In the winter of 1999, a 49-year-old psychologist was struck by
nausea —a queasiness so sudden and strong that it seemed as if it had
been released from a catapult.
More puzzled by her symptoms than alarmed — this nausea came without any
aura of pain — she saw her internist. She was given a diagnosis of
gastroenteritis and sent home to bed rest and Gatorade.
But the nausea persisted, and then additional symptoms appeared out of
nowhere. Ghostly fevers came and went. She felt perpetually full, as if
she had just finished a large meal. Three weeks later, she returned to
the hospital, demanding additional tests. This time, a CT scan revealed a nine-centimeter solid mass pushing into her stomach. Once biopsied, the mass was revealed to be a tumor, with oblong, spindle-shaped cells dividing rapidly. It was characterized as a rare kind of cancer called a gastrointestinal stromal tumor, or GIST.
A surgical cure was impossible: her tumor had metastasized to her liver,
lymph nodes and spleen. Her doctors halfheartedly tried some chemotherapy,
but nothing worked. “I signed my letters, paid my bills and made my
will,” the patient recalled. “I was told to go home to die.”
In June, several months after her diagnosis, she stumbled into a virtual
community of co-sufferers — GIST patients who spoke to one another
online through a Listserv. In 2001, word of a novel drug called Gleevec
began to spread like wildfire through this community. Gleevec was the
exemplar of a brand-new kind of cancer medicine. Cancer cells are often
driven to divide because of mutations that activate genes crucial to
cell division; Gleevec directly inactivated the mutated gene driving the
growth of her sarcoma, and in early trials was turning out to be
astonishingly effective against GIST.
The psychologist pulled strings to enroll in one of these trials. She
was, by nature, effortlessly persuasive, and her illness had made her
bold. She enrolled in a Gleevec trial at a teaching hospital. A month
later, her tumors began to recede at an astonishing rate. Her energy reappeared; her nausea vanished. She was resurrected from the dead. Her recovery was a medical miracle, emblematic of a new direction in
cancer treatment. Medicine seemed to be catching up on cancer. Even if
no cure was in sight, there would be a new generation of drugs to
control cancer, and another when the first failed. Then, just short of
the third anniversary of her unexpected recovery, cancer cells suddenly
began multiplying again. The dormant lumps sprouted back. The nausea
returned. Malignant fluid poured into the cisterns of her abdomen.
Resourceful as always, she turned once more to the online community of
GIST patients. She discovered that there were other drugs —
second-generation analogues of Gleevec — in trial in other cities. Later
that year, she enrolled in one such trial in Boston, where I was
completing my clinical training in cancer medicine.
The response was again striking. The masses in her liver and stomach
shrank almost immediately. Her energy flowed back. Resurrected again,
she made plans to return home. But the new drug did not work for long:
within months she relapsed again. By early winter, her cancer was out of
control, growing so fast that she could record its weight, in pounds,
as she stood on the hospital’s scales. Eventually her pain reached a
point when it was impossible for her to walk.
Toward the end of 2003, I met her in her hospital room to try to
reconcile her to her medical condition. As usual, she was ahead of me.
When I started to talk about next steps, she waved her hand and cut me
off. Her goals were now simple, she told me. No more trials. No more
drugs. She realized that her reprieve had finally come to an end. She
wanted to go home, to die the death that she expected in 1999.
The word “relapse” comes from the Latin for “slipping
backward,” or “slipping again.” It signals not just a fall but another
fall, a recurrent sin, a catastrophe that happens again. It carries a
particularly chilling resonance in cancer — for it signals the
reappearance of a disease that had once disappeared. When cancer recurs,
it often does so in treatment-resistant or widely spread form. For many
patients, it is relapse that presages the failure of all treatment. You
may fear cancer, but what cancer patients fear is relapse.
Why does cancer relapse? From one perspective, the answer has to do as much with language, or psychology, as with biology.
Diabetes and heart failure,
both chronic illnesses whose acuity can also wax and wane, are rarely
described in terms of “relapse.” Yet when a cancer disappears on a CT
scan or becomes otherwise undetectable, we genuinely begin to believe
that the disappearance is real, or even permanent, even though
statistical reasoning might suggest the opposite. A resurrection implies
a previous burial. Cancer’s “relapse” thus implies a belief that the
disease was once truly dead.
But what if my patient’s cancer had never actually died, despite its
invisibility on all scans and tests? CT scans, after all, lack the
resolution to detect a single remnant cell. Blood tests for cancer also
have a resolution limit: they detect cancer only when millions of tumor
cells are present in the body. What if her cancer had persisted in a
dormant state during her remissions — effectively frozen but ready to
germinate? Could her case history be viewed through an inverted lens:
not as a series of remissions punctuated by the occasional relapse, but
rather a prolonged relapse, relieved by an occasional remission?
In fact, this view of cancer — as tenaciously persistent and able to
regenerate after apparently disappearing — has come to occupy the very
center of cancer biology. Intriguingly, for some cancers, this
regenerative power appears to be driven by a specific cell type lurking
within the cancer that is capable of dormancy, growth and infinite
regeneration — a cancer “stem cell.”
If such a phoenixlike cell truly exists within cancer, the implication
for cancer therapy will be enormous: this cell might be the ultimate
determinant of relapse. For decades, scientists have wondered if the
efforts to treat certain cancers have stalled because we haven’t yet
found the right kind of drug. But the notion that cancers contain stem
cells might radically redirect our efforts to develop anticancer drugs.
Is it possible that the quest to treat cancer has also stalled because
we haven’t even found the right kind of cell?
Even the earliest theories of cancer’s genesis had to
contend with the regenerative power of this illness. The most enduring
of these theories was promulgated by Galen, the Greek physician who
began practicing among the Romans in A.D. 162. Galen, following earlier
Greek physiologists, proposed that the human body was composed of four
cardinal fluids: blood, phlegm, yellow bile and black bile. Each
possessed a unique color (red, white, yellow and black) and an essential
character, temperature and taste. In normal bodies, these fluids were
kept in a perfect, if somewhat precarious, balance. Illness was the
pathological overabundance or depletion of one or more fluids. Catarrh,
pustules, tuberculotic glands — all boggy, cool and white — were
illnesses of the excess of phlegm.
Jaundice
was obviously an overflow of yellow bile. Heart failure arose from too
much blood. Cancer was linked to the most malevolent and complex of all
fluids — black bile, imagined as an oily, bitter fluid also responsible
for depression (melancholia takes its name from black bile).
Fantastical as it was, Galen’s system nonetheless had one important
virtue: It explained not just cancer’s occurrence but also its
recurrence. Cancer, Galen proposed, was a result of a systemic malignant
state, an internal overdose of black bile. Tumors were the local
outcroppings of a deep-seated bodily dysfunction, an imbalance that
pervaded the entire corpus. The problem with treating cancer with any
form of local therapy, like surgery, was that black bile was everywhere
in the body. Fluids seep back to find their own levels. You could cut a
tumor out, Galen argued, but black bile would flow right back and
regenerate cancer.
Galen’s theory held a potent grip on the imagination of scientists for
centuries — until the invention of the microscope quite literally threw
light on the cancer cell. When 19th-century pathologists trained their
lenses at tumors, they found not black bile in overabundance but cells
in excess — sheet upon sheet of them that had divided with
near-hyperactive frenzy, distorting normal anatomy, breaking boundaries
and invading other tissues. The crucial abnormality of cancer was
unbridled cellular proliferation, cell growth without control.
We now have a vastly enriched understanding of how this runaway growth
begins. Cancer results from alterations to cellular genes. In normal
cells, powerful genetic signals regulate cell division with exquisite
control. Some genes activate cellular proliferation, behaving like
minuscule accelerators of growth. Others inactivate growth, acting like
molecular brakes. Genes tell a limb to grow out of an embryo, for
example, and then instruct the limb to stop growing. A cut prompts the
skin to heal itself, but heaps of skin do not continue to grow in
excess. In a cancer cell, in contrast, the accelerators of growth are
jammed permanently on, the brakes permanently off. The result is a cell
that does not know how to stop growing.
Uncontrolled cell division imbues cancer cells not just with the
capacity to grow but also with a crucial property that often accompanies
growth: the capacity to evolve. Cancer is not merely a glum cellular
copying machine, begetting clone after clone. Every generation of cancer
cells produces cells that in turn bear additional mutations, changes
beyond those already present in the accelerator and brake genes. And
when a selective pressure like chemotherapy is applied to a cancer,
resistant mutants escape that pressure. Just as antibiotics can give rise to resistant strains of bacteria, anticancer drugs can produce resistant cancer cells.
This process — evolution’s slippery hand driving cancer’s adaptation and
survival — provided biologists with an explanation for cancer’s
recurrence after treatment. Relapse occurs because cancer cells that are
genetically resistant to a drug outgrow all the nonresistant cells.
Chemotherapy unleashes a ruthless Darwinian battle in every tumor. A
relapsed cancer is the ultimate survivor of that battle, the direct
descendant of the fittest cell.
And yet this theory seemed incomplete. Some cancers relapse months or
even years after a chemotherapeutic drug has been stopped — a delay that
would make little sense if relapse were simply due to resistance. In
other instances, treating a recurrent cancer with the same drug can lead
to a second remission — an outcome difficult to explain if the
recurring cancer has acquired resistance to the original drug. Could
there be a deeper explanation for cancer’s persistence and regenerative
power beyond simple mutations and resistance?
In 1994, a researcher at the University of Toronto
named John Dick performed a striking experiment that would upend the
received wisdom about cancer relapses. Trained as a stem-cell biologist,
Dick was particularly interested in blood stem cells.
Stem cells, regardless of their origin, are defined by two cardinal
characteristics. The first is hierarchy, or potency. A stem cell is the
originator of the many different cell types in a tissue; it sits, like
the founder of a massive clan, at the tip of a pyramid of growth. The
second is self-renewal: even as stem cells create the cells that make up
a tissue, they must also renew themselves. This dynast doesn’t just
produce a clan; in each generation, it rebirths itself. The perpetual
rebirth of a founding cell yields a virtually inexhaustible supply of
cells in a tissue, a reservoir of growth that can be tapped repeatedly
on demand.
In humans, all circulating blood cells — white cells, red cells and
platelets — arise from a population of blood stem cells exclusively
dedicated to the genesis of blood. In their normal, unperturbed state,
these blood-founding cells hibernate deep in the cavities of the bone
marrow. But when circulating blood cells are killed — by chemotherapy,
say — the stem cells awaken and begin to divide with awe-inspiring
fecundity, generating millions of cells that gradually mature into blood
cells. A defining feature of this process is its regenerative
capacity: in generating blood, the blood stem cells also regenerate
themselves. Each round of blood formation restores their supply. If the
entirety of blood is again depleted, by another round of chemo, it can
be regenerated again and yet again — theoretically, an infinite number
of times — because the stem cells replenish themselves in every cycle.
Blood, in short, is hierarchically organized. Its reservoir of renewal
is concentrated in a rare population of highly potent cells. As long as
these cells exist in the marrow, blood can be regenerated. Eliminate
this reservoir, and the vast organ-system of blood gradually collapses.
Now imagine that cancer is also hierarchically organized — with a secret
cellular reservoir dedicated to its renewal. Typically, cancer is
envisioned as a mass of dividing cells, with no difference between one
cell and its neighbor. But what if some cells in a tumor are dedicated
“founders,” capable of infinite regeneration, while others are limited
in their capacity to divide and unable to continuously generate new
cells? Cancer cells bear mutations that enable rapid growth, but what if
only some cells within a tumor possess
indefinite growth? Such
a model of cancer would still retain the essential pathological
features of the disease — distorted growth, invasiveness, the capacity
to mutate and evolve. Yet the driver of regeneration would be different:
as with blood, only a certain subpopulation of cells in the tumor would
be responsible for a cancer’s regeneration. Might such cells lie at the
root of relapse?
John Dick had an obvious place to begin looking for such
cancer-regenerating cells — in leukemia, or cancers of white blood
cells. Dick implanted human leukemia cells into immune-paralyzed mice
and found that these leukemias could survive and grow in these mice. But
not every leukemia cell could. Dick and his students implanted fewer
and fewer leukemia cells — one million, 100,000, 1,000 and so on — to
determine the smallest number of cells required to cause cancer in a
mouse. The answer was surprising: one needed to implant between a
quarter-million and one million cells to be sure of implanting at least
one cell that could generate leukemia. The rest could not; the other
999,999 cells, in short, had evidently grown out of that single cell —
but were themselves incapable of regenerating the cancer.
When Dick’s team focused on defining the characteristics of this
one-in-a-million cell, there was another surprise. All cells express
subsets of proteins on their cell surface that correspond to their
identity like tiny bar codes. The bar codes on the surface of the
leukemia-generating cell bore a familiar mark: of all cell types found
in blood, it most closely resembled the blood stem cell. And when Dick
transplanted this cell from one mouse to the next, he found that he
could generate and regenerate the leukemia — just as a blood stem cell
would generate blood cells.
Dick’s leukemia-forming cell was, in effect, the normal stem cell’s
malignant doppelgänger. It possessed the blood stem cell’s incredible
regenerative ability — but unlike a normal stem cell, it could not stop
regenerating, dividing and producing more cells. It, too, was an
inexhaustible reservoir of growth, but of unstoppable growth. Noting the
analogy between this cell and the blood stem cell, Dick called the
one-in-a-million cell the “leukemia stem cell.”
In time, biologists began to see the implication of Dick’s experiment.
If leukemia possessed stem cells, then — much like normal blood — its
regenerative capacity may be contained exclusively within that select
population. And if so, it was this rare stem cell — not the other
999,999 — that had to be attacked by a new generation of drugs.
Traditional chemotherapy, of course, makes no distinction between a
cancer’s stem cells and any other of its cells, between the roots and
the shoots of a tumor. All cells are treated equal — and what is poison
to one growing cell is largely poison to another. Indeed, most forms of
chemotherapy in use today are derived from enormous chemical hunts begun
in the 1970s, decades before the birth of the cancer-stem-cell theory.
Many of these chemicals came into use because of their ability to kill
dividing cancer cells in a petri dish. The fact that most such drugs
turn out to be nearly indiscriminate poisons of cellular growth should
hardly come as a surprise: they were selected to be generic cell
killers.
But if tumors contain dedicated stem cells, then delivering maximal
doses of poisons to kill the bulk of the tumor might achieve one
response — a shrinkage of the tumor — but have no effect on relapse. If
the rare stem cell lurking within a tumor somehow escapes death, then it
will reassert itself and grow again. Cancers will come back like a
garden that has been cleared by hacking at its weeds while leaving the
roots behind.
The publication of John Dick’s paper eventually
produced an avalanche of interest in cancer stem cells. In 2003, another
laboratory, led by Michael Clarke at the University of Michigan,
isolated a rare population of cancer-regenerating cells from human
breast cancers, thereby extending Dick’s model beyond leukemia to a
“solid” tumor. In 2005, a Harvard professor named Martin Nowak used
mathematical modeling to demonstrate that another human leukemia known
as CML
also possesses a rare subpopulation of regenerating cells. In the
winter of 2006, Dick’s lab and an Italian team independently discovered
cancer stem cells in colon cancers. Laboratories around the United
States rushed to extract cancer stem cells from brain, prostate, lung
and pancreatic cancers. Pharmaceutical companies joined the bandwagon,
spending millions, and then tens of millions, on mammoth chemical
searches for drugs that might destroy cancer stem cells. The National Institutes of Health
issued dozens of grant requests to study and isolate cancer stem cells.
The paradox of this moment was not lost on researchers. For decades,
cancer had been imagined as a degenerative disease — an illness caused
by the corruption of genes and cells over time, often a side-effect of
aging. Yet in the search for a new generation of anticancer drugs, it
was to the science of regeneration — to embryos and stem cells — that
the field turned.
In 2005, by the time I finished my training, the cancer-stem-cell model
had acquired an overheated quality. The boil and froth inevitably
brought challenges. In Michigan, a stem-cell biologist named Sean
Morrison returned to John Dick’s original test for stem cells — diluting
and rediluting cells to find the cells that could regenerate a cancer.
Morrison first tested the model in mouse leukemias and confirmed Dick’s
results in human leukemias. He subsequently tried the experiment with
another type of cancer — melanoma,
deadly blue-black cancers that arise in the skin and metastasize often
to the lungs and brain. Others had suggested that only a few cells —
about one in a million — could regenerate the tumor in mice. But when
Morrison tested the melanoma cells’ regenerative capacity by conducting a
variation of Dick’s experiment, he found that some 25 percent of the
cells from a melanoma could grow a tumor in a mouse. If stem cells were
this common in tumors — if one in four cells could grow cancer — then
their very definition might be reduced to semantic oblivion. How could a
tumor have a stem-cell-like “hierarchy” if every cell occupied the
primary spot?
New questions emerged again in May this year at the Wistar Institute in
Philadelphia. A group there was working on melanoma, the cancer that
Morrison studied. As previous studies had, the Wistar study also
identified a subpopulation of self-renewing cells marked by a distinct
bar code within human melanomas. But when these cells were studied more
deeply, they appeared to possess no greater ability to regenerate
cancers in mice than the nonrenewing cells — thus seemingly
disconnecting the link between self-renewal and cancer regeneration.
The Wistar and the Morrison studies are among the many that have begun
to challenge the universality and the reliability of the
cancer-stem-cell model. “Look,” Morrison told me, “this is all going to
become more complicated. Some cancers, including myeloid leukemias,
really do follow a cancer-stem-cell model. But in some other cancers,
there is no meaningful hierarchy, and it will not be possible to cure a
patient by targeting a rare subpopulation of cells. The field has a lot
of work to do to figure out which cancers, or even which patients, fall
in each category.”
Even Morrison, however, acknowledges that the existence of such cells
would have a transformative impact on cancer. “For a model to be useful,
it need not be universal,” he says. “Even if the stem-cell model
applies only to certain forms of cancer, it would be absolutely
worthwhile studying the biology of these stem cells. Universal cures and
theories of cancer have so often failed that we may as well spend time
talking about specific theories for specific forms of cancer. And it’s
in specific cancers that the stem-cell theory might still apply — and
powerfully so.”
My patient, the psychologist, returned to her hometown
in the South. “No bed like your own bed,” she told me in parting,
smiling her pointed, distinctive smile. A week later, when I called her,
there was no answer on the phone. I assume that she died — in her own
bed, on her own terms — with the same dignity with which she lived. I
finished my clinical fellowship in Boston in 2005 and then moved to New
York four years later to set up a laboratory. Our lab studies leukemia
stem cells. We, too, have joined the quest to create drugs that will
wipe out malignant stem cells while sparing normal stem cells.
How might someone go about finding such a drug? Traditionally, three
strategies have produced anticancer drugs. The first relies on
serendipity: someone hears of a chemical that works on some cell, it is
tested on cancer and — lo! — it is found to kill cancer cells while
sparing most normal ones.
The second approach involves discovering a protein present or especially
active in cancer cells — and relatively inactive in normal cells — and
targeting that protein with a drug. Gleevec, the drug used against GIST,
was designed to destroy the functioning of a family of proteins that
are uniquely hyperactive in GIST and in certain leukemias. (There are
only a few drugs with such exquisite specificity for cancer cells.)
The final strategy involves identifying some behavior of a cancer cell
that renders it uniquely sensitive to a particular chemical. Most
traditional chemotherapies, for instance, attack the rapid division of
cells. These drugs kill cancer cells because those divide the most
rapidly, resulting in a narrow discrimination between cancer cells and
normal cells.
Nearly every drug in oncology’s current pharmacopeia can trace its
origins to some variation or combination of these three approaches. But
notably, while each method depends crucially on discriminating between
normal cells and cancer cells, almost none make any distinction
among the cells of any cancer.
The stem-cell hypothesis of cancer poses new challenges for all three
modes of drug discovery. To start, cancer stem cells might be fleetingly
rare — one in a million, in Dick’s original case. A serendipitous
discovery involving a rare cell demands an unusual confluence of luck —
chance multiplied by chance. Defining specific targets in cancer stem
cells might work, but here again there is a battle against numbers.
Finding such genes unique to cancer stem cells first requires isolating
and extracting these rare cells from real tumors, a formidable technical
hurdle.
The most difficult challenge for drug discovery, though, lies perhaps in
modeling the self-renewing behavior of cancer stem cells. To create
drugs, researchers typically begin with a simple cell behavior — say,
its growth or death, or its capacity to change shape. Chemicals are then
tested for their ability to alter this behavior. But in order to reach
cancer stem cells, we might need to devise assays far more complex than
conventionally used. The most traditional metric by which an anticancer
chemical is judged — its ability to reduce the size of a tumor, or to
kill cancer cells in a petri dish — won’t work, of course. If a chemical
kills only the one-in-a-million cell that drives relapse, then it may
not register as a tumor-shrinking or cancer-killing agent. A traditional
drug hunt would most likely miss this kind of chemical — and yet this
is precisely what is needed to attack the roots of cancer. To find drugs
for cancer stem cells, then, we will need not just to find new
chemicals, but also to find new strategies to test these chemicals.
Still, for cancer researchers, the stem-cell hypothesis is as exciting
as it is vexing. The capacity to tear out the roots of a tumor, and
thereby dispel the specter of relapse, represents a sea change in our
thinking about cancer. Indeed, the effort to isolate and target cancer
stem cells is central to a much larger paradigm shift sweeping through
cancer biology. Until recently, much of the field was focused on
understanding the most salient feature of the cancer cell: its ability
to divide uncontrollably. But our understanding of cancer has reached
far beyond distorted cell division. Cancer cells co-opt neighboring
blood vessels to supply themselves with oxygen. They enable their own
movement through the body by hijacking genes that allow normal cells to
move. When some cancers metastasize and punch holes in the bone to
support their survival, they imitate an accelerated form of osteoporosis — in effect, recapitulating the aging process in bone.
Cancer, it seems, is not merely mimicking the biology of rapidly
dividing cells, but that of organs — or even organisms. At its cellular
core, a tumor might nourish itself with its own supply of oxygen; it
might organize its environment to fuel its growth; it might regenerate
itself from a dedicated population of stem cells. Perhaps if we looked
at cancers using appropriate conceptual lenses, we might find that
tumors possess their own anatomy and physiology — a parallel universe to
that of normal cells and organs. Such a tumor can hardly be described
as a disorganized group of cells. It is a cellular empire, with its own
sustenance, grammar, logic and organization. It is a growing being
within a growing being.
Hence the quest to discriminate between normal and malignant cells is
progressively beginning to resemble one of those devastating surgical
operations to separate conjoined
twins.
Every drug that kills cancer stem cells might also kill the normal stem
cells. This operation, too, might end in tragedy for both twins.
But it might not — and therein resides the hope for a next generation of
drugs. If stem cells can be found for certain forms of cancer, and if a
drug can be found to kill these cells in humans, then the clinical
impact of such a discovery would obviously be enormous. And its
scientific impact would be just as profound. Centuries after the
discovery of cancer as a disease, we are learning not just how to treat
it — but what cancer truly is.
Siddhartha Mukherjee is an assistant professor of medicine in
the division of medical oncology at Columbia University. This article is
adapted from his book “Emperor of All Maladies: A Biography of Cancer,”
which will be published by Scribner next month.
http://www.nytimes.com/2010/10/31/magazine/31Cancer-t.html?pagewanted=all&_r=0
