SummaryBackgroundStaphylococcus aureus bacteraemia is a common cause of severe community-acquired and hospital-acquired infection worldwide. We tested the hypothesis that adjunctive rifampicin would reduce bacteriologically confirmed treatment failure or disease recurrence, or death, by enhancing early S aureus killing, sterilising infected foci and blood faster, and reducing risks of dissemination and metastatic infection.MethodsIn this multicentre, randomised, double-blind, placebo-controlled trial, adults (≥18 years) with S aureus bacteraemia who had received ≤96 h of active antibiotic therapy were recruited from 29 UK hospitals. Patients were randomly assigned (1:1) via a computer-generated sequential randomisation list to receive 2 weeks of adjunctive rifampicin (600 mg or 900 mg per day according to weight, oral or intravenous) versus identical placebo, together with standard antibiotic therapy. Randomisation was stratified by centre. Patients, investigators, and those caring for the patients were masked to group allocation. The primary outcome was time to bacteriologically confirmed treatment failure or disease recurrence, or death (all-cause), from randomisation to 12 weeks, adjudicated by an independent review committee masked to the treatment. Analysis was intention to treat. This trial was registered, number ISRCTN37666216, and is closed to new participants.FindingsBetween Dec 10, 2012, and Oct 25, 2016, 758 eligible participants were randomly assigned: 370 to rifampicin and 388 to placebo. 485 (64%) participants had community-acquired S aureus infections, and 132 (17%) had nosocomial S aureus infections. 47 (6%) had meticillin-resistant infections. 301 (40%) participants had an initial deep infection focus. Standard antibiotics were given for 29 (IQR 18–45) days; 619 (82%) participants received flucloxacillin. By week 12, 62 (17%) of participants who received rifampicin versus 71 (18%) who received placebo experienced treatment failure or disease recurrence, or died (absolute risk difference −1·4%, 95% CI −7·0 to 4·3; hazard ratio 0·96, 0·68–1·35, p=0·81). From randomisation to 12 weeks, no evidence of differences in serious (p=0·17) or grade 3–4 (p=0·36) adverse events were observed; however, 63 (17%) participants in the rifampicin group versus 39 (10%) in the placebo group had antibiotic or trial drug-modifying adverse events (p=0·004), and 24 (6%) versus six (2%) had drug interactions (p=0·0005).InterpretationAdjunctive rifampicin provided no overall benefit over standard antibiotic therapy in adults with S aureus bacteraemia.FundingUK National Institute for Health Research Health Technology Assessment.

No abstract

BackgroundHuman tissue is increasingly being whole genome sequenced as we transition into an era of genomic medicine. With this arises the potential to detect sequences originating from microorganisms, including pathogens amid the plethora of human sequencing reads. In cancer research, the tumorigenic ability of pathogens is being recognized, for example, Helicobacter pylori and human papillomavirus in the cases of gastric non-cardia and cervical carcinomas, respectively. As of yet, no benchmark has been carried out on the performance of computational approaches for bacterial and viral detection within host-dominated sequence data.ResultsWe present the results of benchmarking over 70 distinct combinations of tools and parameters on 100 simulated cancer datasets spiked with realistic proportions of bacteria. mOTUs2 and Kraken are the highest performing individual tools achieving median genus-level F1 scores of 0.90 and 0.91, respectively. mOTUs2 demonstrates a high performance in estimating bacterial proportions. Employing Kraken on unassembled sequencing reads produces a good but variable performance depending on post-classification filtering parameters. These approaches are investigated on a selection of cervical and gastric cancer whole genome sequences where Alphapapillomavirus and Helicobacter are detected in addition to a variety of other interesting genera.ConclusionsWe provide the top-performing pipelines from this benchmark in a unifying tool called SEPATH, which is amenable to high throughput sequencing studies across a range of high-performance computing clusters. SEPATH provides a benchmarked and convenient approach to detect pathogens in tissue sequence data helping to determine the relationship between metagenomics and disease.

The identification of microbiological infection is usually a diagnostic investigation, a complex process that is firstly initiated by clinical suspicion. With the emergence of high-throughput sequencing (HTS) technologies, metagenomic analysis has unveiled the power to identify microbial DNA/RNA from a diverse range of clinical samples (1). Metagenomic analysis of whole human genomes at the clinical/research interface bypasses the steps of clinical scrutiny and targeted testing and has the potential to generate unexpected findings relating to infectious and sometimes transmissible disease. There is no doubt that microbial findings that may have a significant impact on a patient’s treatment and their close contacts should be reported to those with clinical responsibility for the sample-donating patient. There are no clear recommendations on how such findings that are incidental, or outside the original investigation, should be handled. Here we aim to provide an informed protocol for the management of incidental microbial findings as part of the 100,000 Genomes Project which may have broader application in this emerging field. As with any other clinical information, we aim to prioritise the reporting of data that are most likely to be of benefit to the patient and their close contacts. We also set out to minimize risks, costs and potential anxiety associated with the reporting of results that are unlikely to be of clinical significance. Our recommendations aim to support the practice of microbial metagenomics by providing a simplified pathway that can be applied to reporting the identification of potential pathogens from metagenomic datasets. Given that the ambition for UK sequenced human genomes over the next 5 years has been set to reach 5 million and the field of metagenomics is rapidly evolving, the guidance will be regularly reviewed and will likely adapt over time as experience develops.

The results published in Poore and Kopylova et al. 2020 revealed the possibility of being able to almost perfectly differentiate between types of tumour based on their microbial composition using machine learning models. Whilst we believe that there is the potential for microbial composition to be used in this manner, we have concerns with the manuscript that make us question the certainty of the conclusions drawn. We believe there are issues in the areas of the contribution of contamination, handling of batch effects, false positive classifications and limitations in the machine learning approaches used. This makes it difficult to identify whether the authors have identified true biological signal and how robust these models would be in use as clinical biomarkers. We commend Poore and Kopylova et al. on their approach to open data and reproducibility that has enabled this analysis. We hope that this discourse assists the future development of machine learning models and hypothesis generation in microbiome research.

Results published in an article by Poore et al. (Nature. 2020;579:567–574) suggested that machine learning models can almost perfectly distinguish between tumour types based on their microbial composition using machine learning models. Whilst we believe that there is the potential for microbial composition to be used in this manner, we have concerns with the paper that make us question the certainty of the conclusions drawn. We believe there are issues in the areas of the contribution of contamination, handling of batch effects, false positive classifications and limitations in the machine learning approaches used. This makes it difficult to identify whether the authors have identified true biological signal and how robust these models would be in use as clinical biomarkers. We commend Poore et al. on their approach to open data and reproducibility that has enabled this analysis. We hope that this discourse assists the future development of machine learning models and hypothesis generation in microbiome research.

No abstract