November 1, 2014 (Vol. 34, No. 19)
Farideh Bischoff, Ph.D. executive director Silicon Biosystems
Nicolò Manaresi, Ph.D. CSO Silicon Biosystems
Chiara Bolognesi biologist Silicon Biosystems
Realizing the Benefit of Homogeneous Cell Population Analysis from FFPE Specimens
Today, clinical decisions regarding cancer therapy are largely informed by histologic evaluation and perhaps analysis of select molecular biomarkers associated with the particular tumor. However, cancer biology is complex and better tools are needed to advance the use of diagnostic biomarkers.
Pathologists can improve the accuracy of tumor biomarker analysis by isolating and enriching samples for tumor cells prior to testing. However, preparing samples enriched with tumor cells for molecular analysis can be challenging. Factors affecting biomarker analyses include the tissue-preparation process, and the initial size, quality, and composition of the biopsy specimen. While fixation of tissue with formalin and embedding in paraffin (FFPE) has become standard practice for pathology laboratories, fixation and extraction processes can change the molecular composition of the specimen in significant ways.
DNA extraction methods do not discriminate genomic content of tumor cells from that of stromal cells, so the accuracy of any biomarker assay is dependent on the selection of tissue sections bearing an adequate representation of tumor cells. For many tumors, finding suitable tissue sections with sufficient tumor cells, or tissue not impaired by the presence of necrotic or damaged cells, can be difficult.
Tumor cells within biopsies are molecularly heterogeneous, consisting of both cancerous cells and an indeterminate number of normal cells. Such composition can result in equivocal or inconclusive test results that may ultimately disqualify patients for certain therapies or confound biomarker analyses in clinical studies. Hence, any method developed to improve on standard DNA extraction methods by enriching for specific target cells must also work effectively with miniscule FFPE sections as starting sample material.
FACS and LCM—Promising but Incomplete Solutions
Suitability of FACS (fluorescence activated cell sorting) for preparing pure tumor cell populations from FFPE tissues was examined by Willem E. Corver and Natalja T. ter Haar, (DOI: 10.1002/0471142956.cy0737s55). Sorting on the basis of antibody recognition of biomarker proteins (keratin, a carcinoma tumor cell marker; vimentin, normal stromal cell marker), the investigators applied a treatment protocol that reverses formalin crosslinking and produces a cell suspension while sparing the marker proteins of interest. However, the separations were incomplete, producing partially enriched rather than pure, separate populations of tumor and normal cells. The method was also inefficient, requiring tissue sufficient to yield several millions of cells as starting material, which is impractical for both routine diagnostic use or as a means of extracting molecular information from large numbers of FFPE tissue samples.
LCM (laser capture microdissection) technology has also been used to address the effects of tumor heterogeneity. LCM enables the selection and separation of defined regions from intact tissue sections. However, with LCM, the purity of cell subpopulations prepared for analysis is compromised by the complexity of tumor cell infiltration patterns recognized by the histologist. Additionally, the scarcity of individuals possessing the combination of histological experience and technical expertise that is required for LCM instrumentation limits the scope of routine applications.
DEPArray™ Technology—Image-Based Cell Sorting
The DEPArray System from Silicon Biosystems is an automated solution for separating and sorting live or fixed cells in suspension. The technology combines image-based fluorescence and brightfield microscopy with dielectrophoretic cage formation to enable recognition and manipulation of individual cells.
Cells of interest recognized automatically according to their immune-fluorescent signatures, or cells chosen by an operator, are automatically moved and collected individually or pooled by droplet formation. Studies have now shown successful DEPArray sorting for circulating tumor cell (CTC) populations (Hodgkinson, et al., DOI: 10.1038/nm.3600), as well as transformed cell lines and disseminated tumor cells (Carpenter, et al., DOI: 10.3389/fonc.2014.00201).
Recovering 100% Pure Tumor Cells from FFPE Tissue
Using a method similar to Corver and ter Haar, a FFPE-derived cell suspension can be sorted with a DEPArray system to produce pure tumor and pure stromal cells in approximately the same amount of time as was required to achieve a less complete cell separation by FACS (Figure 1). In addition, the DEPArray sorting method uses far less starting material, requiring approximately one percent as many cells from an FFPE specimen.
Although the initial density of biomarker-positive cells is very low (Figure 2), the DEPArray method produces a pure population of cells for analysis. The bulk pool of tumor cells has sufficient DNA material to be used directly in next-generation sequencing (NGS) without requiring whole-genome amplification.
Overview of Method
Tissue specimens are subjected to deparaffinization, then dissociated using .1% dispase/.1% collagenase 1a to create a single-cell suspension containing both stromal and tumor cells. The cell mixture is then subjected to immunofluorecence labeling of specific biomarkers (i.e., cytokeratin and vimentin) as well as nuclear stain (DAPI) for DNA quantitation. Cells are loaded directly onto the DEPArray cartridge and sorted based on immunofluorescent and morphological criteria.
Implications of DEPArray Technology for Molecular Pathology
By comparison to existing methods of tissue preparation, the DEPArray technology-enabled method results in higher quality molecular analysis because allelic frequencies or expression signatures are now not confounded by an indeterminate percentage of normal cells. Subtyping of tumors based on molecular analysis is less ambiguous. The method described here provides finer resolution of the target cell population and improves biomarker sensitivity and specificity, enabling the development of better companion diagnostic methods to benefit patients.
Additionally, since the method requires significantly less starting tissue, many samples initially rejected due to sample size, insufficient tumor cell infiltration, partial necrosis, or poor fixation become suitable for molecular analysis.
In research laboratories and pharmaceutical drug development, obtaining pure tumor cell subpopulations for DNA/RNA analysis will help investigators eliminate a source of research and patient selection bias that has favored the study of tumors composed of high percentages of cancer cells. Preselection protocols have had the unfortunate effect of limiting many programs to the study of advanced-stage cancers where clinical intervention is of least consequence to patients.
It is estimated that as many as 1 billion FFPE tissue specimens are archived worldwide. The ability to derive pure cellular subpopulations from individual FFPE samples will enable meaningful retrospective correlative studies aimed at connecting cell genotypes to clinical histories. The addition of whole-genome sequencing to these studies will produce an evergreen resource of biological information and research opportunities.