One of the most common comments I’ve read on Twitter from my northern hemisphere buds these past weeks is – “hey look at those bad influenza (flu) numbers in Australia…that means we’re in for a bad season up here soon!”
But does it mean that?
When it comes to flu seasons, which is the inFLUencing hemisphere, north or south? Or is it neither?
12 per cent of an answer
I’m going to partly answer this using just one timespan and by looking at just one genetic part of one subtype of seasonal flu virus, the A/H3N2. A limited study but it will hopefully make the point that like all things flu, who gets what viruses when, if at all, is complex. In the end, I’ll highlight a larger and better scientific study as well.
But first, we need to frame all the possibilities that are packed into that phrase above,”a bad season”.
Whatdya mean “bad”?
We might each mean different things when we talk about a bad flu season so let’s pick that apart first up.
When a flu season begins may indicate badness to you
If the question is just about when the flu season starts – d’uuh. That’s easily seen by looking over past years to see when flu peaks happened. They usually have reliable timing. It’s this reliability that is how we know that in the southern hemisphere during late 2018 and in 2019 (#flunami), something unusual has been happening. Flu cases started to climb waaay earlier than usual, and our seasonal peak seemed to have shifted many weeks earlier than was normal (last happened in 1995 ).
Bad may be about the size of the flu season
If the question is about the size of the flu season remember: one large season in one hemisphere may follow one in the other, but they could each be made up of very different viruses.
Take the example of Chile right now. They, like New Zealand and South Africa, are having earlier than normal high levels of flu, just like Australia. Chile is predominantly experiencing an A/H1N1 season, whereas at the start of Australia’s early season we had A/H1N1 and A/H3N2 in almost equal proportions, then A/H3N2 booted A/H1N1 out, and then the numbers were further bolstered by FluB/Victoria.
To me, a different viral mix is a different underlying cause. These are both early seasons, but they’re being driven by different viruses – so it’s really two concurrent but distinct early seasons.
Size and timing are interesting and they are what we see, but what virology lies beneath? This is a major driving factor that can control the season’s pace, virulence, size and length.
I don’t think we can say – “ohhh, the south is going to give us northerners a bad flu season” – if the viruses differ between the hemispheres.
To know this level detail – and inform public and professional understanding of what’s going on in their communities – we have to look at the viruses involved to see where the first ones originated, track their movements and report all of that clearly, using simple language.
We don’t sample flu viruses very densely, nor do we sample from all nations, so we might miss some of the important origins and steps. Also, if we sample with bias (culturable viruses or only viruses present at high levels or causing severe disease), we might be missing significant viral populations. And on communication, of course we don’t talk about flu viruses in much digestible detail.
It might be the dominant flu viruses that are “bad” to you
You might hear “A/H3N2” and think that’s going to hit the elderly hard.
So if the question is whether one hemisphere will get the same genotypes of flu virus as the other hemisphere just had, the figure we’re about to look at suggests almost certainly kinda maybe, yes, no. Probably not. Don’t bet on it.
The “Australian / British / US / Chinese / Canadian killer flu”
There’s a lot to unpack in the next bit. Let’s start with that example I mentioned. In the figure below we’re looking at the Australian flu season of 2018, the following 2018/19 flu season in the United States and then the lay of the land in Australia after that in 2019.
The aim is to see if we can reliably state whether the US season followed the Aussie one.
We’re going to use trees in the example.
There are three trees in the next image. They are made solely using the genetic sequences of the hemagglutinin (H) gene. The images were lifted from the nextstrain platform’s seasonal flu page. On the right of those trees, I’ve added in colour-coded boxes. These are to more clearly identify where the same genetic clades occur between seasons.
Before reading on, if some terms were new to you, I recommend the introductory post I wrote recently.
These three trees show a few things:
- Australia saw viruses of the 3c2 A1b/135K clade in its earlier season, as did the US later. Levels were about the same. They remained about the same in Australia afterwards. This clade looks well spread and well embedded at each site. Or so it simplistically seems in this example (see the caveats below).
- Australia initially had a little 3c2 A1b/131K but the US went on to develop quite a lot; perhaps it was seeded from the south? Australia has since grown its A1b/131K clade and it’s doing exceptionally well in2019. Thanks to the US?
- 3c2 A1b/135N was present in Australia in 2018 but didn’t appear in the US and subsequently disappeared from Australia in 2019.
- 3c2 A2/re had a big Australian presence in 2018 but was barely seen in the following US season or in 2018/19 Australia
- 3c2 A3 was just barely present in Australia in 2018 and 2019 but didn’t show up among US sequences in 2018/19
- The 3c3.A clade was small and genetically distinct (if that can be inferred from the branching patterns) in Australia’s 2018 season. It was a huge and diverse presence in the 2018/19 US season. Australia later saw some of those strains in 2019, but not many.
This all suggests that there is no reliable harbinger of, or direction to, the H3N2 makeup of the northern flu season based on what happened in the south. There might be some direction among some viruses and clades – but it’s not a unidirectional tide. We are constantly shuttling travellers back and forth and some will be infected.
What dominates depends on which viruses are already there, what the population is immune to, and how distinct (related to immunity) an imported strain is. The flu season usually parallels a particular set of climatic conditions that likely produce more favourable conditions for flu survival and transmission. Flu is always circulating, but it reaches epidemic levels when the conditions are right.
The world is round and has neither beginning nor end
There is a lot of complexity to flu virus circulation. Flu viruses replicate and spread all the time in the tropics, with wet-season peaks. There are also known to be major hubs for fluA viruses which include China, the US and Southeast Asia.
While it would be great to accurately predict what’s coming we can still only perfectly do that, based on what’s already been.
Knowing which antigenic virus types are present at the start of flu season could help predict which strains of the virus will predominate later on. Also, how well they match the pred
What does it mean?
We reliably and regularly do not see the exact same genetic variant of A/H3N2 travel from Australia to the US and drive the US season.
If you think about that for a minute, it shouldn’t be a big surprise. If we did see this pattern, we wouldn’t need the annual February and September World Health Organization consultation of global experts to decide the flu vaccine components each year. The North could just use the vaccine that the South had! But modelling, analysis, prediction and revision is necessary every single year, for each hemisphere. This is because each country harbours its own “stocks” of flu viruses into which imported cases feed.
Flu viruses enshroud the planet, are and always changing wherever they find purchase. They’re unpredictable.
In amongst those A/H3N2 clades we looked at, there were identical strains alongside strains that were different. Some seemed to fade away and some rise up. Detailed genetic analysis is the only tools we have to track which ancestor they might have branched from in the past.
To answer questions about where any given flu virus we have in the lab came from, we need to refer to up-to-the-moment and detailed genetic surveillance data.
A much more detailed approach looking at flu virus movement on a grander scale has already been described in the paper Phylogenetic Analysis Reveals the Global Migration of Seasonal Influenza A Viruses. That study used complete flu virus genomes – all 8 genetic segments, not just the haemagglutinin gene as was used above. That study concluded global flu virus movement contributes to seasonal emergence of fluA epidemics, but that this migration doesn’t have any clear pattern of direction.
I couldn’t agree more. So when next you see someone write:
- “Australian trends are a good approximation of what will happen in the Northern hemisphere”
- “CountryX may follow this trend”
- “Australia is often used as the predictor for the upcoming influenza season in CountryX”
- “This does not bode well for us in CountryX this coming winter!”
Remember – they’re mostly wrong.
Caveats and extra stuff
This post uses just one flu subtype, H3N2 and a limited number of hemisphere flu seasons. It also only looks at the H gene and not complete viral genomes. And it just eyeballs the whole thing – hardly a detailed genetic comparison! These trees also represent what people have sequenced and uploaded to GISAID and there is no detail about how biased the selection of viruses was. From just 👀 the clades, we can’t tell for certain whether one group of viruses in one hemisphere arrived in the other, further evolved through mutation and expanded that clade or vice versa. Comments about “seeding” are untested.
H3N2 is the fastest evolving of the flu viruses. It gains a “new look” by altering the proteins on its coat that are recognised as foreign and attacked by our immune responses. In doing so, A/H3N2 viruses replace their old look. Or at least, the newly unfashionable viruses don’t show up causing noteworthy illness at large scale again.
This gradual ongoing change to the coat proteins (which are called “antigens”) means that a once strong anti-flu immune response no longer finds any target to attack. The process of such change is called “antigenic drift”. It happens through the selection of genetic mutations which created changes that the virus could use to its advantage; in this example they let it get away from attack.
- Integrating influenza antigenic dynamics with molecular evolution
- Phylogenetic Analysis Reveals the Global Migration of Seasonal Influenza A Viruses
- Integrating influenza antigenic dynamics with molecular evolution
- Recommended composition of influenza virus vaccines for use in the southern hemisphere 2019 influenza season and development of candidate vaccine viruses for pandemic preparedness
- Global migration dynamics Underlie Evolution and Persistence of Human Influenza A (H3N2)
- NATIONAL INFLUENZA SURVEILLANCE 1995