Must H5N1 moderate its virulence as it evolves?

[Warning: this post is fairly long and has a reasonable geek factor. It explores the question whether the virulence of H5N1 "must" moderate as the virus evolves.]

The high case fatality ratio of H5N1 (currently around 60%) is a reflection of how virulent this virus appears to be at the moment. Virulence is the ability to cause severe disease. The common cold virus and H5N1 both cause disease (both "pathogenic") but H5N1 is highly virulent, the common cold virus is not. The only thing a virus does is make copies of itself. It doesn't grow, it doesn't eat, it doesn't eliminate waste or move. It just makes copies of itself. A virus hijacks the copy machinery of the host's cell to make copies. It is a true parasite, living at the expense of the host. It doesn't care if it hurts the host or not, as long as it can replicate.

Hosts fight back and have developed elaborate mechanisms to avoid being harmed by viral freeloaders. Some of these mechanisms are non-specific (the so-called innate immune response) and some are adapted to attack particular viruses so precisely that small changes allow a virus to escape this immune antibody system. Thus the host and the virus engage in a complicated dance, with one move answered by a countermove. The virus's countermoves are mediated by genetic changes, which are one determinant of whether the virus will cause disease, how severe the disease will be, what kinds of hosts the virus can infect (e.g., birds or humans in the case of H5N1) and how transmissible from host to host it will be. We describe the whole complex of such features that allow a virus to replicate (its only task) as its genetic fitness. We don't yet understand enough of the relationship of the genetics to the biology of fitness to be able to predict what kinds of gene or protein changes to look for regarding any of these important features. But there are still tools that can be used to get some idea of what might happen. And the results are interesting and surprising.

As the virus evolves, adapting to its various hosts, the genetic changes that are selected for will change host range, disease capacity, virulence and transmissibility. It is a commonplace to say that a virulent virus like the current strains of H5N1 will evolve to moderate the seriousness of the disease it causes, i.e., to become less virulent. This is based on the commonsense idea that there is no advantage to a virus to kill the host it needs to replicate. Moderation of virulence does indeed happen in many cases, the most famous example being rabbit myxoma virus in Australia. The virus was initially so effective at killing rabbits it moderated its virulence because too few hosts were left for it to replicate efficiently. If all the rabbits had been killed, the virus would either have to find a new host or become less deadly. It became less deadly. Many parasites and their hosts have reached non-aggression pacts after millions of years of co-evolution. Influenza viruses and aquatic birds are probably a good example. This kind of reasoning and experience have led many people to expect that H5N1 will also moderate its virulence were to change genetically to a pandemic strain.

But there are many reasons why this doesn't have to happen. First of all, there are many well-known examples of diseases that have remained highly virulent for humans despite co-habitation with us throughout recorded history. Smallpox and malaria are prime examples and probably HIV as well. These are highly virulent parasites that have not moderated their virulence. It is easy to see where the logic fails. Suppose, by way of caricature, there was a parasite that produced no symptoms, so that a person did not know he was infected. Then one day, after a suitable incubation period, the person explodes in a shower of infectious blood and body fluids. In this case there would be no advantage to moderating virulence. A pretty extreme example, but it turns out it illustrates one way virulence need not moderate, by having a positive correlation between virulence and transmissibility.

Trying to figure out the evolution of virulence can be done with methods of molecular biology, but it can also be approached via mathematical modeling. The mathematical biologist Martin Nowak has just published a book, Evolutionary Dynamics, which I am reading now. It has an interesting chapter on the evolution of virulence. The tools used are systems of ordinary differential equations, which isn't as bad as it sounds (the math isn't that hard), but it does tie the results to some crude assumptions, for example random mixing of the population. This means any person has an equal chance of coming in contact with any other person. There are many ways to relax this assumption (although Nowak doesn't in this book), but the more important point is that models like this can reveal underlying dynamics one wouldn't guess in other ways. Whether these dynamics apply in the real world is a matter for study, but it suggests places to look, like a big sign in the ground that says, "Dig here." It also shows that simple logic doesn't always lead to simple behavior. So let me give a brief sketch of Nowak's exposition.

He first points out that the simplest models do not minimize virulence (this is a simple, non-linear mass-action model in a steady state population). These models evolve to maximize the basic reproductive rate, R0. R0 is the average number of new infections that would be caused by introducing an infected person into a susceptible population. We have discussed it before, here and here, at the Flu Wiki. If R0 is less than one, the parasite doesn't spread, but dies out (sooner or later). If R0 is greater than one, the virus can spread. This makes intuitive sense, since an R0 greater than one means that each infected person is producing more than one new infected person. Estimates of R0 for influenza go from 1.3 to 4 or more. It's probably somewhere around 2, but the point is that it evolves in this simple model to maximize whatever it can achieve, given the constraints of host reaction and environment. Vaccination programs work by making enough of the population immune that the effective R0 of the virus falls below 1 because most of the contacts are ineffective. Having a lot of immune people in the population is called "herd immunity."

This model is an old analysis of a simple system. I regularly teach it and its consequences to my students. It can be easily made only slightly more complicated by considering another co-circulating viral strain. We allow each strain, say the current clade 2 H5N1 circulating in southeast Asia and a mutated version, to have its own rate of transmission, virulence and overall, different R0s (call them R1 and R2). Mathematical analysis shows both strains can coexist in the population only if R1 = R2. Regardless of virulence, transmissibility, etc., the important requirement for both to co-circulate over the long run in a population is for their reproductive rates to be exactly the same. In any other circumstance, the one with the larger R eventually out competes the other unless both have R0s of less than 1, in which case they both disappear. In the special circumstance where virulence and transmissibility are independent of each other, a little further analysis shows that in all cases transmissibility will tend to increase and virulence decrease. This is the classical picture usually given of an obligatory "moderation of virulence."

But that's a pretty special case. Usually transmissibility will be related to virulence. Sometimes when transmissibility goes up virulence does, too, and often when one goes up the other goes down. They are related and not independent. This is the more usual case. For example, transmissibility of a respiratory virus that makes us cough and sneeze is greater than transmissibility of one that leaves us without symptoms. Here transmissibility increases with virulence. But the relationship might be complicated, so Nowak only presents some idealized cases tp spw what can happen. In one case virulence and transmissibility increase together, roughly proportionately (if you double transmissibility you double virulence, for example). It is obvious, even without mathematical analysis, this will tend to increase virulence because the increased transmissibility will cause more cases with increased virulence. A second idealized case is considered where transmissibility and virulence increase together up to a point but then virulence stops increasing even if transmissibility increases. This is the case where a virus changes from asymptomatic to a maximum of cough and sneeze intensity. The common cold might be a good example. In this case, Then it turns out that virulence will increase to an intermediate level, neither the maximum nor the minimum.

In these models a person was infected with only one strain at a time, although different people might be infected with different strains. The analysis was aimed at figuring out which viral strain would eventually be left and what its virulence would be. The most interesting examples, however, come from the case of superinfection, i.e., where we allow a host to be simultaneously infected by strains of varying virulences and transmissibilities. This can happen two ways, with influenza. A person can be co-infected by two different sources; or the fast mutating virus can produce more than one strain in the same person. It is important to be clear we are not talking about genetics, here, so questions of reassortment or recombination are not being considered except where they might produce a new strain of different transmissibility and/or virulence. The mechanism of multiple infection does not concern us, in other words, just the fact that it happens.

Nowak includes in the model a (mathematical) "knob" (a parameter) that he can turn to express how likely a host infected by one strain is to be infected by another. It might be true that once infected by one strain, a person is less likely to be infected by another (for example, if there were immune cross-reactivity). But it might also be that infection by one strain weakened a person so they were more likely to be infected by another strain (or another parasite; this is the case of secondary infections). He then makes a crucial, but reasonable, assumption: at some point the more virulent strain outcompetes the less virulent one in a multiply infected host. If I am infected with a mild strain and a virulent strain, the virulent strain wins. This does not guarantee it will spread throughout the population because changes in virulence are also accompanied by changes in transmissibility. I might be unlucky enough to get one of the virulent strains, but I might die too soon to pass it on so it eventually disappears from the population because the effective R0 is below one.

Nowak also uses an idealized version of the relation between transmissibility and virulence, using a relationship where virulence increases with transmissibility at low virulence but then flattens out, like the common cold example but allowing it to flatten out at a much higher level of severity (like death by cytokine storm, for example). He then runs the system for various settings of the superinfection knob, from zero (where no superinfection is allowed, i.e., essentially the previous model where you are infected by one strain at a time) and increasing it from levels where superinfection is possible but discouraged to those where superinfection is actually encouraged. Where the knob is set in the real world is a matter of biology, not mathematics, so Nowak looks at a range of cases.

With the knob at zero, we are in the previous model and the eventual virulence is at an intermediate level determined by the largest R0. But as soon as the knob moves off zero, i.e., as soon as some superinfection is allowed, we see a whole range of strains coexisting in the population of different virulences. More interesting, the lowest virulence strain that survives in the population has a virulence which is always greater than the virulence of the non superinfection model, and other virulences range upward from there. In other words, many strains are co-circulating and the virulence of even the lowest of them is higher than would occur if superinfection weren't possible. It is possible to give formulas for the lowest and highest virulences, but this is somewhat meaningless because the whole analysis is tied to some specific assumptions. The more important point is to note the qualitative behavior of virulence evolution in this simple, but biologically plausible, description of how a virus spreads in a population.

These results were obtained by computer simulation. The mathematical analysis of the general case (allowing different transmissibilities) is quite difficult. With two strains he can show that it is possible to have co-existence of two strains of unequal virulence and unequal transmissibility, or a situation where one or the other winds up winning out. Which of these possibilities occurs depends on the starting point, i.e., how many of each strain of infected people were initially seeded into the population. Different starting points can produce different of these three outcomes (co-existence or one strain or the other). Things can get even more complicated when the superinfection knob is set so that a second infection is more likely after a first infection but one of the strains is so virulent that it could not sustain itself in the population if it were there by itself (its effective R0 would be less than one). If there can be co-infection with another strain of less virulence but R0 greater than one, the increased susceptibility of the hosts infected with the less virulent strain is sufficient to boost the more virulent strain's effective R0 above 1. Thus superinfection can stabilize in the population strains of very high virulence.

When three or more strains are involved the dynamics of virulence becomes much more complicated, still, with the possibility of oscillating levels of strains for various virulences or other situations where the population switches suddenly from one strain to another or combinations thereof. This will produce sudden changes in the average level of virulence. This all happens without requiring any change in the genetics of the strains, only their co-existence and ability to superinfect in a population where the influx of fresh susceptibles balance those taken out by infection and a few other simple assumptions.

What about superinfection? We know it occurs for H5N1 and influenza generally (that's how reassortment occurs), but this analysis shows that superinfection is not advantageous for the influenza virus generally as it leads to smaller numbers of infected hosts (another feature of the analysis). So we shouldn't expect situations where there are people co-infected with large numbers of different flu viruses, since, at least in this plausible model, this would not be favored. But multiple infection does occur and the analysis suggests it can stabilize in some segment of the infected population a steady source of people infected with a highly virulent virus.

The bottom line here is that the conventional wisdom might not be so wise. Even simple interactions can produce remarkably complicated virulence dynamics. In particular, the idea that virulence "must" moderate is certainly not true. We know this by common examples (smallpox, malaria, HIV) and now by a more sophisticated analysis of how this can happen.

More like this

This is an interesting piece of work, Revere. We know that the 1918 pandemic was composed of at least 2 and probably 3 co-circulating subtypes, so we must presume that if H5N1 ever gets out of the gate, this will likely occur. Evolutionary speaking, is there any advantage to a situation in which multiple subtypes co-circulate? Possibly. It would allow for more possibilities of mutation and recombination in a situation in which further adaptation would be advantageous. Against this, is the downside of superfection.

3rd paragraph: "This is based on the commonsense idea that there is advantage to a virus to kill the host it needs to replicate."

Based on the context, I suppose this should read "there is no advantage".

In a pandemic would there not be hundreds or thousands of co-circulating subtypes for a virus as mutable as influenza...and in the end would the resulting biological chaos not lead to an end of the pandemic but not the virus. Maybe we could consider the virus environment in the same way we consider the populations it affects.

Revere. I find this material very interesting but difficult to digest...I for one would appreciate a broadening of the potential field and scientific implications of this study...Thanks.

Edmund: Quite right. Thanks. Correction made.

Tom: Nowak's analysis suggests that superinfection leads also to reduced numbers of infected hosts, so that for the virus as a whole, more and more superinfection is not advantageous. I haven't worked out that analysis in detail (and Nowak doesn't provide it) but my guess is that what happens is that there is a distribution of people in the environment, most with a single infection, the next most with a double infection and that it tails off quickly after that with multiple co-infections. Thus some subpopulations would be able to maintain high virulence strains in a minority of that subpopulation.

The bottom line here is that with strains of multiple virrulence it is not necessarily true that one strain, whether of low or intermediate virulence will necessarily outcompete the others. There will be pockets of cases of people infeced with a range of virulences, some very high. Thus if a pandemic strain evolves we shouldn't expect it to be homogeneous with respect to virulence. This is something we sort of know intuirively because there is always a range of severity of illness, but we usually think of this as being on the host side, i.e., that some people are better able to resist the effects of "the" pandemic virus. But analyses like this show that it can also be because some of the circulating bugs are much worse and they are the ones that kill some people while most of the pandemic bugs have lesser virulence. And that there will naturally be this distribution of virulence in the population, independently of host resistance. Said another way, if we were all identical, this analysis suggests some of us would get a lot sicker than others because the bug is more virulent.

Since we don't know the factors that go into virulence (and we are not all identical) we don't know if current strains exhibit a range virulences, and virulence itself is a function of the host as well. So . . .

Does that help?

Thanks for such an interesting perspective on this issue. It has seemed to me, and to many others concerned about the preparedness planning of our health authorities, that the "expected" CFR values used for planning purposes have been absurdly low. Just intuitively, it seems wildly optimistic to expect that the Indonesian CFR of say 75% can moderate down to the levels that the planning is based upon. If it turns out that this is in fact correct, that the virulence remains anywhere near it's present value, these "plans" better be revisited as a matter of utmost urgency.

don't you mean Thailand?

Peter, Thailand, not Indonesia, according to that link!

Revere, can you recommend a textbook for the basic mathematics of virulence and transmissibility, the kind of thing you were saying you teach to students? Or is Nowak's book a good source for all that material?

So it's really a race to see whether the human or the virus can evolve self-protecting features with greater speed and agility. Evolution in real time - very interesting.

Your point about not expecting homogenous virulence from this virus is well taken. Unfortunately, it brings to mind those stories from 1918 of all but one of a group of riders on a streetcar being suddenly and violently stricken with the virus, and sucumbing before their journey across town was completed. That phenomenon may be an outlier (hopefully) but so may what some suspect is a low virulence form of H5N1 now circulating in parts of Asia.

News today of yet another "restrospective confirmation" of a mild H5N1 infection, this time in Iraq. WHO has issued a number of retrospective confirmations recently. I understand they have changed their H5N1 confirmation criteria, but the Reuters article makes reference to "repeated tests using different methods." Does anyone know if the WHO has new tests that they are using? Or, are these "retrospective confirmations" like a company restating earnings due to new accounting rules?

Actually what appears to be commonsense--don't be so virulent that you kill your host--depends to a large extent on how easily you can get to a new host. Paul Ewald has written a book, Evolution of Infectious Disease, in which he spends a significant number of pages addressing the evolution of virulence. I think the jist of his argument is that the type of transmission, for example airborne vs vector-borne, goes a long way towards explaining levels of virulence. Essentially, vector-borne diseases tend to be more virulent. Further, he advocates taking the infectious organism's perspective when addressing the cost and benefits of various life history strategies.

Yes, Thailand. Long day. Sorry all, especially to the Indonesian Army who I am sure are doing a great job upholding the establishment and public health.

Whether or not this study attaches, strictly, to observed H5N1 progression, over the last few years, it seems clear to me that it is a remarkable preamble to any future trajectory of the pathogen. Nice presentation. Well stated. The most interesting approach to resolving the issue of virulence/transmissibility that I've seen to date. I feel so much better, now, Revere.

The virulence of the 1918 pandemic virus can indeed be exceeded, from what we seem to be seeing here. Would it be wide of the mark to assume that H5N1's R0 -- when it goes pandemic -- might well exceed the 1.4 to 2 that I've seen cited for previous flu epidemics and pandemics? Essentially ensuring that a pandemic is unstoppable? Average incubation/infectious periods, coupled with the general tenacity of the virus in various environmental milieus, and other associated phenomena that can enhance local and global dispersion (expanding host range, for example), all seem to be characteristics that H5N1 has accrued -- and compounded -- over the last few years. It's evolved from a tropical depression to a tropical storm, over the last few years; and all appearances continue to suggest that it has the capacity to become a hyper-Category-Five global storm, in the not too distant future.

We know that we already have high mobility, and high population density. So, it seems to me that high infectivity is the clincher, here. Coupled with a virulence level that has no compelling biological reason to adjust significantly downward from observed levels. We've never observed a pathogen of this sort working its way through a global population of this size. This study seems to conflate very nicely with -- and complement, in a sense --Taubenberger's, research on the 1918 H1N1 virus. He demonstrated -- conclusively, I believe -- that not only was an antigenic shift not necessary for H1N1 to achieve fully efficient h-to-h transmissibility, it just didn't happen, at all. H1N1 produced an extremely vicious pandemic, without it. That persuaded me that we are simply not going to experience a 1957 or 1968 event, this time around, either. All of the observed -- and thoroughly catalogued -- attributes of H5N1 seem to suggest that we are standing nearly eyeball-to-eyeball with one of the most monumentally ferocious creature's that the planet has ever encountered. We're not making any visible progress in any area that might indicate that we have any sort of handle on how to deal with this thing. The overwhelming, prevailing mindset is obvious, though: ignore it, and it will just go away. Or, maybe, throw in a prayer, or two.

Dylan, it's taken me over a year to be able to read this stuff without getting chills and you just went and messed that up completely.

Revere, please tell me he's blowing this way out of proportion.

Name: I wouldn't have put it this way, but in the couple of years we've been doing this blog, I take Dylan very seriously. I am not as dire in my outlook but the real truth is that neither of us knows. We're both guessing. But it's educated guessing. And Dylan's guesses are always very well informed.

Still, there is plenty of room for different takes on this and mine tends to be less dire than Dylan's. Probably a personality thing.

Revere: Dylan expresses some of what I've been wondering, and wanting to ask you, a learned epidemiologist, to comment on. In comparing the possible effects of an H5N1 pandemic today with the 1918 pandemic, there are of course a number of differences to consider. Besides the virulence of the flu virus itself - which you now point out will not necessarily drop appreciably to become more transmissible - there are the issues of much greater population densities now as compared to 1918, increased environmental pollution and its related detrimental effects on general health now as compared to then, as well as the potential for rapid spread through travel, especially via commercial airlines, now as compared to then. Re peripheral economic effects of a pandemic, there is a much greater global economic interdependence as well as a more futures based (credit based) economy now as compared to then. On the upside, there is better health care now compared to 1918, better worldwide communication and tracking, and (for now) no world war throwing quite so many soldiers together in filthy conditions as in 1918. So, when you factor all these things together (plus I'm sure a score that I haven't thought of) how do you as an epidemiologist predict a pandemic today with H5N1 would compare with the 1918 pandemic?

By mary in hawaii (not verified) on 19 Sep 2006 #permalink

MiH: Yes, I am an epidemiologist with many decades of experience, but I don't really know more about when a pandemic will occur, how bad it would be or if it will happen than the rest of you. There are some things we can't predict. As you point out, too, the world today is very different than the world in 1918 and the virus isn't the same, either. Some factors are positive, some are negative and they all interact in ways we can't predict. So we are all in this together in more ways than one. We are also trying to read the tea leaves together, and I am more like an ancient priest or seer than a scientist in that regard. Anything I told you would be on a par with what you might say, or Dylan or the DG of WHO or Tom or MRK. We are all idiosyncratic individuals.

I see as my task here a little bit of reality check on the more extreme views that tend to be generated by the more extreme possibilities. But you ask me about some of the central questions, not the extreme ones and I just don't know.

Let's also remember the likely collateral damage from a severe pandemic. People who are dependent on drugs and treatments and won't be able to get them: diabetics, transplants, renal dialysis patients, people who have high blood pressure, cardiac conditions, etc.

Indeed, Marissa, one more factor adding to the perfect storm. We're supposedly far ahead in terms of health & quality of life, but then you look at all the people who are only being kept alive by technology and drugs in the west and the billions existing in the filthy mega-slums of the developing world, where life is no more advanced that 1918 (apart from the TV).

Better communication is one of our preciously few advantages. But we've become very sophisticated at using it for reasons other than to benefit humanity, so even then, one wonders...

why did the 1918 pandemic 2nd wave lose in virulence
over time as it spread in almost all locations distant
to each other ? Was it an exception or did this happen
in other pandemics too ? Would we expect the same with H5N1 ?

History folks isnt on our side and while I think William may be a little to the wild side on this stuff, he isnt too terribly wrong. Some may not know that there is a whole division inside of the UN that I call the "end of days" group that monitors the disasters of the planet. They use the past, the present to maybe predict the prologue/future.

In 2000 and some change they posted up a graphic and I'll send it over to Revere here in a day or two when I find it, but it wasnt a good forecast for humanity. Each time the worlds population has surged in the last 100 years, nature has responded. I say nature rather than God because I can see nature and while a believer, I havent ever physically seen him/her. So what does the graphic show? Since the 1970's there has been a 500% (no thats not a typo) increase in disasters and that includes the diseases. It started just before Carter took office and was off the scale by the time Reagan was in. AIDS/HIV is included.

While I dont know what the criteria is to make the list it has occurences that are inexplicably imbedded in our minds. Andrew, WWII and I, H1N1,polio, pandemic 57-68, Camille, etc. They are all there. I am not a defeatist by any stretch but shinola this thing is BAD! Anyone who watches the line oscillate from the 1900's to 2000's can see its almost a straight line from beginning of the last to this one until 30 years ago. Then BANG...it starts going to Hell and surging in straight up lines that make only short valleys and then it goes again.

Divine intervention, natural selection, intelligent design and/or more intelligent destruction I dont IMHO think we have have anything to look forward to for the next 50 years or so. If H5N1 or one of its cousins comes dropping in and it does 15% mortality of all that get it then the balance of power on this world will change. Sheer numbers and the ability to grow food might be the deciding factor.

I would submit that everyone here would be a survivor including I wont prepare Revere as each of us are already trying to survive via information gathering. It was the smart rodents and/or the fast moving Adam and Eve that didnt get eaten depending on your bent that got us here. The same is going to apply again. If Revere posts it only goes to like 2000 and doesnt take into account Iraq, Lebanon, H5N1, or the WTC's, Katrina. I would be interested in an update myself but I already know the answer. Be prepared.

By M. Randolph Kruger (not verified) on 21 Sep 2006 #permalink

Anon: I think this is one of the $64,000 questions. We know that it is possible to trade virulence for infectivity but we don't know if this always the case. it could be it's a function of an evolutionary path. For example, if one strain mutates, or recombines, and ends up being more infective but less virulent, that's the way it will go. But it's possible that a strain might end up being more infective and just as virulent. When we understand how polymorphs in one gene affect the functions of the same gene, and other genes, we'll know more about the answer.

This is a question that I have given considerable thought to over the last few years in regard to H5N1. The idea that the virulence would significantly decrease so it does not kill off the host it needs to replicate has been put forth repeatedly in various forums.

"It is a commonplace to say that a virulent virus like the current strains of H5N1 will evolve to moderate the seriousness of the disease it causes, i.e., to become less virulent. This is based on the commonsense idea that there is no advantage to a virus to kill the host it needs to replicate."

If the HPAI H5N1 is infectious for possibly 48 hours before the individual shows any symptoms and it has reached a very efficient level of H2H transmission would not the first wave of infection just continue to spread with close to the original level of virulence until it began to find a sparsity of new hosts?

Even the rabbit example given above, "Moderation of virulence does indeed happen in many cases, the most famous example being rabbit myxoma virus in Australia. The virus was initially so effective at killing rabbits it moderated its virulence because too few hosts were left for it to replicate efficiently. If all the rabbits had been killed, the virus would either have to find a new host or become less deadly. It became less deadly."

Wouldn't this indicate that the virus does not have anyway of knowing how virulent it actually is until it experiences a lack of new hosts and then it has a biological imperative to become less deadly as the above quote seems to indicate.

To put it another way, if the CFR is 50%, the virus does not have a conscious awareness of the fact that it is killing off a high percentage of it's potential hosts. At the risk of making a simplification, isn't the two most common human responses to this type of Avian Influenza that 1) the host dies or 2) the host develops immunity to that specific strain of the disease.

Would the virus know the difference between these two responses and would it care? In either case it has effectively lost that individual as a future host unless it undergoes a significant mutation.

Furthermore wasn't one of the key points in John Barry's book "The Great Influenza," the idea that the virus became even more lethal in the second wave. "Now all over the world, the virus had gone through the same number of passages through humans. All over the world, the virus was adapting to humans, achieving maximum efficiency. And all over the world the virus was turning lethal." page 193

The implication being that the virus adapts to the new host and for a while becomes even more efficient at killing them as it makes an increasing number of human to human transmissions.

Back to the rabbit example, "The virus was initially so effective at killing rabbits it moderated its virulence because too few hosts were left for it to replicate efficiently." What is the implication here for humans? How few of us, Homo sapiens, have to be left to cause the virus to be unable to replicate efficiently?

Historically, all avian influenza pandemics die out as people develop immunity to the current strain, and other factors. No matter how lethal the virus is or is not, they all have the same end. Does the virus know or care how many humans are left alive? It's sole purpose is to replicate itself as efficiently as possible. And it is entirely possible this one will do so with increasing virulence once it develops an efficient human to human transmission and makes multiple passages through people.

John Barry again, page 231, "Influenza is a special instance among infectioius diseases. This virus is transmitted so effectively that it exhausts the supply of susceptible hosts."

If this question was strictly an academic one without all the extreme emotional attachments, a sustained or increasing CFR would seem to be a very high probability outcome. Although anything can happen, based on the available evidence, this may well be a scenario worth exploring.

Or as Dylan said at 07:40 PM on 09-19-06, "It's evolved from a tropical depression to a tropical storm, over the last few years; and all appearances continue to suggest that it has the capacity to become a hyper-Category-Five global storm, in the not too distant future."

By James in MT (not verified) on 22 Sep 2006 #permalink

James: The rabbit myxomatosis example is given because it is the classical example for a mechanism whereby virulence is moderated with viral evolution. This does actually happen this way. The point of the post, however, is that it doesnt have to happen that way, that there are examples where high virulence has been maintained over millenia (smallpox, malaria) and where plausible mechanisms can be given whereby high virulence could be maintained in the population for evolutionary reasons.

Since we don't know the virulence factors for H5N1 we don't know what to expect. But we shouldn't expect that only one of the possibilities must happen.

James, please read also pages 370ff, chapter 31

Thank you for your response. The reference you gave me was a bit surprising. Chapter 31, the second paragraph reads, The influenza virus is different. Since birds provide a natural home for it, influenza does not depend upon civilization. In terms of its own survival, it did not matter if humans existed or not. That makes the virus must lose some of its initial virulence in H2H transmission to insure its survival argument even less valid than even I had thought.

Your statement, Since we don't know the virulence factors for H5N1 we don't know what to expect. But we shouldn't expect that only one of the possibilities must happen. is exactly what I was attempting to express; only you did it much better. Since we dont know how virulent this will turn out to be, we can only look at the CFR to date. In the 1997 Hong Kong episode it was 6 out of 18 or 33%. Now nine years later based on the latest WHO data, with 247 confirmed cases and 144 deaths the CFR is about 58%. Based on this data it seems unreasonable to expect the only outcome must be a match for the averaged 1918 figure of 2.5%.

I am suggesting that when we make personal and family pandemic plans we might well consider Plan C, where the CFR is 25% or higher. Pandemic planning for the long term may well be appropriate. Of course we would expect the virulence to lower over time, but if the initial CFR is above a certain undetermined level then the global technological just in time system would collapse first anyway. I appreciate this opportunity to clarify my own thinking and get feedback from others.

By James in MT (not verified) on 23 Sep 2006 #permalink

interesting, that you summarize chapter 31 as:
"That makes the virus must lose some of its initial virulence in H2H transmission to insure its survival argument even less valid than even I had thought."
while it is indeed all about, how the 2nd wave
clearly _lost_ in virulence in almost all places
worldwide.

anon: I'm not sure to whom your comment is addressed, but the second wave was the worst 1918 but it was other waves in other pandemics. There doesn't seem to be a consistent pattern about which wave is the worst in a pandemic and we have very few data points.

no, I mean the 2nd wave alone in 1918. The virus lost in virulence during the 2nd wave. Almost everywhere.
Very significant. And this wave was way more virulent
than any other known influenza pandemic wave.
The fear is, that H5N1 could be similar to 1918.
(Osterholm,Taubenberger) So we should examine why
virulence went down in the 2nd wave in 1918 and whether
the same could be expected with H5N1.