Heterogeneity Mapping of Protein Expression in Tumors using Quantitative Immunofluorescence

Here we describe a method to quantify molecular heterogeneity in histological sections of tumor material using quantitative immunofluorescence, image analysis, and a statistical measure of heterogeneity. The method is intended for use in clinical biomarker development and analysis.

The overall goal of the following experiment is to accurately quantify biomarker expression across whole tissue sections of human cancer in order to assign a numerical value to the degree of heterogeneity in protein expression. This is achieved by first scanning immuno fluorescently labeled sections of cancer tissue using a whole slide fluorescent scanner. Next regions of interest within the tumor are selected for evaluation.

Then the expression of the protein of interest within the invasive tumor areas is quantified using the aqua analysis automated image analysis system. Ultimately, this method can be used to map biomarker expression in cancer tissue, which is often very heterogeneous across the tissue section. The main advantage of using these techniques over traditional methods such as immunohistochemistry using color and metric visualization reagents is that immunofluorescence is more quantitative and more sensitive.

Coupling these analytical methods with a whole section, immunofluorescence scanning enables fine-grain mapping of subtle variations in biomarker expression across tissues. For the first time, demonstrating these procedures is Mark Gustafson a direct director of operations at the Histo RX laboratory. After microtome, sectioning of a cancer tissue biopsy and immunofluorescent staining of the specimen for expression of the biomarkers of interest, load up to five slides of the stained sections into the slide cassette of the slide scanner.

Then using the interface in the scan scope console. For each slide, select a region to encompass all the tissue on the slide, regardless of the staining pattern or other observable aspects of the sample. Next, using the scan scope console, first, select a region of tissue to evaluate ideally within the tumor region of the tissue to provide the best representation.

Now using the scan scope console auto exposure feature, determine the exposure time for each channel along with the image focus, use the blue diamond on the scan scope console image to select an additional region away from the tissue, yet still within the region of the slide under the cover slip to define the background area where the flat field correction images will be acquired. For each filter add focus points to the tissue area represented by yellow squares to optimize image capture. If the images appear to have high background or other image artifacts, the image region should be moved and new images acquired.

The tissue is then scanned and the acquired digital slide images are then uploaded automatically into the Aperio spectrum database. Double click the thumbnail image for the spectrum program to select the slide for analysis from the digital slide list in the spectrum database. Then using the accompanying h and D annotated image, annotate the region of interest on the fluorescent image.

Multiple regions of interest delineated with individual circles can be selected on a single sample. The annotated images are saved automatically to the spectrum database. Now select Analyze from the menu bar at the top of the screen when the analysis window appears, select histo RX aqua analysis from the pull down menu.

When ready to begin, press the analyze button, A minimized console window will appear in the Windows task bar displaying the progress of the transfer of the images from the Spectrum database to the local computer. Once the data has been transferred, aqua analysis will launch to generate an aqua score using an unsupervised aqua scoring algorithm based on the data clustering of a sample first threshold, the pan cytokeratin image signal above the background image signal. This will allow use of the pixels to define a tumor mask that then can be used for subsequent calculations.

Use the high expression pan cytokeratin pixels to also define the cytoplasmic or non-nuclear regions of the tumor cells. As seen here, the pixels with a high signal in the DPI channel that are within the tumor mask are identified as nuclear in nature. If upon review fields that were not suitable for scoring due to sample or imaging artifacts are identified, those fields are redacted from scoring and also will produce a final result of fail.

The results of the aqua scoring are represented as a table of scores associated with each 512 by 512 pixel tile in the region of interest within a whole tissue section sample, which can be saved and used for subsequent statistical analysis. These hemat toin, and eoin stained photo micrographs demonstrate how different areas of the same ovarian cancer can exhibit variable morphology. The high power view on the top right of the Figure B highlights cells with cytoplasmic clearing and a solid pattern of growth in the upper part of the tumor.

The lower part of the tissue magnified in the bottom right section of the Figure C, however, has a more homogeneous eosinophilic cytoplasm and a papillary growth pattern. Here, heterogeneity heat maps of ER expression in a tissue section of ovarian cancer before and after a chemotherapy treatment are shown in the upper and lower panels respectively. Note the change from variable ER expression across the section with regions of both high and low expression in the upper figure to an even distribution of high ER expression illustrated in red in the lower figure.

The numerical representation of this change is denoted on the left by the decrease in the Simpsons index. In this figure heterogeneity heat maps of HER two in a tissue section of ovarian cancer are shown again before and after therapy as seen in the upper and lower panels respectively. Note again, the change from variable HER two expression across the section with regions of both high and low expression in the upper figure to an even distribution of high HER two expression in the lower figure.

Once again, the Simpsons index is seen to decrease. While attempting this procedure, it's important to remember that the quality of the data generated by the computational analysis is only as good as the quality of the staining, and this must be of a high technical standard.