Jeffery K. Taubenberger
In 1918 and 1919, an influenza pandemic of unprecedented virulence swept the globe, leaving up to 40 million people dead in its wake (Crosby, 1989). Although the virus responsible for this catastrophe was not isolated at the time, it is now possible to study the genetic features of the 1918 virus using fixed and frozen tissue specimens recovered from victims of the pandemic (Taubenberger et al., 1997; Reid et al., 1999). The study of the 1918 virus is not just one of historical curiosity. Because influenza viruses continually evolve by mechanisms of antigenic shift and drift, new influenza strains —as emerging pathogens—continue to threaten human populations. Pandemic influenza viruses have emerged twice since 1918, in 1957 and 1968, and the risk for future influenza pandemics is thought to be high. An understanding of the genetic makeup of the most virulent influenza strain in history may facilitate the prediction and prevention of future pandemics.
The influenza pandemic of 1918 was exceptional in both its breadth and its depth. The first wave of influenza in the spring and summer of that year was highly contagious but caused few deaths. But in late August a virulent form of the disease emerged and swept the globe in 6 months. The main wave of the global pandemic occurred in September through November of 1918, killing more than 10,000 people each week in some
U.S. cities. Outbreaks of the disease spread across not only North America and Europe but also as far as the Alaskan wilderness and the most remote islands of the Pacific. Large proportions of the population became ill, with 28 percent of the U.S. population estimated to have been infected. The disease was also exceptionally severe, with mortality rates greater than 2.5 percent of the U.S. population, compared to less than 0.1 percent in other influenza epidemics. Incredibly, some isolated populations had mortality rates of over 70 percent (Crosby, 1989).
Furthermore, in the 1918 pandemic most deaths occurred among young adults, a group that usually has a very low death rate from influenza. Influenza and pneumonia death rates for 15 to 34 year olds were more than 20 times higher in 1918 than in previous years, with 99 percent of excess deaths occurring among people under age 65. It has been estimated that the 1918 influenza epidemic killed 675,000 Americans, including 43,000 servicemen mobilized for World War I. The impact was so profound that it depressed the average life expectancy in the United States by more than 10 years (Reid and Taubenberger, 1999; Linder and Grove, 1947).
Influenza A viruses are negative-strand RNA viruses with a segmented genome with eight gene segments coding for 10 proteins. They are known to infect a wide variety of warm-blooded animals, including birds and mammals. Antibodies against the hemagglutinin (HA) protein prevent receptor binding and are very effective at preventing reinfection with the same strain. HA can change in order to evade previously acquired immunity either by antigenic drift, whereby mutations of the currently circulating HA gene disrupt antibody binding, or by antigenic shift, in which the virus acquires an HA of a new subtype. Fifteen HA subtypes of influenza A are known to exist in wild birds (H1 through H15) and provide a source of HAs that are novel to humans (Webster, 1997). Both the 1957 and 1968 pandemics resulted from a shift in HA, and in both cases the hemagglutinins of the pandemic strains were closely related to avian strains (Bean et al., 1992; Schafer et al., 1993). For a pandemic influenza strain to emerge, the virus must have a hemagglutinin antigenically distinct from the one currently prevailing; this hemagglutinin cannot have circulated in humans for the past 60 to 70 years; and the virus must be transmissible from human to human (Kilbourne, 1997). The other surface protein, neuraminidase (NA), which has nine described subtypes (N1 through N9), also has been shown to shift in pandemics.
The natural reservoir for influenza virus is thought to be wild water-fowl. Periodically, genetic material from avian strains emerges in strains
infectious to humans, and because pigs can be infected with both avian and human strains, they are thought to be an intermediary in this process. In 1979 an avian influenza A virus (without reassortment) entered the swine population in northern Europe, forming a stable viral lineage (Scholtissek et al., 1983). Influenza strains with recently acquired genetic material were responsible for pandemic influenza outbreaks in 1957 and 1968. Until recently there was no evidence that a wholly avian influenza virus could directly infect humans. However, in Hong Kong 18 people were infected with an avian H5N1 influenza virus in 1997, and six died of complications after infection (Subbarao et al., 1998; Claas et al., 1998). To understand the emergence of pandemic influenza strains, it is extremely important to determine how influenza viruses move between species. Learning more about the relationship of the 1918 influenza virus to swine and avian viral strains is one of the primary goals of this project.
The broad goal of research in this area has been twofold: (1) Where did the 1918 influenza virus come from, and how did it infect people? (2) Are there any genetic features of the sequence that would provide insight into the virulence of this strain? In total, 78 autopsy cases of victims of the lethal fall wave of the 1918 pandemic were examined for this study. Seventy-four of these consisted of fixed tissues. The majority of these individuals died of secondary bacterial pneumonia. Because they often had clinical courses longer than 1 week, it was extremely unlikely that any of the tissue samples from these cases would still retain influenza RNA. However, a subset of individuals died within 1 week with very unusual and characteristic lung pathology, including massive pulmonary edema or hemorrhage. While these pathological changes have occasionally been observed in other influenza outbreaks (including the 1957 flu), their prominence in 1918 is one of the cardinal features of the “Spanish” flu. It was this subset of patients on which research efforts were concentrated. Three influenza RNA positive cases were identified (Taubenberger et al., 1997; Reid et al., 1999).
While the serological data suggested that the 1918 hemagglutinin would resemble the H1-subtype swine flu isolated in 1930, the avian origin of the 1957 and 1968 HAs made it possible that the 1918 H1 would more closely resemble an avian H1. The complete coding sequence of the gene was generated from the South Carolina case and was confirmed using RNA from the New York and Alaska cases. Out of the 981 bases of the HA1 domain of the gene, only two nucleotide differences were noted between the cases. One of these differences (in the New York case) would change an amino acid as compared to the sequence of the South Carolina
and Alaska cases. Interestingly, that change occurs at one of the critical amino acids involved in receptor binding. The overall receptor binding pattern for the 1918 hemagglutinin is most similar to those of classic swine influenza strains, and it is possible that the New York case could specifically bind both avian- and mammalian-type receptors, a property it would share with classic swine influenza viruses (Reid et al., 1999).
The full-length sequence of the 1918 HA shows that it is most closely related to the human and swine influenza strains of the 1930s. While it is more closely related to avian strains than any subsequent mammalian H1, it is phylogenetically distinct from current avian H1s (Reid et al., 1999). It is probable that the HA involved in the pandemic did not pass directly from an avian source to its pandemic form but rather spent some unknown amount of time adapting in a mammalian host, although whether that host was human or swine is unclear. Phylogenetic analyses of the full-length 1918 HA sequence consistently place it in the mammalian clade, sometimes near the root of swine strains and sometimes near the root of human strains. Differences in placement probably reflect differences in mutation rates between swine and human strains. The most likely interpretation of these results is that the 1918 strains are most closely related to the common ancestor of all subsequent human and swine H1 strains.
An alternative hypothesis is that the 1918 virus may have acquired its HA by shift immediately before the pandemic from an avian virus. If avian influenza genes are in evolutionary stasis, as has been suggested (Murphy and Webster, 1996), such a virus would not resemble current avian strains phylogenetically. For the 1918 strain to be the result of the direct introduction of an avian hemagglutinin, antigenic drift would have to have occurred within the avian clade over the last 80 years. However, because there are no known samples of avian H1 viruses from 1918, this hypothesis cannot be tested.
The different mortality patterns among influenza pandemics indicate that influenza viruses might emerge in different ways. The pandemic of 1890 demonstrated a mortality pattern different from that of 1957 or 1968. While morbidity was high in each year from 1890 to 1892, mortality was low in 1890, rose in 1891, and peaked in 1892 (Glezen, 1996). In 1957 and 1968 mortality was highest in the first year of the pandemic. In light of the 1890 pattern it is intriguing that prior to the 1918 pandemic the mortality rate from influenza and pneumonia began to rise in 1915 and 1916. The rate dipped slightly in 1917 and then rose sharply with a classic “herald” wave in the spring of 1918; it finally skyrocketed with the most virulent form in the fall and winter of 1918 to 1919. Is it possible that a poorly adapted H1N1 was beginning to spread in 1915 that caused some serious illness but that was not yet perfectly adapted? The best argument against this scenario is that, if a strain with a new hemagglutinin was causing
enough illness to affect the national death rates from pneumonia and influenza, it should have caused a pandemic sooner, and significant numbers of people should have been immune, or at least partially immunoprotected, in 1918. Eighty years later it is impossible to distinguish fluctuations in mortality caused by drift in the previous strain (H2N2, H3N8?; Webster et al., 1992) from early waves of a newly emerging virus. Both the 1957 and the 1968 pandemics were preceded by mild waves early in the same year, and there is evidence that the 1968 pandemic virus had begun to circulate several years earlier (Monto and Maassab, 1981). The reassorted viruses of 1957 and 1968 had human-adapted internal proteins; perhaps after a surface protein shift neither virus required a long adaptation period before causing a pandemic.
A number of scenarios for the origin of the 1918 flu are possible. First, it could have been an entirely avian virus that entered the human population in 1918 already capable not only of infecting people —as was the 1997 Hong Kong “chicken flu”—but also of spreading from human to human with extreme efficiency. Second, a wholly avian virus could have entered the human population some years before 1918, gradually establishing itself and adapting toward efficient replication and transmission in humans. Third, it could have been a reassortant virus with some genes of avian origin and some of human origin, as were the pandemics of 1957 and 1968. Fourth, the pandemic may have resulted from mutations in a previously circulating human virus that made it completely unrecognizable antigenically. The extent of the pandemic and the fact that it arose in a mild wave followed by more severe waves make the fourth possibility the least likely. The supreme efficiency with which the 1918 influenza spread and the evidence that many different genes contribute to host specificity argue against an immediate avian origin. Distinguishing between a wholly avian virus that had been adapting in humans for some years and a reassortant virus where some genes were of immediate avian origin and some from the previously circulating human strain will be difficult, given the lack of any contemporary avian or pre-1918 human strains for comparison. Reassortment has been the mechanism for two well-characterized pandemics, while a gradually adapting avian virus has never been detected.
Little is known about how genetic features of influenza viruses affect virulence. Virulence of a particular influenza strain is complex, involving several features that include host adaptation, transmissibility, tissue tropism, and replication efficiency. The genetic basis for each of these features is not yet fully characterized but is most likely polygenic in nature. There are, however, several identified mutations that do radically change the behavior of a given flu strain. In the case of the 1918 virus, neither the HA cleavage site mutation (Taubenberger et al., 1997; Reid et
al., 1999) nor the △146 mutation of WSN/33 was present (Goto and Kawaoka, 1998; Taubenberger, 1998). Both of these changes have been shown to affect tissue tropism of the viral strain.
Fragmentary sequences of all the remaining gene segments of the 1918 virus have already been deciphered, and full-length segment sequences will be completed. Such sequences will allow complete phylogenetic analyses of each segment and will help elucidate the origin of the 1918 virus. Whether any particular genetic features of the virus can be related directly to its exceptional virulence is yet unclear. Even as the genetic structure of the Spanish flu virus is becoming fully known, other questions, such as the role that differences in immunity in different age groups played in the 1918 mortality, may be forever lost to study. It is hoped that knowledge gained by studying this very successful human pathogen can be applied to prevent or at least predict the emergence of new influenza viruses that have pandemic potential.
The answer to this question is unclear. Because whatever viral, human, and historical factors existed in 1918 that allowed the emergence of such a pandemic, it remains possible, given a similar set of circumstances, that a pandemic of similar magnitude could reoccur. Additionally, the rapid movements of people around the globe make containment of such a newly emerging flu virus inconceivable.
Three factors were associated with virulence of the 1918 flu: viral, historical, and human. In terms of viral factors, the strain probably reflected an HA shift to H1 before or during 1918. The 1890 pandemic is thought by archeserological analysis to have been an H2N2 strain. Similarly, it has been proposed that an H3 subtype circulated in humans in 1900 (Webster et al., 1992). There is no evidence of pantropicity by viral mechanism and none is reflected in the pathology. Sequence analysis has revealed that the 1918 strain lacked the HA cleavage site mutation and the △146 mutation in NA. Little is understood about the genetic basis of virulence in influenza. Therefore, questions such as what virulence factors are in other genes and which genes shifted to form the pandemic virus will be difficult to address unless pre-1918 viral samples are recovered.
In terms of historical factors, did enhanced virulence occur solely because of World War I? This is doubtful because morbidity and mortality rates for both civilian and military populations were similar across the globe. However, the density of troops in camps and the increased movement of people because of the war did facilitate spread of the virus. The lack of effective vaccines, antibiotics, and antiviral drugs as well as inad-
equate nursing and hospital care certainly played a role in the morbidity and mortality of the pandemic.
Finally, purely viral factors are unlikely to account for markedly different age mortality rates. Therefore, human factors, both immunological and physiological, may have played a role in virulence: What was the pre-1918 exposure (by age group) to H1N1 influenza strains and other sub-types of influenza A? What other physiological age differences exist? Possibilities include changes in ligand specificity and distribution in the respiratory tree, changes in protease levels, and age-related response to primary influenza infection.
A three-pronged strategy is proposed to meet the challenge of predicting or averting a future influenza pandemic: (1) studying the basis of virulence of the 1918 flu directly by genetic and functional analyses, (2) studying the basic biology of flu and host responses in other flu strains to identify features associated with virulence, and (3) increasing surveillance of influenza strains in humans and domestic and wild animals.
Much larger inventories of data are needed for an increased understanding of influenza ecology. Because birds are thought to be a natural reservoir for influenza viruses, the following questions need to be addressed more thoroughly: What is the subtype distribution among species and locations (both domestic and feral)? What is the drift rate of avian viruses? What is the incidence of genetic shift between/among avian strains? Longitudinal sampling studies are needed to answer these questions.
Because influenza is a zoonotic disease, massively increased surveillance of human exposure to animal-adapted flu strains would help elucidate the mechanism for the emergence of novel (potentially pandemic) strains in humans. Were the recent Hong Kong H5N1 and H9N2 outbreaks and human/swine H3N2 reassortant outbreak in United States swine unique and unusual events or were they “surveillance artifacts” resulting from limited sampling and limited reagents? What is the incidence of human exposure to avian, swine, or equine viruses (e.g., in poultry workers and swine and horse farmers)? What are the clinical and subclinical infection rates? What is the incidence of interpandemic shifts among circulating human strains (e.g., shift of internal genes)? Current testing procedures using reagents developed for known circulating flu strains are not adequate to address these questions. Again, longitudinal sampling studies are needed.
The opportunity exists to collect and analyze large amounts of data
currently unavailable for analysis to help elucidate the mechanisms by which influenza viruses can adapt and cause pandemic infections in humans. This challenge is clearly worth attempting, given the legacy of the 1918 Spanish flu.
Bean, W., M. Schell, J. Katz, Y. Kawaoka, C. Naeve, O. Gorman, and R. Webster. 1992. Evolution of the H3 influenza virus hemagglutinin from human and nonhuman hosts. Journal of Virology, 66:1129-1138.
Claas, E. C., A. D. Osterhaus, R. van Beek, J. C. De Jong, G. F. Rimmelzwaan, D. A. Senne, S. Krauss, K. F. Shortridge, and R. G. Webster. 1998. Human influenza A H5N1 virus related to a highly pathogenic avian influenza virus. Lancet, 351:472-477.
Crosby, A. W. 1989. America's Forgotten Pandemic: The Influenza of 1918. Cambridge: Cambridge University Press.
Glezen, W. 1996. Emerging infections: Pandemic influenza. Epidemiologic Reviews, 18:64-76.
Goto, H., and Y. Kawaoka. 1998. A novel mechanism for the acquisition of virulence by a human influenza A virus. Proceedings of the National Academy of Sciences USA, 95:10224-10228.
Kilbourne, E. 1977. Influenza pandemics in perspective. Journal of the American Medical Association, 237:1225-1228.
Linder, F. E., and R. D. Grove. 1947. Vital Statistics Rates in the United States: 1900-1940. Washington, D.C.: Government Printing Office, p. 254.
Monto, A. S., and H. F. Maassab. 1981. Serologic responses to nonprevalent influenza A viruses during intercyclic period. American Journal of Epidemiology, 113:236-244.
Murphy, B. R., and R. G. Webster. 1996. Orthomyxoviruses in Virology, (3rd ed.), B. N. Fields, ed. Philadelphia: Lippincott-Raven, p. 1419.
Reid, A. H., T. G. Fanning, J. V. Hultin, and J. K. Taubenberger. 1999. Origin and evolution of the 1918 “Spanish” influenza virus hemagglutinin. Proceedings of the National Academy of Sciences USA, 96:1651-1656.
Reid, A. H., and J. K. Taubenberger. 1999. The 1918 flu and other influenza pandemics: “Over there” and back again. Laboratory Investigation, 79:95-101.
Schafer, J. R., Y. Kawaoka, W. J. Bean, J. Suss, D. Senne, and R. G. Webster. 1993. Origin of the pandemic 1957 H2 influenza A virus and the persistence of its possible progenitors in the avian reservoir. Virology, 1 94:781-788.
Scholtissek, C., H. Burger, O. Kistner, and K. F. Shortridge. 1983. Genetic relatedness of hemagglutinins of the H1 subtype of influenza A viruses isolated from swine and birds. Virology, 129:521-523.
Subbarao, K., A. Klimov, J. Katz, H. Regnery, W. Lim, H. Hall, M. Perdue, D. Swayne, C. Bender, J. Huang, M. Hemphill, T. Rowe, M. Shaw, X. Xu, K. Fukuda, and N. Cox. 1998. Characterization of an avian influenza A (H5N1) virus isolated from a child with a fatal respiratory illness. Science, 279:393-396.
Taubenberger, J. K., A. H. Reid, A. E. Krafft, K. E. Bijwaard, and T. G. Fanning. 1997. Initial genetic characterization of the 1918 “Spanish” influenza virus. Science, 275:1793-1796.
Taubenberger, J. K. 1998. Cleavage of influenza virus hemagglutinin into HA1, HA2: No laughing matter. Proceedings of the National Academy of Sciences USA, 95:9713-9715.
Webster, R. G., W. J. Bean, O. T. Gorman, T. M. Chambers, and Y. Kawaoka. 1992. Evolution and ecology of influenza A viruses. Microbiological Reviews, 56:152-179.
Webster, R. G. 1997. Predictions for future human influenza pandemics. Journal of Infectious Disease, 176(Suppl. 1):S14-S19.