Download Making Sense of a Heart Gone Wild

SPECIAL SECTION
M AT H E M AT I C S
IN
BIOLOGY
Bayesians” notwithstanding, the agency so
far has approved only one drug on a Bayesian
platform. It’s a pill that combines pravastatin
(Pravachol), an existing cholesterol-lowering
drug, with aspirin. Approval was based in
part on a Bayesian analysis that made it easier to synthesize information from five previous trials, and to allow for diverse sets of patients within each of those studies. FDA approved the combination drug in June 2003.
Scott Berry tells the pravastatin story
with pride. Father and son launched Berry
Consultants in 2000, and it worked with the
drug’s manufacturer, Bristol-Myers Squibb,
to shepherd it through approval. “Most of
our meetings take place between 12 and 1
a.m.,” says Scott, who’s the company’s sole
full-time employee.
Because of potential conflicts of interest
with Anderson, Berry Consultants rarely advises on cancer. The overwhelming majority
of its business is medical, however, such as
helping the device firm Medtronic gain approval for an improved shunt for infants with
hydrocephalus. Some companies seek out
Berry Consultants in the wild hope that a
drug or device that’s performed poorly in traditional trials can somehow undergo a
Bayesian resurrection. (Such a “rescue analysis” is rarely a possibility, both Berrys agree.)
His colleagues may be nearing retirement,
but Don Berry isn’t ready to slow down anytime soon. He’s been a workaholic for as long
as Scott can remember. All four of the Berry
boys played ice hockey as children, and Scott
remembers his father attending games back in
the 1970s with work and a clipboard in hand.
Goalie Scott would see his father watching as
the puck slid toward his net. But once it glided
safely away, Scott would glance up and spot
his father braced against the clipboard, scrib–JENNIFER COUZIN
bling away.
NEWS
Armed with computer models, interdisciplinary teams of researchers are
studying what triggers life-threatening fibrillation—and the even deeper
mystery of why it can be stopped
Richard Gray, a biomedical engineer at the
University of Alabama, Birmingham, studies
the heart for a living, but last year the heart’s
mysteries struck close to home. Gray’s
68-year-old father called 911, complaining of
chest pain. The paramedics were already on
the scene when he suddenly collapsed. He
had gone into ventricular fibrillation—his
heart running amok, its muscle fibers all
marching in time to their own drummers instead of beating in unison.
Ventricular fibrillation is a death sentence if not treated within 10 minutes, but
John Gray was in luck. A member of the
rescue squad applied the paddles of a
defibrillator to his chest and with a
whomp of electricity shocked his
heart back into its normal rhythm.
Hundreds of times a day, defibrillation resuscitates people who
would otherwise die in minutes.
For implantable cardioverter defibrillators (ICDs), the success
rate exceeds 99%. (External
defibrillators, like the one used
4
by the rescue squad, have a
lower success rate, primarily
because they are not always
applied in time.) It’s a true
medical miracle—and as befits a miracle, no one can explain why it works. “We don’t
even know how the electric current
goes into the heart,” says Gray. Nor
does anyone really know how ventricular
fibrillation gets started, or why a big shock
brings it to an end.
Gray is one of many bioengineers and
heart specialists who expect the answers to
786
emerge from mathematical models of the
heart. Researchers are experimenting with
virtual hearts in part because it is easier than
tinkering with a living, beating one. And
there is no way to look beneath the surface
of a real animal heart. As Alan Garfinkel, a
cardiologist at the University of California,
Los Angeles, puts it, “You can’t get the light
into the meat.”
So far, mathematics has answered some
questions but raised others. James Keener, a
mathematician at the University of Utah,
Salt Lake City, says that if defibrillation
worked the way most experts think it
does, then we
1
would have a lot more dead patients. “If we
invoke the prevailing theory, the probability
of success is no greater than 20% in our numerical simulations—regardless of the amplitude of the shock,” says Keener. “Yet defibrillators have a success rate that approaches 100% as the shock gets larger. So
we have a problem.”
Some people might argue that this is a
good problem to have. If the treatment
works, who cares that no one understands
why? Garfinkel, for one: “I would urge
that electrical defibrillation, the delivery
of a huge, painful shock by an implanted
$40,000 device, is neither a medically satisfactory solution, nor does it represent
any scientific insight into the phenomenon,” he says. If cardiologists could understand fibrillation from first principles, he
argues, they might be able to improve the
treatment with less expensive equipment,
less painful and damaging shocks, and potentially with antiarrhythmic drugs, which
have until now been an
embarrassing flop.
2
3
Raging current. In a fibrillating heart, the
normal bottom-to-top electrical activity of
the ventricles (above) is replaced by spiraling scroll waves (right).
6 FEBRUARY 2004
VOL 303
SCIENCE
www.sciencemag.org
The mathematical
heart
Mathematical models
showed long ago that
there is some method
to the apparent madness of the fibrillating
heart. Ventricular fibrillation is f irst and
foremost a malfunction in the heart’s electric circuitry. In a normal heartbeat, electrical activity starts near
CREDIT: (INSET) F. XIE, Z. QU, J. YANG, A. BAHER, J. WEISS, A. GARFINKEL, JOURNAL OF CLINICAL INVESTIGATION (IN PRESS, MARCH 2004)
Making Sense of a Heart Gone Wild
IN
BIOLOGY
the top, in the atria; shoots to the bottom of them is the “virtual electrode” theory. to wipe it out (see figure). It’s like fighting a
along special highly conductive muscle About a decade ago, John Wikswo, a bio- forest fire with fire: You have to completely
cells; and rises through the ventricles be- medical engineer at Vanderbilt University surround it, or it might escape. Even then,
fore dying away. When ventricular fibrilla- in Nashville, Tennessee, noticed that an the shock has to be timed just right, or it
tion sets in, however, one or more “spiral electric current applied to the heart creates will just trigger another wave of fibrillation.
waves” of electrical activity start pin- several spots of positive and negative And as a f ibrillating heart has several
wheeling around in the cardiac muscle, voltage—not just a single spot under each “fires” burning at once, it is very unlikely
like the “Mexican wave” in a soccer stadi- electrode of the defibrillator. Wikswo ex- that they will all be doused at one stroke.
um. Just like sports fans, the cells in the plained this observation with a model that
Keener thinks the real key to defibrillation
ventricles start paying more attention to treats the heart as if it were a coaxial ca- lies at the level of the cells themselves. The
the wave than to the game. That is, they ig- ble. Like a coaxial cable, the heart has two cell membrane, he says, responds more vignore the normal pacing signals coming different conductors: the insides of the orously to a positive voltage than a negative
from the atrium.
cells and the outsides. The positive and one. Applying a positive voltage to it is like
Heart modelers who study fibrillation negative spots, or “virtual electrodes,” are pouring gasoline on a fire: It helps release the
are more interested in the electrical behavior places where current is flowing through energy already stored inside the dry wood.
of the heart than its mechanical pumping, the cell membranes, from one conductor to The energy propagates from cell to cell until
because the electricity drives the pump. So the other. Those transmembrane currents, the “fire”—the region of positive electric
they begin with what mathematicians
potential—engulfs the whole heart.
call a reaction-diffusion equation,
Then the fire burns out,
which expresses the two ways that elecand the scroll waves
“Virtual electrodes”
trical signals travel through the heart:
are gone. If the individPositive
by diffusion of ions from cell to cell
ual cells do indeed
Negative
through gap junctions, and by currents
act as batteries, then
that pass through the cell membranes.
the electric potential
Spiral wave
Valentin Krinsky, a Russian bioshould follow a “sawphysicist, and Arthur Winfree, an Ameritooth” profile, with one
Wave core
can mathematician, were the first to reend of each cell posialize that rotating spiral waves arise nattive and the other negaurally as solutions to reaction-diffusion
tive. Earlier experiments failed to deequations. Two interwoven spirals
tect such a sawtooth, but Arkady
emerge from an inactive core or “phase
Pertsov, a biophysicist at SUNY Upsingularity,” like the two flavors in a spistate Medical University, is gearing up
ral lollipop: a spiral of active cells and a
to search with new technology.
spiral of resting cells. The spirals rotate
So are the mathematical models
as individual cells take turns charging up
getting closer to a solution or just
or relaxing. In three-dimensional mod- Calming the waves. Models say the heart’s “virtual elec- adding to the confusion? “I think
els, the spirals become scrolls wrapped trodes” must cover spiral-wave cores—a near impossibility.
we’re getting pretty close,” says Brad
around a “f ilament” that meanders
Roth, a biophysicist at Oakland Unithrough the heart muscle.
Wikswo and others believe, are responsi- versity in Rochester, Michigan. “I don’t
It took experimenters a while to catch ble for calming the spiral waves and re- think the cardiologists will be convinced by
up, because it is not easy to map the heart’s turning the heart to normal.
mathematical models, but they will be inelectric field as it races around the organ
Unfortunately, the coaxial-cable ap- trigued enough to do experiments. I see our
five times per second. Voltage-sensitive proach predicts that transmembrane currents role as motivating the experiments.”
dyes solved the problem, and by 1993 re- should exist only in the outer millimeter of
Certainly Medtronic is taking models sesearchers at the State University of New the heart—not deep enough to stop fibrilla- riously. Belk and fellow engineer Paul de
York (SUNY) Upstate Medical University tion. Natalia Trayanova, a biomedical engi- Groot use them to determine the best placein Syracuse had produced the first video im- neer at Tulane University in New Orleans, ment for the electrodes of an ICD. They are
ages of spiral waves on the surface of a rab- Louisiana, has shown that by taking into ac- also working on “pain-free” defibrillators
bit heart: ghostly yin-yang shapes twirling count the electrical effects of the twisting that restore the heart’s normal rhythm by a
on a heart-shaped radar screen, many times fibers that make up heart muscle, a 3D gentle pacing signal instead of a giant jolt.
per second. Scroll waves have not been seen model can create virtual electrodes extend“The best thing about a model is that
yet, because no one has imaged electric ing into the interior of the heart. Trayanova’s you can see exactly what’s going on,”
fields inside the muscle.
simulation gets rave reviews from some of says Belk. “You can stick 500 electrodes
Although the basic reaction-diffusion her colleagues: “I can’t say enough good on the heart. [Animal experiments typimodels explain why spiral waves exist, they things about Trayanova’s model,” says Paul cally manage 30.] You can induce tachydo not explain how to start or stop them. In Belk, a biomedical engineer at Medtronic cardias on the computer and play around
fact, Keener says, “the heart is the only Inc. in Minneapolis, Minnesota, which man- with different pacing schemes. But at the
medium in which we know how to get rid of ufactures defibrillators. But Wikswo warns end of the day, you don’t have to trust
these waves. The mechanism that works in that some of the assumptions in the model your model very much. You ask yourself,
cardiac tissue doesn’t work in any chemical have not been tested in the lab.
‘Does that make sense?’ If the result is
oscillators. It’s special, it’s unique—and that
Keener, however, thinks virtual elec- valid, almost invariably the answer is
makes it a mystery.”
trodes alone can’t explain defibrillation. A ‘Jeez, I should have thought of that.’ ”
–DANA MACKENZIE
As in any mystery, there is an abun- virtual electrode, Keener says, would have
dance of suspects—perhaps too many. One to cover a scroll wave filament completely Dana Mackenzie is a writer in Santa Cruz, California.
www.sciencemag.org
SCIENCE
VOL 303
6 FEBRUARY 2004
SPECIAL SECTION
CREDIT: ADAPTED FROM J. KEENER/UNIVERSITY OF UTAH
M AT H E M AT I C S
787