NEW YORK – Real-time whole-genome sequencing surveillance for healthcare-associated infections (HAIs) can be "affordable and time-efficient," according to a new analysis by researchers at the University of Pittsburgh.
In a preprint published in MedRxiv last month, the Pitt researchers described their laboratory and analysis methods for identifying HAIs as well as their associated costs.
"For any new program that would be implemented by a hospital system or laboratory, cost is one of the primary concerns," said Lora Lee Pless, director of the microbiology and genetic epidemiology laboratory at the University of Pittsburgh and the corresponding author of the paper. "The purpose of this current study is to show just how cost-efficient we have become."
Dubbed Enhanced Detection System for Healthcare-Associated Transmission (EDS-HAT), the program kicked off in 2016, initially as a retrospective study to explore the utility of WGS for curbing HAIs.
Encouraged by the pilot results, in November 2021, the Pitt researchers transitioned to using EDS-HAT for prospective surveillance within the university health system and sharing the results with the hospital infection prevention and control team in real time for outbreak identification.
In their preprint, Pless and her team described the protocols to carry out EDS-HAT, spanning from bacterial isolate collection from the medical center's clinical microbiology laboratory to automated DNA extraction, library preparation, and sequencing.
To maximize cost-efficiency, the researchers deployed the Illumina MiSeq for runs that had between 32 to 40 samples and the NextSeq 550 for runs with more than 40 isolates. Overall, they were able to achieve a maximum data output of 52 Gb, or 100 million reads, for an average run of 48 samples on the NextSeq 550, according to the study.
After sequencing, the data were processed with an automated bioinformatics analysis pipeline developed in-house to confirm bacterial species, determine sequence type, and compare the genetic relatedness of isolates.
Subsequently, the electronic health records of patients with genetically similar isolates were reviewed for potential epidemiological links, which were then communicated to the hospital infection prevention and control staff for appropriate mitigation.
The average turnaround time to complete the EDS-HAT workflow from sample collection to bioinformatic analysis is around 10 days, Pless said.
Additionally, the researchers conducted a cost analysis for EDS-HAT, based on data collected between March 2022 and March 2023, during which they sequenced 2,070 bacterial isolates, averaging 48 isolates per week. The per-sample sequencing cost fluctuated based on the sequencing platform and batch size, ranging from $48 to $83, the authors noted, or $65 on average.
According to Alexander Sundermann, an infectious disease professor at the University of Pittsburgh and a coauthor of the paper, people have previously criticized that labor costs were not included in the cost calculations of EDS-HAT.
To address that, the authors added a 90 percent effort from a laboratory technician and a 50 percent effort from a bioinformatics research analyst, arriving at an average of $91 per sample for EDS-HAT, a mean weekly cost of $4,293, or $223,236 per year.
The study "really provides other institutions a way forward," said David Gaston, an infectious disease physician and medical director of the molecular infectious disease laboratory at Vanderbilt University Medical Center.
The detailed protocols described in the paper, together with the other data, can serve as a roadmap for other hospitals interested in implementing a similar WGS-based workflow for HAI surveillance, he said.
Additionally, the cost analysis "gives laboratory directors like me a way to be able to talk with our senior leadership to say, 'we need to implement this' … and provide a reference point of how much this might cost," he said.
While Gaston and others are hoping to follow in the footsteps of the Pitt team, he noted that the wide implementation of next-gen sequencing for infectious disease control and clinical microbiology applications is not without obstacles.
Sequencing costs were previously a barrier, he said, but have come down noticeably during the last few years, especially with the emergence of new platform providers in the field.
Additionally, gaining experience and expertise can be a learning curve for traditional microbiology labs. However, for technologists who are experienced with other molecular techniques, NGS is "not hard to pick up," he said.
Furthermore, the lack of adequate bioinformatic resources and skills can be another barrier. "As clinical microbiologists, we are really good at anything that happens in the laboratory," he pointed out. "When it comes to what to do with that data afterwards, there are very few people that are well trained in bioinformatics."
Adding to that, few third-party bioinformatics tools are commercially available to clinical microbiology labs, Gaston said.
"I think it is an excellent paper," said Susie Jerwood, chief medical officer at London-based bioinformatics company Genpax and a former consultant microbiologist for the UK’s National Health Service. "It is going to be really useful when this comes out in print, and I will be sharing it with some of the people I know who are quite keen to get started."
Genpax previously published a model study that estimated a WGS-based surveillance strategy could prevent about 74,000 infections annually in England, saving the NHS around £480 million ($604.5 million) per year. The analysis also established a US model using available data, estimating that WGS surveillance can help prevent over 169,000 HAIs in the US, resulting in net savings of roughly $3.2 billion.
As Genpax's analysis helped demonstrate the benefits of WGS for preventing HAIs, the company has also launched a WGS analysis software called Idem that is cloud-based, automated, and "fine-tuned" for looking at transmission events, she noted.
"One of the markets that we think should be using more whole-genome sequencing is healthcare [associated infections]," Jerwood said. "We believe that if people actually did proactive sequencing, then they would find a lot more transmissions."
While the preprint is focused on tabulating costs, Jerwood noted that she would also like to see how much the process has saved.
"How many infections do they think they have [stopped]? How many deaths do they think they have [prevented]? How many people did not need antibiotics because of this? Things like that would be really nice to see," she said.
Sundermann said his group is working on another paper to model how many infections EDS-HAT has helped prevent and the associated healthcare costs it has saved.
He also pointed out that the current cost analysis did not include upfront instrument and maintenance costs.
While nanopore sequencing may enable an even faster turnaround for real-time outbreak detection, Pless said "there seems to still be a little bit of hesitation to rely on the [technology] alone."
"For single nucleotide polymorphism calling using Oxford Nanopore [data], the accuracy over time is not necessarily where we want to see it," Sundermann said, adding that the group is currently exploring to incorporate nanopore sequencing for hybrid assembly with Illumina short-read data to help look at plasmid transmission.
Currently, EDS-HAT only sequences organisms of interest collected from patients admitted to the hospital for at least three days or who had a previous hospital exposure in the prior 30 days. Meanwhile, as NGS becomes more accessible, Sundermann said the scope of sequencing could be expanded beyond the selected pathogens to make sure all possible outbreaks are captured.
"The hope is that as the cost does keep coming down, we can become more efficient at optimizing our runs, and even reduce that cost further in the coming years," he said.