Utility of FISH in Anatomic Pathology (Modified from Reference # [1]) Arie Perry

Although in situ hybridization (ISH) has been around for over 30 years, its application to the study of DNA alterations in solid tissue has only recently become popular. Unique among molecular techniques due to its morphologic basis, it involves direct microscopic visualization of probe-specific, intranuclear signals utilizing either chromogenic (CISH) or fluorescence (FISH) detection. Given that nuclei in any phase of the cell cycle may be analyzed and metaphase chromosomes are not required for interpretation, this technique has also been referred to as interphase cytogenetics. In clinical cytogenetics laboratories, it is currently utilized most often for either prenatal detection of germline alterations (e.g. aneusomy or microdeletion syndromes) or the detection of somatic cancer-associated alterations that have known diagnostic, prognostic, or therapeutic implications. In anatomic pathology, the oncology-associated applications are of greatest interest, though the detection of constitutional alterations may be useful in selected situations as well (e.g. monosomy or trisomy in a fetus, triploidy in a partial mole, etc.). In essence, FISH provides data on intranuclear target DNA localization and copy number. Therefore, with the exception of some sex-chromosome determinations, two signals per nucleus would normally be expected and four common alterations are amenable to detection: aneusomy (gain or loss of a chromosome or chromosomal region), gene deletion, amplification, and translocation. Advantages, Limitations, and Artifacts of FISH Analysis FISH is applicable to a variety of specimen types, including fresh or frozen tissue, cytologic preparations, and formalin-fixed paraffin-embedded (FFPE) tissue. The latter provides a particularly rich source of archival material and may be performed using either thin (4-6 m ) sections, such as those cut for immunohistochemistry or intact nuclei extracted from thick sections (e.g. 50 m), such as those normally prepared for flow cytometry. Although adjustments must be made for nuclear truncation (see below), we prefer the thin sections because it preserves architecture, is simpler to prepare, and wastes less tissue. For pathologists, morphologic preservation is probably the greatest advantage, particularly attractive for studies on heterogeneous tissue samples without the need for microdissection. For example, a morphologically mixed tumor recently studied by FISH is the gliosarcoma [2, 3] . The finding of identical genetic alterations in both components refuted the notion of a collision tumor and supported the hypothesis that both elements originate from a single clone, with the mesenchymal component arising from metaplasia. An extension of this morphologic advantage comes from the possibility of combining FISH with immunohistochemistry, wherein separate counts are rendered in immunopositive and negative cells. For example, this approach coined FICTION (fluorescence immunophenotyping and interphase cytogenetics as a tool for investigation of neoplasms), was required to demonstrate numerical chromosomal alterations in the CD30-positive Reed-Sternberg cells of Hodgkin's lymphoma [4, 5] . Because these neoplastic cells typically constitute only a minor fraction of the lymphoid population, clonal alterations are not amenable to detection by "averaging" techniques such as flow cytometry and PCR. Using this dual FISH-immunohistochemistry approach, we similarly demonstrated that NF1 deletions are restricted to the S-100 protein positive, Schwann cell elements of the cellularly and immunophenotypically heterogeneous plexiform neurofibroma [6]. Lastly, Tubbs et al. demonstrated the ability to simultaneously visualize HER-2/neu gene amplification and protein overexpression using this technique in breast carcinomas as a method of clarification for ambiguous cases [7]. They coined the acronym CODFISH (concomitant oncoprotein detection with FISH) for this particular application. Another distinct advantage is the similarity of FISH to immunohistochemistry, which is already familiar and widely applied in pathology laboratories. In many ways, the techniques are analogous, except that FISH utilizes DNA probes, rather than antibodies. Unlike the typical qualitative or semiquantitative immunohistochemical interpretation schemes (e.g. +/- or 0-3+), FISH provides quantitative results and is therefore more objective and straightforward in terms of interpretation. In comparison to classic metaphase cytogenetics (i.e. karyotyping), FISH has several advantages, most importantly, the lack of requirements for mitotically active cells and culturing. Given that only the cells capable of proliferating in vitro are assessable on karyotype, there can be significant growth selection biases, including overgrowth of non-neoplastic elements, particularly when analyzing benign or low-grade neoplasms. On the other hand, FISH is not a genomic screening tool, only providing a more targeted approach for alterations that have been initially identified by more global assessments, such as conventional cytogenetics, comparative genomic hybridization (CGH), or gene expression microarray chips. In terms of sensitivity and resolution, FISH is better than karyotyping and CGH, but worse than PCR-based assays for detecting small alterations. The former is limited to alterations of several Mb in size, whereas the latter can be designed to detect even single base mutations. Since FISH probes are typically at least 30 Kb in size, alterations need to be equally large for reliable detection. For this reason, FISH cannot detect small intragenic mutations and is best reserved for alterations that occur at the "cytogenetic" level. PCR is also more sensitive than FISH for the detection of abnormal fusion transcripts resulting from translocations, picking up as few as one per million cells. This is particularly useful when attempting to detect "minimal residual disease" or early recurrence, though the biologic relevance of such small fractions is not always clear and it is possible that in certain situations, PCR is "too sensitive". In contrast, FISH is more sensitive than PCR at identifying gene deletions or amplifications from samples of mixed cellularity, such as neoplasms with clonal heterogeneity or contaminating non-neoplastic elements [8]. It is estimated that sample purity must reach at least 70% tumor for such quantitative PCR-based assays and this is sometimes difficult to achieve in highly infiltrative neoplasms or tumors with abundant stroma and/or inflammation. FISH, on the other hand, can typically identify amplification in as few as 5% and deletion in 15-30% of cells within a sample. In comparison to LOH studies using microsatellite markers, FISH results are often similar, but not identical and each has its advantages and disadvantages. A common misconception is to equate the two techniques, stating that "FISH demonstrated LOH" for a region of interest. Since FISH measures absolute copy number rather than allele status, such a statement is inaccurate. Although LOH most often results from simple deletion, this is not always the case. For example, mitotic recombination of chromosome 17p may lead to loss of the wild type p53 allele and duplication of the mutated allele. Although one "allele" (maternal or paternal) would be lost in this scenario (i.e. LOH), there would still be two copies of the p53 gene, simulating the normal situation on FISH analysis. This was in fact, found to be the most common mechanism for p53 inactivation in gliomas [9] and therefore, FISH is not a suitable assay for detecting this type of loss in these tumors [10]. Another advantage of the LOH studies is the ease of evaluating large numbers of markers spanning the entire length of a chromosome or chromosomal arm. As emphasized above however, morphologic correlation is not possible unless regions of interest are microdissected first. LOH also requires matching germline DNA from the patient's leukocytes or microdissected normal tissue and this is not always available. Another recent application of FISH is high-throughput analysis via tissue microarray (TMA). This technology takes advantage of multispecimen paraffin blocks constructed from up to 1000 0.6-mm neoplastic, non-neoplastic, and control tissue cores of interest. Therefore, hundreds of specimens can be simultaneously evaluated on a single slide using TMA-FISH, markedly increasing efficiency and reducing data acquisition time, probe, reagent, and storage space requirements. A recently popularized approach is to initially screen a small number of tumors with gene expression profiling and then verify the resulting candidate genes in a large number of tumors, utilizing TMA-immunohistochemistry and TMA-FISH [11, 12] . Recent TMA studies have shown excellent morphologic, antigenic, and genomic preservation with high levels of concordance compared to the traditional whole slide approach [12, 13, 14, 15, 16] . For gene amplifications, TMA-FISH is particularly appealing, since interpretations are rapid, typically requiring only seconds per tissue core. For aneusomies and deletions, manual signal counts still remain tedious and time consuming, though automated spot counting software is currently under development and promises to further increase the efficiency of this technique. Recent technical advances have greatly enhanced the applicability of FISH. However, a number of limitations remain. One of the main disadvantages of FISH as a clinical tool is signal fading. By storing hybridized slides in a freezer and avoiding prolonged exposure to light, hybridization signals remain visible for up to a year or longer. However, a permanent record is not currently possible, unless chromogenic detection is used. Unfortunately, multicolor CISH is not as simple as multicolor FISH and currently available chromogens lack the spectral versatility, sensitivity and spatial resolution attainable with fluorochromes. Some are currently working on alternatives by developing non-fading fluorochromes (e.g. Bobrow MN and Roth KA; US patent pending) or improved protocols for multicolor CISH [17]. Alternatively, software solutions now make it possible to archive high volume FISH results through digital imaging. Other limitations include a variety of artifacts, particularly common in paraffin sections. It is for this reason that while the FISH protocol itself is often mastered quickly, interpretation requires significantly more experience. Most troublesome are truncation artifacts, aneuploidy, autofluorescence, and partial hybridization failure. Truncation artifact refers to the underestimation of target copy numbers due to incomplete DNA complements in transected nuclei. Therefore, it is important to include controls cut at the same thickness. In our lab, we usually set cutoffs for deletion based on median percentages of control nuclei with <2 signals plus 3 standard deviations. However, a number of other approaches have also been applied and are acceptable. Aneuploidy and polyploidy are particularly common in malignant neoplasms and can result in confusing signal counts. The inclusion of reference probes is most informative in such situations. Although the simplest approach is to interpret absolute losses (<2 copies) and gains (>2 copies), one may opt to delineate "relative" losses and gains compared with a reference ploidy, obtained either by flow cytometry or the assessment of multiple chromosomes by FISH. For example, the finding of 3 copies would be considered a relative gain in a diploid tumor, normal in a triploid tumor, and a relative loss in a tetraploid tumor. Lastly, one may combine a centromere and locus-specific probe from a single chromosome and determine their ratios. For example, cells with 4 chromosome 9 centromeres and 2 copies of the p16 region on 9p21 would be interpreted as having polysomy 9 and a hemizygous p16 deletion. A similar tumor with 4 centromere and no p16 signals would be interpreted as polysomy 9 with homozygous p16 deletion. Similarly, cells with 6 copies of EGFR might be interpreted as low-level amplification if there were only 2 chromosome 7 centromeres, but would represent polysomy 7 without gene amplification if there were 6 centromeres. Autofluorescence is a particularly pesky problem in paraffin sections. Since autofluorescent tissue fragments are typically larger and more irregular than true signals, they can often be disregarded. However, some fragments present at just the right size to simulate signals. In this case, the use of multiple filters is helpful, since autofluorescence will typically appear on both green and red filters, whereas true signals only fluoresce under one or the other. The problem of partial hybridization failure can be minimized by counting only in regions where the majority of cells have clear signals. FISH Assays A number of FISH protocols have been published and vary depending on individual preferences and specimen type. In general, the simpler protocols are preferable, since they require less hands on time, have fewer steps in which errors may be introduced, and are easier to troubleshoot. The basic steps are similar to those of immunohistochemistry and include deparaffinization, pretreatment / target retrieval, denaturation of probe and target DNA, hybridization (usually overnight), post-hybridization washes, detection, and microscopy/imaging. This is typically a 2-day assay, which requires roughly 3-4 hours the first day and 30 minutes the second day. Alternatively, same day protocols are possible with robust probes and automated systems are now available to reduce the required tech time to a minimum (e.g. www.vysis.com, www.ventanamed.com). A few technical caveats should be kept in mind. Similar to immunohistochemistry, microwave or heat-induced target retrieval often works better than chemical forms of pretreatment and significantly improves hybridization efficiency [8, 18, 19] . When this step is included, protein digestion may often be reduced or eliminated altogether. Nevertheless, optimal pretreatment and digestion varies from one specimen to another, depending on methods of fixation/processing. We have also found that some hybridization buffers are significantly more efficient than others and therefore work with lower probe concentration requirements (e.g. DenHyb from Insitus, www.insitus.com). This is particularly useful when utilizing expensive commercial probes, because they may last 5-20 times as long as they would when using the manufacturers recommended dilutions. This same company now offers a product called "SkipDewax", which allows one to deparaffinize and pretreat all in one step. Lastly, a variety of amplification steps are available for cases with weak signals. However, we have rarely found this necessary and in our lab, we prefer the simpler protocol and cleaner background associated with directly labeled fluorochrome probes (e.g. FITC, rhodamine), in contrast to indirectly labeled probes (e.g. digoxigenin, biotin) that require an additional step (e.g. fluorochrome-labeled secondary antibody) with or without further amplification. Nevertheless, dramatic levels of signal amplification are now achievable, particularly with tyramide signal amplification (TSA) or catalyzed reporter deposition (CARD) [17, 20, 21, 22, 23] . This technique takes advantage of peroxidase-mediated deposition of haptenized tyramine molecules, not only in the precise site of hybridization, but also in the nearby vicinity. This results in increases of signal size and up to 1000-fold or greater amplification. Although one possible application is marked reductions of probe concentration requirements, the more exciting potential is the use of smaller probes, perhaps down to the level of 1 Kb or less [24]. Therefore, TSA could potentially increase the sensitivity for small alterations, such as those detectable by PCR, while maintaining the morphologic advantage of FISH. Types of FISH Probes and Probe Development A number of different probe types are currently available for FISH. Centromere enumerating probes (CEPs) were among the first to be developed and are ideal for detecting whole chromosome gains and losses, such as monosomy, trisomy, and other polysomies. Because they target highly repetitive 171 bp sequences of a -satellite DNA, they are associated with excellent hybridization efficiencies and typically yield large, bright signals. However, sequence similarities in some pericentromeric regions result in cross-hybridization artifacts. Because of the inevitable cross-hybridization between centromeres 13 and 21 or between centromeres 14 and 22, these CEPs have been previously utilized as probes with four expected signals rather than two. A better solution however, is to use locus-specific or painting probes (see below) to enumerate these individual chromosomes. Anecdotally, we have also encountered cross-hybridization problems with CEP9, though the non-specific signals are usually dimmer and the utilization of either more stringent washes or lower probe concentrations sometimes resolves this problem. Also, an interesting phenomenon in non-neoplastic brain is that certain chromosomes are packaged into interphase nuclei with paired centromeres in close proximity, a concept referred to as somatic pairing [25, 26] . This is most dramatic with CEP17, but may be encountered to a lesser extent with other centromeres as well, including CEP1 and CEP8. Because of this close proximity, FISH yields an unexpected fraction of cells harboring a single large signal rather than two smaller ones, an artifact that could potentially lead to overinterpretation of monosomy. This somatic pairing is typically not encountered in brain neoplasms, such as gliomas. Therefore, interpretations with these CEPs are more difficult, if utilizing normal brain controls to establish cutoffs for monosomy. Despite these technical limitations, CEPs remain extremely useful for detecting aneusomies and are still among the best FISH probes available. The presence of similarly repetitive DNA sequences in subtelomeric regions has now led to the development of commercially available telomere probes for each chromosomal arm as well (e.g. www.vysis.com). Another chromosome-specific probe is the whole chromosome paint (WCP), in which a cocktail of DNA fragments is created to target all the non-repetitive DNA sequences in an entire chromosome. Because they cover such a large region of DNA, they yield more diffuse signals in interphase nuclei and are primarily utilized on metaphase spreads for resolving complex structural alterations. However, some of the smaller, acrocentric chromosomes yield sufficiently discrete signals that enumeration is possible in interphase nuclei. The WCPs also form the basis for advanced applications such as spectral karyotyping (SKY) and M-FISH, where each chromosome is painted with its own unique mixture of fluorescent colors. In contrast, another advanced application, comparative genomic hybridization (CGH), utilizes entire genomes as the "probe". Genomic tumor DNA is labeled in one color, normal DNA is labeled in another color, and equal quantities of both are competitively hybridized to a normal human metaphase in order to screen for regions of relative tumoral losses and gains. Techniques, such as FISH, CGH, M-FISH, and SKY are sometimes referred to as "molecular cytogenetics". Today, some of the most versatile and commonly used FISH probes are the locus-specific (LSI) or gene-specific probes. These probes target specific regions of interest and utilize single copy rather than repetitive DNA sequences. Therefore, in order to yield signals large enough to be detected in tissue sections, the probe typically needs to be at least 30 Kb in size. The largest FISH probes are often >1 Mb and most fall into the 100-300 Kb range. Until recently, commercially available LSI probes have been extremely limited in scope. Therefore, cloning vectors have been exploited for creating homemade FISH probes, including cosmids, bacterial artificial chromosomes (BACs), P1 artificial chromosomes (PACs), and yeast artificial chromosomes (YACs). Whereas in the past, this required a rather lengthy and tedious screening of vector libraries with PCR primers, the recent human genome initiative and mapping of entire BAC libraries has enabled rapid internet screening, utilizing DNA sequences of interest, gene names, or physical maps of chromosomes (e.g. http://genome.ucsc.edu, http://gdbwww.gdb.org). Similarly, mapped BAC clones spread throughout the human genome at 1-Mb intervals have also become available (www.ncbi.nlm.nih.gov/ncicgap; www.resgen.com). Therefore, it is now relatively simple to obtain a BAC clone localizing to virtually any region of interest, label the DNA with commercially available kits, and utilize it as a FISH probe. This recent development should greatly enhance the applicability of FISH to anatomic pathology. References Fuller CE, Perry A (2002) Fluorescence in situ hybridization (FISH) in diagnostic and investigative neuropathology. Brain Pathol 12:67-86 Paulus W, Bayas A, Ott G, Roggendorf W (1994) Interphase cytogenetics of glioblastoma and gliosarcoma. Acta Neuropathol 88:420-425 Boerman RH, Anderl K, Herath J, Borell T, Hohnson N, Schaeffer-Klein J, Kirchhof A, Raap AK, Scheithauer BW, Jenkins RB (1996) The glial and mesenchymal elements of gliosarcomas share similar genetic alteration. J Neuropathol Exp Neurol 55:973-981 Weber-Matthiesen K, Deerberg J, Poetsch M, Grote W, Schlegelberger B. (1995) Numerical chromosome aberrations are present within the CD30+ Hodgkin and Reed-Sternberg cells in 100% of analyzed cases of Hodgkin's disease. Blood 86:1464-68 Nolte M, Werner M, Vonwasielewski R, Nietgen G, Wilkens L, Georgii A (1996) Detection of numerical karyotype changes in the giant cells of Hodgkins lymphomas by a combination of FISH and immunohistochemistry applied to paraffin sections. Histochem Cell Biol 105:401-404 Perry A, Roth KA, Banerjee R, Fuller CE, Gutmann DH (2001) NF1 deletions in S-100 protein-positive and negative cells of sporadic and neurofibromatosis 1 (NF1)-associated plexiform neurofibromas and MPNSTs. Am J Pathol 159:57-61 Tubbs RR, Pettay J, Roche P, Stoler MH, Jenkins R, Myles J, Grogan T (2000) Concomitant oncoprotein detection with fluorescence in situ hybridization (CODFISH). A fluorescence-based assay enabling simultaneous visualization of gene amplification and encoded protein expression. J Mol Diag 2:78-83 Perry A, Nobori T, Ru N, Anderl K, Borell TJ, Mohapatra G, Feuerstein BG, Jenkins RB, Carson DA (1997) Detection of p16 gene deletions in gliomas: Fluorescence in situ hybridization (FISH) versus quantitative PCR. J Neuropathol Exp Neurol 56:999-1008 James CD, Carlbom E, Nordenskjold M, Collins VP, Cavenee WK (1989) Mitotic recombination of chromosome 17 in astrocytomas. Proc Nat Acad Sci USA 86:2858-2862 Perry A, Anderl KA, Borell TJ, Kimmel DW, Wang CH, O'Fallon JR, Feuerstein BG, Scheithauer BW, Jenkins RB (1999) Detection of p16, RB, CDK4, and p53 gene deletion / amplification by fluorescence in situ hybridization (FISH) in 96 gliomas. Am J Clin Pathol 112:801-809 Sallinen S-L, Sallinen PK, Haapasalo HK, Helin HJ, Helen PT, Schraml P, Kallioniemi O-P, Kononen J (2000) Identification of differentially expressed genes in human gliomas by DNA microarray and tissue chip techniques. Cancer Res 60:6617-6622 Moch H, Kallioniemi O-P, Sauter G (2001) Tissue microarrays: What will they bring to molecular and anatomic pathology? Adv Anat Pathol 8:14-20 Kononen J, Bubendorf L, Kallioniemi A, Barlund M, Schraml P, Leighton S, Torhorst J, Mihatsch MJ, Sauter G, Kallioniemi O-P (1998) Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nature Med 4:844-847 Schraml P, Kononen J, Bubendorf L, Moch H, Bissig H, Nocito A, Mihatsch MJ, Kallioniemi O-P, Sauter G (1999) Tissue microarrays for gene amplification surveys in many different tumor types. Clin Cancer Res 5:1966-1975 Camp RL, Charette LA, Rimm DL (2000) Validation of tissue microarray technology in breast carcinoma. Lab Invest 80:1943-1949 Hoos A, Urist MJ, Stojadinovic A, Mastorides S, Dudas ME, Leung DHY, Kuo D, Brennan MF, Lewis JJ, Cordon-Cardo C (2001) Validation of tissue microarrays for immunohistochemical profiling of cancer specimens using the example of human fibroblastic tumors. Am J Pathol 158:1245-1251 Speel EJM (1999) Detection and amplification systems for sensitive, multiple-target DNA and RNA in situ hybridization: looking inside cells with a spectrum of colors. Histochem Cell Biol 112:89-113 Henke R-P, Ayhan N (1994) Enhancement of hybridization efficiency in interphase cytogenetics on paraffin-embedded tissue sections by microwave treatment. Analytical Cell Pathol 6:319-25 Shi S-R, Cote RJ, Taylor CR (2001) Antigen retrieval techniques: Current perspectives. J Histochem Cytochem 49:931-937 Macechko PT, Krueger L, Hirsch B, Erlandsen SL (1997) Comparison of immunologic amplification vs enzymatic deposition of fluorochrome-conjugated tyramide as detection systems for FISH. J Histochem Cytochem 45:359-363 Schmidt BF, Chao J, Zhu Z, DeBiasio RL, Fisher G (1997) Signal amplification in the detection of single-copy DNA and RNA by enzyme-catalyzed deposition (CARD) of the novel fluorescent reporter substrate Cy3.29-tyramide. J Histochem Cytochem 45:365-373 Van Gijlswijk RPM, Zijlmans HJMAA, Wiegant J, Bobrow MN, Erickson TJ, Adler DE, Tanke HJ, Raap AK (1997) Fluorochrome-labeled tyramides: use in immunocytochemistry and fluorescence in situ hybridization. J Histochem Cytochem 45:375-382 Speel EJM, Hopman AHN, Komminoth P (1999) Amplification methods to increase the sensitivity of in situ hybridization: Play CARD(S). J Histochem Cytochem 47:281-288 Schriml LM, Padilla-Nash HM, Coleman A, Moen P, Nash WG, Menninger J, Jones G, Ried T, Dean M (1999) Tyramide signal amplification (TSA)-FISH applied to mapping PCR-labeled probes less than 1 Kb in size. Biotechniques 27:608-613 Arnoldus EPJ, Peters ACB, Bots GTAM, Raap AK, van der Ploeg M (1989) Somatic pairing of chromosome 1 centromeres in interphase nuclei of human cerebellum. Hum Genet 83:231-234 Arnoldus EPJ, Noordermeer IA, Peters ACB, Raap AK, van der Ploeg M (1991) Interphase cytogenetics reveals somatic pairing of chromosome 17 centromeres in normal human brain tissue, but no trisomy 7 or sex-chromosome loss. Cytogenet Cell Genet 56:214-216