‘Virtual’ flow cytometry: objective, yet accessible
A new method could provide an important quantitative tool for hematopathologists, who diagnose diseases such as lymphoma and leukemia by looking at immunohistochemically stained cells, for example. The practice is traditionally subjective, with the clinicians typically estimating the percentage of stained cells either visually or manually. Consequently, the possibility of error is relatively high.
To minimize it, researchers with the H. Lee Moffitt Cancer Center and Research Institute at the University of South Florida in Tampa, and at New Brunswick, N.J.-based IHCFlow-GreenGreat have developed a method and instrument that enable clinicians to perform image analysis of immunohistochemically stained cells. As reported in the Jan. 15 issue of Cytometry Part B, the method converts immunohistochemistry data to a two-parameter dot-plot display, much like those found in flow cytometry. Thus it provides a sort of “virtual” flow cytometry, offering the objectivity of the latter method but with relatively inexpensive tools that are readily available to most hematopathologists.
The software developed by the researchers generates a two-parameter dot-plot display, similar to those found in flow cytometry, offering the objectivity of the latter method but with tools generally available to hematopathologists and other clinicians.
Immunohistochemistry techniques typically do not provide population-based statistics, with positive and negative cell counts. “The current paradigm is to calculate the antigen density in particular unit areas, or ‘hot spots,’” said Hernani D. Cualing, a researcher with the cancer center and the first author of the paper. “This method departs from that paradigm by looking at tissue cell by cell. It provides cell-based analysis.”
Researchers have developed a method for converting immunohistochemistry data into objective, flow cytometrylike information. The method could help minimize the possibility of error when diagnosing lymphoma and leukemia using immunohistochemically stained slides — such as the one shown here, with red AEC staining of cancer lymphocytes with Hematoxylin counterstain.
To be able to extract flow cytometry-like results from immunohistochemically stained cells, the researchers had to address a number of issues; for example, avoiding the inherent variability of manual immunostaining techniques and determining the appropriate sampling resolution to obtain both size and staining information for each cell object.
The most challenging task, Cualing said, was resolving the overlap between populations of cells: between two immunostained or two nonimmunostained cells, or between individual immunostained and nonimmunostained cells. One way to address this issue was to cut the samples thinner. In fact, the researchers sectioned the tissue with a microtome set at 2 μm. However, they also employed an algorithmic approach: a patent-pending advanced adaptive method using a combination of spectral, contrast, size and shape properties of individual images/cells.
They demonstrated the new method by analyzing 14 lymph node samples, using a panel of seven monoclonal antibodies. After fixing the samples in formalin, they cut sections 2 μm thick and oven-dried them on positively charged slides. Finally, they performed immunohistochemistry using an automated system made by Ventana Medical Systems of Tucson, Ariz.
They examined the immunohistochemically stained slides with a brightfield microscope made by Leica of Westlar, Germany, outfitted with a 20×, 0.4-NA objective. A color CCD made by Diagnostic Instruments, also of Westlar, captured the images. Each image was 512 × 474 pixels, with 1.5 pixels per micron. They analyzed the images with software prototyped and developed by IHCFlow-GreenGreat.
The same samples were analyzed with flow cytometry. The results of these measurements correlated well with those obtained using the virtual flow cytometry technique, suggesting that hemato-pathologists could confidently use the latter technique for diagnosis or prognosis when only tissue immunohistochemistry is available — that is, when fresh sample is not.
Other groups have reported methods of separating positive stain from nonstained pixels, thus obtaining objective information from stained tissue samples. Some have described hybrid microscope flow cytometers, for example; others have applied immunofluorescence techniques on slide substrates and used laser scanning instrumentation to acquire information about cell populations. Such approaches typically require complex —and expensive — technology not commonly found in pathologists’ and oncologists’ offices and labs.
The investigators continue to develop the technique and are preparing to present results demonstrating its efficacy with different microscopes and under different lighting conditions, using additional antibodies. “We were able to show excellent correlation between these variables,” Cualing said.
He formed IHCFlow and partnered with GreenGreat to develop the algorithm and software used in the study and is working to bring the software to market. “With my partners, I am currently in communication with a technology company to commercialize the technology for clinical use,” he said, “to help hospitals and hematopathologists do immunohistochemistry analysis more accurately, using ‘evidence-based medical diagnosis,’ and to try to port the tools and standards of flow cytometry to immunohistochemistry.” He added that they have produced a prototype of the software in Java.
Contact: Hernani D. Cualing, H. Lee Moffitt Cancer Center and Research Institute at the University of South Florida, and IHCFlow-GreenGreat; e-mail: email@example.com.
MORE FROM PHOTONICS MEDIA