A team of scientists at Vanderbilt University have achieved the first “image fusion” of mass spectrometry and microscopy, which, they say, could dramatically improve the diagnosis and treatment of cancer, among other advancements.
Microscopy yields high-resolution images of tissues, but “it really doesn’t give you molecular information,” said Richard Caprioli, PhD, the Stanford Moore Professor of Biochemistry and director of the university’s Mass Spectrometry Research Center. Mass spectrometry provides a very precise accounting of the proteins, lipids, and other molecules in a given tissue, but in a pixelated manner. Combining the best features of both imaging modalities, he says, allows scientists to see the molecular make-up of tissues in high resolution. (See Figure 1)
One of the ways this hybrid technology is a game-changer, Caprioli said, is in cancer detection, where the technique could redefine the surgical margin—the surgical line between cancer cells and normal cells.
Currently, the margin is determined by histology, the appearance of cells examined under a microscope. But many cancers recur after surgery. That could be because what appear to be normal cells, when analyzed for their protein content using mass spectrometry, are actually cancer cells in the making.
“The application of image fusion approaches to the analysis of tissue sections by microscopy and mass spectrometry is a significant innovation that should change the way that these techniques are used together,” said Douglas Sheeley, ScD, senior scientific officer in the National Institute of General Medical Sciences (NIGMS).
“It is an important step in the process of making mass spectrometry data accessible and truly useful for clinicians,” he said. Using a mathematical approach called regression analysis, the researchers mapped each pixel of mass spectrometry data onto the corresponding spot on the microscopy image to produce a new, “predicted” image.
It’s similar in concept to the line drawn between experimentally determined points in a standard curve, Caprioli said. There are no “real” points between those that were actually measured, yet the line is predicted by the previous experiments. In the same way, “we’re predicting what the data should look like,” he said.