COVID-19 patients who have high levels of coronavirus in their bodies when admitted to the hospital are four times more likely to die than those with lower amounts of virus, according to a new analysis from the University of Washington.
But viral loads, which could help identify patients most at risk, are not generally measured or reported for COVID-19, says the report published this week in the journal Open Forum Infectious Disease.
“I would like to see it more routinely used,” said Dr. Alex Greninger, assistant director of the UW Medicine Clinical Virology Laboratory, which conducts much of the coronavirus testing for the state. “Unfortunately, we haven’t had this information for most of the pandemic.”
Measuring viral levels might also help contact tracers identify and focus on the most infectious people, said Dr. Andrew Bryan, medical director of the clinical laboratory at UW Medicine – Northwest and lead author of the analysis.
The UW study adds to growing evidence that high amounts of the novel coronavirus translate into more serious illness, which is true for many viral infections.
“These differences are not subtle,” Greninger said. One recent sample analyzed at the lab had a level of virus 100 million times higher than the test’s detection threshold.
But diagnostic tests for novel coronavirus are only approved by the U.S. Food and Drug Administration (FDA) to give “positive,” or “negative,” answers — not to measure viral loads, Greninger said. An FDA spokesperson confirmed in an email, “there are currently no authorized quantitative molecular tests for COVID-19.”
The type of test used at the UW, considered the gold standard for diagnosis, doesn’t directly measure the amount of virus in a specimen. Instead, it copies and amplifies bits of viral genetic material. But that can provide a good approximation of viral load based on the number of amplification cycles it takes for the test to turn positive. The fewer the cycles, the higher the concentration of virus, Bryan explained.
The trigger for higher mortality appeared to be about 22 cycles or fewer, reflecting a viral concentration at least 50,000 times higher than the test’s detection threshold.
Knowing a patient’s viral load could be helpful in managing patients even though treatment options are limited, said Dr. Jennifer Ross, an infectious disease specialist at UW Medicine who also treats COVID-19 patients at the Seattle Veterans Affairs Medical Center.
“It would let the clinician know that if someone is coming in with a very high viral load … that is a bad sign, a poor prognostic sign, as they have shown in this paper,” said Ross, who was not involved in the UW analysis.
Viral load measurements are common for many other respiratory infections and are particularly valuable for patients with compromised immune systems, Ross said.
To understand the relationship between viral loads and death, the UW team analyzed data from UW Medicine’s Northwest campus. For 109 patients admitted to the hospital, the odds of dying within a month of their initial positive test was 4.2 times higher for patients who had high viral loads on admission compared to patients who had low viral loads.
Knowing that threshold might help doctors determine when patients are out of the woods, said Bryan.
“One of the tricky things about COVID-19 is that early on, patients might seem to be doing fine and could be discharged, only to be readmitted later.”
It’s not clear why viral loads vary so widely among people infected with the novel coronavirus. “It’s the difference between a drop or an ocean,” Greninger said. “And we’re not taking that into consideration.”
Weak immune systems probably allow the virus to flourish. And how much virus a person was initially exposed to might be important, but that’s still not clear, he added.
The researchers also found patients whose bodies had already started producing antibodies against the novel coronavirus by the time they were admitted were about half as likely to die as patients without antibodies — though because of the small numbers of those patients, the results were not statistically significant.