In the early, uncertain days of the coronavirus pandemic, scientists delivered one comforting pronouncement: The virus that caused COVID mutates rather slowly. If that remained true, the virus would not change much to become more dangerous soon, and any vaccine could provide years of durable protection.
What actually happened was that SARS-CoV-2 began mutating quickly, first to be more transmissible and then to evade our immunity, causing breakthrough infections and reinfections. Five years and an alphanumeric soup of variants later, most of us have gotten COVID at least once. The vaccine is still being updated to match new circulating variants. And the virus itself is still changing.
In truth, scientists were both right and wrong about the speed at which SARS-CoV-2 mutates. The rate of mutations as this virus jumps from person to person is indeed unimpressive. But scientists were not aware of a second, accelerated evolutionary track: When SARS-CoV-2 infects a single immunocompromised patient, it can persist for months, accumulating countless mutations in that time.
And if we are unlucky, that highly mutated virus might spread to others. This is the likely origin of Omicron, which appeared in fall 2021 with more than 50 mutations—an astounding evolutionary leap. Omicron looked like it had achieved four or five years’ worth of expected evolution in just months, Jesse Bloom, who studies viral evolution at the Fred Hutchinson Cancer Center, told me at the time. These mutations enabled Omicron to cause a massive wave of infections even among the vaccinated.
Scientists now believe that chronic infections in immunocompromised patients are a key driver of variants in Omicron and beyond. Even as COVID surveillance has faded in urgency, researchers are watching chronic infections for signs of what’s to come.
In retrospect, clues were there from the beginning. At the start of the pandemic, researchers in New York, including Harm van Bakel, a geneticist at the Icahn School of Medicine at Mount Sinai, began sequencing viruses from cancer patients who tested positive for SARS-CoV-2 in March and April 2020—and then kept testing positive for up to two months. The patients couldn’t clear the virus because their immune systems had been weakened by disease and by cancer treatments they received. The study, published in December 2020, concluded that immunocompromised patients with COVID might need long isolation periods, lest they unwittingly spread the virus. (These chronic infections in people who are immunocompromised are distinct from long COVID, which doesn’t necessarily involve continual shedding of virus.)
That same month, a preprint from a group led by Ravindra Gupta in the U.K. connected more of the dots. Gupta and his colleagues had found an immunocompromised patient with a lingering infection who was treated with antibodies from COVID survivors, only for the virus to acquire curious new mutations. Two mutations in particular gave the virus a slight edge in infectivity and antibody evasion. An immunocompromised host, the authors suggested, could provide the ideal viral training ground: A weakened immune system cannot wipe out the virus but can put up just enough defense for the virus to learn its tricks. In this case, the infused antibodies from COVID survivors likely contributed to whatever defenses the patient himself had, but even together they were not enough to completely clear the infection.
The virus from that particular patient probably didn’t spread far, if at all; most do not. But countless chronic infections all around the world subjecting the virus to similar immune defenses could ultimately lead to the same battle-tested mutations showing up over and over again. Indeed, mutations similar to the two in the U.K. patient soon showed up in variants such as Alpha and Omicron that did sweep around the world, Gupta told me recently. And in 2021, multiple alarming variants were found to have a different mutation that researchers in New York first observed in immunocompromised patients way back at the beginning of the pandemic. (Researchers at Mount Sinai, led by van Bakel and Viviana Simon, did match a minor variant from an immunocompromised patient to other infections in the New York City area, though it didn’t seem to spread much beyond that.)
None of the more notorious COVID variants has been directly traced to a single immunocompromised patient. But indirect evidence has accumulated over time that many variants do develop this way. Chronic infections, scientists have now observed over and over, create a distinct pattern of mutations: an overabundance of changes in the spike protein (which helps penetrate human cells) but not in the rest of the virus. This pattern is clearly found in both BA.1 and BA.2 versions of Omicron, as well as the variant that gave rise to JN.1, which drove last winter’s COVID surge. Bloom now says he has “very high confidence” that these variants came from chronic infections. The evidence is not as clear with other variants, he told me, but they could very well have evolved in the same way.
Long before COVID, Bloom had tracked the evolution of influenza during chronic infections in four immunocompromised patients; some mutations in these patients eventually showed up in the seasonal flu. I wrote about the study when it was published in 2017, intrigued by the possibility that chronic infections could predict changes in flu from year to year. At the time, this was quite a novel idea.
Flu and COVID evolution do differ in important ways, but chronic COVID infections, too, are now being examined as harbingers of the future. “Those will actually teach us a lot about the future tricks SARS-CoV-2 will come up with,” says Simon, a microbiologist at the Icahn School of Medicine at Mount Sinai. To discover what those might be, she and van Bakel are now leading a research project to create better tools for sequencing chronic infections and to better understand which immunocompromised patients are most at risk for carrying them. What they find could be a preview for the future of COVID.
