NEW YORK (360Dx) – Scientists at IBM Research have developed a microimmunohistochemistry technique that could offer improved quantitation and multiplexing compared to traditional IHC assays.
The approach, detailed in a study published this month in Nature Biomedical Engineering, makes use of a microfluidic probe developed by the IBM researchers that can apply antibodies to tissue samples across areas as small as 400 microns.
According to Govind Kaigala, a research scientist at IBM and senior author on the study, this allows researchers to apply antibodies in adjacent portions of the tissue using a variety of different staining conditions (for instance, different incubation times or different antibody concentrations), which then enables them to collect data on the kinetics of the antibody-target reaction that can be used to improve an IHC assay's quantitative accuracy.
IHC is commonly used in pathology to detect biomarkers that can be used for diagnosing disease or stratifying patients and guiding their therapy. Typically, antibodies are used to detect these markers, but, Kaigala noted, these assays are not rigorously quantitative.
He cited the example of scoring guidelines for the breast cancer protein marker HER2, which is a measurement to determine whether patients should be treated with drugs like trastuzumab (marketed by Genentech as Herceptin) that target the protein. Patient tissue is scored on a scale ranging from 0 to +3 according to the intensity of HER2 staining or the percentage of cells with HER2 present in the tissue. A score of +3 indicates that a patient should be treated with an anti-HER2 therapy; a score of +2 is typically equivocal, requiring further testing; and scores of +1 and 0 indicate a patient is not a good candidate for anti-HER2 drugs.
These scores, however, "are only semi-quantitative," Kaigala said, noting that a number of factors, ranging from sample preparation to scoring and interpretation introduce variability that limits the ability of these assays to provide quantitative information.
In their paper, Kaigala and his coauthors observed that IHC assay standardization efforts and tools like digital pathology are helping to address this issue, but they suggested that collecting information on the kinetics of the antibody-antigen reactions used in an IHC experiment might further drive it toward being a quantitative technology.
Kaigala said he and his colleagues realized that the microfluidic probe they had previously developed could enable such an approach. Initially, they used the probe's ability to precisely deliver antibodies or other reagents to small sections of tissue to study the optimal staining conditions for different assays.
"We started out looking at whether we could do parameter optimization to create a range of conditions in one section of the tissue and then use those [optimized] parameters to process the rest of the section so that you get better signal to noise," he said.
They saw, though, that by staining several small sections of tissue, each according to different parameters (varying, for instance, the antibody concentration used or the incubation time), they could collect data on the kinetics of the antibody-antigen reaction under these varying conditions, from which they could extract more accurate information on the quantity of the target antigen present in the tissue sample.
In traditional IHC, "you do [for instance] 10 minutes of incubation with the primary antibody, then you do the secondary antibody incubation, the [signal] amplification, and then the pathologist looks at the slide," Kaigala said. "But the problem with this is you lose information on the dynamic variation [of the experiment], of what happens during the binding."
He noted that in theory researchers could use fluorescently labeled primary antibodies to monitor the evolution of the signal over the course of the incubation period but that this approach would require changes in the standard pathology equipment and workflow, making it unlikely to be adopted.
"If we want to influence the way pathology is implemented in the diagnostic facilities in hospitals and in pathology labs, we can't come with a scientific concept that is incompatible with the infrastructure they currently have," he said. "We asked whether we could leverage the tools that already exist but just makes some fine tweaks to the workflow."
While the researchers' microfluidic probe is not part of the standard pathology repertoire, "everything else — the type of chemicals used, the antibodies used, the post-processing, the downstream analysis — is identical," to current IHC techniques, Kaigala said.
In the Nature Biomedical Engineering study, the researchers used the approach to analyze 30 breast cancer samples using the standard IHC breast cancer panel of HER2, estrogen receptor (ER), and progesterone receptor (PR), building a machine-learning based algorithm that used the kinetic data to place samples into the traditional scoring HER2 categories of 0, +1, +2, or +3. They noted that in the future the algorithm could be trained against a larger number of classes, potentially providing a more finely graded analysis of a patient's HER2 status.
Additionally, the researchers used the approach to profile different portions of patient samples, finding that the kinetic information varied across these different regions. This, they noted, suggests the approach might be useful for assessing tumor heterogeneity, which they wrote "is often lots in whole-tissue IHC methods that run reactions until saturation."
Kaigala said the microfluidic probe also offers the possibility of improved multiplexing. In a 2012 paper in Lab on a Chip introducing the device, he and his colleagues used the device to multiplex measurements of ER, PR and p53 in breast cancer tissue.
Moving forward, he said it would be necessary to automate and industrialize the microfluidic probe device and process to make it suitable for use in a clinical setting, though he noted that this was not something his group would pursue itself.