Search Results

Bacterial HAIs are a substantial source of global morbidity and mortality, resulting in increased length of hospital stay and high healthcare costs. At Genpax, we're often asked what the benefit of adopting a WGS led intervention strategy could have on hospitals systems, so we developed at thorough model to work it out. Our findings will be presented at the forthcoming ASM Microbe 23 meeting in Houston in June. If you would like further information, a copy of the poster, the model, or materials submitted for publication, please contact us directly. Economic and health impact modelling of a Whole Genome Sequencing-led intervention strategy for bacterial Healthcare-Associated Infections for England and for the USA. J. M. Fox, N. J. Saunders & S. H. Jerwood Bacterial healthcare-associated infections (HAI) are a substantial source of global morbidity and mortality. The estimated cost associated with HAI ranges from $35 to $45 billion in the United States of America (USA) alone. The costs and accessibility of whole genome sequencing (WGS) of bacteria and the lack of sufficiently accurate, high-resolution, scalable and accessible analysis for strain identification are being addressed. Thus, it is timely to determine the economic viability and impact of routine diagnostic bacterial genomics. Additional Information

At ECCMID 2023 in Copenhagen, we displayed some of the findings regarding our near-zero error capabilities generated from our platform which will be launched commercially this summer. Calling Zero: A new foundation for diagnostic bacterial genomics James C. Littlefair, Benedict J. Uttley, Dan G. Frampton, Gareth M. Linsmith, John F. Peden, & Nigel J. Saunders Reference ECCMID

The same features of the Genpax analysis that mean that all strains of a species can be compared in one location with a common analysis resource (as opposed to others, such as those that must use multiple reference genomes or where resolution reduces with increasing numbers and diversity of strains), and its open scalability means it will generate equally comparable and integratable findings from information generated and submitted from different laboratories. This is enabled by its high accuracy/low noise, which is associated with very high reproducibility and the same result is obtained repeatedly from the reanalysis of the same sequence files and different independently sequenced DNA samples of the same strain. Analyses of published data can readily show the benefits of this. There are examples of common strains split into different study and analysis groups in single hospitals, of transfers of MRSA between different UK centers, and of connected Listeria strains spanning three labs in two countries, just from a limited set of early studies using publicly available read sets. This can form the basis of a new approach to public health surveillance. A bottom-up strategy empowered by the sequencing of strains in real clinical time that provides direct clinical and economic value, leading to more sequence information being collected in a more clinically relevant timeframe. The sequencing can be distributed close to the point of isolation but with the comparability of a centralized sequencing model. Equally, the data can be fully integrated when generated by hybrid models combining distributed and central sequencing. As more laboratories use the common analysis, the greater value the sequencing information can have locally and more widely.

Genpax analysis was created for the real-world problem from the bottom-up with scalability as an essential requirement. Scalability is a constraint for existing solutions, the three main reasons being: • The statistical nature of relationship determinations that are impossible to scale (they are ‘NP-hard’ problems) • The loss of resolution with increasing diversity and number of compared strains with common genome SNP • Data generated with more than one reference genome when analyzing a species cannot be readily integrated A key aspect of scalability is deliverability. The time and computational costs of analysis using unscalable solutions rapidly grow with increasing resolution and numbers of strains being compared. Typically, high-resolution studies are limited to sets of highly related strains, to local and recent isolates (for example, over a 3-month window in a large hospital setting), or to restricted sets of highly related strains. This is because the time needed to compare hundreds or more strains rapidly become too slow when the information is needed quickly, and the process has to be re-run each time additional strains are added to the analysis. Even when restricted to Sequence Types with a good reference genome, the number of isolates can be too many; an extreme example is ST22 MRSA, which represent around 50% of the isolates in the UK. Whether you wish to compare 5 strains, or five thousand, Genpax will deliver in real clinical time, and do so for multiple new strains in parallel, comparing each against every other previously and co-analyzed strain within a real-world clinically relevant turn-around target of 2 hours. Not at the resolution or with the errors of Sequence Typing (that must be followed up with further analyses); at full SNP resolution, with greater accuracy than other methods, and with more sequence addressed than ‘common genome SNP’.

When strains are evolving at 1 to 10 nucleotides per year per genome, it is important to address as much of the genome sequence information as possible within the limits of what can be achieved accurately. This cannot be 100% of the genome with current sequencing technologies. For example, regions that are repeatedly present in the genome either identically or with minor variations (for example the ribosomal RNA sequences) are inherently difficult to address. Exactly how much of a genome can be robustly addressed varies by species and differences in the underlying composition of their genomes. Sequence Typing using a widely used method called MLST typically addresses 3000 to 4000 nucleotides derived from sections of 5 to 7 highly conserved genes or around 1% of an average genome sequence. Its WGS-based successor, cgMLST, typically addresses the whole sequence of 50 to 70% of genes present in individual strains, representing up to 50 or 60% of the sequence. wgMLST is a more inclusive version of this analysis, which is used to assess specific groups of strains rather than the whole species and can increase this by around 5 to 10%. However, it should be noted that in addition to excluding genes that are not universally present, non-protein-coding sequences, and genes in closely related families, none of these methods uses the SNP-level resolution and information available. Instead, they give arbitrary numbers to different versions of the genes that are included; thus, their real resolution is less than the sequence length that they address. SNP-level resolution analysis requires separate and additional bioinformatic analysis of strains that have been identified as related at the level of the Sequence Types and Clonal Clusters. This is perfectly reasonable if it is remembered that the original purpose of Typing was either to provide supporting evidence for an already suspected set of connected strains or to exclude connection by showing that strains were unrelated. The original purpose of Typing was never to proactively detect outbreaks and direct infection prevention and control. Analysis of SNPs within genomes is more complicated. Using previous methods its resolution is limited by whether there is a high-quality highly related reference to work with or not; and for some species this is not possible. If there is a reasonable reference, how similar that reference is, how many strains are compared, and how diverse they are, all impact resolution. Whole genome sequencing information might be used, but that doesn’t mean the whole genome is addressed. As stated above, some regions cannot be safely analyzed and, in addition, it is only the shared genome from the analyzed stains which is included in what might be more clearly described as ‘Common genome SNP’ analysis. (Rather than ‘Whole genome SNP analysis’ or ‘Core genome SNP analysis’ – which is the SNP-level analysis of what was included in cgMLST). The result for a small analysis using a closely related reference strain is that only 70 to 80% of the genome is addressed. When addressing more diverse strains this can fall to to as low as what is addressed by cgMLST. The real-world challenge is to reduce noise, increase accuracy, and at the same time address the maximum amount of the safely analyzable genome – and to maintain that resolution regardless of the availability of a highly related reference genome or the number and diversity of strains compared. Genpax achieves this. At the same time as reducing the noise and increasing the accuracy, the analysis achieves a resolution of typically 5 to 10% above whole genome SNP analysis at its best (ranging from 75 to 90% of the genome, depending upon the species and its sequence composition), and maintains this regardless of the number and diversity of strains that are addressed. It is also equally good at analyzing species for which reference genomes for most strains do not exist (e.g., Salmonella) or where recombination means that they cannot exist (e.g., Campylobacter).

With some variation by species and genome size, most bacterial pathogens normally change at a rate of up to 10 SNP changes per genome per year. A clinical outbreak spanning weeks or months will include strains diverging at double this rate but also contain strains that differ by less because 99% of strains replicate without error. Meanwhile, sequencing, assembly, and mapping errors can generate tens of errors per MB, resulting in noise that exceeds the essential signal to identify recently connected strains, source attribution, and transmission inference. Various partial solutions exist, such as focusing on the most accurately addressed parts of the genome, using thresholds that allow a certain number of differences while still typing a strain similarly, and labelling gene versions rather than using the detail of the several differences that they may contain. All involve a sacrifice of resolution and have their own issues. None provide a solution that delivers what is required for proactive detection and direction of outbreak and infection prevention and control responses. The accuracy issues create another problem. In order to not to exclude genuinely connected strains, the distance threshold to identify them is loosened (for example from 10 to 15, to 20 to 30, sometimes more). However, strains that are not connected by transmission can differ within this range, especially the highly related strains that tend to exist within ‘local healthcare microbiomes’ (that exist within hospitals, associated care facilities, and healthcare workers). It is essential to be able to discriminate between these unconnected but very similar strains from those that are genuinely transmission- or common source-associated in order to target infection prevention responses properly, and not to over-identify and over-extend the strains that are considered to be part of an outbreak.. Both the under- and over-detection of outbreaks can have significant costs and impacts. The third issue is that the noise is not constant, nor completely random. As a result, it frequently re-orders the connections that can be inferred with respect to the chain of transmission between patients or other sources. Being able to reliably determine the detail of the distances and connections between strains is necessary for clinical genomics to be able to be used for proactive infection prevention and control as a leading source of information, rather than for confirmation or as an adjunct to epidemiological and other indicators. Where the noise is coming from. Error has always been tolerated, and it is unavoidable in genome sequencing. Partly it is generated in the creation of the raw sequence data from the sequencing machine, predominantly it is generated and assimilated in the process of genome assembly, and both are the result of differences in the data which is analyzed from the FASTQ file (the universal output file format from DNA sequencers). The properties of this file are influenced by multiple things: the quality of the DNA extracted from the bacteria, the preparation of the sequencing library, the flow cell it is run on (even the lane of the flow cell), the sequencing protocol (e.g., fast or slower with more wash steps), and also what reads from the output are used in assembly. This is why some try to demonstrate analysis performance using the same FASTQ file rather than genuine replicate sequencing files; or even simulated read-sets – which generate the fewest errors possible but do not represent the real-world situation. Some of these issues are overcome by using mapping-based sequence analysis, but this is dependent upon having a good quality highly-related reference genome, which isn’t always available; and for some species it is a practically impossible solution. And some genomes and parts of genomes are inherently more accurately assembled and analyzed than others. Lowest possible noise with Genpax Genpax has developed strategies and analysis resources that address the sources of sequencing noise and that recognize the essential nature of a Near-Zero Noise solution. Samples that reflect real-world conditions of independently generated sequences; including replicate cultures in true technical replicate DNA sequencing prepared from the same DNA sample, different DNA preparations from co-cultures, and very closely epidemiologically linked independent cultures consistently generate Zero distances. No previous platform has achieved this for strains representing clinical isolates and diversity. This is the necessary starting point for an impactful bacterial genomics solution for infection prevention and control to detect connected strains and direct necessary responses.

While the costs and ease of sequencing have been falling, one common obstacle to adoption is the requirement for local genomics staff and infrastructure. High costs of IT staff, bioinformaticians, and genomic epidemiologists (plus sufficient staff to provide continuity of service) and the initial and ongoing costs of computational infrastructure often provide a barrier to entry. Initial capital and equipment costs that are commonly encountered, particularly in resource-limited health environments, are also a hindrance. Furthermore, people with the relevant qualifications and experience are in high-demand and command good salaries. With Genpax, all that is required are the laboratory skills and environment to generate the sequence information, which is progressively more accessible with the latest sequencers that are increasingly easy to use. Once a sequence of sufficient depth and coverage has been generated, this is transferred to Genpax through the Cloud, where the analysis is performed automatically. This only requires the input of basic sample information, after which the necessary reports needed for the labora tory to monitor their performance, for the patient records, and for the infection prevention and control team are returned in easy-to-understand formats that highlight the clinically important and actionable information. No additional computer infrastructure or specialist staffing is required. Charges are per-test and post-analysis, so there are no up-front costs to adoption nor disproportionate charges while services are being established.

Session CPHM01 - NGS for Near Real-Time Surveillance and Outbreak Response by Public Health and Clinical Labs. Title: A Novel Multi-Scale Outbreak Detection and Strain Identification Capability for Genome Sequence-Based Infection Control: An MRSA Example. Authors: J. C. Littlefair, B. Uttley, D. Frampton, J. F. Peden, N. J. Saunders; GenPAx SA, London, United Kingdom

A Novel Multi-Scale Outbreak Detection and Strain Identification Capability for Genome Sequence-Based Infection Control: An MRSA Example. J. C. Littlefair, B. Uttley, D. Frampton, J. F. Peden, N. J. Saunders The exclusion (as well as inclusion) of strains is vital for outbreak investigation and infection prevention and control of healthcare-associated infection, for both health resource management and patient-care. Current genome-sequence based diagnostics strategies have limitations in their analytical capabilities and scalability which restrict their utility for proactive diagnostic surveillance and to direct real-time infection control. MLST is of limited value in the context of dominant epidemic or more virulent clones; such as MRSA where up to 90% of isolates fall into a small number of common STs. High-resolution SNP-based analysis can be pursued when there is an appropriate reference genome, but the greater the diversity between the sample and reference genome the lower the coverage, resolution, and accuracy. Further, the similarity of members of clonal clusters within hospital and surrounding community and health-care environments can be high. Thus, the detection and differentiation of hospital-associated acquisition and transmission from strains entering the hospital is non-trivial and requires improved sequence analysis to resolve. GenPAx has developed a novel analysis pipeline that can perform multi-scale analysis for both outbreak detection and high-resolution determination of strain identity and relatedness (and implied transmission connections). This can be run without prior knowledge of potential outbreaks or associated strains, and can detect single instances of known high-virulence and highly-resistant clones. Using a well-established dataset of over 600 London-based MRSA we achieved clearly higher resolution and improved determination of connected isolates than using traditional methods: the number of sites used for identification within the study was more than doubled, and the dynamic range of differences and resolution of associated strains was greatly increased. While highly similar strains some with less than 10 nt difference were still identified, other previously linked strains became clearly separated, indicating that the number and size of hospital and health-care associated cross-infections was substantially lower than previously thought. There was also a better match between the size of the largest cluster and the number of epidemiologically determined interactions. Thus, we have shown that this next generation of bacterial genomics resources can substantially increase the future diagnostic utility of genome sequencing for hospital infection control. Session CPHM01 - NGS for Near Real-Time Surveillance and Outbreak Response by Public Health and Clinical Labs Reference ASM

It is repeatedly stated that we face a global challenge of antimicrobial resistance and catastrophic health and economic impacts, and that this is the product of misuse, abuse, and essentially over-use of antibiotics. It is hard to argue, and I would not wish to, that this is at least partly the case, but it is both dangerous and misleading to present this as the whole of the explanation. What we are seeing is an inevitable consequence of many pressures and their inevitable evolutionary outcomes. That mis-use, over-use, and various pressures other than human health have contributed to the current situation, but there are plenty of other contributory factors – and we cannot genuinely address the problem without naming and facing them. We are concerned about AMR because it challenges human health and some human’s personal survival. But it should be realized that significant drivers for the AMR challenge we now face had nothing to do with human health. You can’t blame a company for wanting to sell its products, or for rolling out new ones as soon as they are available, by keeping up with the arms race against the bugs rather than seeking to restrict and change practice to focus and minimise the development of new resistances. I remember when we used to have what is now regarded as a new policy of zero tolerance towards hospital infections with MRSA, and then it was stopped, intentionally! A case of MRSA in hospital triggered a response of containment, isolation, and effective infection control precautions. Possibly infected contacts were screened, contained, and, if necessary, treated. Then, with excuses such as practice in Australia (where such infections were not contained and were subsequently completely out of control) were cited to show that it wasn’t necessary; while a lack of funding for microbiology and infection control staff to maintain recommended ratios to patients meant that it was practically hard work to deliver. So, what is now our new policy of ‘zero tolerance’ was abandoned in the 1990s. As COVID-19 should have illustrated for everyone, pathogens are without borders. Our Brazilian AIDS patients had histoplasma lung infections, not the local versions, and our patients who had been to Spain had Spanish AMR infections - 20 years ago. Many countries allow you to buy antibiotics over-the-counter (and STILL do), and this is associated with societies with far greater emergence of AMR than those with more restrictive practices. That reducing the use of antibiotics for viral sore throats by UK GPs has positive effects isn’t up for debate, and the campaign for ‘Antibiotic Stewardship’ is good and wise. But, the types of antibiotics that were over-used in these settings aren’t the ones that the really challenging ones being lost in the battle with AMR, and unless you not only stop other places continuing with much poorer practice (and stop our population going on holiday and collecting them from such places) it isn’t going to make a big difference in the end. So, two points could be usefully made with respect to the use of antibiotics in human care: First: AMR isn’t caused by misuse; at the most, it is accelerated by local use and practices. The emergence of AMR is an inevitable consequence of the use of a selective pressure on an evolutionary process. How many tons are used per year in human and other contexts will affect how fast, but not if it will happen. The scale and impact of AMR within hospitals – that’s a separately identifiable issue, but the doctors were put in an impossible position, and the resources and recognition now associated with the new ‘zero-tolerance policy didn’t exist when it was originally practised. Second: nobody should blame GPs and other UK-based doctors for a situation which was being more effectively created and imported from elsewhere. And where the antibiotics that were being marginally over-used locally are not the ones where resistance presents the real challenges we face. Dr Nigel Saunders, Chief Scientific Officer Contact: Nigel@genpax.co

At one level, the numbers look good, but if you look again, maybe not. Once upon a time, TB started out as something else, probably some environmental predecessor that lived in watery or other environments, perhaps starting out in another animal, but eventually, it got to humans. There was a ‘patient zero’ for TB, just as there is for all new and emergent pathogens, and for TB, it is guessed that this happened around 5,000 to 10,000 years ago. There is evidence that it has been with us since the Neolithic, so it has been with us for most of what we consider to be human cultural history. But that’s not that long ago in evolutionary time, and it didn’t keep itself to people; it infected cows, goats, seals, and probably other things too. In all likelihood, it was us who gave it to the cows, where it evolved into the bovine version, and them to the badgers – since there are no cow or badger hosts crossing the oceans to get it to them. From patient zero … it spread to infect. Now numbers are falling overall, and the control of the TB epidemic is one of the 17 United Nations Sustainable Development Goals that were ratified in 2015. An interim progress report was given in 2020 (A/75/236 - E - A/75/236 -Desktop (undocs.org)). The only region to achieve an interim target of a 20% reduction by 2020 was Europe, where the majority of countries are amongst the 54 with low incidence. Elsewhere reduction is happening but far more slowly than intended, except for the WHO African region, while the incidence in the WHO Region of the Americas was actually increasing slightly. These numbers aside, the drug resistant strain numbers aren’t decreasing at all – they are increasing. This includes a specific increase in strains called Beijing because of where it was first characterized. A combination of higher virulence (ability to spread, which is related to having symptoms and coughing) and a greater association with drug resistance (meaning infectious patients stay infectious for longer) is combining to see remarkable increases in the proportion of cases it is associated with. For example, in Mumbai, a global centre of infection for drug-resistant TB, the proportion of cases caused by this one strain (or group of related strains) has increased from 4% to 40% in just 10 years. It has emerged to cause around 50% of infections in East Asia and at least 13% worldwide. The point is that looked at separately, rather than with all the other strains together, there is effectively a new pandemic with rapid spread and expansion of this version of TB. The expansion of this pathogenic strain demonstrates that current methods of control are inadequate to contain it, let alone eradicate it. It also illustrates the importance of being able to identify new emergent challenge strains long before they get to spread globally, as this one has. Dr Nigel Saunders, Chief Scientific Officer Contact: Nigel@genpax.co

As Sun Tzu famously said: ‘know your enemy’, or more correctly: ‘If you know the enemy and know yourself, you need not fear the result of a hundred battles.’ This is true not only for human conflict but in what is probably the greatest and longest running battle the human race has ever engaged in: that between humans and their pathogens. Ultimately, COVID-19, our current enemy of focus, might kill 1%, but hopefully, many fewer now that there is a vaccine and better treatments established – and given that the elderly and most vulnerable populations exist in (and partly because of) the locations of the better health-care systems, this is probably not misplaced optimism. But, despite its impact, this is, in historic terms, a minor skirmish in a relatively short war. Before drug treatment, TB killed by some estimates around 1% every year. That’s across many age groups, not just the elderly. It killed my wife’s fathers’ sister at 2 years, her great aunt as an adult, and 1% scaled for the current world population would be around 70 million, compared to an estimated death toll from its start in 2019 to early 2021 of 2.5 million. And TB killed its 1% every year, year after year, but even TB is not as bad as it can be. Consider the plague called the Black Death? In less than 10 years, it is estimated to have reduced the European population by between 30 and 60%, and the world population from around 475 to between 350 and 375 million. An equivalent today would be a death toll in the billions. When medical students learn about infection and pathogens, it is normal for them to be introduced as pathogenic species. For a few, normally just one, the idea is introduced that things can be more complex so that some strains of a species can be dangerous and others not. An example of this is E. coli which can be a harmless resident of the gut, perhaps even providing protection against other bugs that could make you ill (and even some strains used as health-promoting probiotics), while others cause increasingly unpleasant gut infections with one particularly challenging example that causes considerable deaths and kidney damage. This is a good example for teaching because the different ‘weapons’ the more dangerous strains are carrying are well known, and can be used as an introduction to how the causes of virulence (dangerousness) are investigated. But, for the rest, it is pretty much a list of villains: the bad bugs you have to learn about: the infections they cause, some of the armaments and defences they carry, the list of enemies we need to know and contend with, and its fairly uncomplicated. If you have some bad bugs, it’s always bad news. If you have other bad bugs where they shouldn’t be or when you are particularly vulnerable – then that’s also something that has to be addressed. But, this model taught generation-to-generation, which is the structure and form of traditional lectures, exam questions, and textbooks – is it right? The paper by Tong and colleagues is one of a number describing applying genomics to pathogen surveillance and control, which could be interpreted as suggesting a new and different perspective. Yes, you need to understand both the host and the pathogen; both ‘the enemy’ and ‘yourself’, because there is most certainly an interplay between the dangerousness of the pathogen and the vulnerability of the patient, but it is no longer necessary to be so simplistic as to how we define the enemy or fight it. What I see in this and other reports is an emergent pattern in which it is not ‘pathogen species’ but ‘more or less pathogenic strains’ that are evident. A subset of strains causes a disproportionate number of cases, and these have persistence and presence over time and in different settings. Of note in this study was that on several occasions, one clone of bacteria was replaced by a more dominant one in individual patients. It may even be that for the normally healthy, the less dangerous strains may be providing protective effects through their competition with the more dangerous versions. We have perhaps missed something else: that we face outbreaks, epidemics, and pandemics that are occurring in ways that we haven’t previously appreciated because they are mixed with each other and with less dangerous relatives; and perhaps exacerbated by travel, extensive population mixing, and food production and distribution systems. I think that this is exactly the situation. And this greater understanding of our enemy has the potential to transform our future strategies for both attack and defence. Dr Nigel Saunders, Chief Scientific Officer Contact: Nigel@genpax.co Reference: Genome sequencing defines phylogeny and spread of methicillin-resistant Staphylococcus aureus in a high transmission setting (nih.gov)