Deepa Bhojwani recalled feeling lucky her 2-year-old cancer patient with COVID-19 bounced back quickly after being seen for a fever in the emergency department at Children’s Hospital Los Angeles. But in the months that followed, an unnerving thing happened.
The toddler’s condition flip-flopped from sick to well to sick again and so on, resulting in six hospitalizations over 196 days — and each time, the boy was positive again for the coronavirus.
Bhojwani, a leukemia specialist, wondered whether the lab results might be a mistake or — terrifyingly — a rare case of reinfection. But when the medical team dived deeper, it found evidence the original virus had been inside the boy all along, evolving into more efficient forms.
“We know viruses mutate,” Bhojwani said. “But we didn’t expect this.”
The child’s case report, shared online last week, is one of about 15 similar ones involving immunocompromised patients and recently published in medical journals or preprint servers that have become an important puzzle piece for researchers seeking to understand the origins of the coronavirus variants taking over the world. Those reports lend support to an intriguing theory that some individuals with weakened immune systems due to cancer, HIV or other illnesses may act as incubators for new mutations — an idea bolstered by the astonishing fact that the B.1.1.7 variant, first recognized in the United Kingdom, and B. 1.351, from South Africa, were found in some hospitalized patients months before their discovery in the outside world.
“The evidence points to these immunocompromised patients as an accelerated cauldron of evolution,” said David Pollock, a professor of genomics at the University of Colorado School of Medicine.
Coronaviruses are generally believed to have strong proofreading capabilities, making them more stable than other viruses and therefore better targets for therapies and vaccines. In fact, when the mystery virus in Wuhan, China, turned out to be a coronavirus, many scientists were initially relieved, pointing out the virus’s rate of change was approximately half that of influenza and one-fourth that of HIV.
“Early in the pandemic, when you talked to virologists, they said evolution would happen — but they thought that it would happen slowly,” said Ghady Haidar, an infectious-diseases doctor at the University of Pittsburgh who treated one of the patients in the case studies. “That it’s happening much faster is extremely surprising.”
New forms of viruses, called variants, arise from “mistakes,” such as substitutions or deletions of genetic material that occur in the copying process as the virus replicates, and they can happen at any time, in any place, in anyone. But the speed of the coronavirus’s evolution in recent months has led scientists to speculate that some kind of unanticipated situation — an especially hospitable environment or, on the flip side, new pressures on the virus — may be at play.
There’s no indication that specific patients spread the variants to the larger community, but researchers speculate that some medical treatments, such as convalescent plasma and monoclonal antibodies, may play a role. In some, but not all, of the patient case reports made public, the number and variety of mutations increased after those therapies were administered, leading some researchers to speculate the virus was changing itself to evade them.
On a clinical level, the cases have triggered intense debate about how to best manage immunocompromised patients with the coronavirus — whether additional testing or sequencing should be more routine, different isolation procedures considered and alternate treatment protocols followed.
The discussion comes at a time when the United States is bracing for the B.1.1.7 variant to become dominant in coming weeks. Studies have confirmed its increased transmissibility, and there is also growing evidence it may cause more severe illness. The United Kingdom, where it was first identified, has remained largely shut down since December. On Feb. 26, the Czech Republic declared a state of emergency as cases surged because of the variant, and Finland followed suit March 1.
As of this week, the B.1.1.7 variant is estimated to account for about 32% of U.S. cases.
Michael Osterholm, who served as a COVID-19 adviser to President Biden during the transition and is director of infectious-disease research and policy at the University of Minnesota, expressed concern the United States is moving in the opposite direction with the loosening of social distancing restrictions in recent weeks and that this may lead to deaths that could be avoided.
“We could not be offering a more welcoming environment for the transmission of B.1.1.7 than we are doing right now,” he said.
— — —
The case reports of persistent coronavirus in immunocompromised patients are eerily similar despite involving different underlying conditions, geographic locations, genders and ages. They include a man in his 70s in London with B cell lymphoma who experienced large shifts in the virus in his body after getting convalescent plasma therapy and died on Day 102; a 73-year-old man seen at the University of Pittsburgh Medical Center, who had multiple myeloma and died on Day 74; a 45-year-old man in Boston with a disorder in which the immune system mistakenly attacks normal proteins in the blood who died on Day 154; a woman in her 70s in Munich with follicular lymphoma who died on Day 156; a 60-year-old man in Ann Arbor, Mich., with a different type of lymphoma who on Day 156 left the hospital for home hospice.
There are also studies involving an adult with HIV, and of a child and a 21-year-old with leukemia who also were observed to have prolonged infection with mutating virus.
William Hanage, an epidemiologist and evolutionary biologist with the Harvard T.H. Chan School of Public Health, said he has been struck by the “remarkable convergence” in the types of mutations in the case studies, suggesting they arose simultaneously in different hosts, in different parts of the world. He believes this is a strong indicator that the changes are associated with properties of the virus that boost its survival — perhaps by elevating its ability to infect, replicate and evade efforts to slow it down.
“It shows there is more than one way to skin a cat,” he said. “The fact that there is overlap indicates that the virus is finding the same solution” to challenges it encounters.
In Boston, for example, one patient who was treated at Brigham and Women’s Hospital over the summer developed both the B.1.1.7 variant, as well B. 1.351, which been found to be “markedly more resistant” to neutralization by convalescent plasma and vaccines, according to a paper published in Nature this month.
Jonathan Li, an infectious-disease specialist and co-author of the case study about the Boston patient, which was published as a letter in the New England Journal of Medicine, said he was surprised to find a host of new mutations — “more mutations than we have seen up to that point anywhere else.”
“Those were not present at the very beginning, but we saw them developing over time in front of our eyes,” he recalled. “This individual really was a harbinger of what was to come.”
Similar mutations were seen about the same time approximately 575 miles away in Haidar’s patient at the University of Pittsburgh Medical Center. Haidar said that the emergence of these variants in so many different places is unlikely to be a coincidence.
“Our minds were blown,” he recalled.
Pollock, a co-author of the study on the London patient, said the fact that most of the people described in the case studies died is “obviously worrisome.”
As he sequenced 23 samples taken from the 70-year-old man with B-cell lymphoma, he recalled watching in distress as the genetic code of the pathogen evolved.
As the virus copied itself, new mutations arose, Pollock said. But as soon as one became dominant, it was pushed aside by another one more powerful, in eight cycles of one-upmanship until the patient — who at a few points had been so well doctors hoped he might recover — was dead.
“One lineage would be dominant and then be replaced by another,” he said. “But then [it would] gain some mutation that made it sort of win more again. So this battle was going back and forth, accumulating mutations.”
Pollock said the adaptations he saw appeared to increase the virus’s rate of replication and protect it from disease-fighting antibodies: “You can’t really say the adaptation caused the deaths, but it’s certainly implied.”
— — —
Millions of Americans — as many as 3% to 4% of the population, according to one estimate — have weakened immune systems for different reasons, and there is growing scientific literature about how their interactions with different bugs can affect those around them. Drug-resistant tuberculosis is believed to be associated with immunocompromised patients, for example, and some have proposed that influenza’s evolution may be as well.
While active infection with the coronavirus for most people lasts only for about eight to 10 days — considered too short for many mutations to occur — it can last for 160 days or longer in those with weakened immune systems, providing more opportunities for change. Scientists have described virus replication in such patients as akin to making photocopies of photocopies, with later copies being more blurred or different from copies of the originals.
On Feb. 18, the Centers for Disease Control and Prevention issued new guidance recognizing that some severely immunocompromised patients may shed the virus beyond 20 days and require additional testing and consultation, but did not offer specific recommendations.
One of the biggest questions facing doctors treating coronavirus patients with weakened immune systems is how the combination of therapies for SARS-CoV-2 and for cancer, or other underlying illnesses, affect viral mutation.
“I will tell you, to be completely honest, we don’t know the best thing to do,” Haidar said.
The ethics are tricky, he said: Even if doctors were to weigh the theoretical risk of contributing to mutations versus doing all that they could to save the life in front of them, the issue is that “we really don’t have anything to offer these patients” beyond convalescent plasma and monoclonal antibodies. So he said he and his colleagues will continue to give the treatments.
Throwing vaccines into the mix is another question. They are not expected to be very effective in immunocompromised people, Bhojwani said, and no one really knows how the virus might react to a combination of cancer therapies, the vaccine and COVID-19 treatments.
“These are all questions we keep asking ourselves,” she said.
One common thread among those reported to have prolonged COVID-19 infections is a deficiency in what are known as B cells, which produce virus-attacking antibodies — a situation that may allow the virus to replicate with abandon.
According to preliminary findings posted March 2 while a paper is under peer review, Jennifer Dien Bard, who runs the virology lab at Children’s Hospital Los Angeles, and her colleagues described seeing two patients with such a condition who kept returning to the hospital and who had persistently positive coronavirus tests with high viral loads.
One was a 21-year-old man and another was the 2-year-old treated by Bhojwani.
Both had been diagnosed about a half-year earlier with B-cell acute lymphoblastic leukemia or ALL, the most common type of childhood cancer. The disease is characterized by the presence of immature white blood cells in the bloodstream and bone marrow that are unable to fight off infection.
The young man’s infection with SARS-CoV-2 was picked up during routine testing after he had undergone chemotherapy. He was initially asymptomatic but then showed up in the emergency room with a fever and a dry cough. X-rays showed fluid in his lungs, and over the next few days his condition worsened, and he was put on supplemental oxygen.
He was discharged on Day 45 — but then fell ill again. He was subsequently hospitalized four more times with persistent fever and respiratory issues. Doctors treated him with weekly infusions of convalescent plasma, and after more than 250 days, he was finally negative for infection.
The toddler was discharged on Day 6, then readmitted on Day 14, discharged again on Day 40, only to be readmitted on Day 43, discharged on Day 51 and readmitted on Day 162. He had been treated with the antiviral drug remdesivir and was negative for infection on Day 196.
The virus inside the 21-year-old developed 14 mutations while the virus inside the toddler developed 21. Similar to the adult patients who have been studied, the pace accelerated later in their illnesses.
Interestingly, Dien Bard said, several of the new mutations involved the spike protein — the area of the virus that binds to our cells — but they are different from the ones reported in most adult case studies. And they are not among the variants of concern circulating around the world.
The evidence strongly suggests that a patient’s inability to mount an immune response “plays a significant role in allowing the virus to really thrive and replicate,” she said. Still, she noted, “these are two interesting cases among hundreds of other cases. The likelihood of this happening even to immunocompromised patients is rare and low.”