Multiscale Computational Approaches in Infectious Diseases

Influenza viruses infect millions of humans every year causing an estimated 400,000 deaths globally. Due to continuous virus evolution current vaccines provide only limited protection against the flu. Several antiviral drugs are available to treat influenza infection, and one of the most commonly used drugs is oseltamivir (Tamiflu). While the mechanism of action of oseltamivir as a neuraminidase inhibitor is well-understood, the impact of oseltamivir on influenza virus dynamics in humans has been controversial. Many clinical trials with oseltamivir have been done by pharmaceutical companies such as Roche but the results of these trials until recently have been provided as summary reports or papers. Typically, such reports included median virus shedding curves for placebo and drug-treated influenza virus infected volunteers often indicating high efficacy of the early treatment. However, median shedding curves may be not accurately representing drug impact in individual volunteers. Importantly, due to public pressure clinical trials data testing oseltamivir efficacy has been recently released in the form of redacted PDF documents. We digitized and re-analyzed experimental data on influenza virus shedding in human volunteers from three previously published trials: on influenza A (1 trial) or B viruses (2 trials). Given that not all volunteers exposed to influenza viruses actually start virus shedding we found that impact of oseltamivir on the virus shedding dynamics was dependent on (i) selection of volunteers that were infected with the virus, and (ii) the detection limit in the measurement assay; both of these details were not well-articulated in the published studies. By assuming that any non-zero viral measurement is above the limit of detection we could match previously published data on median influenza A virus (flu A study) shedding but not on influenza B virus shedding (flu B study B) in human volunteers. Additional analyses confirmed that oseltamivir had an impact on the duration of shedding and overall shedding (defined as area under the curve) but this result varied by the trial. Interestingly, treatment had no impact on the rates at which shedding increased or declined with time in individual volunteers. Additional analyses showed that oseltamivir impacted the kinetics of the end of viral shedding, and in about 20–40% of volunteers that shed the virus treatment had no impact on viral shedding duration. Our results suggest an unusual impact of oseltamivir on influenza viruses shedding kinetics and caution about the use of published median data or data from a few individuals for inferences. Furthermore, we call for the need to publish raw data from critical clinical trials that can be independently analyzed.

As whole genome sequencing is becoming more accessible and affordable for clinical microbiological diagnostics, the reliability of genotypic antimicrobial resistance (AMR) prediction from sequencing data is an important issue to address. Computational AMR prediction can be performed at multiple levels. The first-level approach, such as simple AMR search relies heavily on the quality of the information fed into the database. However, AMR due to mutations are often undetected, since this is not included in the database or poorly documented. Using co-trimoxazole (trimethoprim-sulfamethoxazole) resistance in Staphylococcus aureus, we compared single-level and multi-level analysis to investigate the strengths and weaknesses of both approaches. The results revealed that a single mutation in the AMR gene on the nucleotide level may produce false positive results, which could have been detected if protein sequence analysis would have been performed. For AMR predictions based on chromosomal mutations, such as the folP gene of S. aureus, natural genetic variations should be taken into account to differentiate between variants linked to genetic lineage (MLST) and not over-estimate the potential resistant variants. Our study showed that careful analysis of the whole genome data and additional criterion such as lineage-independent mutations may be useful for identification of mutations leading to phenotypic resistance. Furthermore, the creation of reliable database for point mutations is needed to fully automatized AMR prediction.