PATIENTS AND METHODS

Patients. Plasma samples were collected from Kenyan children who were admitted to Kilifi District Hospital with a primary diagnosis of severe malaria, from those with mild malaria who attended the hospital outpatient department, and from a set of age-matched control subjects within the community. Malaria case subjects were defined as those children with P. falciparum parasitemia (asexual P. falciparum parasites detected on a peripheral film stained with 10% Giemsa)—(1) cases of CM were defined in children who could not localize a painful stimulus and for whom other causes of encephalopathy had been excluded [6], (2) cases of malaria plus seizures (M+S) were defined in children with a history of ⩾2 convulsions during the 24-h period before admission, and (3) cases of mild malaria (MM) were defined in children attending the outpatient clinic in Kilifi district hospital who received a primary diagnosis of malaria but had none of the above complications and were not admitted to the hospital. In addition, antibodies were measured in British children with bacterial meningitis (n = 10). All experiments involving humans followed published guidelines of conduct of clinical research [7] and were performed after informed consent of the children's parents or guardians had been obtained. The present study was approved by the Kenya National Ethical Committee.

Methods. Plasma samples were assayed for antibodies to P-/Q- and N-type VGCCs, VGKCs, and muscle nicotinic acetylcholine receptors (nAChRs), as described elsewhere [8, 9]. In brief, human extracts of brain or muscle (5–10 γmol) were labelled with [¹²⁵I]-neurotoxins specific for the channels to be labelled ([¹²⁵I]-q-conotoxin MVIIC, [¹²⁵I]-ω-conotoxin GVIA, [¹²⁵I]-dendrotoxin, and [¹²⁵I]-α-bungarotoxin, which label P-/Q-type VGCCs, N-type VGCCs, VGKCs, and muscle nAChRs, respectively). The labelled extract was incubated overnight in patients' serum samples (5–10 γL), and the complex was precipitated by the addition of goat anti-human IgG serum. The samples were centrifuged, washed, and counted. Antibodies to the γ-aminobutyric acid synthesizing enzyme, GAD, were measured by use of a radioimmunoassay kit developed by RSR. To account for the higher levels of immunoglobulin found in African children, each assay was initially standardized with plasma (n = 10) from healthy age-matched community control subjects. These samples were run in all subsequent assays, and their mean value was subtracted from that of the test samples. The results are expressed as the mean of 3 separate determinations. All assays were performed by investigators who were blinded to the clinical status of the children.

RESULTS

Initially, 149 samples from children with malaria were screened for antibodies to VGCCs (P/Q and N type), VGKCs, GAD, and muscle nAChRs, and their values were compared with the values obtained for community control subjects (n = 40) and British children with bacterial meningitis (n = 10). The number of samples with positive levels of antibody, defined as 3 SD greater than the mean levels for the community control subjects, in each of the assays performed is shown in table 1. The proportion of children with antibodies to both P-/Q- and Ntype VGCCs and the level of specific antibody increased with malaria severity. However, between the different groups, there were no significant differences in the patterns of the other antibodies (VGKCs or GAD). No patients had antibodies to the muscle nAChRs, and no antibodies were detected in children with bacterial meningitis (table 1 and figure 1).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 1.

Anti-voltage-gated calcium channel (VGCC) antibodies in children with malaria and in healthy community control subjects. Plasma samples from patients with cerebral malaria (CM), with malaria and seizures (M+S), or with uncomplicated malaria without seizures (MM) and from community control subjects with (CC+P) or without (CC-P) parasitemia were assayed for anti-VGCC antibodies [9]. For comparison, British children with bacterial meninigitis (BM) were included as an infection control. In brief, [¹²⁵I]-ω-conotoxin MVIIC ([¹²⁵I]-ω-MVIIC), a toxin specific for the P-/Q-type VGCC, was used to label solubilized human cerebellar extracts. The labelled extract was then incubated in plasma (5 γL) overnight, and the complex was immunoprecipitated. The amount of [¹²⁵I]-ω-MVIIC precipitated was expressed as picomoles per liter of plasma, after subtraction of the mean value obtained from the community control subjects (n = 10). Each result represents the mean of 3 separate determinations. Bars show significance values between groups; **P < .001; * P < .002 (Mann-Whitney U test).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 2.

Correlation between levels of anti-P-/Q-type voltage-gated calcium channel (VGCC) antibody and parasitemia in African children with and without malaria. Plasma samples from all children (n = 199) were assayed for anti-VGCC antibodies, and parasitemia was assessed by standard methods. A weak correlation between levels of antibody and parasitemia was detected (r = 0.2; P = .004 [2-tailed Spearman's rank correlation]).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Figure 3.

Correlation between levels of anti-P-/Q-type and N-type voltage-gated calcium channel (VGCC) antibody in children with cerebral malaria (CM). Children with CM were assayed for both P-/Q-type and Ntype VGCCs, by use of [¹²⁵I]-ω-conotoxin MVIIC, a toxin specific for P-/Q-type VGCCs, and [¹²⁵I]-ω-conotoxin GVIA, a toxin specific for N-type VGCCs. A strong correlation between the 2 levels was observed (r = 0.79; P < .0001 [Spearman's rank correlation]).

View larger version:

• In this window
• In a new window

• Download as PowerPoint Slide

Table 1.

No. of children positive (3 SD greater than the mean levels for healthy Kenyan control subjects) for antibodies, with medians, lower confidence limits (LCLs), and upper confidence limit (UCLs).

Samples from an additional 100 children with malaria were then assayed for anti-VGCC antibodies; the combined results for all 249 samples are shown in figure 1. As before, the median titer of each malaria subgroup increased with increasing clinical severity (8, 33, and 53 pmol/L, for MM, M+S, and CM, respectively) and was significantly different from those of community control subjects. In addition, the median titer (38 pmol/ L) of anti-VGCC antibodies was significantly higher in children with malaria and seizures (CM and M+S) than in those without seizures (MM; 8 pmol/L) (P < .002, Mann-Whitney U test). If the community control subjects were further subdivided into those with and those without parasitemia at the time of sampling, then those with parasitemia had a higher median titer than those without parasitemia (-3 and -15 pmol/L, respectively), although the difference did not reach significance. The results for the community control subjects were not significantly different from the results for the British children with bacterial meningitis (0.2 pmol/L). To investigate this further, we compared individuals' levels of parasitemia with their anti- P-/Q-type VGCC antibody titers (figure 2), which, overall, showed a weak but significant correlation (r = 0.2; P = .004 [Spearman's rank correlation]).

Of the 49 children with CM, 35 recovered (mean ± SD antibody titer, 81.5 ± 121 pmol/L), 8 had residual neurological sequelae (mean ± SD antibody titer, 148 ± 133 pmol/L), and 6 died (mean ± SD antibody titer, -29 ± 46 pmol/L). None of the 6 children who died had a positive antibody titer. There was no significant difference between the antibody titers of children with CM who either recovered, had residual sequelae, or died.

A number of different subtypes of VGCCs have been described on the basis of their biophysical and pharmacological properties. We have assessed the samples for P-/Q- and N-type VGCCs. In the present study, all patients with N-type VGCC antibodies also had P-/Q-type VGCC antibodies, and there was a strong correlation between P-/Q-type and N-type VGCC antibody titers (figure 3).

DISCUSSION

Antibodies to VGCCs, but not to other ion channels, are detected, at increased frequency and concentration, in children with malaria. This is not due to a general increase in circulating antibodies or general polyclonal activation; this trend is not observed in any of the other antibody assays. There is a trend for increasing levels of antibody, first with infection per se and subsequently with disease severity. This suggests that the detectable levels of these antibodies may be a marker of current infection and that higher levels of antibody are associated with disease severity. Furthermore, the levels of anti-VGCC antibody were significantly higher in children with malaria and a history of seizures (CM and M+S groups) than in those without seizures (MM). Although the antibody titers were highest in the children with CM, within that group, none of the children who died had detectable levels of anti-VGCC antibodies; however, the numbers of children in each group were too small to be significant.

Autoantibodies to VGCCs are seen in a limited range of autoimmune disorders, including Lambert-Eaton myasthenic syndrome (LEMS) and paraneoplastic cerebellar ataxia [10], but, at present, have not been reported in association with infection. In LEMS, the antibodies are highly specific to P-/Qtype VGCCs: >90% of patients have positive titers, whereas only 30% have anti-N-type VGCC antibodies [9]. It is the P-/ Q-type VGCC that is considered to be important in modulation of neurotransmission in the CNS. Anti-P-/Q-type, but not anti-N-type, VGCC antibodies have been shown to produce pathogenic effects [11]. In contrast, the present study has shown that anti-VGCC antibodies recognize P-/Q-type and N-type VGCCs almost equally, and this correlation may indicate crossreactivity with a common epitope, either in the channel itself or in a closely associated protein. Alternatively, the VGCC is known to exist in a large complex of proteins, including other molecules involved in neurotransmitter release, and, so, the possibility occurs that the antibodies are in fact recognizing other molecules, such as syntaxin and synaptotagmin, that coprecipitate with the labelled VGCCs [12].

The present study has shown a weak correlation between parasitemia and levels of antibody. Peripheral parasitemia is not a good measure of total parasite load in malaria, because of the stage-specific sequestration of infected cells within the deep vasculature. This weak but significant correlation may therefore reflect a much tighter association between parasite load and levels of antibody. The lack of a clear correlation between the antibody titer and the severity of clinical symptoms—for example, some children with CM have low antibody titers—has also been seen in a number of different established autoimmune disorders. For example, in LEMS, the absolute antibody titer does not correlate with clinical severity in the population as a whole, although there is often an inverse relationship between antibody titer and disease severity in an individual.

Several possible mechanisms may be involved in the generation of these seemingly specific antibodies. Since they were not detected in patients with bacterial meningitis, they are unlikely to be a generalized feature of CNS infection. P. falciparum has been reported to induce polyclonal B cell activation, but the lack of antibodies to other channels makes this an unlikely explanation. Alternatively, it is possible that these antibodies are generated to cross-reactive P. falciparum epitopes. Although we cannot eliminate the possibility of cross-reactive conformational epitopes, exhaustive searching of the entire P. falciparum genome sequence (available at: href=http://www.PlasmoDB.org) revealed no primary sequence homology to either P-/Qor N-type channels. Finally, it may be that the impairment of the blood-brain barrier in African children with cerebral malaria [13] and increased levels of IgG [14, 15] may allow an antibody response to normally inaccessible determinants to which the host has not developed tolerance. The seizures themselves may also result in cell damage and subsequent generation of antibodies. However, the absence of antibodies to other CNS components, such as VGKCs and GAD, does not support this. The detection of autoantibodies in the cerebrospinal fluid (CSF) is problematic because of the sensitivity of the assay systems involved. We have assayed CSF from 10 patients with elevated anti-VGCC titers. A titer of 28 pmol/L, which was significantly different from the mean ± SD antibody titer from CSF from African children without serum antibodies (-14 ± 10 pmol/L), was detected in one patient with a serum antibody titer of 240 pmol/L. In a recent study [10] of paired serum/CSF samples from patients with anti-VGCC antibodies and paraneoplastic cerebellar ataxia, anti-VGCC antibodies were detected for 7 of 15 patients; however, the levels of antibody in the CSF (range, 20–400 pmol/L) were ~1/100 those in the serum (range, 100–30,000 pmol/L).

We hypothesize that these autoantibodies, induced by P. falciparum, may be involved in the generation of some of the neurological complications of acute falciparum malaria and/or the subsequent development of complications, such as cerebellar ataxia [16] and epilepsy [2]. Further studies are required to investigate the pathogenicity of these antibodies and to study the relationship between their generation and the subsequent development of epilepsy.