There’s still a lot of uncertainty about how exactly the immune system responds to the SARS-CoV-2 virus. But what’s become clear is that re-infections are still very rare, despite an ever-growing population of people who were exposed in the early days of the pandemic. This suggests that, at least for most people, there is a degree of long-term memory in the immune response to the virus.
But immune memory is complicated and involves a number of distinct immune features. It would be nice to know which ones are engaged by SARS-CoV-2, since that would allow us to better judge the protection offered by vaccines and prior infections, and to better understand whether the memory is at risk of fading. The earliest studies of this sort all involved very small populations, but there are now a couple that have unearthed reasons for optimism, suggesting that immunity will last at least a year, and perhaps longer. But the picture still isn’t as simple as we might like.
Only a memory
The immune response requires the coordinated activity of a number of cell types. There’s an innate immune response that is triggered when cells sense they’re infected. Various cells present pieces of protein to immune cells to alert them to the identity of the invader. B cells produce antibodies, while different types of T cells perform functions like coordinating the response and eliminating infected cells. Throughout this all, a variety of signaling molecules modulate the strength of the immune attack and induce inflammatory responses.
Some of those same pieces get recruited into the system that preserves a memory of the infection. These include different types of T cells that are converted into memory T cells. A similar thing happens to antibody-producing B cells, many of which express specialized subtypes of antibodies. Fortunately, we have the means to identify the presence of each of them.
And that’s the focus of a major study that was published a couple of weeks ago. Nearly 190 people who had had COVID-19 were recruited, and details on all these cells were obtained for periods as long as eight months after infection. Unfortunately, not everyone donated blood samples at every point in time, so many of the populations were quite small; only 43 individuals provided the data for six months after infection, for example. There was also a huge range of ages (age influences immune function) and severity of disease. So the results should be interpreted cautiously.
Months after infection, T cells in this population still recognized at least four different viral proteins, which is good news in light of many of the variants in the spike protein that have been evolving. T cells that specialize in eliminating infected cells (CD8-expressing T cells) were present but had largely been converted to a memory-maintaining form. The number of cells declined over time, with a half-life of roughly 125 days.
Similar things were seen with T cells that are involved in coordinating immune activities (CD-4-expressing T cells). Here, for the general population of these cells, the half-life was about 94 days, and 92 percent of the people who were checked six months after infection had memory cells of this type. A specialized subset that interacts with antibody-producing B cells seemed to be the relatively stable, with almost everyone still having memory cells at over six months.
So overall, as far as T cells go, there are clear signs of the establishment of memory. It does decline over time, but not so rapidly that immunity would fade within a year. However, for most of the cell types examined, there are some individuals where some aspects of the memory seems to be gone at six months.
The B side
Like T cells, antibody-producing B cells can adopt a specialized memory fate; cells can also specialize in producing a variety of antibody subtypes. The first paper tracked both antibodies and memory cells. Overall, the levels of antibodies specific to the viral spike protein dropped after infection with a half-life of 100 days, the number of memory B cells increased over that time, and stayed at a plateau that started at about 120 days post-infection.
A second paper, published this week, looked at the trajectory of the antibody response in much more detail. Again, it involved a pretty small population of participants (87 in this case), but monitored for over six months. A bit under half of them had some long-term symptoms after their initial infections had cleared. As with the earlier study, the levels of antibodies the researchers found declined in the months following the infection, dropping by anywhere from a third to a quarter, depending on the antibody type. Intriguingly, people with ongoing symptoms tended to have higher levels of antibodies across this period.
But when the team looked at antibody-producing memory cells, they noticed that the antibodies were changing over time. In memory cells, there’s a mechanism by which parts of the genes that encode the antibody pick up a lot of mutations over time. By continuing to select those cells that produce antibodies with a higher level of affinity, this can improve the immune response in the future.
That seems to be exactly what is happening in these post-COVID patients. At the first sampling time, the researchers identified the sequences of many of the genes that encode antibodies against coronavirus proteins. At the time of the second check months later, they were unable to find 43 of these initial antibody genes. But 22 new ones were identified, arising from the mutation process—by six months, the typical antibody gene had picked up between two and three times the number of mutations. In some cases, the authors were able to identify the ancestral antibody gene that picked up mutations to create the one present at six months.
The system seems to be working. One of the early antibodies was unable to bind some of the variants of the spike protein that have evolved in some coronavirus strains. But the replacements with more mutations could, suggesting it had a higher affinity for the spike protein than the earlier version. While the average antibody had similar affinities at the early and late time points, specific antibody lineages saw their ability to neutralize the virus increase.
The immune system has ways of preserving the spike protein to select for improved antibody variants after infections are cleared, and that may be part of what’s going on here. But in a number of participants (under half of those tested), there were still indications of active SARS-CoV-2 infections in the intestine, even though nasal tests came back negative. So it’s possible that at least some of the improved binding comes from continued exposure to the actual virus.
The big picture
Let’s emphasize again: these are both small studies, and we really need to see them replicated with larger populations and more consistent sampling. But at least when it comes to antibodies, the consistencies between these two studies are a step toward building confidence in the results. And those results are pretty good: clear signs of long-term memory and that the immune system’s ability to sharpen its defenses seems to be working against SARS-CoV-2.
Beyond that, the T cell results, while more tentative, also seem to hint at a long-term immunity. But there, the results aren’t as consistent, with different aspects of T cell immunity persisting in different patients. The researchers divided the different aspects into five categories and found that fewer than half their study population still had all five categories of memory present after five months. But 95 percent of them had at least three categories present, suggesting the persistence of at least some memory. At this point, though, we don’t really understand what would provide protective immunity, so it’s difficult to judge the meaning of these results.