The Emperor of All Maladies: A Biography of Cancer (52 page)

*
No relation of Sidney Farber’s.

You’re under a lot of stress
, my dear. You haven’t really got anything wrong with yourself. We’ll give you an antidepressant.

—Barry Marshall on the treatment of women
with gastritis, a precancerous lesion,
in the 1960s

The classification of tobacco smoke as a potent carcinogen—and the slow avalanche of forces unleashed to regulate cigarettes in the 1980s—is rightfully counted as one of cancer prevention’s seminal victories. But it equally highlighted an important lacuna in cancer epidemiology. Statistical methods to identify risk factors for cancer are, by their very nature, descriptive rather than mechanistic: they describe correlations, not causes. And they rely on a certain degree of foreknowledge. To run a classical “case-control” trial to identify an unknown risk factor, paradoxically, an epidemiologist must know the questions to ask. Even Doll and Hill, in devising their classic case-control and prospective studies, had relied on decades of prior knowledge—centuries, if one counts John Hill’s pamphlet—about the possible link between tobacco and cancer.

This does not diminish the incredible power of the case-control method.
In the early 1970s, for instance, a series of studies
definitively identified the risk factor for a rare and fatal form of lung cancer called mesothelioma. When mesothelioma “cases” were compared to “controls,” this cancer appeared to cluster densely in certain professions: insulation installers, firefighters, shipyard workers, heating equipment handlers, and chrysolite miners. As with Pott and scrotal cancer, the statistical confluence of a rare profession and a rare tumor swiftly identified the causal agent in this cancer: exposure to asbestos. Tort litigation and federal oversight soon followed, precipitating a reduction in the occupational exposure to asbestos that, in turn, reduced the risk of mesothelioma.

In 1971, yet another such study identified
an even more unusual car
cinogen, a synthetic hormonal medicine called diethylstilbestrol (DES). DES was widely prescribed to pregnant women in the 1950s to prevent premature deliveries (although it was of only questionable benefit in this regard). A generation later, when women with vaginal and uterine cancer were questioned about their exposures to estrogens, a peculiar pattern emerged: the women had not been exposed to the chemical directly, but their
mothers
had been. The carcinogen had skipped a generation. It had caused cancers not in the DES-treated women, but in their daughters exposed to the drug in utero.

But what if the behavior or exposure responsible for the cancer is completely unknown? What if one did not even know enough about the natural history of mesothelioma, or the link between estrogen and vaginal cancer, to ask those afflicted about their occupational history, or their exposure to asbestos and estrogen? Could carcinogens be discovered a priori—not by the statistical analysis of cancer-afflicted populations, but by virtue of some intrinsic property of all carcinogens?

In the late 1960s, a bacteriologist named Bruce Ames
at Berkeley, working on an unrelated problem, stumbled on a test for chemical carcinogens. Ames was studying mutations in
Salmonella
, a bacterial genus.
Salmonella
, like any bacteria, possesses genes that allow it to grow under certain conditions—a gene to “digest” galactose, for instance, is essential for a bacterium to survive on a petri dish where the only sugar source is galactose.

Ames observed that mutations in these essential genes could enable or disable the growth of bacteria on a petri dish. A strain of
Salmonella
normally unable to grow on galactose, say, could
acquire
a gene mutation that enabled this growth. Once growth-enabled, a single bacterium would form a minuscule colony on a petri dish. By counting the number of growth-enabled colonies formed, Ames could quantify the mutation rate in any experiment. Bacteria exposed to a certain substance might produce six such colonies, while bacteria exposed to another substance might produce sixty. This second substance, in other words, had a tenfold capacity to initiate changes in genes—or a tenfold rate of mutation.

Ames could now test thousands of chemicals to create a catalog of
chemicals that increased the mutation rate—mutagens. And as he populated his catalog, he made a seminal observation:
chemicals that scored as mutagens in his test tended to be carcinogens as well
. Dye derivatives, known to be potent human carcinogens, scored floridly, causing hundreds of colonies of bacteria.
So did X-rays, benzene compounds, and nitrosoguanidine
derivatives—all known to cause cancers in rats and mice. In the tradition of all good tests, Ames’s test transformed the unobservable and immeasurable into the observable and measurable. The invisible X-rays that had killed the Radium girls in the 1920s could now be “seen” as revertant colonies on a petri dish.

Ames’s test was far from perfect.
Not every known
carcinogen scored in the test: neither DES nor asbestos sprinkled on the disabled
Salmonella
caused significant numbers of mutant bacteria. (In contrast, chemical constituents of tobacco smoke did cause mutation in the bacteria, as noted by several cigarette manufacturers who ran the test and, finding it disconcertingly positive, quickly buried the results.) But despite its shortcomings, the Ames test provided an important link between a purely descriptive approach toward cancer prevention and a mechanistic approach. Carcinogens, Ames suggested, had a common, distinctive functional property: they altered genes. Ames could not fathom the deeper reason behind this observation: why was the capacity to cause mutations linked to the ability to induce cancer? But he had demonstrated that carcinogens could be found experimentally—not retrospectively (by investigating cases and controls in human subjects) but by
prospectively
identifying chemicals that could cause mutations in a rather simple and elegant biological assay.

Chemicals, it turned out, were not the only carcinogens; nor was Ames’s test the only method to find such agents. In the late 1960s, Baruch Blumberg, a biologist working in Philadelphia, discovered that a chronic, smoldering inflammation caused by a human hepatitis virus could also cause cancer.

A biochemistry student at Oxford
in the 1950s,
Blumberg
had become interested in genetic anthropology, the study of genetic variations in human populations. Traditional biological anthropology in the 1950s mainly involved collecting, measuring, and categorizing human ana
tomical specimens. Blumberg wanted to collect, measure, and categorize human
genes
—and he wanted to link genetic variations in humans to the susceptibility for diseases.

The problem, as Blumberg soon discovered, was the lack of human genes to be measured or categorized. Bacterial genetics was still in its infancy in the 1950s—even the structure of DNA and the nature of the genes was still largely undiscovered—and human genes had not even been seen or analyzed. The only tangible hint of variations in human genetics came from an incidental observation. Proteins in the blood, called blood antigens, varied between individuals and were inherited in families, thus implying a genetic source for this variation. These blood proteins could be measured and compared across populations using relatively simple tests.

Blumberg began to scour far-flung places
in the world for blood, drawing tubes of serum from Fulani tribesmen in Africa one month and Basque shepherds the next.
In 1964, after a brief tenure at the NIH
, he moved to the Institute for Cancer Research in Philadelphia (later renamed the Fox Chase Cancer Center) to systematically organize the variant blood antigens that he had cataloged, hoping to link them to human diseases. It was a curiously inverted approach, like scouring a dictionary for a word and then looking for a crossword puzzle into which that word might fit.

One blood antigen that intrigued him
was present in several Australian aboriginal subjects and found frequently in Asian and African populations, but was typically absent in Europeans and Americans. Suspecting that this antigen was the fingerprint of an ancient genetic factor inherited in families, Blumberg called it the
Australia antigen
or
Au
for short.

In 1966, Blumberg’s lab set out to characterize
the aboriginal antigen in greater detail. He soon noted an odd correlation: individuals carrying the
Au
antigen often suffered from chronic hepatitis, an inflammation of the liver. These inflamed livers, studied pathologically, showed signs of chronic cycles of injury and repair—death of cells in some pockets and compensatory attempts to repair and regenerate liver cells in others, resulting in scarred, shrunken, and burnt-out livers, a condition termed chronic cirrhosis.

A link between an ancient antigen and cirrhosis suggested a genetic susceptibility for liver disease—a theory that would have sent Blumberg off on a long and largely fruitless tangent. But a chance incident overturned that theory and radically changed the course of Blumberg’s studies. The lab had been following a young patient at a mental-disability clinic in New Jersey. Initially, the man had tested negative for the
Au
antigen. But
during one of the serial blood draws in the summer of 1966, his serum suddenly converted from “
Au
negative” to “
Au
positive.” When his liver function was measured, an acute, fulminant hepatitis was discovered.

But how could an “intrinsic” gene cause sudden seroconversion and hepatitis? Genes, after all, do not typically flicker on and off at will. Blumberg’s beautiful theory about genetic variation had been slain by an ugly fact.
Au
, he realized, could not mark an inherent variation in a human gene. In fact,
Au
was soon found to be neither a human protein nor a blood antigen.
Au
was a piece of a viral protein floating in the blood, the sign of an infection. The New Jersey man had been infected by this microbe and thus converted from
Au
negative to positive.

Blumberg now raced to isolate the organism responsible for the infection. By the early 1970s, working with a team of collaborators, his lab had purified particles of a new virus, which he called hepatitis B virus, or HBV. The virus was structurally simple—“
roughly circular . . . about forty-two nanometers
in diameter, one of the smallest DNA viruses that infect humans”—but the simple structure belied extraordinarily complex behavior. In humans, HBV infection caused a broad spectrum of diseases, ranging from asymptomatic infection to acute hepatitis to chronic cirrhosis in the liver.

The identification of a new human virus set off a storm of activity for epidemiologists.
By 1969, Japanese researchers
(and subsequently Blumberg’s group) had learned that the virus was transmitted from one individual to another through blood transfusions. By screening blood before transfusion—using the now familiar
Au
antigen as one of the early biomarkers in serum—the blood-borne infection could be blocked, thereby reducing the risk of hepatitis B.

But another illness soon stood out
as linked to HBV: a fatal, insidious form of liver cancer endemic in parts of Asia and Africa that arose out of scarred, ashen livers often decades after chronic viral infection. When cases of hepatocellular cancer were compared to controls using classical statistical methods, chronic infection with HBV, and the associated cycle of injury and repair in liver cells, stood out as a clear risk factor—at about five- to tenfold the risk for uninfected controls. HBV, then, was a carcinogen—although a live carcinogen, capable of being transmitted from one host to another.