November 12, 2004 —
The Food and Drug Administration (FDA) recently approved two new drugs -- Enbral,
for treating adult and juvenile rheumatoid arthritis, and Xolair, the first monocloncal
antibody for treating allergy-related asthma.
The testing of both drugs --- and many others before them -- relied heavily on
the work and products of one of the USC Viterbi School's longest standing and
most productive research efforts, the Biomedical Simulations Resource (BMSR).
Dean C. L. Max Nikias calls the BMSR "a brilliant model for what creative university
engineering research should be."
Once a daring vision, the BMSR has developed the classic, basic engineering tool
of "systems modeling" into well-used technologies applied by front-line medical
researchers and industry to biomedical problems. All 15 of the world's largest
pharmaceutical companies now use BMSR-developed methods as part of their process
for testing new drug candidates.
The BMSR’s astonishing track record was validated by this year's five-year renewal
of its funding from the National Institutes of Health. The new grant of $5.9 million
running through 2008, will bring the total funding to $21.4 million and like
the previous ones, the renewal was competitive.
“It is difficult to get NIH center grant funding,” says BMSR co-director David
D’Argenio, “and even more difficult to maintain it, let alone for more than 23
The record achieved by BMSR co-directors Vasilis Z. Marmarelis and D’Argenio
of the Department of Biomedical Engineering is the result of negotiating a promising
but very difficult path. In many areas of engineering, success in understanding
physical processes and improving devices and techniques has come from devising
mathematical models based on theoretical principles, and then refining these models
by checking their predictions against observations, in an iterative process. The
key is to precisely quantify predictions, so that the model's fit to reality can
Engineers have long used the technique to improve aircraft, chemical plants or
automobile engines, or the construction of structures ranging from the atomic
to the skyscraper level. But applying the technique to biology and medicine was
still fragmentary and unsystematic when Marmarelis and D’Argenio made their initial
NIH application in September 1984. They proposed a research unit with a mission
“to advance the state of the art in biomedical systems modeling,” working on both
basic research methods and practical software development.
Now, as the project approaches the midpoint of its third decade, biomedical modeling
has become established and applied to conditions such as Alzheimer’s disease,
stroke, diabetes, cancer, AIDS, and sleep disorders.
“The FDA has recognized the importance of an expanded role of systems modeling
in drug development,” notes D’Argenio, “and in a recent white paper has called
for the creation of a new product development toolkit that includes computer-based
predictive models to improve and accelerate the drug approval process.” D’Argenio
was recently appointed to the FDA Clinical Pharmacology Subcommittee, the only
engineer on this 12 member panel that provides advice to the FDA on drug response
modeling, pharmacogenomics and pediatric clinical pharmacology. “It is gratifying
to see that the systems modeling framework, as well as the methods and tools created
and refined at the BMSR are playing an increasingly important role in drug development”
This great success was not achieved overnight.
“The challenge in biomedical systems modeling,” explains D’Argenio, “is biological
processes are nonlinear, with complex feedback mechanisms operating within and
between at least five interconnected, but separate levels: the molecular level,
the cellular level, the intercellular level, the organ level, and the whole body
level. Accordingly, the methods and tools we develop must be able to describe
the autoregulatory consequences of these feedback processes at all levels, both
in heath and disease.”
Diabetes, which Marmarelis, a professor of biomedical engineering and electrical
engineering, has recently been studying, is a classic example. The condition results
from a single, simple well-understood cause – the failure of cells in the pancreas
to manufacture enough of the blood sugar regulating substance insulin. The results
of a shortage of insulin, however, are anything but simple and Marmarelis has
played a key role in better understanding diabetes. The condition affects almost
every organ in the body differently, and in different degrees (in non-linear fashion).
Because the functioning of different organs is interconnected, the initial effects
of reduced insulin production triggers a cascade of other effects, creating the
bewilderingly broad clinical picture of diabetes.
Marmarelis uses modeling methods to analyze the complex dynamic interactions
between insulin, glucose and free fatty acids and with the goal of achieving the
long-held dream of an "artificial pancreas. " Such a device would maintain the
level of blood glucose fairly constant throughout the life of diabetics in an
automatic fashion. This advanced scientific understanding also reveals critical
aspects of the causes of obesity and offers clinical solutions for this growing
“Ultimately, our goal is to understand the full complexity in the dynamic nonlinear
interactions among the metabolic, the endocrine, the cardiovascular and the nervous
system, thereby ushering in a new era in quantitative medicine," Marmarelis asserts.
Clinical research has given physiologists a general, qualitative overview of
what the mechanisms are but in order to better control the disease, it is crucial
to have a dynamic, unified, quantitative model. Such a model must integrate the
multiple interconnected variables that enables an experimenter to see, in a simulation,
what happens to the whole system when individual inputs chantge.
Marmarelis’ BMSR work serves as the basis for a new book, “Nonlinear Dynamic
Modeling of Physiological Systems,” just published by Wiley Interscience and IEEE
D’Argenio's has applied the BMSR systems modeling philosophy to therapeutic
drug development through “in silico methods,” using it to study and evaluate candidate drugs to treat cancers,
viral, autoimmune and other diseases. He has developed sophisticated methods to
model the way drugs work, from their action on molecular targets through clinical
disease response. He has created software embodying the methods and models that
other researchers can use in their work, whether for drug development or physiological
“To date,” says D’Argenio, “more than 4500 biomedical researchers worldwide have
used BMSR developed software.” The titles include the drug effect modeling software
ADAPT, the general biological system modeling system LYSIS, the neural modeling
system EONs and PNEUMA, a set of programs describing cardiorespiratory function.
BMSR techniques and software require training and since 1985, 2400 researchers
from all over the world have attended 23 short courses and 19 workshops on BMSR
In addition to Marmarelis and D’Argenio, University professor Theodore W. Berger
and Michael C.K. Khoo, professor chair of the Department of Biomedical Engineering
serve as main BMSR investigators.
Khoo’s research focuses on improving the modeling of biological control systems
and how they feedback on each other, with particular emphasis to the intricate
teamwork between heart and lungs. His applications to cardio-respiratory coupling
in sleep disorders were highlighted in the Spring/Summer 2003 issue of USC Engineer.
Berger’s longterm collaboration with Marmarelis centers on modeling neural activity
in the brain, specifically the hippocampus. This research has been the springboard
from which he embarked on his now widely known efforts to create sophisticated
microelectronic components that can emulate the physiological function of parts
of the central nervous system. He hopes to eventually repair damaged neural tissue
and is also attempting to better understand the processes leading to Alzheimer’s
Marmarelis is more enthusiastic than ever about the prospects for BMSR as a leader
in promoting a new quantitative medicine that will be more flexible and effective,
and give more bang for the experimental buck.
“I believe the best is yet to come in our area,” he said. “Over the past twenty
years, we established elements of biomedical modeling using the computers and
software tools of the time,” he notes. “But I believe the accelerating improvement
in computers, and advanced methodologies of biomedical data analysis and collection
is going to have an explosive impact on our specialty, and on medicine as a whole.
And I think when the history of this impact is written, USC Biomedical Engineering
will have a striking role.”
- Eric Mankin