Histopathology and Imaging - Services
Routine histologic preparation for laboratory investigators: These services include tissue processing such as fixation, embedding, tissue sectioning, and histochemical staining. Central to this service is proper specimen acquisition in coordination with the Biorepository Resource Service Shared Resource, recording, reporting and storage.
Immunohistochemistry: These studies are divided into two major types: (1) localization studies using novel investigator-generated antibodies; or (2) studies using commercially available antibodies. The former generally requires a greater expenditure in time and effort in determining proper antibody dilution, tissue preparation, and defining the conditions for optimal tissue localization. In both cases, the localization is available by immuno-fluorescence or immuno-peroxidase with current development of quantum dots. The automated immunohistochemical stainer equipment allows for testing with a variety of antigen-retrieval methods. Protocols have been developed to perform double and triple staining.
Fluorescent In Situ Hybridization/In Situ Hybridization (FISH, ISH): Quality staining using chromogenic or fluorescent in situ hybridization is offered. These highly sensitive staining techniques use the programmable software purchased as part of an automated immunohistochemistry stainer. This feature allows for protocols that optimize the various procedural steps necessary for in situ hybridization, such as pre-treatments, probe application, denaturation, hybridization and the stringency washes before detection and counterstaining.
Construction of tissue microarrays: Array construction involves retrieval of formalin-fixed paraffin-embedded tissue blocks containing tissue samples of interest, along with the corresponding hematoxylin-&-eosin (H&E)-stained slides, from surgical pathology archives or BRS; preparation of H&E-stained slides from each donor block and identification of areas of interest (e.g., normal tissue, in situ carcinoma, invasive carcinoma, or lymph node metastasis) on the H&E stained slide; array block layout design, taking into consideration the number of cores that will be used to represent each donor block (we typically use 2- to 4-fold redundancy), the spacing between rows and columns (we typically group cores in small squares of 5 x 5), and the size (total cores/cases) of the recipient block. Core biopsies of 0.6 mm in diameter are taken from each donor block and arrayed into a recipient paraffin block (30mmL X 24mmW X 5mmH) using a tissue puncher or automated arrayer (Beecher Instruments) as described by Kononen et al. (Kononen, et al., Nat Med,1998). Once an array block has been constructed, H&E stained sections are re-examined by the reference pathologist to check the histology represented in each tissue disk and to ensure quality control. Although TMAs have traditionally been assembled using paraffin-embedded tissues, the technology is now available to construct arrays from frozen specimens.
The guiding principle in developing TMAs is to design them to be useful for as wide a range of scientific objectives as possible and for biomarkers that might have different underlying biology, mechanisms, or functions. The process is intended to avoid the situation of having to design a new TMA for every new biomarker. To accomplish this goal as guided by the needs of Cancer Institute investigators, we construct four different types of TMAs:
- Test TMAs: General purpose arrays of anonymized de-identified human tissue specimens to optimize utilization of reagents, test new techniques, etc.
- Tumor-type-specific screening TMAs: Anonymized, de-identified, non-annotated collections of human malignant tissue specimens of a particular organ or organ system, with normal adjacent tissues whenever available. These arrays can be used to rapidly test for expression of biomarkers across different tumor types.
- Tumor-specific validation TMAs: Annotated collections of human malignant tissue specimens of a particular organ or organ system, with normal adjacent tissues if available. These arrays might also contain tissues from intermediate stages of tumor development. These arrays can be used to test for expression of biomarkers in particular tumor types, their changes during tumor development, and their validation as prognostic or predictive markers. Depending on the P.I. for the array and the study for which they were constructed, the clinical data may be supplied to the investigator in anonymized and/or de-identified fashion.
- Study-specific TMAs: Collections of tissue samples with full annotation assembled by a P.I. for a specific study.
Histologic preparation of TMA sections for laboratory investigators: These services include array block sectioning and histochemical or immunohistochemistry staining. Typically, 4-5µm sections of tissue microarray blocks are mounted onto glass slides using a paraffin sectioning aid system (adhesive coated slides PSA-CS4x, adhesive tape, UV lamp, Instrumedics Inc) to support the adhesion of the tissue disks to the glass Superfrost Plus slides (Fisher Scientific, Cat# 12-550-15). Microarray sections are processed within 2 weeks of cutting to avoid oxidation of antigens. Alternatively, cut sections can be soaked in xylene (2 min.), oven dried at 60°C for 20-30 min, then air-dried at room temperature for 30 min before being coated with liquid paraffin and stored at room temperature for up to 3 months prior to processing. Central to this service is proper specimen acquisition through the Biorepository Shared Resource.
Digital imaging, archiving and distributed microscopy of histopathology: All stained TMA sections are digitally imaged at 200x magnification using an Olympus AX70 microscope equipped with a Prior 6-way robotic stage and motorized turret. The server workstation consists of a Pentium IV computer with 1 GB of RAM running Windows 2000 operating system. Microscopic images are captured using a high-resolution, Olympus DP70 digital camera with RGB liquid crystal filter. Middleware was developed to allow communication between the robotics, the image processing modules, and an Oracle8i Database Management System. The software we developed locates, delineates and indexes each tissue disk using proper column and row indices. A low-resolution first-pass scan is performed to create a quilted digital version of the array. Each disk is then imaged at 100 and 200x magnification and archived in the database described below. An example can be viewed at “http://pleiad.umdnj.edu/~wjc/tma_example/CTMA3A2-15-3-4-10x.jpg”.
Investigators can view and access the image files via client software that allows them to access the image database remotely for viewing and downloading, or by requesting the image files on DVD or VCD. In most cases, the jpeg format is sufficient for viewing the images and for publication purposes. However, computer-drive image analysis (see below) will be performed on uncompressed images.
Database: The Oracle 8i database used to house TMA data consists of a Physical Specimen Layer (PSL), a Digital Sample Layer (DSL), and a Quantification Layer (QL). The physical specimen layer (PSL) of the database relates to the construction and preparation of the actual TMA sample. The specific data housed in this layer are referred to as the “array profile” and the specific fields include a) recipient array format information e.g. array dimensionality, cylinder diameter and interval; b) donor block information; and c) array construction information that records the correspondence between the specific cylinder grid location and its donor. A visually intuitive array profile editor has been developed to facilitate the design, editing and managing of array profiles. The structure of the database closely follows the TMA data exchange standards proposed by the Automated Information Management in the Clinical Laboratory (AIMCL) conference. The digital sample layer (DSL) of the database stores archived digital images and image maps as well as the corresponding images of each tissue disc at multiple resolutions. High-resolution images of tissue discs are automatically streamed and stored at 2 mirror sites while an additional copy is replicated onto a RAID5 system. Distributed databases are automatically updated with a pointer for each image to indicate its location and the scanning settings that were used during digital acquisition. Data regarding quality of each disc and each image, and surgical- and histopathological descriptors and diagnosis, are entered by the reference pathologist following the SNOMED convention. A web-based SNOMED viewer was designed to facilitate this process. Since the TMA technique results in a standardized set of tissue samples, it provides an ideal data set for developing and evaluating image processing and computer vision protocols, which can be used to perform quantitative immunohistochemistry. The third layer of the database, the quantification layer (QL), supports automated segmentation and computation of protein expression levels across each disc as specified below. As this database is accessible to individual investigators using any web browser, this service enables investigators to view their own array images at remote locations, subject them to analysis, and generally utilize the information to understand the molecular forces underlying malignant transformation. Normal appearing epithelium, premalignant cells, and the frank invasive carcinoma can be analyzed separately within the same tissue specimen or paired specimens from the same patient enabling investigators can compare the fluctuation of expressed proteins that correlate with the transition from one disease stage to the next. Moreover, potential prognostic or predictive biomarkers can be validated because the TMA database will be linked to the clinical research database in the Office of Human Research Services Shared Resource.
Image analysis software development and implementation: We have designed, developed, and validated a reliable means for unsupervised imaging, archiving, and analysis of TMAs and other histological specimens. The key computational and imaging tools that were developed in the first phase of the project include a color decomposition algorithm for analyzing the staining characteristics of histologic sections; image analysis tools for automatically computing the integrated staining intensity, effective staining area and effective staining intensity of expression patterns; image archiving and caBIG compliant data management tools; a reference library of expression signatures for more than 120,000 imaged tissue specimens and a novel active contour algorithm based upon robust estimation; a chromatic gradient model for use in antigen localization; and a new texture descriptor based on region covariance, which was shown to provide quick, reliable performance for identifying and delineating the boundaries of tumor regions throughout a given imaged specimen. We are expanding the scope to include a wider range of biomarkers and developing tools to support multi-spectral image datasets including antibody-conjugated quantum dot probes. We added a Trestle/Zeiss, high-resolution whole slide scanner and a high-resolution, liquid-crystal tunable multi-spectral camera and have begun to integrate a module to support automated analysis of specimens which have been prepared with quantum dot technology. This technology will enable investigators to evaluate and quantify multiple antibodies and/or FISH probes simultaneously on a single microscopic slide thereby preserving precious tissue, improving assessment of multiple proteins relative to each other and hastening the pace of our experiments. By combining the multi-spectral camera with the fluorescent microscope we can exploit quantum dot technology and offer this resource and associated analytical services to Cancer Institute members.
Expert consultation: Expert consultation is available for developing projects through Drs. Goodell, Chekmareva, and Foran on the proper selection of antibodies, processing, analysis of TMA slides and optimal use of image analysis and archiving.