Piercing the body with precision
How imaging is aiding the fight against cancer
New imaging technologies are raising hopes that doctors soon will be able to locate tumors with pinpoint accuracy, and track their hour-by-hour response to treatment—without the need for surgery.
Coupled with recent advances in genetics and molecular biology, imaging is speeding the discovery and evaluation of safer, more effective treatments that can stop tumors in their tracks.
“In the past 10 years we’ve made tremendous strides in improving imaging of cancer,” says Dennis E. Hallahan, M.D., chairman of Radiation Oncology at Vanderbilt University Medical Center. “In the near future we will be using functional imaging to image pre-cancer.”
At Vanderbilt, for example, a sophisticated technique called dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is being tested in women with breast cancer.
The goal: to see whether new “targeted” therapies are shrinking tumors by disrupting their blood supplies. If successful, the technique could avoid the need for repeat biopsies, says Tom Yankeelov, Ph.D., director of Cancer Imaging in the Vanderbilt University Institute of Imaging Science.
“Particularly in breast cancer there’s currently no adequate, some would say there’s no non-invasive method at all, to determine whether or not a tumor is responding to treatment,” Yankeelov says.
“It’s really a sad state of affairs,” he says. “Repeat biopsies are not really an option because you have to go under the knife each time—who wants to do that?”
In addition, “biopsies by definition only sample a portion of the tumor. It is entirely possible that you could sample a section of tissue that is free of active disease and miss the sections that are actively proliferating.
“That is why imaging is so powerful,” Yankeelov says. “You can get a more complete description of the tumor status, and you can do it non-invasively.”
That’s the aim of DCE-MRI, a modified MRI technique in which a contrast agent is injected into the patient to outline the profusion of fragile, leaky blood vessels that spring up to feed growing tumors.
Nearly 40 anti-angiogenic drugs, which inhibit the growth of these vessels, are now in clinical trials. Advanced imaging technologies like DCE-MRI, by detecting changes in blood flow and vessel permeability or “leakiness,” for example, may help doctors determine whether the tumor is responding—even after the first course of chemotherapy.
More work needs to be done, however, before DCE-MRI will be ready for the clinic. “I personally think (it) is really just another tool in the toolbox,” Yankeelov adds.
Newer imaging techniques can measure glucose metabolism, hypoxia (lack of oxygen) and the diffusion of water molecules in and out of cells—indicators of how big the tumor is, how “healthy” it is, and whether it is surviving attempts to kill it.
By tracking various markers that have been tagged with a radioisotope, PET also can tell whether a tumor is dying—or proliferating. Similarly, mass spectrometry techniques can detect changes in the expression of various proteins by tumors in response to treatment. Whether these techniques can predict the outcome of therapy and its impact on patient survival remains to be proven clinically.
Another technical challenge: “registering” the different images—mapping coordinates representing the same anatomical point so that the same “voxel,” or three-dimensional piece of data, lines up in each of them.
In the same patient after chemotherapy (B), a drastic reduction in perfusion/permeability indicates treatment is successfully "starving" the tumor by disrupting its blood supply.
(C) and (D) are single-slice images taken from the center of the 3-D volume renderings before and after treatment. The hope is that this kind of analysis will enable doctors to determine early on whether the tumor is responding to therapy.
Guiding the scalpel
Registration already is an integral part of stereotactic surgery and radiosurgery, the precise guidance of scalpels and radiation beams to remove abnormalities, including tumors, with minimal damage to surrounding tissue.
New techniques developed by Vanderbilt engineers and computer scientists are extending the reach of the neurosurgeon and radiation oncologist even further. Their contributions are proving to be invaluable, especially for treatment of aggressive, infiltrating glioblastomas of the brain.
“The visual cues we have at surgery are really poor,” explains Reid C. Thompson, M.D., director of Neurosurgical Oncology at Vanderbilt. “There isn’t often a discrete edge… maybe there’s a slight discoloration… maybe the tumor just feels a little different.”
As a result, he says, “you either don’t take out enough tumor in the brain, and we know that’s probably not good in terms of prognosis, or you take out too much, which is an obvious problem.”
To further define the margins of the tumor during surgery, Anita Mahadevan-Jansen, Ph.D., and her colleagues in the Department of Biomedical Engineering have developed an optical probe that within 30 seconds can differentiate between normal and abnormal brain tissue based on the spectra of light bounced off of them.
A recent clinical study concluded that the instrument can achieve what amounts to an “optical biopsy” with “near-instantaneous feedback,” improving the percentage of tumor that is removed during surgery and reducing operating time and expense.
Michael I. Miga, Ph.D., assistant professor of Biomedical Engineering and director of the Biomedical Modeling Laboratory, has harnessed a widely used commercial technique, laser range scanning, to adjust for changes in the surface of the brain as the surgeon cuts into it.
By repeatedly sweeping a laser beam across the brain surface, the scanner produces “point clouds” or sets of three-dimensional points that—in clinical studies—have accurately predicted the changing locations of the tumor as well as nearby blood vessels and other delicate structures during the operation.
The development of these techniques owes much to the rich, longtime collaboration between Vanderbilt engineers, computer scientists and neurosurgeons.
Leaders in this effort include J. Michael Fitzpatrick, Ph.D., and Benoit M. Dawant, Ph.D., professors in the Department of Electrical Engineering & Computer Science; Robert L. Galloway, Ph.D., professor of Biomedical Engineering; and Neurosurgery Department chair George S. Allen, M.D., Ph.D.
During the past 15 years, Vanderbilt researchers have improved the registration of preoperative images with anatomical information collected by an optical probe during surgery. The combined image, projected onto a computer screen in the operating room, helps the surgeon hold a true course through tissue topography that changes with each the touch of the scalpel.
Dawant, who with Fitzpatrick co-directs the Medical Image Processing Laboratory, has developed computational “atlases” of the brain and liver—marked by recognizable anatomical guideposts such as blood vessels—that can be “warped” to fit individual patient cases.
Similarly, Galloway and his colleagues have teamed up with surgeons to create intraoperative guidance systems that use optical tracking or articulated “arms” to track the surgical position in three dimensions.
The primary software platform, called ORION, for Operating Room Image-Oriented Navigation, can be modified to support neurosurgical and surgical applications, including ones such as liver surgery, where the target moves with patient breathing.
All this would not have been possible were it not for the astronomical increase in computing power, Miga adds.
“We solve about 80,000 coupled equations in a model, and we do that in about 15 to 18 seconds,” he says. “And when I say 80,000 equations, I’m talking about essentially a spatial understanding of how the organ we’re looking at is deforming or moving.”
A tracked surgical probe collects data from a brain tumor during surgery (left). The information is used to index pre-operative images to the correct slice of the tumor (right) and to display the surgical position on that slice on a monitor in the OR. The CT image (upper left), PET image (lower right) and two MR images (upper right and lower left) update in real time as the probe is moved.
Hitting the target
Thompson says the next logical step in cancer imaging is to develop a molecular imaging agent, “some unique marker in the tumor that you could target… that would allow the tumor to fluoresce” as the surgeon peers through the operating microscope.
With Vanderbilt chemistry professor Darryl J. Bornhop, Ph.D., Thompson is investigating another class of fluorescent molecules, the rare-earth lanthanide chelates, which potentially could be used to delineate tumors both in MRI scans and under the operating microscope.
Many cancer cells express a high density of peripheral benzodiazepine receptors (PBRs), named for their ability to bind anti-anxiety drugs like Valium and Xanax. To PK-11195, another compound that binds tightly to PBR, Bornhop attaches fluorescing complexes of lanthanide chelates.
In animal studies, the injected marker shows unique, dual capabilities: Bornhop’s hybrid shows up in MRI scans, and its fluorescent tag can be observed through the microscope. If it works in humans, the surgeon could match the fluorescence seen during surgery to the pre-operative MRI.
While PBR targeting also may be useful in the treatment of other tumors, including those of the breast and colon, it alone may not be enough, Bornhop cautions. A “cocktail” of chemicals will probably be needed—especially to monitor how a tumor is responding to therapy.
One possible avenue: cell adhesion molecules, which play important roles in inflammation, cell migration, cell signaling—and cancer.
About a decade ago, Hallahan and his colleagues at the University of Chicago observed that the inner linings of tumor blood vessels sprouted these distinctive glycoproteins (carbohydrate-protein complexes) when zapped by a dose of radiation. He wondered how he could capitalize on this phenomenon.
After moving to Vanderbilt to chair the Department of Radiation Oncology in 1998, Hallahan assembled a diverse team that included Todd D. Giorgio, Ph.D., associate professor of Biomedical Engineering and Chemical Engineering.
The researchers began searching for fragments of proteins—short sequences of amino acids called peptides—that would hone in on tumor blood vessels.
Hallahan hoped the peptides would bind specifically to these radiation-induced markers inside tumor blood vessels. When tagged with radioisotopes, these tiny guided missiles could be used to monitor the effectiveness of drugs designed to shut down the tumor’s blood supply. They also could deliver their own toxic payloads.
The researchers found an amino-acid sequence—arginine–glycine–aspartic acid or RGD—that bound specifically to the markers.
In a preliminary feasibility study, they labeled the peptide with a gamma ray-emitting radioisotope, injected it into patients receiving high-dose radiation to treat metastatic brain tumors, and watched the tumors light up in images taken by a gamma camera. “This study shows that it is feasible to guide drugs to human neoplasms by use of radiation-guided peptides,” they reported in 2001.
Next, the researchers coated “nanoparticles” (about the size of a virus) with fibrinogen, a blood-clotting protein that contains the RGD sequence, tagged the particles with a radioisotope, injected them into tumor-bearing mice, and blasted the tumors with radiation. Not only did the blood vessels light up, but the fibrinogen coating apparently triggered clots to form inside the vessels, blocking blood flow and causing the tumors to shrink.
Currently the researchers are searching for peptides and antibodies that zero in on tumor blood vessels following low-dose irradiation in combination with Sutent (SU11248).
“That’s why it takes so long to get a drug or an antibody to market,” explains Raymond L. Mernaugh, Ph.D., director of the Molecular Recognition and Screening Facility in the Vanderbilt Institute of Chemical Biology who is participating in the research. “You go through all these steps to prove that you have something that’s very specific, doing exactly what you want.”
Sutent is a “targeted” cancer therapy, now in clinical trials, which blocks an enzyme key to the development of tumor blood vessels. Not all tumors or patients respond to targeted therapy, however. Hallahan’s goal is to develop a way of determining within hours, rather than weeks, whether the drug is working.
Meanwhile, Giorgio and his colleagues have identified peptides capable of penetrating the nuclei of cells in the breast, and which potentially can differentiate normal cells from tumors. By attaching gold nanoparticles to the peptides, this method could generate an early and extremely precise view of breast cancer.
Similarly, the recent discovery of neural stem cells could lead to improvements in the early detection and treatment of gliomas. Scientists believe these stem cells, the source of normal brain tissue, under some circumstances can be transformed into tumors.
“Let’s say you could image the stem cells,” Thompson imagines. “Then you could see that your therapy made (the abnormal cells) go away… I hope we would get to a point where if somebody came in with a tumor we’d be able to simply flip the switch and shut it off and keep it from progressing… without having to do surgery.
“It is absolutely changing the way we think about these kinds of cancers.”