The 1918 influenza pandemic killed an estimated 50 million people worldwide, making it the deadliest epidemic in human history. And despite the passage of nearly a century, a number of unexplained mysteries remain concerning the 1918 pandemic virus. A mystery important to our story is that the 1918 virus suddenly and inexplicably disappeared from the world in the early 1920s. And, since influenza virus was not even identified until the 1930s, no samples of the 1918 influenza strain were isolated at the time of the pandemic. Therefore, the 1918 pandemic virus did not exist in the world until it was “resurrected” nearly 80 years later by Jeffery Taubenberger and his colleagues, who used new, state-of-the-art molecular techniques to accomplish that feat.
More recently, in 2011, two independent research groups, one led by Yoshihiro Kawaoka and the other by Ron Fouchier, modified an H5N1 bird flu (see Aside 5) from a form that does not spread between humans, to forms that very well might. The unmodified avian virus has thus far infected only about 600 humans, in almost all instances by close contact with a diseased bird. But, and importantly, the avian virus killed more than half of the infected humans; a fatality rate far greater than that of even the 1918 pandemic virus. Thus, the resurrection of the 1918 pandemic virus, and the creation of transmissible H5N1 avian influenza, may have brought into the world pathogens with the potential to unleash extraordinary devastation.
These stories are compelling scientifically, historically, and for the public policy issues that they raise. As usual, we begin with some background.
The initial outbreak of the 1918 influenza pandemic occurred in March of that year, at an Army training camp outside of Boston. Yet by the fall of 1918 it was being referred to as the “Spanish” flu, probably because Spain, as a non-combatant in World War I, then in its final year, did not censor news of the pandemic. The combatants, on the other hand, fearing that news of the pandemic might cause panic that might undermine their war efforts, repressed news of it.
By the end of the winter of 1918-1919, two billion people around the world contracted the pandemic influenza strain and, as noted above, estimates of the total number of fatalities range as high as 50 million. That amount is about twice as many as would die of AIDS worldwide during the entire first twenty years of the AIDS epidemic. Moreover, the 1918 influenza pandemic killed more people in a single year than the four-year bubonic plague that ravaged Europe from 1347 to 1351. In the United States alone there were an estimated twenty million cases (out of a population of 100 million at the time) and 850,000 dead, including 196,000 people killed during the single month of October1918.
Red Cross workers remove victims of the 1918 influenza pandemic from a house in St. Louis. St. Louis Post-Dispatch
[Aside 1: Influenza virus pandemics also occurred in 1957 (the “Asian” flu) and 1968 (the “Hong Kong” flu). However, those pandemics were much less devastating than the pandemic of 1918. The number of deaths in the United States from those latter flu pandemics is estimated to be 70,000 and 50,000, respectively.]
Bearing in mind the sheer devastation of the 1918 pandemic, consider Taubenberger’s following comments from 1997: “It is curious that the (1918) pandemic doesn’t seem to be part of the cultural memory, at least in the United States, although it was a huge event with a huge impact. Everyone hears about the Black Death in the 1300s, yet here was an infectious disease only 85 years ago that killed 40 million people and for some reason we don’t know about it.”
It also is rather curious that while the 1918 influenza pandemic killed an astonishingly large number of people, it did not cause any public panic. Apropos that, in my last posting, Jonas Salk and Albert Sabin: One of the Great Rivalries of Medical Science, I noted that the annual poliovirus outbreaks, in the pre-vaccine days of the 1940s and 1950s, did cause widespread public panic. Moreover, that was so despite the fact that poliomyelitis actually caused fewer fatalities than were caused by seasonal influenza, to which the public then and now seems rather indifferent.
Another of the mysteries associated with the 1918 pandemic is that the first cases in March 1918 were relatively benign. Then, in August, the mild infection suddenly changed into something astonishingly lethal. Initial outbreaks of the new lethal variant of the virus occurred almost simultaneously in three locations; France, Sierra Leone, and Boston, and then spread worldwide. The changed virus struck with a ferocity that stunned medical professionals.
Influenza’s genetic variability is a well known characteristic of the virus. [Indeed, it is the reason why the flu vaccine needs to be re-formulated each year.] Regardless, it is not clear how the 1918 pandemic virus suddenly became so deadly. Many of the fatalities resulting from our yearly seasonal influenza epidemics are due to pneumonia caused by opportunistic bacterial pathogens. And, while bacterial pneumonia also killed many during the 1918 pandemic, the 1918 virus itself was quickly lethal in many individuals. Some patients had massively hemorrhaged lungs, and were effectively drowning in their own blood; a scenario more reminiscent of the pathology of Ebola virus than of the fevers and aches typically associated with seasonal influenza infections.
Indeed, the 1918 pandemic virus was utterly unique in how quickly it could kill; literally overnight. There are anecdotes of people leaving for work in the morning feeling fine, and then succumbing on their way. One story tells of four women in a bridge group playing together until 11:00 in the evening. By morning, three of them had died.
Another puzzling feature of the 1918 virus was that it tended to kill the hale and hearty; individuals between the ages of 25 and 34, in the primes of their lives. In contrast, seasonal influenza epidemics cause the most fatalities in the elderly, the very young, the chronically ill, and people with weakened immunity.
The lower mortality rates among the elderly during the1918 pandemic is possibly explained by their prior exposure to an influenza strain serologically related to the 1918 pandemic virus, thus providing them with a measure of protective immunity against the pandemic virus. [On this point, and others related to the biology of influenza virus, see chapter 12 of Virology: Molecular Biology and Pathogenesis.]
The higher mortality rate among individuals between the ages of 25 and 34 is sometimes attributed to the fact that the pandemic occurred during the last year of World War I; a time when many individuals in this most susceptible group were living in crowded army camps, which predisposed them to the opportunistic bacterial infections responsible for many of the influenza fatalities in the pre-antibiotic era. Yet the virus itself was extraordinarily lethal, as noted above. Moreover, the “crowded army camp theory” can not explain why the same pattern of disease was seen in the populations of countries that did not participate in the war. So, these mysteries remain.
Recalling that the 1918 pandemic virus was absent from the world after the early 1920s, we now tell the story of Jeffery Taubenberger. In March of 1997, Taubenberger and his colleagues at the Armed Forces Institute of Pathology (AFIP) in Washington, D.C. startled virologists when they reported the sequence of the hemagglutinin (HA) gene of the 1918 pandemic virus. So, how was Taubenberger able to sequence the HA gene of a virus that was nonexistent for nearly 80 years?
[Aside 2: Influenza HA proteins are located in the viral envelope. They bind to the receptor on the target cell, and then promote fusion of the viral envelope with the plasma membrane of the target cell.]
[Aside 3: Jeffery Taubenberger is currently at the National Institute of Allergy and Infectious Diseases. The Armed Forces Institute of Pathology closed its doors in September 2011. It was founded in 1862 as a museum for specimens taken from American Civil War casualties. Over the years, the Institute’s specimen collection became legendary, and it became known for its role in diagnosing difficult civilian, as well as military cases. Moreover, its staff has included some of America’s greatest pathologists.]
Taubenberger was hired by the AFIP to create a state-of-the-art molecular pathology laboratory. Towards that end, his unit, which included molecular biologist Ann Reid, developed new procedures to recover nucleic acids from tissue samples that were fixed in formaldehyde and embedded in paraffin. Although pathologists routinely examine fixed tissues, molecular analysis of those specimens had not been possible, since the fixation can destroy nucleic acids.
Taubenberger’s initial involvement with influenza was not based on an interest in influenza per se. Instead, his intention was merely to showcase his Institute’s new procedures, and also its vast collection of specimens that had been assembled over the past century. With those purposes in mind, Taubenberger and Ann Reid put in a request for fixed tissue samples from soldiers who had succumbed during the 1918 flu pandemic.
Expecting a long wait, Taubenberger and Reid were themselves surprised when the Institute’s automated recovery system successfully retrieved their samples from the 3 million others in the AFIP collection, within a few seconds of receiving their request. The samples contained flecks of tissue from soldiers killed by the flu pandemic 80 years earlier. They were taken by doctors who, of course, had no knowledge at the time of what might be causing the soldiers’ illness.
Their interest now aroused, Taubenberger and Reid began to screen paraffin-embedded, formaldehyde-fixed patient specimens for influenza sequences, using then new, extremely sensitive molecular techniques (reverse- transcription polymerase chain reaction [RT-PCR] amplification of HA gene fragments). They hoped to increase their chance of success by focusing on specimens that showed severe lung disease. The rationale was that these samples would have come from victims who died quickly, before the virus might have cleared. [Influenza generally clears the lungs within days of the infection.] Regardless, they looked in vain for a year, until they came to a sample from Private Roscoe Vaughn, who died in September 1918 at Fort Jackson, SC., during the peak of the pandemic. In Private Vaughn’s fixed cells they found small segments of influenza-like RNA. Then, to be certain that these RNA segments were indeed from the 1918 pandemic virus, they resumed their search for positive samples until they found one from a soldier who died at Camp Upton, NY, also in September 1918. After thus confirming that their samples contained RNA segments from the actual 1918 pandemic virus, they were able to generate the complete sequence of it’s HA gene. Interestingly, the HA gene of the 1918 pandemic virus was unlike that of any other influenza HA gene that had been sequenced to date.
Having thus succeeded at reconstructing the HA gene of the 1918 virus, the next step would be to reconstruct its entire genome. However, from the very small amounts of tissue in the formaldehyde-fixed autopsy samples, Taubenberger doubted ever being able to do so. What follows is my favorite part of the story.
Dr. Johan Hultin, a 73-year-old retired pathologist, unexpectedly provided a solution to the AFIP group’s dilemma. Years earlier, in 1951, when Hultin was a graduate student at the University of Iowa, he attempted to grow live influenza virus from Alaskan Inuit victims of the 1918 pandemic, whose bodies remained buried in the Alaskan permafrost over the subsequent years. It was Hultin’s hope that the virus might have been preserved in those frozen victims. However, all his attempts to grow the virus were unsuccessful.
Hultin’s failure caused him to abandon his graduate studies and, instead, become a pathologist. Then, in 1997, after he was already retired, he happened to read the report from Taubenberger’s group describing how they reconstructed the HA gene of the 1918 pandemic virus. The report rekindled Hultin’s memories of his own earlier attempts in 1951 to grow the virus. Now, excited by his thought that the frozen bodies of the Alaskan victims might contain influenza genome fragments, from which it might be possible to reconstruct the entire genome, he wrote to Taubenberger, offering to return immediately to Alaska to obtain fresh specimens. Taubenberger agreed and, thus, Hultin eagerly returned to Alaska in 1997. There, he deliberately took tissue samples from a particularly obese woman, hoping that the combination of her fat and the permafrost might have preserved the influenza genomes. Hultin’s reasoning may indeed have saved the day, since Taubenberger’s group was able to generate the entire genome of the 1918 virus from these samples and, subsequently, to grow up the virus itself.
After Taubenberger and his co-workers successfully brought the 1918 pandemic virus “back to life,” they then tested its virulence in mice. Not surprisingly, the pandemic virus was extraordinarily lethal in the mouse model. However, the explanation for the exceptional virulence of the virus was not revealed by its genetic sequence per se. But, once the technology was available to recover gene sequences of the 1918 virus, it became technologically feasible to identify which genes of the 1918 virus accounted for its extreme virulence. Some readers may need to read the following brief aside to fully appreciate this part of the story.
[Aside 4: Most viruses contain all of their genes on a single chromosome. In contrast, the influenza genome is comprised of eight distinct single-stranded RNA segments. Five of these segments encode a single protein, while three of these segments encode two different proteins. The segmented nature of influenza genomes has important consequences in nature, as follows.
If a cell were simultaneously infected with two different influenza strains, then the genomic segments of the two strains might randomly re-assort to produce brand new strains. Indeed, this is precisely how pandemic strains are believed to arise in nature. In those instances, a human influenza genome re-assorts with the genome of a zoonotic virus, usually an avian one. In fact, the 1918 pandemic virus is at least partly avian in origin.]
Bearing in mind that influenza viruses contain segmented genomes (Aside 4), and that re-assortment of genomic segments between different strains occurs in nature, several research groups, each working independently, sought to determine which of the genomic segments of the 1918 pandemic virus might be responsible for its extraordinary virulence. In brief, it was possible to experimentally substitute each of the genomic segments of a benign influenza strain with the corresponding genomic segment of the 1918 pandemic virus. [The individual influenza gene segments were reverse transcribed and then inserted into individual plasmids. Recombinant viruses were then generated by microinjecting different combinations of these plasmids into cells in culture.] These viruses were then screened for their virulence in mice.
The results of these experiments showed that several different genes from the 1918 virus contributed to its virulence. These included the viral genes that encode two of the envelope proteins; the HA protein described above and the neuraminidase (NA), which promotes virus release from cells. The viral polymerase also contributed to its virulence.
An early hypothesis to explain the virulence of the 1918 pandemic virus was based on the contention that the virus acquired and expressed the HA gene, and perhaps the NA gene as well, of an avian influenza strain. Consequently, there might have been little if any immunity in the human population against the pandemic virus. However, Taubenberger’s group found that laboratory-generated recombinant viruses, which contained both the HA and the NA proteins of the 1918 pandemic virus, induced higher levels of inflammation in the mouse model than were induced by more benign influenza viruses. That is, the laboratory-generated recombinant viruses were actually more immunogenic than benign influenza strains. While this finding might not have been predicted, it actually is consistent with the extreme lung pathology seen in humans during the pandemic. At any rate, more research still needs to be done to better understand the virulence of the 1918 virus.
Taubenberger’s group also found some important differences between the viruses in samples from individuals infected early in the 1918 pandemic, when the virus was relatively benign, and the viruses in individuals infected after the virus became vastly more virulent. In the earlier cases, the HA protein was more like that found in avian influenza strains, while later cases had an HA protein somewhat more like that found in human influenza strains. Presumably, the avian HA gene underwent changes that adapted the virus to disseminate and spread more easily in its human host.
[Aside 5: There are 16 known serologically distinct types of the influenza HA protein in nature; only three of which, H1, H2, and H3 are found in human influenza strains. There are nine known serologically distinct types of the NA protein, of which N1, N2, and N3 are most commonly found in human strains. The 1918 pandemic virus was an H1N1 strain. Pandemic viruses generally arise when a current seasonal human strain acquires a new HA gene from an avian influenza. Other genes also may be acquired from the avian virus in addition to the HA gene. Thus, the 1957 Asian flu was H2N2, and the 1968 Hong Kong flu was H3N2. See the following aside.]
[Aside 6: In April 2009, a novel H1N1 virus (see the above aside), which originated in swine, was found in humans in the United States, Mexico, Canada, and elsewhere. Although this virus turned out to be relatively benign, its emergence caused widespread panic, due in part to the non-stop updates of new cases in the media, which created the false impression that a killer pandemic was sweeping through the country.
In May, 2009, Vice President Joe Biden told a national TV audience that he would tell members of his own family not to go anywhere where they might be in a confined space, such as an airplane, subway or classroom. But, in fairness to Biden and the media, it was net yet clear that the virus was relatively mild.
Initially, the virus was referred to as the swine flu. But, Biden’s boss, President Barack Obama, in deference to the U.S. pork industry (people were afraid they might catch the virus by eating pork), began to deliberately call this virus “the H1N1 virus.” The new designation stuck. And while it does characterize the 2009 swine flu, it likewise characterizes the vastly more lethal 1918 pandemic virus, as well as a current seasonal influenza strain. Thus, the 2009 virus was hardly the H1N1 virus.
The world was of course fortunate that the 2009 H1N1 swine flu outbreak turned out to be relatively mild. Many millions of people might have been killed. Will the public remember the episode and, consequently, be complacent in the face of a future outbreak, doubting the credibility of government warnings?]
An earlier influenza outbreak, which indeed startled virologists, took place in 1997, when the first cross-species transmission of an avian H5N1 influenza to a human was documented. The patient, a child succumbed, and there were additional lethal human infections that followed. Indeed, the H5N1 virus killed about half of the individuals it infected; a fatality rate far greater than that of even the 1918 pandemic virus. Fortunately, during the past 17 years, the virus has not adapted to spread readily from person to person. Instead, the vast majority of the 600 humans, who were estimated to have been infected, acquired the virus by close contact with diseased birds.
Next, in September 2011, Yoshihiro Kawaoka at the University of Wisconsin and Ron Fouchier of Erasmus Medical Center in Rotterdam, shocked virologists when they announced that they and their colleagues had created variants of the H5N1 virus that could be transmitted between ferrets; often considered a good model for transmission in humans. What’s more, Fouchier’s group deliberately modified the virus so that it might be transmitted through the air; a very significant modification, since transmission of avian influenza viruses between their avian hosts is via the fecal-oral route, whereas mammalian influenza viruses are transmitted via the respiratory route.
Kawaoka’s group randomly mutated the HA gene of the H5N1 virus, until they found mutations that caused it to attach to human receptors, instead of to bird receptors. Then, they replaced the HA gene from the 2009 H1N1 “swine flu” strain (Aside 6) with the mutated H5 HA gene, thereby creating a virus that contained the mutated avian HA gene, and the remaining genes from the 2009 H1N1 virus. In contrast, Fouchier’s group examined the possibility that the H5N1 virus might acquire the ability to transmit via the respiratory route by mutation alone; without re-assortment. They began by giving the H5N1 virus three mutations previously identified in the HA genes of the 1918, 1957, and 1968 pandemic viruses.
Fouchier’s virus indeed was lethal in ferrets. In contrast, Kawaoka’s virus did not kill the animals, and was no more pathogenic in ferrets than the 2009 H1N1 swine virus. But, recombinant viruses that that arise in nature might have unpredictable and very different pathogenicities. And, bear in mind that both research groups in fact demonstrated that H5 avian viruses might acquire the ability to infect mammals.
Now, consider that the resurrected 1918 pandemic virus is essentially identical to the virus that claimed up to 50 million lives during the 1918 pandemic. Moreover, consider that up to now H5N1 viruses have not been able to readily transmit between humans. But, if either of the H5N1 viruses developed in Wisconsin and Rotterdam is indeed transmissible between humans, while retaining a measure of its virulence, it might be even more life-threatening than even the 1918 pandemic H1N1 virus.
In view of the above, one may well ask what reasons could possibly justify creating such potentially dangerous viruses. A common rationalization is that these experiments provide insights into the genetic changes that might happen in nature to generate deadly pandemic viruses. A potential benefit of that knowledge might then be to enable surveillance against the emergence of such viruses, thus providing a window of opportunity to develop strategies to cope with the threat and minimize its consequences.
But regardless of the possibly enormous benefits that might result from the types of experiments described above, one could easily imagine important arguments against doing these experiments. Clearly, resurrecting the 1918 pandemic virus brought an extremely deadly pathogen back to life. And, the experiments in Rotterdam and Wisconsin may likewise have given rise to very lethal viruses. Moreover, the accidental release of these viruses, even from the most secure facility, is not all far-fetched. In this regard, in 2003 and 2004 the SARS virus “escaped” from three different Asian laboratories. Furthermore, while these experiments might be done safely in a very few laboratories in the United States and Europe, there is no global mechanism to insure that they would be done safely elsewhere. What’s more, there is concern that terrorist groups might gain possession of these viruses, or perhaps even replicate the work that gave rise to them.
So what is the bottom line? The issue is not simply whether the research is dangerous. It clearly is. And, the issue is not simply whether the research holds the promise of real and important benefits. While some potential benefits may have been overstated, they yet may one day be considerable. Thus, the real question is whether the potential benefits of the research outweigh its here-and-now risks. Experts have taken opposite positions on this question, and a heated debate goes on.
Yet a new issue arose with regard to the H5N1 experiments; specifically, whether or not the work ought to be reported in scientific journals. This issue arose over concern that the transmissible H5N1 variants might fall into the hands of individuals or groups with evil intentions or, perhaps, even be made by them. Consequently, in December 2011, the U.S. National Science Advisory Board for Biosecurity (NSABB) made the unprecedented recommendation to censor the papers that reported the work of the Rotterdam and Wisconsin groups. The papers were, at the time, under review at Nature and Science. The NSABB worried that publication of “the methodological and other details could enable replication of the experiments by those who would seek to do harm.” Thus, the NSABB recommended that the general conclusions of the papers, but not their methodologies, might be published. Later, in February 2012, a World Health Organization committee recommended that the studies be published in full.
As might be expected, there is no consensus in the scientific community over this censorship issue. On the one hand, constraints on communication are inherently incompatible with free scientific inquiry and would hinder progress in a field that significantly impacts human health. Moreover, would scientists devote years to investigating dangerous viruses, only to have their work censored in the end? On the other hand, should not the scientific community bear at least some responsibility for keeping the fruits of its research from being misused by those who would do harm? Few scientists would prefer to have individuals who are not practicing scientists, and who don’t always understand the science, making these judgments in their place. [I find it interesting that these other individuals are often referred to as bioethics, biosecurity, or bioterror “experts,” and wonder what makes them so.]
Herfst S, Schrauwen EJ, Linster M, Chutinimitkul S, de Wit E, Munster VJ, Sorrell EM, Bestebroer TM, Burke DF, Smith DJ, Rimmelzwaan GF, Osterhaus AD, Fouchier RA. 2012. Airborne transmission of influenza A/H5N1 virus between ferrets. Science 336:1534-1541.
Imai M, Watanabe T, Hatta M, Das SC, Ozawa M, Shinya K, Zhong G, Hanson A, Katsura H, Watanabe S, Li C, Kawakami E, Yamada S, Kiso M, Suzuki Y, Maher EA, Neumann G, Kawaoka Y. 2012. Experimental adaptation of an influenza H5 HA confers respiratory droplet transmission to a reassortant H5 HA/H1N1 virus in ferrets. Nature 486:420-428.
Taubenberger JK, Reid AH, Krafft, AE, Bijwaard, KR, and Fanning TG. 1997. Initial genetic characterization of the 1918 “Spanish” influenza virus. Science 275:1793-1796.