The three COVID-19 vaccines available in the U.S. have proven to be astoundingly safe and effective for people with healthy immune systems, enabling them to shed masks and enjoy new freedoms. But for the millions of Americans who are immunocompromised, vaccination does not ensure protection.

A large, ongoing study by researchers at the University of Pittsburgh Medical Center finds that the level of protective antibodies generated by vaccination varies dramatically, depending on the reason for the weakened immune response. Only 37% of solid organ transplant patients and 55% of blood cancer patients had antibody responses, compared to 82 % of solid tumor patients, 84% of autoimmune disease patients and more than 94% of patients with well-controlled HIV infection.

The study, which builds on research published in April, is one of the first to show that blood antibody levels correlate with the ability to kill, or “neutralize,” the coronavirus in lab dishes. Yet many questions remain about which immune-compromised patients are most vulnerable and why. The “autoimmune” category, for example, covers a vast variety of mysterious diseases in which immune cells attack healthy cells.

Lindsay Ryan, an emergency room physician at San Francisco General Hospital and the San Francisco VA Medical Center, recently wrote and talked to Slate magazine about living with an autoimmune disease during the pandemic. She expected her disease and the medication she takes to treat it would lessen her response to vaccination. Instead, tests showed she made “no neutralizing antibodies.”

» READ MORE: The COVID-19 vaccine doesn’t fully protect people like me who are immuno-compromised. Science is working on it.

“I can’t live the way I do now forever, with my ambitions collapsing in on me at the age of 37,” she wrote in the Journal of the American Medical Association. “I want to work in South Africa again, hear live music, and laugh with friends at the Burmese restaurant we all love. It’s hard to conceive of the fact that if COVID-19 never abates, the guy sitting next to me in the movie theater could transmit a virus that might kill me.”

The Pitt researchers, led by infectious disease specialist Ghady Haidar, acknowledged the “urgent need to optimize and individualize COVID-19 prevention” in such patients.

“Our findings also have implications for public health guidance,” they wrote on medRXiv, a website for sharing data before publication in a peer-reviewed journal, “particularly given revised recommendations permitting vaccinated individuals to abandon masking and social distancing in most settings.”

Immune function tends to wither with age, but an estimated 4% of Americans have abnormal function because of immune-damaging diseases, such as HIV and cancer, or treatment of those diseases, or because they take immune-suppressing drugs to prevent rejection of an organ transplant.

Compromised patients were excluded from the clinical trials of vaccines because developers were rushing to get clear-cut results.

The Pitt study tested blood samples from 107 healthy medical center workers and 489 immunocompromised patients. Both groups were fully vaccinated.

While 98% of the healthy volunteers mounted a robust antibody response, vaccine failure was common among the compromised patients, except for those with HIV.

Among organ transplant patients, those with new lungs had especially poor responses, with only 22% producing antibodies, while those who got livers fared best, with 61% making antibodies to the coronavirus. Vaccine failure was more common among patients who received their transplant less than a year ago, cancer patients undergoing radiation, and patients taking certain drugs that deplete immune cells.

This lack of vaccine response is, as Ryan put it, “personally devastating,” because compromised patients are at risk of more severe, persistent COVID-19. But it also has implications for controlling the pandemic. There is evidence from cases at Pitt and elsewhere that such patients can serve as a breeding ground for more dangerous variants of the virus because it can rapidly mutate in their defenseless bodies. That, Ryan pointed out, adds “the stigma of whether certain people are risky” to be around.

Vaccine studies are now underway around the world to see how compromised patients respond, and whether they should be monitored with antibody tests or given booster shots, or both. Studies are also analyzing other types of immune response, such as T-cells, which can “remember” an invader and reactivate protection long after antibodies fade.

“Although we anxiously await the results ... it is not surprising that many immunocompromised patients will never mount an antibody response,” the Pitt researchers wrote. “Patients who fail to respond to re-vaccination should be referred to clinical trials of other preventive measures” such as synthetic antibodies and antiviral drugs.