The role of cytogenetic analysis in the diagnosis and management of haematological malignancies is undisputed.The accuracy of cytogenetic diagnosis has improved steadily over the past 20 years, primarily due to a series of technical developments. However, despite improvements in high-resolution banding and culture methods to detect the chromosomally abnormal cells, many haematological malignancies are retractable to conventional cytogenetic analysis. This may be due to the presence of multiple abnormal clones, complex rearrangements, a low mitotic index, or poor chromosome morphology. Since the late 1980s a range of techniques based around fluorescence in situ hybridization (FISH) have greatly enhanced cytogenetic analysis. These use a variety of nucleic acid sequences as probes to cellular DNA targets and serve to bridge the gap between molecular genetic and conventional cytogenetic methods. Virtually any genomic DNA can now be used as a probe with which to investigate a wide variety of DNA targets, from metaphase chromosomes to mechanically stretched DNA fibres. The simultaneous detection of multiple target regions is also possible, using differentially labelled probes detected by different colours. In research, FISH has played a pivotal role in the identification of non-random translocations and deletions, pinpointing regions which contain genes involved in leukaemogenesis. Now, at the cutting edge, a new set of resources and technical innovations herald a new era for molecular cytogenetics, with colour karyotyping, comparative genomic hybridization (CGH) microarrays and mutation detection using padlock probes providing the promise of the future. The number of applications for FISH is almost unlimited (see Table I for some pertinent examples). This review will concentrate on the most recent developments in FISH which have had a considerable impact on the cytogenetic diagnosis and study of haematological malignancies, with some insight into the possible future roles for this flexible technology. The application of FISH to metaphase chromosomes provides unequivocal evidence of chromosome rearrangements. There are many different types of cloned or uncloned DNA which can be used for as a probe for FISH (reviewed in Buckle & Kearney, 1994). However, the most commonly used probes in cytogenetic analysis of haematological malignancy are: (i) repetitive sequence centromeric probes, (ii) whole chromosome paints, and (iii) locus-specific probes. Chromosome-specific centromeric probes which target tandemly repeated alpha (or beta) satellite sequences present in the heterochromatin of the chromosome centromeres are used to detect numerical chromosome abnormalities. Centromeric probes are commercially available for all human chromosomes and these provide a rapid and simple way of enumerating specific chromosome pairs, both in metaphase and interphase. This type of analysis is useful in many types of leukaemia where the chromosome morphology is poor and banding indistinct, such as in hyperdiploid acute lymphoblastic leukaemia (ALL). However, centromeric probes only give information on the number of centromeres of a particular type present; they cannot tell whether the chromosome is structurally abnormal. Whole chromosome painting probes are complex mixtures of sequences from the entire length of a specific chromosome. These are also available for all human chromosomes, and can be used to delineate chromosome pairs (Cremer et al, 1988; Pinkel et al, 1988). Whole chromosome painting probes (paints) can be derived from chromosome-specific libraries, PCR amplification of flow-sorted chromosome fractions, or microdissected DNA specific for each chromosome (Collins et al, 1991; Telenius et al, 1992; Vooijs et al, 1993; Guan et al, 1996). Chromosome paints are most useful for identifying the components of highly rearranged and marker chromosomes, where the banding pattern cannot be relied upon. However, their usefulness is limited to metaphase analysis, as the extended chromosome domains in interphase are often diffuse and difficult to quantitate. In addition, chromosome painting is a relatively insensitive technique and cannot detect small interstitial deletions, duplications or inversions. The resolution for the detection of small telomeric translocations is also limited. Single locus probes detect specific sequences present in only one copy. When using these probes the efficiency of hybridization needs to be considered; the larger the target sequence the more efficient the hybridization. Single-copy probes cloned in cosmid, YAC, P1, PAC and BAC vectors all give reliable FISH signals, with a fluorescent signal on both chromosome homologues in >90% of metaphases. Structural rearrangements detected using this type of probe include translocations, inversions and specific deletions (Dauwerse et al, 1990; Tkachuk et al, 1992; Sacchi et al, 1995; Jaju et al, 1998). The use of specific gene probes for chromosomal translocations has simplified the process of identifying known translocations, especially in complex or masked versions of the translocation (e.g. BCR/ABL, PML/RARα fusions), and has particular applications for interphase analysis. One of the greatest advances in cytogenetic analysis facilitated by FISH has been the ability to use non-dividing cells as DNA targets, referred to as interphase FISH (Cremer et al, 1986). This enables the screening of large numbers of cells and provides access to a variety of sources of haemopoietic cells including blood and bone marrow smears and haemopoietic progenitor cells from colony assays (Bentz et al, 1993; Poddighe et al, 1993; Mühlmann et al, 1998). This has considerable advantages for some haemopoietic malignancies, where the proliferative activity is low, or when the mitotic cells do not represent the neoplastic clone, for example chronic lymphoblastic leukaemia (CLL), Hodgkin's disease, multiple myeloma. Interphase FISH permits the identification of both numerical and structural chromosome abnormalities both as an aid to cyto-genetic diagnosis and for monitoring disease progression. Interphase FISH has had a major impact on the cytogenetic analysis of B-CLL, revealing a much higher incidence of trisomy 12 than found by conventional cytogenetic analysis (Anastasi et al, 1992; Garcia-Marco et al, 1997). An examination of the relationship between clinical stage and trisomy 12 showed an association with atypical morphology, advanced stage of disease and low proliferative activity. In addition, immunophenotyping and FISH showed that the +12 is present in only a proportion of clonal B cells (Garcia-Marco et al, 1997). All of this data suggests that trisomy 12 is a secondary event in the development of CLL. For chromosome deletions, specific locus or region-specific probes have been used to demonstrate a high frequency of mono-allelic deletions of the RB1 and p53 genes in B-cell malignancies (Stilgenbauer et al, 1993, 1995; Döhner et al, 1995; Cano et al, 1996). Interphase FISH was also instrumental in identifying the critical region of deletion on 11q13 associated with B-cell lymphoid malignancy, which consequently identified mutations of the ATM gene in T-prolymphocytic leukaemia (PLL) (Stilgenbauer et al, 1997). DNA probes for the fusion genes involved most specific chromosomal translocations and inversions in leukaemia are now commercially available. The differential labelling and detection of these probes in different colours enables a direct visualization of the fusion gene. The simplest scheme is to use two probes (one from each of the fusion genes), differentially labelled and detected with two different-coloured fluorochromes (see Fig 1A). An interphase cell positive for the translocation will exhibit a red–green fusion signal representing the translocation, and a single red and green signal corresponding to the normal chromosome homologues. However, the false positive rate using this approach is quite high (approximately 5%). In addition, the presence of variant translocations or translocations in which the breakpoints are spread over a large distance (e.g. Burkitt's lymphoma), means that the false negative rate can also be quite high. There are several more complex strategies to overcome this (see Figs 1B and 1C). Firstly, if a series of probes spanning both translocation breakpoints are used, this will result in splitting of both fluorescent signals, and the presence of two red–green fusions. Another, more complex, strategy is to employ three or even four different colours, so that the incidence of false positives and false negatives is reduced (Ried et al, 1993; Sinclair et al, 1997). However, the more complicated the colour scheme, the more difficult and complex the analysis. At present, this analysis is done manually, so this is a serious consideration. . Schematic representation of the detection, by FISH, of the Philadelphia translocation in interphase nuclei. In each case the left-hand panel shows the location of the FISH signals on metaphase chromosomes (partial karyotype), and the right-hand panel the interphase FISH signals. In (A) two probes from the flanking regions of the BCR and ABL genes are labelled and detected in different colours: BCR in red and ABL in green. The BCR/ABL fusion results in co-localization of the red and green signals on the der(22) (Philadelphia) chromosome, with a single red and green signal separated, corresponding to the normal chromosomes 22 and 9, respectively. A BCR/ABL negative cell would show two separate red and two green signals. The scheme in (B) uses two probes, this time spanning both the BCR and ABL breakpoint regions. In this case, two red/green fusion signals are formed: one corresponding to the der(9), and the second to the der(22). A positive cell would therefore exhibit one red, one green and two red/green fusions (from Dewald et al, 1998). In (C), a third probe from the region just proximal to ABL on 9q34 is used, labelled in a different colour (represented here in yellow). A translocation positive cell exhibits one green/yellow doublet, one red/green and a single red and yellow signal (from Sinclair et al, 1997). The possibility of using interphase FISH as screening test for specific abnormalities found in acute myeloid leukaemia (AML) subtypes was recently described by Fischer et al (1996). This study used 23 different probes and six to eight hybridizations per patient. They found that interphase FISH was more sensitive for the detection of t(8;21), inv(16), +8q, +11q, +21q, +22q and −Y, and obtained a cytogenetic result in a proportion of cases with no evaluable metaphases. However, this kind of analysis may eventually be replaced by disease-specific DNA chips (see Matrix-CGH below). The detection of residual Philadelphia-positive cells is important after allogeneic bone marrow transplant or interferon (IFN) treatment. In particular, the degree of response to IFN treatment has been shown to be an independent prognostic indicator. The sensitivity of conventional cytogenetics is around 5%, and may be difficult due to low mitotic rate of cells after treatment. RT-PCR is the most sensitive method for detection of BCR-ABL (approximately 10−6) but quantification is difficult. Interphase FISH offers the prospect of using peripheral blood samples, reducing the need for frequent bone marrow aspirates. However, ‘in house’ cut-off levels must be established for each probe set. Conventional FISH probes for the detection of BCR-ABL gene fusion in interphase cells have suffered from a high false positive rate (Tkachuk et al, 1992). The development of three-colour/three-probe FISH protocols for BCR-ABL detection has significantly lowered the false positive rate, and also increased the sensitivity of detection (Sinclair et al, 1997; Dewald et al, 1998). Sinclair et al (1997) used a third probe (for the ASS gene) 200 kb proximal to ABL, such that when a true BCR-ABL fusion was present, there was one co-localization for BCR-ABL, and a separate ASS signal corresponding to the der(9). In cells where the BCR and ABL signals co-localized due to chance, the ASS signal co-localized with the red ABL signal on both chromosomes 9 (see Fig 1C). This three-colour approach resulted in a low false positive and false negative rate. Dewald et al (1998) used a similar strategy, with probes spanning both the BCR and ABL breakpoints. This resulted in two different co-localizations: one representing the der(22) and the other the der(9) chromosomes (see Fig 1B). Strict scoring criteria, experienced operators and scoring of >3000 cells all enabled the detection of residual disease in 0.079% of cells. This skilled and time-consuming approach was also successful in detecting variant translocations. Although the sensitivity of dual-colour interphase FISH is less than for RT-PCR, PCR is not a possibility in a number of cases, for example for the detection of deletions, monosomy or trisomy. Interphase FISH has been used for the detection of residual disease after allogeneic bone marrow transplantation (Anastasi et al, 1991; Wessman et al, 1993). Kasprzyk & Secker-Walker (1997) studied hyperdiploid karyotypes in ALL to detect minimal residual disease. Using three-colour interphase FISH, targeting three chromosomes simultaneously, they were able to achieve a sensitivity of 10−4, and predict relapse in a number of cases. The ability to combine interphase FISH analysis with immunological staining for cell surface antigens provides a powerful method to combine cell by cell analysis with morphology or immunophenotype. Simultaneous immunophenotyping and FISH analysis has been used to investigate lineage involvement in myelodysplastic syndrome (MDS), chronic myeloid leukaemia (CML) and other myeloproliferative syndromes (Price et al, 1992; Nylund et al, 1993; Torlakovic et al, 1994; Soenen et al, 1995; Haferlach et al, 1997, reviewed in Knuutila, 1997). Concurrent immunophenotype and FISH analysis has also been used to demonstrate that the leukaemia which emerged 5 years after sex-mismatched allogeneic bone marrow transplant occurred in donor cells (Katz et al, 1993). In CML, three-colour detection of the Philadelphia translocation and immunophenotype enabled the identification of the translocation in CD20-positive B cells (Torlakovic et al, 1994) and more recently CD3-positive T cells and CD34-positive precursor cells (Haferlach et al, 1997). This supports the belief that CML is a disorder of an early progenitor cell, capable of differentiating into myeloid and some lymphoid lineages (reviewed in Knuutila, 1997). There are also reports of the clonal involvement of B cells in MDS, using del(20q) and monosomy 7 as clonal markers (White et al, 1994; van Lom et al, 1995). In Hodgkin's disease the low percentage of Hodgkin and Reed-Sternberg (HRS) cells means that even interphase FISH may not detect clonal abnormalities. In a recent study the combination of CD30+ staining and FISH with pairs of centromeric probes revealed numerical abnormalities in 100% of HRS cells (Weber-Matthiesen et al, 1995). Surprisingly, clonal abnormalities found in metaphase analysis were not consistent with the interphase FISH analysis, indicating that metaphase analysis of Hodgkin's disease may not be informative. FISH has proved an invaluable aid in the mapping of translocation breakpoints, resulting in the identification of many fusion genes (reviewed in Rabbitts, 1994). A recent addition to the repertoire of FISH techniques now provides significant advantages over other molecular methods for mapping breakpoints which are dispersed over large distances. The term Fibre-FISH is used to describe a collection of methods for performing FISH to extended DNA stretched out on a glass slide (Wiegant et al, 1992; Parra & Windle, 1993; Bensimon et al, 1994; reviewed in Raap, 1998). vandraager et al (1996) have demonstrated the usefulness of this technique for mapping breakpoints of the cyclin D1 gene in mantle cell lymphomas. Using a series of overlapping probes from the 11q13 breakpoint region labelled in alternating red and green fluorochromes creates a colour bar code for the region. Translocations are recognized by the disruption of this bar code into its two complementary parts. The advantages of this method over Southern blotting or pulsed field gel electrophoresis are its simplicity and speed: only a few images need to be examined, and chromosomal breaks over a distance of 250 kb can be visualized. However, the parameters underlying the technique are poorly understood, and at present it remains a research rather than diagnostic tool, confined to a few specialist laboratories. The strength of conventional (G-banded) cytogenetic analysis has always been the ability to survey the entire genome for clues to pathogenesis. However, the poor chromosome morphology and low mitotic index of many leukaemias and lymphomas means that conventional cytogenetic analysis is often limited. In addition, the analysis of banding pattern in highly rearranged karyotypes is difficult and unreliable. One of the remaining challenges for the new FISH techniques is to identify cryptic rearrangements, particularly involving telomeric regions, in apparently normal karyotypes. A significant proportion (15–20%) of bone marrow karyotypes in leukaemia are reported as normal by conventional (G-banded) cytogenetic analysis. Despite significant improvements in the quality of leukaemic metaphase preparations over the past decade, the abnormality rate has not improved. The t(12;21)(p13;q22) remained undetected until 1994, despite the fact that it accounts for 25% of childhood B-cell ALL cases (Romana et al, 1994). This translocation still remains undetectable by conventional cytogenetic analysis. The difficulty in detecting chromosome abnormalities such as this lies in the fact that there is a reciprocal exchange of terminal, pale staining (G-band negative) regions of a similar size. The recent development of multicolour whole chromosome painting provides the promise of identifying cryptic chromosome rearrangements, allowing a full screen of all chromosome abnormalities in a single metaphase. Multicolour FISH using the method of ‘combinatorial’ probe labelling was first described by Nederlof et al (1990). In this approach, probes are labelled with mixtures of fluorochromes such that no two probes have the same combination. The theoretical number of targets which can be discriminated in this manner is 2n−1, where n = number of fluorochromes available. Multicolour FISH with up to seven different colours has been available for a number of years, using probes labelled with three fluorochromes (Dauwerse et al, 1992; Ried et al, 1992). Increasing the number of fluorochromes to five enables the identification of all 24 pairs of human chromosomes. The latter goal has finally been achieved, due in part to the availability of new fluorochromes in the far and infra-red range, and to two ingenious detection methods to discriminate mixtures of fluorochromes (Schröck et al, 1996; Speicher et al, 1996). Both of these used a set of whole chromosome paints, combinatorially labelled with different mixtures of five fluorochromes. The first detection method, multiplex FISH (M-FISH), relied on capturing separate fluorochrome images for each of five fluorochromes, using specifically selected narrow bandpass filter sets (Speicher et al, 1996). The unique labelling combination for each chromosome was computed and displayed in pseudocolours using dedicated software. The second approach, called spectral karyotyping (SKY), used a single exposure of the image, and employed a combination of CCD imaging and Fourier transform spectrometry (Schröck et al, 1996). An interferometer was used to measure the spectrum at each pixel of the image. Both of these techniques have already demonstrated hidden chromosome rearrangements in complex karyotypes in tumour cell lines and in haematological malignancies (Speicher et al, 1996; Veldman et al, 1997; see also Fig 22A). However, the sensitivity of both M-FISH or SKY remains to be established. The limitations of this technology are the reliance on metaphase analysis, and the resolution of painting probes. All of the available whole chromosome paints are deficient in some areas of the genome, particularly the telomeric regions. Our preliminary studies indicate that the sensitivity of multicolour painting for the detection of translocations involving subtelomeric regions may be as low as 2–3 Mb (unpublished observations; see also Fig 2A). In addition, whole chromosome painting will not detect intrachromosomal deletions, duplications or inversions. In practice, both M-FISH and SKY still require reference back to the G-banded karyotype and a combination of FISH approaches is still required to identify all abnormalities in complex karyotypes. . Complementary FISH approaches to the analysis of a complex karyotype in the myeloid leukaemia-derived cell line GF-D8. (A) Colour karyotype after M-FISH analysis. The structural abnormalities identified (arrows) are: t(5;15), der(7)t(7;15), der(8)ins(8;11), der(11)t(8;11), t(11;17). A cryptic der(Y)t(Y;12) was also present, but difficult to identify by M-FISH analysis. (B) A representative metaphase after CGH analysis. Regions of amplification are represented in green and deletions in red. Regions of the genome which are balanced appear white. Amplified regions were identified as: the entire chromosome 7, except 7q22-q33, 8q22.3-qter, 11q21-qter, 13q. Regions of deletion: 5q, 11p, 12p11.2-p13.3, 15q14-q24, 17p and Yq12-qter. A deletion of 7q22-q33 appears white due to the additional copy of an inv(7). The M-FISH analysis identified the origin of a marker chromosome, as well as revealing several cryptic translocations in GF-D8. The value of the CGH analysis was to reveal the unbalanced nature of the karyotype, with large deletions accompanying translocations in most cases. Müller et al (1997) have recently described an innovative approach which makes use of the regions of homology between different species to approach a form of colour bar-coding of chromosomes. The idea of a colour ‘bar-code’ for each chromosome was first described by Lengauer et al (1992), using a series of YACs from the length of the chromosome, labelled differentially and detected in a different colour. Cross-species banding relies on the blocks of homology between species and has been useful in building comparative maps of syntenic regions (reviewed in O'Brien et al, 1997). The Cambridge group, led by Professor Ferguson-Smith and Dr Johannes Wienberg, have now developed a set of regional paints derived from gibbon (genus Hylobates) cell lines. Chromosome-specific painting probes were derived from Hylobates concolor and Hylobates syndactylus by chromosome sorting and DOP-PCR. When used for FISH to human metaphase chromosomes, this resulted in the differentiation of each chromosome into between two and six subregions. Combinatorial labelling using three fluorochromes (generating seven colours) resulted in a unique colour banding pattern for each chromosome. In conjunction with specific software, an automated colour karyotype can be generated. Although at present the number of colour bands is limited, the power of this approach is clear, with the potential to identify chromosomal inversions and insertions. Indeed, colour banding has already been used to identify cryptic translocations in CML (Dr Christine Harrison, personal communication). A set of chromosome-specific subtelomeric probes which uniquely identify the ends of all human chromosomes (except the p arms of the acrocentrics) is now available for FISH (National Institutes of Health and Institute of Molecular Medicine collaboration, 1996). These contain subtelomeric DNA sequences cloned in cosmid, P1 and PAC clones, the majority of which have been localized to between 100 and 300 kb from human chromosome ends. These probes have been validated in a multiprobe FISH assay for subtelomeric rearrangements on a series of individuals with cryptic unbalanced constitutional chromosome rearrangements (Knight et al, 1997). This is essentially 24 dual-colour hybridizations, enabling an assessment of all chromosome subtelomeric regions on a single microscope slide. However, the multiprobe approach requires a high mitotic index and is most suitable for the analysis of constitutional karyotypes, which rely on PHA-stimulated peripheral blood lymphocytes, or where a cell line is available. In our experience, the value of these probes for leukaemic karyotypes has been to identify the specific subtelomeric region in rearrangements found initially by multicolour painting (see Fig 3). An alternative approach, a multicolour FISH assay, would enable the assessment of all chromosome subtelomeric regions in a single metaphase. However, the extension of combinatorial labelling and multiple colour detection methods for cosmid, or even P1 and PAC clones represents a technically challenging series of developments. Firstly, the simultaneous analysis of all chromosome ends in a different colour requires 41 colours; the maximum number of targets achievable with five fluorochromes is 31. Secondly, the interpretation of such an assay would rely on specifically designed computer software. Finally, it is not certain whether the targets of such small probes combinatorially labelled with several different fluorochromes can be discriminated, due to limits of spatial resolution of the microscope. It is hoped that the development of additional spectrally separable fluorochromes will ease this type of analysis considerably. . The use of chromosome-specific subtelomeric probes to identify the origin of additional chromosome material on two add(7q) chromosomes. In each case dual-colour hybridization was carried out with the p arm probe labelled in biotin and detected in Texas red (red fluorescent signal), and the q arm probe labelled with digoxigenin and detected with fluorescein (green fluorescent signal). In (A) subtelomeric probes for 8p and 8q identified the additional material on the add(7q) as derived from 8q (arrow). In (B) the additional material on 7q was identified as deriving from 4q. Although these abnormalities were detected by M-FISH, painting gives no information of the chromosomal region involved. It is also important to note that the abnormality in (A) was originally described by G-banding as del(7q). The limitation of all of the new multicolour karyotyping approaches is that they still require metaphase chromosomes. The advantages of CGH are that it bypasses the need for dividing cells and does not require any prior knowledge of the chromosome constitution. The concept is simple: genomic tumour and reference DNA are labelled with different coloured fluorochromes, mixed in equal amounts and hybridized onto normal metaphase chromosomes. The differences in copy number between the normal (reference) and tumour is reflected by differences in red and green fluorescence along the length of the chromosome (see Fig 2B). Over the 5 years since its introduction (Kallionemi et al, 1992) CGH has produced a steady stream of publications, identifying new regions of amplification and deletion in a wide variety of tumour types (reviewed in Forozan et al, 1997). The use of CGH for analysing haematological malignancies is more limited (Bentz et al, 1995a, b; Joos et al, 1996; Monni et al, 1996; El-Rifai et al, 1997). The disadvantages of CGH for haematological malignancies are the inability to detect balanced rearrangements, and the requirement for >50% cells with the clonal abnormality. However, CLL and some lymphomas have clearly benefited from the application of CGH (Bentz et al, 1995a; Joos et al, 1996; Monni et al, 1996). One study of B-CLL identified gains and losses not identified or not detected by G-banding. Importantly, clonal aberrations were identified in six out of 13 cases with a normal karyotype (Bentz et al, 1995a). The main reasons for discrepant results between G-banding and CGH were a complex karyotype, the wrong clone analysed, or a lack of metaphases. This study indicates that banding analysis may miss relevant karyotypic abnormalities and may explain why clinically important chromosome changes have not been identified in CLL. In contrast, a study of 10 myeloid leukaemias found a good correlation between G-banding and CGH results (Bentz et al, 1995b). The only discrepancies were a failure of CGH to detect minor clones (of 8% and 35% respectively). The major limitation of CGH is its resolution, due to the reliance on metaphase chromosomes. For deletions, the resolution of CGH has been estimated at > 10 Mb (Bentz et al, 1998). Perhaps the most promising future for CGH technology lies in CGH to cloned DNA arrays, so-called matrix-CGH (see below). This technology promises to overcome the limitation of using metaphase chromosomes as a target for CGH by replacing condensed metaphase chromosomes with cloned DNA arrayed in small spots and immobilized to the surface of a glass slide. Solinas-Toldo et al (1997) successfully used matrix CGH for the detection of high copy number c-MYC amplification using cosmids as target. For low copy number amplifications, larger cloned probes (P1 or PACs) were necessary. For deletions, matrix CGH achieved a resolution of 75–130 kb, although it could not distinguish between mono- and bi-allelic deletions. The other limitations of metaphase CGH also apply: matrix-CGH requires at least 50% of clonal cells and will not detect balanced translocations. One could envisage various types of matrix-CGH containing: (i) disease-specific probe sets, (ii) oncogenes or tumour suppressor genes, (iii) DNA clone contigs for specific regions, or, eventually, (iv) clones spaced at 1 Mb over the whole genome. PNA probes are DNA analogues in which the deoxyribose backbone has been replaced by 2-aminoethyl glycine. The rapid kinetics and stable formation of duplexes with complementary DNA sequences make them attractive for a number of applications, including FISH. Oligonucleotide PNA FISH probes have been developed for the human telomeric repeat sequences, enabling the fluorescent detection of all telomeres in a single metaphase. These signals appear to be a better substrate for quantitative fluorescence measurements than conventional FISH signals, allowing an assessment of telomere length (Lansdorp et al, 1996). This technology may also be extended to other repeat sequences such as centromeric alphoid repeats. It may also be feasible to combine PNA telomeric probes with the chromosome-specific DNA probes to provide a correlation of telomere shortening and specific chromosome type. This stunning new technology holds the ultimate promise of detecting single base changes in individual cells. Padlock probes consist of two different oligonucleotides, each approximately 20 nucleotides long, separated by a 30-40mer spacer sequence. When the probe sequence exactly matches the target, the 3′ and 5′ ends of the probe are juxtaposed, and the probe is immobilized. Nilsson et al (1997) used two different oligonucleotide probes, each labelled with a different fluorochrome, to detect single base pair differences in a centromeric alphoid repeat. The sensitivity of this technology may be improved by the incorporation of new sensitive labelling techniques such as the use of fluorescent tyramides (tyramide signal amplification) (Kersten et al, 1995; Raap et al, 1995). Another option to increase the sensitivity is amplification of the circular DNA, so-called rolling circle amplification. This extension to the technology would potentially enable fluorescent detection of mutations in individual nuclei. Rolling circle amplification has been achieved with some success using extended DNA from halo preparations, but at this stage does not work well or reproducibly on nuclei (Lizardi et al, 1998). In the relatively short time since its inception, FISH has had a major impact on cytogenetic analysis, due to the speed, sensitivity and flexibility of its use. Although some of the applications will remain research tools, the technology and probes for most diagnostically relevant cytogenetic abnormalities are now well within the reach of most clinical laboratories. However, conventional FISH can only provide answers to the specific questions posed and requires some preconceived knowledge of the karyotype. The recent updating of multicolour FISH to enable the visualization of the entire human genome in 24 different colours has caught the imagination of scientists and clinicians alike. The power of this approach is the ability to interrogate the whole genome in a single hybridization experiment, combining the screening potential of cytogenetics with the accuracy of molecular genetics. The belief that cytogenetics is more an art than a science has never been further from the truth. With the aid of new colour karyotyping techniques (SKY and M-FISH), cytogenetic analysis now approaches a molecular definition of karyotype. The major impact of this development in field of haematological oncology is likely to be the identification of new and non-random chromosome rearrangements and clinical correlations. Many of the most recent innovations lend themselves well to automation. The most basic of these, fluorescent metaphase finding, is now being addressed by microscope and imaging companies, and will not only relieve the boredom of such tasks, but should increase the sensitivity of interphase FISH analysis. With many of the tedious aspects of cytogenetic analysis removed by automated approaches, the future for cytogenetics has never looked brighter. The author thanks all of the members of the Cytogenetics Group, MRC Molecular Haematology Unit, particularly Sabrina Tosi, for the CGH and multicolour analysis of the GF-D8 cell line. Much of the work described here was supported by the Medical Research Council and the Leukaemia Research Fund, U.K.