Flu vaccine 'match' not super predictor of effectiveness

TORONTO -- From the start of this year's influenza season, public health officials assured people the flu vaccine was a good 'match' for circulating strains of the virus, which suggests it should offer good protection this winter.

But an early assessment of how well the vaccine is doing at preventing illness in those who are vaccinated suggests this year's shot reduces one's risk of being infected with influenza A -- in other words H3N2 or H1N1 -- by about 55 per cent and influenza B by 70 per cent. (Those are U.S. estimates; a corresponding Canadian analysis is underway.)

Those figures raise two related questions: Is that the best currently designed flu vaccines can do, even when the strains in the vaccine are closely related to those making people sick? And does the concept of a good match carry as much weight as has been suggested?

The answer to the first question, flu experts admit with some reluctance, is likely yes.

And a growing chorus within the flu community is also suggesting that a good match doesn't necessarily guarantee better protection. The welcome flip side is the awareness that a bad match doesn't mean the vaccine is waste of time.

"The more we've peeled back this onion called the match, the more we realize we have many questions about it that we can't answer," says Dr. Michael Osterholm, who last year published a major report on influenza vaccine and its shortcomings.

The report is an argument for redesigning flu vaccine to produce more effective products. Among its recommendations is the suggestion that until the relationship between vaccine effectiveness and vaccine virus match is cleared up, public health officials should not overstate its importance.

Nor should people use the questions surrounding the match as an excuse to reject the vaccine, says Osterholm, who is the director of the Center for Infectious Diseases Research and Policy at the University of Minnesota.

"I would not base whether I get influenza vaccine on what the match is. I would just get it," he says.

Until relatively recently, experts estimated that flu vaccine reduced one's risk of being infected by between 70 to 90 per cent, depending on a person's age, health, and the closeness of the vaccine strains to the ones making the rounds.

But a mounting pile of studies has forced down those estimates.

Studies of the vaccine used in the 2009 H1N1 pandemic, for instance, showed a median effectiveness rate of 69 per cent. And that was in the situation where the virus in the vaccine was as well matched to the ones circulating as vaccine manufacturers could have hoped for.

Now the U.S. Centers for Disease Control typically estimates flu vaccine effectiveness at between 50 and 70 per cent. The estimates for this season, released Friday by the CDC, fall comfortably within that range, giving an combined effectiveness rate of about 62 per cent.

"I think having vaccine effectiveness of 60 per cent overall is very reasonable," says Dr. Nancy Cox, head of the CDC's influenza division.

"And it's what we've come to expect with an observational study design," -- the type of study that generated the new estimate -- "and a good match."

Dr. Danuta Skowronski of the British Columbia Centre for Disease Control is in the process of generating flu vaccine effectiveness estimates for Canada for this season.

There will likely be some variation, she cautions, because the mix of viruses circulating north of the border is different than what is being seen south of it. There has been more influenza B in the U.S. so far this season than in Canada, and more H3N2 -- the more severe of the strains -- in both Canada and the U.S. than was captured in the CDC study.

Skowronski, an influenza expert, is among those trying to figure out how to make sense of the seeming disconnect between vaccine effectiveness and vaccine virus match.

She says it comes down to two options: there could be a problem with the way science is monitoring the evolution of flu viruses (the match), or with the way vaccine protection is being measured.

In the 2010-11 flu season, she notes, the BCCDC heard reports that a lot of vaccinated staff working in long-term care facilities were falling ill with flu, even though the word was the match that year was good.

So she and colleagues did a vaccine effectiveness study and found that the H3N2 component of the 2010-11 vaccine offered only around 40 per cent protection.

Skowronski suggesting that doing genetic sequencing of the main protein on the surface of flu viruses, the hemagglutinin, could offer better insights into whether the viruses in circulation have changed enough from those in the vaccine to render it less protective.

The current methods for assessing whether the vaccine viruses are a match involve tests that show whether antibodies in the blood of ferrets -- animals which are considered the best model for how flu acts in humans -- or sometimes humans will inhibit or neutralize flu viruses. These tests are meant to show how close or how distant a virus is to the protection offered by the vaccine or by previous infection.

Cox admits that the tests are "blunt instruments" and suggests better results might be available if a test involving human respiratory tract cells could be developed. But in the meantime, she says, these tests are not useless, and a combination of information from the two types of tests, plus vaccine effectiveness studies, provides a good understanding of how well the vaccine is working in a particular year.

"It's a gestalt," she said.

But this year's estimate of the effectiveness of the vaccine's protection against influenza B provides a good example of some of the shortcomings of the concept of match.

There are two families or lineages of influenza B, named Yamagata and Victoria. The lineages are sufficiently different that it is thought vaccination against one doesn't offer much protection against the other and flu vaccines currently in use only include one B virus type.

That has created a situation where in many years there has been a B component mismatch. The experts who advise on the composition of influenza vaccine choose Yamagata and Victoria surges instead -- or vice versa.

This year Yamagata is in the vaccine, but it only makes up about 67 per cent of the circulating viruses in the U.S. And at one of the sites studying vaccine effectiveness for the CDC, the Marshfield Clinic in Marshfield, Wisc., there was about a 50-50 split between Yamagata and Victoria viruses.

Still, nationally the B component of the vaccine is showing about 70 per cent effectiveness. And in the Marshfield portion of the study, vaccine effectiveness for the B component was 69 per cent, says Dr. Edward Belongia, director of the clinic's epidemiological research centre.

"So it's not sufficient to say there's a good match or there's less than a good match," says Belongia. "And in fact there are some seasons where the match is reported to be not ideal, and yet they still get vaccine effectiveness estimates in the same general range."

Belongia points to work in Australia which showed that over several years -- 2007 to 2011 -- vaccine effectiveness and vaccine virus match didn't correlate.

The study, by Dr. Health Kelly, head of the epidemiology unit of the Victorian Infectious Diseases Reference Laboratory in Melbourne, showed a vaccine effectiveness estimate of 58 per cent in 2007, a poorly matched year, and 59 per cent in 2011, when the vaccine was deemed well matched to the circulating strains.

"I think we're moving away from the concept that we just report out how good the match is and we're done, and everybody knows what they need to know," Belongia says.

"It's become clear that it's actually more complicated than that.... I'm not saying the match isn't important. It is certainly a factor. But it's not the only factor. And I think we don't understand all the factors that are driving clinical vaccine effectiveness."