The Cancer Chronicles Read online

  Some of our fears were rooted in a misunderstanding. Epidemiologists define “environment” very broadly to include everything that is not the direct result of heredity—smoking, eating, exercise, the bearing of children, sexual habits, any kind of behavior or cultural practice. Viruses, exposure to sunlight, radon, cosmic rays—these are all defined as environmental. To get a sense of how strongly cancer was influenced by heredity and how strongly by these extrinsic factors, scientists in the 1950s studied populations of black people whose ancestors had been captured by slave traders and moved to the United States and compared them with their relatives who remained in Africa. Liver cancer and Burkitt’s lymphoma turned out to be very high among the Africans but not among the American blacks. Lung, pancreatic, breast, prostate, and other cancers were far higher among the black Americans than the Africans. Other researchers found similar patterns. Japanese men were known to have higher rates of stomach cancer but lower rates of colon cancer compared with their counterparts in the United States. When they moved to this country, the situation changed. They tended to adopt the cancers of their hosts and leave the native cancers behind. Since their genes remained the same, factors beyond heredity had to be involved.

  By the late 1970s decades of these migrant studies had come to the same conclusion: For 90 percent of cancer cases some kind of external influence was required. Something “environmental.” There was a chance of a person getting a head start on cancer by inheriting a damaged gene. But most of the mutations that triggered a malignancy were those acquired during life. That was encouraging news for public health and prevention. But it was often misconstrued to mean that almost all cancer was brought on by pollution, pesticides, and industrial waste. That fit so perfectly with the rest of our worldview that there was little incentive to look deeper. Calmer voices called for a more balanced perspective, but it was the direst warnings that became enshrined in public perception. If we or someone we knew ever got cancer we were quick to wonder whether corporate America was to blame.

  There was more to the story than politics and semantics. In 1973, not long after Richard Nixon declared the War on Cancer, the government Surveillance, Epidemiology and End Results Program, called SEER, began collecting data from state cancer registries on incidence and mortality—how frequently people got cancer and how often it killed them. For years the mainstream view had been that except for lung cancer, overall rates were holding steady. But in 1976 when the new SEER data were compared with earlier surveys by the National Cancer Institute, the number of new cases seemed to be escalating abruptly, even when the aging of the population was allowed for. This appeared to be the vindication so many people sought.

  Combining two sets of statistics, compiled from different sources according to different rules, is bound to cause trouble. Early on epidemiologists warned that the comparisons were invalid and no conclusions should be drawn—that there was no evidence of a cancer epidemic. To get a clearer idea of what the public was facing, the U.S. Office of Technology Assessment commissioned a study by Richard Doll and Richard Peto, two Oxford University epidemiologists who had made names for themselves by establishing the link between cigarettes and cancer as well as the carcinogenic effects of asbestos. It would have been hard to find two more accomplished scientists in their field.

  To begin with, they had to decide which numbers to trust. Although they were improving, statistics on cancer incidence—the number of new cases occurring in a population—were not yet dependable. What appeared to be more new cancers might be the result of better diagnostics, more accurate medical records, and an ever-increasing proportion of the population seeking and receiving medical care. Death certificates from earlier in the century were also suspect. Doctors might acquiesce to a family’s request that the stigma of cancer not be entered into the public ledgers. Mistakes in both record keeping and diagnosis were often made. Someone who died of lung cancer might be listed as a victim of pneumonia. A death from an undiagnosed brain tumor might be attributed to senility. A patient might be recorded as dying of cancer when the cause was really something else. The situation improved in 1933 when states began reporting deaths to a central registry, and midway through the century a standardized classification scheme was put in place. (Cervical and uterine cancer had been lumped together, and Hodgkin’s lymphoma, a blood cell malignancy, was misconstrued as an infectious disease.) Starting with 1950 and using death rates as the best available approximation for the prevalence of cancer, the authors produced an intricate analysis spanning more than a hundred densely filled pages of words, tables, and graphs and six meticulous appendices. In addition to their own calculations they also reviewed the findings of more than three hundred other studies.

  Since it was published in 1981, Doll and Peto’s “The Causes of Cancer” has become one of the most influential documents in cancer epidemiology. It concluded that most cancer, by far, is “avoidable”—brought on by factors that, to a great extent, are within the grasp of human control. In 30 percent of cancer deaths, tobacco was a cause. For diet the proportion was 35 percent, and for alcohol it was 3 percent. Some 7 percent of fatalities involved “reproductive and sexual behavior,” which included delaying or forgoing the bearing of children and promiscuous sex. (Having multiple partners was recognized as a risk for cervical cancer, although it was not yet known that the agent was human papillomavirus.) Another 10 percent of cancer was tentatively attributed to various infections and 3 percent to “geophysical” phenomena: exposure to the ultraviolet components of sunlight and naturally occurring background radiation from soil and cosmic rays. For deaths by artificially produced carcinogens, including radioisotopes, the percentages came out very low: 4 percent from occupational exposure, 2 percent from air, water, and food pollution, 1 percent from the side effects of medical treatment (including x-rays and radiotherapy), and less than 1 percent from either industrial products like paints, plastics, and solvents or food additives. The remainder was of unknown origin with the suggestion that psychological stress or a compromised immune system might be involved. Except for lung cancer, Doll and Peto concluded, “most of the types of cancer that are common today in the United States must be due mainly to factors that have been present for a long time.”

  What a hard conclusion this was to swallow. Any specific case of cancer will have multiple causes—environmental (in the broadest sense) along with hereditary dispositions and the elusive influence of bad luck. But for the public at large, chemicals spewed from factories or the polysyllabic additives found in foods were apparently only minor parts of the equation. They were a component—“there is too much ignorance for complacency to be justified,” the authors wrote—but far more important was how we lived and the effect that had on a cell’s natural tendency to break loose and assert its Darwinian imperative. Most telling of all, Doll and Peto found that cancer had not been increasing rapidly, as one would expect if we were being subjected to an efflorescence of newly invented assaults. When lung cancer and other malignancies closely associated with smoking (oral, laryngeal, esophageal, and others) were removed from consideration, and the aging of the population adjusted for, cancer mortality among people under sixty-five had been steadily decreasing in almost every category since 1953. (That also appeared to be largely true for older Americans, but those figures, relying on earlier medical and census reports, were considered less reliable.) The lower mortality was not because we were getting much better at curing cancer, the authors concluded, but because the number of new cases was not escalating. Once SEER became better established and the quality of data improved, they confirmed that there was no alarming rise in the incidence of cancer.

  Doll and Peto were not alone in their findings. Two smaller studies, one in the United States and one for the industrial city of Birmingham, England, had come up with similar percentages—with most cancer attributed to smoking and a mix of other so-called lifestyle factors and with occupational exposure responsible for only a few percent. But “The Causes of Canc
er” was the most wide-ranging study that had been undertaken. Its conclusions were, of course, what the leaders of industry wanted to hear, and people committed to fighting the problems of industrial pollution began challenging the report. The lifestyle argument was dismissed as a diversion—blaming the victims instead of the perpetrators. While cigarettes were clearly an important influence, maybe a significant number of smokers wouldn’t have gotten lung cancer without additional help from polluted air or synthetic carcinogens—some knotty synergistic effect. Whatever was happening with overall rates, the incidence of some cancers appeared to be rising, especially among the aged and minority groups. Maybe what Doll and Peto laid to better diagnosis were really hints of carcinogenic poisons that were steadily accumulating and would erupt in coming years in a devastating outbreak of cancer. When lung cancer rates began rising earlier in the twentieth century, that was also dismissed as an artifact of better diagnostics. Only with time would the true horror we were inflicting on ourselves become clear.

  While epidemiologists kept watch for the appearance of a delayed epidemic, Bruce Ames, the inventor of the Ames test, was also coming to question whether synthetic substances were a significant threat. It was Ames who, back in 1973, had used experiments with bacteria to show that carcinogens, most of them anyway, caused cancer by inducing genetic mutations. (Not all carcinogens are mutagens. Some can work more indirectly. By killing esophagus cells and increasing their rate of replacement, alcohol raises the odds of random copying errors.) As his test became established, Ames worried at first about the hazards of what modern man was releasing into the world. His early research helped lead to bans on carcinogens that were being used as flame retardants in children’s pajamas and in hair dyes. He helped persuade California to strengthen its regulation of an agricultural fumigant. He became something of an environmental hero. Then he began testing chemicals that occurred in nature, finding that a surprising number also appeared to damage DNA.

  It made good evolutionary sense. Throughout time plants have evolved the ability to synthesize chemicals that ward off predators—bacteria, funguses, insects, rodents, and other animals. Ames described some of these natural pesticides in a paper in Science in 1983. The black pepper used to spice our food contains safrole and piperine and causes tumors in mice. Edible mushrooms carry hydrazines that are carcinogenic. Celery, parsnips, figs, and parsley have carcinogenic furocoumarins. In chocolate there is theobromine, and pyrrolizidine alkaloids are found in various herbal teas. Over the years Ames continued to keep count. In 1997, he reported that of sixty-three natural substances found in plants, thirty-five tested as carcinogenic. His most striking example was a cup of coffee—nineteen different carcinogens, including acetaldehyde, benzene, formaldehyde, styrene, toluene, and xylene. Altogether, he estimated, people were imbibing ten thousand times as many natural pesticides as manufactured ones. Those seeking chemical causes of cancer, he said, were looking in the wrong place.

  In fact he doubted that nature’s poisons were really causing much cancer. Often forgotten is that his paper in Science also listed numerous antioxidants and other elements in plants that might conceivably provide some protection. It was possible, Ames proposed, that the good outweighed the bad, that on balance eating fruits and vegetables might reduce the incidence of cancer. But no one really knew.

  Ultimately Ames’s message was that we were worrying too much about both kinds of chemicals, natural and artificial. Half of everything tested, he wrote, was registering as carcinogenic, but that didn’t necessarily mean that the substances were dangerous. Suspected carcinogens are administered to rodents using what is called the maximum tolerated dose—as much as the animals can take without debilitating effects. This is many times the exposure that people might receive in the world. There is a logic to this approach. Suppose that exposing 10,000 people to some chemical results in a single instance of cancer. For a population of 10 million that is 1,000 potentially preventable cases. To demonstrate the danger you would have to give the chemical to tens of thousands of mice—an experiment costing tens of millions of dollars. The alternative is to give megadoses to many fewer animals and see if a significant portion of them are affected. The problem, Ames said, was that big concentrations of any foreign substance can throw an animal into physical turmoil. Sensing the damage to its tissues, the body reacts as though it has been wounded, unleashing the healing process. That involves the acceleration of mitosis—rapidly generating new cells to replace damaged ones. With so much DNA being duplicated, the odds of random mutations would be higher and so would the possibility of acquiring one of the deadly combinations. In technical terms, mitogenesis increases mutagenesis.

  Toxicologists defended the tests as a reasonably good compromise. And like Doll and Peto, Ames was condemned by his harsher critics for giving comfort to polluters and diverting attention from a genuine problem. Perhaps environmental poisons are collecting in the human bloodstream—barely noticed but still adding incrementally to the background cancer rate. A recent report by a White House advisory group suggested that animal tests are actually understating carcinogenicity—the opposite of what Ames has long contended. The tests are generally done on adolescent rodents that are sacrificed when the experiment is over. That would miss the effects of prenatal and childhood exposures as well as late-developing tumors. The alternative would be to administer chemicals to pregnant animals and follow the health of their babies as they grow into adults and die of their own accord. Also overlooked would be synergistic interactions. It has been estimated that more than eighty thousand novel substances have been introduced into the world in modern times. The number of combinations is endless. Only a small fraction of new compounds are tested—after they have already been suspected of causing cancer. Taking these factors into consideration, the panel gravely concluded that the number of cancer cases associated with industrial carcinogens “has been grossly underestimated.”

  While many scientists criticized the report for seriously exaggerating the threat of synthetic chemicals and giving unwarranted credence to a maverick view, few would disagree that toxicology testing needs to be improved. The National Academy of Sciences has described how advances in cellular biology and computer science are opening the way to rapid high-throughput assays allowing many more chemicals and combinations of chemicals to be analyzed. Instead of animals, the tests can be done on cells kept alive in laboratory dishes. The hope is that new carcinogens will be identified quickly and measures taken to reduce their prevalence. If all that should come to pass then cancer rates might be lowered further. That can only be good. But it is hard to make the case that the effect would be very large.

  As the years have passed, no epidemic has appeared. Adjusted for the aging of the population, the statistics amassed by SEER show that death rates from cancer did rise gradually by half a percentage point a year from 1975 to 1984—smoking no doubt was a factor—and at a slower pace until 1991, but then they began decreasing modestly and have been doing so ever since. Incidence rates tell a similar story, though the picture is a bit more complex. Like death rates they gradually rose from 1975 until the early 1990s with a burst of newly reported cases from 1989 to 1992, when the rate increased by 2.8 percent a year. The biggest driver for the spike appears to have been more assiduous screening for two of the most common cancers. The number of cases of prostate cancer that were detected shot up by 16.4 percent per year before sharply dropping and breast cancer by 4.0 percent. Then incidence rates, like death rates, began their slow decline.

  Every year when the National Cancer Institute publishes the “Report to the Nation on the Status of Cancer,” the story has been the same. Evidence that a large percentage of cases can be attributed to lifestyle has also prevailed. Opinion continues to vary on just which elements are the most important, with specific foods—how much red and processed meat is bad, how many fruits and vegetables are good—giving way to the suspicion that lack of exercise and excess weight are far more to blame. A twenty-fiv
e-year retrospective on “The Causes of Cancer” still attributed 30 percent of cancer to tobacco. Obesity and inactivity were believed to account for 20 percent, diet for 10 to 25 percent, alcohol for 4 percent, and viruses for 3 percent. A study by the World Health Organization’s International Agency for Research on Cancer found comparable numbers in France. Far down on the lists are occupational exposure and pollutants. Other studies have shown similar proportions in the United Kingdom and other industrialized countries.

  Throughout all of this, neighborhood cancer clusters, like the ones I’d read about in Los Alamos and on Long Island and saw fictionalized in Erin Brockovich, continue to be reported. But in almost every instance they turn out to be statistical illusions, more examples of the Texas sharpshooter effect. Of those that do not, only a rare few have been associated with an environmental contaminant. Over the decades unusual occurrences of cancer among workers have led to the identification of some carcinogens—the link between mesothelioma and asbestos, for example, and between bladder cancer and aromatic amines (substances also found in cigarette smoke). But even occupational clusters are uncommon.

  As the rest of the world develops, the same patterns are appearing as those in the West. Poorer countries tend to be dominated at first by cancers that spread through sexual intercourse and overcrowding—those induced by viruses. There is human papillomavirus and cervical cancer, hepatitis B and C and liver cancer, Helicobacter pylori and stomach cancer. With better hygiene and the growing use of Pap smears (and more recently HPV vaccine), cervical cancer may begin to recede. But then new cancers will arise to take its place. As women choose to have fewer children and their better nourished daughters begin menstruating at an earlier age, there may be more estrogen-driven cancers of the uterus and breasts. Education, vaccines, better sanitation—these also push down cancers of the liver and stomach, but at the same time colorectal cancer increases as more people move from the fields to the cities and become slothful. They go from being undernourished to overnourished with all the nutritional imbalances that can come with a modern diet. The cancers of poverty give way to the cancers of affluence. Prostate cancer, a disease of old men, becomes a problem when life expectancy rises into the seventies and eighties. Lung cancer increases as the cigarette companies migrate to less discriminating markets. Industrialization brings with it new dangers of occupational exposure.