|
|
NOW YOU SEE IT…:
Before treatment, and
just seven days after.
These FDG-PET scans
gave a very early signal
that Pfizer’s Sutent was dramatically inhibiting
tumor metabolism. |
More and Better
Experiments
After a lot of internal mulling at pharma companies and some prodding by the FDA, everyone seems to have come to the same conclusion: Drug makers need more and better data to make go/no-go decisions. That means doing more research studies, including some that would have seemed extreme, or at least unusual, not that long
ago.
Various terms are used to describe this new approach. At Wyeth it's called
"learn and confirm." Others call it translational medicine or proof of concept. Everyone is after the same thing
-- a much higher degree of certainty about how things
work.
"Proof of concept is framed by two extremes," explains Novartis' Trevor Mundel, head of exploratory clinical development. On one hand, it can mean having everything, even the data needed to make specific market claims. The other,
"minimalist" definition is just "showing your drug hits the target," he
says.
Mundel has resisted writing down his own definition. He doesn't want scientists
"bogged down" by some preconceived notion. For each project, he wants researchers
"looking at how far can we push the envelope, and be confident we are seeing something that is clinically meaningful," he says. Although a huge number of new tools are available for this, the most important is still basic experimental
design.
At companies like Wyeth and Merck, this has led to reorganization. Speaking at this year's BIO CEO meeting, Merck's Peter Kim described their new proof-of-concept program as a
"parallel track" that runs alongside early-stage development. As a lead compound series is being tested, simpler, better-understood molecules against the same target are being simultaneously run through a battery of additional studies. The goal is to really understand the target's
biology.
At Wyeth, the old divide between discovery and development has been reengineered.
"Learn" teams will now take projects through to proof of concept
-- traditionally Phase II -- where they pass them on to
"confirm" teams. "It's real exploration during early clinical development," says Gombar.
"In the past we sometimes locked onto a disease target based on preclinical data, even when the models weren't that predictive.
"The goal is not to prove pet theories, but to answer the critical questions and then live with those answers even if that means killing more
projects.
"This is not easy to do," says Julian Adams, president and CSO of Infinity Pharmaceuticals. Adams has seen firsthand the ups and downs that even successful drugs face: He invented Velcade (Millennium Pharmaceuticals), which is the first proteasome inhibitor to hit the market.
"Everybody would like to do those killer experiments early that say,
'If the drug survives it will be good and if it fails I will kill the
project,'" he says. Unfortunately, many of the experiments provide inconclusive data; or worse, it turns out the data were
bad.
By doing more of the right experiments, drug developers will face fewer surprises as drugs move into those expensive late stages. So they are thinking ahead and trying to prepare for those studies. For example, if a drug works on the brain's serotonergic signaling system, Mundel's group wants a PET-ligand probe for the target made before it's time to put that drug in human studies. That way, they can do in vivo imaging and quickly see how the drug works in the human pathway. The serotonergic system has been associated with a host of CNS disorders, including addiction, cognitive disorders, anxiety, and depression.
"You can't do Phase IIb studies in all these conditions, so where do you go?" Mundel asks. Small, well-defined studies using PET with the right tracer may point the
way.
Imaging
Explosion
That example illustrates the easiest new trend to spot
-- biomarkers. Drug developers now want biomarkers from start to finish. Markers might be used to tell how well the drug hits the target, whether it's being given in sufficient dose, or if it poses a toxicity risk. The marker can be a plain old immunohistochemistry test or even an X-ray. Or it could be something more sophisticated and expensive such as a PET scan or a gene expression
readout.
Imaging is becoming particularly important.
"Tracer-based technologies have always been favored to look at things in hidden structures," says York Haemisch, of Philips' preclinical imaging arena. Now the tools are even better, providing higher resolution and allowing scientists to look at smaller and smaller structures.
"The biggest trends are looking at metabolic processes and receptor imaging," says
Haemisch.
There has been an explosion in imaging research, leading to many new tracers and even new modalities. Popular imaging techniques include dynamic contrast enhanced magnetic resonance imaging (DCE-MRI), diffusion MRI, magnetic resonance spectroscopy (MRS), and
FDG-PET.
The level of detail available for viewing keeps growing. In PET scanning, gains in resolution have brought even tiny animal structures into view. In the past, better resolution meant sacrifices in sensitivity, that is, the ability to detect the tracer. But now,
"Companies are discovering ways to get more of both using new concepts," Haemisch says. The next generation of PET coming from Philips will have spatial resolution of 1.5 mm and sensitivity of 3.5 percent; that's three times the sensitivity and almost 40 percent better spatial resolution than the previous generation of our preclinical PET."
Even molecules inside the cell can be seen with more precise detail. Sidec Technologies and FEI, for example, have adapted electron microscopy to provide biomolecular tomograms. (See
"Protein Tomography Opens New Vistas" below.)
Bridge
Work
The biggest divide of all is the one between animal and human studies.
"In many areas we have relied a bit too much on animal models and fooled ourselves into thinking they tell us exactly what happens," says Gombar. Now scientists are taking a more skeptical look at their models and trying to find better ways to correlate results in animals to early human
findings.
For imaging studies, achieving spatial resolution and sensitivity becomes even more challenging when you are looking at much smaller bodies, such as those of rats or mice. So major adjustments and specialized instrumentation is required. But drug makers have clearly seen the value, and most are just diving in and making the change.
"Five years ago, I was anticipating it would be 2008 before any pharmaceutical companies had their own preclinical imaging labs," says Philips' Haemisch.
"But today, almost all the big ones already have them," Baum concurs.
"People are using imaging more and more in early-stage trials," he
says.
Tools such as biophotonic imaging provide an extraordinary look at molecular goings-on in the living animal.
"But you can't do those studies in humans," explains Caliper CEO Kevin Hrusovsky, whose company is scheduled to acquire Xenogen, a leader the biophotonic model field.
"The big thing we see are bridging modalities," says Hrusovsky. These are new ways of combining several data sets obtained from different measurement techniques, even from different types of subjects (i.e., animals and humans).
"Imagine if you had an MRI of a human heart, and the molecules you are interested in are lighting up," he says. The data from that study could be more easily compared to the biophotonic study done in a mouse. Mice can also be genetically engineered so that when certain molecular events occurred, they are illuminated.
"Drug discovery researchers want to see more experimentation per organism," Hrusovsky
says.
Some early studies with transgenic animals helped Pfizer pick between Sutent and another promising kinase inhibitor.
"Sutent targets VEGF and PDGF," explains Baum. "The other candidate only targeted VEGF." Because of its dual action, Sutent stems growth of both endothelial cells and pericytes, which surround blood vessels in the early stages of their growth. A group led by Douglas Hanahan at the University of California in San Francisco used RIP1Tag2 transgenic mice to determine which of the two Pfizer drugs most effectively blocked tumor blood vessel growth. These mice have played a pivotal role in angiogenesis research because their tumors develop in a distinctive stepwise fashion that is easy to observe. Hanahan's studies showed that just hitting VEGF was sufficient to inhibit early-stage tumors, but for well-developed tumors, Sutent's dual action made a crucial difference. Later, the Pfizer scientists tested actual patient tumor samples and saw the same effect.
"That gave us a higher level of confidence that we were hitting the mechanism," Baum
says.
Early proof-of-concept trials can also be done in patients that may not have the actual disease being targeted, but are still genetically relevant. At Novartis, for example, they tested an arthritis candidate in patients suffering with Muckel-Wells Syndrome. The drug candidate inhibits IL-1, which is
"100 percent the driver" of that disease," says Mundel. After seeing a
"fantastic" effect in the four Muckel-Wells patients, "We had a good sense of doses we need to get suppression of IL-1," he says. Knowing the drug inhibits IL-1 well, they can now truly test whether that mechanism is relevant in rheumatoid arthritis. The researchers also now have molecular markers in hand to correlate drug activity and effect in
patients.
More
Herceptins?
Some researchers are still frustrated by how slowly this revolution seems to be going.
"Why aren't there more Herceptins already," one attendee asked UCLA's Dennis Slamon during a recent CHI biomarker meeting.
"I'm confident there will be more, and fairly soon," Slamon responded. Genentech's Herceptin is the golden child of biomarker research. By understanding HER2's role, being able to measure its expression, and then targeting that molecule directly, Slamon and others turned a very aggressive form of breast cancer into something curable. Now there is a strong clue about what leads to the cardiovascular side effects that can occur when the drug is taken along with chemotherapy. A genetic test to determine who is most likely to suffer those effects is now
feasible.
Clearly, a lot is known about how this drug works. Unfortunately, even Slamon admits that Herceptin is likely to be an unusual case. It's hard to imagine all the pieces falling into place so neatly very many more times. With the new proof-of-concept approach, however, better drugs are already being made, treatments are being better targeted, and it is easier to predict powerful
combinations.
Back at Pfizer, researchers are still turning over stones. One of the most important compounds in its pipeline is a new cholesterol-lowering drug candidate, torcetrapib. The company is betting that torcetrapib combined with Lipitor will be better than anything currently available to actually help reverse the buildup of arterial plaque. Pfizer recently reported on some NMR studies that looked at the sizes of HDL and LDL particles in patients taking
torcetrapib.
Those NMR studies showed that patients on the combination had HDL and LDL particles that were larger than what's usually seen. A Pfizer press release notes that
"It's not clear yet what that data means." In the new way of doing things, that doesn't matter, at least not right now. It's more information to fill out the picture. If it turns out particle size does make a difference, at least the Pfizer scientists will be among the first to
know.
|
|
Protein Tomography Opens New
Vistas
M.B. |
| When scientists at AstraZeneca wanted to know more about how a particular ion channel forms, they turned to a new technology called Protein Tomography. Visualizing the channel, right in the cell membrane,
"We could see it was clearly a tetramer," says Margaretha Gadnell of Sidec Technologies. Even more interesting,
"We could see subunits of the channel in the cytoplasm." Those subunits were dimers that were pair-wise attached to an assisting protein. The results from the study could shed light on questions related to cell-based assay
development.
Sidec and FEI have collaborated to make Protein Tomography available both as contract research and instrumentation and software. Samples are usually flash frozen, as they must be specially prepared prior to analysis. Gold-labeled antibodies are used as tracers. Multiple transmission electron microscopy pictures are taken of the molecules from different angles, and then the 3-D image is reconstructed.
"The big issue is how the analysis and reconstruction is done," says FEI's Jack Elands.
"Data acquisition and reconstruction has to be as efficient as possible to get all the information available into the
images."
As Gadnell points out,
"There have not been too many technologies available to study disease mechanisms at a protein level."Sidec and FEI have seen
"a big jump" in business as researchers demand a better understanding of how drugs work. According to Elands, at first some scientists did not believe that looking at single protein molecules in this way would be useful, but that perception has completely changed.
"There are many things that biochemical studies don't allow you to measure, that you can actually see with this technology."
|
|
|
Additional Material
Online
|
