Tissue microarray: A simple technology that has revolutionized research in pathologyS Avninder1, K Ylaya2, SM Hewitt2
1 Institute of Pathology-ICMR, New Delhi, India
2 Tissue Array Research Program, Laboratory of Pathology, Center for Cancer Research, National Cancer Institute, NIH, Bethesda, USA
Tissue microarray (TMA) technology is a high-throughput research tool, which has greatly facilitated and accelerated tissue analyses by in-situ technologies. TMAs are amenable to every research method that can be applied on the standard whole sections at enhanced speed. It plays a central role in target verification of results from cDNA arrays, expression profiling of tumors and tissues, and is proving to be a powerful platform for proteomic research. In this review article, primarily meant for students of pathology and oncology, we briefly discuss its basic methodology, applications and merits and limitations.
Keywords: High-throughput, immunohistochemistry, pathology, proteomics, tissue microarray
In the post-genomic era, with the availability of vast information on the simultaneous expression levels of thousands of genes by high-density cDNA microarrays, a platform for high-throughput analysis of tissue specimens was inevitable. TMA technology though initially utilized primarily for basic research, typically in target validation of cDNA and other high-throughput approaches, is now becoming an invaluable tool for clinical research. , In cancer research, with rapid increase in discovery of new genes, and their involvement in multiple pathways, there is an increasing demand for analyses of these genes in different stages, precursor lesions as well as metastatic tissues. 
We review this new technology and discuss its design, construction, applications, and future prospects of this in basic and applied pathology research.
Although a manual or automatic device is used to manufacture TMA, but most of the work involved in construction of tissue cores is standard histopathology work. For maximizing the benefits of TMA, extensive planning and attention to details are required prior to construction. 
Construction requires collection of relevant clinical data and tissue after appropriate ethical approval for collection and use of tissue. The vast majority of TMAs are constructed from archival formalin-fixed paraffin embedded tissue. Hematoxylin-eosin stained slides with adequate representative tissue are required before starting as eventually the quality of TMA will be represented by the quality of each tissue selected.
Mapping of slides
The goal of the TMA is to present the pertinent tissue on the array. The optimal method to select a tissue for arraying from a donor block is to map the area of interest with ink on the fresh HandE stained slide on a scanning power. It is important to remember that the target is three-dimensional and the depth of tissue is as important as the width.
For constructing a TMA cylindrical cores of paraffin embedded tissue are removed from preexisting donor paraffin blocks that may be from surgical pathology, autopsy or research material. These tissue cylinders are then inserted into a blank recipient block made from low melting point paraffin (52-56°C) and care should be taken to prevent bubble formation within the block as it may cause problems during the arraying process. , It should ideally measure at least 45×20mm in dimension with practical array field of 22×18mm. This area will easily hold 500 or more 0.6mm cores. Sub-arrays should ideally be large enough so that each sub-array can entirely be seen at 4× objective. 
The present day tissue arrayer emerged from improvements on its previous methodologies, the precursor 'sausage roll' of tissue described by Baltifora  and its refinement, the checkerboard pattern was represented by squares of tissue.  Kononen etal. ,  in 1998 developed the current methodology at National Institutes of Health, USA and since then many modified versions have appeared including the fully automated ones. The basic prototype manual TMA developed by Beecher Instruments shown [Figure - 1] was built to create holes in the recipient blank block (smaller punch) and then taking cylindrical cores of tissue from the donor blocks (larger punch) and then delivering them in these holes for precise arraying according to a pre-set array design. The arraying needles move vertically with a stylet to release the extra cores and a pendulum called turret allows the alternating movement of recipient and donor needles. Also a precision guide with a digital micrometer attached in X and Y-axis provides successive locations on the array. In this way hundreds of existing formalin-fixed tissues can be arrayed on a single block for analyses. Tumor TMAs are broadly classified into three types, multi-tumor array, progression array (based on stage of tumor) and prognostic arrays when tumors with known clinical endpoints are arrayed.
Array design and sampling
The essential idea in arraying is to match the core size to the number of cores and put these cylinders of tissue into the user-friendly recipient block. The standard needles come in 0.6,1.0,1.5,2.0mm diameter [Figure - 2]. The maximum practical array area in the recipient block with each of these needles will be approximately 500,200,100,50 cores respectively. A template grid with selected area size and individual sub-arrays should be made prior to construction of tissue array. The choice of selecting a needle size is crucial depending on the quantity and quality of available tissue. A general dictum is that larger the sample size, smaller should be the core diameter. As the needle diameter increases, the spacing between cores must increase. Certain tissues and situations merit noting. There are certain tissues that are not representable on TMA like normal renal glomeruli or the portal tracts of liver while it is better to take multiple cores for optimal representation in sampling prostate tissue.  Bony tissues are difficult to array and lesions like osteosarcoma are best cored with 1.0mm needles.  It is important to check that there are no air bubbles, the needles are clean and the stylet moves freely. It is imperative that in each cycle a recipient hole is made first and then donor tissue is cored. If a series of recipient holes are made beforehand, the holes will get deformed due to the elasticity of the paraffin. It is always useful to make a 'pointer' core on the top left of the array area of the recipient block for providing orientation [Figure - 3]. There should be adequate space between the sub-arrays and normal controls should always be included in the array.  Once the recipient block is arrayed with desired tissue cores, it is important to 'temper' it by keeping it in an incubator at 37°C overnight and then allowing it to cool at room temperature so that the recipient block holds all the cores firmly. After this 100-150 sections of 3-5 microns can be cut with ease by a trained histotechnologist.
The conventional approach of formalin-paraffin embedded and fresh frozen tissue is to time consuming and less cost effective to be applied to the characterization of hundreds or thousands of genes or gene clusters associated with distinct tumor entities. TMA technology has facilitated high-throughput analyses on a large series of tissues in a single experiment. Virtually all the tissues are suitable to be arrayed in the TMA and therefore the range of applications is as broad as those for any standard tissue sections with immunohistochemistry and in-situ hybridization as the favorite methodologies.  A spectrum of TMA applications are discussed below
Validation of diagnostic biomarkers
The most popular use of TMA is in basic/translational research in validation of diagnostic markers in annotated clinical samples. Given the rapid pace by which new genes are being discovered by high-throughput DNA microarrays, TMAs hold an immense potential to validate this genomic and proteomic data across multitumor types in a limited timeframe. Diagnostic biomarkers are tissue markers that can serve as surrogate endpoints for clinical studies, as screening tools for certain diseases or for identification of specific subclass in a disease. Any method of profiling a large number of pathology specimens at one time, will be extensively used to develop novel diagnostic biomarkers. For example, Nishizuka et al. ,  screened a panel of cell lines and identified genes that were differentially regulated between ovarian and colon carcinoma. On TMA validation, they identified genes villin and moesin that could be used to differentiate colonic from ovarian adenocarcinoma.
Validation of prognostic biomarkers
TMAs are becoming the standard for the validation of prognostic biomarkers. As tumor banks and clinical sample cohorts are maturing with regard to clinical follow-up data, it is becoming possible to perform correlations of biomarker expression with clinical endpoints like disease-free survival or overall survival.  Using TMA validation, Khanna et al. ,  demonstrated that high ezrin expression was associated with earlier metastasis and poor outcome. Similarly, Fong et al. ,  showed on a TMA based study of squamous cell carcinoma of oral cavity that TROP2 overexpression was significantly correlated with decreased overall survival.
TMA and response to therapy
TMAs are of increasing utility in defining the feasibility of developing particular therapeutic targets. They are now being constructed from samples of patients receiving specific treatments providing a unique tool for validating predictive response to these treatments. This approach is becoming popular in translational cancer research. Recent results of this approach include heat shock protein (HSP27) expression as a marker for drug resistance in prostate cancer  and in renal cell carcinoma, where overexpression of carbonic anhydrase 9 correlated with response to interleukin-2 therapy. 
Although TMA technology was initially described and used in cancer research, the horizons of this technique are now widening and are being reported in non-neoplastic pathology research such as brain tissue microarrays for research in neurodegenerative diseases.  TMA-based studies in other non-neoplastic tissues like dermatological, cardiac and placental diseases are also beginning to emerge. ,,
TMAs are being used for clinical applications in Pathology departments for testing of new antibodies and probes, or determining optimal staining conditions using small test TMAs. Van de Rijn et al. ,  have described the utility of 29-sample 'mini-TMA' of breast cancer blocks as a useful control for estrogen receptor instead of a single positive control slide as a spectrum of positivity from weak to strong can be analyzed. Application of TMA is emerging in clinical trials where the TMA is constructed from pathology tissues taken before initiation of therapeutic trial. The objective is to define the functional status of the tumor prior to therapy.
Quality control (QC)
Immunohistochemistry analyses have been justly criticized for their subjective and semiquantitative means of determining the level of protein expression. QC in IHC is one of the major problems in diagnostic pathology. There is high variability in intralaboratory and interlaboratory results mainly due to interlaboratory differences in antigen retrieval, staining protocols, antibodies used and in interpretation of results. TMAs can facilitate the IHC staining procedures and interpretation of external and internal quality control assurances. Recently, Fitzgibbon et al. ,  have shown in a College of American of Pathologists HER-2 TMA-IHC survey, that over 90% of the tested laboratories correctly scored the staining. This way TMAs now enable individual pathology laboratories to perform their own QC against a validated benchmark.
Novel TMA-based platforms
(a) Frozen TMA: The fixatives in the paraffin embedded tissue affects the quality of RNA thereby giving sub-optimal results for RNA hybridization. Though challenging, there has been progress in the development of frozen TMAs. Here TMAs were constructed from unfixed frozen tissues and embedded in a recipient blocks made of optimal cutting temperature (OCT) media.  The brittleness of frozen OCT renders coring difficult and smaller number of samples should be arrayed to prevent cracking of the blocks. Although there is distortion of morphology in frozen TMAs and they are difficult to work on, they provide excellent target material for study of RNA, DNA and proteins. They are useful when antibodies do not work on paraffin embedded tissue and when studies for fluorescent in-situ hybridization (FISH) are required. FISH on TMAs have been frequently used to validate findings of gene amplifications discovered by genome-wide screening
(b) Cell line microarrays (CMA): Increasingly, TMAs are being used for alternative tissue sources rather than archival human tissue and most of them are focused on drug development. Techniques have been described wherein both suspension grown and adherent grown cells are placed in agarose gels, fixed and then embedded in paraffin.  They are low cost, high throughput means of exploring proteins using cell lines and offer the capacity to multiplex the assay by studying the activated forms of protein on individual slides.
(c) Xenograft tumor arrays (XMA): XMAs are utilized as preclinical model system in which assays are developed that measure either susceptibility or response to a drug. There will be increasing demand that these assays be translated to clinical practice.
(d) Tissue immunoblotting: Although proteomic profiling can be carried out on IHC on TMA, there are inherent limitations of this approach. Chung et al. ,  used a tissue transblotting approach in which they transferred the proteins from a paraffin embedded TMA section from a glass slide to a nitrocellulose membrane, resulting in a functional protein array for quantitative protein analysis.
Tissue repository and education
TMAs can act as a valuable tissue repository with pooling of pathology specimens and tumors from high volume pathology departments across the country. This valuable research material can be used for organization of long-term tissue banking, for education purposes and facilitating multicentric studies.  In fact, with so much data being generated on TMA experiments a TMA database has been developed to collate all aspects of information related to TMA. 
The most obvious drawback of this technology was initially believed to be the possibility that small tissue cylinders might not adequately represent the whole sections, particularly in case of tumors, because of intratumor heterogeneity of protein expression. Kononen et al. ,  found the same frequencies of HER-2, c-myc, cyclinD1 and 17q23 amplifications in breast cancer as were expected from previous published literature from whole tissue sections. Addressing the problem of heterogeneity, Camp et al. ,  reported that two cores of 0.6mm from the same block of breast cancer provided equivalent information on estrogen receptor indices as that of whole sections. Thus, increasing the number of cores collected from the sample and/or increasing the core diameter to 1.0-2.0mm can resolve this limitation.
Other practical limitations of TMA are related to loss of tissue cores during processing and the unreliability of IHC staining. The former can be prevented by using the adhesive tape transfer method and the later by frequent interlaboratory QC checks.
In the current world of high-throughput technology, TMA has revolutionized the histopathological analysis in research. It has proved advantageous in comparison to standard whole sections as hundreds of tissue samples can be analyzed in a single experiment using 0.6mm disks of tissue. It is amenable to a wide range of techniques including histochemistry, IHC and in-situ hybridization. TMA provides a judicious use of precious tissue, gives experimental uniformity, and analyzing a large number of samples improves statistical precision in addition to enhanced speed and quality of analysis.
It is likely that in future, automation in TMAs will be available not only to construct better quality of arrays but also sophisticated tools will be developed to automate TMA scoring and facilitate even better and more efficient data analysis
The authors acknowledge the UICC for the technology transfer fellowship to SA
[Figure - 1], [Figure - 2], [Figure - 3]