TORONTO -- A new study sounds a cautionary note for work that is being done to try to develop vaccines to protect against all subtypes of influenza.

The research describes a phenomenon in which vaccination against one strain of flu actually seems to raise the risk of severe infection following exposure to a related but different strain, an effect called vaccine-associated enhanced respiratory disease.

The scientists say it's not currently known why the effect happens. Nor is it clear that it would be seen in other species -- this research was done in piglets -- or with the kinds of flu vaccines used to protect people. But they suggest the findings should be considered during the development and assessment of experimental universal flu vaccines.

"We need to ask questions and make sure the appropriate experiments and studies are done to ensure this will or will not happen in different vaccine scenarios," one of the senior authors, Dr. Amy Vincent, said in an interview Wednesday.

The study was published by the journal Science Translational Medicine.

Vincent is a research veterinary medical officer for the U.S Department of Agriculture's agriculture research service, based in Ames, Iowa. Her co-authors are also with the USDA or with the U.S. Food and Drug Administration.

The authors cautioned against drawing a line between what happened to the pigs in the study and what might happen with people. For one thing, the animals were "naive" -- they had never before been exposed to flu viruses or flu vaccines. As well, the study used vaccine made from whole, killed flu viruses. The injectable flu vaccines used to protect people are made from parts of killed flu viruses.

Still, the finding is reminiscent of something that was observed in people in Canada during the 2009 H1N1 pandemic.

Dr. Danuta Skowronski, a flu expert at the B.C. Centre for Disease Control in Vancouver, spotted a surprising and unsettling trend when she looked at data on people in British Columbia who contracted H1N1 in the spring and summer of 2009. People who had received a seasonal flu shot the previous autumn were more likely to contract the new pandemic strain.

Her findings, which were initially dismissed by many in the global influenza research community, were later replicated in studies done in other provinces as well, leading some to dub the phenomenon "the Canadian problem."

Later, Skowronski and colleagues ran an experiment in ferrets that was similar to the work Vincent is now describing, with comparable results.

Asked about Vincent's study, Skowronski said she's been watching results from that group for years, and even cited vaccine-associated enhanced respiratory disease when she was asked by skeptics to describe what might explain her unexpected results in 2009.

"I think ... what they're showing is a biological mechanism that warrants further evaluation in terms of its relevance to the use of seasonal vaccines in humans and what that may mean for the next pandemic threat," Skowronski said.

"It's concerning obviously because if this is the mechanism, then it means there needs to be a lot more ... attention paid to these universal vaccine candidates that are targeting that stalk antibody."

Her comment about the stalk antibodies refers to a popular target for flu vaccines currently in development.

Most flu vaccines aim to stimulate antibodies to the main protein on the outer shell of flu viruses, the hemagglutinin. Shaped like a lollipop, the hemagglutinin is what attaches to the cells a virus is attempting to invade.

It's a frustrating target for flu vaccine designers. There are 17 known hemagglutinins, which give flu viruses the H in their name. (Most don't currently infect people.) The hemagglutinins on H1 viruses look different than those on H3 viruses, and antibodies to one don't protect against another.

Even within a subtype -- H1, for instance -- there are different strains, and a vaccine against one might offer lots, some or no protection against another. And all these hemagglutinins are constantly changing, which is why flu vaccines have to be updated almost every year.

But these differing and mutating targets are the head of the lollipop. By contrast, the stalk or stem of the protein is relatively unchanged across viruses, a recent discovery that has led to renewed optimism that a universal flu vaccine could actually be made.

"There is a very high enthusiasm in the field right now about the possibility of a universal vaccine," says Dr. James Crowe, a viral immunologist with the Vanderbilt Vaccine Center at Vanderbilt University in Nashville, Tenn.

"The rationale is that if you removed the head, then the stem would be either exposed or just be the dominant protein being presented (to the immune system) and therefore the stem could more frequently induce antibodies and those antibodies are very cross-reactive."

But Vincent's study raises a red flag. She and her colleagues immunized piglets with an H1N2 vaccine, then exposed them to the H1N1 virus that caused the 2009 pandemic. Though both viruses were H1s, they were quite different genetically.

Instead of being protected, the H1N2-vaccinated pigs developed more severe disease than exposed pigs that hadn't been pre-vaccinated. When the researchers tested the blood of the vaccinated pigs, they found high levels of antibodies that attached to the stalk of the H1N1 hemagglutinin, but not to the head of the protein.

Vincent said she and her colleagues are still trying to figure out why this produces more severe disease. But the theory is that while the stalk antibodies can't neutralize or kill the invading viruses, they do bind to them. And that may actually help the viruses enter the cells and multiply to higher levels -- the paper calls them "fusion enhancing."

Vincent said her group got its first inkling of this phenomenon about 10 years ago and has done several studies like this one with different pairs of vaccines and viruses, including different H3N2 viruses.

Asked if this mechanism might explain "the Canadian problem," Vincent said more research would need to be done to answer that question. "I think it's not in my area of expertise and I think it's sort of outside of the data that's available. But I certainly think it's a legitimate question to ask."

Skowronski and others suggested the work demonstrates the complexity of influenza immunology -- the science of how the viruses interact with immune systems. "The problem is everybody wants influenza to be simple and be like other vaccine-preventable diseases. And it's not," Skowronski said.

Infectious diseases expert Dr. Michael Osterholm said with influenza, there is always a complicated interplay between the virus and the person the virus infects, one that is influenced by what viruses and vaccines the person's immune system has previously encountered.

"I think as we move forward with vaccine for influenza, we need to understand ... the subsequent host-virus interaction with any response we get," said Osterholm, who is director of the Center for Infectious Diseases Research and Policy at the University of Minnesota.

"It really drives home the need to be very cautious about what are we actually accomplishing."

Both Crowe and Osterholm stressed that this phenomenon probably isn't exclusive to vaccination against influenza. Sequential infection with the right -- or wrong -- combination of flu viruses could also probably trigger this, they said.