3.1 Sample Characterization
A total of 225 samples from the Northern region of Ghana were used and of these, 36 (16.0%) had co-infection of P. falciparum and HBV (Co-infected), 57 (25.0%) were P. falciparum mono-infected (Malaria only) and 68 (30.0%) were HBV mono-infected (Hep B only) as by PCR (Figure 1) (Anabire et al., 2019b). These groups were compared to 64 (29.0%) control samples that were negative by PCR for both pathogens (uninfected). Age, gestation period and gravidity of the pregnant women were found to not have any significant difference across the various disease groups (supplementary table 2). The demographics of the study participants listed in Supplementary table 2 are also indicated in our previously reported study (Anabire et al., 2019b).
3.2 P-falciparum infection alters sphingolipid serum concentrations of pregnant women
The pathophysiology of malaria diverges in pregnant cases, which implies that infected pregnant women are at an increased risk of infection severity and its complications (Mills et al., 2021). As a result of the increased risk of malaria morbidity, we determined the serum metabolite signatures in P. falciparum infected pregnant women compared to uninfected pregnant women. Principal component analysis (PCA) was performed to visualize how well the samples separated based on the disease groups (malaria only vs uninfected). From the PCA plot it shows that in addition to the variation based on the disease groups, there were also some observed levels of inter-individual variations within the groups as well (Figure 2a). Univariate analysis (T-test and PLS-DA) accounting for false discovery rate was also performed to identify metabolites significantly altered in the malaria mono-infected group compared to the uninfected. From these analyses, a total of 22 out of the 188 metabolites analyzed were found to be significantly altered (Figure 2b), with 13 metabolites (80.4%) out of the 22 being sphingolipids (Figure 2b). Only two of these metabolites; PC aa C32:0 (phospholipid) and C16:1-OH (acylcarnitine) were increased in the malaria group; the rest including the oxidized metabolites indicated with “(OH)”, were all found to be markedly decreased in the malaria group (Figure 2c). The most significantly altered metabolites in the malaria group were the sphingolipids SM (OH) C22:2, SM C24:0, SM C18:0, and SM (OH) C22:1 (Figure 2d). To identify the various pathways impacted, the significantly altered metabolites were further analyzed using a hypergeometric test; an over representation analysis method. Among the malaria group, it was found that most enriched metabolic pathways included pathways relating to lipid and amino acid metabolism such as the phospholipid metabolism and serine metabolism (Figure 2e). Since serine is one of the primary components required for the synthesis of sphingolipids (Wigger et al., 2019), its increased metabolism augments or complements the significantly altered sphingolipid species identified in the univariate analysis.
3.3 Pregnant women infected with HBV have altered amino acid and phospholipid concentrations
The algorithms used in identifying the significantly altered metabolites in the malaria group was also employed in the hepatitis B group as well. An initial PCA analysis revealed that similar to the malaria group, there was some overlap between the hepatitis B group (Hep B only) and the uninfected group (uninfected) (Figure 3a). Upon analysis of the altered metabolites in the hepatitis B group, 74 metabolites were found to be significantly altered (Figure 3b). The 74 significantly altered metabolites in the hepatitis B group, compared to only 22 altered metabolites found in the malaria group, could be due to the fact during infection HBV majorly occupies the liver, the major metabolic organ of the body and causes more liver damage. Whereas P. falciparum, only inhabits hepatocytes at certain stages of its life cycle (Iannacone and Guidotti, 2022; Maier et al., 2019). Of the 74 significantly altered metabolites, phosphatidylcholines were the most significantly altered metabolite group (Figure 3b). All the 22 topmost altered metabolites were found to be lowered in the hepatitis B group (Figure 3c). The most significantly reduced metabolites in the hepatitis B group were SM C24:0; a sphingolipid, and aspartate; an amino acid (Figure 3d). Aspartate concentrations in the body is predominantly regulated by aspartate amino transferase, a liver damage biomarker which we have already reported to be severely altered in the serum of hepatitis B infected pregnant women (Anabire et al., 2019b). Based on the aforementioned- identified metabolites, further enrichment analysis was also done to explore the metabolic pathways most impacted in the hepatitis B group (Figure 3e). As expected, metabolic pathways involving phospholipid biosynthesis was found to be altered. In addition, most of the significantly enriched pathways were those involving amino acid metabolism such as glycine, serine and lysine.
3.4 Co-infection of P. falciparum and HBV induces reduced metabolite serum concentrations of pregnant women
After establishing the metabolites altered in the mono-infection groups and their corresponding pathways, we proceeded to identify the metabolites and pathways most significantly altered in the pregnant women who had P. falciparum and HBV co-infection (co-infected). Compared to the PCA analysis from the mono-infected groups, the co-infected vs uninfected samples were better separated into their groups (Figure 4a). This indicates a unique metabolic profile in the co-infected group induced by both P. falciparum and HBV. Interestingly, the co-infected group vs uninfected group yielded more significantly altered metabolite classes than any of the mono-infected groups (Figure 4b). Comparison of the co-infected group to the uninfected group yielded six significantly altered metabolite classes which comprised predominantly of phosphatidylcholines and sphingolipids (Figure 4c) which were mostly found to be reduced in the co-infected groups (Figure 4d). Further evaluation of the altered metabolites in the various disease groups revealed that there were some commonly altered metabolites among the 3 disease groups. Out of the 188 metabolites analyzed, only 13 were found to be commonly altered in all the three disease groups, which proves the heterogeneity of the three disease groups (Figure 4e). However, majority of the altered metabolites in the co-infected group were found to be altered in the Hepatitis B group as well (Figure 4e; Supplementary table 3). This is not surprising, as it might imply that the co-infected and hepatitis B groups share a dominant liver damage phenotype. Overall, the four topmost metabolites found to be altered in the co-infected group comprised metabolites whose concentrations were also altered in the hepatitis B group: PC ae C40:1, PC ae C42:3 SM C24:1 and SM(OH) C22:2 (Figure 4f). The latter two metabolites (SM C24:1 and SM(OH) C22:2) were altered in the malaria group as well.
Tryptophan metabolism was found to be the most enriched pathway in the co-infected group. (Figure 4g). Previously, we have found increased inflammatory cytokines in the co-infected group which we hypothesized to be caused by an elevated oxidative stress response due to the co-infection of the P. falciparum and the HBV (Anabire et al., 2019b) . The enriched tryptophan metabolism could be due to the intricate crosstalk between the immune response and tryptophan metabolism pathways. (Sorgdrager et al., 2019). Other pathways associated with oxidative stress such as Warburg effect, glycolysis, gluconeogenesis, and lactose metabolism were all found to be enriched. Oxidative stress induces upregulation of the glycolytic pathway, in a metabolic phenomenon known as Warburg effect. Although this is usually observed in cancer cells (Asantewaa and Harris, 2021), it has also been observed to occur in normal proliferating cells as well (Sun et al., 2019a). Both P. falciparum and HBV infections have been associated with increased oxidative stress. Serum measurements of malondialdehyde (MDA), a biomarker for oxidative damage, in HBV patients shows markedly increased levels (Namiduru et al., 2012; Shaban et al., 2014). Likewise, in Plasmodium infections, oxidative stress has been associated with alterations in nutrient availability, iron and heme metabolism which occurs as part of the host defenses to fight infections (Carlos et al., 2018; Gozzelino et al., 2012). These alterations impact the redox balance in vivo and induce oxidative stress, which could account for the enriched redox metabolism pathways found in this study.
3.5 PC ae C40:1 is a good biomarker for P. falciparum and HBV co-infection
Area under Receiver Operating Characteristic (ROC) curve was utilized in evaluating the significance of a metabolite as a good biomarker for the various disease groups. In the malaria group compared with the uninfected group, 15 such potential biomarkers were identified (Supplementary table 4). The most significant biomarker based on the area under curve (AUC) was the ratio of two phosphatidylcholines: PC aa C28:1/ PC aa C32:0. This had an AUC value of 0.76 (Figure 5a). Whiles in the hepatitis B group, out of 9 potential biomarkers identified (supplementary table 5), Aspartate was found to be the most significant, with an AUC of 0.75, (Figure 5b). A total of 56 potential biomarkers were identified in the co-infected group (supplementary table 6) with p-values < 0.01. The metabolites involved included phosphatidylcholines, acylcarnitines and amino acids. The topmost potential biomarker was PC ae C40:1 (Figure 5c).