High prevalence of co-factor independent anticardiolipin antibodies in malaria exposed individuals

Abstract

Anticardiolipin antibodies (aCL) were investigated in 137 individuals chronically exposed to malaria and living in Africa and Asia. They belonged to several groups according to parasite (Plasmodium falciparum or vivax) and clinical manifestations (i.e. asymptomatic parasite carriers, acute uncomplicated attack or severe malaria episodes). aCL were measured in an enzyme immunoassay (ELISA) performed in the presence of either goat serum (aCLs) or gelatin (aCLg). In a group of 53 patients with autoimmune manifestations (i.e. antiphospholipid syndrome and/or lupus), detection of IgG but not IgM aCL was markedly reduced in the presence of gelatin. In malaria donors, high prevalence of serum co-factor-independent IgG and IgM were detected, and the presence of goat serum in the assay consistently decreased their detection. aCLg levels were found to be related to the clinical/endemic status of donors. IgG aCLg were found to be higher in asymptomatic P. falciparum carriers than in patients with uncomplicated acute or cerebral malaria. IgM aCLg were higher in the cerebral malaria group than in groups with uncomplicated acute malaria patients or asymptomatic individuals. Data suggest that using a serum co-factor independent, sensitive ELISA, aCL are commonly detected during malarial infections and related to malarial infection status.

INTRODUCTION

Antiphospholipid antibodies (aPL) are directed against phospholipid components found in cell membranes [1,2]. Anticardiolipin antibodies (aCL), which figure among the best characterized aPL, were reported in a number of autoimmune conditions, and in infections including parasitic diseases such as African trypanosomiasis, schistosomiasis, filariasis and malaria [3–6]. aPL have also been found in some apparently healthy individuals [7,8].

aPL-associated clinical features observed in autoimmune conditions such as the antiphospholipid syndrome (APLS), which includes thomboses and recurrent fetal losses, are usually not present during infections. Differences in antigen recognition by aPL may play a role in clinical presentation, since epitopes recognized in autoimmune conditions differ from those characterized during infectious diseases. In the first group, antibody targets correspond primarily either to complexes associating cardiolipin (CL) and beta2 glycoprotein-I (β2 GPI) or other serum peptide co-factors, or epitopes present on peptide co-factor(s) in a native or PL-modified form [9–12]. During infectious diseases, aPL were generally found to be peptide co-factor-independent and directed against PL alone [6,10,13]. In malaria patients, since early findings of false positive serological tests for syphylis, a limited number of studies have addressed prevalence and role of autoantibodies including aPL, with debatable conclusions [3,5,14–16].

In this context, the present study was aimed at characterizing aCL found during asymptomatic and symptomatic malarial infections by differentiating serum co-factor-dependent and -independent aCL in an ELISA performed with either goat serum or gelatin, respectively. Sera obtained from patients with defined autoimmune diseases were studied under the same experimental conditions. Association of aPL with clinical manifestations was addressed by comparing donors from endemic areas of West and Central Africa and South-east Asia (Myanmar) with defined clinical statuses corresponding to different protection levels. Data suggest that co-factor-independent aCL are common during malaria infections, and related to the severity of clinical symptoms.

PATIENTS AND METHODS

Malaria-exposed individuals

One hundred and thirty-eight patients and/or asymptomatic blood donors were included. All donors had lived from birth in areas with significant malarial endemicity and exhibited parasitaemia at the time of sampling. They belonged to several groups defined according to clinical symptoms [17]. Group A: 36 pa-tients with an uncomplicated acute P. falciparum malaria attack, sampled in Mlomp, Sénégal, a malaria mesoendemic area, during the rainy season (i.e. period of maximal malaria transmission) [18]. Group B: 19 patients with a documented monospecific P. falciparum infection obtained in Kanbauk, Myanmar, a malaria mesoendemic area where transmission is perennial with an increase during the rainy monsoon season, and prevalence of P. falciparum is higher than that of P. vivax. Group C: 20 patients with a documented monospecific P. vivax infection sampled in the same Kanbauk area. Group D: 35 healthy (asymptomatic) inhabitants of Dielmo, Sénégal, a malaria holoendemic area [19]. They were age- and sex-matched with Group A. Group E: 20 patients from the Hôpital Principal, Dakar, Sénégal, with severe (cerebral) malaria defined according to the WHO criteria [20]. All were of African descent and had lived from birth either continuously or during a large part of the year in holo- or mesoendemic areas. Group F: 8 inhabitants of Djoumouna, Congo, an holoendemic area where transmission is perennial and intense, with uncomplicated P. falciparum malaria attack. Additional characterization of malaria-exposed groups are given in Table 1. To explore influence of malarial status and/or endemicity further, patients from all groups were pooled according to their malarial status, i.e. Group 1: asymptomatic carriers, Group 2: acute uncomplicated infection, Group 3: severe malaria attack. Blood was obtained by venous puncture, serum was prepared within 6 h from the time of sampling, and immediately frozen at – 80°C until informed consent was obtained from all participants or children’s guardians.

Table 1.

Characteristics of malaria exposed subjects

Group no.	Geographical origin/ Plasmodium species	No. of subjects	Mean age/ age range (years)	Sex ratio (males/ females)	Asymptomatic donors (no.)	Patients with acute uncomplicated attack (no.)	Patients with cerebral malaria (no.)
A	Mlomp/P. falciparum	36	17·3/3–62	0·59	0	36	0
B	Kanbauk/P. falciparum	19	20·8/8–40	n.a.	12	7	0
C	Kanbauk/P. vivax	20	17·6/2–36	n.a.	8	12	0
D	Dielmo/P. falciparum	35	17·5/4–61	0·59	35	0	0
E	Dakar hospital/P. falciparum	20	n.a.	n.a.	0	0	20
F	Djoumouna/P. falciparum	8	n.a.	n.a.	0	8	0

n.a. = Not available

Patients with autoimmune manifestations

This group consisted of 53 patients with primary or secondary APLS or lupus with or without APLS, according to previously defined criteria [21].

Measurement of anticardiolipin antibodies

ELISA were performed according to previously described procedures, with some modifications concerning (1) the medium used for microplate saturation and serum dilution which consisted of PBS with either 10% goat serum or 1% gelatin and (2) the dilution of sera, i.e. 1: 100 or 1: 400 (or more, up to 1/3200), when goat serum or gelatin was used, respectively [22]. Briefly, microplates (Polysorp, Nunc, Denmark) were coated overnight with CL (Sigma, St Louis, USA) diluted at 50 μg/ml in ethanol (or ethanol alone in controls) by solvent evaporation at 4°C, and blocked with either 10% goat serum or 1% gelatin in pH 7·4 PBS for 1 h at room temperature. After two PBS washes, serum samples (100 μl) diluted as stated above were incubated for 1·5 h at room temperature. After three PBS washes, alkaline phosphatase-conjugated goat antihuman IgG or IgM antibody (Biosys, Compiègne, France) was added at a 1: 1000 dilution. After 1 h incubation at room temperature followed by PBS washing, 100 μl of a p-nitrophenylphosphate (PNPP, Sigma) solution (1 mg/ml) was added. Plates were set at 37°C and colour development monitored at 405 nm. O.D. observed in control wells without CL was substracted from O.D. obtained in CL-coated wells to account for non-specific binding. For assays in the presence of goat serum, a log/log plot of O.D. versus IgG aCL units (GPL) or IgM aCL units (MPL) with six dilutions of a subsidiary standard preparation calibrated with an international standard preparation was used in each microplate (IgG/IgM calibrators, LAPL-GM-001, Louisville APL Diagnostics, Louisville, USA) [23]. For gelatin assays, serum from one patient (PM) which contained 624 GPL units and 5 MPL units was used as a standard preparation to define arbitrary units (AU) (O.D. = 2·5 = 60 AU for both IgG and IgM).

Measurement of anti-β2 glycoprotein-I antibodies

ELISA quantification of β2 GPI antibodies was performed in irradiated polystyrene plates (Maxisorp, Nunc) coated overnight at 4°C with 100 μl per well of a 10μg/ml β2 GPI (β2 GPI, Stago) solution in PBS. After three washes with 0·05% Tween 20 PBS (PBS-Tween), plates were saturated with 1% BSA PBS (PBS-BSA) for 2 h at 37°C, and washed three times with PBS-Tween. Serum samples (100 μl) diluted 1:100 in PBS-BSA were added to antigen-coated (or control, antigen-free) wells and incubated overnight at 4°C. Plates were washed with PBS-Tween, and 100 μl of alkaline phosphatase-conjugated goat antihuman IgG or IgM (Biosys) diluted 1:400 or 1:2000, respectively, was added to each well. After 1h incubation at 37°C, plates were washed, 100 μl PBS with PNPP was added, and O.D. measured as above. Mean O.D. values of control wells were substracted from O.D. observed in β2 GPI-coated wells. For IgG anti-β2 GPI antibodies, standard curves were performed with 1:400–1:25600 dilutions from of control serum from one patient (MA) yielding mean O.D. from 2·5 to 0·2, respectively, corresponding to 800 AU. For IgM anti-β2 GPI antibodies, standard curves were obtained with 1:100–1:6400 dilutions from another patient serum (NB) yielding mean O.D. from 2·5 to 0·1, respectively, corresponding to 200 AU.

Statistical analyses

aPL values were not normally distributed, and for this reason non-parametric tests were used (SIMSTAT software, version 1·3). For variance analysis, Kruskal–Wallis and Mann–Whitney U-tests were used. Wilcoxon’s signed rank test was used to compare data from nine patients tested twice. Correlation was estimated using the Spearman rank coefficient.

RESULTS

Detection of anticardiolipin antibodies in goat serum and gelatin ELISA in patients with APLS and/or lupus

ELISA detection of aCL in the presence of goat serum (aCLs) and gelatin (aCLg) were compared in the group of 53 patients with autoimmune conditions (Table 2). In 47 patients with primary or secondary APLS, detection of IgG aCL (and not IgM aCL) was markedly reduced in the presence of gelatin (5/47, i.e. 10·6%versus 31/47, i.e. 66%, P < 0·001, Table 2). While documenting the presence of aCLg, the low number (six) of patients with lupus in the absence of APLS precluded any conclusion.

Table 2.

Distribution of anticardiolipin antibodies (aCL) detected in ELISA with goat serum (aCLs) or gelatin (aCLg) in 53 patients with APLSand/or lupus

Diagnosis (no. of donors)	Mean age/age range (years)	Sex ratio (males/females)	No. sera with IgM aCLs/No. sera tested (%)	No. sera with IgG aCLs/No. sera tested (%)	No. sera with IgM aCLg/No. sera tested (%)	No. sera with IgG aCLg/No. sera tested (%)
Primary APLS (35)	45/16–81	0·75	10/35 (29)	20/35 (57)	1/35 (3·3)	2/35 (5·5)
APLS secondaryto lupus (12)	33/16–50	0·2	3/12 (25)	11/12 (92)	3/12 (25)	3/12 (25)
Lupus withoutAPLS (6)	36/20–50	0·2	3/6 (50)	2/6 (33)	3/6 (50)	3/6 (50)

ELISA detection of aCLs, aCLg and antiβ2 GPI antibodies in malaria (P. falciparum or vivax) exposed individuals

The distribution of aPL in 137 malaria-exposed donors is summarized in Table 3. Prevalence of aCL detected using goat serum (aCLs) was lower than aCL detected using gelatin (aCLg) for both IgG and IgM (P < 0·001). While no IgG or IgM aCLs was detected in sera from P. vivax-exposed donors in contrast with donors from P. falciparum-prone areas, the ratios of individuals with IgG or IgM aCLg were close in both groups. IgG and/or IgM antiβ2 GPI antibodies were found in less than 20% of sera, with limited influence of Plasmodium species (P < 0·05).

Table 3.

Distribution of anticardiolipin antibodies (aCL) detected in ELISA with goat serum (aCLs) or gelatin (aCLg) 137 malaria-exposed subjects

	IgM aCLs	IgG aCLs	IgM aCLg	IgG aCLg	IgM antiβ2 GP I antibodies	IgG antiβ2 GP I antibodies
No. sera with aCL/No. sera tested * (%)	3/137 (2·2)	4/137 (2·9)	101/119 (84·9)	112/132 (84·8)	9/129 (6·9)	18/129 (13·9)
No. sera with aCL/No. sera tested * (%)(P. falciparum-exposed subjects)	3/117 (2·6)	4/117 (3·4)	83/100 (83)	97/113 (85·8)	7/109 (6·4)	14/109 (12·8)
No. sera with aCL/No. sera tested * (P. vivax-exposed subjects)	0/20 (0)	0/20 (0)	18/19 (94·7)	16/19 (84·2)	2/20 (10)	4/20 (20)

^*Variations in the numbers of tested sera are due to inadequate sample volumes.

Distribution of aCLg antibodies in patients with defined malarial status

A borderline correlation was found between aCLg values and age for IgG (P = 0·06) but not for IgM (P > 0·06). In Groups A and D, no influence of sex of donors was found for both IgG and IgM aPLg (P > 0·2). No correlation was observed between initial P. falciparum or vivax parasitaemia (data not shown) and IgG or IgM aCLg. Univariate analysis of the influence of the country of origin (Groups A, D and E: Sénégal, Group B: Myanmar, Group F: Congo) showed higher IgM, but not higher IgG aCLg levels in Sénégal (P < 0·001), as shown in Fig. 1. In Sénégal, higher levels of IgG aCLg were found in Group D (asymptomatic) than in Group A (acute uncomplicated attacks) or Group E (severe malaria) (P < 0·005). In contrast, IgM aCLg values were lower in asymptomatic than in symptomatic donors (P < 0·025). As shown in Fig. 2, a higher level of median IgM aCLg was found in cerebral malaria (Group 3) than in uncomplicated cases (Group 2; p < 0·005) or in asymptomatic individuals (Group 1; P < 0·0001). IgM aCLg levels were higher in Group 2 than in Group 1 (P < 0·05). At variance with this situation, an opposite trend was observed for IgG aCLg with higher values in asymptomatic (Group 1) than in acute non-complicated cases (Group 2; P < 0·015) or in cerebral malaria patients (Group 3; p < 0·015). No difference in IgG aCLg was found between Group 2 and Group 3 (P > 0·03).

Fig. 1.

Distribution of (a) IgM and (b) IgG aCLg in malaria-exposed individuals. Groups A, B, C, D, E, F as defined in the Patients and methods section. One bar represents the median value. Significant differences: IgM aCL: A versus D (P < 0·025), D versus E (P < 0·0001); IgG aCL: A versus D (P < 0·005), D versus E (P < 0·002), A versus E (P < 0·005).

Fig. 2.

Distribution of (a) IgM and (b) IgG aCLg in P. falciparum malaria-exposed individuals. 1: asymptomatic individuals; 2: uncomplicated P. falciparum malaria attack; 3: severe (cerebral) malaria. One bar represents the median value. Significant differences: IgM aCL: 3 versus 2 (P < 0·005), 3 versus 1 (P < 0·0001), 2 versus 1 (P < 0·05); IgG aCL: 2 versus 1 (P < 0·015), 3 versus 1 (P < 0·015).

DISCUSSION

In this work, detection and potential significance of aCL were investigated in individuals chronically exposed to malaria. Presence or absence of goat serum in the aCL ELISA was found crucial in distinguishing serum co-factor-dependent (aCLs) from co-factor-independent (aCLg) aCL. aCLs were detected in the majority of APLS and/or lupus patients, and a minority of malaria exposed donors. Conversely, serum co-factor independent aCL detected in the presence of gelatin (aPLg) were found in more than 79% of malaria-infected subjects, and less than 20% of autoimmune patients. Results are consistent with the restriction of co-factor-dependent aPL to autoimmune diseases with some exceptions [13,15,24–29]. Presence of goat serum, which contains β2 GPI, considered as largely homogenous to human β2 GPI and is used commonly in aPL ELISA, decreased notably the detection of aCL in malaria sera. This is consistent with inhibition of PL binding to aPL described previously during protozoal infections and presumably due to the interference of peptidic co-factors such as β2 GPI with antigenic interactions of aCL [11,12]. Anti-β2 GPI antibodies were found in a small number of malaria-exposed donors, which confirms that they are usually not detected during infectious diseases, while in APLS they are considered as more specifically associated with clinical manifestations [10,30–33].

A comparatively limited number of studies are currently available on malaria-associated aPL [16]. High prevalence was reported in naive donors and inhabitants of P. falciparum or vivax endemic areas (35% and 34% in patients from Nigeria and South Africa, respectively) [5,32]. Travellers returning to London from Africa presented a 79% prevalence, close from present results [6]. In contrast, prevalence of IgM aPL was lower (50%), and this was mostly marked among P. vivax patients (44%). Highest aPL values were observed for anionic PL such as CL, phosphatidylserine (PS), phosphatidic acid and for neutral PL such as phosphatidylethanolamine (PE) [6]. Discrepancies seem dependent on variations in methodology and assessment of lower aPL detection limits. For this reason, the present methodological choice was to discard from ELISA animal serum which potentially interferes with aPL measurement. In malaria, increase in aPL is due partly to polyclonal B cell activation resulting from lack of T cell control of B cell proliferation and differenciation [34]. A role for Plasmodium-altered membrane PL as antigen has also been advocated, since phosphatidylcholine (PC) is internalized and used by the parasite for its own metabolic needs, while PE and PS are exposed and may induce aPL production [35–37].

In HIV-infected patients, aPL were shown to increase the clearance of apoptotic lymphocytes via macrophage-mediated antibody-dependent cell cytotoxicity, and the same mechanism has been suggested as playing a role in autoimmune diseases where apoptotic cells express early high PS densities which may trigger aPL [38–40]. In addition, PS can associate with plasma peptides such as β2 GPI, to form neoepitopes and, bound to apoptotic cells, aPL could opsonize and help their recognition by macrophages [41]. Similar processes could be triggered by antigenic modifications of the surface of Plasmodium-infected erythrocytes and the generation of aPL may mediate the increase in apoptotic peripheral blood mononuclear cells observed during P. falciparum malaria [42,43].

A malaria-protective role for aPL remains ill-defined [16]. In a mouse model of P. chabaudi chabaudi infection, a dramatic decrease in parasitaemia was observed after immunization with PC but not PE [44]. In man, a role for aPL in antimalarial immunity has been proposed in several studies [6,14,15]. An attractive explanation consists of inhibition of mechanisms leading to TNF-α production, especially during severe malaria [45,46]. In the present study, definite association was found between aPL and clinical/endemic statuses of donors. ACL IgG were found higher in asymptomatic than in acute and uncomplicated and severe malaria patients, and for IgM higher levels were found in the cerebral malaria group. Since all donors had been previously exposed for their whole life to Plasmodia in malaria endemic areas, IgM versus IgG response was not related to a primary versus secondary exposure. Present findings are consistent with a previous report of 65 Gambian children with different symptomatologies in which the levels of antiphosphatidylinositol and anti-PC IgM antibodies were found higher in cerebral than in mild malaria [14]. In 29 Indian patients with cerebral malaria, anti-PC antibody levels were higher in patients who survived than in patients who died, and IgG1 anti-PC antibodies were inversely related to circulating TNF-α, suggesting that they could have a role in inhibiting toxic malarial phospholipid antigens [15]. In contrast, aPL were found lower in cerebral malaria cases than in acute uncomplicated attacks in other studies, suggesting that aPL are responsible for some of malarial complications [6,47]. As stated above, such discrepancies may be due to differences in sensitivity of aPL detection depending on methodology. Present data indicate that, using a serum co-factor-independent sensitive method, aPL appear more frequent than usually stated during malarial infections, and related to malarial infection status.