Siddhartha Mukherjee - The Emperor of All Maladies: A Biography of Cancer

by Siddhartha Mukherjee

  But she had made it; a charring, private war had just ended. In acute leukemia, the passage of five years without a relapse is nearly synonymous with a cure. I handed her the azaleas and she stood looking at them speechlessly, almost numb to the enormity of her victory. Once, earlier this year, preoccupied with clinical work, I had waited two days before calling her about a negative bone marrow biopsy. She had heard from a nurse that the results were in, and my delay had sent her into a terrifying spiral of depression: in twenty-four hours she had convinced herself that the leukemia had crept back and my hesitation was a signal of impending doom.

  Oncologists and their patients are bound, it seems, by an intense subatomic force. So, albeit in a much smaller sense, this was a victory for me as well. I sat at Carla's table and watched her pour a glass of water for herself, unpurified and straight from the sink. She glowed radiantly, her eyes half-closed, as if the compressed autobiography of the last five years were flashing through a private and internal cinema screen. Her children played with their Scottish terrier in the next room, blissfully oblivious of the landmark date that had just passed for their mother. All of this was for the best. "The purpose of my book," Susan Sontag concluded in Illness as Metaphor, "was to calm the imagination, not to incite it." So it was with my visit. Its purpose was to declare her illness over, to normalize her life--to sever the force that had locked us together for five years.

  I asked Carla how she thought she had survived her nightmare. The drive to her house from the hospital that morning had taken me an hour and a half through a boil of heavy traffic. How had she managed, through the long days of that dismal summer, to drive to the hospital, wait in the room for hours as her blood tests were run, and then, told that her blood counts were too low for her to be given chemotherapy safely, turn back and return the next day for the same pattern to be repeated?

  "There was no choice," she said, motioning almost unconsciously to the room where her children were playing. "My friends often asked me whether I felt as if my life was somehow made abnormal by my disease. I would tell them the same thing: for someone who is sick, this is their new normal."

  Until 2003, scientists knew that the principal distinction between the "normalcy" of a cell and the "abnormalcy" of a cancer cell lay in the accumulation of genetic mutations--ras, myc, Rb, neu, and so forth--that unleashed the hallmark behaviors of cancer cells. But this description of cancer was incomplete. It provoked an inevitable question: how many such mutations does a real cancer possess in total? Individual oncogenes and tumor suppressors had been isolated, but what was the comprehensive set of such mutated genes that exists in any true human cancer?

  The Human Genome Project, the full sequence of the normal human genome, was completed in 2003. In its wake comes a far less publicized but vastly more complex project: fully sequencing the genomes of several human cancer cells. Once completed, this effort, called the Cancer Genome Atlas, will dwarf the Human Genome Project in its scope. The sequencing effort involves dozens of teams of researchers across the world. The initial list of cancers to be sequenced includes brain, lung, pancreatic, and ovarian cancer. The Human Genome Project will provide the normal genome, against which cancer's abnormal genome can be juxtaposed and contrasted.

  The result, as Francis Collins, the leader of the Human Genome Project describes it, will be a "colossal atlas" of cancer--a compendium of every gene mutated in the most common forms of cancer: "When applied to the 50 most common types of cancer, this effort could ultimately prove to be the equivalent of more than 10,000 Human Genome Projects in terms of the sheer volume of DNA to be sequenced. The dream must therefore be matched with an ambitious but realistic assessment of the emerging scientific opportunities for waging a smarter war." The only metaphor that can appropriately describe this project is geological. Rather than understand cancer gene by gene, the Cancer Genome Atlas will chart the entire territory of cancer: by sequencing the entire genome of several tumor types, every single mutated gene will be identified. It will represent the beginnings of the comprehensive "map" so hauntingly presaged by Maggie Jencks in her last essay.

  Two teams have forged ahead in their efforts to sequence the cancer genome. One, called the Cancer Genome Atlas consortium, has multiple interconnected teams spanning several labs in several nations. The second is Bert Vogelstein's group at Johns Hopkins, which has assembled its own cancer genome sequencing facility, raised private funding for the effort, and raced ahead to sequence the genomes of breast, colon, and pancreatic tumors. In 2006, the Vogelstein team revealed the first landmark sequencing effort by analyzing thirteen thousand genes in eleven breast and colon cancers. (Although the human genome contains about twenty thousand genes in total, Vogelstein's team initially had tools to assess only thirteen thousand.) In 2008, both Vogelstein's group and the Cancer Genome Atlas consortium extended this effort by sequencing hundreds of genes of several dozen specimens of brain tumors. As of 2009, the genomes of ovarian cancer, pancreatic cancer, melanoma, lung cancer, and several forms of leukemia have been sequenced, revealing the full catalog of mutations in each tumor type.

  Perhaps no one has studied the emerging cancer genome as meticulously or as devotionally as Bert Vogelstein. A wry, lively, irreverent man in blue jeans and a rumpled blazer, Vogelstein recently began a lecture on the cancer genome in a packed auditorium at Mass General Hospital by attempting to distill the enormous array of discoveries in a few slides. Vogelstein's challenge was that of the landscape artist: How does one convey the gestalt of a territory (in this case, the "territory" of a genome) in a few broad strokes of a brush? How can a picture describe the essence of a place?

  Vogelstein's answer to these questions borrows beautifully from an insight long familiar to classical landscape artists: negative space can be used to convey expanse, while positive space conveys detail. To view the landscape of the cancer genome panoramically, Vogelstein splayed out the entire human genome as if it were a piece of thread zigzagging across a square sheet of paper. (Science keeps eddying into its past: the word mitosis--Greek for "thread"--is resonant here again.) In Vogelstein's diagram, the first gene on chromosome one of the human genome occupies the top left corner of the sheet of paper, the second gene is below it, and so forth, zigzagging through the page, until the last gene of chromosome twenty-three occupies the bottom right corner of the page. This is the normal, unmutated human genome stretched out in its enormity--the "background" out of which cancer arises.

  Against the background of this negative space, Vogelstein placed mutations. Every time a gene mutation was encountered in a cancer, the mutated gene was demarcated as a dot on the sheet. As the frequency of mutations in any given gene increased, the dots grew in height into ridges and hills and then mountains. The most commonly mutated genes in breast cancer samples were thus represented by towering peaks, while genes rarely mutated were denoted by small hills or flat dots.

  Viewed thus, the cancer genome is at first glance a depressing place. Mutations litter the chromosomes. In individual specimens of breast and colon cancer, between fifty to eighty genes are mutated; in pancreatic cancers, about fifty to sixty. Even brain cancers, which often develop at earlier ages and hence may be expected to accumulate fewer mutations, possess about forty to fifty mutated genes.

  Only a few cancers are notable exceptions to this rule, possessing relatively few mutations across the genome. One of these is an old culprit, acute lymphoblastic leukemia: only five or ten genetic alterations cross its otherwise pristine genomic landscape.* Indeed, the relative paucity of genetic aberrancy in this leukemia may be one reason that this tumor is so easily felled by cytotoxic chemotherapy. Scientists speculate that genetically simple tumors (i.e., those carrying few mutations) might inherently be more susceptible to drugs, and thus intrinsically more curable. If so, the strange discrepancy between the success of high-dose chemotherapy in curing leukemia and its failure to cure most other cancers has a deep biological explanation. The search for a "universal cure" for can
cer was predicated on a tumor that, genetically speaking, is far from universal.

  In contrast to leukemia, the genomes of the more common forms of cancer, Vogelstein finds, are filled with genetic bedlam--mutations piled upon mutations upon mutations. In one breast cancer sample from a forty-three-year-old woman, 127 genes were mutated--nearly one in every two hundred genes in the human genome. Even within a single type of tumor, the heterogeneity of mutations is daunting. If one compares two breast cancer specimens, the set of mutated genes is far from identical. "In the end," as Vogelstein put it, "cancer genome sequencing validates a hundred years of clinical observations. Every patient's cancer is unique because every cancer genome is unique. Physiological heterogeneity is genetic heterogeneity." Normal cells are identically normal; malignant cells become unhappily malignant in unique ways.

  Yet, characteristically, where others see only daunting chaos in the littered genetic landscape, Vogelstein sees patterns coalescing out of the mess. Mutations in the cancer genome, he believes, come in two forms. Some are passive. As cancer cells divide, they accumulate mutations due to accidents in the copying of DNA, but these mutations have no impact on the biology of cancer. They stick to the genome and are passively carried along as the cell divides, identifiable but inconsequential. These are "bystander" mutations or "passenger" mutations. ("They hop along for the ride," as Vogelstein put it.)

  Other mutations are not passive players. Unlike the passenger mutations, these altered genes directly goad the growth and the biological behavior of cancer cells. These are "driver" mutations, mutations that play a crucial role in the biology of a cancer cell.

  Every cancer cell possesses some set of driver and passenger mutations. In the breast cancer sample from the forty-three-year-old woman with 127 mutations, only about ten might directly be contributing to the actual growth and survival of her tumor, while the rest may have been acquired due to gene-copying errors in cancer cells. But while functionally different, these two forms of mutations cannot easily be distinguished. Scientists can identify some driver genes that directly goad cancer's growth using the cancer genome. Since passenger mutations occur randomly, they are randomly spread throughout the genome. Driver mutations, on the other hand, strike key oncogenes and tumor suppressors, and only a limited number of such genes exist in the genome. These mutations--in genes such as ras, myc, and Rb--recur in sample upon sample. They stand out as tall mountains in Vogelstein's map, while passenger mutations are typically represented by the valleys. But when a mutation occurs in a previously unknown gene, it is impossible to predict whether that mutation is consequential or inconsequential--driver or passenger, barnacle or engine.

  The "mountains" in the cancer genome--i.e., genes most frequently mutated in a particular form of cancer--have another property. They can be organized into key cancer pathways. In a recent series of studies, Vogelstein's team at Hopkins reanalyzed the mutations present in the cancer genome using yet another strategy. Rather than focusing on individual genes mutated in cancers, they enumerated the number of pathways mutated in cancer cells. Each time a gene was mutated in any component of the Ras-Mek-Erk pathway, it was classified as a "Ras pathway" mutation. Similarly, if a cell carried a mutation in any component of the Rb signaling pathway, it was classified as "Rb pathway mutant," and so forth, until all driver mutations had been organized into pathways.

  How many pathways are typically dysregulated in a cancer cell? Typically, Vogelstein found, between eleven and fifteen, with an average of thirteen. The mutational complexity on a gene-by-gene level was still enormous. Any one tumor bore scores of mutations pockmarked throughout the genome. But the same core pathways were characteristically dysregulated in any tumor type, even if the specific genes responsible for each broken pathway differed from one tumor to the next. Ras may be activated in one sample of bladder cancer; Mek in another; Erk in the third--but in each case, some vital piece of the Ras-Mek-Erk cascade was dysregulated.

  The bedlam of the cancer genome, in short, is deceptive. If one listens closely, there are organizational principles. The language of cancer is grammatical, methodical, and even--I hesitate to write--quite beautiful. Genes talk to genes and pathways to pathways in perfect pitch, producing a familiar yet foreign music that rolls faster and faster into a lethal rhythm. Underneath what might seem like overwhelming diversity is a deep genetic unity. Cancers that look vastly unlike each other superficially often have the same or similar pathways unhinged. "Cancer," as one scientist recently put it, "really is a pathway disease."

  This is either very good news or very bad news. The cancer pessimist looks at the ominous number thirteen and finds himself disheartened. The dysregulation of eleven to fifteen core pathways poses an enormous challenge for cancer therapeutics. Will oncologists need thirteen independent drugs to attack thirteen independent pathways to "normalize" a cancer cell? Given the slipperiness of cancer cells, when a cell becomes resistant to one combination of thirteen drugs, will we need an additional thirteen?

  The cancer optimist, however, argues that thirteen is a finite number. It is a relief: until Vogelstein identified these core pathways, the mutational complexity of cancers seemed nearly infinite. In fact, the hierarchical organization of genes into pathways in any given tumor type suggests that even deeper hierarchies might exist. Perhaps not all thirteen need to be targeted to attack complex cancers such as breast or pancreatic cancer. Perhaps some of the core pathways may be particularly responsive to therapy. The best example of this might be Barbara Bradfield's tumor, a cancer so hypnotically addicted to Her-2 that targeting this key oncogene melted the tumor away and forced a decades-long remission.

  Gene by gene, and now pathway by pathway, we have an extraordinary glimpse into the biology of cancer. The complete maps of mutations in many tumor types (with their hills, valleys, and mountains) will soon be complete, and the core pathways that are mutated fully defined. But as the old proverb runs, there are mountains beyond mountains. Once the mutations have been identified, the mutant genes will need to be assigned functions in cellular physiology. We will need to move through a renewed cycle of knowledge that recapitulates a past cycle--from anatomy to physiology to therapeutics. The sequencing of the cancer genome represents the genetic anatomy of cancer. And just as Virchow made the crucial leap from Vesalian anatomy to the physiology of cancer in the nineteenth century, science must make a leap from the molecular anatomy to the molecular physiology of cancer. We will soon know what the mutant genes are. The real challenge is to understand what the mutant genes do.

  This seminal transition from descriptive biology to the functional biology of cancer will provoke three new directions for cancer medicine.

  The first is a direction for cancer therapeutics. Once the crucial driver mutations in any given cancer have been identified, we will need to launch a hunt for targeted therapies against these genes. This is not an entirely fantastical hope: targeted inhibitors of some of the core thirteen pathways mutated in many cancers have already entered the clinical realm. As individual drugs, some of these inhibitors have thus far had only moderate response rates. The challenge now is to determine which combinations of such drugs might inhibit cancer growth without killing normal cells.

  In a piece published in the New York Times in the summer of 2009, James Watson, the codiscoverer of the structure of DNA, made a remarkable turnabout in opinion. Testifying before Congress in 1969, Watson had lambasted the War on Cancer as ludicrously premature. Forty years later, he was far less critical: "We shall soon know all the genetic changes that underlie the major cancers that plague us. We already know most, if not all, of the major pathways through which cancer-inducing signals move through cells. Some 20 signal-blocking drugs are now in clinical testing after first being shown to block cancer in mice. A few, such as Herceptin and Tarceva, have Food and Drug Administration approval and are in widespread use."

  The second new direction is for cancer prevention. To date, cancer prevention has relied on two dispa
rate and polarized methodologies to try to identify preventable carcinogens. There have been intensive, often massive, human studies that have connected a particular form of cancer with a risk factor, such as Doll and Hill's study identifying smoking as a risk factor for lung cancer. And there have been laboratory studies to identify carcinogens based on their ability to cause mutations in bacteria or incite precancer in animals and humans, such as Bruce Ames's experiment to capture chemical mutagens, or Marshall and Warren's identification of H. pylori as a cause for stomach cancer.

  But important preventable carcinogens might escape detection by either strategy. Subtle risk factors for cancer require enormous population studies; the subtler the effect, the larger the population needed. Such vast, unwieldy, and methodologically challenging studies are difficult to fund and launch. Conversely, several important cancer-inciting agents are not easily captured by laboratory experiments. As Evarts Graham discovered to his dismay, even tobacco smoke, the most common human carcinogen, does not easily induce lung cancer in mice. Bruce Ames's bacterial test does not register asbestos as a mutagen.*

  Two recent controversies have starkly highlighted such blind spots in epidemiology. In 2000, the so-called Million Women Study in the United Kingdom identified estrogen and progesterone, prescribed in hormone-replacement therapy to women to ease menopausal symptoms, as major risk factors for the incidence and fatality from estrogen-positive breast cancer. Scientifically speaking, this is an embarrassment. Estrogen is not identified as a mutagen in Bruce Ames's test; nor does it cause cancer in animals at low doses. But the two hormones have been known as pathological activators of the ER-positive subtype of breast cancer since the 1960s. Beatson's surgery and tamoxifen induce remissions in breast cancer by blocking estrogen, and so it stands to reason that exogenous estrogen might incite breast cancer. A more integrated approach to cancer prevention, incorporating the prior insights of cancer biology, might have predicted this cancer-inducing activity, preempted the need for a million-person association study, and potentially saved the lives of thousands of women.