High throughput genomic and transcriptomic approaches to investigate causes of a Chronic Kidney Disease of unknown etiology (CKDu) in Sri Lanka

Chronic Kidney Disease of unknown etiology (CKDu) was first identified several decades ago as a progressive kidney disease affecting young to middle aged agricultural workers in several locations throughout the world including Guatemala, El Salvador, Sri Lanka and others. Despite a grim prognosis, its underlying cause largely remains a mystery. Multiple hypotheses have been explored (Table 1), yet none currently have substantial support from existing data. 
To further explore (i) whether a (yet-to-be-identified) pathogen may be contributing to CKDu, or (ii) whether a defined physiological/pathological response in kidney tissue might be central in the disease, here we develop a pipeline of high throughput sequencing and data analysis approaches to look for bacterial and viral pathogens that may contribute to CKDu.  We extracted and sequenced DNA and RNA from 45 kidney biopsy FFPE samples and then compared taxonomic classification of the species found in each sample.  Patient samples included cases of CKDu, and AIN (Acute Interstitial Nephritis) from Sri Lankan agricultural communities [1], as well as a variety of non-CKDu kidney diagnoses from Sri Lanka and Stanford (Table 2). 
Our initial observations did not identify a strong bacterial or viral candidate for CKDu. Requiring a minimum of 5 RNA sequencing reads to positively identify a species, we did not observe any species present in Sri Lankan CKDu or AIN patients that were not also observed in negative controls. We were however able to detect human polyomavirus 1 in a single sample (4.5 reads/million) that was independently confirmed as a BK infection.  The low frequency of human polyomavirus 1 reads in this case suggest that we can detect some human pathogens using our methods, but that additional sequencing depth would be needed to detect a pathogenic presence in many of our samples. Patterns of altered native gene expression also evidence the nature of ongoing cellular processes and responses to a pathogenic situation.  The current data (and future more comprehensive sequencing) should begin to provide a window for such analysis and we are beginning to explore this area. 
Biological agents could be associated with disease either directly via human infection, or indirectly via environmental changes and/or specific toxins.  Future work includes environmental sequencing of reservoir water used extensively in farming practices in the Hasakala and Wilgamuwa regions of Sri Lanka to determine if there is a link between specific agents (e.g. individual Cyanobacterial species) and incidence of CKDu cases in patients living and farming nearby.  
This pilot study shows that we can identify some but likely not all pathogens from patient FFPE samples and sets up a framework for additional sequencing to shed light on both host responses to pathogenic events and any bacterial or viral contributors to CKDu.