Preliminary results of this study were presented, in part, in a poster presented at the December 2006 meeting of the American Society of Hematology, Orlando, FL.
Antiphospholipid antibodies in adults with immune thrombocytopenic purpura
Article first published online: 28 MAY 2008
© 2008 The Authors. Journal Compilation © 2008 Blackwell Publishing Ltd
British Journal of Haematology
Volume 142, Issue 4, pages 638–643, August 2008
How to Cite
Pierrot-Deseilligny Despujol, C., Michel, M., Khellaf, M., Gouault, M., Intrator, L., Bierling, P. and Godeau, B. (2008), Antiphospholipid antibodies in adults with immune thrombocytopenic purpura. British Journal of Haematology, 142: 638–643. doi: 10.1111/j.1365-2141.2008.07228.x
- Issue published online: 21 JUL 2008
- Article first published online: 28 MAY 2008
- Received 28 January 2008; accepted for publication 28 March 2008
- anticardiolipin antibodies;
- antiphospholipid antibodies;
- immune thrombocytopenic purpura;
- lupus anticoagulant;
To determine the clinical significance of antiphospholipid antibodies (aPL) in patients with immune thrombocytopenic purpura (ITP), anticardiolipin (aCL) (IgG and IgM) and lupus anticoagulant (LA) were sought at diagnosis in 215 ITP adults with platelets <50 × 109/l. aPL (aCL and/or LA) were detected in 55 patients (26%): aCL alone in 39 (18%), aCL and LA in 15 (7%) and LA alone in one (0·5%). LA was significantly associated with high IgG-aCL levels (P = 0·001). Among age, sex, initial platelet count, bleeding score, acute or chronic ITP outcome, only younger age was significantly associated with LA-positivity (mean age 29 ± 14 years vs. 45 ± 20 years, P = 0·002). After a median follow-up of 31 months, 14/215 (7%) patients developed thrombosis (four arterial, 10 venous and/or pulmonary embolism); four of them (29%) had high aCL levels and LA. Multivariate analysis significantly associated thrombosis events only with age [hazard ratio (HR) = 1·6; 95% confidence interval (CI): 1·2–2·4], LA (HR: 9·9; 95% CI: 2·3–43·4) or high IgG-aCL level (HR: 7·5; 95% CI; 1·8–31·5). Although the thrombosis rate was low, the significant associations between thrombosis and LA or high aCL level suggest that aPL should be tested at ITP diagnosis.
Antiphospholipid antibodies (aPL), usually comprised lupus anticoagulant (LA) and/or anticardiolipin antibodies (aCL), represent a heterogeneous family of immunoglobulins directed against various protein–phospholipid complexes (Levine et al, 2002). In clinical terms, aPL have been associated with venous and/or arterial thrombosis and/or recurrent fetal loss, which define the Antiphospholipid Syndrome (APS) (Wilson et al, 1999; Miyakis et al, 2006). Thrombocytopenia is found in 20–40% of APS patients (Cuadrado et al, 1997; Wilson et al, 1999; Cervera et al, 2002; Krause et al, 2005; Miyakis et al, 2006).
High percentages of patients with isolated immune thrombocytopenic purpura (ITP) also have aPL. Harris et al (1985) first reported aPL at the time of diagnosis in 30% of ITP patients, but they have been detected in up to 75% of cases (Stasi et al, 1994; Arfors et al, 1996; Funauchi et al, 1997; Lipp et al, 1998; Diz-Kucukkaya et al, 2001; Bidot et al, 2005, 2006), which can partly be attributed to technical differences, for example the search for antibodies specific to different phospholipid antigens.
The clinical significance of aPL in ITP patients remains controversial and international experts have not established clear recommendations regarding the impact of aPL in ITP patients [British Committee for Standards in Haematology, (BCSH) 2003; George et al, 1996]. A panel of 13 American Society of Hematology experts concluded that the routine search for aPL was probably not justified (George et al, 1996). The BCSH guidelines only recommended testing for children and pregnant women (BCSH, 2003). The objective of our study was to determine the frequency and clinical relevance of aPL in a large single centre cohort of adults at the time of ITP diagnosis detected with tests routinely used in hospitals.
Materials and methods
We retrospectively collected data from patients with ITP diagnosed and followed in our institution over 15 years (1990–2005). ITP patients referred to our hospital for complementary evaluations were excluded. ITP diagnosis was based on the American Society of Hematology guidelines for ITP (George et al, 1996). Only patients over 15 years old at diagnosis with platelet counts <50 × 109/l were included. Patients with systemic lupus erythematosus (SLE) satisfying the American College of Rheumatology criteria (Tan et al, 1982) and/or APS fulfilling the international consensus statement criteria (Wilson et al, 1999; Lackner et al, 2006) were excluded from the study, as were pregnant patients or those with concomitant human immunodeficiency (HIV) or hepatitis C virus infection.
Lupus anticoagulant and aCL were sought at the time of ITP diagnosis. Anti-ß2-glycoprotein-1 antibodies (anti-β2GP1) were also sought in aCL-positive patients seen in the last 5 years. IgG and IgM isotypes of aCL were detected using anti-ß2GP1-dependent enzyme-linked immunosorbent assay (BMD, Marne-la-Vallée, France) (Harris et al, 1987). Results, expressed as IgG antiphospholipid or IgM antiphospholipid units, were considered negative (<15 μg/dl), weakly positive (15–39 μg/dl) or strongly positive (≥40 μg/dl). A solid-phase immunoassay on human β2GP1-coated plates (BMD) was used to detect anti-β2GP1.
Lupus anticoagulant was sought using standard coagulation assays, according to the guidelines of the International Society of Thrombosis and Hemostasis (Brandt et al, 1995). Briefly, LA were considered present when activated thromboplastin time and/or kaolin-clotting time were prolonged and the prolongation was not corrected by 1/1 mixture of patient plasma/normal pooled plasma. LA presence was then confirmed by shortening or correcting the prolonged coagulation time by addition of excess phospholipids.
The severity of ITP-related bleeding was evaluated at diagnosis using a previously reported bleeding score (Khellaf et al, 2005). Responses to medical and/or surgical treatment, clinical outcome [acute (i.e. ≤6 months without splenectomy) or chronic ITP] and venous and/or arterial thrombotic events were analyzed. A therapeutic response was defined as a sustained platelet count >50 × 109/l with at least a twofold increase from baseline for ≥3 months. Arterial thrombotic events, consisting of stroke or transient ischaemic attack, myocardial infarction or peripheral arterial thrombosis, were confirmed by arteriography or computed-tomography scan (for stroke) or electrocardiogram and cardiac enzymes (for myocardial infarction). Venous thromboses, consisting of superficial and deep vein thromboses, and pulmonary embolism, were confirmed by Doppler ultrasonography, ventilation–perfusion lung scintigraphy and/or computed tomography, while cerebral vein thrombosis was confirmed by angiography or magnetic resonance imaging.
Variables are expressed as mean ± standard deviation (SD), number (%) or median with interquartile ranges (IQR) according to their distribution. Parameters studied included: age, sex, platelet count at diagnosis, aPL, aCL (IgG and IgM), LA, anti-β2GP1, antinuclear antibodies (ANA), bleeding score, therapeutic responses and clinical outcome (acute or chronic). Characteristics of aPL-positive and -negative patients were compared using a chi-squared or Fisher’s exact test as appropriate for categorical variables and Student’s t-test for continuous variables. Pearson’s coefficient was used to test for correlations between variables.
All parameters were analyzed for their association with time of thrombosis, the reference time being defined as the date of ITP diagnosis. Time until thrombosis was measured from the date of ITP diagnosis to the date of thrombosis or last follow-up. These analyses used univariate and multivariate regression with Cox’s proportional hazards model. If a patient underwent splenectomy, the latter was considered a time-dependent covariate.
We used a graphic method based on logarithms of cumulated survival function to verify the proportional hazards model assumption. Lastly, we assessed the final model validity using the global test of Harrel and Lee based on Schoenfeld residuals. All P values were two-sided and P < 0·05 was considered significant. Analyses were conducted with sas software version 8.2 (SAS Institute, Cary, NC, USA).
Among the 293 patients, newly suspected of having ITP with platelet counts <50 × 109/l seen in our institution during the study period, 78 were excluded because of incomplete data (n = 33) or associated disease (24 Evans syndrome, 11 definite SLE, six primary APS and four HIV infections). Our analyses concerned 215 patients, whose baseline characteristics are listed in Table I.
|Characteristic||All patients (n = 215 )||aPL-positive (n = 55)||aPL-negative (n = 160)||P-value|
|Ratio, male/female, n (%)||73/142 (34)||20/35 (36)||53/107 (33)||0·66|
|Age, years, mean ± SD||44 ± 20||41 ± 20||45 ± 20||0·21|
|Platelet count (×109/l), mean ± SD||13 ± 12||11 ± 11||14 ± 12||0·20|
|Bleeding score*, mean ± SD||3·4 ± 3·6||3·0 ± 2·6||3·5 ± 3·9||0·37|
Antiphospholipid antibodies (aCL and LA) were detected at ITP diagnosis in 55 patients (26%), as detailed in Fig 1. IgG-aCL were detected in 42 patients; IgG-aCL were 15–39 and ≥40 μg/dl in 30 and 12 patients respectively. Among the 15 patients with LA and aCL, nine had only IgG-aCL, one had IgM-aCL alone and five had IgM- and IgG-aCL.
Age, sex, initial platelet counts and bleeding scores were not associated with aPL (Table I), aCL or LA. Only age differed for the subgroup of patients with LA (mean age: 29 ± 14 years vs. 45 ± 20 years, P = 0·002). LA was correlated with the presence of high IgG-aCL level (≥40 IU/l) (r = 0·63, P = 0·001). Anti-β2GP1 were sought in 43 aCL-positive patients and detected in 11 (26%) (IgM, n = 7; IgG, n = 6); six anti-ß2GP1-positive patients also had LA.
Median follow-up was 30 (IQR: 10–73) months. Acute ITP (complete recovery within 6 months) was diagnosed in 60 (28%) patients. The outcomes (acute or chronic ITP), responses to steroids or splenectomy and mortality rates did not differ as a function of aPL status (Table II).
|Outcome||All patients (n = 215)||aPL-positive (n = 55)||aPL-negative (n = 160)||P-value|
|Acute/chronic (>6 months) ITP||60/155 (28)||13/42 (24)||47/113 (29)||0·41|
|Transient response to steroids||153/176 (87)||40/46 (87)||113/130 (87)||0·99|
|Sustained response to steroids (>3 months)||87/176 (49)||22/47 (47)||65/129 (50)||0·67|
|Splenectomy||30/36 (83)||7/7 (100)||23/29 (79)||0·32|
|Associated disease during follow-up|
|Definite SLE†||6 (3)||4 (7)||2 (1)||0·039|
|Antiphospholipid syndrome†||4 (1)||4 (7)||0||0·004|
|Death||5 (2)||1 (2)||4 (3)|
Thrombosis during follow-up
Fourteen patients (7%) experienced one or several thrombotic events; four of them developed a second thrombotic event (Patients 3, 8, 12 and 13) and one a third episode (Patient 14). Their detailed characteristics and associated cardio-vascular risk factors are summarized in Table III. The first thrombosis was arterial in Patients 4, 6, 10 and 13 and venous in the remaining 10 (superficial vein thrombosis, deep vein thrombosis, pulmonary embolism). Only one patient was taking dapsone for ITP at the time of thrombosis. None of the patients received intravenous immunoglobulins, danazol or underwent splenectomy during the 2 weeks preceding the thrombotic event. The median platelet count at the time of thrombosis was 126 × 109/l (IQR: 80–200 × 109/l) and was below the normal value in eight patients. Patients 4, 6 and 13 were thrombocytopenic when their arterial thromboses occurred. Thrombosis Patients 4, 7, 8 and 14 had LA and high aCL levels and were subsequently diagnosed with definite APS; Patient 8 developed concomitant SLE. The 10 other patients with thrombosis were aPL-negative. At the time of thrombosis, four aPL-positive patients (100%) were thrombocytopenic in contrast to only three of the 10 aPL-negative patients (30%) who developed a thrombosis; these percentages did not reach statistical significance.
|Patient no.||Age, sex||At ITP diagnosis||Thrombosis and cardiovascular risk factors||First thrombosis||Definitive diagnosis|
|Platelet count (×109/l)||aPL||ANA||Type||Months after diagnosis||Platelets (×109/l)||Acute ITP||SLE||APS|
|1||77, M||3||-||0||-||Pulmonary embolism||50||184||-||-||-|
|2||69, M||2||-||0||-||Db, HT, Hc||Deep venous thrombosis||39||180||-||-||-|
|3||61, M||25||-||0||-||Pulmonary embolism||38||135||-||-||-|
|4||44, M||10||+||127||-||Mesenteric occlusion||56||70||-||-||+|
|5||72, F||1||-||0||-||Deep venous thrombosis||2||360||+||-||-|
|7||15, F||11||+||80||+||Pulmonary embolism||69||102||-||-||+|
|8||21, F||4||+||74||+||Oral contraceptive||Deep venous thrombosis||66||117||-||+||+|
|9||52, F||6||-||0||-||Superficial venous thrombosis||6||469||-||-||-|
|10||47, M||5||-||0||-||Db, tobacco, Hc||Stroke||55||200||-||-||-|
|11||33, F||1||-||0||-||Deep venous thrombosis||1||500||+||-||-|
|12||71, M||13||-||0||-||Obesity||Deep venous thrombosis||1||80||-||-||-|
|13||82, M||14||-||0||+||Myocardial infarction||10||32||-||-||-|
|14||16, F||23||+||80||-||Deep venous thrombosis||48||126||-||-||+|
According to our multivariate analysis (Table IV), thromboses were associated with age [hazard ratio (HR), 1·6; 95% confidence interval (CI), 1·2–2·4], LA (HR, 9·9; 95% CI, 2·3–43·4) and high IgG-aCL level (HR, 7·5; 95% CI, 1·8–31·5). It must be emphasized that all the patients with aPL who developed a thrombosis were less than 45 years of age whereas nine of the 10 patients without APL who did have a thrombotic event were over 45 years of age (P = 0·005). In contrast, no association was found between thrombosis and sex, platelet count at ITP diagnosis, acute or chronic outcome, anti-ß2GP1, ANA or therapeutic responses.
|Factor||Univariate analysis||Multivariate analysis|
|HR (95% CI)||P-value||HR (95% CI)||P-value|
|Male sex||2·2 (0·7–6·6)||0·16||1·1(0·3–3·7)||0·93|
|Age||1·4 (1·0–1·8)||0·02||1·6 (1·2– 2·4)||0·002|
|IgG-aCL ≥40 μg/dl||3·3 (1·0–10·9)||0·05||7·5 (1·8–31·5)||0·006|
|LA*||3·1 (1·0–10·4)||0·06||9·9 (2·3–43·4)||0·002|
|Danatrol||2·5 (0·8–7·5)||0·12||1·6 (0·5–5·1)||0·45|
|Splenectomy||3·4 (0·9–13·4)||0·08||2·2 (0·5–9·8)||0·31|
An ITP-associated disease was diagnosed during follow-up in nine patients (SLE, n = 5; primary APS, n = 3; SLE associated with secondary APS, n = 1). SLE was diagnosed significantly more frequently in aPL-positive patients during follow-up (P = 0·04) (Table II).
Antinuclear antibody were sought in 153 (71%) patients and were present in 44 (29%). Among the six patients who developed definite SLE, five had ANA at the time of ITP diagnosis but anti-double stranded DNA and anti-extractable nuclear antigen antibodies were not detected at that time. aPL were detected more frequently in ANA-positive than ANA-negative patients [19/44 (43%) vs. 26/109 (24%) respectively; P = 0·02].
Among the 101 women of child-bearing age, 29 had become pregnant after a median follow-up of 42 (IQR: 10–86) months; 10 were aPL-positive (34%). Five of the women suffered six miscarriages with comparable frequencies for aPL-negative and -positive groups (2% vs. 4%, P = 0·6).
The exact frequency and clinical meaning of aPL in ITP patients are still controversial. In our homogeneous single centre population with newly diagnosed ITP and platelet counts <50 × 109/l, aPL were found in 26% and their presence was not associated with a particular clinical manifestation of ITP.
This frequency is lower than those previously reported (Harris et al, 1985; Stasi et al, 1994; Arfors et al, 1996; Funauchi et al, 1997; Lipp et al, 1998; Diz-Kucukkaya et al, 2001; Bidot et al, 2005, 2006). Pertinently, it must be emphasized that the present investigation is the largest conducted to date for aPL tested at the optimal time of ITP diagnosis, when thrombocytopenia is the most severe, as previously demonstrated by Bidot et al (2005) who showed a tendency of aPL to emerge during ITP exacerbations and decline during remission.
We only searched for aCL, LA and anti-β2GP1 in the sera of recently diagnosed ITP patients because they are the only clinically relevant aPL taken into account when diagnosing APS (Wilson et al, 1999; Miyakis et al, 2006). A higher aPL rate would be observed if other antigens, such as phosphatidylethanolamine, phosphatidylserine, phosphatidylcholine or other protein cofactors, for example prothrombin, protein C, protein S and annexin V, were tested as substrates (Bidot et al, 2005, 2006), but they are not considered clinically meaningful.
The clinical significance of aPL in ITP patients remains controversial. Stasi et al (1994) found aCL and/or LA in 46% of 149 patients with newly diagnosed ITP but were unable to identify any differences between the outcomes of patients with or without aPL. Notably, treatment responses were not affected by the presence of aPL and the rate of thrombotic events was not higher in aPL-positive patients after a median follow-up of 31 months. Using a similar methodology, Diz-Kucukkaya et al (2001) obtained conflicting results in their prospective study on 81 adults, who were followed for a median of 32 months. aPL were present in 38% of their patients at the time of ITP diagnosis. No differences were found between aPL-positive and -negative patients in terms ITP characteristics and responses to treatments, but aPL-positive patients had a much higher thrombosis rate after 5 years (60% vs. 2%). Notably, the risk of thrombosis was strongly associated with LA. These authors concluded that aPL measurement, especially LA, in ITP patients might be helpful to identify a subgroup of patients at high risk of thrombosis (Diz-Kucukkaya et al, 2001). The present study confirmed that aPL were not associated with any particular ITP manifestations at diagnosis, as platelet counts and bleeding scores were comparable in aPL-positive and -negative patients. These results indicated that aPL do not prevent or even lower the risk of bleeding. Responses to steroids and/or splenectomy were also similar regardless of aPL status.
Our observations do not support the hypothesis raised by others (Stasi et al, 1994), that thrombocytopenia could protect patients against thrombosis, since several thrombocytopenic patients developed thromboses. Clinicians should keep in mind that venous and mainly arterial thromboses can occur during ITP, even in an aPL-negative patient.
One of the main questions raised by the detection of aPL in ITP patients is whether their presence puts the patient at higher risk of thrombosis. Reports have been conflicting concerning this issue. Several investigators found little or no risk of a thrombotic complication (Stasi et al, 1994, 1995, BCSH 2003), while others demonstrated that almost half of the aPL-positive patients subsequently developed thromboses, particularly the subgroup of patients who were also LA-positive (Diz-Kucukkaya et al, 2001). Similarly to Diz-Kucukkaya et al (2001), we found that thrombosis was associated with LA and high IgG-aCL level (≥40 μg/dl). Moreover, younger age (i.e. <45 years) at time of thrombosis was significantly correlated with the presence of aPL and a higher risk of developing SLE was established in the subgroup of aPL-positive patients. Even though the risk of thrombosis was low for our population and thrombosis can occur in aPL-negative ITP patients, the strong associations in multivariate analysis among LA, high IgG-aCL level and thrombosis suggest that aPL should be tested in every ITP patient and monitored very closely.
The authors would like to thank Janet Jacobson for editorial assistance, and Drs Catherine Matheron and Chantal André for performing the biological assays.
- 1996) Prevalence of antibodies to cardiolipin in chronic ITP and reactivity with platelet membranes. European Journal of Haematology, 56, 230–234. , & (
- BCSH (2003) Guidelines for the investigation and management of idiopathic thrombocytopenic purpura in adults, children and in pregnancy. British Journal of Haematology, 120, 574–596.
- 2005) Antiphospholipid antibodies in immune thrombocytopenic purpura tend to emerge in exacerbation and decline in remission. British Journal Haematology, 128, 366–372. , , , , , , & (
- 2006) Antiphospholipid antibodies (APLA) in immune thrombocytopenic purpura (ITP) and antiphospholipid syndrome (APS). American Journal of Hematology, 81, 391–396. , , , , & (
- 1995) Criteria for the diagnosis of lupus anticoagulants: an update. On behalf of the Subcommittee on Lupus Anticoagulant/Antiphospholipid Antibody of the Scientific and Standardisation Committee of the ISTH. Thrombosis and Haemostasis, 74, 1185–1190. , , & (
- 2002) Antiphospholipid syndrome: clinical and immunologic manifestations and patterns of disease expression in a cohort of 1,000 patients. Arthritis and Rheumatism, 46, 1019–1027. , , , , , , , , , , , , , , , , , , , , , , , , , , & (
- 1997) Thrombocytopenia in the antiphospholipid syndrome. Annals of the Rheumatic Diseases, 56, 194–196. , , , & (
- 2001) Antiphospholipid antibodies and antiphospholipid syndrome in patients presenting with immune thrombocytopenic purpura: a prospective cohort study. Blood, 98, 1760–1764. , , , , , & (
- 1997) Characteristics of the clinical findings in patients with idiopathic thrombocytopenic purpura who are positive for anti-phospholipid antibodies. Internal Medicine (Tokyo, Japan), 36, 882–885. , , , , , & (
- 1996) Idiopathic thrombocytopenic purpura: a practice guideline developed by explicit methods for the American Society of Hematology. Blood, 88, 3–40. , , , , , , , , , , , , , & (
- 1985) Anticardiolipin antibodies in autoimmune thrombocytopenic purpura. British Journal of Haematology, 59, 231–234. , , , , , , , & (
- 1987) Evaluation of the anti-cardiolipin antibody test: report of an international workshop held 4 April 1986. Clinical and Experimental Immunology, 68, 215–222. , , & (
- 2005) Assessment of a therapeutic strategy for adults with severe autoimmune thrombocytopenic purpura based on a bleeding score rather than platelet count. Haematologica, 90, 829–832. , , , & (
- 2005) The association of thrombocytopenia with systemic manifestations in the antiphospholipid syndrome. Immunobiology, 210, 749–754. , , , , , & (
- 2006) Revision of the Sapporo criteria for the antiphospholipid syndrome: coming to grips with evidence and Thomas Bayes? Thrombosis and Haemostasis, 95, 917–919. , & (
- 2002) The antiphospholipid syndrome. The New England Journal of Medicine, 346, 752–763. , & (
- 1998) Antibodies against platelet glycoproteins and antiphospholipid antibodies in autoimmune thrombocytopenia. European Journal of Haematology, 60, 283–288. , , , & (
- 2006) International consensus statement on an update of the classification criteria for definite antiphospholipid syndrome (APS). Journal of Thrombosis and Haemostasis, 4, 295–306. , , , , , , , , , , , , , & (
- 1994) Prevalence and clinical significance of elevated antiphospholipid antibodies in patients with idiopathic thrombocytopenic purpura. Blood, 84, 4203–4208. , , , , , , , , , , & (
- 1995) Long-term observation of 208 adults with chronic idiopathic thrombocytopenic purpura. American Journal of Medicine, 98, 436–442. , , , , , , , , , & (
- 1982) The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis and Rheumatism, 25, 1271–1277. , , , , , , , & (
- 1999) International consensus statement on preliminary classification criteria for definite antiphospholipid syndrome: report of an international workshop. Arthritis and Rheumatism, 42, 1309–1311. , , , , , , , , , , & (