[1631] Characterization of Tumor Heterogeneity Using High-Throughput Morphometric Assays (hTMA)

Alexander Baras, Pei-Hsun Wu, Zev Binder, Toby C Cornish, Denis Wirtz. Johns Hopkins, Baltimore, MD

Background: The notion of tumor heterogeneity has received much attention. To attempt to better characterize this notion, we are developing high-throughput morphometric assays and supporting computational algorithms, which we believe will allow for better characterization of tumor heterogeneity and how this could be related to clinicopathological parameters such as tumor stage/grade, therapeutic response, and others.
Design: Various previously established pancreatic cell lines derived from both primary and metastatic components of pancreatic tumors were examined by digital microscopy. From these digital images, over 200 morphologic parameters per cell were extracted from a representative microscopic field from a given specimen. Subsequently, the resulting data were processed by a recently developed context specific clustering algorithm to reveal the distinct patterns of cellular morphology present and the distribution of these patterns within a given specimen.
Results: The analysis of nine cell lines derived from pancreatic tumors revealed that multiple distinct and consistently present cellular morphological phenotypes existed within the context of cell lines grown in the laboratory setting (Figure 1A). Interestingly, certain cellular morphological phenotypes were found to exist in consistently different portions when comparing cell lines derived from the metastatic component of pancreatic tumors as opposed to cell lines derived from the pancreatic tumors that did not exhibit metastatic disease clinically at the time of surgical resection (Figure 1B, Kruskall-Wallis p-value = 0.002).

Conclusions: The utilization of the proposed high throughput morphometric assay allowed for the examination of adherent cells and the identification of distinct cellular morphological patterns whose distributions were significantly associated to relevant clinicopathologic parameters. These preliminary results support the overall hypothesis of tumor heterogeneity and reveal that examination of this heterogeneity may have implications toward better understanding of tumor biology. Our future plans including extending this approach to the evaluation of histologic preparations of tissue specimens.
Category: Informatics

Tuesday, March 20, 2012 1:00 PM

Poster Session IV # 216, Tuesday Afternoon


Close Window