ANALYSIS OF CEREBROSPINAL FLUID TRANSCRIPTOME TO ELUCIDATE THE AETIOLOGY OF PEDIATRIC ACUTE NON- TRAUMATIC COMA OF UNKNOWN CAUSE IN KILIFI, KENYA

Abstract

Background: Acute non-traumatic coma is a prevalent symptom in childhood illnesses, often linked to high mortality and neurological consequences. While conditions like bacterial meningitis and cerebral malaria contribute, a significant portion of cases, termed "coma of unknown cause," lack clear etiology, posing greater risks. Identifying causes, especially infectious origins, holds vital public health significance. Omics techniques, surpassing traditional diagnostics, drove this study to categorize such cases using transcriptomics and RNA-based metagenomics. Materials and methods: RNA sequencing was utilized to analyze CSF samples from children admitted to KCH between 2002 and 2018, aged two months to 13 years, and diagnosed with aNTC. A pilot sequencing of (n = 145) CSF samples was done, followed by quality control filtration, leaving a total of (n = 100) aNTC cases. Of these, there were: CUC (n = 72), ABM (n = 15), and CM (n = 13) comas cases. The edgeR software was used to identify a transcript profile that could distinguish ABM from CM. Using Kmeans clustering, the transcript profile was then used to detect CM or ABM cases within the CUC group. Clustering results were subsequently validated using plasma PfHRP2, CRP, and LTF. Furthermore, clinical characteristics were compared between and within cases that clustered together. Finally, the CZID platform was used to identify neuro invasive pathogens present in CUC patients. Results: Differential gene expression analysis identified 534 DEGs between acute bacterial meningitis and cerebral malaria. These genes, through K-means clustering, identified a subset of cerebral malaria-like comas of unknown cause within the clinically diagnosed comas of unknown cause. From the 72 clinically diagnosed comas of unknown cause, 40, (55%) had a transcriptome profile like that of children with CM. Notably, the genes upregulated in non-CM-like CUCs were associated with T-cell function, suggesting possible viral infection, while those of CM-like CUC were associated with neurological function. Hemoglobin was significantly lower in the CM- like CUC patients when compared to other CUC cases. The metagenomics approach identified infectious pathogens among children categorized as CUC (48/72), 31 in the CM-like CUC group and 17 in the non-CM-like CUC group. There were (24/40) eukaryotic pathogens in the cerebral malaria-like coma of unknown cause group and (15/32) in the non-CM-like CUC. Interestingly, there were (n = 11) Plasmodium falciparum cases in the CM-like CUC group as compared to (n = 3) in the non-CM- like coma of unknown cause group. For viruses, there were (12/40) cases in the CM- like CUC group and (15/32) cases in the non-CM-like CUC group. The most common viral infections were Cytomegalovirus (n=6), Herpes simplex virus (n=7), and Lymphocryptovirus (n=9), while P. falciparum (n=14) and Aspergillus spp (n=18) were commonly identified eukaryotic pathogens. Conclusion: This study unveils that many coma of unknown cause cases lack bacterial etiology and may involve viral and eukaryotic coinfections, highlighting RNA sequencing's diagnostic potential. Further validation is needed in larger cohorts.