For more than two months, a 73-year-old cancer patient fought COVID-19 at the University of Pittsburgh Medical Center.

His doctors did genomic sequencing on blood and respiratory samples collected as he got better and then, suddenly, worse. They chronicled the emergence of a slew of coronavirus mutations — including an ominous one that is part of strains from the United Kingdom and South Africa now sweeping around the globe, threatening the effectiveness of new vaccines.

The thing is, the UPMC researchers did the sequencing after the patient died, months before those countries identified the variants and sounded alarms in December.

Similar reports of patients who succumbed or barely survived worrisome new coronavirus strains – months before scientists were on the lookout for those particular strains – have been published by researchers in Boston and London, and at the National Institutes of Health.

These patients all had severely weakened immune systems, long-running COVID-19 infections, and treatments intended to boost their immune response. For the virus, they were the perfect hosts for speeding up natural evolution and accumulating “escape” mutations – DNA changes that would give the microbe a survival advantage and enable it to outcompete other variants if it started spreading in the community.

What this suggests, researchers say, is that mutations that make the virus more transmissible and more intractable are arising repeatedly, spontaneously, independently — and largely undetected — around the world. Scientists call it “convergent evolution.”

“It’s pretty remarkable that many of the mutations that characterize the U.K. and South African variants actually first emerged in the spring and summer of 2020, months before news reports of these new variants began to circulate,” said Ghady Haidar, a transplant infectious disease physician at UPMC and senior author of a case report on the 73-year-old cancer patient. “This is exactly what happened with our patient: We discovered all these mutations and were stunned at how quickly the virus evolved, but it wasn’t until the winter of 2020 that we realized that these are the same mutations that the U.K. is reporting.”

This phenomenon has implications for treating immune-compromised patients, for new waves of the pandemic, and, most of all, for vaccinating our way out of it.

A variant grows dominant

Viruses invade cells, hijack the cell’s molecular machinery, replicate, then break out and infect more cells.

The novel coronavirus uses its thorn-like “spike” protein to bind to a receptor called ACE2 on the surface of cells and then break in.

Cells lining the nose and lungs are particularly laden with ACE2.

As the virus copies itself, it occasionally makes errors in its genetic code – a typo here, a deleted letter there. Most mistakes are harmful or insignificant to the virus, but a few are beneficial.

The British variant, known as B.1.1.7, has a collection of mutations, including eight that make the spike protein bind more tightly to ACE2. As a result, the virus infects more cells, the viral load grows, and transmission increases. The British variant is now in 60 countries, including the U.S., where the Centers for Disease Control and Prevention predicts it will become dominant next month.

The South African strain and a similar one that emerged independently in Brazil are potentially even more threatening. Not only do they attach to airway cells efficiently, but they also block the action of certain disease-fighting antibodies — including some antibodies marshaled by the new COVID-19 vaccines.

Mutant strains are also developing ways to elude the hard-won arsenal of COVID-19 treatments, as the Pittsburgh patient’s chronic infection showed.

First cancer, then COVID-19

The patient had been controlling his multiple myeloma, an incurable but treatable blood cancer, until 2018, when he relapsed and underwent a stem cell transplant. The procedure is drastic, basically destroying and then rebuilding the immune system to get rid of the part making diseased blood cells.

Even that gave him only a respite from the cancer. In early 2020, as the pandemic hit, his UPMC specialists recommended an experimental but promising approach that would involve bioengineering his own immune system T cells to target and kill his malignant cells.

T-cell therapy had produced near-miraculous cures in other types of blood cancer, but at the cost of short-term and permanent immune system damage. Healthy B cells, a key part of protective immunity, are collateral victims of T-cell therapy, so patients need lifelong shots of antibodies to compensate for the loss.

Thirteen days after his T-cell infusion, the patient went home, his cancer on the wane.

But 12 days later, he was back in intensive care, coughing and struggling to breathe. He had COVID-19.

Besides oxygen, he received blood plasma from a recovered COVID-19 patient, and a five-day course of the antiviral remdesivir. Both were only marginally effective in clinical trials, but at that point, no other COVID-19 treatment was approved.

And no one understood yet that certain therapies might make things worse for immune-compromised patients, and better for the virus.

The virus evolves and thrives

Two major evolutionary forces are driving changes in the spike protein, as Wired magazine explains: “Interacting with ACE2, and getting clobbered by neutralizing antibodies.”

Since humans don’t evolve fast enough to alter the ACE2 receptor to foil the spike protein, that leaves the job to antibodies — and later on, the “cellular immunity” conferred by T cells and B cells.

For a century, various diseases had been treated with a low-tech therapy: antibody-laden blood plasma donated by recovered patients.

In the early, stab-in-the-dark days of the pandemic, hospitals embraced “convalescent plasma” for COVID-19. Plasma remains controversial, and early this month, the Food and Drug Administration restricted its use, because updated data show benefit only if the plasma has very high antibody levels, and only in newly hospitalized recipients with impaired antibody immunity.

The Pittsburgh patient — who had no B cells and depleted T cells because of cancer treatment — received two infusions of plasma, the last one three days before his death of respiratory failure. He also received the steroid dexamethasone, which tamps down life-threatening inflammation — by inhibiting normal immune response.

Nonetheless, he kept testing positive for COVID-19. At the time, experts believed that such lingering positivity did not mean the patient was still infectious. The Pitt researchers’ data indicated otherwise.

“Our findings suggest that certain severely immunocompromised patients with COVID-19 may require isolation longer than the 20-day period currently proposed by” the CDC, they wrote.

Equally chilling, the patient eventually had no antibodies, despite the plasma therapy. The virus, meanwhile, evolved to have numerous mutations in the spike protein, including the one that tightly binds it to the ACE2 receptor. The researchers concluded that the patient, with practically no biological defense, had “uncontrolled viral replication and overwhelming [COVID-19] pneumonia.”

They also urged “caution in use of convalescent plasma in patients with immune suppression of both T cell and B cell arms,” and suggested plasma should be limited to studies that follow such patients’ outcomes.

‘They are the victims’

So far, no one has traced transmission of a mutant coronavirus strain from an immune-compromised patient to a close contact — say, a hospital worker or family member — and then into the community. But it is plausible, considering the vast pool of immune-weakened people, including those with HIV infections, autoimmune diseases, and organ transplants.

Paul Duprex, a coauthor of the UPMC case report, hastened to say such patients should not be subjected to a horrible side effect of the pandemic: prejudice.

“We have to remember not to see them as the people who are generating these new viruses,” said Duprex, a molecular geneticist and director of Pitt’s Center for Vaccine Research. “They are the victims. They already have trouble clearing infections, so they are doubly vulnerable. Variants are also arising in normal individuals.”

There is evidence that the latest immune-boosting COVID-19 therapies, synthetic “monoclonal” antibodies, may also supercharge viral evolution in immunocompromised patients. Lab experiments, published in the journal Cell by researchers in Scotland, showed that a coronavirus variant evaded monoclonal antibodies.

What about the new Pfizer and Moderna vaccines? These and other COVID-19 immunizations are the linchpin of global hopes for building enough immunity in enough people that transmission is stopped, snuffing out the pandemic. But these inoculations work primarily by generating a torrent of antibodies that variously disable the spike protein. Immune-suppressed people were excluded from clinical trials, so how well the vaccines work in them is not clear.

So far, lab experiments show Pfizer and Moderna remain potent against the worrisome variants, although the South African strain reduced Moderna’s antibody levels sixfold. Moderna announced it is developing a booster shot, just in case it’s needed.

And it might be. The virus is ever-evolving. While that Darwinian process appears to be foreshortened in immune-compromised people, it may also be abetted by other factors: repeated resurgence in community transmission. People who reject precautions such as masking. Inept vaccine rollouts. A lack of rollouts in some poor countries.

“I am quite convinced that there are dozens, if not hundreds, of variants with similar mutations emerging around the world right now,” Tulio de Oliveira, a geneticist and bioinformatician at the University of KwaZulu-Natal in South Africa, told Wired magazine.