NEW YORK – Having secured a $6.2 million award from the global non-profit Combating Antibiotic-Resistant Bacteria Biopharmaceutical Accelerator (CARB-X), infectious disease startup Day Zero Diagnostics (DZD) is pushing forward with commercial plans to transition its core technology to a diagnostic platform that can be deployed in hospital microbiology labs.
The Boston-based company said its assay will quickly detect a bacterial pathogen in a patient's bloodstream and establish its drug-resistance profile, which it believes will help guide antibiotic treatment in hospitals.
Initially founded in 2016 by a Harvard-affiliated team and as a spin-off from a Massachusetts General Hospital research lab, DZD joined the Harvard Life Lab in 2017 and began exploring methods to determine bacterial antibiotic susceptibility profiles.
According to DZD CEO and cofounder Jong Lee, the firm's diagnostic platform consists of three major parts; the Blood2Bac sample prep technology, the Keynome machine-learning algorithm, and the MicrohmDB pathogen database.
Instead of using bacterial swab cultures to identify a pathogen and establish its antibiotic susceptibility profile, which could require up to two to three days while a patient develops septic shock, DZD believes its technology could quickly dictate which antibiotics doctors should administer to patients suffering from bacterial infections.
Lee explained that the Blood2Bac sample prep tool allows a clinician to process and enrich a patient's blood sample for bacterial genomic information for downstream whole-genome sequencing. Collecting about 16 ml of blood from a suspected septic patient, Blood2Bac increases the bacterial-to-human cell ratio from 1 in 1 billion to 1 in 100 cells.
"When you have a clinical sample of blood from a septic patient, they may have a ratio as small as one bacterial cell for every billion human cells in the bloodstream," Lee explained. "Blood2Bac's level of enrichment allows us to get rid of [as] much of the human DNA as possible so we can sequence the sample quickly and cost-efficiently, which is currently impossible in a clinical lab setting."
Following Blood2Bac enrichment, DZD sequences the sample using Oxford Nanopore sequencing technology. While DZD has used both long- and short-read sequencers for different portions of the workflow, Lee explained that his team is developing the clinical diagnostic version to work with Oxford Nanopore's sequencing device because he believes it offers cost advantages for small-scale applications, such as "including single sample-capable flow cells." Oxford Nanopore's sequencers do not yet have US Food and Drug Administration clearance for clinical diagnostic testing.
The company then analyzes the sequenced data with the Keynome machine learning algorithm. Keynome is trained on the company's MicrohmDB internal database of sequencing information and antibiotic resistance and susceptibility profiles. As the amount of bacterial genetic information from clinical samples in Keynome grows, Lee claims that it will continue to improve accuracy calls on several species that are important to antibiotic testing.
Lee said that the firm plans to provide answers for clinicians needing to make antibiotic treatment decisions "during the first cycle of therapy," or within 8 to 12 hours.
"The system we're trying to develop is something that is relatively comprehensive in the way we analyze bacteria, despite the diversity of morphologies and cellular properties between species," Lee said. "So, we're trying to develop a process that can comprehensively identify a diverse range of species."
Lee believes DZD will be able to prioritize clinically relevant bacterial pathogens by partnering with US hospital microbiology labs to perform its data collection. The company initially plans to target at least 50 bacterial species, but has not yet released a list of species.
Acknowledging that viral and fungal infections can also lead to sepsis, Lee said that his team will focus on bacteria because they are the most common cause of sepsis. Compared to biomarker-based infectious disease diagnostic assays, he highlighted that DZD's platform will allow clinicians to treat patients for a wide range of species because it will be able to perform whole-genome sequencing.
"While other companies are looking to apply biomarkers in specific bacteria, they're a bit limited in the species they go after, as some [only] look at five to 20 species," Lee argued. "None of those biomarker-based systems are able to recover whole-genome sequences, which we think is important for determining antibiotic resistance."
Without enriching a blood sample prior to WGS, researchers spend most of their sequencing time and investment producing DNA reads that are not important to the diagnostic tool. Therefore, the cost and time needed for sequencing increases substantially to get enough pathogen DNA to identify the species.
Instead, Lee emphasized the level of enrichment that Blood2Bac provides will allow users to shift the majority of their sequencing costs and minimize the time used to generate bacterial DNA data. He believes that the platform will be able to sequence samples for a shorter period of time at a "far less expense to generate bacterial genomes."
Commercial growth
In addition to the $6.2 million from CARB-X, DZD has raised $12 million in equity funding since launching in 2016, and previously raised $8.6 million in Series A funding in January 2019. DZD also received a $224,000 Small Business Innovation Research grant from the National Institutes of Allergy and Infectious Disease last July to improve its "ksim" algorithm, which it said will help determine infection relatedness in suspected hospital-acquired infection outbreaks.
Lee said that CARB-X provided funding because it was convinced by DZD's "ambitious" plans to not only provide diagnostic technology for "everyday use," but that Keynome would also allow for further stratification of patients with antibiotic-resistant bacterial infections.
"Day Zero is one of the first whole-genome sequencing and machine learning-type technologies that CARB-X has added to their portfolio," Lee said. "During the CARB-X evaluation process, we were able to show them data that demonstrated that ultra-high enrichment of bacterial DNA was possible at the front end, and that being able to do antibiotic resistance prediction from the sequencing data was possible on the back end."
DZD also has the potential to receive an additional $18.7 million from CARB-X after achieving certain development milestones related to technical performance and commercialization objectives.
DZD has received one patent from the US Patent and Trademark Office and has filed additional IP related to Blood2Bac sample enrichment technology.
Depending on the results of planned validation studies, which Lee said have been delayed due to the COVID-19 pandemic, DZD eventually anticipates seeking 510(k) clearance from the FDA for the diagnostic assay. Lee envisions the commercial test eventually being used in a hospital microbiology lab setting, where clinicians could run a patient's blood sample to identify the pathogen within hours.
"Our immediate plans are to transition from core technology development to product engineering," Lee said. "At the moment, we have developed the capability to do this in our lab, but [the] next stage is to do the engineering required to have an instrument that can be deployed to hospital microbiology labs."
DZD may also eventually analyze bacterial genetic information in other liquid samples such as cerebral spinal fluid and urine, where the Blood2Bac process could serve as a universal sample prep approach. Lee argued that CSF and urine usually have higher amounts of colony forming units (CFUs) — used to estimate the number of viable bacteria in a sample — than in blood, and less human DNA that would hinder bacteria detection.
"We first chose bloodstream infection because we think it's the place with the biggest unmet need, especially for sepsis, but it's also one of the hardest sample types to work with," Lee said. "The clinical sensitivity required for bacteria is in the 1 CFU per ml scale, which makes achieving bacterial isolation and enrichment from blood particularly challenging."