 (44)
Figure 1.

 Average number of days until response among survivors (95% Confidence Interval).

The overall 6-month all-cause mortality rate was 19% (95% CI, 11% to 26%) (n = 107 patients); mortality did not change significantly over the time-period of the study. The 6-month all-cause mortality rate for patients with idiopathic versus secondary TTP/HUS was 12% (95% CI, 4% to 21%) and 26% (95% CI, 14% to 38%), respectively. Patient mortality rates compared across the four TTP/HUS sub-groups are presented in Table II. There were no mortality events among the 17 relapsed patients. The average time to relapse was 1·9 years (min: 1·9 months; max: 7·5 years).

Table II.   Six-month all-cause mortality rate for diagnostic categories of TTP/HUS (n = 107).
Diagnostic category% (95%CI)
  1. *Upper events = 3/n (Hanley & Lippman-Hand, 1983).

 Idiopathic17 (6, 29)
 Relapsed0 (0, 18)*
 Responsive21 (7, 35)
 Unresponsive35 (13, 58)

As shown in Table III, relapsed patients had less severe presenting characteristics compared to primary idiopathic patients. Statistically significant differences were observed for haemoglobin, comas, and focal motor defecits. There were 14 relapsed patients with complete data from at least two TTP/HUS presentations. While age significantly increased between time-points, severity of clinical features decreased with statistically significant improvements observed for LDH and haemoglobin (Table IV).

Table III.   Presenting characteristics of idiopathic primary and relapsed TTP/HUS patients (n = 59).
 Primary (n = 42)Relapsed (n = 17)P-value
% or mean (SD)% or mean (SD)
Female  66·7 70·61·00
Age (years)  45·2 (18·2) 44·7 (12·4)0·911
Fever  23·1 23·11·00
Coma  30·0  0·00·012
Focal motor deficit  22·5  0·00·048
Seizures  20·0 12·50·707
Platelet count (×109/l)  40·6 (49·8) 49·5 (57·3)0·552
Haemoglobin (g/l)  88·4 (20·5)105·6 (24·0)0·007
Creatinine (μmol/l) 249·4 (250·2)132·4 (114·5)0·071
Lactate dehydrogenase (U/l)1363·1 (1545·2)763·8 (729·0)0·133
Table IV.   Presenting characteristics of TTP/HUS patients who relapse: a comparison of first and final presentation (n = 14).
 First presentationFinal presentationP-value
Mean (SD)Mean (SD)
Age (years) 39·5 (10·1) 42·4 (10·6)0·002
Time to response (d) 12·3 (17·0)  5·9 (3·8)0·196
Platelet count (×109/l) 23·1 (16·5) 50·8 (52·9)0·050
Haemoglobin (g/l) 96·4 (25·0)110·4 (28·1)0·009
Creatinine (μmol/l)151·2 (142·7)133·7 (125·5)0·097
Lactate dehydrogenase (U/l)1192·5 (746·1)852·7 (858·6)0·049

The average number of risk factors did not differ significantly between those who died (3·6) versus those who survived (3·5); however, patients with a greater number of risk factors at presentation received more plasma exchange than those with fewer risk factors. Patients with ≥5 risk factors received 15·7 ml/kg per day more plasma exchange than those with 3–4 risk factors (P = 0·01) and 10·4 ml/kg per day more plasma exchange than those with 1–2 risk factors (P = 0·22). Although not statistically significant, survivors received slightly larger volumes of plasma exchange during the first cycle of therapy compared to those who died, 47·9 ml/kg per day (SD 22·1) vs. 42·8 ml/kg per day (SD 13·4) respectively; P = 0·36.


In this 25-year retrospective analysis of TTP/HUS patients at the LHSC plasmapheresis unit (Canada), we observed a zero-event 6-month mortality rate among TTP/HUS patients who relapsed. This result is a novel and important finding, particularly when compared to overall mortality rates for idiopathic TTP/HUS patients. Moreover, prognosis in the relapsed group is less well established in cohort studies published to date (Bell et al, 1991; Bobbio-Pallavicini et al, 1992; Shumak et al, 1995; Dervenoulas et al, 2000; Shamseddine et al, 2004; Gurkan et al, 2005). No report was explicit with outcome measure referenced to time, the importance of which was demonstrated by Rock et al (1991). The two largest cohorts of relapsed patients, with 69 and 17 patients (Bell et al, 1991; Shumak et al, 1995), reported mortality rates of 3% and 12%, respectively. Given our extremely low 6-month mortality in relapsed patients, we examined our data for plausible explanations. In our study, the clinical features of relapsed patients were less severe and time to response was faster at the final versus primary presentation with TTP/HUS, despite the significantly older age of patients at the final time-point. These observations lead us to hypothesize that the survival advantage observed within this group was the result of earlier presentation and treatment initiation, which previous studies have shown to improve the odds of survival (Pereira et al, 1995; Colflesh et al, 1996; Dervenoulas et al, 2000). A delay in initiating plasma exchange has been associated with treatment failure (Pereira et al, 1995), and another study reported that first plasma exchange occurred sooner in the survivor group than the deceased group (1·3 vs. 9·0 d, P < 0·01) (Colflesh et al, 1996). Dervenoulas et al (2000) observed a longer time to diagnosis among patients who died (9·3 d) compared to patients who survived (7·3 d), although this difference was not statistically significant. We speculate that patient education and regimented follow-up could improve mortality. Although our study did not collect data on the number of days between symptom recognition, presentation and treatment initiation, it is plausible that relapsed patients presented earlier due to symptom recognition and also due to the follow-up system at LHSC, where standard laboratory surveillance of TTP/HUS patients involves regular monthly platelet count and LDH determinations and annual physician assessment. Patients are given instructions to obtain platelet and LDH testing after discharge if they experience symptoms similar to their presentation. Test results are faxed to the on-call physician. It is not clear whether the excellent outcomes observed in relapsed patients relate to a biological difference between patients that will or will not relapse, or relates to a difference in how these patients are managed. The hypothesis that patient education and regimented follow-up could improve mortality may be important, because it represents a safe and low cost intervention that may not be presently exploited by many centres. We are not aware of any published literature on this management strategy, and this represents an important gap in the literature.

The mortality rate observed for idiopathic cases of TTP/HUS falls within the range reported from single-centre retrospective studies (9% to 46%) (Bell et al, 1991; Bobbio-Pallavicini et al, 1992; Bandarenko & Brecher, 1998; Dervenoulas et al, 2000; George et al, 2004; Gurkan et al, 2005) and randomized clinical trials (16% to 29%) (Rock et al, 1991; Bobbio-Pallavicini et al, 1997; Zeigler et al, 2001; Mintz et al, 2006). The 6-month all-cause mortality rate for idiopathic patients at LHSC was 10% lower (95% CI, –4% to 24%) than that reported in a Canadian multicentre trial with a similar patient composition and health care system (Rock et al, 1991). Although the multicentre trial included patients with a history of TTP (n = 11), mortality rates were not reported separately for primary and relapsed patients. While the difference in mortality rates was not statistically significant, this test was severely underpowered (Power = 17%). Comparing mortality outcomes across studies is difficult because of differences in study methods, selection criteria, definition of mortality (eg, all-cause versus disease-specific, and time-to-event), patient and diagnostic heterogeneity, disease-specific care, time-frame and health care system. However, the latter comparison with the Canadian trial is free of most of these biases. Both studies used the same inclusion criteria, mortality definition (all-cause mortality at 6 months from presentation), and were limited to idiopathic TTP/HUS patients. Since both studies were Canadian and overlap in time frame, this comparison is matched in terms of access to care and general available resources. While it is possible that the difference in the proportion of relapsed patients between studies may be responsible for some of the difference in the observed mortality rates, differences in patient therapy should also be considered. In the multicentre trial (Rock et al, 1991), the plasma exchange volume and frequency was fixed, whereas in this study, plasma exchange volume and frequency were both flexible, and patients with a greater number of risk factors received significantly more plasma exchange than those with fewer risk factors. Since our centre does not have a protocol determining exchange volumes based on the presence of risk factors, this difference probably reflects the judgement of the treating physician. This observation is important in light of the fact that previous studies have recommended more aggressive treatment for patients with more severe presenting characteristics because these patients are at greater risk for death and have a greater number of risk factors at presentation compared to those who survive (Rose et al, 1993; Wyllie et al, 2006). Treating sicker patients more aggressively may also explain why no association was observed between number of risk factors and mortality among LHSC patients. These results provide indirect evidence that an increase in plasma exchange volume confers a survival benefit. In our study, the average daily exchange volume was slightly higher among survivors than the deceased, though this difference was not statistically significant. In other observational studies, a non-significant trend towards lower mortality with greater exchange volumes was observed (10·5% versus 8·0% for plasma volume 1·0–1·3 versus plasma volume 1·5–2·0, respectively; P > 0·45) (Bandarenko & Brecher, 1998), and Bobbio-Pallavicini et al (1992) observed an association between more frequent exchanges and improved survival (P < 0·005). Also, twice daily plasma exchange has been successfully used in some cases of refractory TTP (Nguyen et al, 2008). Plasma exchange dose and frequency may very well be important determinants of survival in TTP/HUS patients; however, these relationships are difficult to study in an observational setting due to two biases acting in opposite directions. In the first case, the volume/frequency of plasma received will be under-estimated among the deceased, although less so if therapy is defined as average volume per day. In the second case, clinicians may increase plasma exchange volumes for sicker patients, as was shown here, or for non-responding patients, thus obscuring a beneficial effect of larger plasma exchange volumes on survival. These questions would be better answered within a randomised controlled trial.

Although our 25-year review does not have ADAMTS-13 protease or antibody inhibitor measurements, the hypothesis that dose of plasma exchange and time to treatment initiation are critically important factors is biologically plausible if idiopathic TTP/HUS in adults mainly relates to depression of the ADAMTS-13 protease by an inhibiting antibody (Furlan et al, 1998; Tsai & Lian, 1998; Lammle et al, 2005; George, 2006) or an antibody directed at another critical target (Noris & Remuzzi, 2005). A delay in initiating plasma exchange would allow a greater degree of microvascular thrombosis to ensue, leading to a greater extent of organ damage and a greater risk of patient mortality. This theory is indirectly supported by the laboratory values of the relapsed patients who presented with significantly higher hemoglobin and platelet count and lower LDH level at their final versus primary presentation (Table IV). This theory would also account for the trend towards greater exchange volumes in survivors, due to an earlier restoration of protease activity sufficient to reverse the microvascular thrombosis and avoidance of more extensive organ damage and a greater mortality risk.

In summary, we observed a relatively low mortality rate for idiopathic patients, and no mortality events amongst relapsed TTP/HUS patients, which has not been reported to date. We may infer from these findings that time to presentation and treatment initiation is critically important in the successful treatment of TTP/HUS, yet management implications, such as formal patient education and laboratory surveillance, have not been well addressed by published literature. A non-significant difference in mortality between our study and a multicentre trial (Rock et al, 1991) may be partially explained by the fact that LHSC patients with a greater number of risk factors received more plasma exchange, which was fixed in the multicentre trial. However, we are unable to draw firm conclusions given the small sample size and limitations inherent to observational data. In conclusion, we present indirect evidence suggesting that improved outcomes in TTP/HUS may be possible through urgent initiation of therapy after diagnosis and more aggressive use of plasma exchange in high risk patients. This important data is hypothesis-generating and we propose a multi-centre randomised controlled trial that compares large volume high frequency plasma exchange with conventional plasma exchange, and also compares a patient education intervention and follow-up strategy to conventional follow-up management.


The authors wish to acknowledge the efforts of Mr. John Enright and Ms. Katie Lockwood, who worked enthusiastically on data collection.