In-silico immune cell deconvolution of the airway proteomes of infants with pneumonia reveals a link between reduced airway eosinophils and an increased risk of mortality

Charles J Sande; Jacqueline M Waeni; James M Njunge; Martin N Mutunga; Elijah Gicheru; Nelson K Kibinge; Agnes Gwela

We then applied the phenotype classification features identified by machine learning to airway proteome data obtained from naso/oropharyngeal samples from children who were admitted to hospital with clinical pneumonia and who either survived and were discharged from hospital(N=61) or ultimately died (N=19) in the course of admission. We also obtained similar data from age-matched well controls, who were sampled from the community (N=10). Using a false discovery rate (FDR) of 5%, we identified >1,000 proteins in the airways of these children. We then compared the expression levels of phenotype-specific feature proteins between these groups. We found that the expression levels of eosinophil-specific classification features were significantly overexpressed in children who survived infection, relative to non-survivors (P<0.0001) and controls ( figure 4B ). This was in contrast to neutrophils, where the expression level of phenotype-specific proteins did not vary by survival status (figure 4B). The expression level of proteins predominantly expressed by naïve CD8 T cells (T8.naive) were also significantly elevated in pneumonia survivors compared to non-survivors (P<0.05). We observed no significant differences for the remaining phenotypes (figure 4C). To validate these findings, we reviewed the clinical records of >10,000 children who had been admitted to Kilifi County Hospital Kenya over an approximate 10-year period with clinical pneumonia and for whom haematological data (including eosinophil and neutrophil counts) had been collected at admission. These data were stratified by survival status and the difference in eosinophils and neutrophil counts between these groups was tested. We found that eosinophil counts in children who survived pneumonia in the retrospective validation cohort (N=10,859) were significantly greater than those of non-survivors (N=1,604, p=0.0004 ) - Figure 5A , first panel. We then evaluated the risk of mortality in the first 10 days after admission in a subset of children who are at an elevated risk of pneumonia mortality (undernourished children), in order to determine the association between eosinophil counts and pneumonia survival in this group. Eosinophils counts in this group were stratified into two groups on the basis of median counts. Undernourished children who were admitted with pneumonia and with low (below median) eosinophil counts, were significantly more likely to die in the first 10 days of admission, relative to those with high eosinophil counts ( Figure 5A , second panel). In contrast, no association with mortality was observed when neutrophils were analysed using a similar strategy ( Figure 5B ).