Clinical correlates and epidemiology of respiratory viruses

The introduction of the polymerase chain reaction (PCR) into the diagnostic setting has provided unprecedented opportunities in the field of respiratory medicine, not only because pathogens need no longer be cultivable for detection but also through improved sensitivity, specificity and turnaround time compared with traditional methods. The recent discovery of several novel respiratory viruses, such as human metapneumovirus (HMPV), human bocavirus and human coronaviruses (HCoVs) NL63 and HKU1 has nevertheless created significant challenges in respiratory diagnostics, as identification of which pathogens should be tested for is increasingly difficult. The recent discovery of two novel respiratory coronaviruses (HCoV-HKU1 and HCoV-NL63) presented the opportunity to undertake large scale clinical and epidemiologic study of these alongside two previously known respiratory coronaviruses, HCoV-229E and HCoV-OC43. Over 12,000 samples collected over three years were screened using a novel four-way multiplex real-time reverse transcription-PCR (RTPCR). Clinically, coronaviruses were similar to viruses currently included in routine diagnostics, with the exception of HCoV-229E which was identified as an opportunistic pathogen in immunocompromised hosts. Variability in detection frequencies of HCoVHKU1 and HCoV-OC43 was evident. The low detection frequencies of HCoVs, comparable to those of parainfluenza viruses 1 and 2 (which are included in the routine diagnostic screening panel) indicate a borderline case for inclusion of these pathogens in routine respiratory diagnostics. To investigate the epidemiology and clinical correlates of HMPV in Edinburgh, large scale retrospective screening of over 7000 respiratory samples collected over two years was conducted. Nucleotide sequencing of HMPV-positive samples was undertaken to determine phylogenetic relationships of circulating HMPV strains. HMPV comprises two genotypes, A and B. Comparisons of the clinical presentations of the two genotypes revealed little difference, with only the observation that sub-genotype B2 was more frequently associated with infection of immunocompromised patients. Detection frequencies and symptomatology associated with HMPV infections were comparable to respiratory viruses currently included in the routine diagnostic panel, mandating its inclusion in future diagnostic screening. A switch of the predominantly circulating genotype of HMPV was observed between respiratory seasons. This is a phenomenon more widely reported for the closely related respiratory syncytial virus (HRSV), which also comprises two circulating groups. To further investigate subtype (HRSV)/ genotype (HMPV) switching, evolutionary analyses of nucleotide sequence data generated from isolates collected from geographically disparate referral centres was undertaken. The fusion and attachment (G) genes were targeted, as these encode major surface proteins and are immunogenic. Analyses were by MCMC analyses using Bayesian Evolutionary Analyses of Sampling Trees (BEAST) software. Identification of positively selected sites was performed using Phylogenetic Analysis Maximum Likelihood (PAML). Switching of the predominantly circulating lineage does not arise for either virus due to emergence of novel strains, but through fluctuating circulation frequencies of pre-existing lineages which have been circulating for several decades, indicated by the time since the most recent common ancestor. Two HRSV-A lineages comprising genotypes undergoing turnover and replacement were identified. This finding is agreeable with serologic studies of the 1970s which reported three HRSV serogroups, two within HRSV-A and one within HRSV-B. HMPV and HRSV have similar mutation rates. Positively selected sites identified within the HRSV G gene were incongruent with those identified in a previous study, generating the hypothesis that immune evasion occurs within linear epitopes rather than at specific sites. A great deal of clinical and epidemiologic data was generated through this work, parallel studies of other respiratory viruses and through diagnostic screening results. To provide a robust indication of where resources should be diverted in terms of diagnostics, therapeutics and vaccine development, and to inform infection control measures and public health policy planning, quantification of the relative disease burden attributable to the most commonly detected respiratory viruses was calculated using the World Health Organization- endorsed Disability Adjusted Life Year (DALY) model. Relative disease burden was calculated in an age stratified manner to reflect the differences in sampling in different age groups. HRSV and influenza A were consistently one of the greatest causes of disease regardless of sampled population, although HRSV caused more disease in children under 5 than influenza A and B combined. Rhinoviruses and PIV-3 were significant pathogens in all groups except those aged 16-64 years; rhinoviruses were the leading cause of disease in the immunocompromised patient group. The potential for patient-specific diagnostic screening and guidance of interventions such as patient cohorting were clear.