Dr. Philip McMillan,  John McMillan

A new preprint shows that naive human B cells are three times less likely to recognise the latest COVID variants than the original Wuhan strain. Surveillance dashboards, meanwhile, are reportedly quiet. The signal lies in the gap between the two.

Something in the COVID picture no longer adds up. Across the United Kingdom, public health dashboards show the virus circulating at very low levels. Yet clinicians describe a steady flow of patients whose symptoms look familiar and refuse to go away, even as their tests come back negative. That mismatch between what surveillance says and what the wards see is too wide to ignore, and a new piece of laboratory research may explain it.

On 16 October 2026, researchers at the Fred Hutchinson Cancer Center posted a preprint to bioRxiv describing a quietly unsettling experiment. Jacob-Dolan and colleagues took blood drawn from seven healthy donors before SARS-CoV-2 ever reached humans, isolated the naive B cells, and asked one direct question: how many of those cells could physically recognise the receptor-binding domain of the virus? They then ran the same test against the receptor-binding domains of newer variants.

The answer, in the authors’ own words, is that the evolution of SARS-CoV-2 has moderately decreased its inherent antigenicity in humans by making the receptor-binding domain less prone to recognition by naive human B cell receptors. Three times fewer naive B cells bound the KP.2 receptor-binding domain than bound the original Wuhan version. The drop traces to two mutations, E484A and F486P, both sitting at sites that commonly drew near-germline human antibodies. Mice showed no equivalent decline, which tells the authors the shift is specific to the human repertoire.

Taken on its own, the paper reads as a careful piece of biochemistry. Set beside the autumn surveillance data, it starts to matter more. In the UK National Flu and COVID Surveillance report for week 45 of 2026, the SARS-CoV-2 line is flattening while influenza and RSV climb. Hospitals, meanwhile, still see patients whose clinical picture looks like COVID but who come back negative across the full respiratory panel. The preprint does not explain those patients, and it is not yet peer reviewed. What it does suggest is narrower and still significant: if each new variant starts from a smaller pool of naive B cells able to see it, the immune response built on a first encounter may begin from a thinner base, which is a question about immunity rather than about the dashboards.

A Prediction Catching Up With Itself

Dr. Philip McMillan has described this trajectory for several years, often alongside the Belgian virologist Geert Vanden Bossche, who coined the shorthand “Hivicron” to label the highly invasive Omicron variant the immune system would struggle to clear. The prediction has been easy to dismiss because its clock was never the point. The point was always direction. The bioRxiv data now lands within that direction.

“In the same way that the virus is evading the immune system, maybe the tests are no longer as effective at picking them up.”

The immunology behind the clinical mismatch pulls in a second strand that has been accumulating evidence since 2022. In people who received multiple mRNA boosters, particularly from the Pfizer platform, antibody class-switching drifts toward IgG4. IgG1 and IgG3 are the inflammatory antibodies that push the immune system to attack. IgG4 is, functionally, a stand-down order. It can bind the virus. It can partially neutralise. What it does not do is call in the clean-up. The vector platforms, J&J and AstraZeneca, do not appear to drive the same class-switch.

The implication is uncomfortable. In a highly vaccinated population sitting on a pool of IgG4 and now facing a receptor-binding domain that fewer naive B cells can even see, the virus does not need to be more lethal to become more persistent. It needs only to stay below the threshold at which the immune system commits fully.

Four Cohorts, Four Risk Profiles

A useful way to read the current picture is through exposure history.

Those previously infected and never vaccinated carry the broadest natural immunity. Their mucosal IgA responds to multiple parts of the virus, not only the spike, which makes evasion harder at the point of entry. The exception is the subset carrying interferon autoantibodies, who remain vulnerable regardless.

Those vaccinated before their first infection carry a narrower, spike-focused response. Protection against severe disease from older strains can be strong at the start, but as the receptor-binding domain drifts, the match deteriorates and breakthrough becomes more likely.

Those infected before vaccination begin with broader immunity, but repeated doses pull the antibody profile back towards the spike and can erode the mucosal advantage.

Those with multiple infections and multiple vaccinations occupy the most ambiguous ground. Some report almost no infections at all, and it is not clear whether that reflects true protection or an elevated IgG4 background quietly suppressing the symptomatic response. What is clearer is that the more plausible shape of risk is persistence of the virus rather than a single acute episode.

“What happens when the immune system can partially recognise the virus, but not effectively control it?”

The Clinical Signature Is Changing

The shape of acute COVID in 2020 required a naive immune system, a precondition that no longer exists in most adults. What appears to be surfacing instead is a subtler picture: mild respiratory symptoms, profound fatigue that is sometimes out of proportion to anything else, subclinical sinus involvement, and headaches that come and go. It is a spectrum that is easy to shrug off in the short term and difficult to attribute in the long term.

Running underneath these presentations is a problem the spike protein itself seems to cause. It deactivates portions of the adaptive immune response, affecting T cells, B cells and natural killer cells alike, while macrophages try to fill the gap and tip into hyperactivation. Over months and years, the pattern looks less like acute viral illness and more like immune dysregulation, one that opens the door to smouldering inflammation and the chronic metabolic complications that follow.

Why Cumulative Exposure Is The Metric

The clinical arithmetic that best fits the current data turns on the running total of spike exposure across infections and injections, rather than the severity of any single infection. Early infections passed with transient symptoms. Repeat infections added fatigue, and further rounds left a plateau of persistently unwell people who are not quite sick enough to be investigated. As receptor-binding domain antigenicity falls and viral persistence rises, the concern is a step-change rather than another slow increment.

The precise shape of what follows is not fixed. The reliable prediction is that the signal will show up in unusual places first: in symptoms that look like something else, and in diagnoses that take longer than they should. The tools that help are humble ones: nasal humming, upper airway hygiene, nitric oxide-supportive habits and careful attention to mucosal health. None of them will reverse the receptor-binding domain changes, but all of them reduce the odds of carrying the virus far enough into the body to compound the problem.

A virus that fewer human B cells can recognise has not gone away. It has learned to travel more quietly, and for the moment the surveillance systems built to find it are looking in the wrong place.