Main

The hallmark of chronic myelogenous leukemia is the t(9;22)(q34;q11). In this translocation, the 3′ segment of the c-abl proto-oncogene on chromosome 9 is juxtaposed with the 5′ segment of the bcr gene on chromosome 22.1, 2, 3, 4, 5 Breaks in the c-abl gene typically occur in the first intron. Breaks in bcr generally occur in one of three regions: the major breakpoint cluster region (M-bcr), the minor breakpoint cluster region (m-bcr) and the micro breakpoint cluster region (μ-bcr).6 Breakpoints occurring in M-bcr involve introns 13 or 14 and join exon 13 (also known as b2) or 14 (also known as b3) with exon 2 of abl (a2) resulting in the fusion transcripts b2a2 and b3a2, respectively. These transcripts lead to the production of an 8.5 kb transcript coding for a 210 kDa (p210) chimeric protein.7, 8 Breakpoints in m-bcr involve the first intron of bcr and join exon 1 (e1) with a2 resulting in a smaller fusion transcript, e1a2, that codes for a 190 kDa (p190) protein.9 Breakpoints in μ-bcr involve intron 19 and result in the joining of exon 19 (e19) of bcr with a2, e19a2, coding for a 230 kDa (p230) protein.10 Although all bcr-abl fusion proteins display activated tyrosine kinase activity, the p190 form has been shown to have more transforming potential than p210 in vitro and in vivo.11, 12, 13 Fusion transcripts b3a2 and b2a2 account for the majority of CML cases. The e1a2 fusion transcript is seen primarily in t(9;22)-positive acute lymphoblastic leukemia and lymphoid blast phase of chronic myelogenous leukemia, but rarely in chronic myelogenous leukemia in chronic phase. The rare e19a2 fusion transcript is found in cases of chronic myelogenous leukemia with prominent neutrophilic maturation.14 These unusual neoplasms also have been reported in the literature by some investigators as chronic neutrophilic leukemia.15

The introduction of fluorescence-based PCR methods has greatly simplified detection and quantification of bcr-abl fusion transcripts. Real-time reverse transcriptase-polymerase chain reaction (RT-PCR) methods enable t(9;22) detection and quantification without the need to open amplification tubes and are target specific, highly sensitive and reproducible. Hence, the real-time RT-PCR approach has become the preferred method for monitoring tumor burden and the effect of therapy in patients with t(9;22)-positive leukemias.16, 17, 18, 19, 20, 21, 22, 23, 24 Two strategies are generally followed to quantify the common of b2a2, b3a2 and e1a2 fusion transcripts. In one approach, a multiplex qualitative PCR is performed initially. Following gel electrophoresis and identification of the specific fusion transcript, real-time RT-PCR is performed using fusion sequence specific primers to measure bcr-abl transcript levels. In the second approach, a multiplex real-time RT-PCR assay that detects the three fusion transcripts is performed directly, without prior analysis or determination of the fusion transcript type. The first approach involves two independent PCR assays, is time consuming and is prone to manual errors and contamination. Using the second approach, one cannot determine the specific fusion transcript present in the sample unless real-time RT-PCR products are subjected to gel electrophoresis and ethidium bromide-based detection, a comparatively less-sensitive technique that is not very useful for monitoring minimal residual disease. A third strategy can also be pursued, that being two (or multiple) separate real time RT-PCR assays for each transcript type separately. Although this approach allows for easy identification of the specific transcripts, it basically multiplies the workload of the second approach.

In a previous study,25 we showed that PCR products labeled during real-time PCR by incorporation of a fluorescent dye-labeled primer can be resolved by capillary electrophoresis and Gene-Scan analysis in follicular lymphoma cases with the t(14;18). We thought this approach could be applied to analysis of bcr-abl transcripts in patients with t(9;22)-positive leukemias. In this assay, we labeled the 5′ end of the abl primer with the fluorescent dye NED (Applied Biosystems), which is included along with e1 and b2 bcr primers during the TaqMan RT-PCR assay and thus is incorporated into the bcr-abl fusion product. This allows specific identification of the fusion transcript type by capillary electrophoresis equipped with highly sensitive laser-based detection of the fluorescently labeled fusion products. This approach, which has a sensitivity of detection equivalent to other real-time RT-PCR assays, requires no further manipulation to confirm or identify the specific fusion transcript in patient specimens.

Materials and methods

Patient Specimens and Controls

A total of 67 bone marrow (BM) aspirate specimens from patients with bcr-abl-positive leukemias were included in this study. These specimens were divided into two groups. One group included 26 BM aspirates from patients selected randomly at various stages in their clinical course. Almost all of these patients were being treated for their disease with chemotherapy. The second group included 41 BM aspirates from nine patients assessed at time of diagnosis, prior to therapy, and then followed over time to assess response to therapy. The diagnosis of chronic myelogenous leukemia and acute lymphoblastic leukemia was based on standard clinical and pathologic criteria as stated in the World Health Organization classification14 and used at The University of Texas MD Anderson Cancer Center. Only those patient samples that had abl transcript values greater than 20 000 copies by quantitative RT-PCR were included in the study. The bcr-abl-positive cell lines KBM7, K562 and B15 that carry b2a2, b3a2 and e1a2 fusion genes, respectively, served as positive controls. The HL60 cell line was used as a negative control. Serial dilutions of a plasmid containing bcr-abl fusion sequences26 were analyzed simultaneously with patient samples to generate standard curves for bcr-abl and abl (normalizer). The standard curves for bcr-abl were similar whether the plasmid was diluted into water or into the HL60 cell line. Hence, plasmid diluted into water was used for construction of standard curves in order to calculate the quantities of bcr-abl and abl transcripts in a test sample.

For RNA isolation, total leukocytes were prepared by erythrocyte lysis of BM aspirate (1–2 ml) specimens collected in ethylenediaminetetraacetic acid (EDTA). Total RNA was isolated using Trizol reagent (Invitrogen Lifetechnologies Gaithersburg, MD, USA) according to the manufacturer's instructions. The integrity of RNA was determined by gel electrophoresis prior to reverse transcription. Total RNA ranging from 3 to 30 μg with intact 28 s and 18 s RNA was converted to cDNA using random hexamers and Superscript II reverse transcriptase (Invitrogen Lifetechnologies) according to the recommendations of the manufacturer. For specimens with less than 7 μg total RNA, the final reaction volume was adjusted to 15 μl, and for specimens with greater than 7 μg of total RNA the cDNA conversion was performed in 60 μl.

Real-time RT-PCR Assay for bcr-abl

Real-time RT-PCR assays were performed using the ABI PRISM 7700 Sequence Detection System (Applied Biosystems, Foster City, CA, USA). A volume of 2.5 μl of cDNA from each sample was subjected to amplification in duplicate for bcr-abl in a multiplex RT-PCR using an abl primer in combination with bcr b2- or e1-specific primers, and an abl probe labeled with 6-carboxyfluorescein (FAM) (Table 1). For size analysis of fusion transcripts by capillary electrophoresis, the abl primer was labeled with NED fluorescent dye at its 5′ end (ABL-NED). As a control, abl transcripts were amplified simultaneously in duplicate, but in a separate reaction, to confirm the quality of extracted RNA and to normalize bcr-abl values.

Table 1 Primers and probes used in the modified quantitative t(9;22) (q34;q11) RT-PCR
Full size table

Both bcr-abl and abl PCR assays were performed in a final volume of 25 μl using 2.5 μl of cDNA and a universal master mix without uracil N-glycosylase (Applied Biosystems). bcr-abl amplification consisted of previously described primers BCR-2F, BCR-cF and ABL-N27 at 400 nM each and 200 nM of FAM-labeled TaqMan bcr-abl probe. Amplification of abl, the normalizer, consisted 100 nM of each primer (ABL2-F and ABL2-R) and 200 nM VIC labeled TaqMan abl probe. Samples were subjected to 40 cycles of PCR, each cycle consisting of 95°C denaturation for 30 s, 57°C annealing for 20 s and 72°C extension for 45 s. The last cycle was followed by a 10-min elongation step at 72°C.

Data Analysis

The fluorescence emission data for each sample were analyzed immediately after PCR using Sequence Detection Software (SDS version 1.7, Applied Biosystems). The threshold cycle (Ct) values representing the PCR cycle number at which fluorescence signal is increased above an arbitrary threshold were exported into Microsoft Excel software for further analysis. The bcr-abl levels for each sample were expressed as a ratio of bcr-abl to abl × 100.

GeneScan Analysis

Following real-time RT-PCR, each amplification product was diluted 10-fold with water and 0.5 μl the diluted products were then subjected to capillary electrophoresis in an ABI Prism 3100 Genetic Analyzer (Applied Biosystems) with 16 capillaries. The 3100 Genetic Analyzer is a laser-based fluorescence detection system that automatically introduces the samples labeled with fluorescent dyes into a polymer-filled capillary for electrophoresis. We used the GenoTYPE-ROX 50-500 DNA ladder (Life Technologies, Gaithersburg, MD, USA) as internal size standards. The size of each amplified fragment is calculated with GeneScan software using the Local Southern sizing option (Applied Biosystems).

Results

Sensitivity of the Multiplex bcr-abl Real-time RT-PCR Assay

We assessed the effect of introducing abl primer labeled with the fluorescent dye NED on the sensitivity of multiplex TaqMan real-time RT-PCR bcr-abl assay by performing RT-PCR with the abl-NED primer and comparing the results with those obtained using an unlabeled abl primer. Serially diluted plasmid clone containing bcr-abl fusion sequences showed similar Ct values with and without NED labeled abl primer at all dilutions including 10 copies/assay, demonstrating that introduction of this fluorescent dye did not alter the efficiency or sensitivity of the bcr-abl assay (Table 2).

Table 2 Correlation between traditional and modified real-time PCR assays for bcr-abl
Full size table

Interassay Variability

The interassay variability of the real-time quantitative RT-PCR for bcr-abl and abl was evaluated in 11 independent assays performed on different days over a period of 2 months. Serially diluted plasmid clone was used to generate standard curves for abl and bcr-abl. A constant amount of cDNA derived from the positive cell lines B15, KBM7 and K562 was also simultaneously amplified for bcr-abl and abl with each assay. To minimize variation, the same stock cDNA was used in all experiments. The mean, standard deviation (s.d.) and interassay coefficients of variation at each dilution of the plasmid and positive cell lines for bcr-abl and abl are shown in Table 3. S.d. of less than 1 Ct was observed with input plasmid copy number of 100 000, 10 000, 1000, 100 and 10. The standard curves plotted using mean threshold cycle (Ct) and mean copy number from 11 experiments for bcr-abl and abl indicate that the modified quantitative bcr-abl RT-PCR assay is linear over 5 logs and that amplification efficiencies for the two targets are similar (Figure 1a and b). The coefficient of variation in Ct values of positive control cell lines tested on 11 different days using a constant amount of cDNA each time varied from 1.64 to 3.16% for bcr-abl and 2.9 to 4.7% for abl (Table 3). These data indicate that the assay is reproducible and that 3–5-fold variation from experiment to experiment can be expected. Interestingly, the e1a2-positive B15 cell line consistently expressed a much lower level of bcr-abl fusion transcripts compared with abl, with an average bcr-abl to abl ratio of 2341/39127(0.06). The b2a2 and b3a2 positive cell lines, KBM7 and K562, expressed either similar or slightly higher levels of bcr-abl fusion transcripts compared with abl.

Table 3 Interassay variability of the modified real-time t(9;22)(q34;q11) RT-PCRa and the real-time abl PCR
Full size table
Figure 1
figure 1

The standard curves plotted using mean threshold cycle (Ct) and mean copy number from 11 experiments for bcr-abl (a) and abl (b) indicate that the amplification efficiencies for the two targets are similar. (a) Also demonstrates that the modified quantitative t(9;22) RT-PCR assay including NED-labeled abl primer is linear over 5 logs.

Full size image

Capillary Electrophoresis of Amplification Products

The amplification products were resolved by capillary electrophoresis and analyzed by GeneScan after real-time RT-PCR. Amplification of the positive control cell lines B15, KBM7 and K562 resulted in products of 264, 132 and 208 bp, respectively for the e1a2, b2a2 and b3a2 fusion transcripts using the primers described in Table 1. These results confirmed that fluorescent signal detected by real-time RT-PCR was due to the presence of fusion sequences (Figure 2). Amplification of the HL60 negative control showed no peak confirming the specificity of the assay. Amplification of serially diluted plasmid standards and analysis of PCR products by capillary electrophoresis and GeneScan revealed that fusion transcripts could be identified even at 10 copies/assay (data not shown).

Figure 2
figure 2

Electropherogram generated by GeneScan software shows the molecular size and fluorescence intensity of NED labeled amplification products generated by real-time PCR of t(9;22)-positive cell lines. Negative control, HL60 shows no peak (bottom panel), indicating the specificity of the assay.

Full size image

In order to determine if all three fusion sequences, when present in a sample at different levels, amplify uniformly and proportionately, we mixed cDNA synthesized using a constant amount of RNA from each of the positive cell line controls and performed RT-PCR as described in the methods. cDNA from the KBM7, K562 and B15 cell lines was mixed in the following ratios: 1:1:1; 1:0.1:1; 1:1:0.1 and 0.1:1:1. Thus, in the latter three mixing experiments, one transcript type was present at 10% of the amount of the other transcript types. The chromatograms generated by GeneScan analysis and shown in Figure 3 demonstrate that the three fusion sequences were amplified proportionately and that the RT-PCR assay allows detection of a fusion transcript present at lower levels compared with other fusion transcript types in the same sample.

Figure 3
figure 3

Cell line mixing experiments demonstrate that the multiplex real-time RT-PCR for bcr-abl allows proportionate amplification of the three transcripts when present at different levels in a sample. cDNA from KBM7, K562 and B15 was mixed in 1:1:1 ratio (a), 1:0.1:1 ratio (b), 1:1:0.1 ratio (c) and 0.1:1:1 ratio (d).

Full size image

Analysis of 26 randomly selected clinical specimens (Table 4) in patients with t(9;22)-positive chronic myelogeneous leukemia (n=24) and acute lymphoblastic leukemia (n=2) showed that the RT-PCR assay can detect a wide range of bcr-abl transcript copy numbers. Two untreated patients with chronic myelogenous leukemia had high copy numbers of b3a2 (case 23) and both b2a2 and b3a2 (case 24) transcripts, respectively. All other chronic myelogenous leukemia patients were being treated for variable intervals of time and had much lower bcr-abl transcript levels, or were completely negative. In positive samples, the lowest tumor burden detected was 0.0003 transcript copies (0.04%). Samples that were negative for bcr-abl by the RT-PCR assay were also negative by capillary electrophoresis and GeneScan analysis. All six patients with negative samples had no morphologic evidence of chronic myelogenous leukemia or cytogenetic evidence of the t(9;22) in the BM aspirate sample assessed by the RT-PCR assay. Two acute lymphoblastic leukemia patients (cases 25 and 26), both on therapy, had low levels of e1a2 transcripts. Thus, this approach is sensitive and reliable for identifying the specific fusion transcript in post-therapy samples and thereby allows monitoring of minimal residual disease.

Table 4 Type of transcript and the number of abl and bcr-abl transcripts present in randomly selected follow-up specimens from patients diagnosed with chronic myelogenous leukemia1–24 and acute lymphoblastic leukemia25,26
Full size table

We then selected nine patients with t(9;22)-positive chronic myelogenous leukemia (n=8) or acute lymphoblastic leukemia (n=1) (Table 5) and assessed bcr-abl transcript type and bcr-abl copy number at time of diagnosis, before onset of therapy at our institution, and then assessed multiple follow-up samples obtained at 3, 6, 9 and 12 months on therapy. Eight patients (cases 1–8) had much higher transcript copy numbers at time of diagnosis than after treatment, with a dramatic drop off in levels within the first 3 months of therapy. In one patient (case 9), the patient had progressive disease with increasing levels of both b2a2 and b3a2 transcripts.

Table 5 Results of the modified bcr-abl real-time RT-PCR of specimens derived at initial diagnosis and at various intervals following therapy from eight patients with chronic myelogenous leukemia and a patient with acute lymphoblastic leukemia (# 8)
Full size table

Discussion

In an earlier study,25 we showed that PCR products labeled during real-time PCR by incorporation of a fluorescent dye-labeled primer can be resolved by capillary electrophoresis and GeneScan analysis in follicular lymphoma cases with the t(14;18). We thought that a similar approach to assess bcr-abl in patients with chronic myelogenous leukemia or acute lymphoblastic leukemia, either at time of diagnosis or for monitoring minimal residual disease, would be faster and more convenient compared with other approaches published in the literature.16, 17, 18, 19, 20, 21, 22, 23 For example, traditional quantitative real-time TaqMan RT-PCR assays used to assess for bcr-abl generally require prior knowledge of the type of fusion transcript present in a given specimen. Thus, a qualitative gel-based PCR is initially needed for this purpose. In contrast, most multiplex TaqMan bcr-abl RT-PCR assays reported previously do not require knowledge of the specific fusion transcript a priori, but one cannot distinguish which specific transcript is amplified. A third approach, that being assessment of each possible bcr-abl transcript by RT-PCR using multiple sets of specific primers, avoids the problems of traditional or previously published multiplex RT-PCR approaches, but is cumbersome.

In the assay we describe, labeling the abl primer at its 5′ end with NED, a fluorescent dye that does not interfere with TaqMan reporter dyes or assay sensitivity, allows easy identification of the specific fusion transcript after real-time PCR by automated high throughput capillary electrophoresis and GeneScan analysis. The latter also corroborates the quantitative PCR results and thus serves as another quality control for the real-time RT-PCR assay. For leukemia patients in whom two transcripts are detected, GeneScan analysis also provides a means for assessing the relative contribution of each transcript to the tumor levels. As the sensitivity of detection of fluorescently labeled amplification products by laser-based capillary electrophoresis is substantially superior to that of ethidium bromide-based detection, coupling real-time RT-PCR with capillary electrophoresis and GeneScan analysis allows easy identification of fusion transcript type even at very low tumor burden.

We have shown the utility of this assay in two separate patient groups. Initially, we randomly selected 26 samples of patients known to have bcr-abl-positive leukemias, 24 patients with chronic myelogenous leukemia and two with precursor B-cell acute lymphoblastic leukemia. As most of these patients were being treated and were at various stages of their clinical course, there was a wide spectrum of bcr-abl copy number, ranging from 16 162 to 0 copies detected. All six patients in whom no bcr-abl copies were detected had no morphologic or cytogenetic evidence of disease. We followed up on these results by assessing nine patients sequentially, at time of diagnosis, and then at regular intervals after onset of therapy at our institution. As expected, the RT-PCR assay detected relatively high levels at time of diagnosis and much lower levels subsequently, with the greatest drop in bcr-abl copy number occurring in the first 3 months after onset of therapy. These results illustrate that this modified RT-PCR method to assess bcr-abl is useful for de novo diagnosis and for monitoring response to therapy and minimal residual disease in patients with bcr-abl-positive leukemias.

Although relatively few studies are currently available regarding the significance of bcr-abl transcript type, some preliminary reports suggest that knowledge of transcript type may have clinical meaning or help us to further understand the pathogenesis of t(9;22)-positive leukemias. For example, Perego et al28 reported that chronic myelogenous leukemia patients with b3a2 transcripts had higher platelet counts than those with b2a2 transcripts. In another study of chronic phase chronic myelogenous leukemia patients, Prejzner29 suggested that patients with b3a2 transcripts had longer survival than those with b2a2 transcripts. Dual transcription of b2a2 and b3a2 transcripts may be linked to a polymorphism within the bcr locus.30 Several investigators have recently reported detection of low levels of e1a2 transcripts, in addition to b2a2 or b3a2 transcripts, in most patients with chronic phase chronic myelogenous leukemia.31, 32, 33 These e1a2 transcripts, thought to be derived from alternative splicing of b2a2 or b3a2 fusion RNA, may be clinically significant. In patients with acute lymphoblastic leukemia, Radich et al34 have shown that detection of bcr-abl transcripts after allogeneic stem cell transplantation is highly predictive of eventual relapse in acute lymphoblastic leukemia patients with e1a2/p190, as compared with patients with b3a2/p210. Thus, identification of the type of fusion transcript, in addition to its utility for ruling out false positives and cross contamination, may prove valuable for assessing the clinical relevance of various transcripts in patients with the chronic myelogenous leukemia and t(9;22)-positive acute lymphoblastic leukemia. Furthermore, identification of the type of fusion transcript in a leukemia patient prior to BM or stem cell transplant is essential for assessing minimal residual disease after transplantation.