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”—though not for the reasons so many people had come to believe. 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 involved “reproductive and sexual behavior,” which included the hormonal effects of delaying or forgoing the bearing of children as well as 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 sunlight and the 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. 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. 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. 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 either incidence or mortality.
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.
A 25-year retrospective on “The Causes of Cancer” still attributed 30 percent of cancer to tobacco. Obesity and lack of exercise were believed to account for 20 percent, diet for 10 to 25 percent, alcohol for 4 percent, and viruses for 3 percent. Far down on the lists are occupational exposure and pollutants.
Throughout all of this, neighborhood cancer clusters, like the one fictionalized in Erin Brockovich, continue to be reported. But in almost every instance they turn out to be statistical illusions. 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.
One day, trying to absorb all of this, I holed up in my office and began unpacking the most recent SEER statistics. Concentrating on overall cancer rates can smear over some interesting details, and I wondered what might be lurking underneath. The prime mover in driving down the numbers has been a decline or leveling off in what are by far the most common cancers—cancer of the prostate in men, cancer of the breast in women, and lung and colorectal cancer in both women and men. At the same time, the cancers that appear to be rising—melanoma, for example, and cancer of the pancreas, liver, kidney, and thyroid—are among the rarest. The annual incidence of pancreatic cancer is 12.1 cases per 100,000, compared with 62.6 cases for lung and bronchial. Year by year the figures fluctuate ever so slightly. With numbers so low it can be difficult to tell if the increases are real or illusory—artifacts created by better reporting and early detection.
Every cancer tells a different story. For many years lung cancer declined among men because of the delayed effects of giving up cigarettes. Women started smoking later in the century and so their rates continued to climb. Only recently have they taken a downward turn. A spike in breast cancer in the last quarter of the 20th century—including the tiny, slow-growing “stage zero” tumors that some doctors don’t think should be classified as cancer—may be explained both by better diagnosis and earlier onset of puberty. (Each menstrual cycle adds incrementally to cancer risk.) The recent improvement in the numbers may be partly because of a drop in the use of hormone replacement therapy during menopause. Rising rates of melanoma, which began long before the discovery of the ozone hole, is often attributed to the popularity of sunbathing, tanning salons, and skimpier clothing that protects less flesh from ultraviolet rays. Another reason may be international travel. People from northern climes with lighter skin are now more likely to spend time in sunnier places. What may appear to be a climb in childhood malignancies, the National Cancer Institute suggests, is probably because of better imaging technologies and the reclassification of some benign tumors as malignant. Childhood obesity may conceivably be involved.
How much easier cancer would be if it were obviously driven by chemical contaminants. Instead there is a muddle of many little influences. High among them is entropy—the natural tendency of the world toward disorder. Of the multiple mutations it takes to start a cancer there is no way to know which was caused by what. Or, in the case of spontaneous mutations—DNA copying errors—if there was a cause at all.
I imagined an army of clones, genetically identical, going through life under the same conditions in the same geographic locales. They would eat the same foods, engage in the same behaviors, and some would die of cancer by the time they were 50 or 60 while others would succumb decades later to something else. As Doll and Peto put it, “Nature and nurture affect the probability that each individual will develop cancer.” But it is luck that determines which of us really do.
Excerpted from The Cancer Chronicles by George Johnson. Copyright (c) 2013 by George Johnson. Excerpted by permission of Knopf, a division of Random House LLC. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.