Simultaneous Detection of Multiple Genetic Aberrations in Single Cells by Spectral Fluorescence in Situ Hybridization¹

Acquired cytogenetic abnormalities have proven to be valuable biomarkers in monitoring a patient’s clinical course because of their robust association with tumor aggressiveness, response to therapy, and overall survival (1, 2, 3). Although recent technological advances in cytogenetics have improved the growth of neoplastic tissue in vitro, a considerable number of cancer cytogenetic studies are characterized by normal or otherwise noninformative karyotypes, the latter mostly because of low mitotic indices or poor chromosome morphology. Molecular cytogenetic and concurrent flow cytometry studies have established that some “karyotypically normal” studies are the result of in vitro cell selection of infiltrating lymphocytes after short-term culturing and do not accurately represent the genotype of the tumor (4). These cell cycle kinetic/in vitro tissue-processing concerns are particularly evident in ALL,³terminally differentiated lymphoproliferative disorders, and most epithelial solid tumors, e.g., 51% of karyotypically normal breast tumors contain a markedly aneuploid clone by flow cytometry(4). The reported frequency range of 30–50%karyotypically abnormal studies in multiple myeloma is discordant with 67–90% incidence of aneuploidy detected by flow cytometry and FISH analyses (5, 6, 7). These data suggest critical aberrations occur at the chromosomal and gene level in these cancers, and their detection in patient specimens is necessary for improved prognostication and MRD detection. Current MRD assays of flow cytometry, PCR, and standard FISH have great value but notable limitations (8). Flow cytometry provides ploidy data for a tumor but lacks the sensitivity to describe specific genetic alterations or to detect small variations in DNA content attributable to partial chromosomal gain or loss. Conversely, PCR is a sensitive method used to detect well-characterized chromosomal translocations;however, like flow cytometry, whole or partial chromosomal gains or deletions cannot be reliably detected. Standard FISH protocols, using chromosome-specific or locus-specific DNA probes, identify and quantify one to two known cytogenetic aberrations in both metaphase chromosomes and interphase nuclei. The sensitivity of a FISH assay is directly proportional to the number of DNA probes used simultaneously, e.g., single color FISH ranges between 10⁻¹ to 10⁻³, whereas the use of two to three DNA probes may increase FISH sensitivity from 10⁻³ to 10⁻⁴(9, 10). As a result, simultaneous detection of more than three chromosomal or gene targets should increase the overall sensitivity of FISH, holding greater promise for improved detection of MRD at the single cell level.

On the basis of this premise and previous reports (11, 12)demonstrating the potential of combinatorial labeling and digital imaging microscopy, we sought to develop a quick and sensitive screening assay to detect cytogenetic markers, particularly less specific, non-PCR amenable cytogenetic markers, in nonmitotic tumor cells. S-FISH allows the simultaneous visualization of 3–10 chromosomal sites in a single cell after a single hybridization. The technique combines the resolution and sensitivity of standard FISH with spectral imaging, thus permitting immediate correlation of cell morphology with genotype. S-FISH may be designed to identify common numeric trisomies observed in prenatal tissue (12), or the probe panel may be comprised of disease and locus-specific probes. In neoplasia, S-FISH raises the possibility of a quick screen for common disease-specific abnormalities at diagnosis or, once clonal aberrations have been identified in a tumor, individualized patient-specific panels to screen for MRD. The objective of this pilot feasibility study was to create a generic “aneuploidy” oncology panel to detect gains or losses of chromosomes 7, 8, 9, 10, X, and Y to ask whether S-FISH could detect clonal cytogenetic aberrations, using a variety of specimens which could ultimately be used to monitor MRD.

The experimental design was a simple feasibility study to determine the basic advantages and limitations, delimit protocol/data collection parameters, and assess interpretative issues of the S-FISH assay in various neoplastic disorders. Peripheral blood samples from two normal volunteers (one male and one female) served as normal controls. The test population was comprised of residual cytogenetic cell pellets from six breast tumors (cases A1-A6), 10 bladder washings (cases B1-B5,B7-B10 and 12), two bladder tumors (cases B6 and B11), and one cryopreserved ALL sample. Each S-FISH assay was performed using a 5-cell analysis to confirm clonality established by the validation genetic assays. Validation was performed by comparing S-FISH results with CC, standard FISH, or flow cytometry data. Discrepant results were verified by additional FISH studies. A standard 200-cell FISH analysis was performed to verify chromosome copy number using the following probes: chromosome 7 α-satellite probe (Oncor, Inc., Gaithersburg,MD; or Vysis, Downers Grove, IL); chromosome 8 α-satellite probe(Oncor); chromosome 9 α-satellite probe (Vysis) in bladder washings B1-B8, B10, and B11; and chromosome 9 classical satellite probe (D9Z1;Oncor) for bladder washings B9-B12.

Cytogenetic preparations were prepared using established methods and residual cell pellets stored at −80°C. Before S-FISH analysis, the residual cytogenetic cell pellets were washed in 3:1 methanol:glacial acetic acid solution and dropped onto precleaned, nonsilanized slides. Slides for S-FISH analysis were pretreated in 2 × SSC at 37°C for 30 min, digested in pepsin (Sigma Chemical Co., St. Louis, MO; 6 μg/ml in 0.01 n HCl, 37°C) for 1 min,washed twice in 1 × PBS and one time in 1 × PBS/50 mm MgCl₂ for 5 min each, fixed in formaldehyde (1% in 1 × PBS/50 mm MgCl₂, room temperature) for 10 min, washed in 1 × PBS for 5 min, dehydrated in ethanol series (70%, 80%, and 95% for 2 min each), and then denatured in 70% formamide/2 × SSC (pH 7.0) at 72°C for 2 min.

Six α-satellite probes were chosen to represent chromosomes 7, 8, 9,10, X, and Y (Table 1). Probes for the α-satellite regions of chromosomes 8 (pJM128), 10(pA10RR8), and Y (DYZ3) were obtained from the American Type Culture Collection (Rockville, MD), and the α-satellite probe for chromosome 9 (CEP 9) was obtained from Vysis. Probes for theα-region of chromosomes 7 (pZ7.5) and X (pDMX1) were generous gifts from Dr. Mariano Rocchi, University of Bari, Bari, Italy. The QIAfilter plasmid midi kit (Qiagen, Valencia, CA) was used to isolate DNA. With the exception of the prelabeled chromosome 9 probe, the probes were labeled by nick translation using standard technique.

Before mixing the S-FISH probe cocktail, each probe was hybridized individually to normal lymphocytes to test fluorescence intensity. On the basis of the intensity results, probe concentration for each chromosome was adjusted to ensure comparable fluorescent intensity among probes and coprecipitated with 1 μg of Cot-1 DNA (Life Technologies, Inc.) and 2 μg of human placental DNA (Oncor)/slide. The probe pellet was dissolved in 1 μl of H₂O,followed by 10 μl of Hybrisol VI (Oncor)/slide. The probe mixture was denatured at 72°C for 7 min, applied to the slides, covered with 22 × 22-mm coverslips, and sealed with rubber cement. The slides were hybridized overnight at 37°C and washed with 0.25 × SSC at 72°C for 5 min. Biotin-labeled probes were detected with 2.5 μg/ml Cy5.5 conjugated avidin (Rockland,Gilbertsville, PA) in 1 × PBS, 1% blocking reagent(Roche Molecular Biochemicals, Indianapolis, IN), and 1% BSA for 5 min. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole.

Images of the individual interphase nuclei were captured and analyzed using the interferometer-based SD200 Spectracube system (Applied Spectral Imaging, Migdal HaEmek, Israel) attached to a Zeiss Axioplan II microscope with a 150-W Xenon UV light source. Before S-FISH analysis, the system was calibrated as described (12). A combinatorial labeling scheme of the four fluorochromes (Table 1),including spectrum red, spectrum green, spectrum orange, and Cy5.5, was used to generate the six spectral images. The spectrum for each fluorochrome is recorded and stored as a reference in the combinatorial table file, and a unique pseudo-classification color is assigned for each target. S-FISH hybridization domains were visualized using a 63 × objective and a triple band pass filter. During the analysis, the spectral profile of the signals in the interphase nuclei recovered by Fourier transformation (Fig. 1,B) was compared with the references in the combinatorial table, determining which fluorochromes were present and creating a classified image to reveal chromosome identity (Fig. 1 C). The combinatorial table was adjusted to classify any signal containing more than two fluorochromes as an overlap, and an inconclusive result was recorded for the overlapping signals. Only compact, nonoverlapping signals in the same focal plane were scored. Determination of ploidy for each chromosome was based on the number of separate domains that matched their respective reference spectrum. As in standard FISH,doublet or split signals were counted as one signal. Each cell took approximately 10 min to analyze.

Sensitivity and specificity of the S-FISH assay and instrumentation were evaluated using normal lymphocytes. In normal controls, DNA probe signals were observed in the same focal plane in ∼95% of nuclei. However, because of the three-dimensional structure of the nucleus,signals in ∼5% of cells were in different focal planes. In this study, cells were excluded from analysis if the nuclei or signals were overlapping or if the signals were found in different focal planes. To determine the sensitivity of the assay, 50 nuclei from normal male and female control samples were analyzed for six chromosomal targets. Eighty and seventy-eight percent, respectively, of the nuclei scored exhibited a normal disomy signal pattern for the tested autosomes and a normal gonosomal complement. This result is in agreement with the 70%of interphase lymphocyte nuclei displaying the expected number of signals in a previous report (12).

Table 2 compares the results of CC, standard FISH, and S-FISH for six breast cancer cases. As mentioned above for normal lymphocytes, 5–10% of tumor interphase nuclei were not scorable because of focal plane distortions. Numerical aberrations were easily detected in all of the six cases. Case A1 was a grade III breast tumor with a 57–62 chromosomal modal range and a DNA index of 1.28. S-FISH correctly identified consistent gains of X, 7, 8, and two additional copies of chromosome 10 with disomy 9 (Fig. 1). Similar to composite karyotypes of solid tumors, S-FISH results were recorded in chromosomal ranges because of intratumor heterogeneity. Taking this into consideration,concordance with minor variability was observed among the three cytogenetic methods used (Table 2).

The cytogenetic morphology of case A2 was marginally acceptable,reported as an incomplete composite karyotype in the near 4n-ploidy range. S-FISH provided valuable cytogenetic data and explained the inconsistency between the 2.5 DNA index (∼5n) reported by flow cytometry and the limited CC information. S-FISH identified a 4n content for chromosomes 9 and 10 with five chromosome 7 signals, 11 chromosome 8 signals, and six copies of chromosome X (Fig. 2). Suspected structural rearrangements were indicated by the smaller signals and the doublet observed in the probe profile.

The S-FISH result was concordant with classical cytogenetics in case A5; however, like standard FISH, S-FISH detected additional copies of autosomes not found by CC. This discordance was related to the overall complexity and suboptimal chromosome morphology resulting in many unclassified marker chromosomes. S-FISH did successfully detect the multiple copies of chromosome 7 that CC and FISH found, although chromosome 7s were not detected in one interphase cell, possibly because of either poor hybridization of the chromosome 7 probe for that cell or the spatial (three-dimensional) orientation of the cell. Standard FISH detected tumor cell populations with tetrasomy and pentasomy for chromosome 7 not detected by the other two methods. This inconsistency was apparently because of the larger sampling size(200 cells) of standard FISH. Although a five-cell screen was proposed for this study, the number of cells to be analyzed by S-FISH should be customized to answer the objectives of the research or clinical investigation, e.g., a minimal residual study would require a larger sample size than a study designed to confirm clonality.

Monosomy X and 10 were easily identified by S-FISH in the hypodiploid breast tumor (case A6); however, the complexity of the tumor karyotype,with structural aberrations for every chromosome with the exception of chromosome 2, was not evident. As a potentially significant clinical tool, a breast cancer S-FISH probe panel should be optimized to evaluate commonly observed genetic aberrations associated with tumor progression or a poor response to therapy such as loss of 1p, gain of 1q, loss of TP53, amplification of Her-2/neu at 17q12, CCND1/EMS1 at 11q13, and 20q13.2 (ZNF217/NABC1; Ref. 3, 13).

FISH has proven to be a powerful technique for predicting recurrence of bladder cancer in noncycling, exfoliated cells from bladder irrigations(14, 15, 16). Accordingly, S-FISH was performed using the residual cytogenetic cell pellets from 10 bladder washings (cases B1-B5, B7-B10, and B12) and two bladder tumors (cases B6 and B11). S-FISH and standard dual color FISH were concordant in 11 of 12 cases(cases B1-B11; Table 3). Neither standard FISH nor S-FISH detected numeric aberrations in cases B1-B4. In case B5, FISH and S-FISH identified the loss of chromosome 9 as the sole numeric aberration. No mitotic cells were observed in case B6 by CC; however, using the residual cell pellet,trisomy 9 was identified by S-FISH and confirmed by standard FISH.

In three concordant cases, S-FISH detected additional cytogenetic aberrations. In case B7, standard dual color FISH detected one to two additional copies of chromosome 7 and a “normal” disomy chromosome 9 complement. Using the same cell preparation, S-FISH identified a 4n tumor with loss of chromosome 9. Both quantitative assays yielded correct information; however, the interpretation was modified to reflect the loss of chromosome 9 in a tetraploid (4n) tumor when multiple probes were analyzed simultaneously, instead of tetrasomy 7 in a diploid (2n) tumor using a dual probe standard FISH assay. The increased sensitivity of S-FISH correctly identified loss of chromosome 9 in this bladder washing, a recurring aberration in transitional carcinoma of the bladder. Additional aberrations were also revealed in cases B8 (+8) and B9 (+10) by S-FISH, underscoring the improved sensitivity of the S-FISH assay by surveying multiple chromosomal targets in a single hybridization. Moreover, the reporting of age-related and acquired genetic changes at the DNA and chromosomal levels, e.g., trisomy 7 and loss of the Y chromosome as sole aberrations in non-neoplastic tissue (17), argues for the requisite to identify a second abnormal genetic marker for suspicious cells containing these specific genetic aberrations.

The signal pattern for chromosome 9 differed between S-FISH and standard FISH in cases B10 and B11. Two different DNA probes represented chromosome 9 in these assays; a classical satellite DNA probe was used in standard FISH, and an α-satellite probe was used in S-FISH. Further investigation of the discrepancy in case 10 revealed a normal disomy content for chromosome 9 using the α-satellite probe with under-representation of chromosome 9 signal when the classical satellite probe (D9Z1) was used. In case B11, one normal signal and one smaller (deleted) signal was observed, a finding consistent with the unbalanced der(9)t(9;20)(q13;q12) identified by CC. The classical satellite DNA probe hybridizes to the heterochromatin region of the chromosome 9 immediately distal to the α-satellite centromeric region. These results suggested the breakpoint in the der(9) occurred between the centromere and D9Z1. On the basis of the results of this study and the reported frequency of chromosome 9 aberrations in bladder cancer, a classical satellite DNA probe to evaluate loss of chromosome 9 in bladder cancer by S-FISH is recommended. In addition, allelic loss in band 9p21, the normal cellular locus for the p15/p16 cell cycle regulatory genes, and the 9q31–34 region is frequent in bladder cancer (18). Incorporation of disease-specific probes to evaluate allelic loss of these regions, in association with a chromosome 9 enumeration probe,should be considered for the next validation phase. This latter revision would also provide valuable control probe data necessary to reduce false positives associated with somatic pairing of classical DNA chromosome 9 sequences (14, 19).

The S-FISH data from the massively aneuploid B12 tumor were incomplete. Although both FISH and S-FISH detected severe intratumor cell heterogeneity, reported as vast ranges in chromosome copy number,standard FISH detected a higher ploidy level (8n); signal overlap precluded precise counts by S-FISH. Using a six-probe panel,ploidy levels higher than 6n are too difficult to resolve by S-FISH analysis.

To determine whether S-FISH has clinical utility in detecting MRD, an ALL sample collected at day 30 after induction chemotherapy was processed. The CC study revealed a normal karyotype in a 20-cell analysis. The corresponding pathology report described a small blast population most consistent with a regenerating marrow; however,residual disease could not be ruled out. Because cytogenetic studies at disease presentation exhibited hyperdiploidy with gains of chromosomes 7 and 8, residual cryopreserved material was evaluated. S-FISH readily identified the presence of trisomy 7 and trisomy 8 in a blast, thus correlating morphology with genotype and confirming the presence of MRD.

The limitations of S-FISH are similar to those described previously with standard FISH (8). Overlapping or signals in close proximity result in fluor-blending, making it difficult, if not impossible, to accurately quantitate numeric changes especially in the highly aneuploid tumors. The three-dimensional aspects of interphase nuclei are problematic with the current Skyvision software version,prohibiting the evaluation of overlapping signals or the capturing of signals at different focal planes. When preparing a specimen for S-FISH, care must be taken to ensure the nuclei are well preserved throughout the cell-processing procedure. In our experience, signals were detected and scorable in 90–95% of interphase nuclei using tissue obtained from the following sources: bone marrow, peripheral blood, bladder washings, isolated nuclei from fresh breast and bladder tumors, and cell lines. However, the percentage of interphase nuclei suitable for analysis was substantially lower in paraffin-embedded tissue and cryopreserved tissue. This three-dimensional focal plane obstacle varied depending on type/timing of fixative and cryopreservation used. In paraffin-embedded material, it was evident in both paraffin-embedded sections and isolated nuclei from embedded tissue, including formalin-fixed material. Assay modifications currently in progress to lessen this inherent three-dimensional effect of paraffin-embedded or inadequately cryopreserved cells include the incorporation of Z-axis stacking or the use of computerized confocal microscopy with merging imaging software. Alternatively, the cost-effective approach of using an agar overlay to flatten nuclei on slides has potential (20); however, our attempts of this latter suggestion have not been successful.

Careful selection of FISH probes is key to providing valuable prognostic information and reducing false negatives, as observed in this study using the α-satellite chromosome 9 probe in bladder irrigations. In samples with suspected loss of a FISH signal, the simultaneous use of a locator probe mapped to another region of the chromosome is highly recommended. As a quick screening tool, a five-cell analysis of the neoplastic cells appeared to be sufficient. However, for cases evaluated for MRD, intratumor heterogeneity, minor focal clones, or clonal evolution of disease, or if multiple or unrelated clones are suspected, additional cells will need to be analyzed. Accordingly, because of the labor-intensive technical component of this analysis, the efficiency of the assay would be greatly enhanced with high throughput automation.

On the basis of the results of this study, S-FISH appears to be a rapid and reliable screening method to detect numeric chromosomal aberrations in nonmitotic cells from a variety of pathological samples. Ploidy,tumor heterogeneity, and MRD data were easily detected. The number of probes used is currently limited by the fluorophore combinations,discriminated by the filters and imaging instrument used, and the ploidy level of the tumor being analyzed. In our experience, using a conventional microscope equipped with an interferometer/CCD camera to capture images and Skyvision software, ploidy levels >6n were not accurately scored using a six-probe panel because of either overlapping signals or the blending of fluorophores when the signals were in close proximity to one another. When evaluating nuclei in the near diploid range, nine probes were successfully assessed with precision and reproducibility. Clinical and research laboratories that do not have access to the specialized equipment described in this study could modify the S-FISH assay to the recently described color-changing technique that simply requires a conventional fluorescent microscope equipped with three fluorescent filters and a digital camera(21).

Single cell analysis allows for simultaneous genotype/morphology correlations. Extending this technology to include other non-PCR amenable chromosomal regions [e.g., del(6q), del(5q),del(9p)] or disease-specific loci could potentially increase the accuracy of genetic diagnoses of neoplastic disorders by not only detecting multiple aberrations in a single interphase cell but also,with concomitant immunophenotyping detection, greatly enhancing its ability to correlate genotype with phenotype (22). As cytogenetic and DNA, RNA, and protein array databases reveal common genetic defects in specific tumors, future S-FISH panels could be designed to analyze tumors for clonal evolution of disease, metastatic potential, or specific biological subsets, e.g., core binding factor leukemias. Moreover, tumor markers could be isolated from a patient and subsequently individualized, or patient-specific panels could be developed to monitor the disease throughout its clinical course. Once the desired assay modifications are achieved, a larger validation trial is needed to assess the sensitivity,specificity, and practical utility for a quantifiable S-FISH assay.