Can genomics lead the fight against antimicrobial resistance?

As antimicrobial resistance becomes a global problem, Ruth Massey and Anita Justice explore whether genome sequencing has a part to play in the on-going battle Antimicrobial resistance (AMR) is a major global health problem. In the European Union (EU) 5–12% of hospital patients acquire an infection during their stay¹. Each year, an estimated 400,000 present with a resistant strain, of whom on average 25,000 die¹. In addition to causing increased morbidity and mortality, AMR has huge economic implications. Multidrug-resistant bacteria in the EU are estimated to cause an economic loss of more than €1.5 billion each year¹. What is clear is that we need to identify and develop new drugs to treat infections. But what is also clear is that we need to improve our diagnosis and prescribing policies and procedures to protect these new drugs by minimising the development of resistance to them. In this article we consider whether genome sequencing can help. The first full genome sequence of a bacterium was published in 1995², and since then the technology has advanced to a stage where it can be performed in a bench-top machine in a matter of hours directly from blood cultures or from a single colony on an agar plate³. The costs associated with this have also plummeted, so the question this raises is: will genome sequencing improve and then replace routine culture-based diagnostics and infection control surveillance? To address this we need to consider the questions a clinician asks when faced with a bacterial infection, the first of which is what is causing the infection? This is relatively straightforward to answer from a genome sequence, where the DNA encoding the 16S ribosomal subunit has all the necessary information. To compare this to today’s approaches, if we consider a suspected bacterial bloodstream infection, it will take between 12-48 hours for the bacteria to grow in a blood cultures, depending on the infecting bacterium’s growth rate. This broth then needs to be streaked on agar plates to obtain single colonies, which again will take a further 24-48 hours. From a single colony MALDI-TOF can be used to identify the bacterial species. As such routine culture based methodology can take between two and four days to identify a bacterial species from a bloodstream infection. Using current sequencing technologies, this is a comparable time scale to genome sequencing directly from blood culture bottles, which would take between 36-48 hours. So as yet, genome sequencing does not represent a clinical advantage. The next question a clinician should ask is whether it is an antibiotic resistant isolate, and what is it resistant to? Using today’s gold standard approaches it takes the same 12-48 hours for the bacteria to grow in a blood culture, then for automated systems such as the BD Phoenix this culture need to be plated (a further 24-48 hours) and up to 10 colonies selected for use in the system which takes approximately 12-16 hours to produce results. For the majority of antibiotic resistances, the genetic mechanisms have been well characterised, and can be determined directly from the genome sequence. So, for these, the time saved will going from three to five days to 36-48 hours. Unfortunately, for some resistances such as heterogeneous and intermediate resistance to vancomycin and fluoroquinolone resistance the genetic determinants are not fully understood, so until these are fully characterised, there are some limits to the use of genome sequence for accurate resistance profiling. The next question that might arise for a clinician if faced with a sudden increase of clinically similar infections is whether it represents the beginning of an outbreak, and if so what is the source. In many countries methods such as pulse-field gel electrophoresis (PFGE), multi-locus sequence typing (MLST) and spa typing are used to determine relationships between the infecting bacterial strains, to investigate outbreaks and identify sources. These additional tests are often performed at reference laboratories, adding additional time (one to six weeks) to the diagnostic process such that results are often available too late to influence clinical management or infection control decisions. Here, again the genome can be sequenced in 36-48 hours, and used to characterise and compare the isolates to establish their relationship and whether an outbreak is underway. What’s useful to consider here is that all of the clinical questions raised so far can be addressed from the genome sequence, so it represents a one-stop procedure to potentially replace many others. Another problem that can arise in the diagnostic lab is with slow growing bacterium such as Mycobacterium tuberculosis, the causative agent of TB. This is becoming increasingly problematic on a global scale with increasing levels of drug resistance. It can take up to 28 days to get detectable growth of this pathogen, which makes determining its antibiotic resistance profile to allow effective prescribing a huge challenge. In a recent study an infected patient’s sputum became culture-positive after three days in the mycobacterial growth indicator tube culture system, and from this there was sufficient DNA to allow the genome to be sequenced, and the antibiotic resistant profile to be determined and effective treatment to be prescribed, greatly reducing this patients suffering and the costs associated with prolonged hospitalisation and the use in ineffective antibiotics⁴. While it’s clear that genome sequencing can increase the speed of current diagnostic and surveillance procedures, there is another clinically important trait it could help diagnose. Toxin secretion is critical to the ability of most bacteria to cause disease. These toxins cause local tissue damage, the release of nutrients and contribute to transmission within and between hosts. If a clinician could determine at an early stage the potential for severe disease represented by an individual infecting microorganism they could tailor their treatment and isolation procedures accordingly. It could also raise their index of suspicion for serious complications, and they may chose to prescribe cocktails of toxin-suppressing antibiotics such as linezolid and clindamycin. However, this type of analysis is not currently undertaken in diagnostic laboratories, due to its complex nature. A recent study focussing on MRSA found that it is possible to predict toxicity from their genome sequences⁵. This study identified genetic signatures (collections of subtle nucleotide changes) that indicated whether an individual isolate was likely to be highly toxic and cause more severe disease. Although focussed on a single clone of MRSA, this study highlights the potential for genome sequencing to revolutionise how infectious disease is diagnosed and controlled. The main issue that will limit the wide-spread use of genome sequencing in diagnostic settings is cost. Over a relatively short space of time, these have plummeted, and in a high-throughput settings it can cost as little as £36 to sequence an isolate. For routine diagnostics this still does not represent good value for money, but in situations where an antibiotic resistant TB infection is suspected, the time saved could be considered valuable in terms of the outcome for the patient and the time and resources wasted prescribing ineffective antibiotics. So, to answer to the question “Will genome sequencing help in the fight against antimicrobial resistance?” It certainly has the potential to be a valuable weapon for us to use, reducing the time to accurate diagnosis, antibiotics profiling, and infection control surveillance. We may even be able to use it to predict how severe an infection might become. But the cost needs to come down and in certain areas, such as toxicity prediction and for complex antibiotic resistance mechanisms, we need more time to study and fully understand the genetics behind them. The methodology of sample preparation and analysis of the sequence data also needs be simplified and automated for use in routine diagnostics, to be confident in the quality and reliability of the results. However, if the rate of dramatic changes that have occurred in genome sequencing over the past 10 years continues, these limitations could soon be addressed, and we will be able to apply genomics with confidence to infectious disease diagnosis and management. References 1. http://ec.europa.eu/health/antimicrobial_resistance/policy/index_en.htm 2. Fleischmann, R.D. et al. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science. 269: 496-512. 3. Köser, C.U. et al. 2014. Rapid single-colony whole-genome sequencing of bacterial pathogens. J Antimicrob Chemother. 69:1275-81. 4. Köser, C.U. et al. 2013. Whole-genome sequencing for rapid susceptibility testing of M. tuberculosis. N Engl J Med. 369:290-2. 5. Laabei, M. et al. 2014.  Predicting the virulence of MRSA from its genome sequence. Genome Res. 24:839-49. Authors Ruth Massey, University of Bath, Bath, UK. and Anita Justice, John Radcliffe Hospital, Oxford, UK.