Could a blood test tell you how well a vaccine will work for you?
A study of more than 4,000 people found that antibodies left over from ordinary past infections predict how strongly someone responds to COVID-19 vaccination, flagging those who need extra doses.
- 26 August 2026
- 4 min read
- by Priya Joi
At a glance
- Researchers measured antibodies in blood samples taken before and after COVID-19 vaccination.
- Antibodies against three common microbes, Staphylococcus aureus, respiratory syncytial virus and human respirovirus 3, consistently predicted who would respond well.
- About 5% to 6% of healthy participants mounted weak responses despite having no known immune condition.
Not everyone is protected equally from a vaccine. While age, sex, genetics and underlying illness all play a part, clinicians have no routine way of knowing in advance whose immune system will respond well and whose will not to vaccination.
Most of what is known comes from measuring antibodies after the shot, by which point the opportunity to do anything differently has passed.
But a new study published in Cell Press Blue suggests that part of the answer may already be circulating in a person’s blood before they are vaccinated at all.
The team found that so-called “sentinel” antibodies, triggered by past previous infections, predicted strong or weak responses to COVID-19 vaccination. This understanding could make for more personalised vaccine strategies that can anticipate the body’s responses to the vaccine.
A record of every infection
Researchers led by Joshua LaBaer at Arizona State University analysed 8,687 blood samples from 4,089 people taking part in the US Serological Sciences Network, a programme set up by the National Cancer Institute during the pandemic to build serology testing capacity.
Samples came from four centres across the USA. The group included 2,445 healthy volunteers alongside 1,644 people with conditions or treatments that suppress the immune system: HIV, multiple myeloma, solid organ cancers, inflammatory bowel disease, autoimmune disease and organ transplantation.
Have you read?
For each sample, the team measured antibodies against 97 viruses, 40 bacteria and 25 of the body’s own proteins. That produced roughly two million data points, and a fairly complete record of which pathogens each participant’s immune system had met over a lifetime. They then compared those profiles against how strongly each person responded to COVID-19 vaccination.
Immune status alone isn’t the whole story
Being immunosuppressed did predict a weaker response, as expected.
After the initial two-dose course, 71.6% of transplant recipients mounted a weak antibody response, along with 48.5% of people with multiple myeloma and 43.0% of those with inflammatory bowel disease, compared with 6.2% of healthy participants.
But the clinical label was relatively blunt – within every immunosuppressed group there were people who responded as strongly as healthy volunteers.
Between 9.9% and 43.1% of participants in those cohorts had strong responses after the initial course. Meanwhile, roughly one in twenty healthy participants responded weakly, and still did so after a booster.
Belonging to a high-risk group, in other words, told doctors something useful about the odds, but very little about the individual in front of them.
Sentinel antibodies
What did separate strong responders from weak ones was the existing antibody profile.
People with higher levels of antibodies against S. aureus, respiratory syncytial virus (RSV) and human respirovirus 3, three microbes that almost everyone has encountered, went on to respond more strongly to the COVID-19 vaccine. The pattern held in healthy and immunosuppressed groups alike.
The single strongest signal came from antibodies to an S. aureus protein that the bacterium uses to scavenge iron. People in the top quarter of vaccine responders had roughly twice the level of these antibodies as those in the bottom quarter, following a booster.
The researchers call these sentinel antibodies. They are not fighting SARS-CoV-2, and they offer no protection against it. What they appear to signal is how vigorous a person’s antibody-producing machinery is in general.
Those levels also proved remarkably stable. Of the antibodies tracked across 18 months, 90% barely shifted and fewer than 2% of participants saw any change greater than four-fold. That stability is what makes them plausible as a baseline measure: a snapshot taken today should still mean something in a year.
Where AI comes in
Individual sentinel antibodies are informative, but noisy. So the team trained a deep-learning model on the whole profile at once, using 98 antibodies alongside age, sex and clinical group, and asked it to identify who would respond weakly to a booster.
The model was less effective in predicting immune response in immuno-compromised people such as those with HIV, inflammatory bowel disease, myeloma and solid tumours, and it failed in the autoimmune and transplant groups, where accuracy dropped below anything clinically useful.
For non-immunocompromised people, however, the researchers say the study could lead to more personalised vaccination strategies.
“What our study found is that certain biomarkers, when analysed with AI, can predict who is likely to respond well to a vaccine, even before they receive it. This suggests that some people may be more immune-ready than others,” said Joshua LaBaer, who led the study.
More from Priya Joi
