Development of Multiplexed 2D and 3D Pathology Techniques

Treatment decisions following surgery for breast cancer rely heavily on immunohistochemistry (IHC) assays for prognostic markers such as estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor (Her2/neu), all of which are performed on a highly limited sampling of tumour tissues. Although secondary treatment directed by the presence or absence of these markers is associated with a reduction in mortality, there is much room for improvement: treatment with Herceptin improves overall survival by 30%, and with Tamoxifen the annual breast cancer death rate is reduced by 31%. Four key factors may be implicated in the gap between receptor status measurements and treatment outcome, although the relative contribution of each is difficult to isolate using conventional sampling-based IHC:

 

  1. Spatial heterogeneity in the distribution of biomarkers and artefacts due to undersampling: Significant intratumoural heterogeneity has been described for ER, PR, and Her2/neu. Adequate sampling is critical, especially for rare cells (e.g., tumour stem cells).
  2. True shift in molecular subtype: There is some evidence that this may occur during the course of disease progression or treatment
  3. Artefacts due to variability in pre-analytical conditions: Factors such as cold ischemic time, formalin quality (concentration, pH) and antigen retrieval affect biomarker expression in an uncontrolled manner and are implicated in some studies, while others suggest that biological variables or cellularity in the spatially-dependent pattern of intratumoural expression may be responsible for inaccuracies in ER, PR, and Her2/neu scoring.
  4. Distinct molecular subtypes within tumours showing the same receptor profile: In Ontario, patients with moderately-differentiated breast cancer currently receive a recurrence score calculated by the multi-gene Oncotype Dx assay (Genomic Health; Redwood City, California, USA), which is prognostic and predictive for ER+ tumours. Recent studies suggest that an IHC biomarker panel comprising ER, PR, Her2/neu along with cell proliferation marker Ki-67, which is recommended for routine IHC in newer practice guidelines, is prognostically equivalent to the expensive, RNA-based Oncotype Dx test. However, before Ki67 or other novel biomarkers that may lead to more patient-specific secondary treatments are adopted clinically on a widespread basis, validation is necessary. In addition to assessing the prognostic benefit, validation must address technical considerations such as artifacts that may arise from under-sampling and cold ischemic time.

To address the issue of secondary treatment failure, we propose to develop and test methods for simultaneous multiple biomarker IHC imaging and analysis, extendable to the ‘3D pathology’ framework. True biological heterogeneity, as it may relate to subclonal diversity, or variations in tumour microenvironment (e.g., ischemia, since heterogeneity is commonly observed in necrotic tumours), or proximity to stroma or vessels (i.e., chemokines, growth factors or other modulators) can be separated from artefact using whole-mount sections. Whole-mount sections are also valuable for accurate validation of novel biomarkers, particularly Ki67 which has shown significant intratumoural heterogeneity, and other markers that may become incorporated into a more ‘personalized’ molecular signature. Multiplexing can be used to provide an essential link to known references by enabling co-localization of the novel marker with established ones (i.e., ER, PR, Her2/neu) and also with tumour morphology. This link is required for providing the complementary information from the same cell that is frequently required for accurate interpretation. True co-localization is also essential for both for accurate identification of established, significant relationships between biomarkers (e.g., ‘triple negative’ breast cancer) as well as the discovery of new ones. Our effort toward improved quality assurance will focus on characterizing and reducing artefacts related to cold ischemic time and incorporating a colour calibration technique that is an essential pre-requisite for computer-assisted detection (CAD). CAD algorithms (e.g., scoring, cell-counting, segmentation and co-localization) will be developed along with statistical methods for the large, multiparametric datasets.