FISH Technology OverviewUse of multicolored, closely linked DNA probe sets to reproducibly detect deletions in human cancer pathology (FFPE) sections using FISH. Introduction to DEL-TECT and FFPE TechnologiesThis section provides an overview on guiding principles and an optimal scoring system for using patent pending Deletion Detection Probe sets (DDPs) that have been designed primarily for FISH analysis of sections of formalin fixed paraffin embedded (FFPE) tissue in human solid tumors where deletions are common events when cancer is present, and loss of tumor suppressor genes (such as PTEN, RB1, p16, p53 etc) may be of crucial clinical significance. There is a compelling need to design robust probe configurations and scoring algorithms that allow for precise detection of deletion of genes such as PTEN, which appears to be of crucial significance in human malignancies such as prostate, colorectal, breast, and glioma. In this technology section we will discuss the general methodology used to identify deletions using interphase FISH analyses of FFPE sections, and we use the example of the detection of PTEN deletion in prostate cancer sections to illustrate the merits of this approach to achieving unprecedented accuracy in (FFPE) tissue sections.
Deletion Detection Probes by Four Color FISHThree color FISH for deletion detection also uses individually labeled probes. Each distinctly colored probe will bind to the region of a chromosome containing a tumor suppressor gene of interest. The probes are designed to span an appropriate genomic interval that is frequently deleted in human cancers. The tumor suppressor gene is placed in the middle region of the three probes and is detected with a specific color. This gene probe is flanked by differently colored probes on either side (Figure 3A).
FIGURE 3 Schematic depiction of DDP probe set configurations (panel A) and typical interpretations (panel B) using three color interphase FISH analysis of nuclei. In this example a BAC probe that encompasses a tumor suppressor gene (Tum Sup) is labeled with red fluorescence and flanking BAC probe "A" is labeled with green fluorescence while flanking BAC "B" with blue fluorescence. The upper "A" panel shows the configuration on a schematic chromosome. In panel B a normal nucleus (left) will have three pairs of each color signal. For simple hemizygous loss (center nucleus) of the Tum Sup gene the red BAC is deleted so only one signal is present, but probes A and B flanking the tumor suppressor gene are present in duplicate. For homozygous loss (right nucleus) both red signals are deleted while flanking probes are retained. Note that in comparison to Figure 2B and 2C in which randomness of signal distribution within the nucleus is apparent, the spot distribution in Panel 3B suggest a spatial relationship, or "traffic signal configuration" (assay implications discussed below).
In interphase nuclei without a deletion or rearrangement of the tumor suppressor gene, two separate clusters comprising three different colored signals will be visible (Figure 3B –left). In the simplest case involving a deletion there will be loss of the gene-specific signal, while the control flanking probes are still present. Sometimes both copies of the gene are deleted and this results in no signals, this is called a homozygous deletion (Figure 3B-right). If only one copy of the gene is lost, this deletion is called hemizygous (Figure 3 B –center). In addition to the simple deletion configurations shown in Figure 3, more complex patterns involving gain or loss of signals may be seen because of additional rearrangements close to the regions containing the flanking control probes. Some of the PTEN gene losses observed in advanced prostate cancer appear to also involve genes closely linked to PTEN that are associated with deletion events (discussed below). In addition complex signal configurations bearing additional spots may also arise from complex chromosomal gains due to unbalanced translocations, polysomy, or polyploidy. Any pattern differing from the simple patterns observed in normal nuclei is usually considered abnormal if it appears in a significant proportion of cells. The number and location of signals in aberrant patterns should be carefully evaluated and interpreted as it can provide valuable information of underlying chromosomal change. Truncation Signal Losses with FFPE SectionsAs discussed above, commercial probes for deletion detection from other companies use two color FISH, which include the gene of interest and a centromere specific reference probe from the same chromosome (Figure 2). The major limitation of simple two color FISH in a widely spaced configuration (centromere and gene of interest) is that detecting copy number changes in tissue sections will lead to higher false positive rates for deletions as nuclei are scored. The distance between the PTEN gene and the centromere control probes (as used by competitors) means that their distribution in the nuclear three-dimensional space is essentially random. Thus the chances of the control probe being excluded from the nucleus during the sectioning process are just as high as for the gene of interest. This "truncation effect" occurs because part of the cell and nucleus can be lost during the sectioning process, leading to loss of signal from the nucleus being scored. This problem is not a major consideration for the detection of amplifications or gain, where extra copies or signal clusters will still be clearly visible (REF). However, if the target chromosomal change is a deletion then the cutoff level for truncation signal losses (i.e. the false positive rate) can be quite high (see Figure 4).
FIGURE 4 Three dimensional schematic depiction of commercial (competitor) two color interphase FISH using normal cells that will have two red and two green spots per nucleus (see Figure 2 for details of probes). Because of the large distance and random distribution of green and red spots in the nucleus the truncation effect will have an equal probability of leading to signal loss of both probes, so that the control green centromeric probe is not as informative in discriminating between real chromosomal deletion events and false truncation exclusions by sectioning. The expected proportion of nuclei exhibiting truncation losses when using two color FISH has to be carefully determined using appropriate negative controls. The influence of both diameter and shape of nuclei is an important consideration when evaluation truncation false positive levels with interphase FISH. Nuclei in sections typically have a mean diameter of 8-10 microns, and confocal analysis has shown that intact nuclei often have an elliptical shape rather than the perfect spheres shown schematically in figures 4 and 5. In a recent study of truncation loss of signal by interphase FISH (in 5 micron histological sections of normal cells) was found to have threshold levels of only ~ 20% or normal bone marrow nuclei exhibiting signal losses due to truncations. In contrast ~60% of nuclei in normal liver sections exhibit truncation losses (Wilkins et al.). This variation in false positive rates for detecting losses is due to the cutting artifacts of truncation, that have a greater affect when the larger and irregularly shaped liver cells are more frequently bisected than bone marrow cells, using the 5 micron sections. It is therefore essential that the false positive, likely to come from truncation, be determined by comparison with normal nuclei to the neoplastic tissue nuclei of interest, for all deletion FISH assays. One practical solution for minimizing the truncation effects is the use of 3 or 4 color FISH and proper probe design in the region subject to deletion (i.e. the genomic distances between control probes and the tumor suppressor gene of interest.) The false-positive rate due to truncation can be reduced by choosing an optimal genomic distance between each of the probes that constitutes the three color FISH assay design. In this way it is possible to determine whether a given gene is deleted or not with a higher degree of certainty. We compare 4 color FISH (DDP in-house probe set) to 2 color FISH (competitor) is presented in Table 1 below.
FIGURE 5 Principle of DDP in house FISH assay design. LEFT PANEL: Position of probes used by our group to detect PTEN deletion events using four color FISH in prostate cancer (REFS). RIGHT PANEL: Three dimensional schematic depiction of four color interphase FISH using normal cells that will be expected to have two red, green, blue and violet spots per nucleus. The blue centromere probe is used to determine if monosomy 10 may be present. The DDP three probe set comprise a red PTEN probe which is flanked by the violet TSPAN15 on the centromeric side, and by the green FAS gene on the telomeric side. Because of the close proximity or flanking control probes, the truncation effect is more likely to impact the entire DDP set so that scoring algorithms for deletion can be designed with improved discrimination between real deletion events and the false truncation exclusions caused by sectioning. In the DDP assay analyses of normal nuclei such as in the schematic section shown here are used to determine threshold levels of "technical loss events" for each probe due truncation exclusion during sectioning.
Spot Count Variability in FFPE SpecimensThe establishment of cutoff values to set the threshold or false positive level is also influenced by other biological variables affecting both FISH signals and relative nuclear volume in tumor sections:
In addition to these nuclear variables:
Thus, one should take into account that these factors cannot be properly modeled in negative controls used to establish cutoff levels for interphase FISH assays. This is of particular importance in samples with low-tumor cell content or additional subclonal genomic changes. Background to PTEN Deletions in Prostate CancerThe phosphatase and tensin homolog (PTEN) is a protein, which in humans is encoded by the PTEN gene. Inactivation of this tumor suppressor gene is a critical step in the development of many types of cancers and in the inherited developmental defect known as Cowden syndrome. Early reports about the PTEN gene focused on small changes of DNA sequence or point mutations that led to inactivation of PTEN protein function (Whang et al, 1998; Kwabi-Addo et al, 2001). In addition it has been shown that, in a subset of tumors, the PTEN gene may be inactivated by epigenetic events such as promoter methylation (REF). However in recent years it has become apparent that genomic deletions of several hundred kb around the PTEN locus are also common in prostate cancer (Yoshimoto et al 2006), and their presence carries an unfavorable prognosis (Yoshimoto et al 2007 and 2008). The simplest PTEN FISH analyses in prostate cancer divides the signal counts into two classes: two (or more) copies means not deleted; and one or zero copies means deleted. Peer review studies show that deletions are present in about 40% of tumors. Genomic deletions appear to be interstitial (around and including cytoband 10q23), varying in both size and starting points centromeric to the PTEN gene on one side and finishing the other (telomeric) side of the locus. Often, the closely linked telomeric FAS gene is deleted along with PTEN. Recent analyses of one study summarizing 330 localized and hormone-refractory prostate cancers identified deletions in 132 tumors and showed that the deletions always center on the PTEN locus (Yoshimoto CSH mtg 2010). This data is consistent with the PTEN gene being the driver of genomic losses in cytoband 10q23 in prostate cancer. Peer review studies with other tumors (renal cell cancer and endometrial cancer) have shown that genomic deletion events are uncommon using PTEN FISH analyses. Interestingly, endometrial cancer has reduced PTEN protein expression and there is some evidence that this tumor has preference for inactivating PTEN by epigenetic methods (unpublished). Additional FISH studies have produced evidence that PTEN genomic deletion events appear to be quite common in triple negative (basal) breast cancer (work in progress). Collectively, published data (Yoshimoto 2006, 2007, 2008, Sircar 2009, Bismar 2010) suggests that prostate cancer has a uniquely high incidence of relatively large deletions and genomic rearrangements affecting PTEN, compared to other human cancers. Various studies, and also work in PTEN mouse models draw attention to the role of PTEN functional losses in both the onset of prostatic neoplasia and in tumor progression to advanced aggressive disease (Squire 2009). PTEN Deletions in Normal Prostate Tissue, Adenocarcinoma and MetastasisWhen a deletion is present, PTEN gene loss is the classification assigned to all examined cells within the area of the gland identified histologically as adenocarcinoma. In contrast areas of the gland identified as histologically benign or normal, are always classified "not deleted" as the cell counts do not exhibit signal loss patterns above technical background levels for the control probes. Only rarely are more complex heterogeneous signal patterns observed with some focal areas of the tumor exhibiting loss and surrounding areas of tumor retaining of both copies of the gene. Peer review studies interpret such findings as an indication that loss during tumor progression in a subset of tumor cells. PTEN gene results can be further subdivided into three classes: two copies mean not deleted; one copy means deleted (hemizygous); and zero copies mean deleted (homozygous). In advanced prostate cancers and metastatic tumors both copies of the PTEN gene are often lost, and the classes frequently observed are homozygous deletions (Yoshimoto et al 2007, Sircar et al 2009). Significantly homozygous PTEN deletions portend an earlier recurrence of disease and a higher likelihood of metastasis, suggesting that haploinsufficiency (reduced protein levels) of PTEN expression together with genomic losses of closely linked genes may underlie more aggressive prostate cancer (CaP-Sircar et al 2009, Bismar 2010, Squire 2009). Hemizygous deletions (one copy of the gene) are more common in primary prostate cancers (Yoshimoto et al 2006, 2007, and 2008). In one study (Yoshimoto et al 2007) comparing primary prostatic adenocarcinomas to nodal metastatic disease as paired samples derived from ten patients found that only one of the 10 patients retained both copies of the PTEN locus in his primary tumor and in his metastatic lymph nodal biopsy. Hemizygous PTEN deletion was found in both the primary and the metastatic nodal tumor samples in four of 10 patients. Homozygous PTEN deletion was found in both the primary tumor and their metastatic lymph nodes in three of the 10 patients. Significantly, two of the 10 patients with a hemizygous PTEN deletion in their primary adenocarcinomas, had positive lymph node biopsies that had acquired a homozygous PTEN deletion. These findings suggest that loss of the remaining PTEN locus is associated with metastasis. Details of Interphase FISH Copy Number Assay for PTEN in Prostate CancerPTEN copy number, using prostate cancer 5-micron tumor tissue sections, is typically evaluated by counting spots for each probe in 50 to 100 non-overlapped, intact interphase nuclei. The nuclei are identified using a blue DNA specific dyeito stain nuclei. The Pathologist identifies areas of pure adenocarcinoma by referring to corresponding specifically stainedii tissue. The identified tissue is suitable for tumor cell enumeration (spot counting and color classification) using the PTEN FISH assay. Using the Abbott Molecular two color FISH PTEN probe set (depicted schematically in Figure 2) a series of interphase FISH studies of PTEN in prostate cancer (Yoshimoto 2007, 2008) were performed. Hemizygous deletion of PTEN based on hybridization and enumeration in 10 control cores (derived from normal non-diseased glands), and were defined as >20% iii of tumor nuclei containing one PTEN locus signal and two CEP 10 signals. Homozygous deletion of PTEN was classified by the simultaneous lack of the both PTEN locus signals and by the presence of two CEP 10 signals in more than 30% of cells.iv This simple statistical approach may be limited under conditions of signal noise or "extreme outliers" in more complex or morphologically complex tumor tissue of interest. More elegant statistical methodsv can be used as a more robust measure of standard deviation when such outliers are present. However, such statistical approaches have a limited role in addressing the high levels of inherent biological and technical noise encountered when scoring FISH signals FFPE tumor sections. The advantages of the multi-probe multicolor PTEN set are its ability to provide a better standard for evaluation of complex samples. Table 2 is an overview flow diagram of the DDP assay. References
|
