Share
Blog
Cite
Print
Email
BookmarkScienceDaily (Sep. 25, 2008) — Researchers have demonstrated a new instrument that makes it possible to detect and quantify multiple different clinically important proteins in a single tumor sample using conventional staining. Currently, pathologists usually need a separate tissue slice for each protein they want to examine, making it impossible to see how molecules interact within individual cells.
See also:
The process is not only fast and automated – up to 180 tumors from different patients can be analyzed in an hour by a computer – it also provides more information than is available today to help therapy development and direct treatment, say a team of scientists led by Cambridge Research & Instrumentation, Inc. (CRi), in Woburn, Mass, which developed the method.
"This technology is designed to be used by pathologists to reveal new data that can help researchers develop targeted therapies, and physicians personalize treatment for individual patients," said Clifford Hoyt, vice president and chief technology officer at CRi. He presented the results at the AACR's Molecular Diagnostics in Cancer Therapeutic Development meeting being held in Philadelphia, September 22 - 25.
For example, researchers say the instrument, which uses multispectral imaging, can detect how many of the cells in a sample of breast cancer display what quantity of any of four different receptors - progesterone receptor (PR), estrogen receptor (ER), HER1 receptors and HER2 receptors. "That might show us, for example, that 25 percent of cells express both ER and PR, 50% either PR or ER, and 25 percent neither."
Now, however, pathologists can only stain a sample for a single marker, and two serial sections might show half of the first sample is positive for ER and half of the second sample is positive for PR, Hoyt says.
"You wouldn't know from these serial sections that some cancer cells express both receptors, and that can have implications for treatment," he said. "This shows us complex protein expression and interaction patterns in a single tumor section."
Even the most advanced microarray-based protein analysis being tested today loses this key molecular phenotype distribution information, Hoyt adds.
Breast cancer markers are just one example of what the instrument can read, he adds. "Pick any four stable proteins found in tissue and we can analyze a cocktail of protein markers all at once," he said.
Currently, researchers and pathologists use immunohistochemical staining to examine tumor samples using a microscope equipped with a color camera. Color cameras detect three visible wavelengths, red, green and blue, which mimics the human eye.
This new instrument uses a CRi NuanceTM multispectral imaging camera, which captures information from multiple wavelengths in the visible and infrared, an automated microscope, and novel machine learning-based software to extract data from images. This is an advance over purely visual analysis, or the use of color cameras, which do a poor job of disentangling multiple colored protein labels when they are spatially overlapping.
"Our camera looks at samples with 10 to 30 different wavelengths, for staining of up to four proteins using different colors," Hoyt said. "With this technique we can unmix multiple different labels from what would otherwise be a muddy mass of color."
In this study, the researchers, which included scientists from Novartis Institutes for Biomedical Research and the British Columbia Cancer Agency, stained samples from 356 patients for, ER, PR and HER2 proteins, and then for PR, HER1 and HER2. Results of protein expression using the multispectral imaging were then blindly compared to results from visual assessments by pathologists on samples that were serially stained. Agreement between the visual and automated analyses was "the same as we see among pathologists, which is a very good result" Hoyt said.
The analysis shows which proteins are being expressed and the expression level. "It is a quantitative read-out, so it provides a very specific molecular profile," Hoyt said.
