On the Centenary of the 1918 Flu: Remembering the Past and Planning for the Future


MARCIA DAY CHILDRESS:
Good afternoon. Welcome to today’s
Medical Center Hour. I’m Marcia Day Childress. I’m from the Center
for Biomedical Ethics and Humanities, which
is happy to bring you these weekly Medical
Center Hour programs often in partnership
with other programs here at the medical center,
or medical school and health system. Today’s program on the
centenary of the 1918 flu, “Remember the Past and
Planning for the Future,” is a combined Medical Center
Hour and medical grand rounds for the express and excellent
purpose of the Department of Medicine’s annual Hayden
Farr lecture in epidemiology and virology. This endowed lectureship
highlights achievements in epidemiology and virology
by esteemed physicians and scientists. The lectureship UVA professor
emeritus Fred Hayden and the late professor and
hospital epidemiologist Barry Farr. What a pleasure it is to
welcome Dr. Haydon here today and to remember fondly our
longtime colleague Dr. Farr. This year’s Hayden Farr
lecturer is the distinguished pathologist Jeffrey
Taubenberger. Taubenberger is chief of
the viral pathogenesis and evolution section
in the laboratory of infectious diseases and
deputy chief of said laboratory of infectious diseases at the
National Institute of Allergy and Infectious Diseases of the
National Institutes of Health. He has done landmark work
with the 1918 influenza virus and so is quite uniquely
qualified to discuss with us past, present, and
future concerns related to pandemic influenza. As you know, 2018 marks the
centennial of the Spanish flu pandemic, the world’s deadliest
event, which killed at least 50 million persons worldwide. The pandemic’s sudden
emergence and high fatality are stark reminders of
the threat that influenza has posed to human
health and human society for more than a millennium. There’s quite a lot we
still don’t understand about that pandemic, but
the recent sequencing and reconstruction of the
1918 virus work accomplished by Dr. Taubenberger
and colleagues represent a significant
breakthrough. So how does this
scientific advance help us better to know the past
and prepare for the future? This fall, as you may
know, Medical Center Hour has partnered with the
School of Nursing Bjoring Center for nursing
historical inquiry and historical collections
in the Claude Moore Health Sciences Library and now with
the Department of Medicine to observe the centennial
of the 1918 flu. We’re marking this
somber anniversary with three programs,
one examining the pandemic in the historical
context of World War I– that happened in
September– one focused on science, which is
today’s presentation, and one addressing the human
toll of the epidemic, program on the 14th of November. Information about it
is in your handout. So please see your
handout, not only for a bio sketch of
Dr. Taubenberger, but also information about
that final program two weeks from today. Also please do visit the
exhibit in the library lobby on the impact of
the 1918 flu right here in Charlottesville,
Albemarle and at UVA. And if you’ve not already
done so, get a flu shot. We’d like to thank the
Department of Medicine and the Division of
Infectious Disease for partnering with
us today and also all those involved in
the Hayden Farr lecture. We also thank historical
collections and the library sponsor of the history of
the health sciences lecture series, of which this
program is a part, and our influenza 1918
2018 joint commemorative project with the
university’s historians of nursing and medicine. And now Dr. Jeffrey Taubenberger
and on the centenary of the 1918 flu. Welcome. Thank you very much. Thank you, Marcia. It’s an honor to
be here and to give this talk on the 100th
anniversary of the 1918 flu. So I will start with
the important thing. I work for the US
government, and I can only dream of having something
financially to disclose. So unfortunately I do not. It’s an honor to give this
combined talk, in which I see that I’m supposed to
talk about epidemiology, virology, medicine, and
history, and I will do my best and show lots of slides and
talk really fast and cover all those things. I didn’t have the opportunity
to meet Dr. Farr, unfortunately, but I have known and been
friends and colleagues with Fred Hayden
for over 20 years and can conclude that 20 years
is a really long time, even for a good bottle of wine. So it’s going to start with
talking about the 1918 flu by thinking about last
year’s flu season, which was less than a year ago. I think you probably all recall
what a really bad flu year we had. There were reports
all over the nation about how bad the flu was,
including here in Virginia. And the numbers of those
impacted are just coming in and are looking to
be very serious. It could be that as many
as a million people were hospitalized in
the United States for flu in the last flu
season and upwards of 70,000 or 80,000 people may have died. Of course, influenza
comes in different forms– not only seasonal flu, but
occasional pandemics in which a new strain gets into people. And it’s here
especially on this day that we think about the
specter of pandemic flu, and the worst one
that we’ve ever seen which happened exactly
100 years ago in October 1918. So if we cast back 100
years and just picked one random example of what
could be a description of what happened almost anywhere in
the world in the fall of 1918, he at Camp Devens at
the end of World War I, the United States
was really gearing up to put men in the field
on the Western Front. And Camp Devens was one
of a couple dozen training camps in the United
States for the army. This was about 50
miles west of Boston. So here at the
beginning of September, a single soldier presents
with an influenza-like illness to the hospital. The next day there were a dozen. A week later there
were three dozen, and a week after that there
were 12,000 US soldiers in this one little
training camp hospitalized for severe influenza. At the end of this outbreak,
a third of the camp became ill and nearly 800 men died. These were extremely healthy
18 to 25-year-old men who dropped dead of influenza. Here’s some pictures
of Camp Devens and their supportive care
only for flu virus infections. They had no antivirals, no
antibiotics, no vaccines, no intensive care. So all they could do was just
generalized nursing care. A letter from one of the
physicians treating patients that he sent to another
one of his physician colleagues was not
known at the time, but in the last couple
of decades was found. And this letter talks about
how these patients developed the most vicious type
of pneumonia that’s ever been seen, and then you get
these patients rapidly develop cyanosis and then struggle
for air until they suffocate. It is horrible. We’ve been averaging
about 100 deaths per day and keeping it up. William Henry Welch– he
was probably the most famous physician in the United
States at the time– was keenly involved in military
medicine in World War I and came to tour this camp
and, after visiting the camp, was quoted to say that this must
be some new kind of infection or plague. And yet it wasn’t plague,
and it wasn’t new. It was just influenza,
a disease that had been recognized for at
least 500 years before that. Here’s some pictures
of US troops being treated for influenza in
a very primitive camp hospital on the Western Front
in October 1918. So if we talk about some of the
numbers, they’re astounding. And I think that we will
never know the full impact of the 1918 flu. The numbers keep rising
as the developing world is looked at more closely. So currently, I think a
reasonable number of deaths would be 50 million
people in about a nine or 10 month period in
1918, 1919, but 100 million is not unreasonable and
is it possible to be closer to the truth. In the US, there were little
under 700,000 people that died. So to try to put
that in perspective, in a city like
Philadelphia, 16,000 people died, 11,000 just in
the month of October. So at its peak, 4,500
people died in Philadelphia in a seven day period. The US military entered
World War I late compared to the
European countries, but of 100,000 troops that died
to all causes in World War I, 40% of them died of flu. There were about 17,000 deaths
in Virginia, which is about 1% of the population at the time. Here’s some work from
colleague Michigan State, Dr. Chandra, who’s been
doing work as a demographer, trying to understand and
model the influenza pandemics mortality in countries where
no mortality statistics were collected at the time by
looking at the divit that occurred in the growth
of the population. And from these work, he suggests
that at least 14 million people died in India of the 1918 flu. Other estimates have gone as
high as 17 million people. Here’s a temporary camp
hospital in Fort Riley, Kansas, again where they recognized that
it was a respiratory disease. They had patients wear masks
and try to separate them. But other than
supportive care, there was nothing that could be done. The pandemic blew up
and was recognizable everywhere in the world in
September through November in 1918 in a major wave. There seems to be
some flu activity that occurred earlier in the late
spring and early summer. Certainly in
northern Europe there was clear evidence of the
pandemic in June and July. But where the virus started and
where it began to be circulated is not known and
possibly will never be known because it’s under
the radar of what happened. And then there were
subsequent waves. Some places had outbreaks again
in the early months of 1919. A very sad and easy way to look
at the impact of the 1918 virus is just to take a look at
pictures of death records where each death record is
a single page bound by month from the state of Oregon sitting
on the shelf of the National Library of Medicine. And here you see the emergence
of the flu in October, November, December,
January, and then going back to a normal amount
of monthly deaths for the state. You can find pictures, not
just in military camps, but in civilian life. Here’s a picture of ill students
in Dartmouth College, people at training camps, people at
Walter Reed Army Hospital. You see Americans
being Americans. They played baseball
during the pandemic. They went to the theater. They had to have
volunteers dig mass graves. This one gravestone, which
is in central Pennsylvania, says in remembrance of the
children buried in this area who died in the flu
epidemic of 1918 and whose names are
known only to God. This is not a sight
that you would expect to see in cities
of the United States, but were relatively common
sights in the fall of 1918. Its impact on World
War I was tremendous. I think that the
combatent countries could think of no good
way to end the war, and I think the flu
ended it for them. The leaders of the United
States, Britain, and Germany all got the flu. Prime minister Lloyd
George almost died of flu. Woodrow Wilson became
extremely ill with flu, and his slow recovery prevented
him from participating fully in the peace conference. And that probably
affected the way that the peace treaty
was negotiated. General Pershing got the flu. Kaiser Wilhelm got the flu. Thinking about its
impact on society, I can’t emphasize again the
enormous impact of the 1918 virus in terms of the personal
tragedy of how many people died. Some of the most famous
artists were affected. Edvard Munch in
Norway got the flu but recovered but almost died. Egon Schiele, the expressionist
artist who was 28 at the time, died and this is actually
a picture of his deathbed. He died in the same
bed with his wife who was six months pregnant. They also all died within
a couple days of each other in that same bed. Gustav Klimt also died
of pneumonia in 1918. The impact of the
virus was so great that it caused a life expectancy
the United States to drop by about a dozen years. And this is due predominantly
to one very remarkable feature, which is still
really unexplained, which is that usually
influenza virus has its highest impact in the extremes
of life– that is, in neonates
and in the elderly, two populations that are
thought to be immunocompromised in some extent with a
little mortality in between. So you get this roughly u shape
mortality curve for typical flu before and after. But in 1918, what you see
was a new hump in the middle, peaking at about age 28. And this is the one cardinal
feature epidemiologically of the 1918 virus that’s
seen everywhere in the world, that young healthy
adults had very high and unexpected mortality. And there are many hypotheses
about them, but none of them adequately explain the data. And it still remains
a big mystery. So let’s talk very
briefly about the biology of the virus itself. The 1918 virus was
an influenza A virus. It’s an orthomyxovirus, and
it’s a single stranded RNA virus with a segmented genome. Not to talk about the biology
of the virus too much, but just to know that it’s
an envelope virus and expresses a couple of
surface proteins, which play key roles in viral lifecycle. One of them, hemagglutinin, or
HA, is involved in viral entry. It binds to the receptors
on the cells, which are glyocoprotein– the sugars
on the tips of glycoproteins on epithelial cells. And it comes in a variety
of subtypes or flavors. And the neuraminidase
has the opposite effect. It helps newly formed viruses
be released from the cell by cleaving off
those same cleavages. The virus is a single
stranded RNA virus, and its polymerase
lacks proofreading. So it has a lot of errors. So the virus lives in its
kind of error prone lifestyle where every single virion
that’s replicated has between 1 and 10 or more mutations. Many of these are deleterious,
but the virus can rapidly select positively for
mutations, for example, that help it escape from
pre-existing immunity, from antiviral drug treatment. Or if for example of a
virus from one animal finds itself in a
non-native host, there might be some mutations
that help select for adaptation to a new host. The kind of continual
change in flu that we see, especially of the
surface protein genes to escape
pre-existing immunity, has been called antigenic drift. And because it’s a
segmented genome, if one host and
specifically one cell is infected with two flu
strains at the same time, you can mix and match
the gene segments to create a completely
novel virus that can have very different
phenotypic properties. And that’s been called
antigenic shift. So you probably are
all familiar with flu, but this is a helpful algorithm,
as we start in flu season, to figure out if
you have the flu. It’s based on a lot of data. So it says, do you
feel like you’ve been hit by a train, yes or no. No, you do not have the flu. Yes, have you been
hit by a train. Then yes, you’ve
been hit by a train. And if no, then
you have the flu. You can look at how rapidly
influenza viruses change by doing sequence analysis
and phylogenetic trees, and what you can see is that
the influenza viruses that circulate in winter
outbreaks every single year form distinct clades,
which can be dated, so that is, if you look
up a sequence in GenBank, you could say, this human
H3N2 virus is from 1998 or one in 1984, so on. It’s that rapid. You can actually look at
molecular evolution in one host, and you can look
at molecular evolution during one flu season. And most of that
change is driven by changing the protein of
the hemagglutinin on the top where the antibody’s
recognized and where actually the hemagglutinin
binds to the receptors that it’s going to recognize. The result of this is
that the flu vaccine has to keep up with this
rapid evolution, and you can see the H3N2 viruses
that have been the dominant flu strains in the last 50 years. The strain has to be changed
almost every single year to keep up with this
mutation rate, H1 slower. So we clearly need
to do a better job at coming up with ways to
make vaccines that are not so sensitive to the kinds
of mutations of flu viruses to make more broadly
protective vaccines. And I’ll touch on that
briefly at the end. So just to mention again. Last year’s flu season
was particularly bad. The numbers are not in
yet, but it’s possible that 80,000 people died
of influenza pneumonia in the US last season and
almost 200 children were documented to have died. Influenza in our world
is a human disease and is obviously a
hugely important medical and public health problem. But influenza
infections of humans is really kind of an accident. It’s really a wild bird virus. It’s a gastrointestinal virus
in over 100 species of birds, but commonly in waterfowl, like
ducks and geese, shorebirds, terns, seagulls, et cetera. But many other species
of birds have flu. So it’s a GI virus. It’s spread by fecal
spread in the water. But influenza viruses
have this ability to adapt to many other
warm blooded animal species, including
other domestic poultry, like chickens and turkeys,
but also mammals, including wild mammals as divergent as
whales and seals and bats, but important agricultural
and economically important animals like horses, dogs,
pigs, and of course to humans. And the complicated ecobiology
of how influenza viruses move and adapt between
these species is still only poorly
understood, but means that eradication of
the virus from humans is never going to be possible,
like was possible for smallpox and could be possible
for polio and measles. Flu is the poster child
for Darwinian evolution and natural selection
with its rapid ability to respond to negative
and positive selection. Here is a phylogenetic tree
of the different subtypes of hemagglutinin, looking
like a sketch that came out of Darwin’s notebook,
looking at the evolution of these different subtypes
that occurred in birds. So if you were to read
a textbook of influenza as to how pandemics form,
you would see something like this paradigm–
that birds like this duck are the natural host for flu,
and that bird viruses adapt to pigs to become swine
flu, and that swine viruses become human pandemics. So outside of Washington D.C.,
there’s very little evidence for this actually
occurring, and it’s just a much more
complicated problem. It turns out that
humans give viruses to pigs much more commonly
than swine give viruses to us. In looking at this picture,
I don’t know which of the two is enjoying their kiss
more, but they both seem to be enjoying it and
sharing viruses probably in both directions. So if we go back and look
at pandemics in history, we know of what’s happened
in the last 100 years. So we’re talking about
the 1918 flu, which was an H1N1 subtype– so
that hemagglutinin subtype 1 and neuraminidase subtype 1. In 1957, there was a
pandemic that was H2N2– in 1968, the Hong
Kong flu, which is H3N2, which is
still circulating. In 1977, the old
lineage of H1 came back. And in 2009, there was
a new H1N1 that emerged, and now this new H1N1
and this H3 virus continue to circulate as
the influenza A viruses. We do not know what
happened before 1918. If you look at mortality as
just one measure of a pandemic, you find that
pandemics are defined as having a novel virus,
presumably an animal derived influenza virus
completely or in parts, that has emerged in humans. But just because
it’s a novel virus does not mean that
it’s equally pathogenic or has equal public
health impact. The 1918 flu had a huge
impact, and the 2009 virus was a particularly
mild pandemic. And yet it was a novel
virus for humans. If you look at
pandemics in history, certainly they have gone back
at least since about 1510. There have been 14 pandemics
in the last 500 years, but there is some evidence of
pandemics going back earlier than that, although
it’s very hard to look at the medical literature
from the Middle Ages. So there were no virus
isolates made in 1918. The concept of viruses
was still rather new. While flu was
recognized clinically, its causative agent was not. Spurred by the 1918
pandemic, researchers isolated the first influenza
A viruses from pigs in 1930 and from humans in 1933. But by that point,
there was no way to actually study the
1918 virus itself. So about 25 years ago at
the Armed Forces Institute of Pathology, I
had this crazy idea that we could perhaps
use PCR based approaches to find gene segments of the
virus that caused the 1918 flu in autopsy tissues of people
who died in the pandemic that were preserved in the
national tissue repository. And this project worked
unfortunately or fortunately. Here are autopsy tissue
sections cut and stained in 1918 that had been sitting on the
shelf at that point for 80 something years and are
still in good shape. And the tissues are
a little bit of lung like this fixed in formaldehyde
and embedded in candle wax about the size of
your fingernail. So using that, we
had a small group of people that took nine
or 10 years to sequence the genome of the
1918 virus, using what were for us pushing the
technology as far as it could go back in the ’90s. Then we expanded our
collection of tissues going from other
locations, including from a person who died of
influenza whose body was buried in the permafrost in northern
Alaska, exhumed by my colleague you Johan Hultin in
Brevig Mission, Alaska. He did an exhumation in
1951 from the same site with the attempt to try to
culture the 1918 flu virus, and that failed. There was no lab containment. They didn’t even have biological
safety cabinets in 1951. So I’m not sure what they
would have done with the virus had they recovered
it, but they did not. They looked at a place
where, in a local outbreak in a small Inuit village 85%
percent of the adult population died in five days and
were buried in permafrost, leaving dozens and
dozens of orphans with the same mortality
pattern that young adults died and children did not. And it was extremely devastating
to these communities. This is right on the tip of
the Seward peninsula of Alaska where Brevig Mission is. This is flying in about 10
years ago with the frozen Bering Sea here. And with apologies
to Tina Fey, this is one of the very few places
in Alaska you can see Russia from your house. This is where the
bodies are buried on a bluff right on the
beach in between these two wooden crosses. So if you look online, you
see all sorts of weird things about how the 1918
virus was sequenced and here this thing says that
we use these complex computer program and supercomputers to
perfectly match the structure. So I just wanted to show
you our supercomputer, which was that there were
between femtogram and attogram amounts of viral
RNA in the tissue. Most of it were
single nucleotides. There was a tiny fraction of
stuff up to about 80 bases or so. You had to retroactively
label the heck out of it to see anything, even
with 40 cycles of PCR, and then do Sanger sequencing. So it took a couple of weeks to
get 20 to 40 bases of sequence of the virus using this
approach and so a 10 year effort to put the virus together. So here’s the supercomputer
where we recorded the sequence of the 1918 flu. And then rebuilding the virus
after the sequences were done was a multicentered
collaboration funded by the NIH that involved a bunch of
different collaborators in a very exciting
program project. And the 1918 virus
could be resurrected to be studied in
high containment labs to model pathogenesis. Recently, we’ve
been able to develop high throughput
sequencing approaches that are much more efficient. And so now in about a week or
two I can sequence in my lab the complete genome of the
1918 virus instead of a 10 year effort, which is good, because
I’m old and impatient now. And we can do this
more efficiently. So that means that we can look
backward in time from 1918. Let me briefly show
you some pathology from people who died in 1918. This is the histology
of normal lung, which is mostly just blank airspace. And here what you
see in 1918 deaths are features of the
primary viral pneumonia with a diffuse alveolar damage
with fulminant pulmonary edema and/or hemorrhage,
sometimes alveolitis with a necrosis of
alveolar cells– again, DAD, abundant thrombi. But the predominant pathology
that you see was of secondary, untreated bacterial pneumonias,
massive bronchi pneumonias with “C”s of neutrophil destruction
of the bronchial tree but with evidence of repair going on
probably from the primary viral cause and then death from the
secondary bacterial pneumonia. You can see abundant bacteria
by tissue gram stain, the nasopharyngeal common
causes of bacterial pneumonia– pneumococcus, group
A strep, Staph. But they were gram
negatives, and a variety of bacteria that cause
secondary pneumonias in 1918. The influenza virus is present
throughout the bronchial tree from nose to alveolus. It replicates in the
tips of the cells lining the respiratory epithelium, not
the full thickness epithelium, which is of interest. And we have cases going back
to as early as May 11th, 1918, the earliest sequence of a
soldier that died of the 1918 virus and a partial sequence. And those sequences are
identical with fall wave cases. So the most important
lesson initially from sequencing the 1918 flu
is that the 1918 virus is truly the mother of all
pandemics in the sense that every single
human influenza infection, whether every year
a seasonal flu for the last 100 years or the
subsequent pandemics we’ve had in 1957,
’68, and 2009, are all genetically descended
from this one founder virus. So if 50 or 100 million people
died of influenza in 1918, tens of millions of people
have died of influenza in the last 100
years, all of them due by the successful
introduction of a single virus into the human population. So we’re 100 years
into a pandemic era that has no signs of stopping. And so just to make this
point– if you take the last 50 years of mortality, and
including the pandemics that have occurred in
the last three pandemics, 75% of the mortality has
been to seasonal flu. So while there’s a huge concern
to prepare for and prevent pandemic flu, it’s
the annual flu that is causing most morbidity
and mortality in the United States and the rest of
the world and something that we should think
about very seriously. So now let me very
briefly talk about what we’ve learned about why the
1918 virus had such an impact, and I’m going to talk about
a variety of factors that come together to provide
more serious illness in 1918 than in other influenza
virus infections. So let’s start with the virus– you can work with the virus. It’s a select agent
once you have approval under very high
containment conditions. And the 1918 virus
is very pathogenic. Unlike most human viruses,
without adaptation in mice, in ferrets,
in non-human primates, the virus basically causes
severe disease and death very rapidly. And it’s very
different than what happens with normal seasonal
viruses or other avian viruses. To make a long story short,
of all the genes encoded by the virus, the hemagglutinin
surface protein gene is probably the major
virulence factor of the virus that just putting this
gene on the backbone of a non-pathogenic
modern human virus is enough to kill
mice and ferrets. And this is not shared
with the HA genes of other pandemic viruses. And so again, to make a lot
of data very compressed, we think that the 1918 virus
is a very avian-like virus and that there was likely an
avian ancestor to the pandemic shortly before the
pandemic emerged and that the 1918
hemagglutinin is a very avians-like hemagglutinin
with just small adaptations to receptor binding. And we know that
interestingly now and scarily that the H1s of
wild bird, duck hemagglutinins that you can find in a pond
just outside of Charlottesville, if put in the context of a
mammalian replicating virus, share the same pathogenic
features of the 1918 virus so that the mutations
that are not pandemic-specific but
in a sense are just inherited from a bird H1. And since we know
that H1 viruses share this pathogenicity, we wondered
about the other subtypes that circulated in birds. So we made a series
of viruses that were identical,
but just different in their hemagglutinin. All of these are
hemagglutinins right out of wild birds with no
adaptation, and all of them replicated in mouse lung. But most of them did not
cause any appreciable disease, no weight loss, but some of them
were very pathogenic like 1918 with death in about a week. And it did not correlate
with their ability to replicate in the lung. Some that were very
pathogenic had lower titers, and some that were
not pathogenic at all had higher titers. But the subtypes
that were pathogenic were H1, 6, 7, 10,
and 15, and they share a number of features in
that they induce a very marked proinflammatory response. But they’re very divergent
on the family tree, and so it’s not
clear that they share structural features that account
for this high pathogenecity. And that’s something that
we’re still investigating. It’s not just an
animal artifact. If you take normal human
bronchial epithelial airway cells, you can
culture them in vitro, and you see that
the viruses that are pathogenic in
mice and ferrets are cytopathogenic in human
cells, like H1 and H7. This is scary to me and others
because some of these viruses have caused outbreaks in humans
and, currently in the last five years in China, there has been
an H7 outbreak of an avian virus that has currently
infected about 1,600 people and caused about 800 deaths. So let’s talk briefly about
host inflammatory factors. One thing that
characterizes the 1918 and these other pathogenic
bird viral infections is this very profound
proinflammatory response that you see with an
abundance of CD45 cells coming into the lung early
on in infection. But especially crucial
here is that 1918 viruses induce this huge neutrophilic
infiltrate, which is very unusual for
a viral infection, that they induce a lot of
proinflammatory responses and cell death. So taking advantage of
this, John Cash in the lab took a series of experiments
in which he infected mice with a lethal dose
of the 1918 virus, waited for them to
be ill by day 3, and then treated them with
a drug that is a super oxide dismutase catalase mimetic
that reduces the production of free radical oxygens. It has no antiviral
properties, and yet mice that were injected
with this drug for a week during the infection
could actually clear virus and recover and have
minimal damage in their lungs and little evidence of
free radical oxygen damage, whereas untreated mice had this
very profound death, suggesting that, while the virus itself
is a very virulent virus, it’s not just the virus
infection itself that’s lethal but the inflammatory
response is contributing to an immunopathology. We’ll talk briefly
about bacterial factors. As I said, secondary
bacterial pneumonias were a crucial feature in 1918. And as you’ll recall, along
the respiratory epithelium down through the bronchial
tree into the lungs, the influence of
virus replicates in the superficial cells but
not the basal cells, which also turn out to be the
local respiratory epithelial stem cells and rapidly
reproliferate and recover and repair after a
typical flu infection. But in the case of damage caused
by a very pathogenic virus like 1918 and then secondary
bacterial pneumonia, you get a complete loss of
these basal epithelial cells, and you actually
lose the ability to reproliferate and repair. And I think this
lack of repair is one of the reasons
for severe disease and the progression of 1918
copathogenic pneumonias. Another interesting
and still evolving story is that the inflammatory
response generated by the 1918 virus with high
levels of neutrophils actually ended up changing the
behavior and gene expression patterns of secondary bacteria
like pneumococcus strains here and actually make
them more virulent. And in the case of the
1918 virus, what we think is that this co activation of
neutrophils by the 1918 virus and then subsequently by
secondary bacterial infections induces a widespread vascular
tree thrombotic picture. And if you back
to 1918 autopsies, you see very frequent
small venule thrombo. And here you see a massive
factor three deposition in 1918 autopsies
that you do not see in autopsies of people
dying of the 2009 pandemic, even those with secondary
strep infections. So this seems to be a
1918 unique thing that is inducing this thrombosis,
and this may have certainly have contributed again to the extreme
pathogenecity of the 1918 virus. Just as a quick aside,
one of the cases showing a widespread thrombi
had sickled red cells, and we ended up just doing
PCR across the globin gene. And this was an
African-American soldier who died in October 1918 who
had the Glu6Val sickle cell mutation and so, in a sense,
is the world’s oldest diagnosed case of sickle cell anemia
four years before the term was described in 1922. So studying influenza in animals
is clearly very important, and it allows us to understand
basic biology and pathogenesis of viruses like 1918 and
certainly are crucial, but ultimately as a
physician, our goal is to understand
influenza in humans and to control
influenza in humans. And so we have been studying
naturally infected patients with influenza in
the last dozen years or so at the NIH
hospital, concentrating on people who are
in high risk groups, people who are
immunocompromised, pregnant women, et cetera. But in the last
five or so years, we have started doing
studies that Dr. Hayden has done for many years in the
past and other groups had successfully done. But then people
stopped doing them, which were volunteer challenge
studies, in which healthy, very carefully screened volunteers
are brought into the hospital and intentionally infected with
circulating wild type influenza viruses to study
basic pathogenesis and immunocorrelates but using
this as a basis for phase two studies that are very efficient
in terms of ability to look at efficacy of novel
drugs or therapeutics and vaccines in small
numbers of patients. So this has been
something that’s been done very
successfully and safely with no adverse consequences
for the last five or six years. Here’s my colleague
Matt Memoli, who runs all of our clinical studies. He’s an infectious disease
physician in my group, inoculating a patient
with influenza using a little atomizer spray here. This is done in a
high containment suite at the NIH Clinical Center
hospital, the same suite where Ebola patients were treated. And we are actively recruiting
for various studies all of the time, and get
a number of recruits to help us with these studies. And that’s extremely crucial. We’ve done over 400 patients
so far with H1 and H3 viruses. We have additional
H1 and 3 viruses and influenza B
viruses in production, and we’ve done a number
of phase two studies with novel vaccines and
novel therapeutics that are helping us try to understand
how to better prevent and treat influenza in humans. So if one looks at
antibodies in the blood, in the serum as correlates of
protection for flu, something that’s been studied for
decades and decades, most of the decision
making that occurs in terms of looking at people’s
protection against influenza as well as vaccine
efficacy has been based on antibodies against
the head of the hemagglutinin protein. So the antibodies that
inhibit hemagglutinin from binding to its receptor,
in this case a surrogate assay binding to red blood
cells and aglutinating them in culture. And so this assay
hemoglutination inhibition test to HAI titers are
the marker that has been used for vaccine
efficacy and their idea that a dilution of this
antibody to 1 to 40 or so is a protective titer. So what we’ve done in our
challenge studies is– that it’s difficult to do in
a natural infected system– is that we know when time
zero is for the infection. We know what their
preexisting titers are. We know the sequence
of the virus. We can follow the development
of their infection, their inflammatory response
and their subsequent immune response during the
course of infection. And we also have been looking
at other antibody titers as correlates. And interestingly the
hemagglutinin inhibition titer is clearly a
correlative protection, but it is not the best
correlative protection that we found. It turns out that the other
surface protein, neuraminidase, is actually a better and
an independent correlative protection against
prediction of shedding duration, symptom duration,
the number of symptoms, and symptom severity. More recently, there’s
been a lot of emphasis in looking at antibodies against
the stock of the hemagglutinin, that I’ll talk about
briefly, the part of the hemagglutinin
that drifts less, that is under less mutation. And so the idea is that,
if you can make antibodies against this conserved
part, you might be able to make a more
universal vaccine. And stock titers are also
correlates of protection, but again not as good
an independent predictor of prevention of illness than
neuraminidase antibodies. Here are some data
from stock antibodies. You find that, rather than what
some people have suggested, that stock antibodies
are actually rare, that we find that every
patient that we’ve looked at has some detectable
stock antibodies in the serum to the 2009 virus,
of course, over a wide range. But if you look, for example, at
those who develop flu infection and develop illness after
infection or inoculation in the challenge system
versus those that do not, there is a difference in
the mean tire in the stock antibody. But importantly that, if
you look at the highest quartile of stock
antibody, you find that about half the patients
at this highest level of stock antibody were in the group that
was protected from developing infection and disease, and
yet that the people with stock anybody titers at the same
level still developed infection and disease, suggesting that
it’s not a perfect correlative protection and that,
while it certainly plays a role in
protection, it might not be the magic bullet that some
people would like it to be. If you look at influenza
shedding and symptom development, you find that
influenza viruses replicate often to a titer of as
high as 10 to the 3 or 10 to the 4 in nasal wash fluid
before the actual development of clinical symptoms,
which is why influenza is so hard to control,
because you can be shedding virus and transmitting
virus to others before you really know that
you’re ill or very ill. So in this little window of
the first couple of days, you have the
possibility of trying to make some
predictive biomarker assessments of who
will develop more severe disease than others. And this is really important
because at the moment, if someone presents with
an influenza like illness in the hospital and even if
a rapid diagnosis is made of influenza, it’s not clear
if this person is going to have a self limited course,
is this person going to develop a severe infection, is going
to need to be hospitalized and given supportive care. And so if there were
a way to assess this, this would be very useful. So using our challenge
system, since we know what time zero is
and we have a nasal wash and peripheral blood
serum, peripheral blood white cell sampling
everyday during infection, we can begin to examine this. And by day two, in studies of
looking at over 100 patients, you can begin to look at gene
expression in peripheral blood mononuclear cells on
day one or day two, that would predict duration
of shedding so you can make an annotated gene set that’s
very predictive of who will shed virus the longest. And you can do the same
thing for symptom severity and how severe the
illness is going to be. And so if these markers
can be further validated in additional
studies that might be possible to provide some
prognostic information to help guide whether
patients might need more intensive therapy
earlier on to try to prevent the development
of severe illness, pneumonia, and death. So in the last
couple of minutes, I will just go
back to the problem that the current vaccine
is based on an idea that we have to perfectly
match the virus that is in the vaccine antigen to
the virus that’s circulating. Since flu viruses continually
evolve and mutate in ways that are unpredictable and that
it takes at least nine or 10 months from selection to
making vaccine and filling vials and distributing
it for vaccination, that we’re always
behind the eight ball, that we’re always trying to
catch up with flu evolution. And of course, as you
know, sometimes the vaccine is a good match and
sometimes it’s a poor match. And last year was an example
of a less than optimal match, and we had a very bad outbreak. Of course, we have
no ability to predict when a pandemic will occur
or what subtype it would be. And so at the moment,
there’s really no way to make a pre-pandemic vaccine. We can only make a
vaccine against a virus after a pandemic
starts, in which case it’s really too late. In this day and age of
interconnected travel, it didn’t take the
SARS virus very long to be on three
continents and the same would happen with a
transmissible novel flu virus. So there has been a huge push
by my institute and others to try to make so-called
universal vaccines, and the word universal
vaccine could mean lots of different things
to lots of different people. This could be perhaps
a vaccine that would give you a better
breadth of protection from seasonal viruses
so that maybe you don’t need to be
vaccinated every year. Maybe you only have to
be vaccinated every five years or every 10 years. A broader one would be a
vaccine that could actually be a pre-pandemic vaccine that,
no matter what bird or horse or swine flu could get
into people of any subtype, that you would have immunity. This is a pretty
high bar given how diverse flu is and how
much it mutates Personally, I don’t think it’s possible to
make a sterilizing vaccine that would prevent infection from
any and all subtypes of flu, but I do think it might be
possible to make vaccines that would at least help
mitigate disease, reduce transmission, and
reduce severe illness. And if you could actually
prevent severe illness and death, that would be a
huge public health improvement. So a number of groups have
very interesting ideas out there, many of them based
on stock antibody epitopes. But we’ve taken a more
generian, general approach, and we have been making
a series of vaccines that are non-infectious,
that present a mixture of avian influenza
virus hemagglutinins. These are the donor source
for all human pandemic and seasonal viruses. And so as a proof of
concept set of experiments that I’ll show you, we
make an inactivated– I mean, a non-infectious
vaccine cocktail that expresses avian
H1, 3, 5, and 7 in a variety of platforms. And we’ve been able
to show in animals that it provides extremely
broad protection against most or all influenza subtypes. So as an example,
the avian virus could protect mice
against lethal challenge against the 1918 virus. So this is within the same
subtype that’s in the vaccine, but not a matched vaccine
to the 1918 virus. And yet you get 100% protection. But more importantly
and interestingly and intriguingly, we get
protection against subtypes that are not in the vaccine. So the vaccine only contains
H1, 3, 5, 7 hemagglutinins. And yet any other subtype that
we can use to infect animals provides protection– for
example, the 1957 H2 pandemic, or an avian H10
virus, for example. So these are very
encouraging results. So in summary, in mice, what we
have is that, between 10 and 50 times lethal dose challenges
with any subtype that causes disease in mice,
we get 100% protection. You make antibodies against
the vaccine antigens, but that does not explain
the heterosubtipic protection because you do not
make antibodies against the head of novel
HAs you haven’t seen. So there is a huge T
cell component to this as well, which
we’re investigating. And you can see that you
get big rises in memory CD8 and effector CD4
cells in the lungs. You get tetramer staining. You get a lot of flu
specific CD8 cells to various T-cell epitopes
on the hemagglutinin in the head and the stock. One of the features
of that induces severe pneumonias like
in 1918 is this influx of neutrophils in the part
of the acute viral infection. And you completely
abrogate and eliminate the influx of neutrophils in
the lungs of mice and ferrets that receive vaccines, and
here in ferret experiments you eliminate the development
of primary viral pneumonia. And you’re just left
with a very mild sort of focal bronchialitis. You reduce lung titers in
ferrets by four to five logs. So these data are looking good,
and we are a GMP manufacturer of some candidates
now, and we hope to be doing phase 1 studies in
humans next year to be followed with small phase 2 studies
in our challenge model in the future. So in summary, influenza
is an incredibly complicated and protean problem. It has been with us
for hundreds of years, and it likely will continue
to be a huge problem. There is ultimately
no way to eliminate influenza A viruses
from human circulation unless we were to eliminate
all warm blooded animals from the planet, which
I think would be bad. And so we have to
cope with the fact that flu is here and continues
to evolve and outwit us, how a virus with the eight
genes and 13,000 bases does a lot better
than human beings. And we still actually
don’t understand very simple questions of
how flu viruses move around between species,
adapt between species, and vary in their pathogenicity. 1918 is an example
of a virus that is helping us understand
some of those questions, but is simply one virus out of
thousands and tens of thousands and millions of flu viruses
that have circulated and will circulate in the future. I think that it’s crucially
important to study influenza in humans. We need to do a better
job of understanding how humans are protected
from influenza, and what we see is the
incredible varied response– I didn’t have time
to talk about. But that some people who are
challenged develop big rises their antibody titers after
they clear their virus, and some people develop
protection against the virus and their antibody
titers never go up. So they’re doing other things. Are they only doing
a T cell response? Are they developing
stock antibodies but not head antibodies? And we’re beginning to
tease some of this out, and it’s possible
that there might not be a one size fits all idea for
new generations of vaccines. It might be that, if
you could work out the different kinds of
responses that you have, there might be in a
sense multiple vaccine choices available in kind
of a personalized medicine approach in a very Star
Treky kind of future look. So with that, I will stop and
try to take some questions, but I want to acknowledge all
the people in the group who’ve contributed to all the
studies I’ve described, both in the group
and our collaborators and then particularly our
funders, of course the NIH and the NIAID funding us. But we’ve also received
extramural funding from DARPA, from BARDA, and
from the Bill and Melinda Gates Foundation. So thank you for your
attention, and I’d be happy to take any questions. MARCIA DAY CHILDRESS:
Wow, thank you very much. JEFFREY TAUBENBERGER:
You’re welcome. MARCIA DAY CHILDRESS:
You’ve taken us through much territory, and
we have quite a nice amount of time in which to
talk with members of the audience about what
you think of this presentation and what questions you have. And Dr. Anthony
Peters and I will have mics available to bring
to you to ask your question or offer your comment. Please do identify yourself
when you ask your question or offer your comment. So the floor is now yours. JEFFREY TAUBENBERGER:
Question, right? AUDIENCE: Jack Waltney
with a little bit of laryngitis, if
you’ll excuse me, but not a rhinovirus infection. You said what we think
of the presentation. I thought it was
wonderful, number one. When you challenge with the
volunteers with the spray, is that a small
particle aerosol? JEFFREY TAUBENBERGER: It
is a large particle aerosol specifically designed
so that we do not get lower respiratory
tract involvement. So it’s greater than
10 micron spray. AUDIENCE: What about a challenge
with drops in the nose? Is the infection rate
in the illness similar? JEFFREY TAUBENBERGER: Most
of previous challenges have been used
with drops, and we thought that perhaps a spray
like this with a large particle aerosol would give
better distribution, that there would be less chance
of the inoculum running out of the nostril. We have not done a
comparison of the technology. In our H1N1 series,
if you have people who have low HAI titers before
challenge, we see 75% or 80% of them shed virus
and develop symptoms. So it’s a very efficient system. In our recent H3 viruses,
we’re seeing a little less than that– about 60%. But still this system seems
to be a pretty good one. Of course we want to
prevent serious illness. Obviously, we do
not want to induce a pneumonia in our volunteers. AUDIENCE: There’s some
old rhinovirus studies that suggest drops give
you a better infection rate that of course– JEFFREY TAUBENBERGER:
That’s interesting. AUDIENCE: –aerosol. But did it appear that
the deaths in the 1918 were mainly due to a
secondary bacterial pneumonia? Is that what you said? Because you hear of these
patients getting sick and dying so rapidly it didn’t
seem like there was time for them to get a
secondary bacterial pneumonia that would be severe
enough to kill them. And I was an intern
when they had the H2N2, and we did see patients
that would come in and died very rapidly. And so I wondered what
you thought about that. JEFFREY TAUBENBERGER: No,
those are great questions. I think that
certainly that there were people who had
very rapid courses, and there must have been
some people who had, in a sense, fatal
primary viral infections. But if you look at
the aggregate data, I think it’s overwhelming that
the vast majority of people died with secondary
bacterial pneumonias. If you look at tens of thousands
of published autopsy studies, including careful post-mortem
microbiology, 95% of them had culture positive secondary
bacterial pneumonias at death. The average course to death
in 1918 from onset of illness to death was 11 days, exactly
like the 2009 pandemic. So certainly there
were a tiny number of people who probably
had very rapid deaths or cardiovascular
deaths or other things. I think the clinical course
of severe illness and death was one of a viral
infection, partial recovery, the secondary bacterial
pneumonia, and then death 10 or 11 days later. AUDIENCE: Your colleague– Bob Chadwick used to talk
a lot about nasal antibody. Is that still part
of the picture? Do you think there’s
anything to that? JEFFREY TAUBENBERGER:
That’s a great– I could have paid you
to ask that question. That’s fantastic. As I said, I think that serum
antibody correlates are clearly some marker of infection. But flu doesn’t have viremic
phase, as of course you know. It’s not a systemic infection. And you’re looking at
antibody at the wrong place. It’s a localized,
mucosal infection. And I think the key
to understanding why flu is such a
problem maybe be why other mucosal only
infections are hard to develop vaccines against– GI viruses, or other
respiratory virus, be it RSV or peri flu or flu– is that there’s something
about the memory response in the
mucosal system that’s different from a
virus like measles that has a systemic phase. So I think we think
that in our group that looking at what’s
happening in the mucosal level is going to be really important. In all of our new
challenge studies, we’re going to be doing mucosal
antibody, mucosal cytokine, mucosal brushings, looking
at cellular responses, as part of that. And we hope that this will help
us guide vaccine development. AUDIENCE: Jeff, thanks. That was a tour de force. Much appreciated. Fred Hayden– two
pathogenesis questions, if I may, in your
autopsy work from 1918, did you look for extra-pulmonary
dissemination of virus? I know you’ve done some
publication on whether there was virus in the central
nervous system as an explanation for some of those
syndromes, but I didn’t know more
broadly if you’d look– and I have a second after that. JEFFREY TAUBENBERGER:
We have not found any evidence of
extrapulmonary replication of the virus by PCR. And I think the pathology–
the really good autopsies that were done at the
time would really support that it was a
respiratory only thing. Of course, if you have a
really bad fatal pneumonia and you’re getting
hypoxic and cyanotic, you have secondary
hypoxic changes so that you would typically see
in the kidneys and the brain and others due to
hypoxic damage, but not a primary viral thing. We do not think the
virus was systemic. AUDIENCE: That’s helpful. There’ve been no reports,
even with seasonal influenza B with myocarditis
and things like that, leading to fairly
rapid mortality in some unfortunate individuals. And then getting
back to this issue of the importance of
the host responses, these immunopathologic
responses, and you showed data mostly on
the reactive oxygen species. And have you examined
any interventions that might influence the influx
of polys or their activation? Have you had a chance to examine
that in the animal models or in the challenge model? I mean, really what
should we be testing? JEFFREY TAUBENBERGER: Yeah,
I think those are really excellent things, and the
neutrophilic influx is really characteristic and cardenal
feature of the 1918 pathogenicity that seems to
be shared with, say, H7 virus infections, not with
H5 as an example. So it could be that that, for
certain viruses with those subtypes that we model,
that pharmacotherapy that could, in a sense,
influence neutrophils influx into the lung or activation
might be important. But there’s a knife’s edge
because, if those people with severe flu
infections develop secondary bacterial
pneumonias, it’s hard to know how you would do
that in a clinical setting. So mostly we were trying to show
that the inflammatory response itself was contributing
to the pathology, but there is the possibility
that you could intervene in a very careful way in a
very severe viral infection in a way that could
modulate immune responses. But we have not
further investigated. AUDIENCE: I’m Costi Sifri,
infectious diseases– so I also want to thank you for providing
just a fantastic lecture. I just was really
curious specifically about your observation
about neuraminidase and its marker as a correlative
immunity in comparison to hemagglutinin and a
two part question that may be settled science. But is there any
implication in terms of that finding with recombinant
hemagglutinin vaccination, which is now on the market? And also– and this may
be settled science– but is neuraminidase
an alternative target for vaccine development? JEFFREY TAUBENBERGER: I think
that neuraminidase should be an important target
for vaccine development, not necessarily by itself,
although you can imagine, for example, that you could
supplement neuraminidase in the vaccine. So a couple of points– the current
vaccines, inactivated vaccines are inactivated
virus vaccines but that then tend to– the
hemagglutinin and neuraminidase are split of the
virus and purified. The process is such
that, while neuraminidase is present in the vaccine, it’s
less likely to be immunogenic than the HA, probably
because it’s just less– the tetramer of
neuraminidase is less stable. So often, even though
neuraminidase is in the vaccine and it’s not
quantitated like HA, you variably get
an immune response. Some vaccines do,
some perhaps don’t. I think that making
NA more immunogenic and quantitating NA would
be a big improvement. So one possibility would be to
have a supplement NA vaccine. We’ve done vaccine
studies where we just make NA expressed on a
viral like particle– so no hemagglutinin. And we can vaccinate
animals and provide really excellent protection
against nasty viruses like H5. So clearly neuraminidase
immunity is really important. These data are very old. I mean, neuraminidase data
go back to the ’50s and ’60s that it’s a really important
correlative protection. But for whatever
reason, the sort of whole government vaccine
industrial complex for flu has just focused on
hemagglutinin titers in the absence of
really good data. So I would strongly encourage
that new vaccine candidates focus on adding at least an
immonogenic neuraminidase. AUDIENCE: Erick
Hewlett over here. You showed that virus shedding
occurs before symptoms begin, but with the course of illness
as a relatively brief period of time, how do you
envision transmission occurring long
distances to India and places like that in
1918 when they didn’t have travel the same way that we do? JEFFREY TAUBENBERGER:
Flu viruses are not nearly as transmissible
as many other viruses like measles for example,
but they’re clearly transmissible enough. So the R0 for flu is
somewhere around 1.51, 1.7. So one person can give it
to 1 and 1/2 people or so. But that’s clearly enough–
and there’s probably contact transmission. There’s large particle and small
particle aerosol transmission. I think one of the reasons
that flu viruses transmit more in the winter months is that
people are more likely to be closer together and indoors. Kids are back in
school, these features. But the spread of virus in
1918 or even before that was the same way. It was just person to person. And in 1918, the world was
still relatively interconnected that the exact same week that
the flu peaked in Philadelphia it was peaking in Auckland, New
Zealand and Town, South Africa. I think the virus had been
seeded everywhere in the world during the wrong
time of the year. It was in the spring
and the summer when it was the wrong
time of the year for flu to circulate because
temperature, humidity, and other things and that, when
the appropriate time came up, it just exploded. So it really was everywhere. If you go back to the first
pandemics of the 1500s that were seen and recorded
in North America, they follow in less than a
year from a pandemic in Europe and being brought by the
tiny number of people coming on the tiny number
of ships coming to the New World in the 1500s. So clearly flu
spreads really well. There’s a paper in the 1700s
of a British physician who talked about how outbreaks
can occur in towns in England faster than you can
get there by coach. So obviously that can’t be
true, but with this idea that they’re seeded everywhere
and then, when the right time comes, they kind of explode. That’s I think the
best explanation. MARCIA DAY CHILDRESS:
So we’re out of time, but I’d like to thank
Jeffrey Taubenberger for an amazing hour. Thank you. JEFFREY TAUBENBERGER: Thank you. [APPLAUSE]

1 thought on “On the Centenary of the 1918 Flu: Remembering the Past and Planning for the Future

  1. Influenza A of 1918 creates such injuries in the lungs that gives a place for secondary pathogens a banquet to thrive. The virus IS the cause of death 💀.

Leave a Reply

Your email address will not be published. Required fields are marked *