If H5N1 Evolves

Hugh Pennington

I worked on bird flu in a laboratory in London in the 1960s. We called it KP, short for klassische Geflügelpest. The boss was an ardent Germanophile, but this wasn’t the only reason. He wanted us to remember Werner Schäfer’s discovery in 1955 in Tübingen that KP, fowl plague, was an influenza virus, and Shäfer’s suggestion that such bird viruses might have the ability to change and infect other species. It was feared by chicken farmers. It spread rapidly in flocks and was a killer. Birds at the near end of a hen-house would all be dead, those in the middle dying, some with swollen heads and diarrhoea and others just falling without warning, while those at the far end looked healthy but were doomed to drop within the day.

It was rumoured that Schäfer had been attached to Rommel’s Camel Corps, so I was apprehensive when I visited his laboratory in 1966. But he was on holiday in the Canary Islands. In his absence, a brick had been put in the middle of the shallow bath of disinfectant in which one was supposed to paddle when going in and out of the lab. The stepping-stone signified the independent spirit of the young scientists in the lab – they had all been to America – as well as their low regard for the virus as a health hazard. Bird flu killed only birds.

Influenza’s secrets were falling fast to molecular biology. It was possible to explain why new strains appeared so often. The genes coding for the protein spikes – the haemagglutin (H) and the neuraminidase (N) – of the virus mutate at very high rates. Most of its genes are physically separate from each other. They reassort randomly when two different influenzas infect a cell simultaneously: a kind of sexual reproduction. The mutations cause small but significant changes in H and N. But every now and then a virus suddenly appears which is so different that people are not protected by the immunity they have from previous flu infections. These viruses cause pandemics. They arise from gene reassortment between an animal virus and a human one, and can evade existing immunity because the H and N proteins are new for humans. The subtype of the virus that caused the biggest pandemic in the 20th century was H1N1. It is thought to have come from pigs. It killed about 40 million people in 1918 and 1919. The next big pandemic – Asian flu – was in 1957. Its subtype was H2N2. Eleven years later came the Hong Kong flu, subtype H3N2. Flu had also been quite busy in the late 1940s, though the virus did not have new H or N proteins. There seemed to be a pattern: a pandemic every decade or so. By the mid-1970s, this was the experts’ consensus.

In February 1976, an 18-year-old US Army recruit at Fort Dix, New Jersey, died of influenza after a forced march. He had been ill for only a few hours. The virus was an H1N1 swine subtype. The virologists were alarmed: a pandemic was due. Ten years had passed since the last one. Perhaps the 1918 virus had returned: it specialised in killing fit soldiers, as well as anyone else between the ages of 20 and 40. The identity of the Fort Dix virus was confirmed on 13 February. Work started on a vaccine on the 17th. There was a press conference on the 19th and the public picked up the 1918 connection on the 20th. David Sencer, Director of the Center for Disease Control at Atlanta, Georgia, prepared a memorandum called ‘Swine Influenza: action’. It went via the secretaries of health, education and welfare to President Ford. On 24 March a ‘Blue Riband Panel’ of experts met at the White House. Sencer’s memo said there was ‘a strong possibility that this country will experience widespread swine influenza in 1976/77’. It talked of the need for federal leadership. Ford was doing badly in the Republican presidential primaries against Ronald Reagan, so his announcement later that day that he was seeking $135 million from Congress to vaccinate ‘every man, woman and child in the United States’ wasn’t surprising. The vaccination programme started on 1 October; Ford was immunised on television on the 14th. One possible vaccination complication was Guillain-Barré syndrome, a rare paralytic illness known to follow infections. It was looked for, and in November and December some post-vaccination cases were found. The number was very small, and had to be balanced against the tens of millions of satisfied vaccinees – but it was greater than zero, the number of people who had caught swine flu from another human. The vaccination programme was suspended on 16 December. Ford had already lost the election; Sencer was sacked three weeks after Jimmy Carter took office.

Virologists no longer believe in ten-year influenza cycles. Also gone is the hope that understanding the structure of the virus and the sequence of its genes would lead quickly to the discovery of all the rules governing its lethality and transmissibility. The virus has only a handful of genes, but they act in concert, and they interact with several human genes. The complexity of this relationship has so far frustrated attempts to develop laboratory tests for virus nastiness. Even the reconstruction of gene sequences of the 1918 virus – from tissue samples recovered from Alaskan Inuits buried in permafrost and from post-mortem samples of US soldiers’ lungs preserved in paraffin blocks – has had only limited explanatory success. But it has confirmed one certainty about flu: its capacity to surprise. The 1918 sequences suggest that its origins were not as simple as once thought: it was believed to have been created by a reassortment event between human and bird viruses that occurred when they simultaneously infected pigs. It now looks as though another host was involved; but it remains to be discovered which one.

A much bigger surprise was the behaviour of bird flu in Hong Kong in 1997. In March, several thousand birds died in rural chicken farms. Fowl plague was back. In May, a three-year-old boy died. It took a long time to identify the virus that killed him because it hadn’t been found in humans before. In August it was found to be the same as the chicken virus, subtype H5N1. In November there were more human cases, and by late December there had been 18. Six people died; all had been in contact with chickens. The infection of people by bird flu was novel, but a 30 per cent mortality rate was frightening, as bad as the worst kind of smallpox. The only comfort was that the virus did not spread from person to person. Hong Kong banned chicken imports, and on 28 December the slaughter began of all 1.6 million chickens already there. It was completed by the end of the year. The outbreak ended.

But H5N1 bird flu had not gone away. It killed another person in Hong Kong in 2003. In 2004 an outbreak in Thailand and Vietnam had a human mortality rate of 60 per cent. It has been infecting birds there ever since and is busy in Cambodia and China too. It has spread to Indonesia and infected pigs there and in China, where it recently killed lots of bar-headed geese at Qinghai Lake. Close contact with chickens is still a prerequisite for human infection. Mortality estimates are probably exaggerated because mild infections are more likely to be missed and so are not being recorded. But even when this is taken into account, they remain outrageously high, about ten times greater than in 1918-19. So far there have been 88 confirmed cases, 51 fatal.

The recent upsurge of bird flu in Asian chickens and its lethal dribble into human populations has got through to public health policy-makers and to politicians. Virologists have made clear that it wouldn’t be a big evolutionary step for H5N1 to acquire the ability to spread from person to person. If it did this while retaining its current ability to kill, the impact of the resulting pandemic would dwarf that of 1918-19, when more than 200,000 people died in Britain. It could be ten times worse. But this is not the only evolutionary possibility: the virus may die out; the status quo might prevail for years to come; or the virus may become transmissible among human populations but with decreased virulence. The difficulty for policy-makers is that evolution is a random process. This means that the likeliest outcome of any prediction is that it will be wrong. The right thing to do is to hope for the best but prepare for the worst.

The news is not all bad. The WHO Global Influenza Surveillance Network has only a small number of co-ordinating staff in Geneva, but they have outstanding networking skills and led the remarkably successful international response to SARS in 2003. (It was fortunate that it was a quiet year for flu.) And safe anti-influenza drugs and vaccines are available. The UK government is stockpiling tamiflu (oseltamivir), which can provide protection if it is taken before infection, and shortens the illness when given at its beginning. Whether it would stop a pandemic is much less certain. It has never been used in one. Its beneficial effect in severe pneumonia, the kind that kills, is also not known. Flu vaccines have been used for many years. They are good, if not as effective as vaccines for MMR, polio or smallpox. But they have the potential to prevent epidemics if they are given early enough on a large scale.

To manufacture the vaccine, DNA copies of the genes required are made and shot into tissue culture cells, which then grow the virus. To make it safer, the small part of the H component which makes a major contribution to its virulence in chickens is cut out. The live vaccine is then tested in chickens (does it cause fowl plague?) and ferrets (which respond to different influenza viruses in a surprisingly similar way to people). In traditional vaccine manufacture there is an emphasis on protecting the vaccine from contamination by the workers making it; with H5N1 it is the other way round. If the live vaccine virus spread to a worker infected with another flu virus, reassortment could lead to the generation of a new virus – and an outbreak spreading from the factory. So the vaccine has to be made under strict microbiological containment by immunised staff taking antiviral drugs. Special rules also apply because the virus is genetically modified. Finally, before being licensed, the vaccine has to be tested in people for safety and its ability to produce immunity.

The UK’s recent relationship with new and evolving infections is an unhappy one. Twenty years have passed since the first case of BSE was diagnosed, in a dairy cow from Kent. Although 2001 was an election year the hustings were postponed because of the foot and mouth disease outbreak. The handling of both infectious agents left much to be desired. The optimism that prevailed during the early years of BSE that it would not affect humans meant that although the right things were done, they were not done as well as they could have been. It was the same with foot and mouth disease: there was a contingency plan, but it was the minimum the government could get away with and still meet European Directive 90/423/EC. It was designed to cope with an outbreak of ten cases. On the day the first case was confirmed, it turned out that 57 farms were infected. Have lessons been learned? Recent government pronouncements indicate that optimism still rules. Its ‘Pandemic Flu Contingency Plan’ says that during the first wave of the pandemic it ‘assumes rising – but probably limited – stocks of antivirals’. So there is no plan to stockpile enough for even a majority of the population. Vaccine orders are to be delayed until there is more certainty about the kind of virus to be protected against (but past experience shows that flu can spread worldwide very quickly). And the publicly announced official estimates of the number of deaths that could occur if H5N1 turns into a human virus are at the low end of the range.

A Ministry of Agriculture official told the BSE Inquiry that there was a policy of making ‘more reassuring-sounding statements than might ideally have been said’. He called it ‘leaning into the wind’: a new infectious agent appears; warnings are given; but years go by and nothing happens, so it seems fine to downplay the worries. Eleven years elapsed between the first case of BSE and the description of its human form, vCJD, as a new disease. And 11 years went by between the emergence of the new Type O pan-Asia strain of foot and mouth disease in North India and its spread to the UK. If it takes as long for a human H5N1 to evolve – 2008 – let’s hope that by then we’ll be ready.