The study was conducted at S.C.B. Medical College, Cuttack, Odisha, India between 2008-2009. Patients (age ≥15 years) admitted to the Department of Medicine with a short history of fever were clinically examined in detail and screened for P. falciparum infection by Giemsa-stained thick and thin blood smears and immune chromatography test (SD Bio Standard Diagnostics India). Detection of P. falciparum was also performed by nested polymerase chain reaction (PCR). Individuals infected only with P. falciparum were included. Clinical categorization was done based on WHO guidelines [¹⁹]. Uncomplicated malaria (UM) was defined as patients with fever and evidence of falciparum infection in the blood. Severe malaria(SM) was categorized into three groups based on distinct clinical features: 1) Cerebral malaria (CM), 2) Non cerebral severe malaria (NCSM) and 3) Multi-organ-dysfuction (MOD). CM was further defined as patients with altered sensorium, GCS (Glasgow Coma Scale) of ≤ 10. NCSM patients had one of the several manifestations of severe malaria without cerebral involvement, namely severe anaemia (haemoglobin <5 g/dl), acute renal failure (serum creatinine >3 mg/dl), jaundice (serum bilirubin >3 mg/dl), acute respiratory distress syndrome (PaO2/FIO2 <200), haemoglobinuria (dark red or black coloured urine positive for haemoglobin) and shock (systolic BP of <80 mm Hg). MOD was diagnosed based on presence of two or more organ involvement like CNS (GCS≤10), respiratory (PaO2/FIO2 <200), renal failure (serum creatinine >3 mg/dl) and hepatic dysfunction (ALT/AST >3 times of normal, prolonged prothrombin time and albuminaemia). 174 healthy controls (HC) of identical ethnicity and hailing from a similar geographical background were enrolled. None of the controls reported history of clinical malaria in the last 5 years. They were essentially healthy and negative for demonstrable P. falciparum infection. The risk of exposure to malaria was similar for both HC and patients. Criteria for analysis of mortality: Since death occurred in the CM and MOD groups' only patients from these groups were included for analysis. The study was approved by the Institutional Ethics Committee of the Medical College and blood samples were collected after written consent of the patients or accompanying person, depending on the clinical scenario.

Genomic DNA was extracted by GenElute™ Blood Genomic DNA Kit (SIGMA) from whole blood according to the manufacturer's instructions. Polymerase chain reaction (PCR), has been used to detect up to species level - falciparum, vivax, ovale and malariae [²⁰,²¹]. In the population under study, P. falciparum is the major cause of malaria (>80%) followed by P. vivax (10-15%) [²²]. With slight modification, genus and species specific nested PCR technique was used to detect P. falcipaurm and/or P. vivax infections [²⁰]. In brief, primary PCR (to detect Plasmodium genus) was performed in a 20 μl PCR reaction containing 3 μL genomic DNA, 1× Taq buffer containg MgCl₂ (SIGMA), 250 μM dNTP (SIGMA)), 200 nM of two genus-specific primer rPLU1 (5'-TCA AAG ATT AAG CCA TGA AAG TGA-3') and rPLU5 (5'-CCT GTT GTT GCC TTA AAC TCC-3') (Integrated DNA technologies) and 1 U of DNA polymerase (SIGMA). The cycle conditions were as follows: an initial de-naturation step at 95°C for 7 minutes, followed by 38 cycles of 95°C for 30 seconds, annealing at 55°C for 30 seconds, extension at 72°C for 1 minute, and a final extension at 72°C for 7 minutes. For secondary PCR (to detect falciparum and/or vivax species), 3 ul of PCR product of previous reaction was used as template and species-specific primers set were used {rFAL1(5'-TTA AAC TGG TTT GGG AAA ACC AAA ACC AAATAT ATT)/rFAL2(5'-ACA CAA TGA ACT CAA TCA TGA CTA CCC GTC-3') and rVIV1(5'-CGC TTC TAG CTT AAT CCA CAT AAC TGA TAC-3')/rVIV2(5'-ACT TCC AAG CCG AAG CAA AGA AAG TCC TTA-3')} in different tubes. Reaction mixture and thermo-cycler condition was similar to primary PCR. PCR products were analysed in agarose gel electrophoresis.

Blood samples of both falciparum-infected individuals and healthy controls were typed by commercial haemagglutination kit (Tulip Diagnostics, Goa, India). In brief, about 20 ul of whole blood was taken on a clean slide and 20 ul of antisera A, B and Rhesus blood group applied, mixed by means of an applicator stick and results were noted.

An extensive PubMed search was performed for articles published on the association of ABO blood group in severe P. falciparum malaria. Cross-references were checked including studies not located in PubMed. The following key words and subject terms were searched: ABO blood group, Plasmodium falciparum, severe malaria and uncomplicated malaria.

Statistical analyses were performed by using GraphPad Prism (version 5.01). Difference between means was analysed by ANOVA. The Fisher exact test was used to analyse the association between blood groups and severe malaria or different clinical manifestation. Odds ratio (OR) were calculated with 95% confidence interval. The 'O' blood group was taken as a reference to compare the prevalence of other blood groups, 'A', 'B', 'AB' and non 'O', defined as A+B+AB combined, in severe and uncomplicated malaria. A probability value of ≤ 0.01 (0.05/4) was considered statistically significant after Bonferroni correction. Meta-analysis and all related statistics were obtained by Comprehensive Meta-analysis V2 software. Variation within and among different study (heterogeneity) was assessed by Q-test with the null hypothesis that all studies have the same effects. The random effects model was used for meta-analysis if the Q-statistic was significant (P < 0.05) which indicate heterogeneity across studies. On the other hand, lineage of all studies to null hypothesis, the fixed effect model, was used for necessary meta-analysis.

A total of 353 falciparum-infected patients were enrolled in the present study, including 247 severe (SM) and 106 uncomplicated patients (UM). Severe P. falciparum malaria patients were further subdivided into cerebral malaria (CM) (n = 59), multi-organ dysfunction (MOD) (n = 80) and non-cerebral severe malaria (NCSM) (n = 108). 174 healthy controls (HC) were included. The mean age between all clinical categories and HC was comparable. Significantly higher levels of haemoglobin were observed in HC than UM and other clinical categories of severe malaria (P < 0.0001) (Table 1).

To assess possible associations of ABO blood groups with severe course of P. falciparum infection, prevalence of each ABO blood group was compared in severe malaria patients (n = 247), uncomplicated malaria (n = 106) and healthy controls (n = 174). Of the 174 blood samples examined in healthy controls, 74 (42%), 36 (21%), 50 (29%) and 14 (8%) individuals were from O, A, B, and AB blood group, respectively. As shown in Table 2, prevalence of B blood group was significantly higher in severe malaria (SM) compared to UM (P < 0.0001; OR = 4.09) and HC (P < 0.0001; OR = 2.79). Furthermore, blood group 'O' was significantly associated with UM compared to the severe categories (P < 0.0001; OR = 2.81) indicating its likely protective role (Table 2). The distribution of blood groups among HC and UM was comparable.

The association of ABO blood group in different clinical phenotypes of severe malaria, categorized as CM, NCSM and MOD, was studied. As highlighted in Table 3, prevalence of blood group 'B' was significantly higher in CM (P < 0.0001; OR = 5.95), MOD (P < 0.0001; OR = 4.81) and NCSM (P = 0.001; OR = 3.02) compared to UM. Furthermore, the protective nature of group 'O' was observed when comparing its prevalence in UM vis-a-vis the different severe groups (CM: P = 0.0003; OR = 3.77, MOD: P = 0.0002; OR = 3.31 and NCSM: P = 0.007; OR = 2.81).

Since the study revealed a significant association of ABO blood group and severe falciparum malaria, possible role of ABO blood groups in disease outcome was explored. Prevalence of blood groups were analysed among subjects who died during treatment and those who survived Out of 247 cases of severe malaria, twenty patients died during course of treatment: five patients from CM category (n = 59), 15 patients from MOD (n = 80) and none from NCSM (n = 108). Therefore, NCSM patients were not included in the analysis. Furthermore, fifteen patients from CM and MOD groups left the hospital against medical advice, which necessitated their exclusion since their survival status was not known. As a result, a total of 124 patients in CM and MOD groups were analysed. Although, frequency of blood group 'B' was higher in patients who died (80%) compared to those who survived (54%), the difference was not statistically significant (P = 0.06). Distribution of other blood groups in patients who died and survived was comparable (Table 4).

Nineteen relevant publications were identified on falciparum malaria and ABO blood group. The three primary criteria for inclusion in this meta-analysis were as follows:1) study must be case-controlled, 2) patients placed under well defined categories as severe and uncomplicated, 3) the sample size should be more than 50 in each category. Out of nineteen studies only five publications satisfied all inclusion criteria and were considered for meta-analysis. Data from the recent study was also included for analysis. Table 5 shows characteristics of all six studies.

A funnel plot has been used to test publication bias in meta-analysis. The funnel plot remains symmetric in absence of publication bias. For 'O', 'B' and 'AB' blood groups the funnel plots were symmetric. However, plot for blood group 'A' was asymmetric. Trim-and-fill technique was used to adjust publication bias of blood group 'A', for recomputing the effect size [²³].

Heterogeneity Q test was performed to evaluate inter- and intra-study variations and based upon significance value; different models were used for the meta-analysis. Blood group 'O' (P < 0.0001), 'B' (P = 0.004) and 'AB' (P = 0.037) showed significant heterogeneity and, therefore, a random-affect model was employed. On the other hand, a fixed-effect model was used for association of blood group 'A' since the Q-test was not significant (P = 0.120).

Results of meta-analysis are shown in Figure 1. Protection against severe malaria was significantly associated with blood group 'O' (P = 0.01). In contrast, blood group 'A' and 'AB' showed significant association to susceptibility ('A': P = 0.04 and 'AB': P = 0.04). Although, 'B' blood group was significantly associated with severe malaria in the present study, the meta-analysis failed to corroborate that association.

SM: severe malaria; CM: cerebral malaria; MOD: multi-organ dysfunction; NCSM: non-cerebral severe malaria; UM: uncomplicated malaria; HC: healthy control; NA: not applicable; NS: not significant

NOTE: Data are no. (%) of participants unless otherwise specified.

SM: severe malaria; UM: uncomplicated malaria; HC: healthy controls; OR: odds ratio; CI: confidence interval

NOTE: Data are no. (%) of participants unless otherwise specified.

CM: cerebral malaria; MOD: multi-organ dysfunction; NCSM: non-cerebral severe malaria; UM: uncomplicated malaria; OR: odds ratio; CI: confidence interval

NOTE: Data are no. (%) of participants unless otherwise specified;

OR: odds ratio; CI: confidence interval

NOTE: Data are no. (%) of participants unless otherwise specified.

SM: severe malaria; UM: uncomplicated malaria;