Pictures of the Future Fall 2008
Early Detection of Diseases – Interview
Detecting Cancer Cells with Light
Interview with John V. Frangioni
John V. Frangioni (44), M.D., PhD, is an Associate Professor of Medicine and Associate Professor of Radiology at the Beth Israel Deaconess Medical Center (BIDMC) and Harvard Medical School, in Boston. His academic training includes an Engineering Sciences degree from Harvard College. Frangioni is co-founder of the BIDMC Center for Molecular Imaging and the Longwood Small Animal Imaging Facility. His laboratory is focused on solving clinical problems through the application of advanced engineering and chemistry.
How do surgeons know which lymph nodes need to be taken out when resecting a tumor?
Frangioni: Lymph nodes are the body’s first line of defense, and are the most likely place for cancer cells to migrate from a primary tumor. The problem is that until now there was no way of knowing which lymph nodes were at highest risk. This is the reason why surgeons prefer to resect all likely nodes. Take breast cancer, for instance. There are around 30 lymph nodes in a woman’s armpit. What the surgeon wants to know before removing the tumor itself is: which of these 30 nodes the tumor drains into directly. If you can identify those so-called sentinel nodes—and a woman typically has two or three per tumor—then you can limit the degree of surgery significantly.
Have you developed an imaging system that promises to do that?
Frangioni: Yes we have. We call it Fluorescence-Assisted Resection and Exploration—or FLARE for short. It consists of two components. First, you bind a fluorescent substance to human albumin to form a molecule that is just the right size to be caught by lymph nodes. When injected near a tumor while the patient is on the operating table, this substance flows to the tumor’s sentinel lymph nodes and concentrates in them. Within seconds of the injection, the nodes light up on a monitor, allowing the surgeon to see them perfectly. To make them visible, FLARE makes use of two cameras: a color video showing the actual surgical area and a near infrared image that sees only the fluorescent material. A third IR camera is available for more advanced applications.
Where does Siemens come into the picture?
Frangioni:The key to the clinical use of this technology is the ability to align these images perfectly in real time. And that’s where Siemens comes in. The company has provided this ability with the software they wrote. The acquisition and fusion of images from three separate cameras in real time is a tour de force. From a clinical point of view, this opens up amazing possibilities, such as being able to see the electrical activity of the heart exactly where it takes place.
What knowledge have you gained from your results?
Frangioni: Based on our animal studies, we believe we will be able to greatly increase the speed and accuracy of many surgical procedures. But this remains to be proven in clinical trials.
Your laboratory has also developed a kind of automated microscope…
Frangioni: After a potentially cancerous tissue has been resected it is sent to pathology. But suppose there are only a few cancer cells hidden somewhere inside it. With today’s staining technology alone, you’d never find them. So what we’ve done is to develop a way of analyzing such slides by using a selection of infrared channels that can identify cancer cells by detecting differences in wavelengths associated with abnormal oxygen and water concentrations. And by marrying this invisible IR light with visible light we bring new information—such as the ability to see individual cancer cells in their anatomical context—into the pathology laboratory’s workflow. The machine itself can be loaded with, say, an entire lymph node or prostate in five-micron-thick slices. It digitally assembles the resulting microscopic images into a virtual macroscopic whole that a pathologist can zoom in to, and out of, with a resolution of one micron. This allows the pathologist to assemble a much more precise report for the patient’s attending physician than is possible at present. And that, in turn, helps to ensure more precise treatment management.
Does the system learn to recognize the patterns that characterize cancer cells?
Frangioni: That’s our goal. A learning algorithm developed by Siemens could potentially grow into a powerful diagnostic decision support tool for pathologists. Imagine a system that supports pathologists by presenting a report to them that allows them to use the stain data they are used to like a Google map to locate nests of cells that have been flagged as potential cancer points. This would result in earlier detection and improved treatment.
The automated microscope could also sharpen MR’s ability to detect cancers at an early stage. We are now at the stage where we are detecting cancer cells in histopathology. The next step will be to develop a correlation between this and corresponding MR data. This will tell us what cancer cells look like in MR images. The resulting knowledge may soon make it possible to detect early-stage cancers with MR alone.
Interview conducted by Arthur F. Pease