Considerations in centralizing whole genome sequencing for microbiology in a public health setting

KEYWORDS:

• public health microbiology
• whole genome sequencing
• WGS
• centralization
• molecular epidemiology
• outbreak control
• surveillance
• transmission investigation

Molecular epidemiology, epidemiology and public health microbiology are used for local, national and international outbreak management and control, surveillance and antimicrobial resistance (AMR) surveillance to identify disease threats. Accurate reporting of antimicrobial consumption that can be linked to changes in resistance patterns seen can lead to changes in prescribing patterns to prevent the overuse and misuse of antibiotics – the reason for current antibiotics to become ineffective.

A more detailed list of uses for molecular epidemiology of infectious disease for public health is shown below.

• Timely analysis of sequence differences of pathogens over time, place, and person across human populations and relevant reservoirs to study host–pathogen interactions and infer hypotheses about transmission [1];

• Early outbreak detection can be achieved by prospectively typing consecutive cases in a population to identify clusters of clonally linked isolates. The more analysed, the earlier the detection;

• National/international surveillance schemes can detect emergence of drug resistance or strains with enhanced virulence and can help identify risk factors associated with transmission of certain strains and predict effectiveness of vaccines;

• Microbiological source attribution of food or waterborne samples implicated in an ongoing outbreak – the faster the response the smaller the outbreak however you need to be certain. Direct action from molecular typing results (e.g. product recall) can be politically and financially very sensitive and has clear implications for public health;

• Presumptive transmission cluster identification for enhanced contact tracing, for example, TB;

• Detection of emergence or monitoring of spread of epidemic (multi)drug resistant pathogen(s), including in animals – antibiotics are widely used in veterinary medicine and subsequently drug residues may persist in foods derived from animals, which may pose an adverse health effect when consumed;

• Identification of epidemics of mobile drug resistance genetic determinant;

• Monitoring of vaccine strain coverage and prediction of vaccine preventability of disease, for example, influenza;

• Detection of novel hyperpathogenic strains;

• Detection of diagnostic escape variants

To carry out these functions for bacteria, traditionally isolates from clinical and environmental samples were cultured and followed by phenotypic testing to define strains (serotyping, phage-typing, or the ability to grow in the presence of an antibiotic), a strain being a genetically distinct subtype of a microorganism. Phenotyping, therefore, measures the physical manifestation of a change in the genome and the effects of the growth environment on the phenotype. More recently, various molecular methods have been employed to sample various small parts of the genome. Many viruses are uncultivable so molecular genome sampling methods have been used more widely in this area.

The case for whole genome sequencing (WGS)

The recognized pillars of any validated microbial typing scheme of typeability, reproducibility, and discriminatory power need to be central to any scheme and some conventional phenotypic microbial typing methods have been useful in describing the epidemiology of infectious diseases. However, these methods can be insufficiently discriminatory, generally not always reproducible, labor intensive, and not timely enough to be of practical value in an epidemiological investigation. Molecular or DNA-based typing methods used in more recent years rely on the analysis of the genetic material of a microorganism. Many different methods have been developed with different advantages and limitations in terms of the features of the scheme and the molecular typing method used in a study will depend on the skill level and resources of the laboratory as well as the aim and scale of the investigation. Several studies have comprehensively reviewed and compared the advantages and disadvantages of the features of different DNA-based molecular typing methods used in the epidemiology of bacterial pathogens [1,2].

The ideal features then for an internationally (ideally globally) used public health molecular tool include the following [3]:

• Be relatively easy and robust – that is, demonstrated reproducibility – and driven by practical needs for effective infection control and surveillance

• Be discriminatory in a meaningful way (e.g. to distinguish isolates associated with particular specific sources from others) and to have high typeability (i.e. allowing typing of 100% of isolates from the pathogen to which the scheme applies).

• Produce comparable and valid data within and between laboratories across the globe – that is, reproducibility and portability

• Be produced rapidly – from patient management (e.g. drug susceptibility) to local outbreak investigation to national/international surveillance. The more rapidly all this information is available (potentially from WGS carried out just once), and in the appropriate format, the quicker those responsible for action can respond, reducing the burden of infectious disease.

• Be appropriately interpreted and assessed by the multidisciplinary expert team of those involved for each pathogen in order to guide public health response and decision-making.

It is clear that only WGS has the capability to fulfill all of these features and has already been used, or shown to be potentially useful, for epidemiological (and potentially diagnostic) investigation [4–12]. In what is a transformation in microbiology, Public Health England (PHE) has sequenced a total of approximately 50,000 bacterial and viral genomes with resulting improved surveillance and outbreak investigations and intends to harness its full potential in supporting the NHS to deliver the best care for patients [13]. In the summer of 2014, an outbreak of Salmonella infection, with almost 250 cases reported across several English regions was investigated using WGS. This was the first time that WGS was used as part of a national investigation of this kind, and confirmed that the cases shared a common source. The rapid availability of the data helped to contain the outbreak while it was still under way. Together with European partners, the evidence indicated that the outbreak was associated with the consumption of eggs from a single source outside the UK [14].

The advances in WGS technology means that it is now capable of replacing many of the traditional methods and procedures used for molecular epidemiology of infectious diseases of public health importance. The instrumentation from the current market leader, Illumina, produces millions of accurate short reads that need to be reassembled, either by mapping to a full reference sequence or assembling de novo without a reference. For bacteria that contain a large number of repeats, this assembly will only generate approximately 90% of the total genome as short reads cannot be used to assemble repeats. Ideally all the repetitive information will also be captured, but this needs to be from longer reads to traverse the repeat regions entirely. Two suppliers of single molecule long-read sequence generation are currently available, PacBio, producing accurate reads of around 10 kb [15] and Oxford Nanopore Technology’s MinION, with the ability to produce much longer reads, around 100 kb, but with lower (but rapidly improving) accuracy [16].

So it is clear that WGS has the potential to examine the entire genome, providing the ultimate resolution and technologies are at a stage where they can be implemented for infectious disease for local, national, and international public health improvement. The instruments required to generate the data needed to fulfill the ideal characteristics of a global public health tool listed above are still developing however but are at a stage where high-income countries can begin the implementation of their use to generate WGS for diagnostics/molecular epidemiology of infectious diseases of public health importance. The introduction of benchtop sequencers makes ubiquitous WGS for microbial molecular epidemiology a possibility. Is centralization of such a facility the most appropriate way forward, or are there other more suitable models for the successful implementation of this opportunity to change the way we carry out public health microbiology? The advantages of improving the speed and accuracy of information generated to inform patient management, outbreak management, and early identification of outbreaks using this technology are clear, but cost and efficiency need to be carefully balanced with informed health economic arguments.

WGS provision models – centralized and localized, their advantages and disadvantages

The case for centralization

A single unified management of WGS will be more easily administered and controlled, facilitating better national coordination (and ideally should include animal and food/water/environment microbiology health). The challenges of coordinating the expertise required to develop complex laboratory and analytical workflow knowledge will be reduced. Staff involved in the production of WGS data and its analysis will benefit from a concentration of shared knowledge and expertise to achieve a spread of key talent skills through standardization. Significant cost advantages could be achieved through equipment purchase including automation instrumentation and the improved efficiency that brings together with the reduced staffing requirement that centralization of sequencers and automation brings generally. Infrastructure and indirect cost savings would be an important consideration for centralization of WGS, but alternative recovery centers would need to be identified in a stringent Business Continuity Plan in the event of unforeseen shutdown.

The case for localized rapid response facilities

There is a clear health-care economic argument for a faster result for improved patient management and/or outbreak control. Local management of public health response may offer advantages in terms of more effective interaction with frontline users with improved response to local needs and priorities. Although more expensive to carry this out locally the advantage(s) to the patient/community of receiving results in a shorter timeframe (the time it takes to send a sample to a centralized facility, 12–24 h) may well be worthwhile. However, these benefits are currently anecdotal rather than documented and the anticipated benefits of sequencing locally in terms of more rapid turnaround time have yet to be realized currently due to the cost saving required necessitating batching (and therefore processing delay) of samples [11].

The case for a mixed model – sentinel centralized facilities and local rapid response facilities

A mixed model of centralized and local rapid response facilities would mitigate many of the disadvantages of the either/or model. Careful attention to the management and reporting lines would be required to maintain standardization and quality management across the facilities and prevent the formation of ‘silo’ working. Samples to be processed through either the local or the central facility would need to be carefully assessed prior to submission, taking into account the cost/time benefit analysis with strong justification based on evidence of impact on patient or outbreak management.

The case for data centralization

The mixed model of both types of facility to carry out WGS is likely to be applicable for some time to come with the technologies and methodologies available currently. Only when WGS is near real time and can be carried out at point of care (POC) will this model change. In this ideal scenario, the initial WGS information taken at POC could be used for patient management, local and national outbreak and control, and national/international surveillance with optimized informatics and infrastructure. To be able to identify a pathogen from a clinical sample, predict its resistance, define and assess its virulence for (a) clinical management; (b) outbreak investigation and transmission intervention; and (c) National/international surveillance will be the ultimate goal. The current diagnostic and molecular epidemiological testing pathway for infectious disease in the UK involves NHS diagnosis at the patient management level and the broader public health areas of outbreak investigation/surveillance at the local/national population level carried out in PHE’s newly formed National Infection Service could become a single WGS test. The most efficient way this can be realized is if the data is centralized. Ultimately the centralization, analysis, organization, reporting, and coordination of the data will have the biggest impact on infectious disease outcomes.

Declaration of interest

The author is a member of the MinION Access Programme and has received travel and accommodation funding from Oxford Nanopore Technologies for a community meeting. The author has no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.