A third of patients who undergo surgery to remove cancerous tumors end up with microscopic pieces left behind. These overlooked remnants can lead to the recurrence of cancer after what was thought to be a successful surgery.

Two surgeons at the University of Pennsylvania have joined forces to try to solve this problem. Their solution: making tumors glow. Using a combination of injectable dyes and high-resolution cameras, the surgeons found a way to image tumors during surgery and more easily identify their margins. They published their findings in July in the journal PLoS One.

"Once you get into surgery, you've only got your eyes and your hands to tell you where the margin of the cancer is," said David Holt, lead author of the study and professor of surgery at the Penn School of Veterinary Medicine. Tumors can be studied before surgery with MRIs or CT scans, but these images are helpful only to a certain extent, he said.

After an initial study in mice, Holt used the new imaging technology on eight dogs with naturally occurring lung tumors. Holt and his team injected the dogs with indocyanine green, a dye that accumulates in tumors, more so than other tissues, due to their leaky blood vessels. The dye can't be seen directly because it emits light in wavelengths that aren't perceivable by human eyes.

But when the doctors shined a near-infrared light on the tumors during surgery, they glowed an "Incredible Hulk green" on computer screens hooked up to their cameras, Holt said. With the help of this real-time image, Holt could make decisions about where to make incisions during surgery on the dogs.

Even in the precision-driven field of surgery, "a picture's worth a thousand words," Holt said.

The successful dog study led to approval for a human clinical trial run by another surgeon, Sunil Singhal, a coauthor on the study and assistant professor of surgery at Penn's Perelman School of Medicine. Using the same technology, which he compared to night-vision goggles, Singhal examined lung and chest tumors in five human patients. Just as in the dog study, the tumors strongly fluoresced under the near-infrared cameras.

Singhal also used this imaging approach in another study published last month in the Annals of Thoracic Surgery. In two out of 18 patients with lung cancer, the imaging system helped surgeons identify cancer in areas that had previously gone undetected.

"For those two people, it was life-altering," Singhal said. In cases like these, other imaging techniques or visual inspection of tissue might have given patients a Stage I diagnosis when they actually have Stage IV cancer. Singhal's goal is to use this technology to more accurately gauge the severity of cancers as well as reduce the recurrence rate after surgery.

Singhal's expertise led him to start exploring this technology in lung cancer, which kills more people in the United States than any other type of cancer. But he is already exploring its applications in other cancers, including breast cancer. He has seven open clinical trials using dyes to image tumors during surgery.

This type of imaging doesn't involve radiation, making it safe to use for most patients, even pregnant women. But this limits how deeply into tissues surgeons can see. The dye Holt and Singhal used can be detected only at depths of 10 to 15 millimeters in tissues, leaving deeper areas unexplored.

Also, the dye does not specifically bind to tumor cells; thus, adjacent tissues can also collect dye, especially if they are inflamed. As tumors get larger, they tend to inflame surrounding tissues, Singhal said, leading to blurred margins on the images.

"That's a major issue" with indocyanine green, said Zhen Cheng, associate professor of radiology and director of the Cancer Molecular Imaging Chemistry Laboratory at Stanford University. Imaging with dyes that selectively target tumor cells is more important, he said.

Singhal and Holt plan to use targeted dyes in future studies that are molecularly selective for tumor cells as well as continuing to improve their camera technology. Their work is not ready for mass use. But Singhal hopes it will ultimately lead to more sophisticated approaches, including three-dimensional holograms of tumors generated in real time during surgery.