The ravages of Alzheimer's disease now afflict four million people in the U.S., a number expected to triple in the decades to come. The current costs of dealing with Alzheimer's-nearly $70 billion-don't have to triple, however. The increasing sophistication of imaging technology and a greater understanding of the molecular underpinnings of the disease are enabling physicians to catch it earlier in its course, even years before symptoms surface, giving patients the opportunity to take preventive and progression-slowing drugs.
Advances of similar magnitude are also stretching the possibilities in detecting and treating other neurodegenerative diseases, cancers of all types, vascular diseases, osteoporosis, and arthritis. Many improvements hinge upon the progress made in genomics and proteomics to pinpoint the biomarkers of genetic mutations and rogue proteins that cause diseases. Scientists are using these discoveries to devise new imaging techniques that radiologists can apply to visualize the precise molecular signature of disease, opening the door to earlier, more accurate diagnoses and drug targets of improved specificity and efficacy.
The medical community stands at a turning point in the comprehension, diagnosis, and treatment of disease, according to Michael Phelps, Ph.D., chair of molecular and medical pharmacology at the University of California, Los Angeles.
"The imaging of biomarkers will provide a diagnosis based on how cells have been reprogrammed into disease," he said. "This helps establish new ways for how molecular therapeutics are developed, drugs are selected for patients, and therapeutic responses are assessed."
HUNTING BIOMARKERS
The most likely biomarkers to image will be proteins because of their status as the actual functioning culprits in disease. Yet imaging biomarkers can comprise any anatomic, physiologic, biochemical, or metabolic parameter that can be detected and measured with an imaging agent, according to Dr. James Thrall, radiologist-in-chief at Massachusetts General Hospital and cofounder with Dr. Gregory A. Sorensen of MGH's Center for Biomarkers in Imaging. To be useful, a biomarker must have a tight coupling to the disease process.
"Many potentially interesting molecular biomarkers are also molecular imaging agents," Thrall said. "They tell us the size of something or the metabolism of something."
They also provide detailed information on location, which is not possible with biomarkers obtained through blood, urine, cerebrospinal fluid, and other bodily fluids.
"Bodily fluids are going to be the most common entry point," said Dr. King Li, chair of the imaging sciences program at the National Institutes of Health. "But if the treatment is site-specific, you need to locate where the disease is, and hopefully you can use the biomarker as a hint to develop a targeted imaging agent."
Even the FDA is embracing efforts to image biomarkers. It sees them as a way to revitalize the stagnation of the "biomedical revolution," which had once offered the promise of new drugs and therapies.
"Properly applied, [genomics, proteomics, bioinformatics systems, and new imaging technologies] could provide tools to detect safety problems early, identify patients likely to respond to therapy, and lead to new clinical endpoints," said the FDA's March 2004 report, Challenge and opportunity on the critical path to new medical development.
The new strategy promises to improve the results of clinical trials that preselect patients based on their biomarkers. Iressa, a kinase inhibitor developed to treat non-small cell lung cancer, shrank tumors in only 10% to 15% of patients during the clinical trials. In those who responded, however, most tumors rapidly shrank 80% to 90%, said Mary Lynn Carver, director of oncology public affairs at AstraZeneca, developer of the drug.
In May, researchers from Dana Farber Cancer Institute and Massachusetts General Hospital announced what they had found in the genes of these "super-responders." Publishing their independent discoveries in Science and The New England Journal of Medicine, respectively, the researchers pointed to a mutation in the gene for epidermal growth factor receptor (EGFR), which made the super-responders "exquisitely sensitive" to Iressa, Carver said.
Expanding the ability to image biomarkers such as mutant EGFR will require contribution from several fields of study and collaboration and communication among researchers.
"The secret to making this work is for groups of people to come together and be good teachers and good students of each other," Phelps said. "Together, they will create a science and medicine that does not exist today."
To bring the various components of molecular medicine together, Phelps laid the framework for UCLA's molecular pharmacology department, its Institute for Molecular Medicine and the Crump Institute for Molecular Imaging.
"Physical and biological sciences are building a new systems biology view of disease-how cells and intercellular connections are programmed from the genome into systems of protein-based cell circuits and networks . . . [and] are the new targets of molecular imaging, which are key to linking imaging from discovery to patient care," he said.
Computational and computer scientists crunch the results of genomics and proteomics research by analyzing its numeric and sometimes unwieldy data. Drug developers use the findings to pursue small molecules or macromolecules that can bind to the uncovered biomarkers. Collaboration of biologists, chemists, and physicists helps tackle the job of developing ways to visualize the biomarkers, first in mice, possibly in nonhuman primates, and then in humans.
"Radiologists are the team leaders," Li said. "We have been looking at biodistribution and contrast agents throughout our entire careers, so we have more insight into drug and contrast delivery."
FREE DELIVERY
Li's insight will serve radiologists well as they lead the way in surmounting one of the biggest challenges facing molecular imaging: the delivery of imaging agents, along with the molecular probe, to the desired biomarker. To reduce background noise and to increase accuracy, probes need to be specific enough to bind only to the target.
"[The imaging probe] becomes subject to all the pharmacokinetic rules and constraints that govern the concentration of 'drugs' in plasma, including absorption, distribution, metabolism, excretion, and other factors within the vascular compartment," said Dr. Sanjiv Gambhir, director of the molecular imaging program at Stanford University, in a review published in Genes and Development.
Cell membranes and the blood-brain barrier can be formidable obstacles for molecular imaging probes of all sizes. To study intracellular biomarkers, the probe must be made lipophilic enough to traverse the lipid bilayer membrane or be transported by a transporter within the cell membrane. Some researchers are investigating the use of translocation signals to stimulate shuttling of the probe across the membrane and of polyethylene glycol to increase lipophilicity.
Lipophilicity is also important for crossing the blood-brain barrier. To visualize neurotransmitter receptors, amyloid accumulations of Alzheimer's disease or other neural biomarkers, researchers are looking at various metabolites and radiotracers as possible solutions.
"Researchers have longed for a way to visualize and quantify amyloid deposition in the living human brain," said Eric Reiman, director of the Arizona Alzheimer's Disease Consortium. "Since Dr. Alois Alzheimer's original discovery in 1907, we have had stains that detect and measure amyloid in the postmortem brain, but we need PET or SPECT radioligands that not only bind like stains but also get past the blood-brain barrier."
Reiman cited promising probe candidates. One, fluoroethyl(methyl)amino-2naphthyl ethylidene malononitrile (FDDNP) labeled with fluorine-18, was developed at UCLA. While preliminary studies have been encouraging, he said that continued refinement will improve precision such that the probes can allow for accurate quantitative measurements of amyloid burden. Research shows that when FDDNP binds to its target, it can block the buildup or accelerate the breakdown of amyloid.
The endothelial cells of the blood vessel walls constitute another daunting obstacle to observing biomarkers tucked within tissue parenchyma such as that of tumors or organs. Li has been working on applying physical force such as ultrasound energy to break the hold of intercellular junctions that glue endothelial cells together. Such force, or "shaking," optimized to do the job without killing the cells or encouraging metastases, produces leaks in the blood vessel and allows probes to seep into the tissue, Li said. With promising results in animal models, clinical trials could begin in the next few years.
BLOOD COMPLICATIONS
Visualizing biomarkers that may be accessible in the lumen of blood vessels incolves other complexities as well. Interstitial pressures and other factors can interfere with the long voyage to the target, mandating that the probe have a half-life that's long enough to allow this process to occur.
To improve the odds of producing an adequate signal at the target site and delivering enough probes there, Li has been studying polymerized vesicles, which are composed of polymerized lipids and lipids that are chelated to gadolinium. These vesicles counteract the instability of pure liposomes while avoiding the clumping and tangling of pure polymers. The resulting vesicle is round and flexible, and it interacts well with surfaces where biomarkers may be located.
Antibodies specific to the biomarker can be attached to the vesicle and injected into the blood. Studies of the polymerized vesicles using MR imaging or gamma scintigraphy and probes targeted to specific vascular receptors of tumor in mice and rabbits have been successful.
Li estimated that by using polymerized vesicles, the amount of imaging agent accumulating at target sites increases by at least five orders of magnitude over antibodies without the vesicles.
"The vesicles have multiple chances to interact with the target as well as the capacity to carry a large amount of contrast agent per particle," he said. "All new vessel formation induces molecular markers on the endothelium, which means that this can be a fairly generalized approach to all tumors."
The liver, spleen, and bone marrow also take some of the polymerized vesicles, highlighting the challenge created by blood's flowing and ebbing. Such nonspecific uptake, which occurs with most delivery vehicles, reduces specificity.
PUMP UP THE VOLUME
Even if successful at binding to the target, a molecular probe has little use unless it can be detected. This is especially a concern when the density of biomarkers, which may be in the form of receptors or other structures, is low.
"One of the big initiatives in supporting molecular imaging is to learn more about signal amplification to improve detectability," Thrall said. "If the basis of localization is the one-to-one docking of an imaging molecular probe to the target molecule, then you don't get amplification."
Antibodies have long been touted for their specificity. But only about 0.001% to 0.01% of monoclonal antibodies that are injected intravenously will reach and bind to target tissues in humans, Li wrote in an editorial published in Academic Radiology. With so few arriving at the target, not enough imaging agent accumulates to be detected.
Imaging protein biomarkers is still more feasible than targeting mRNA biomarkers with labeled oligonucleotide probes, however. Mutated proteins could be present in amounts of 100 to one million copies per cell, while using the antisense approach would mean attempting to image merely 10 to 100 copies of mRNA.
The solution lies in the accumulation of the imaging agent at the target site or requires that the target biomarker, such as an enzyme, cause amplification of the imaging agent, according to Thrall. In the case of PET, the target cells can take in multiple tracers such as FDG because they are phosphorylated and then unable to leave the cell. The rate at which the probe accumulates and is processed by glycolysis enzymes produces information on pathology.
Another approach also relies on enzyme action. Gambhir works with enzymes whose substrate is the molecular probe. The herpes simplex virus-1 enzyme thymidine kinase, which inhibits cell replication, amplifies the signal by phosphorylating the tracers, which are then trapped in the cell.
"When each enzyme molecule in turn traps a large number of substrate probes, [it leads] to the greatest degree of signal amplification," he said.
Gambhir and his colleagues at UCLA and in Spain are conducting a clinical trial with the procedure to detect and treat pancreatic and hepatocellular cancer. By injecting cells containing F-18 fluoro-penciclovir as a gene construct, imaging studies can provide information about the expression and presence of HSV1-tk and its tumor-killing activity.
Other methods of signal amplification are also under study. Laboratory researchers have long relied on avidin-biotin amplification for in vitro signal amplifications. Now it is being harnessed for use in human subjects.
INTO INNER SPACE
Improvements in signaling will continue to boost advances in the machinery behind targeted imaging. CT, MR, ultrasound, and PET represent launching pads into the cellular cosmos.
"These techniques are complementary," said Simon Cherry, Ph.D., a professor of biomedical engineering at the University of California, Davis. "You can do things with PET that you can't do with optical and MR imaging, and vice versa."
PET/CT provides information on both biochemical and anatomic pathology at the same time. Efforts to combine MR with PET and to merge optical imaging with PET are ongoing. Phelps described the combination of these technologies as a great example of bringing imaging disciplines and multiple types of information together, explaining that PET/CT delivers molecular imaging in the context of an anatomic image, which is more familiar to radiologists.
The problem of penetration must be solved to make other techniques more amenable to molecular imaging. Advanced contrast agents, such as fluorescence, allow for the simultaneous imaging of multiple biomarkers in cells and small animals. But detecting optical signals through the thick layers of human flesh is still a challenge, according to Cherry, who developed microPET for imaging small animals.
REVOLUTIONIZING CURES
Such efforts may bring radiologists' eyes straight into the cells of their patients to provide information about the most appropriate therapies. No longer will physicians and their patients need to waste time and limited finances testing therapies of uncertain efficacy. Moreover, the refinement of molecular imaging will give direct feedback about how patients are faring during their treatments, including gene therapy, chemotherapy, and drug regimens.
Ongoing innovation will provide the expertise to tether imaging agents to most small molecules, antibodies, other macromolecules, or even nanoparticles that are designed to bind and hinder disease-causing biomarkers. Physicians could then accurately assess treatment effectiveness.
Molecular imaging may make its most significant impact in the continuing struggle to solve the challenge of gene therapy, which requires a way to follow the therapeutic gene construct and determine its arrival at the target site. Including a reporter gene that encodes for a molecule that can be imaged has proven to be a good tactic in visualizing both the arrival and expression of the therapeutic gene.
Chief among findings in the genomics world are the single nucleotide polymorphisms that dictate how individual patients respond to a drug or an environmental insult or their susceptibility for developing disease. Instead of formulating drugs to treat relatively broad disease categories and symptoms, the pharmaceutical industry will find itself designing therapies aimed at smaller populations who possess certain genotypes and patients in evolving stages of disease.
"You don't get most diseases, you develop them over time," Phelps said.
Study Assesses Lung CT-Based AI Models for Predicting Interstitial Lung Abnormality
September 6th 2024A machine-learning-based model demonstrated an 87 percent area under the curve and a 90 percent specificity rate for predicting interstitial lung abnormality on CT scans, according to new research.