Diagnosing Vicariously: Karius Uses AI to Advance Infectious ... - Genetic Engineering & Biotechnology News
A baby admitted to a hospital in Texas goes through a complete workup for infectious disease and other potential causes of disease. The clinicians at that children's hospital concluded that there is no medical reason why the baby isn't thriving. In many parts of the country, if you have a baby that's failing to thrive and a complete workup has been done, you're mandated to call Child Protective Services (CPS) so that the home can be evaluated and investigated.
A clinician then ordered a test for that child which identified several pathogens causing a disease that, over the course of multiple days or weeks in that hospital, had not been identified. The baby had a twin who was also struggling to thrive, and in that child, the test discovered numerous pathogens that were causing disease within 24 hours, preventing CPS from looking into the family and possibly separating them.
That's a story that Alec Ford, CEO of Karius, likes to tell about their microbial DNA diagnostic test.
"As a parent, there is a part of you that will never come back from that," Ford told GEN Edge. "It would've been a wrecking ball for this family. To think that an infectious disease DNA sequencing tool and a machine learning AI tool prevented a family from a human tragedy, which was easily treated with standard-of-care medicines, a family literally would've never come back the same way had this tool not been used to identify the cause of their infection."
From conception to cancer: the magic of cell-free microbial DNA
Karius began at the lab of Stephen Quake, PhD, at Stanford University, where co-founder Mickey Kertesz, PhD, was studying transplant patients and realized that among all the molecules that appear in these patients' plasma, there was another compartment of biology that was somewhat underappreciated. That separate DNA was coming from the microbes—microbial cell-free DNA. Similar to human cells, when these microbes die, they shed DNA into the bloodstream, where it can be collected and harvested for analysis. But the challenge was that there was very little DNA in the microbial cell-free DNA that could be used for analysis.
With Stanford's permission, Kertesz got the exclusive rights to the patent and started Karius, which is named after a Norwegian children's story about brothers Karius (caries) and Baktus (bacteria) who live in the mouth and cause cavities. In December 2016, Karius received its first sample. Today, Karius receives hundreds of samples from hundreds of hospitals and thousands of providers and looks at tens of thousands of samples coming from patients in active treatment through the core chemistry and computational technology developed by Bercovici.
According to Bercovici, who is CTO at Karius, it all starts with chemical innovations for harvesting cell-free microbial DNA.
"The ability to harvest microbial DNA is quite unique," said Bercovici. "You must also plan for various events that may occur during the process and the number of answers you may get. That limits the clinical value. Then, we had to pair that up with where these microbes were found and what the relevant concentrations of these microbes were."
The process starts with a hospital sending Karius a plasma sample, which they receive around 8:30 in the morning. The chemistry component takes about 18 hours or so, which then gets analyzed. End-to-end, it takes just over a day (26 hours) from collecting the sample to receiving the report. The involvement of the Karius team in that process is limited, with much of the process being automated to achieve such a quick turnaround. At the end of the process, the physicians get a simple report that lists which microbes were found and the absolute concentration of these microbes.
"Many other NGS-based technologies out there for clinical applications have turnaround times of days, weeks, or even months," said Bercovici. "We are providing this service with a next-day turnaround time."
With this technology, Karius has achieved a non-invasive way to study infections; it could be an infection in the brain, heart, or lung. We can add another layer of clarity in terms of the quantity of microbes. Commercially, the Karius test stands alone, with competition only coming from the broad range bacterial PCR and sequencing tests at Mayo Clinic Labs and the University of Washington as well as the mNGS Pathogen Dx from the University of California, San Francisco, Clinical Microbiology Laboratory, which are qualitative and not quantitative. Previously, many tests spoke to whether or not microbes existed. But the quantity is a very relevant measure because it allows you to study over time how well the patient is actually responding to treatment.
Due to some shocking healthcare statistics about infectious diseases, Karius has applied its technology to the role of infectious diseases in prenatal care, with non-invasive prenatal screening and oncology. Ford said that claims data show that more than 3 million immunocompromised patients are admitted to hospitals each year because they may have an infection. The Infectious Disease Society of America (IDSA) says that infectious disease is the main or contributing cause of death in about 50% of cancer patients. "One of the top causes that kill a cancer patient is an infection," said Ford. "If you have cancer in the US and people are talking about genomics, it's all about identifying mutations and personalizing people's treatments. For us, the degree to which people focus on infectious diseases as a lever for changing cancer outcomes is something we try to bring up and talk about as much as we can. Our test can be applied in tandem with many existing tests, whether it's cancer screening tests like Cologuard or the one from GRAIL."
Ford said that if you go talk to hematologists and oncologists, they have hundreds of these stories in 20 or 30 years of clinical practice. "It's just one tragedy after another that had nothing to do or very little to do with the underlying malignancy and everything to do with the fact that people in hospitals are still stuck with the same tools they had 20, 30, 40, or 50 years ago in terms of diagnosing," said Ford. "We try to spend as much time as we can increasing people's focus on [the infectious disease component] because we think that curve can change much more dramatically than the cancer curves have changed over the last three or four decades."
Ford said that if you go talk to hematologists and oncologists, they have hundreds of these stories in 20 or 30 years of clinical practice.
"It's just one tragedy after another that had nothing to do or very little to do with the underlying malignancy and everything to do with the fact that people in hospitals are still stuck with the same tools they had 20, 30, 40, or 50 years ago in terms of diagnosing," said Ford. "We try to spend as much time as we can increasing people's focus on [the infectious disease component] because we think that curve can change much more dramatically than the cancer curves have changed over the last three or four decades."
Beyond infectious diseases
Early on, Bercovici realized that microbial DNA was a new data type that might be useful for infectious disease diagnostics but perhaps for other human diseases as well. Bercovici said that they can examine the connection between microbial cells and inflammatory bowel disease or liver disease.
"We have this unique ability to look at microbial cell-free DNA in the blood," said Bercovici. "When you look at the human microbiome, the only place people have really looked in depth so far is in the stool. So, we have a chance to describe, measure, and find links between the human microbiome and how a disease gets worse, how it is diagnosed, and how a person is likely to react to treatment.
There are these fascinating points of intersection where certain diseases change the amount of microbial cell-free DNA that shows up in your blood. For example, the ulcerative colitis patient's blood- or plasma-based microbiome looks very different from that of a healthy person. Ford thinks that the Karius diagnostics can contribute to mucosal barrier injury patients, such as those with irritable bowel, Crohn's disease, or ulcerative colitis, and also those with pulmonary diseases where barriers are injured.
"There are other diseases where this becomes really important as a biomarker that can either predict someone's response to therapy or identify the presence of a disease," said Ford. "We're early in that journey, but the data are really strong already that we can do some good there and help people who aren't being helped in the right way today."
Karius wants to use its cell-free microbial DNA diagnostics in a couple of different areas. One is the first adoption of genomic technology for infectious disease in a hospital. "We have a couple of published data sets that show we can identify a bacteria up to a week before the first sign of illness and up to a month before the first sign of an invasive fungal infection," said Ford.
For example, the typical workup for a pneumonia patient who is going through stem cell treatment is that they'll have anywhere from 15 to 18 tests in the first seven days. Those tests will cost upwards of $18,000–$20,000. Most people will have to have an invasive test, like a biopsy or bronchoalveolar lavage, to find out what's wrong. Even for these immunocompromised patients, the cause of their infection will often never be identified. Karius wants to get clinicians to use their tests so that patients can get rapid results in terms of the cause or etiology of that infection.
Second, Karius wants to have the test available on an outpatient basis to the same population, which Ford laid out with the example of febrile neutropenia—the most common and life-threatening complication of cancer therapy. "Febrile neutropenic patients are often admitted to hospitals in the United States because they need to be able to do a thorough infectious disease workup, and often less so because of their medical fragility," said Ford. "If we can take all these folks with suspected infections and keep them at home as often as possible and out of the hospital, that's going to serve a lot of folks and give them an answer on an outpatient basis in 24 hours."
And it doesn't need to be limited to human diseases. In January 2022, the Karius team got a call from the University of Maryland asking if they were willing to participate in the efforts to treat the world's first cardiac xenotransplantation patient. "We already had the Karius platform equipped to identify human infections in such scenarios because we are often dealing with transplant patients," said Bercovici. "All of a sudden, a new emerging opportunity came up where the question was not about human pathogens but porcine pathogens."
Beyond the U.S. healthcare system, Ford wants to bring this to the point of care globally and see people across the world get answers about infectious diseases with a one-day turnaround time. "One of the technological challenges is that there's an extraordinary amount of sample preparation that goes on that has a lot of rigor and complexity to it," said Ford.
"The sequencing isn't necessarily extraordinarily different than you might find in a typical NGS workflow, but the third piece that's critical to making the quick turnaround happen is the bioinformatics pipeline and AI," Ford added. "So, even if sometime in the future someone is able to build an instrument that can do both the sample prep and the sequencing, you're still going to need the bioinformatics pipeline so that when physicians see a quantified pathogen, they can also determine if it is in fact the cause of infection."
Ford thinks that there are going to be a lot of opportunities. "We are in discussions today with health organizations in other parts of the world that want to be able to advance this quickly, first in the hospital and then elsewhere," said Ford. "Hopefully some of those will lead to very tangible partnerships, so that Singapore, the NHS in the U.K., or anybody in Burkina Faso can get a test like this. But it is still early in terms of that journey."
Comments
Post a Comment