For weeks, Americans looked on as other countries grappled with case reports of rare, sometimes fatal blood abnormalities among those who had received the AstraZeneca vaccine against COVID-19. That vaccine has not yet been authorized by the FDA, so restrictions on its use throughout Europe did not get that much attention in the United States. But Americans experienced a rude awakening this week when public-health officials called for a pause on the use of the Johnson & Johnson vaccine, after a few cases of the same, unusual blood-clotting syndrome turned up among the millions of people in the country who have received it.

The world is now engaged in a vaccination program unlike anything we have seen in our lifetimes, and with it, unprecedented scrutiny of ultra-rare but dangerous side effects. An estimated 852 million COVID-19 vaccine doses have been administered across 154 countries, according to data collected by Bloomberg. Last week, the European Medicines Agency, which regulates medicines in the European Union, concluded that the unusual clotting events were indeed a side effect of the AstraZeneca vaccine; by that point, more than 220 cases of dangerous blood abnormalities had been identified. Only half a dozen cases have been documented so far among Americans vaccinated with the Johnson & Johnson vaccine, and a causal link has not yet been established. But the latest news suggests that the scope of this problem might be changing.

Whether the blood issues are ultimately linked to only one vaccine, or two vaccines, or more, it’s absolutely crucial to remember the unrelenting death toll from the coronavirus itself—and the fact that COVID-19 can set off its own chaos in the circulatory system, with blood clots showing up in “almost every organ.” That effect of the disease is just one of many reasons the European Medicines Agency has emphasized that the “overall benefits of the [AstraZeneca] vaccine in preventing COVID-19 outweigh the risks of side effects.” The same is true of Johnson & Johnson’s. These vaccines are saving countless lives across multiple continents.

But it’s also crucial to determine the biological cause of any vaccine-related blood conditions. This global immunization project presents a lot of firsts: the first authorized use of mRNA vaccines like the ones from Pfizer and Moderna; the first worldwide use of adenovirus vectors for vaccines like AstraZeneca’s, Johnson & Johnson’s, and Sputnik V; and the first attempt to immunize against a coronavirus. Which, if any, of these new frontiers might be linked to serious side effects? Which, if any, of the other vaccines could be drawn into this story, too? How can a tiny but disturbing risk be mitigated as we fight our way out of this pandemic? And what might be the implications for vaccine design in the years to come?

Theory 1: Platelet Problems

The leading theory behind the blood abnormalities associated with the AstraZeneca vaccine traces to a case from late February. A 49-year-old nurse in Austria who had received the vaccine developed clotting in her veins and arteries, along with a low platelet count. That’s an odd combination: Platelets are like the bricks of the clotting system, held together with a mortar of molecules called coagulation factors; when your body has a shortage of these bricks, it typically has a hard time building clots.

There are a couple of exceptions, though, as the science journalists Kai Kupferschmidt and Gretchen Vogel explained in March. In a condition called disseminated intravascular coagulation, so much clotting occurs that circulating platelets get used up. A rare reaction to a blood-thinning drug called heparin, which is safely prescribed to millions of people in the U.S. every year, can also produce the same signal. When that occurs, heparin attaches itself to a molecule called platelet factor 4, which is released by damaged platelets. The body mistakenly treats this complex as a pathogen before going on to attack the platelet cells directly, which sometimes leads to clots.

A hematologist named Sabine Eichinger first treated the nurse in Austria. She had an inkling that the problem resembled the heparin reaction, so she reached out to Andreas Greinacher at the University of Greifswald in Germany, who has published extensively on that topic over the past three decades. Other, similar cases following immunization with the AstraZeneca vaccine had emerged, and Greinacher rallied his team. They put in 18-hour days in the lab to analyze the patients’ blood samples.

Last Friday, Greinacher and his team published a paper on their findings in the New England Journal of Medicine. In a press briefing, he said they’d analyzed blood from several dozen people who had experienced blood abnormalities after exposure to the AstraZeneca vaccine, and that every single person tested positive for antibodies against platelet factor 4, and against platelet factor 4 joined with another molecule.

On the same day, a separate group in Norway published similar findings from five patients there who had received the AstraZeneca vaccine. Then, in a meeting this week of the Advisory Committee on Immunization Practices, which helps the CDC make vaccine recommendations, it was reported that five of the six American patients who developed this same blood condition after receiving the Johnson & Johnson shot had been tested for antibodies to platelet factor 4—and all were positive. “It is, in my opinion, absolutely clear that there’s a causal relationship” between the presence of these antibodies and the abnormal clotting, Greinacher had said at last Friday’s briefing. “There’s no doubt about this.”

Greinacher has given the blood condition the tongue-twisting moniker of “vaccine-induced immune thrombotic thrombocytopenia” or VITT. A lot remains to be confirmed, but if this theory is borne out, scientists will have to determine whether the AstraZeneca vaccine is the only one—or whether the AstraZeneca and Johnson & Johnson vaccines are the only two—that can generate this haywire autoimmune response. Greinacher’s theory has a vital, near-term implication too: Patients who present with blood clots are often given heparin as treatment; if what they have is really VITT, then this standard treatment may only make things worse.

Theory 2: The Spike’s the Problem

If the theory above is correct, then some unknown component of a COVID-19 vaccine can, in very rare situations, spur an autoimmune reaction against platelet factor 4. One potential culprit is the all-important spike protein—the one used by coronaviruses to infect cells, and involved in every available COVID-19 vaccine as a training target for the immune system. As it happens, the structure of the platelet factor 4 molecule has some similarities, in its curves and crevices, with that of the spike protein. In other words, training up the body to go after the spike protein could yield antibodies that stick to clotting-related proteins too.

Greinacher does not believe this to be true. In an article posted online, which has not yet been published in a scientific journal, he and others found no evidence that the troubling antibodies found in people with VITT were also binding to the spike protein. That’s “fantastic news,” he said in last Friday’s press briefing, because “otherwise there would have been the risk that many other vaccines might have caused the same problem.”

But others think the role of the spike protein warrants further examination. Michel Goldman, an immunologist at the Free University of Brussels and a former director of Europe’s Innovative Medicines Initiative, told me that unvaccinated people with severe COVID-19 have also experienced strange clotting events that mimic what happens in the heparin-related autoimmune disorder. He speculates that the spike protein could be the source of this problem—but not because it looks a bit like platelet factor 4.

It’s possible, he says, that the spike protein interacts with platelets directly and causes them to secrete platelet factor 4. One mouse study published last year by scientists in China found that the spike protein could indeed activate platelets that had been engineered to carry a cell receptor for the coronavirus. Meanwhile, another study from December, not yet published in a journal, suggests that the spike protein can damage the cells that line our blood vessels. Goldman noted that these cells, when perturbed, are known to release molecules that bind to platelet factor 4, which might in turn become a target for rogue antibodies. All of this could potentially contribute to the symptoms seen in VITT.

If the spike protein is involved in the blood disorders that are now under investigation—and that is a gigantic “if”—that isn’t necessarily bad news for all COVID-19 vaccines. While it’s true that each one deploys the coronavirus spike protein, they don’t all use the same version. The Pfizer, Moderna, and Johnson & Johnson vaccines, for example, have a modification in the spike protein that keeps it stabilized and elongated. “The only EMA-approved vaccine that includes a nonstabilized form of the spike is the AstraZeneca vaccine,” Goldman said.

Theory 3: A Suspect Sequence

The AstraZeneca vaccine is built from a non-replicating version of a virus that is thought to be harmless to people, and has been engineered to produce the coronavirus spike protein. That’s the “adenovirus vector” approach, also used in the Johnson & Johnson vaccine, and in Russia’s Sputnik V. But AstraZeneca’s differs from the others’ in that its engineered virus has also been fitted with a small portion of the genetic sequence for a molecule called tissue plasminogen activator, or tPA. When naturally produced in the body, tPA is thought to have a role in regulating normal blood-clotting processes. A manufactured form of the same molecule is often given as a clot-busting drug to people who have suffered strokes. The AstraZeneca vaccine uses a noncoding snippet of the gene for tPA that boosts production of the spike protein, with the aim of producing a stronger immune response.

Some scientists in Europe have wondered whether AstraZeneca’s inclusion of a partial tPA sequence could throw off normal clotting. The European Medicines Agency said in March that there is insufficient evidence to support this theory. And if tPA were somehow involved in the strange blood disorder, there’s no reason to expect that the same problem would show up among people who received vaccines that don’t include the sequence, such as Johnson & Johnson’s.

It’s possible that some other aspect of the adenovirus technology might be in play, which would explain why the only two vaccines to raise major red flags so far both use that delivery system. A couple of past laboratory studies of blood cells have found evidence that adenoviruses can trigger platelets to secrete clotting-related molecules. Still, it’s important to remember that regulators have not yet confirmed a causal link between the unusual blood disorder and the Johnson & Johnson vaccine.

Theory 4: The Mixed Bag

It could be that lots of things are going on at once. We shouldn’t assume that the same thing is happening in every affected patient, says David Lee, an emergency-medicine doctor at NYU Langone Health. “When you just say, ‘Hey, everybody who had a clotting issue, yeah, let’s pile up all those cases,’ you’re mixing a bag,” he told me. Lee and others have shown that COVID-19 patients produce many different antibodies that go after the body’s own proteins, including one that gloms onto a clotting molecule called annexin A2. If platelet factor 4 is involved in some cases of abnormal clotting after vaccination, Lee said, then annexin A2 or other factors might play important roles in others. He pointed, for example, to another molecule involved in clotting, called fibrinogen. The molecule shows up at abnormal levels both in people with severe COVID-19 and some of those who have suffered rare blood disorders following the AstraZeneca vaccine.

Lee is not alone in thinking that this story might have many different plots. Saskia Middeldorp, a vascular internist at Radboud University Medical Center in the Netherlands, says she was initially convinced by the VITT theory. “I was really, really excited because I figured, ‘Wow, now we’ve got a case definition. It’s not great, of course, that this is occurring, but we know better what to do.’” But when she and her colleagues analyzed blood samples from one Dutch patient who developed strange blood issues after receiving the AstraZeneca vaccine, they did not find antibodies against platelet factor 4. (Greinacher says there might be differences in the laboratory assays used to detect the antibodies, which could explain the different results.)

At any rate, it’s clear that multiple mechanisms could lead to the formation of clots when people get sick with COVID-19. That’s one reason it’s so much safer to develop immunity from vaccination than from being naturally infected, according to David Kaslow, the vice president for essential medicines at PATH, a Seattle nonprofit. When your body is presented with one specific portion of a virus—the spike protein, for example—there may be some tiny risk of immune dysfunction. But this is far, far better than the alternative, where the body is exposed to a full pathogen. Natural infections may lead to all sorts of unexpected effects, including platelet damage and abnormal blood clots. A vaccine that whittles down the virus to a single subcomponent is a huge benefit, Kaslow says, because it provides fewer opportunities for something to go wrong.

Setting aside the question of whether it was good policy to pause the use of the Johnson & Johnson vaccine in the U.S., or to limit the use of AstraZeneca in Europe, more information is likely to emerge in the coming weeks.

The Vaccine Safety Datalink, a collaboration between the CDC and nine large health-care organizations tracking data from anonymized medical records of millions of patients, was already watching for some clotting events, according to Ed Belongia, an investigator with the program and the director of the Center for Clinical Epidemiology and Population Health at Marshfield Clinic Research Institute, in Wisconsin. He told me the Vaccine Safety Datalink had not identified a risk from any of the three vaccines now authorized for emergency use in the U.S. The program is in the process of adding an alert for cerebral venous sinus thrombosis, the form of clotting that turned up in the six Americans who received the Johnson & Johnson vaccine.

The pandemic is still raging around the globe, and there is a real urgency to solve the riddle of the rare blood abnormalities linked to the AstraZeneca shot, and to figure out whether other vaccines carry the same risk. But recent history suggests that health officials might have to decide whether to restrict or resume vaccinations before scientists reach agreement. During the swine-flu pandemic of 2009, for example, a vaccine known as Pandemrix was given in Europe, and then some evidence emerged in Finland that it might be linked to narcolepsy. (The vaccine never became available in the U.S., and it’s no longer manufactured.) Although many scientists consider the narcolepsy risk well established, there is still no consensus on this point.

The wrath of COVID-19 will likely stretch into the years ahead, and it’s plausible that the debate over risks related to COVID-19 vaccines will do the same. The 2009 swine-flu strain never really went away, and many scientists now believe that this coronavirus is here to stay as well. But the current pandemic is unlike any public-health crisis in our lifetimes, and it has led to an unprecedented pace of scientific progress. This could lead us to a quicker, more definitive resolution of the vaccine-related clotting issue. If that happens, what’s been learned will help us continue to navigate the promises and pitfalls of new vaccine technologies, and it might even shed light on how clotting goes awry because of COVID-19 itself.