Evaluation of next-generation sequencing technology in determining infectious causes of fever

Undifferentiated fever (UDF) is a common complaint in clinical practice, but its aetiology is not always determined due to non-specific symptoms and laboratory findings. While fever of unknown origin (FUO) is a common medical term for fever without obvious cause, this condition is distinguished from acute undifferentiated fever (AUF) in terms of duration, progression of illness and underlying causes. In FUO, fever must exist for more than 3 weeks and can persist for a very long period unless the underlying cause is found and eliminated. In contrast, AUF is more limited in duration and many episodes spontaneously resolve, presumably due to self-limiting infectious diseases.\ud \ud The problem of determining the infectious causes of fever has received considerable attention, particularly in tropical countries. Previous studies in South and Southeast Asia reported high prevalence of infection as the main cause of AUF. This prompted the hypotheses that infection-related AUFs are common in the tropical region of Far North Queensland, Australia, and that a significant proportion of AUFs in this region are undiagnosed.\ud \ud Diagnosing infectious causes of fever is a challenge for clinicians. With hundreds of possible agents and a limited number of specific tests that can be performed, it is very likely that doctors will miss the true cause of fever. Moreover, current diagnostic approaches rely on prior knowledge of the pathogens being sought, thus precluding the detection of unsought or novel pathogens. Thus, the prevalence of undiagnosed undifferentiated fever (UUDF) indicates either that clinicians are failing to order appropriate tests, that current diagnostic methods are inadequate, or that there are causes of fever that are yet to be discovered.\ud \ud This diagnostic challenge necessitates good clinical skills, knowledge of the pattern of signs and symptoms associated with particular infections and a broad diagnostic tool for determining infectious causes of fever. Since there is a wide range of pathogens, the diagnostic tool should be able to distinguish one pathogen from another as well as identify multiple pathogens simultaneously. This can be achieved if the diagnostic tool can 'read' unique characteristics of pathogen(s) present in the sample as distinguishable nucleic acid sequences so it can facilitate the identification of organism(s).\ud \ud Next-generation sequencing (NGS) technology has the capability to produce large amounts of nucleic acid sequences in a relatively short time. This technology has been applied to the study of biological diversity in environmental samples. The extent of diagnostic problems and the capability of NGS technology to detect the presence of nucleic acids in any environment have brought the theory into clinical application. The potential of NGS as a broad-scale diagnostic tool prompted the hypothesis that NGS is a practical method for investigating pathogens causing fever.\ud \ud There are several NGS platforms available, but little is known about their effectiveness and efficiency with regards to pathogen detection in clinical samples. The primary aim of this study was to assess the capacity, sensitivity, and more importantly, the specificity of the Illumina HiSeq platform for broad-scale characterisation of pathogens associated with fever. Secondary aims included: to describe the epidemiology of AUF and UUDF in Far North Queensland, Australia, and to optimise the preparation procedure for samples that will be subjected to NGS assay.\ud \ud The study was conducted in three stages, comprising two preliminary studies and the main study. The first preliminary study was a retrospective study of fever patients presenting to Cairns Hospital over the three-year period between 1 July 2008 and 30 June 2011. The findings suggest that AUF is common in the population of Far North Queensland, Australia. A robust definition of UUDF is proposed based on the clinical and laboratory characteristics of patients reviewed in this study, including the following criteria: 1) a fever of ≥ 38.0°C or symptoms suggestive of fever; 2) a duration of fever of ≤ 21 days; 3) a failure to reach a diagnosis after performing clinical evaluation and laboratory investigations, including complete blood count, serum biochemistry, urinalysis, blood culture, chest X-ray; 4) a request by the clinician of specific test for at least one infectious agent and; 5) a failure to make a specific diagnosis. The proportion of UUDF was 56.8% (193/340), indicating the need for a broad diagnostic tool to determine infectious causes of AUF. In general, the findings provide valuable information regarding the feasibility of conducting a fever study using NGS technology at Cairns Hospital.\ud \ud The second preliminary study was conducted to determine the most suitable type of blood specimen for NGS analysis. It was anticipated that there would be small quantities of pathogen nucleic acids present among abundant human nucleic acids background. Therefore, it was important to minimise the quantity of human nucleic acids in order to increase the sensitivity of detection of the pathogen. Six healthy volunteers participated in this second preliminary study, which aimed to measure levels of double-stranded DNA in plasma and serum. Specimens were taken using different methods of blood collection: with a syringe and with a vacuum system, and with and without applying a tourniquet. DNA concentration in the samples was measured using microplate fluorescence assays using SYBR Green I as the fluorescent dye. This study found that DNA concentration in plasma was significantly lower than that in serum (p < 0.05). However, the method of blood collection did not significantly affect DNA concentration.\ud \ud A main component of this thesis was a prospective study involving the use of a NGS platform to determine infectious causes of AUFs. Isolation of DNA and RNA from plasma/serum samples was performed using QIAamp® DNA Mini Kit (Qiagen) and TRIzol® LS reagent (Life Technologies) respectively. Following nucleic acid extraction, amplification of DNA was conducted according to the SeqPlex Enhanced DNA Amplification Kit (SEQXE) protocol (Sigma-Aldrich). DNase treatment, cDNA synthesis and amplification were performed on RNA samples according to SeqPlex RNA Amplification Kit (SEQR) protocol (Sigma-Aldrich). Sequencing was conducted on 22 DNA/cDNA samples that met the standard input determined by the sequencing company. These samples originated from 17 patients, comprising seven positive control samples from patients who had specific diagnoses and 10 samples from patients for whom diagnoses had not been achieved. Data analysis was conducted using the Kraken program and the traditional assembly-alignment pipeline.\ud \ud The study findings demonstrate the limitation and utility of NGS technology in determining the aetiology of AUF. Various viruses and bacteria were found in every sample, so considered selections were made on pathogens for which there were supporting reads consistent with clinical data and pathology findings. Aetiological diagnosis was verified in 85.7% (6/7) of controls. Among the undiagnosed participants, deep sequencing identified some plausible causes of fever in 60% (6/10) of subjects, including Escherichia coli bacteraemia and scrub typus that eluded conventional tests. Further, the NGS technology generated valuable information for studying microbial diversity in human blood.\ud \ud Further work could be directed at optimising sample preparation and improving sequencing efficiency, as well as developing efficient bioinformatics tools for analysing sequence data. It is hoped that NGS technology can be adopted in clinical practice at more affordable costs and with timely delivery of results