American Journal of Hematology 57:139–143 (1998)

Anticardiolipin Antibodies in Acute Myeloid Leukemia:

Prevalence and Clinical Significance
Izidore S. Lossos,* Vered Bogomolski-Yahalom, and Yaacov Matzner
Department of Hematology, Hadassah University Hospital, Jerusalem, Israel

This prospective study was designed to explore the prevalence and the clinical and
prognostic significance of anticardiolipin (ACL) antibodies in patients with acute myeloid
leukemia (AML). The study includes 37 consecutive AML patients >15 years old without
previous history of thromboembolism, recurrent fetal loss, or autoimmune disease and
with no evidence of infection at the time of enrollment. ACL antibodies were found in 25
patients (68%). None of the patients had high positive titers; 8 had moderately positive
while 17 had low positive ACL antibody titers. ACL antibody positivity did not predict
response to chemotherapy and was not correlated with age, gender, FAB class, platelet
and white blood cell counts at presentation, and the risk of thromboembolism. ACL
antibody titers did correlate, however, with AML activity in the majority of patients (93%)
during 4–19 months of follow up. These results demonstrate a high prevalence of ACL
antibodies in AML patients and suggest that serum ACL antibodies may be a useful
adjunct in predicting relapse and documenting disease activity and therapy response.
Am. J. Hematol. 57:139–143, 1998. © 1998 Wiley-Liss, Inc.

Key words: anticardiolipin antibodies; acute myeloid leukemia; lupus anticoagulant

INTRODUCTION small series of patients with acute myeloid leukemia

(AML) was reported [12]. The purpose of this prospec-
Antiphospholipid (APL) antibodies are a heteroge- tive study was to determine the prevalence and the clini-
neous family of autoimmune and alloimmune immuno- cal and prognostic significance of ACL antibodies in
globulins, directed predominantly against protein- patients with AML.
phospholipid complexes [1,2]. APL antibodies include
lupus anticoagulant (LA), which prolongs phospholipid
dependent coagulation in vitro and anticardiolipin (ACL) MATERIALS AND METHODS
antibodies detected by solid phase immunoassay. Be- Patients
cause in some patients there is a lack of concordance
between the results of these tests, most investigators now The study population consisted of consecutive patients
agree that LA and ACL antibodies define separate sub- with AML hospitalized in the Hematology Department,
groups of antibodies [2–4]. APL antibodies have been Hadassah University Hospital, Jerusalem, between June
detected in patients with systemic lupus erythematosus 1995 and October 1996; follow-up continued through
and other autoimmune diseases, infections, drug-induced March 1996. Patients were eligible for this study if they
conditions, and in approximately 10% of the healthy suffered from untreated, de novo or secondary, AML, or
population without predisposing factors [5,6]. ACL an- relapsed AML following at least 6 months of unmain-
tibodies have been found in 22% of patients with solid tained remission. Diagnosis of AML was established ac-
malignancies who also suffered from an increased risk of cording to the French-American-British (FAB) criteria
malignancy associated thrombosis [7]. Only sporadic re- after examination of hematoxylin-eosin stained smears of
ports of the association of APL antibodies (e.g., LA or
ACL) with hematological malignancies such as non-
Hodgkin’s lymphoma [8], plasma cell dyscrasia [9], *Correspondence to: I.S. Lossos, M.D., Department of Hematology,
Hadassah University Hospital, P. O. Box 12000, Jerusalem IL-91120,
hairy cell leukemia [10], and acute lymphoblastic leu- Israel. E-mail: [email protected]
kemia [11] were published. Also a single investigation
describing prevalence of 26% of APL antibodies in a Received for publication 16 July 1997; Accepted 27 August 1997
© 1998 Wiley-Liss, Inc.
 10968652, 1998, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/(SICI)1096-8652(199802)57:2<139::AID-AJH8>3.0.CO;2-X by CochraneChina, Wiley Online Library on [05/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
140 Lossos et al.
peripheral blood and bone marrow and cytochemical TABLE I. Characteristics of Anticardiolopin (ACL) Antibody
Positive and Negative Patients With Acute Myeloid Leukemia
studies. Immunophenotypic flow cytometry and cytoge-
netic studies were also performed. Patients were ex- ACA-positive ACA-negative
cluded from the study if they had a previous history of Characteristic (25 patients) (12 patients)
thromboembolism, recurrent fetal loss, or an autoim- Age
mune disease, if their age was less than 15 years; if they Range 15–77 19–71
had been treated with immunosuppressive agents, im- Mean 44 47
muno- or chemotherapy within the previous 6 months or Sex
Male 12 8
if they had clinical, microbiologic, or roentgenographic
Female 13 4
evidence of infection at the time of the ACL antibody test. FAB
Patients up to the age of 65 years were treated by an M–0 2 –
induction regimen consisting of continuous infusion of M–1 5 2
cytarabine 100 mg/m2/day for 7 days and daunorubicin M–2 5 1
M–3 1 1
45 mg/m2/day for 3 days. If remission was achieved, two
M–4 2 3
consolidation courses consisting of high-dose cytarabine M–5 5 1
(3 g/m2 twice daily for 6 days) and a course of mitox- Secondary 5 4
antrone 12 mg/m2/day for 3 days with etoposide 100 Thrombosis 1 1
mg/m2/day for 5 days were administered. Patients aged Chemoresistance 4 1
51–65 years were treated with a reduced dose of cy-
tarabine (2 g/m2). Patients with acute promyelocytic leu- Statistical Analysis
kemia were treated with all-trans retinoic acid 45 mg/m2
until remission induction, before institution of the above The differences between the ACL antibody-positive
regimen. Patients who failed to achieve complete remis- and ACL antibody-negative patients were evaluated us-
sion (CR) or patients with relapsing AML were treated ing the chi-square or Fisher’s exact tests when appli-
with a salvage protocol consisting of cytarabine 3 g/m2/ cable. A P value of less than 0.05 was defined as statis-
day for 5 days and mitoxantrone 20 mg/m2/day for 2 tically significant.
days. Eligible patients with poor prognostic factors or
following disease relapse were referred for bone marrow RESULTS
transplantation (BMT). The therapeutic regimen in pa-
tients aged 66–77 years varied according to their general Thirty-seven AML patients were eligible for this
health condition. study. Their clinical characteristics are presented in
Table I. ACL antibodies were found in the serum of 25
Antibody Assay patients (68%). They were of IgG, IgM, or both types in
Laboratory evaluation of serum ACL antibody titers 11, 6, and 8 patients, respectively (Table II). None of the
was performed at diagnosis, before every chemotherapy patients had high positive ACL titer. Moderately positive
course, at 3-month intervals following completion of IgG- and IgM-ACL antibody titers were detected in 3 and
chemotherapy, and at relapse. Ig G and Ig M ACL anti- 5 patients, respectively, while the majority of patients
bodies were measured in serum stored at −20°C by an had low positive titers. ANAs were found in a single
enzyme linked immunosorbent assay (ELISA), using a patient with ACL antibodies. As shown in Table I, no
commercially available kit (Selisa™, Cambridge Life correlation was found between the presence of ACL an-
Sciences plc, Cambridgeshire, England). The assay was tibodies and the following parameters: age, gender, FAB
calibrated against an international standard, one unit be- class, chemoresistance, platelet and white blood cell
ing defined as the cardiolipin-binding activity of 1 mg of count at presentation, and the length of remission (data
affinity-purified IgG/IgM-ACL antibody [13]. All ACL not shown). Thrombotic phenomena (cerebrovascular ac-
antibody determinations were performed blindly at two cidents) were observed in one ACL-positive and one
different runs, including other unknown sera. The normal ACL-negative patient.
levels were defined by measuring IgG- and IgM-ACL Fifteen ACL antibody positive patients were followed
antibody titers in controls. Results were considered posi- for 4 to 19 months (Fig. 1). Ten patients had chemo-
tive when the values exceeded the mean controls by 3 therapy-responsive leukemia and had achieved remission
standard deviations (>7.5 u/ml for both IgG- and IgM- lasting for at least 3 months (Fig. 1R), while 5 patients
ACL antibodies). Titers between 7.5 and 20 u/ml were had chemotherapy-resistant (4 patients) or rapidly relaps-
considered low positive, between 20 and 80 u/ml mod- ing (1 patient) AML (Fig. 1U). The number of follow-up
erate positive, and over 80 u/ml high positive. Anti- ACL antibody measurements in Figure 1 is 19, since 4
nuclear antibodies (ANAs) were determined by indirect patients (two in Fig. 1R and two in Fig. 1U) had both
immunofluorescence on Hep-2 cell slides. IgG- and IgM-ACL antibodies and are shown in both
 10968652, 1998, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/(SICI)1096-8652(199802)57:2<139::AID-AJH8>3.0.CO;2-X by CochraneChina, Wiley Online Library on [05/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Anticardiolipin Antibodies in AML 141
TABLE II. Serum Anticardiolipin (ACL) Antibody
Concentrations in Acute Myeloid Leukemia Patients

Antibody titer
ACL Number of
antibodies patients Mean ± SD Range

IgG 11 12.6 ± 5.1 8.0 − 25.0

IgM 6 27.3 ± 16.2 8.0 − 45.0
IgG and IgM 8
IgG 15.9 ± 4.2 9.0 − 21.5
IgM 19.0 ± 13.7 9.0 − 54.0

curves. In 5 patients with chemoresistant or rapidly re-

lapsing leukemia (within 3 months of therapy termina-
tion), ACL antibodies were persistently positive (Figs.
1U, 2A). One of these patients with chemoresistant-
leukemia had both IgG- and IgM-ACL antibodies at di-
agnosis. During chemotherapy IgM-ACL antibodies dis-
appeared, while IgG-ACL antibodies persisted.
Ten patients achieved remission, lasting at least 3
months following completion of chemotherapy. In eight
patients ACL antibody titers fell below 7.5 u/ml follow-
ing achievement of morphologic and cytogenetic CR
(Figs. 1R 2B, C). Five of those patients suffered from
AML relapse, accompanied by reappearance of ACL an-
tibodies in four (Fig. 2B, representative case). Three pa-
tients have remained in CR for 4 to 13 months with
persistently negative ACL antibodies (Fig. 2C, represen-
tative case). An additional two patients, in chemo-
therapy-maintained second CR following initial relapse,
had persistently positive ACL antibodies for 8 and 10
months, respectively (Fig. 1R). One of these patients had
AML relapse, while the other is still in maintained sec-
ond CR.
Among the patients with baseline negative ACA anti-
bodies, one patient with secondary AML refused chemo- Fig. 1. Serum IgG (A) and IgM (B) anticardiolipin (ACL) an-
therapy. During the follow-up she became persistently tibody titers at diagnosis, before treatment of acute myeloid
ACL-positive. In four patients, a transient ACL positivity leukemia (BT) and following chemotherapy (AT), in 5 pa-
coincidental with infection was observed. tients with either chemoresistant or rapidly relapsing leu-
kaemia (U) and in 10 chemotherapy responding patients (R).
Four patients had both IgG- and IgM-ACL antibodies. Pa-
tients with both IgG- and IgM-ACL antibodies are presented
DISCUSSION
by open rhombi, patients with IgM-ACL antibodies by
Increased incidence of certain autoantibodies was closed circles, and patients with IgG-ACL antibodies by
closed squares.
documented in patients with malignancy [14]. For ex-
ample, in patients with carcinoma of breast, the presence
of autoantibodies is associated with a less favorable results of different laboratory methods even in the same
prognosis [15]. There is also growing evidence on the patient [2–4]. Therefore, patients with LA antibodies
association of APL antibodies (e.g., LA and ACL anti- who are ACL antibodies negative and vice versa, may be
bodies) with malignancy [7,8]. These antibodies, directed identified.
predominantly against complexes of negatively charged A 22% prevalence of ACL antibodies was described in
phospholipids with protein, are commonly found in au- malignancy with an increased risk of thromboembolism
toimmune disorders and are characteristic of the primary [7], but ACL antibodies were uncommonly reported in
anti-phospholipid syndrome [5,16]. It has to be empha- hematological malignancies. To the best of our knowl-
sized, that APL antibody population is very heteroge- edge, there is a single study demonstrating LA and ACL
neous (e.g., LA, ACL, b2 glycoprotein 1 [b2GPI] anti- antibody prevalence of 26% in a small group of 19 pa-
bodies), resulting in a lack of concordance among the tients with de novo AML [12].
 10968652, 1998, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/(SICI)1096-8652(199802)57:2<139::AID-AJH8>3.0.CO;2-X by CochraneChina, Wiley Online Library on [05/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
142 Lossos et al.

Fig. 2. Follow-up of serum anticardiolipin (ACL) antibody levels in representative cases receiving chemotherapy for acute
myeloid leukemia (AML). A: Patient with chemoresistant AML. B: Patient who achieved a complete remission (arrow)
following chemotherapy, but subsequently relapsed (*). C: Patient who achieved a long-standing complete remission
(arrow) following chemotherapy.

To further elucidate the possible association between coincidental association between ACL antibodies and
ACL antibodies and AML and its clinical significance, AML.
we performed this prospective study in a larger group of A similar pattern of ACL antibody response to treat-
AML patients. Our study demonstrates a 68% prevalence ment of underlying malignancy was reported in patients
of ACL antibodies in this population. This prevalence with non-Hodgkin’s lymphoma and solid tumors [7,8].
markedly exceeds the reported 10% prevalence in the However, the etiopathogenetic relationship between AML
healthy population [6]. Patients with disorders known to and ACL antibodies is not clear. Several mechanisms
be associated with ACL antibodies, such as autoimmune may be suggested for ACL antibodies production: (1) a
diseases, infections, thromboembolism, and recurrent fe- change in the membrane of leukemic cells may result in
tal loss, were excluded from the study, thus suggesting exposure of protein-phospholipid complexes leading to
that ACL antibodies in our patients are related to the immune system response. Plasma membrane alterations
basic disease, AML. Furthermore, the persistence of have been reported in leukemic patients, especially in
ACL antibodies in patients with resistant disease and chronic myeloid leukemia [17,18]; (2) leukemic cells
early relapse, its disappearance following achievement of may secrete protein-phospholipid complexes or its com-
CR and resurgence following relapse may suggest a non- ponents with subsequent complex assembly in plasma,
 10968652, 1998, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/(SICI)1096-8652(199802)57:2<139::AID-AJH8>3.0.CO;2-X by CochraneChina, Wiley Online Library on [05/08/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Anticardiolipin Antibodies in AML 143
initiating an immune response; (3) production of mono- REFERENCES
clonal immunoglobulins with LA or ACL activities by
1. Arnout J: The pathogenesis of the antiphospholipid syndrome: A hy-
malignant cells. Although such immunoglobulins have pothesis based on parallelisms with heparin-induced thrombocytope-
been reported in patients with lymphoma [8] and plasma nia. Thromb Haemost 75:536, 1996.
cell dyscrasia [9,19], this mechanism is unlikely to apply 2. Harris EN: Antiphospholipid antibodies. Br J Haematol 74:1, 1990.
to a non-B cell malignancy like AML. 3. Derksen RHWM, Hasselaar P, Blokzijl L, Meyling FHJG, DeGroot
PG: Coagulation screen is more specific than the anti-cardiolipin an-
The structure of the protein-phospholipid antigen com-
tibody ELISA in defining a thrombotic subset of lupus patients. Ann
plexes to which ACL antibodies are directed in AML Rheum Dis 47:364, 1988.
patients is unclear. Whether ACL activity in these pa- 4. Sarice M, Cicella A, Griggi T, Garofalo T, Nicodemo G, Pittoni V,
tients is dependent on the presence of prothrombin, Pontieri GM, Lenti L, Valesini G: Anticardiolipin and anti-b2-GPI are
b2GPI, or other proteins is unclear. Several studies had two distinct populations of autoantibodies. Thromb Haemost 75:303,
1996.
demonstrated that the b2GPI-dependent binding to phos- 5. Love PE, Santoro SA: Antiphospholipid antibodies: Anticardiolipin
pholipids can be used to discriminate between autoim- and the lupus anticoagulant in systemic lupus erythematosus (SLE)
mune ACL antibodies and those found in patients fol- and in non-SLE disorders. Ann Intern Med 112:682, 1990.
lowing infections [20]. The presence of b2GPI in AML 6. Shi W, Krilis SA, Chong BH, Gordon S, Chesterman CN: Prevalence
patients was not yet tested. Further studies are needed to of lupus anticoagulant and anticardiolipin antibodies in a healthy popu-
lation. Aust NZ J Med 20:231, 1990.
elucidate the mechanisms and the antigenic nature of 7. Zuckerman E, Toubi E, Dov Golan T, Rosenvald-Zuckerman T, Sabo
ACL antibodies in AML patients. E, Shmuel Z, Yeshurun D: Increased thromboembolic incidence in
The association between thrombosis and ACL anti- anti-cardiolipin-positive patients with malignancy. Br J Cancer 72:
bodies is well established in patients with primary anti- 447, 1995.
8. Ciaudo M, Horellou MH, Audouin J, De Carbonnieres C, Conard J,
phospholipid syndrome and autoimmune diseases [5].
Samama M: Lupus anticoagulant associated with primary malignant
Recently, an increased risk of thromboembolism had lymphoplasmacytic lymphoma of the spleen. Am J Hematol 38:271,
been observed in ACL-positive patients with solid ma- 1991.
lignancies [7]. In our and Stasi et al.’s [12] AML pa- 9. Bellotti V, Gamba G, Merlini G, Montani N, Bucciarelli E, Stoppini
tients, the presence of ACL antibodies was not associated M, Ascari E: Study of three patients with monoclonal gammopathies
and lupus-like anticoagulants. Br J Haematol 73:221, 1989.
with an increased risk of thromboembolism. This may 10. Duncombe AS, Dalton RG, Savidge GF: Lupus-type coagulation in-
result from either low ACL antibody titers or due to hibitor in hairy cell leukemia and resolution with splenectomy. Br J
thrombocytopenia, frequently found in AML patients. Haematol 65:120, 1987.
The relatively small study population size can also be an 11. Donner M, Bekassy NA, Garwicz S, Holmberg L, Wiebe T: Cerebral
explanation, since thromboembolism is a rare phenom- infarction in a girl who developed anticardiolipin syndrome after acute
lymphoblastic leukemia. Pediatr Hematol Oncol 9:377, 1992.
enon in the AML population. 12. Stasi R, Stipa E, Masi M, Oliva F, Sciarra A, Perroti A, Zaccari G,
Our study suggests that the repeated measurement of Papa G: Antiphospholipid antibodies: prevalence, clinical significance
serum ACL antibodies may be a useful adjunct in docu- and correlation to cytokine levels in acute myeloid leukemia and non-
menting disease activity and response to therapy in AML Hodgkin’s lymphoma. Thromb Haemost 70:568, 1993.
13. Harris EN, Gharavi AE, Patel SP, Hughes GRV: Evaluation of the
patients. Disappearance of ACL antibodies accompanied
anti-cardiolipin antibody test: Report of an international workshop
achievement of CR, while its resurgence predicted AML held 4 April 1986. Clin Exp Immunol 68:215, 1987.
relapse. ACL antibody persistence was observed in pa- 14. Tannenberg AEG, Muller, HK, Cauchi MN, Nairin RC: Incidence of
tients having early relapse or suffering from resistant autoantibodies in cancer patients. Clin Exp Immunol 15:153, 1973.
leukemia. In addition, ACL antibodies persisted in two 15. Wasserman J, Glas U, Blomgren H: Autoantibodies in patients with
carcinoma of breast. Clin Exp Immunol 19:417, 1975.
patients during second CR, a situation with a high relapse
16. Devine DV, Brigden ML: The antiphospholipid syndrome. Postgrad
rate. It has been suggested that sustained antitumor im- Med 99:105, 1996.
mune response may be due to ‘‘dormant’’ tumor cells 17. Ackerman GA: Ultrastructural histochemical alteration of the plasma
that provide continued antigenic stimulation [21]. membrane in chronic myelocytic leukemia. Blood 63:869, 1975.
Persistence of ACL antibodies during CR in patients 18. Baker MA, Taub RN, Whelton CH, Hindenburg A: Aberrant sialyla-
with early relapse or in second CR may be a result of tion of granulocyte membranes in chronic myelogenous leukemia.
Blood 63:1194, 1984.
such low-grade antigenic stimulation due to a minute 19. Gleicher M, Friberg, J: IgM gammopathy and the lupus anticoagulant.
number of tumor cells that still have potential for relapse JAMA 253:3278, 1985.
but are morphologically undetectable. 20. McNally T, Purdy G, Mackie IJ, Machin SJ, Isenberg DA: The use of
In conclusion, our findings demonstrate a high preva- an anti-b2-glycoprotein-I assay for discrimination between anticardio-
lence of ACL antibodies in the serum of AML patients. lipin antibodies associated with infection and increased risk of throm-
bosis. Br J Haematol 91:471, 1995.
ACL antibodies were not correlated with particular clini-
21. Khazaie K, Prifi S, Beckhove P, Griesbach A, Russell S, Collins M,
cal features, but might be a reliable marker of disease Schirmacher V: Persistence of dormant tumor cells in the bone marrow
activity. Further studies clarifying the etiopathogenetic of tumor cell vaccinated mice correlates with long term immunologic
relation between AML and ACL antibodies are needed. protection. Proc Natl Acad Sci USA 91:7430, 1994.