Layers of ’ifs’ bedevil antiangiogenesis
Close your eyes when you listen to antiangiogenesis researchers
talk about their work, and it’s easy to imagine yourself drinking
in the words of presidential candidates as they stump their way
cross-country and over the airwaves, sharing their dreams for a
better America: So many hopes. So many visions for a better future.
So many possibilities for arriving at that golden tomorrow.
Then, with a jolt, you realize
the election is nearly a year away, and all but two candidates will
fall by the wayside before voters head for the polls. Even then,
little may change despite the new office-holder, who typically sees
his grand ideas ground down by Congress, politics, or scandals.
Likewise, the roseate promises of antiangiogenesis therapy may
pale over the long haul. No one knows whenor even ifthe
best candidates will emerge, how these agents should be monitored,
or whether they will prove useful in the clinical setting. While
study after study demonstrate their ability to shrink or eliminate
tumors in animal models, the leap to humans is enormous.
Or, as pioneering angiogenesis researcher Judah Folkman, MD, has
been known to point out, antiangiogenesis therapy at this point
means, "We could treat a lot of mice really nicely."
Good for the mice. But what about humans?
The legion of researchers trying to answer that question tend
to exude an air of cautious optimism as they wend their way through
the many layers of "ifs" that attend their work.
"We’ve had many ’miracle’ compounds come along that have failed
by the time they get to phase II trials," acknowledges Laurence
M. Demers, PhD, distinguished professor of pathology and medicine
at Penn State University College of Medicine. "Maybe you see no
change, or you extend the patient’s life by a weekthere are
tons of studies out there showing results like that."
"But," he quickly adds, "I truly think we are on the verge of
finding out this is an entirely new way to treat malignancy."
"We’re on the cusp of something great," agrees Edward Gubish,
PhD, senior vice president for research and development at EntreMed.
The vast media attention given to the work of Dr. Folkman
and his colleagues at Harvard Medical School has made the basic
concepts of angiogenesis familiar to many. For a tumor to grow,
it has to create vessels to feed itself and to carry away toxic
waste. "Tumors that can’t do that essentially just remain dormantthey
don’t grow at all. And as soon as they switch to an angiogenic phenotype,
that is, the ability to induce vessels, then they metastasize and
grow," explains Noel Weidner, MD, professor and director of anatomic
pathology at the University of California, San Diego, and author,
along with Dr. Folkman and two others, of a seminal paper on tumor
angiogenesis and metastasis, published in The New England Journal
of Medicine a decade ago (1991;324:1-8).
Dr. Weidner (who at the time was affiliated with Brigham and Women’s
Hospital, Boston, and Harvard Medical School) and his colleagues
were the first to demonstrate microvessel density was predictive
of outcome in invasive breast cancer, and they suggested assessing
tumor angiogenesis could be used to select patients for aggressive
therapy. As interest in and acceptance of angiogenesis have grown,
so have efforts to identify antiangiogenic agents that would take
advantage of this mechanism. While no one is suggesting such therapy
would defenestrate traditional cancer treatments, combining antiangiogenic
agents with radiation or chemotherapy might stop many types of malignant
tumors before they metastasize. Dr. Weidner likens it to maintaining
a bonsai tree, whose dwarf stature is achieved through careful cultivation.
It’s an intriguing notion. "We’ve all struggled for years with
the fact that a patient can have a tumor surgically removed and
be free of disease for two, five, even up to 10 years, then suddenly
relapse," says Michael S. Gordon, MD, associate professor of medicine,
Section of Hematology and Oncology, Indiana University School of
Medicine. What used to be called tumor dormancy is now explained,
under the concept of angiogenesis, as the existence of microscopic
deposits of tumor whose extremely high proliferative rate is balanced
by apoptosis, or programmed cell death.
This yin and yang keeps these tumors microscopic in size; for
them to grow, new blood vessels must start feeding them. "When this
occurs, it’s as if a switch has been flipped; this switch is angiogenesis,"
Dr. Gordon says. "Then the proliferative rate stays up, apoptotic
rate drops, and the tumor starts to grow and metastasize."
Preventing the switch from being flipped, or circumventing it,
thus seems a reasonable target and has become the Holy Grail of
countless researchers and drug manufacturers. Some 20 agents are
in various phases of clinical trials, and another 20 to 30 are under
serious scrutiny. Though that sheer number alone points to the palpable
excitement in the field, it is also indicative of the confusion
"This is no small issue," says Dr. Gordon. "It’s difficult not
only for patients, who are pinning a lot of hopes on these drugs,
but for investigators. How do you know which one to focus on?"
To answer that question, researchers are taking aim at the more
recondite aspects of tumor angiogenesis, trying to untangle its
underlying mechanism. Evidence to date suggests the answer will
"There are so many pieces of data out there, and so many players
involved, there’s going to be no one star in the game," predicts
John Berryman, cancer products trials coordinator for American Diagnostica
Inc. "It’s unlikely that any one mechanism is predominant in the
angiogenic process, and therefore there will be no single agent
that will be predominant."
"If we look at angiogenesis as a cascade of events, rather than
one event, we can look at many different points at which one could
interrupt that cascade," adds Dr. Gubish.
That line of thinking is mirrored in the current clinical trials
for antiangiogenic agents. No hard-and-fast schema for their classification
exists, although the current understanding of the angiogenesis mechanism
hints at some broad categories.
Grabbing the most headlines, of
course, are the endogenous proteins endostatin and angiostatin.
The former is part of collagen 18, which in turn is part of the
endothelial cell framework; the latter is a normal fragment of plasminogen.
Both proteins were isolated by Michael O’Reilly, MD, of Dr. Folkman’s
laboratory, and have been developed into drugs by Entre-Med.
Phase I solid tumor studies with endostatin began last year and
are taking place at Dana-Farber Cancer Institute, Boston; M.D. Anderson
Cancer Center, Houston; and the University of Wisconsin Comprehensive
Cancer Center, Madison. Phase I trials with angiostatin and 2-methoxyestr-diol
may get under way as early as the first quarter of this year.
These and similar agents, which are in various stages of phase
I and phase II trials, essentially starve the tumor by inhibiting
growth of endothelial cells. Such cells are an important target
in preventing new blood vessel growthand, given their normal
status, perhaps a slightly more vexing one as well. "Tumor cells
are actually a little easier to attack," says Dr. Weidner. "They’re
weak cells, injured cells, and about a half step away from being
dead in their own right. But cells with an intact physiologic system,
without any injury, are a different ballgame."
Agents taking aim at normal endothelial cells could also potentially
attack such cells throughout the body, not just at the tumor site,
although Dr. Weidner suggests such risk may be low. "Normally, endothelial
cells in adults are restingthere’s very slow turnover. It’s
the endothelial cells in the tumor that are growing rapidly, and
they may be more susceptible to these agents than resting endothelial
A related, though different, approach involves inducing angiostatin
in vivo. Gerald A. Soff, MD, reports that by combining two classes
of drugsincluding a plasminogen activatorit is possible
to convert plasminogen to angiostatin. "We can increase the angiostatin
levels from less than 10 nmol/L to 200 nmol/L, and we have demonstrated
clinical efficacy of this approach in several initial patients who
have been treated on the compassionate-use basis," says Dr. Soff,
who is assistant professor of medicine, Division of Hematology/Oncology,
Northwestern University Medical School. He is also affiliated with
Northwestern’s Robert Lurie Cancer Center, Chicago.
Drugs that block extracellular matrix breakdown represent another
loose category of agents, although some argue that because they
are anti-invasive compounds, they are not truly antiangiogenic.
These matrix metalloproteinase (MMP) inhibitors are probably the
furthest along in clinical testing, with approximately half a dozen
agents in phase II and phase III trials, and perhaps another dozen
under review. MMPs include enzymes such as collagenase (MMP-1),
elastase, and gelatinase A and B (MMP-2 and MMP-9, respectively).
Nonselective MMP inhibitors, such as those aimed at MMP-1 and possibly
MMP-13, have been linked with joint-related side effects; some of
the newer, more selective inhibitors, including those that block
MMPs-2 and -9, may cause fewer side effects.
"Metalloproteinase inhibitors give you the ability to shut down
the network of vessels leading to and from the tumor," explains
Penn State’s Dr. Demers. "It’s almost like dealing with an extensive
highway system, where, if you want to cut off traffic to a particular
town, you basically close down the roads leading into it."
Another category is specific agents that inhibit a variety of
growth factors. Some are aimed directly at vascular endothelial
growth factor, including anti-VEGF antibody, interferon-alpha, and
SU5416. Othersincluding the recently resurrected thalidomide
(N Engl J Med. 1999;341:1565-1571)appear to operate
on growth factors ranging from transforming growth factor b to fibroblast
"What you have is this complex network of cross-talk between these
different growth factors, which normally allows for normal skin
growth, collagen growth, and so on," says Dr. Demers. Malignancy,
of course, revs up this cross-talk to another level. Drugs that
lower the availability of VEGF, TGFb, or FGF can inhibit vascular
and tumor growth.
In addition, notes Dr. Gordon, many drugs in development, such
as tyrosine kinase inhibitors, are homing in on specific growth
factor receptors; still others, such as EMD121974, appear to act
by inhibiting endothelial-specific integrins. And, he notes, "There
are also a lot of drugs that appear to have antiangiogenic activity,
but we don’t how they work." TNP-470, for example, may work on basic
FGF or directly against endothelial cells.
What is becoming apparent, researchers say, is that many of these
mechanisms are occurring simultaneously. "You could argue that there’s
a rationale for taking an MMP inhibitor and a VEGF inhibitor and
an integrin inhibitor and using them all. It makes sense," Dr. Gordon
says. Sorting out which combinations work bestor not at all"will
be the focus of [research] laboratories in the next few years."
It’s possible that each tumor may have its own angiogenic profile,
suggests Lee M. Ellis, MD, associate professor of surgery and cancer
biology at M.D. Anderson, although trying to discern the optimum
slot of each agent leaves researchers looking a little like U.N.
ambassador Richard Holbrooke plotting the seating at one of his
"If, for instance, in the endostatin trial, we see a decrease
in VEGF, that may not necessarily mean that endostatin decreases
VEGF expression from the tumor. It could mean that the tumor mass
is getting smaller and therefore there’s less VEGF being released
into circulation," says Dr. Ellis, who is overseeing the studies
correlating basic science endpoints with clinical observations in
the phase I endostatin trial and another antiangiogenesis trial
at M.D. Anderson looking at SU5416 and an agent called CPT11.
"If a tumor is driven by VEGF and you use anti-VEGF therapy,"
he continues, "you may see some beneficial effect. However, it is
possible that you may knock off the heterogenous portion of the
tumor where angiogenesis is driven by VEGF; the remaining cells
of the tumor may be overexpressing another angiogenic factor." In
such a case, initial response to therapy would be good; however,
it might be followed by tumor progression as angiogenesisnow
driven by another factorcontinues. A nonspecific antiangiogenic
agent, perhaps one that targets activated endothelium, might obviate
the need for an angiogenesis profile, but such compounds appear
to have mixed results, even in animal models, Dr. Ellis observes.
Culling the most effective antiangiogenic agents or combination
of agents from the current field is at best a herculean task, one
matched in complexity only by the search for appropriate testing methods.
"One of the limitations in doing antiangiogenic trials is that
there are no good surrogate markers for efficacy besides the ultimate
clinical response," says Northwestern’s Dr. Soff. "At this stage
in the game, there are no well-developed, standardized assays, and
this is a major limitation of the clinical trials that are ongoing
and will be ongoing."
In fact, suggests one observer, lack of standardized assessment
methods may be part of the reason antiangiogenesis theory has come
under attack from some quarters.
"One of the common criticisms in this field is that study results
are often reproducible only with the original observers," says Douglas
S. Harrington, MD, president and chief executive officer of ChromaVision
Medical Systems Inc. "And a lot of the studies are difficult to
compare because of the significantly different number of methods
that have been used to assess results."
Antiangiogenesis research will not grind to a halt because of
this gap, but the burden of proof will only grow heavier as time
goes on. Says Dr. Demers: "Now that we’re moving from animal models
to human models, there’s a real onus on the drug manufacturers to
determine that an agent is having an effect, apart from just measuring
tumor shrinkagewhich in itself is not all that easy."
So what are the possibilities? Current clinical trials may provide
a glimpse, although they raise as many questions as they answer.
At M.D. Anderson, Dr. Ellis and his colleagues will be measuring
angiogenic factor levels in serum of patients enrolled in the endostatin
trial as well as the SU5416/CPT11 trial. They will also obtain a
preoperative biopsy near or at the tumor normal interface, tissues
frozen in OCT solution, and formalin-fixed, paraffin-embedded tissues.
Vessel density will be measured using CD34 staining on paraffin-embedded
tissues and counting the number of vessels in a high-powered field.
Staining for apoptosis will be done on either the frozen or paraffin-embedded
tissue using a commercially available stain.
The researchers also plan to use an immunohistochemical method
developed by Dr. Ellis and his colleagues and recently reported
on in Cancer Research (1999;59:5412-5416), in which apoptotic
cells stain green, endothelial cells stain red, and overlapping
areasstained yellowdenote apoptotic endothelial cells.
This double-staining technique is particularly interesting, Dr.
Ellis says, because if endostatin induces endothelial cell apoptosis,
as some researchers have demonstrated, "We would then be able to
determine endothelial cell apoptosis pre- and post-therapy to see
if endostatin increases the number of endothelial cells undergoing
Dr. Demers and his colleagues at Penn State are developing tests
to assess patient response to a variety of MMP inhibitors. "We’re
especially interested in those MMPs present in the circulation,
such as MMP-3, MMP-9, and MMP-13," he says. In one trial they are
testing MMP inhibitors in patients with metastatic breast cancer,
monitoring the effects of the inhibitor on circulating metalloproteinase
levels and serum VEGF levels, as well as more generalized tumor
markers (CEA and CA 15-3) and biochemical markers of bone metastases.
"We’re looking at a whole panel of new tests, such as the MMPs,
to assess patient response to metalloproteinase and other antiangiogenic
inhibitor therapy in the cancer patient with metastatic disease,"
Dr. Demers says. "These tests have been developed for research,"
he adds, "and simply are not part of anyone’s standard test panels."
At Northwestern, Dr. Soff has used a simple Western blot to demonstrate
the efficacy of in vivo angiostatin induction. "We have the advantage
of actually being able to measure angiostatin in the plasma as our
surrogate marker," he says.
Clearly, no one method of assessment will suffice, just as no
one therapy will be the magic bullet. "That’s why the war on cancer
has been such a long and complex battle, because every mechanism
is happening at once," says American Diagnostica’s Berryman. "You
can target one mechanism, and 10 more occur. You can target 10 mechanisms,
and dozens more happen."
The good news for laboratories is that none of the tests they
eventually may be called on to perform are likely to be particularly
esoteric. "In most cases, we’re talking about fairly simple tools:
serum assays, microvessel density determinations, advanced imaging
techniques. In other words, not power tools, but screwdrivers and
wrenches," Berryman says.
And the bad news? "At this point, we’re talking about a thousand
different wrenches and screwdrivers," he says with a laugh.
No laboratory is large enough to accommodate so many tools,
nor, realistically, will it have to. As with any parvenu enterprise,
angiogenesis research is far from fully developed. Once (or if) the
field matures, a handful of practical testing methods likely will
emerge. "Right now we’re working with assays that are set up for research
purposes but not for the clinical lab," says Dr. Demers. "That will
At one end of the spectrum lie relatively simple serum
tests. Speaking at a plenary session at last year’s annual meeting
of the American Association for Clinical Chemistry, Dr. Folkman
suggested pathologists will be asked to perform "ELISA immunoassays
that will tell us the blood levels of these angiogenesis inhibitors."
A research ELISA is available for endostatin, and one for angiostatin
may not be far behind.
While such tests are straightforward, their practical applications
are less so. "The information is never as simple as you want it
to be," says Berryman. You can almost hear him sigh as he says this.
His company has developed a testwhich is being further assessed
by Dr. Demers’ laboratoryto measure urokinase-type plasminogen
activator receptor levels in plasma. It appears that the urokinase
family, including the cell surface receptor, protease enzyme, and
its type-1 inhibitor (PAI-1), is involved in tumor spread via matrix
destruction. The receptor likely also plays a role in signaling
other events leading to metastasis.
"Levels of this cell surface receptor seem to correlate, by some
initial data at least, with prognosis, overall survival, and disease-free
survival in breast cancer patients," Berryman says. "But we don’t
know how it’s related. Is it related to meta-static spread? When
does it come into play? Is it involved in the angiogenic process
in a way that’s distinct from tissue remodeling?"
"Remember, angiogenesis can occur for reasons unrelated to metastasis,"
he says. "Angiogenesis happens all the time in various locations,
for various reasons. Anyone who has cancer is also undergoing therapies
that may be causing damage by themselves, which may cause an angiogenic
response regardless of whether there are metastatic lesions."
What would be most valuable, Berryman says, is a marker that could
specifically indicate the occurrence of metastatically linked angiogenesis.
"But I haven’t seen or heard of anything like that yet," he adds.
The serum marker issue "is a lot like what we dealt with 10 years
ago with cytokines and interleukins," says Dr. Gordon. "They’re
upregulated or downregulated because of the physiologic response,
but there are so many that interweave and interrelate that it’s
hard to know what a change in any one factor means. If VEGF goes
down, what happens to basic fibroblast growth factor? Or does a
drop in VEGF change some native, internal antiangiogenic factor,
like MMP inhibitors? And what happens to MMPs in a setting where
you change VEGF levels?" In fact, he notes, "There have been some
real concerns that MMP inhibitors, for example, may actually cause
upregulation of VEGF. If that’s the case, is there a negative side
to inhibiting certain MMPs rather than others?
"It may be that, down the road, we have the ability to interpret
the angiogenic factor and antiangiogenic factor analyses," Dr. Gordon
continues. "But for now, though we can monitor them and measure
them, I don’t think we can interpret them."
Dr. Soff raises another issue. It may be, he suggests, that antiangiogenesis
therapy will resemble the administration of an antibioticdosing
will be within a given range for all patients, with no need to monitor
plasma levels. On the other hand, antiangiogenic agents could turn
out to be more like warfarin in terms of their monitoring requirements,
necessitating careful titration to achieve efficacy and avoid complications.
"In that case, we’re going to desperately need a blood marker to
guide the dose adjustment," he says.
Even more complicated is the matter of micro-vessel density determinations.
As Dr. Weid-ner points out, perhaps unnecessarily, "This is not
straightforward. It takes a trained pathologist who has a good,
clear understanding of what tumors are all about, what they look
like, and with the ability to pick out the right areas to count
It will also take a certain amount of agreement among pathologists,
as well as more standardized methods, both of which are lacking,
says Dr. Ellis.
"I think it is important for pathology societies to determine
whether or not they’re going to use vessel density as a prognostic
factor," he suggests. "And if they are, they will need to set some
standards. How are you going to quantitate the microvasculature?
What antibodies are you going to use? What are your cut-offs going
The picture becomes even cloudier, he adds. "First, you have to
prove that vascular density is a better prognostic factor than other
factors studied. And it’s going to be different from tumor to tumor.
Not only that, it’s going to be different within individual areas
of the tumor. Vessel density at the normal tumor interface, for
instance, is going to be higher than it is in the center of the
"The other problem is, even if you have high vascular density,
it doesn’t alter your therapy," he notes. "You would intuitively
think that if patients have high vascular density, then they would
be the ones most likely to benefit from adjuvant therapy, or the
current adjuvant therapies that we have. However, there aren’t any
good data out there to show that patients who have high vessel density
benefit from adjuvant chemotherapy."
Trying to clear away at least some of the fog, ChromaVision is
developing a method to automate assessment of microvessel density.
The platform uses image analysis software that permits pathologists
to capture an entire stained specimen and store it on a computer,
"as though it were an aerial photograph of the slide," explains
Dr. Harrington. "This makes identifying hot spots"a frequently
cited area of concern"extremely easy."
The company has been performing retrospective studies to identify
which parameters or combination of parameters, as well as which
stains, offer the best predictive value, Dr. Harrington reports.
"This is definitely a multiparameter problem," he says.
The software’s developers are focusing on vessel counts, though
future plans call for comparing numbers of vessels with areas of
stain. "In time, when we have developed both ways of looking at
the microvessel density, we’ll be able to compare the two and see
if one provides better information as far as patient outcome, response
to therapy, and so on," says ChromaVision collaborator Roscoe D.
Atkinson, MD, assistant professor of clinical medicine, Department
of Pathology, Keck School of Medicine, University of Southern California.
Initially, Dr. Harrington suggests, the technology might be used
to stratify patients in clinical studies to understand their response
rates, as well as for prognostic purposes. "From that, we’ll potentially
derive a means of using the technology to guide who should and who
shouldn’t get a specific therapy. It’s highly likely that some patients
will not be candidates for an antiangiogenesis therapy, based on
an initial evaluation of the tumor. They may have such low vessel
density that it won’t be clinically useful to treat them."
Beyond the laboratory, it’s possibleeven probablethat
noninterventional radiology tests will be an influential part of
antiangiogenesis therapy in areas ranging from blood flow to tumor
necrosis. Toward that end, the National Cancer Institute is sponsoring
a working group, scheduled to meet in late February or early March,
to assess potential imaging modalities in biologic therapies, including
In short, observers say, almost any detection method might prove
practicaland will be used if it is. "Anyone who has a consistent,
reliable tool is going to have a place at the table," Berryman predicts.
In all likelihood, that will be a large table indeed.
Karen Titus is CAP TODAY contributing editor and co-managing