How lethal is ebolavirus?

Ebola seropositivity GabonAfter we discussed newly discovered entry factors for ebolavirus and hepatitis C virus on TWiV 166, the New York Times covered part of the story in Key protein may give Ebola virus its opening. Given my recent interest in the case fatality ratio of avian influenza H5N1, I was intrigued by the following introductory statement:

Of the pathogens that keep worried scientists awake at night, few rival Ebola for ruthless efficiency. The virus contains just seven genes, yet it manages to kill up to 90 percent of the people it infects.

Is it true that the fatality rate of ebolavirus is ‘up to 90 percent’? According to the WHO page on Ebola haemorrhagic fever,

Zaïre, Sudan and Bundibugyo species have been associated with large Ebola haemorrhagic fever (EHF) outbreaks in Africa with high case fatality ratio (25–90%) while Côte d’Ivoire and Reston have not. Reston species can infect humans but no serious illness or death in humans have been reported to date.

There have been roughly 1850 recorded cases with over 1200 deaths since ebolavirus was discovered, an average fatality rate of 65%. But have there been only 1850 human infections?

The answer is clearly no. The results of several serological surveys have shown that many individuals have antibodies against Zaire ebolavirus – purportedly the most lethal. The results of one study revealed antibodies in 10% of individuals in non epidemic regions of Africa. A similar seroprevalence rate (9.5%) was reported in villages near Kikwit, DRC where an outbreak occurred in 1995. In addition, a 13.2% seroprevalence was detected in the Aka Pygmy population of Central African Republic. No Ebola hemorrhagic fever cases were reported in these areas.

A more recent study examined sera from 4,349 individuals in 220 villages in Gabon. Antibodies against Zaire ebolavirus were detected in 15.3% of those tested, with the highest levels in forested regions (see map). The authors believe that the seropositive individuals had mild or asymptomatic ebolavirus infection:

The high frequency of ‘immune’ individuals with no disease or outbreak history raises questions as to the real pathogenicity of ZEBOV for humans in ‘natural’ conditions.

These findings indicate that the fatality rates of Zaire ebolavirus that are quoted widely are likely to be vast overestimates. Why the infection is more lethal during outbreak conditions is not known. One possibility is related to the size of the viral inoculum received. During outbreaks the virus is spread by contact with the blood, secretions, organs or other body fluids of infected individuals, which contain very large quantities of virus. In contrast, infections in nature – by contact with contaminated fruit, for example – may involve far less virus.

Whether we are discussing avian H5N1 influenza, ebolavirus, or even the fictitious MEV-1, do not assume that widely quoted fatality rates are correct – check the scientific literature!


Should we fear avian H5N1 influenza?

Becquart P, Wauquier N, Mahlakõiv T, Nkoghe D, Padilla C, Souris M, Ollomo B, Gonzalez JP, De Lamballerie X, Kazanji M, & Leroy EM (2010). High prevalence of both humoral and cellular immunity to Zaire ebolavirus among rural populations in Gabon. PloS one, 5 (2) PMID: 20161740

Riding the influenza pandemic wave

1973927918_ce00011ef5_mOne notable characteristic of the four previous influenza pandemics is that they occurred in multiple waves. The 1918 pandemic began with outbreaks of low mortality in the spring and summer, followed by a more lethal wave in the winter. This pattern has fueled speculation that the current H1N1 pandemic strain will undergo mutation that leads to the emergence of a more lethal virus. What is the evidence that pandemic waves of increasing virulence are a consequence of viral mutation?

The only virus available from the 1918 pandemic was rescued from an Alaskan influenza victim who was buried in permafrost in November of that year, when higher mortality was already evident. This makes it impossible to correlate any genetic changes in the virus with increased virulence. Furthermore, as discussed on ProMedMail,

…there are many different ways of interpreting these differences other than more virulent virus. Some of these are differences in populations affected, more circulation of pneumococci and staphylococci during cold weather, more circulation of other viral pathogens, more virulence and larger inocula with the crowding and cold air inhaled.

The November 1918 influenza virus certainly has genetic and phenotypic properties expected of a virulent virus. These include the ability to multiply in the absence of trypsin*, lethality in mice and embryonated chicken eggs, and efficient replication in human bronchial epithelial cells. But we don’t know if these properties were absent from the virus that circulated in the spring of 1918.

Do the pandemics of 1957 and 1968, which also occurred in waves of increasing lethality, provide any information? Viruses are available from different stages of these pandemics, but to my knowledge the virulence studies have not been done.

This uncertainty makes it impossible to conclude that the 2009 H1N1 pandemic strain will become more virulent. Nevertheless, speculation is rampant, and accompanied the recent release of the Brazilian isolate. Another example is an amino acid change in the viral PB2 protein observed in some 2009 H1N1 isolates. According to Recombinomics,

Acquisition of E627K is a concern because it allows for optimal replication at 33 C, the temperature of a human nose in the winter, in contrast to E627, which is in the avian version of PB2 and allows for optimal replication at 41 C, the body temperature of birds. The appearance of E627K raises concerns that the level of swine flu with E627K will markedly increase in colder months. In 1918, the flu in the spring was mild, but the fall version of the virus, which had E627K, was much more virulent and targeted young, previously healthy adults…

If the amino acid at 627 is an important determinant of virulence, we would expect to find E627 in viruses isolated early in the 1918 pandemic – but such viruses are not available. Therefore the role of this amino acid change in virulence in humans cannot be tested. Further complicating the situation is that other amino acids in the viral PB2 protein can influence viral replication at low temperatures.

Fortunately, new H1N1 isolates are obtained every week, which provide a very accurate sampling of the entire pandemic. Should the new H1N1 strain become more virulent, it will be a relatively straightforward task to determine the genetic changes that accompany this property. Finally we will be able to determine if  pandemic waves of increasing virulence are a consequence of specific changes in the viral RNA.

*We’ll discuss the requirement of proteases for influenza virus replication next week.

Miller, M., Viboud, C., Balinska, M., & Simonsen, L. (2009). The Signature Features of Influenza Pandemics — Implications for Policy New England Journal of Medicine, 360 (25), 2595-2598 DOI: 10.1056/NEJMp0903906

Tumpey, T. (2005). Characterization of the Reconstructed 1918 Spanish Influenza Pandemic Virus Science, 310 (5745), 77-80 DOI: 10.1126/science.1119392.