3 OCTOBER 2014 • VOL 346 ISSUE 6205 25 SCIENCE sciencemag.org
The worst day of Kenneth Olive’s career began unremarkably. He woke up in his two-bedroom apart- ment in Harlem and tag-teamed breakfast for his 1-year-old son as he and his wife raced to get ready for work. At the 116th Street sub- way stop nearby, Olive hopped on a C train uptown to 168th Street.
His lab is about a block away, in the cancer center at the heart of Columbia University’s medical complex. There, the young
assistant professor was banking his career
on a mouse—one he hoped would pinpoint
new drugs for pancreatic cancer, among the
deadliest diagnoses in oncology.
Sitting on the train, Olive opened an
e-mail message he had downloaded earlier
but hadn’t yet read. It was from
an executive at a biotech company
with which he was working closely.
What he read sucked the air right
out of him. The company, Infinity
Pharmaceuticals Inc., had been
running a clinical trial in 122 people
with advanced pancreatic cancer;
about half were getting a new drug
thanks to impressive results in
Olive’s genetically engineered mice.
Days earlier, data monitors had noticed a
striking disparity between patients who
were on the experimental treatment and
those who were not—but because the trial
was blinded, they did not know which group
was which. Now, the blinds had been lifted.
To everyone’s horror, the patients dying
more quickly were on the new drug. It
was the first big test for Olive’s mice—and
the animals had, by all appearances, failed
“I remember being numb,” Olive says.
It was January 2012, 2 years since he had
joined Columbia’s faculty. He was 35 years
old. “We had built our entire laboratory based
on this mouse,” he says. Intended to mimic
human cancer with unusual precision, the
animals were even being monitored and
treated in a “mouse hospital” custom-built
just for them. Rushing into work, Olive called
a lab meeting for 10: 30. After his group
gathered, he broke the news. “It was a very
quiet and heavy room. I told them that we
were allowed to be mopey for 3 hours, and
then we were going to have a meeting that
afternoon to brainstorm” what went wrong.
Olive is trying to shift a dismal statistic
that plagues his field: About 90% of cancer
drugs that enter clinical trials based on
upbeat mouse data fail. For some tumors,
the need for new therapies is especially
acute. In pancreatic cancer, the 5-year
survival rate is about 6%. “These patients,”
Olive says, “don’t have any options.”
To change that, he and a growing number
of other cancer researchers are trying to
build a better mouse. Their approaches are
a radical break from the past: taking human
cancer cells grown on plastic over many
years and injecting them into an animal.
Dozens if not hundreds of drugs have
subdued cancer in these mice. A handful
have done the same for people.
Olive believes he can do better with a
genetically engineered mouse. At conception,
he endows the animal with the same gene
mutations that show up in most human
pancreatic cancer cells; the mutations
become active during early development, in
cells destined to form the pancreas. Just like
certain people, these animals spontaneously
develop pancreatic tumors.
Changing the paradigm requires not just
a superior rodent. It also takes rethinking
how and how thoroughly the animals are
studied. To wring as much information as
possible from his mice, Olive schedules
every one of them for ultrasounds twice
a week to check for tumors and monitor
existing disease. The “treatment room”
includes boxes of surgical gloves, sterile
drapes, a surgical microscope, and a venting
hood to protect researchers from inhaling
the anesthetic they use on the animals.
But that day in 2012 underscored just how
difficult it is to mirror human cancer in a
wriggling 30-gram ball of fur, where tumor
size is measured in millimeters and lifespan
in days. Every couple of weeks Olive hears
from a scientist or company eager to test a
favorite therapy. “Can you just throw this
into a few mice for me?” they ask. That isn’t
quite how it works, Olive says. The risks and
the challenges sometimes feel overwhelming.
But the payoff, he believes, will make it all
worthwhile—if his mice pan out.
OLIVE STEPPED INTO THE WORLD of
pancreatic cancer by chance. In 2005, fresh
out of graduate school, he was recruited to
join the lab of David Tuveson, an oncologist and cancer biologist then at the University of Pennsylvania. Tuveson had
just published a paper describing an
uncommonly accurate mouse model
for pancreatic cancer.
To create it, Tuveson mutated
two of the mouse’s genes. One,
Kras, shows up in about 95% of
human pancreatic tumor samples.
The second, p53, appears in 70%.
The model isn’t an exact replica of
humans. Unlike people, the animals
have these mutations throughout their
pancreas, not just in their tumor cells, and
throughout their life rather than just when
cancer makes its appearance.
Yet the mice captivated Tuveson and
Olive. The disease’s choreography closely
matched its dance in people. It metastasized
to the same sites—the liver, the lymph nodes,
the lungs. The animals developed many of
the same complications as people, including
fluid collections in the abdomen and a
muscle-wasting syndrome. When Tuveson
slipped a slide from a mouse’s tumor into
a stack of slides from people, a pathologist
“couldn’t tell the difference,” Olive says.
But would a mouse whose cancer generally
Cutting-edge mouse models fuel hope
for understanding and treating cancer
By Jennifer Couzin-Frankel, in New York City
Kenneth Olive holds one of his mice engineered
to develop pancreatic cancer. He tests drugs on the
animals in hopes of identifying the best ones for
“I told them that we were allowed to be
mopey for 3 hours, and then we were
going to have a meeting that afternoon
Kenneth Olive, Columbia University
patient The littlest