p53

by Sue Armstrong

  CHAPTER FOURTEEN

  The Smoking Gun

  In which we discover that the component in tobacco smoke that causes lung cancer, benzo(a)pyrene, does so by sticking to DNA and damaging the p53 gene, leaving a mutation that is as unique and incriminating as a fingerprint.

  ***

  When a risk factor for a disease becomes so highly prevalent in a population, it paradoxically begins to disappear into the white noise of the background . . . If nearly all men smoked, and only some of them developed cancer, then how might one tease apart the statistical link between one and the other?

  Siddhartha Mukherjee

  In the mid-1990s, p53 emerged briefly from the rarefied environment of academia to play a dramatic role in the battle between public-health authorities and the tobacco industry by decisively nailing the connection between smoking and lung cancer. The connection itself was not news. Nearly half a century earlier, in 1950, the Oxford-based epidemiologist Richard Doll – a maverick and somewhat controversial figure among the medical establishment because of his membership of the Communist Party – had drawn attention to the association between cigarettes and lung cancer in a paper for the influential British Medical Journal. In it, Doll and his collaborator, Austin Bradford Hill, reported their findings from a research project aimed at figuring out what was causing the sudden explosive rise in lung-cancer cases in the UK.

  In the quarter-century from 1922 to 1947, deaths from lung cancer in England and Wales rose from 612 to 9,287 per annum – ‘one of the most striking changes in the pattern of mortality recorded by the Registrar-General’, Doll said in his report. Tobacco had been a rare source of comfort to men fighting in the trenches and on the high seas of World War I and cigarettes were included among a soldier’s or sailor’s rations in both the British and American armed forces. Indeed, General John Pershing, leader of the American Expeditionary Forces in World War I, is reported as saying, ‘You ask me what we need to win this war. I answer tobacco as much as bullets . . . We must have thousands of tons without delay.’

  Thus a generation of young men returned to civilian life addicted to tobacco, entrenching a habit that had been growing since the invention in the late 1880s of a machine for rolling cigarettes that enabled mass production. By the time Doll and Hill were doing their studies in the late 1940s, approximately 80 per cent of British men were regular smokers and the death rate from lung cancer in England and Wales had increased more than sixfold among adult men and approximately threefold among women in less than 20 years.

  The two scientists did not, however, suspect tobacco of being the cause of the premature deaths. Their initial hunch was that it was atmospheric pollution – dust from the new road-building material, tarmac; exhaust fumes from cars, power stations and the coal fires burning in everyone’s front rooms. But the picture of smoking habits that emerged from the questionnaires conducted with 1,732 cancer patients and 743 controls in 20 London hospitals pointed compellingly to tobacco exposure as the culprit. Doll and Hill’s BMJ paper concludes with the statement that ‘above the age of 45, the risk of developing [lung cancer] increases in simple proportion with the amount smoked, and that it may be approximately 50 times as great among those who smoke 25 or more cigarettes a day as among non-smokers’.

  Though he had no idea what the carcinogenic substance in tobacco could be, Doll himself gave up smoking on the strength of the epidemiology. As the evidence of its deadly effects accumulated, governments in many countries adopted measures aimed at curbing the habit. But as long as the link between smoking and lung cancer remained circumstantial, without a physical mechanism to back it up, cigarette companies had ample wriggle room to fight the public-health case against them. They continued to promote their product aggressively, particularly in developing countries with potentially massive markets and poor controls. Even when scientists, following on from Doll and Hill’s work, showed that painting tar – the sticky black residue of tobacco that coats smokers’ lungs – on to the skin of laboratory mice causes tumours to grow, Big Tobacco was wont to scoff that the experiments were irrelevant: these were mice, not people.

  In fact, the first person to do such an experiment was Angel Honorio Roffo, an Argentinian oncologist who flagged up the link between smoking and cancer nearly two decades before Doll and Hill’s paper in the BMJ. By painting rabbits’ ears and the skin of mice repeatedly with either nicotine or tobacco tar, Roffo identified the latter as being the carcinogenic substance – nicotine alone had no effect on the animals’ skin no matter how long he left it. But the results of his extensive research into the disease that afflicted his patients went almost unnoticed by anyone except the tobacco industry. This was at least partly because his most important scientific papers were published only in German and partly because he was way ahead of his time in his understanding of cancer biology. So far ahead, in fact, that a New York physician who had heard of Roffo’s studies and written to the American Tobacco Company (AT) in May 1939 to ask if his findings were valid was able to be reassured by the cigarette manufacturer’s own research director. Hiram Hanmer replied to the doctor that AT had been following Roffo’s work for some time and felt that it had left the literature on tobacco ‘in a very beclouded condition’. He assured his correspondent that ‘the use of tobacco is not remotely associated with the incidence of cancer’.

  Rubbishing the science has always been the tobacco industry’s modus operandi, but it became infinitely more difficult – and finally impossible – once the molecular biologists were on the case. Curt Harris is a bear of a man with a shaggy grey beard and a deep bass voice as rich as a pint of brown ale; he heads the Laboratory for Human Carcinogenesis at the National Cancer Institute in Bethesda. He is also a pioneer of the relatively young field of molecular epidemiology, the science that traced the origins of the AIDS pandemic back to chimpanzees and green monkeys in the forests of Africa by following the genetic footprint of the virus that became HIV. It’s the science that uncovers the sources and assesses the virulence of regular flu outbreaks; and it’s a science used widely today to try to pin down the multiple causes of cancer.

  In the 1980s, Harris’s lab was involved in studying the activity of carcinogenic substances, including components of tobacco tar, when they get into cells. They found that a number of these substances stick themselves firmly to the DNA, and Harris knew from the literature and from his own studies that this would lead to mutations in the genes, the first step along the road to cancer. He became especially fascinated by Bert Vogelstein’s work with p53, and the two men decided to collaborate on a research project to assess how commonly p53 might be mutated in cancer. This was 1989, just after Suzy Baker in Vogelstein’s lab had made her sensational discovery that p53 was a tumour suppressor, not an oncogene. Harris and Vogelstein found p53 mutations in many of the common tumour types they saw in their clinics, including breast, lung, brain and colon. They found too that the mutations showed a pattern and were clustered in four particular positions along the length of the gene, which they dubbed ‘hot spots’.

  The two researchers published their findings in Nature in 1989. Soon afterwards Harris, who remained intrigued by the pattern he and Vogelstein had revealed, decided to collect systematically the information on the different mutations mentioned in the steady stream of papers that appeared on p53. In 1990, in collaboration with Monica Hollstein, then working at the World Health Organization’s International Agency for Research on Cancer (IARC) in Lyon, France, he formalised his collection into the p53 database, a rare resource for scientists and clinicians that can provide clues to the identity of carcinogens in the environment, as well as to the likely course of a patient’s disease or the best options for treating certain tumours.

  In 1994, Hollstein left IARC and her place at the helm of the database was taken by Pierre Hainaut, who expanded it, increased the level of detail recorded for each mutation and managed it until mid-2012. Today the database contains information on tens of thousands of mutations, toget
her with the tumours in which they occur and as much information as possible about the lifestyles and personal characteristics of cancer patients, including their response to treatment and the final clinical outcome if available.

  ‘We started the database because we had this idea that there was going to be a relationship between environmental causes of cancer and the p53 mutation spectrum,’ Harris told me. And indeed to the disease detectives – the molecular epidemiologists – the database has proved invaluable. Its role in the tobacco story is one of its most notable successes.

  TOBACCO’S FINGERPRINT FOUND ON p53

  In the early 1990s, Gerd Pfeifer, who heads a lab at the City of Hope medical centre in Duarte, California, was exploring DNA damage in relation to cancer, looking to see if the causative agents left distinctive patterns of damage, or ‘fingerprints’, that identified them as culprits. His lab had developed a tool that enabled the scientists to home in on individual genes among the thousands along a strand of DNA – a process akin to finding needles in haystacks – and the general buzz surrounding p53 at the time persuaded Pfeifer and his fellow researcher, Mikhail Denissenko, that this gene would be an interesting focus for their studies. They would look at the effects of tobacco smoke on p53.

  ‘The early tobacco work had clearly suggested there is high tumour-causing activity in the tar – the black stuff you can collect on filters when you burn tobacco, and that you can see in the lungs of heavy smokers when you operate on them. It looks gross,’ Pfeifer commented when I phoned him in his lab. He and Denissenko knew that the most damaging components of tar are the polycyclic aromatic hydrocarbons, or PAHs, of which one, benzo(a)pyrene (BaP), is particularly noxious. This, they decided, would be an ideal damaging agent to use in their experiments.

  Others had already investigated what happens to PAHs when they get inside cells. Their research showed that these substances are not water soluble, so the body has difficulty ridding itself of them. In an effort to transform the compounds into something that can then be excreted, the machinery of the cells creates a dangerously reactive substance that sticks itself to the DNA. In the case of BaP, this transformed substance has a mind-numbing formula represented simply as BPDE, which is considered to be one of the most potent carcinogens yet discovered.

  Pfeifer and Denissenko took BPDE and added it to various cell types, including human lung, and then left the cells to their fate. After an hour or two they returned to their lab benches to isolate the DNA and to apply their special technique to identifying the exact positions on the p53 gene that had sustained damage. This, they discovered, was not random. BPDE always attached itself to the DNA next to the guanine base – one of the four basic building blocks of DNA, represented by G in the DNA code – at three very specific ‘hot spots’ along the gene, at codons 157, 248 and 273 (to recap, a codon is a segment of a gene just three bases long that codes for one of the building blocks of the protein, and the codon’s number defines where that building block should go).

  Here at last was proof that a defined product of cigarette smoke damages DNA. But the clincher for the case against Big Tobacco came when Pfeifer and Denissenko compared their lab’s results with the p53 mutation database, which by 1996 had more than 500 entries for lung cancer drawn from the literature worldwide. The great majority of the mutations described in the database among smokers, but rarely among non-smokers, corresponded precisely with their results: they occurred in the same hot spots targeted by the BPDE and they showed that the building blocks of the gene had been scrambled such that the guanine (G) was replaced by a thymine (T). Crucially, while codons 248 and 273 are mutation hot spots in many types of cancer, the database revealed that codon 157 is found exclusively in lung tumours. In other words the fingerprint of BPDE was all over the p53 database. Pfeifer and Denissenko’s paper, published in Science in October 1996, concluded, ‘Our study thus provides a direct link between a defined cigarette smoke carcinogen and human cancer mutations.’

  BIG TOBACCO QUESTIONS THE SCIENCE

  This was bad news for Big Tobacco. It meant not just that smoking was implicated in a generalised threat to public health, but that it could be linked to individual people with lung cancer. Tobacco companies were now much more vulnerable to lawsuits from customers looking for compensation for ruined lives and, as ever, they set about trying to refute the evidence. In an address to investors, analysts and journalists soon after the Science paper came out, the Chief Executive of British American Tobacco Industries (BAT), Martin Broughton, stated, ‘There is still a lack of understanding of the mechanisms of disease attributed to smoking . . . The importance of this Science magazine study may lie, not least, in the recognition that there are important missing links in the understanding of causation . . . It may lead to further research . . . into the complex process by which a cell becomes cancerous – a process we and others have spent millions in trying to understand for many years now.’

  The R J Reynolds tobacco company was even more blatantly dismissive. A public statement issued by the company said, ‘That BaP will cause a mutation has been known for a long time . . . The authors themselves describe these findings as a coincidence. The press release’s conclusion that these findings are the key to lung cancer is an overstatement.’

  Interestingly, R J Reynolds’ statement came out on the eve of publication of Pfeifer and Denissenko’s study being published in Science – clearly suggesting the company had been tipped off in advance.

  The following year, Pierre Hainaut and a colleague at IARC, Tina Hernandez-Boussard, carried out a detailed analysis of the spectrum of p53 mutations in smokers as recorded in their database. Their paper for Environmental Health Perspectives came to the same conclusion as Pfeifer and Denissenko’s: that these mutations carried the fingerprint of BPDE. ‘We thought at the time, that settles it,’ said Hainaut as we sat together in the front room of his home in Lyon, with its distant views of the snow-covered Alps, talking about the database’s role in revealing the causes of cancer. ‘You have experimental data; you know what the mutagenic substance is; you can demonstrate its effect very well in the lab and you can show that the people who are exposed to the same substances in real life get the mutation at exactly the same place.’

  Hainaut and Pfeifer believed their two papers made an irrefutable case, and they were taken aback two years later to see first one and then a second paper challenging their results as being ‘over-interpretations’. ‘People had done some analysis themselves of our database to try to prove that we were wrong – that the connection was not there; that maybe tobacco was helping mutations occur, but not actually causing them,’ explained Hainaut. ‘I was shocked. But also I have to say that we’re used to trust within the scientific community, and you expect people to be fair. So the first reaction when you see a paper like this attacking your work is, oh my God, I’ve missed something very important! I’ve made a big mistake! In fact when the first of these papers came out my director called me into his office and said, “What’s this about? You have three days to bring all your data to me for review, because if you’ve mishandled or misrepresented an issue like that it’s a serious matter.” So I was in the hot seat somehow!’

  To Hainaut’s relief, his interpretation stood up to review and he began to wonder about the author of the second paper, published in Mutagenesis – a scientist named Thilo Paschke, working for a Munich-based institute, Analytisch-Biologisches Forschungslabor. At this point in our conversation, Hainaut disappeared into his study and returned with a sheaf of yellowing papers held together in a clear plastic-covered folder – a record of the murky drama he found himself being dragged into. ‘I had never heard of this guy, never seen a paper by him and I had no idea about the institute he worked for,’ he continued. ‘So I tried looking up the institute on the internet and I didn’t find any website. All I had was an address in Munich, so I called the German telecommunications people and they told me, “It’s the address of the German Association of Tobacco Manufacturers.”


  ‘My first reaction was relief. I realised, okay, that’s a completely unfair attack on our work and I can probably forget it. All I have to do is demonstrate that it comes from people who have a bias they haven’t declared.’

  But then the plot began to thicken. As part of the general settlement of lawsuits between the American States and Big Tobacco in 1998, the industry was required to make publicly available all internal documents used during the trial. These were organised systematically and posted on the internet at www.tobaccodocuments.org. In early 2001, Hainaut opened his computer at the website in search of information about the journal Mutagenesis, whose behaviour he and Pfeifer had found ‘strange’ when they raised the issue of Paschke’s conflict of interests and asked for space to respond to his critical paper. Their request had been turned down and, intrigued to know if anything suspicious lay behind the rejection, Hainaut typed the name of the journal’s founding editor, Jim Parry, into the site’s search engine and watched wide-eyed as a series of references popped up on screen.

  Here was evidence of research and consultancy contracts between Parry and BAT and Philip Morris running over more than a decade. To the fascinated Hainaut, a letter dated 3rd August 1994, from Parry to his main contact at Philip Morris, the company’s chief scientist Ruth Dempsey, about a proposed research project was particularly revealing. In it Parry, then a Professor of Biochemistry at the University of Wales at Swansea, advised Dempsey that ‘the overhead figure of 40 per cent I have given in the costs can be overcome if money is given to me as a gift, “as a contribution to my research” without specifying how it should be spent. My colleagues tell me that some companies are increasingly appreciating the financial advantage of giving money in this way compared to contracts which specify components of the project.’