Sections cut from the array allow parallel detection of DNA (fluorescence in situ hybridization [FISH]), mRNA (mRNA in situ hybridization), or protein (immunohistochemistry [IHC]) targets in each of the hundreds of specimens in the array (1). This allows consecutive analyses of a large number of molecular markers and construction of a database of correlated genotypic or phenotypic characteristics of uncultured human tumors. The power of this TMA technique is the capability of performing a series of analyses of thousands specimens in a parallel fashion.
A related technology, termed multitissue or sausage block was introduced more than 10 years ago by Battifora (6). In this method, large tissue fragments are brought into a recipient block in a less organized way than in TMAs. The TMA technology has a series of substantial advantages over the sausage technique. The arrays are constructed in a highly precise, regular formation that is amenable to automation and that facilitates the analysis of TMA data. The number of sections that can be generated from a series of tissue blocks using the TMA technology is several thousands, much more than can be reasonably retrieved with the “sausage” approach. The cylindrical form and the small diameter (0.6 mm) of the specimen taken out of one paraffin block maximize the number of samples that can be taken out of one block and minimize the tissue damage inferred to the donor block. The latter is important for pathologists because they can now more readily give researchers access to their material and at the same time retain their tissue blocks in a condition that is compatible with verification of clinical diagnoses or performing new immunostainings for future diagnostic evaluation. The morphology of the punched tissue cylinders is well preserved (Figure 2).
RECENT EXAMPLES OF APPLICATIONS IN BASIC, TRANSLATIONAL, AND CLINICAL CANCER RESEARCH
A number of different applications of TMAs have been described in recent publications. The first paper, by Kononen et al. (1), described the potential of the array technology by assembling a breast cancer tissue array with 645 samples. Six gene amplifications, as well as p53 and estrogen receptor protein expression, were detected by FISH and IHC. The frequencies of ERBB2, CMYC, CCND1, and CyclinD1 gene amplifications agreed with published results. Two newly discovered regions of DNA amplification in breast cancer, 17q23 and 20q13, were also analyzed. The prevalence of these gene amplifications and their association with other clinical and molecular parameters were rapidly uncovered.
A consecutive paper by Schraml et al. (7) also demonstrated that the results from the literature can be reproduced on minute tissue samples of the tumor arrays. In this study, three FISH experiments were performed to analyze amplifications of three oncogenes (CCND1, CMYC, and ERBB2) in a TMA consisting of 397 samples from 17 different tumor types in a single paraffin block. The study was performed in only 1 week and could confirm and extend existing FISH data in the literature.
The value of TMA for analysis of genetic alterations in different stages of tumor progression was demonstrated by Bubendorf et al. (8). To obtain a comprehensive survey of gene amplifications in different stages of prostate cancer progression, they constructed a TMA containing 371 specimens from benign prostates and different stages of prostate cancer progression, including therapy hormone–refractory local recurrences and metastases. Five different gene amplifications (androgen receptor, MYC, ERB-B2, CyclinD1 and N-myc) were studied by FISH on two consecutive formalin-fixed TMA sections. The high-throughput TMA based screening by FISH identified distinct patterns and interrelationships among the different gene amplifications.
The cDNA–microarray technology allows the screening of thousands of over-or underexpressed genes in a cell line or in a tumor (3–5). Tumor and control RNA are reverse transcribed to cDNA, which is then labeled and hybridized either on filters (radioactive labeling) or on glass slides (fluorescent labeling) that contain an array of thousands of different gene targets. The high number of candidate genes emerging from such experiments results in the practical problem to test all candidate genes. Using conventional methods, such analyses are time consuming and labor intensive. The TMA method is the ideal tool to test these candidates in a high-throughput manner (9,10).
In a proof of principle study (9), we performed a cDNA array experiment with radioactive-labeled cDNA from a normal kidney and from a kidney cancer cell line (CRL-1933). The experiment showed that 89 genes were significantly differentially expressed in the cancer. Thirty-eight of the transcripts consisted of 26 previously cloned genes, and 12 were unknown expressed sequence tags (ESTs), which were overexpressed in CRL-1933. The sequence of one of the upregulated genes in the tumor cell line was identical to vimentin. Vimentin protein expression was examined in primary tumors by immunohistochemistry using a tumor TMA with 532 formalin-fixed, paraffin-embedded renal cancer specimens. The prevalence of vimentin expression matched previous findings on conventional renal tumor sections. Interestingly, our TMA study showed that vimentin was significantly linked to poor patient prognosis. This example indicates that one can identify a potential biomarker from a cDNA microarray experiment and rapidly test its clinical significance by immunohistochemistry, provided that an antibody is available. The combination of TMA and cDNA microarrays is clearly a powerful approach to rapidly identify and further evaluate genes that play a role in tumor biology or that may have clinical relevance as diagnostic, prognostic, or therapeutic markers.
Bubendorf et al. (10) provide another example of a combined approach using cDNA and TMAs in an investigation of molecular mechanisms of hormone therapy failure in human prostate cancer. Gene expression changes between a hormone-refractory prostate cancer xenograft and its primary, hormone-sensitive parenteral strain were analyzed with the cDNA-MTA. Of 5184 genes surveyed, there were 37 transcripts whose expression was systematically elevated in the hormone-refractory xenograft as compared with the hormone-sensitive primary tumors. These differentially expressed genes included the insulin-like growth factor binding protein 2 (IGFBP2) and heat-shock protein 27 (HSP27). The immunohistochemical analyses of the prostate TMAs confirmed high expression of the IGFBP-2 protein in 100% of hormone-refractory clinical tumors, compared with only 37% in primary tumors.
These two studies (9,10) showed that the strategy of combining expression (cDNA) arrays and TMAs is a feasible approach for translating novel gene findings from experimental model systems to clinical applications.
In a recent paper by Mucci et al. (11), analysis of a TMA was used to study the neuroendocrine differentiation of prostate cancer. This study compared results from standard slides with TMAs in 50 primary and metastatic prostate tumors from men with hormone-refractory prostate cancer. The study showed only focal expression in advanced prostate tumors. This result was unexpected, given data from prostate tumor cell lines and animal models suggesting that progression to a neuroendocrine phenotype parallels tumor progression. The study also demonstrated the use of high-density TMAs to screen for protein expression, even when expression is focal.
EXAMPLES OF DIFFERENT TMA TYPES
Typical TMAs constructed so far include multitumor arrays, progression, and prognosis arrays.
Multitumor TMAs are composed of samples from multiple tumor types. These arrays are used to screen different tumor types for molecular alterations of interest. The first example of a multitumor TMA contained 397 samples from 17 different tumor types, including specimens of the most frequent cancer types (7).
Progression TMAs have been used to study molecular alterations in different stages of one particular tumor, e.g., breast, urinary bladder, kidney, and prostate cancer (8,10). For example, a prostate cancer progression TMA described by Bubendorf and coworkers contained samples from normal prostate or benign prostatic hyperplasias, prostatic intraepithelial neoplasia, incidental carcinomas (stage pT1), organ-confined carcinomas (stage pT2), carcinomas with extraprostatic growth (pT3-4), as well as metastases and recurrences after androgen-withdrawal treatment.
Prognostic TMAs contain samples from tumors of patients for whom clinical follow-up data and clinical endpoints are known (9,12). With the help of such prognosis array, novel prognostic parameters can be identified, or the value of molecular alterations for prediction of chemotherapy response can be tested.
Currently, TMAs have been used only in cancer research, but the technology is not limited to cancer research. We predict that TMAs will be constructed and widely used in other fields, such as in inflammatory, cardiovascular, and neurologic diseases. Similarly as for patient tissues, TMAs can be used for cell lines and other experimental tissues such as xenograft tumors or tissues from animal model systems.
THE PROBLEM OF TUMOR HETEROGENEITY
Because of the small size of the individual arrayed tissue samples (diameter 0.6 mm), the question arises as to whether these specimens are representative of their donor tumors. It is unavoidable that some alterations are not detected if the analysis of heterogenous tumors is restricted to samples measuring 0.6 mm (7). However, it is important to realize that the TMA approach has been designed to examine tumor populations and not to survey individual tumors. Meanwhile, we have analyzed the impact of tissue heterogeneity on TMA data and compared results obtained from TMA with results from large sections in multiple different studies. In two studies, sets of breast and bladder carcinomas were used to construct four replica TMAs (manuscripts in preparation). Different prognostic parameters were analyzed on all replica arrays and corresponding large sections. The results showed heterogeneity within tumors but suggested that this heterogeneity did not influence the identification of prognostic parameters. Every single association of morphologic or clinical parameters that was found on large-section analysis was also detected in TMA analysis. Importantly, this was not only true for the combined data from four replica TMAs but also for every individual TMA constructed during these studies. This suggests that TMA studies will provide meaningful data, even if only one sample is analyzed per tumor. Based on these data, we assume that at least most associations between molecular changes and clinical endpoints can be detected on TMA, especially if the number of tumors included in a TMA is large enough. Up to now, all TMA studies have confirmed all clinicopathologic correlations that were previously established by examining larger tissue samples. Tissue microarray studies may therefore replace most large-section studies in the research setting in the near future. TMAs will be especially valuable for a rapid initial evaluation of the potential utility of a novel biomarker. If a marker with a potential value is identified, the results can be confirmed by a large-section study.
ENVISIONED IMPACT FOR MOLECULAR AND ANATOMIC PATHOLOGY
The TMA technology will have a dramatic impact not only on basic cancer research but also on anatomic pathology. Tissue microarray may serve as quality-control instruments because standardized slides that include normal tissues, positive controls, and fixation controls can be efficiently used to evaluate sensitivity and specificity of antibodies, tissue fixation methods, and antigen retrieval methods and to optimize staining protocols. It is evident that TMAs help to save reagents, manpower, and money because one can now investigate up to 1000 tumors with the same amount of reagent that was previously necessary to analyze one tumor. There are several important differences between immunohistochemical analyses on TMAs and large sections, both on the technical and interpretation side. Even though an important source of variability, fixation, cannot be controlled using TMAs composed of archival tissues, TMAs allow a high level of standardization for immunohistochemical stainings because all tumors are pretreated and stained under exactly the same conditions. In contrast to the reading of large sections (which always is an attempt to integrate the observations in multiple different regions of a tissue section), morphologic classification and interpretation of immunoreactivity are based on the findings within one small, highly defined tissue area in TMAs. The criteria for diagnostic decisions are therefore much easier to establish between the individual samples on the array and to compare among different observers.
Further advantages of TMAs are that the minute tissue samples obtained from valuable materials or rare tumors will ensure that precious research materials are not destroyed. This makes it easier to start collaborative studies or set up collaborative networks.
In summary, we foresee many applications for TMAs, including:
- testing and optimization of probes and antibodies, improved utilization of pathology archives and tissue banks,
- international and other large-scale collaborations, e.g., studies of rare tumors or molecular profiling of tissues from multicenter clinical trial materials,
- teaching and quality-control tool to improve standardized interpretation of morphologic, immunohistochemical, and molecular analyses in pathology,
- standardized molecular or immunohistochemical detection of targets, and
- rapid translation of results from cell lines, xenografts, and animal models to human cancer.
Major changes that can be expected in the field of TMA technology include automation of TMA construction and analysis. One of the advantages of TMAs for automated image analysis is the exact X-Y-positioning of each specimen. This precise arrangement of arrayed samples facilitates interpretation of the stainings but also serves as an ideal basis for automation of the analysis. Even without automation, it is possible to perform up to 1000 uncomplicated analyses within a few hours. This also contributes to a high intraobserver consistency because all tumors of one study can usually be analyzed within 1 day. It is likely that the speed and the objectivity of immunohistochemical analyses will be substantially improved by automated analyses. Already at the present state, there are promising results in the automated analysis of immunohistochemical staining results or FISH signals (13).
It is not only expected that the interpretation of staining results on TMA will be completely automated, but automation could help to archive the raw image data and to support the decisions by pathologists. Attempts are underway to construct automated tissue arrayers that would permit simultaneous construction of multiple tumor array blocks with identical sample coordinates. With automated array construction, the quality of TMAs will improve if all samples can be automatically positioned with high precision. The selection of representative areas for analysis has always been a major obstacle, requiring a person with experience in tissue morphology. To assure optimal TMA quality, the selection of the areas of interest must always be done by an experienced pathologist.
In the near future, TMA will become more available for researchers, from both academic and commercial sources, for those individuals who do not have access to tissues or who do not have the technical expertise and pathology manpower to construct these arrays. It is anticipated that the speed and costs of tissue-based molecular pathology research will eventually be substantially lower than today. It makes sense to develop several large-scale arraying resources and centers that collaborate with a large number of investigators. The application of large-scale, standardized TMAs simultaneously in many different laboratories would generate a new paradigm for clinical cancer research. Currently such efforts are very decentralized, with each institution working on their own, often relatively modest tissue specimens, often not linked to proper clinical and follow-up information. Furthermore, each laboratory working with its own material can perform only a limited number of molecular analyses. Constructing a database where thousands of different molecular analyses would have been analyzed from the same set of thousands of clinically characterized tissues would be highly valuable to the research community.
In summary, the TMA technique has a number of distinct advantages over traditional pathology methods using large sections. Improved standardization, capacity, and speed of analysis, as well as the potential of automatization of array construction and analysis, are strong advantages of this technology. Studies showing that clinicopathologic associations with molecular markers can reliably be identified on TMAs will be important to validate and increase the acceptance of TMA in the near future. The genetic code of the human genome is now being uncovered, and this forms a new basis of linking genes with diseases. This will have a major impact for diagnosis, prediction of prognosis, and informed treatment of patients. A challenge for the pathology services in this new era should be to support molecular medicine to ensure that new innovations and novel developments can be implemented quickly as routine service in a timely, efficient, cost-effective manner. In anatomic pathology of the new millennium, TMAs will constitute a fundamental basis for this goal. Pathologists have to integrate this new technology to successfully establish their discipline as an integral component of the contemporary multidisciplinary approach to medical and surgical treatment of diseases. The emerging area of pathology informatics will be increasingly important and will enable pathologists to manage and disseminate data and information obtained from thousands of molecular assays on thousands of tumors in an efficient manner.
1. Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med 1998; 4:844–7.
2. Peters R, Sikorski R. Tissue cork borer. Science 1998; 281:1163.
3. Schena M, Shalon D, Davis RW, et al. Quantitative monitoring of gene expression patterns with a complementary DNA microarray. Science 1995; 270:467–70.
4. Schena M, Shalon D, Heller R, et al. Parallel human genome analysis: microarray-based expression monitoring of 1000 genes. Proc Natl Acad Sci U S A 1996; 93:10614–9.
5. DeRisi J, Penland L, Brown PO, et al. Use of a cDNA microarray to analyse gene expression patterns in human cancer. Nat Genet 1996; 14:457–60.
6. Battifora H. The multitumor (sausage) tissue block: novel method for immunohistochemical antibody testing. Lab Invest 1986; 55:244–8.
7. Schraml P, Kononen J, Bubendorf L, et al. Tissue microarrays for gene amplification surveys in many different tumor types. Clin Cancer Res 1999; 5:1966–75.
8. Bubendorf L, Kononen J, Koivisto P, et al. Survey of gene amplifications during prostate cancer progression by high-throughout fluorescence in situ hybridization on tissue microarrays. Cancer Res 1999; 59:803–6.
9. Moch H, Schraml P, Bubendorf L, et al. High-throughput tissue microarray
analysis to evaluate genes uncovered by cDNA microarray screening in renal cell carcinoma. Am J Pathol 1999; 154:981–6.
10. Bubendorf L, Kolmer M, Kononen J, et al. Hormone therapy failure in human prostate cancer: analysis by complementary DNA and tissue microarrays. J Natl Cancer Inst 1999; 91:1758–64.
11. Mucci NR, Akdas G, Manely S, et al. Neuroendocrine expression in metastatic prostate cancer: evaluation of high throughput tissue microarrays to detect heterogeneous protein expression. Hum Pathol 2000; 31:406–14.
12. Richter J, Wagner U, Kononen J, et al. High throughput tissue microarray
analysis of cyclin E gene amplification and overexpression in urinary bladder cancer. Am J Pathol 2000; 157:787–94.
13. Bubendorf L, Kononen J, Barlund M, et al. Tissue microarray
FISH and digital imaging: towards automated analysis of thousands of tumors with thousands of probe [abstract]. The American Society of Human Genetics 1999, 49th Annual Meeting.
Keywords:© 2001 Lippincott Williams & Wilkins, Inc.
Tissue chips; Tissue microarray; Quality control; High throughput analysis; Immunohistochemistry; Molecular biology