Children between one and twelve years of age with severe forms of malaria as described previously 30, were consecutively recruited into an unmatched case control study in July-August, 2000 at the Emergency Unit of the Department of Child Health (DCH), Korle-Bu teaching Hospital, Accra, Ghana. Apparently healthy children, randomly selected from a database of children in the same age range from a nearby community, Dodowa were enrolled as control subjects (AC) 2930. Malaria transmission in the study area, a coastal savannah, is perennial with considerable seasonal variation, peaking during and immediately after the rains (May-October). Residents are estimated to receive about 20 infective bites per year and P. falciparum constitutes 98% of all infections 31. The Scientific and Technical Committee of the Noguchi Memorial Institute for Medical Research and the Ethics and Protocol Review Committee of the University of Ghana Medical School, Korle Bu, Accra approved the study. All patients and control subjects were enrolled in the study only after signed, informed, parental consent was obtained.

Patients with axillary temperature > 37.5°C of no other obvious cause than malaria were screened for inclusion by a project physician. Clinical parameters were documented on standard written forms. Spleen enlargement was assessed by palpation and quantified as cm below the left costal margin along the mid-clavicular line. Patients with P. falciparum parasitaemia ≥ 10,000 parasites/μL were enrolled into the study if they fell into one of the following patient categories 29: SA: haemoglobin (Hb) < 5 g/dL, fully conscious with no episodes of severe bleeding, reported or observed convulsions; CM: Blantyre coma score ≤ 3 and duration of coma > 60 minutes, any Hb value and no record of recent severe head trauma and other cause of coma or neurological diseases; IVH: evidence of haemoglobinuria detected by the urine dipstick test (Roche Diagnostics Ltd, Great Britain) followed by microscopy as described elsewhere 21; RD: rapid breathing plus one or more of the following: alar flare, chest recessions, use of accessory muscles for respiration, or abnormally deep breathing; UM: Hb > 8 g/dL, fully conscious, no other features of SM. Patients with Hb between 5 and 8 g/dL were only included if they fell into one or more of the patient categories: IVH, CM and RD.

Patients who were sickling positive were excluded. Other haemoglobinopathies were not taken into consideration. Patients were also excluded from the study based on evidence of other infectious disease such as typhoid or upper respiratory tract infections or any other identified cause of anaemia than malaria. Patients with a history of antimalarial treatment within 2 weeks prior to admission were excluded 1314.

Based on institutional practice at the time, all patients with UM were treated with a standard chloroquine regime at a total dose of 25 mg/kg body weight, given over three days. In the event of treatment failure, treatment with amodiaquine (10 mg/kg body weight per day, as single daily doses for three days) was instituted. All patients with SM were treated with either amodiaquine syrup via nasogastric tube at the same dosage as described, or intramuscular quinine dihydrochloride (10 mg/kg body weight, 8-hourly). Parenteral quinine was changed to syrup at the same dosage when patients regained full consciousness or after 72 hours (which ever was earlier), to complete a 7-day course. Patients with SA or RD were given humidified oxygen and children with Hb < 5 g/dL received blood transfusions 29.

Immediately after admission, 5 ml of venous blood was collected into EDTA tubes from all patients screened for inclusion. Haematological profile was determined using an 18-parameter, automatic haematology analyzer (Sysmex KX-21, Japan). Sickling test was done using the sodium metabisulphite test. Giemsa-stained thick and thin blood films were used for microscopic detection and identification of Plasmodium parasites. Parasites were counted against 300 WBC, and the value was converted into parasites per μL of peripheral blood, based on individual WBC count. Quality control (QC) was ensured by checking the autoanalyser daily and testing of 10% slides by second look. Errors were below 10% variation or disagreement and without systematic deviations.

DCT was done using the method described by Goka et al. 21. RBC suspensions were washed 4 times in comparatively large volumes (4 ml per wash) of 0.9% saline by centrifugation and reconstituted with saline to 5% PCV. One drop of this suspension was mixed with two drops of poly-antiglobulin sera (DIAGAST laboratories, Cedex, France), followed by centrifugation at 1,200 rpm for 1 minute. The samples were immediately inspected macroscopically for agglutination and negative or doubtful positive/weak results were re-examined by light microscopy. The positive samples were re-tested using mono-specific antiserum against IgG and C3d. Test results were graded as +4 (complete agglutination), +3 (several large agglutinates, few cells), +2 (large agglutinates in a sea of smaller clumps and free cells), +1 (many small agglutinates) or 0 (no agglutination). Positive and negative controls were run in parallel.

Packed RBC were washed twice by centrifugation and resuspended in PBS (pH 7.2) to 2.0 × 10⁷ cells/ml. The RBC suspension was stained with ethidium bromide (50 μg/ml final concentration), followed by surface staining with FITC-conjugated antibodies in the dark 32. The following antibodies were used: Mouse IgG₁ FITC (Pharmigen International, 33814X, San Diego, CA, USA); CD35 (Pharmigen, 30961A); IgG (Becton Dickinson, San Jose, CA, USA, 345140{5140}); Mouse IgG_2a FITC (Pharmigen, 33034X); CD55 (Pharmigen, 33571A); Mouse IgG₁ pure (DAKO, Glostrup, Denmark, X 0931); C3bαβ (Cymbus Biotechnology Ltd., Hampshire, UK, CBL 189 for α; CBL 190 for β); Rabbit IgG pure (Zymed Laboratories, INC., San Franscisco, California, USA, 81245172); C3d pure (DAKO, A 0063); C3d FITC (DAKO, F 0323) and controls (Goat anti-mouse FITC (DAKO, F0479); Swine anti-rabbit FITC (DAKO, F0054). After incubation, cells were washed twice with cellwash (Becton Dickinson) by centrifugation and samples were stored at 4°C in the dark till acquisition on a FACScan flow cytometer (Becton Dickinson, Japan). A minimum of 10,000 RBC were acquired and analysis was done by the CellQuest programme. The cut-off point was 10¹ on both axes.

Statistical analysis was performed using SigmaStat software package (Jandel Scientific). Continuous variables were compared between groups using the student t-test or one way analysis of variance (ANOVA). For data that were not normally distributed and could not be normalised by logarithmic transformation a one-way ANOVA on ranks was used. Correlation between parameters was determined by Spearman rank order test. Proportions were compared using the Chi-square test, or Fisher's exact test. Variables that showed significant or near-significant differences between groups by univariate analysis or that were otherwise considered relevant for the study were entered in a conditional logistic regression analysis and in a multiple regression analysis. P-values < 0.05 were considered significant.

A total of 484 parasitaemic patients were screened for enrolment of which 87 patients with distinct clinical presentations of SM were included, 36 SA, 18 IVH, 27 CM and 6 RD. Their clinical characteristics together with those of UM patients and AC are summarised in Table 1. SA patients were younger (mean age = 2.5 yrs.) than the other categories of patients and the AC group (mean age 5.7 yrs.) (p < 0.001). SA had the lowest mean Hb (4.1 g/dL), followed by RD (5.5 g/dL) and IVH (6.5 g/dL). The highest mean Hb was found in AC (10.9 g/dL). All the patient categories had high parasitaemia whilst AC had the lowest parasite density (geometric mean, 1.4 × 10³ parasites/μL). The geometric mean parasite densities of SA and UM cases (63.0 and 57.3 × 10³parasites/μL respectively) were significantly lower than those of the IVH, CM and RD cases (98.7 × 10³, 114.9 × 10³ and 80.9 × 10³ parasites/μL, respectively, p < 0.001). In order to study the effect of complement activation on RD, the data were re-analysed for all patients with SM, including those with overlapping clinical manifestations. The characteristics of these 104 children is summarised in Figure 1. Forty-six children had Hb < 5 g/dL, 34 had coma score ≤ 3, 22 had haemoglobinuria and 20 had RD.

Of all the parameters tested, only age and spleen size showed a significant correlation with Hb (Table 2). Age was positively correlated with Hb (r = 0.379, p = 0.002), and spleen size was inversely correlated to Hb (r = -0.402, p = 0.001).

Of all the patients screened (484), 27.0% (131) were DCT positive. Most of the sensitisation was due to C3d alone (87.8%), with a small proportion being ascribed to IgG alone (2.3%) or to both (9.9%). Children below 5 years formed 84.7% (111/131) of the DCT positives relative to those of 5 years and older, OR = 3.8 (95% CI, 2.2–6.7, p < 0.001). There was a high prevalence of DCT positive cases in the RD (50.0%), IVH (38.8%) and SA (27.8%) patients whereas UM and CM had 9.5% and 11.8% prevalence respectively. Surprisingly, 26.3% of the AC patients had positive DCT (Table 1). When overlapping cases were included in the analysis, RD patients still had the highest DCT positive rate with (50.0%, (10/20)), followed by IVH (45.5%, (10/22)); SA (34.8%, (16/46)) and CM (20.6%, (7/34)), Figure 1. Of all the children with severe forms of malaria (SM, n = 104), 33 (31.6%) were DCT positive (Table 3). There were no significant differences in age and parasitaemia between the DCT positive and the DCT negative patients of the SM (p > 0.05, Table 3), whereas the mean Hb was significantly lower in the DCT positive patients (p < 0.001). Although there was a trend toward higher prevalence of a positive DCT in RD, IVH, and SA, compared with CM, these associations were not statistically significant (Table 3).

Out of 20 RD cases, 50.0% (10/20) were DCT positive compared with 27.4% (23/84) non RD cases, p = 0.092. Furthermore, the strength of the DCT result (0–4) correlated significantly with RD (standard coefficient β = -0.304, p = 0.006). Since children with RD appeared to be more likely to have DCT positive results compared to those without RD, the relationship between DCT result and RD was further investigated using multiple linear regression where the strength of the DCT result, Hb, coma score, and age were tested for their contribution to RD. In this model, Hb (β = -0.341, p = 0.012) and coma score (β = -0.256, p = 0.031, Table 4) emerged as the strongest predictors of RD compared with DCT grade (β = 0.228, p = 0.061) and age (β = 0.208 and p = 0.820). These results suggest that the association between RD and DCT may be related to the role of DCT in precipitating anaemia.

The association between the DCT method and flow cytometry was confirmed by a positive correlation between DCT grade and MFI after surface staining of RBC with anti-C3d (r = 0.53, p < 0.001, Figure 2). In order to study the influence of complement regulatory proteins on clinical outcome, binding of C3d and C3bαβ and expression of CD35 and CD55 on the RBC surface were measured by flow cytometry in a subset of the patients (Table 5). As shown in Table 5, there were no differences in any of the tested parameters between the main clinical groups (p > 0.7). The binding of C3bαβ correlated significantly with CD35 and CD55 (p < 0.001) in children with SM, Figure 3. There was no correlation between C3d and either CD35 or CD55 (p > 0.05, data not shown).