Molecular evolution of acute myeloid leukaemia in relapse: unstable N-ras and FLT3 genes compared with p53 gene


Dr Tomoki Naoe, Department of Infectious Diseases, Nagoya University School of Medicine, Nagoya 466-8560, Japan. e-mail:


Relapse is a major cause of treatment failure in acute myeloid leukaemia (AML), and is usually accompanied by resistance to chemotherapy. To study whether relapse is accompanied by genetic alterations, we compared N-ras, p53 and FLT3 gene mutations in paired samples obtained at initial diagnosis and first relapse. 28 patients with relapsed AML were studied, and their duration of complete remission ranged from 133 to 989 d (mean 318 d). Karyotype changes were observed at relapse in 11 patients. Point mutations of the N-ras gene were positive at both stages (+/+) in three patients, positive at initial diagnosis and negative at relapse (+/−) in three patients, and negative at initial diagnosis and positive at relapse (−/+) in two patients. Internal tandem duplications of the FLT3 gene (FLT3/ITD) were +/+ in five patients, +/− in one patient, and −/+ in six patients. The p53 gene mutations were +/+ in two patients, +/− in one patient, and −/− in 25 patients. FLT3/ITD and mutant p53 at relapse were associated with short survival after relapse. These results indicate that relapse is frequently accompanied by molecular alterations that include the loss and/or acquisition of mutations. Thus relapse can be understood as clonal shift or collateral succession rather than clonal progression.

Progress in chemotherapy has brought about a high complete remission (CR) rate and an increase in the number of long-term survivors of acute myeloid leukaemia (AML) (Burnett & Eden, 1997; Bishop, 1997). However, nearly half of the patients who obtain CR eventually relapse, which remains a major cause of treatment failure. Relapsed leukaemia is generally resistant to chemotherapy, and even if a second CR is achieved, the period of the remission is short (Ohno et al, 1994). Thus relapse does not simply mean reappearance of leukaemia at initial diagnosis. Expression of MDR-1 or an accelerated cell-cycle has been suspected as an underlying mechanism for the resistance at relapse (Ivy et al, 1996; Scuderi et al, 1996). Over half of relapsed AML patients had karyotype changes (Garson et al, 1989; Estey et al, 1995), frequently to a more complex karyotype, often to a less-complicated or normal karyotype, and less frequently to a clonally unrelated karyotype. Numerical abnormalities involving +8 and/or +21 and structural abnormalities such as 5q, 7q, 9q and 12p were relatively common at relapse, although it is difficult to find a general rule in the karyotype development. These findings suggest that relapse is accompanied by clonal evolution. However, there have been few studies on the molecular alterations associated with relapse of AML.

As to the molecular mechanism of leukaemogenesis, non-random chromosomal translocations, which usually target and deregulate the genes coding transcriptional factors, are thought to mediate the initiation rather than progression of leukaemia (Look, 1997; Caligiuri et al, 1997). On the other hand, gain-of-function mutations of the signal-transducing molecules, such as RAS and FLT3, are frequently observed during progression of myelodysplastic syndrome (MDS) (Hirai et al, 1987; Horiike et al, 1997). Loss-of-function mutations of cell-cycle associated molecules such as p53 are also found as a late event in chronic myeloid leukaemia (CML) and MDS (Ahuja et al, 1991; Jonveaux et al, 1991). However, the association of these genes with relapse remains unclear.

In this study we compared the presence of the above gene alterations in paired AML samples obtained at initial diagnosis and at progression/relapse.


Patients and leukaemia cells

28 patients from whom leukaemia cells at initial diagnosis and at relapse were preserved were analysed. The patients with AML were diagnosed according to the French–American–British (FAB) classification from 1984 to 1994, and were treated by the AML87 protocol (Ohno et al, 1993) from 1987 to 1989, the AML89 protocol (Kobayashi et al, 1996) from 1989 to 1992, or the AML92 protocol from 1992 to 1994 (unpublished observations). Each protocol fundamentally consisted of N4-behenoyl-1-beta-d-arabinofuranosyl-cytosine (BHAC), daunorubicin, 6-mercaptopurine, and prednisolone. After achieving CR each induction therapy was followed by three courses of consolidation chemotherapy and four to 12 courses of intensification chemotherapy. After relapse, patients received either the same therapy as the induction chemotherapy or a combination therapy including mitoxantrone, VP-16 and Ara-C.

Bone marrow samples were taken from the leukaemia patients at the time of initial diagnosis and relapse after obtaining informed consent. Leukaemia cells were isolated by Ficoll-Hypaque gradient centrifugation, suspended in RPMI 1640 medium with 10% dimethylsulphoxide and 10% fetal calf serum, then cryopreserved in liquid nitrogen until use. All samples contained > 80% viable leukaemia cells.

CR was determined when there were <5% blasts in normo-cellular bone marrow with normal levels of peripheral neutrophil and platelet counts. CR duration was measured from the date of CR to relapse.

Analysis of the internal tandem duplication of the FLT3 gene

High-molecular weight DNA was extracted from AML cells. Genomic PCR amplification was performed using the primers 11F, 5′-GCAATTTAGGTATGAAAGCCAGC, and 12R, 5′-CTTTCAGCATTTTGACGGCAACC-3′. The PCR protocol was described previously (Kiyoi et al, 1997). To confirm the sequence of FLT3 cDNA, reverse transcriptase-PCR was performed using the primers R5, 5′-TGTCGAGCAGTACTCTAAACATG-3′ and 12R. Abnormal-sized transcripts were cloned and sequenced as previously described (Kiyoi et al, 1997).

N-ras gene amplification and dot blot hybridization

To amplify the sequences spanning codons 12 and 13 and codon 61, the following oligonucleotide primers were prepared: NA12: 5′-GACTGAGTACAAACTGGTGG-3′, NB12: 5′-CTCTATGGTGGGATCATATT-3′, NA61: 5′-GGTGAAACCTGTTTGTTGGA-3′, and NB61: 5′-ATACACAGAGGAAGCCTTCG-3′. PCR and oligonucleotide probe hybridization were performed as described previously (Kubo et al, 1993).

Single-strand conformation polymorphism (SSCP) and sequence analysis of p53 gene

The PCR technique was employed to amplify the genomic DNAs corresponding to exons 5–8 of the p53 gene. Four pairs of sense and anti-sense primers were used: 5-5′:5′-TGTTCACTTGTGCCCTGACT-3′, 5-3′: 5′-CAGCCCTGTCGTCTCTCCAG-3′, 6-5′: 5′-GCCTCTGATTCCTCACTGAT-3′, 6-3′: 5′-TTAACCCCTCCTCCCAGAGA-3′, 7-5′: 5′-ACTGGCCTCATCTTGGGCCT-3′, 7-3′: 5′-TGTGCAGGGTGGCAAGTGGC-3′, 8-5′: 5′-TAAATGGGACAGGTAGGACC-3′, 8-3′, 5′-TCCACCGCTTCTTGTCCTGC-3′. The PCR mixture (10 ml) for SSCP analysis contained 100 ng of genomic DNA, 5 pmol of primer pairs, 5 mm of each dNTP, 18.5 MBq of α[32P]d-CTP and 0.25 units of Taq polymerase (Boehringer Mannheim) in buffer as recommended by the manufacturer. The amplification was carried out in a thermocycler (Perkin-Elmer/Cetus) with an initial denaturation step (5 min, 94°C), followed by 30 cycles consisting of three steps: 94°C for 30 s, 55–62°C for 1 min and 72°C for 1 min. An additional cycle was performed at 72°C for 7 min. The SSCP technique was described by Orita et al (1989). Amplification products were diluted 10-fold in loading dye solution and heat-denatured in boiling water for 5 min, chilled on ice and loaded on non-denaturing polyacrylamide gel. Electrophoresis was performed at 25°C for 3 h. Gels were exposed to X-ray films at −70°C. If mobility shift was observed the corresponding region of the p53 gene was amplified again. The amplified products were cut out from the gel, purified with a Qiaex gel extraction kit (Qiagen Inc., Chatsworth, Calif., U.S.A.), and cloned into the pMOSBlue T-vector (Amersham, Buckinghamshire, U.K.). 10 recombinant colonies were chosen and cultured in LB medium. Plasmid DNA was prepared using a QIAprep spin plasmid miniprep kit (Qiagen Inc.), and both strands were sequenced using fluorescein-conjugated 21M13 and T7 primers on a DNA sequencer (377; Applied Biosystems, Foster City, Calif., U.S.A.).

Statistical methods

Survival probabilities were estimated by the Kaplan-Meier method, and differences in the survival distributions between the mutation-positive and -negative groups were evaluated by the log-rank test. Statistical analyses were performed with StatView software (Abacus Concepts Inc., Berkeley, Calif., U.S.A.). For these analyses, the P values were two-tailed, and a P value of <0.05 was considered to indicate statistical significance.


Detection of mutations of the N-ras, p53 and FLT3 genes

Point mutations of the N-ras gene were studied in 19 patients (Table I). The N-ras gene was mutated in both stages (+/+) in three patients, positive at initial diagnosis and negative at relapse (+/−) in three patients, negative at initial diagnosis and positive at relapse (−/+) in two patients, and negative in both stages (−/−) in the remaining 11 patients.

Table 1. Table I. Summary of 28 patients with relapsed AML.Thumbnail image of
  • a

    Mutations of the N-ras, FLT3 and P53 genes are denoted as follows: +/+, positive at both stages; +/−, positive at initial diagnosis and negative at relapse; −/+, negative at initial diagnosis and positive at relapse; −/−, negative at both stages; ND, not done.* Patient 7 achieved a partial remission.

  • The FLT3 and p53 genes were analysed in the 28 patients (Table I). FLT3/ITD were +/+ in five patients, +/− in one patient, −/+ in six patients, and −/− in 16 patients. Representative PCR data are presented in Fig 1. The tandem duplications frequently involved a Y-rich stretch from codon 589–599 (data not shown), the same position as found in previous studies (Kiyoi et al, 1997; Yokota et al, 1997).

    Figure 1.

    . PCR analysis of the FLT3 gene and transcripts. Germline products were detected at 329 bp on PCR and at 456 bp on RT-PCR. Lanes 1–4 correspond to patients 2, 3, 7 and 14. Lane M, 100 bp size marker; D, at diagnosis; R, at relapse.

    The p53 gene mutations were +/+ in two patients, +/− in one patient, and −/− in the remaining 25 patients. In patient 1, CAC (codon 168) was changed to CGC. In patient 6, CAG (codon 167) and CGG (codon 248) were mutated to TAG and CAG, respectively. In patient 7, TCT (codon 227) and ATC (codon 254) were altered to TCG and AGC, respectively. Since these mutations in patient 7 were not detected at relapse (Fig 2A), we studied the p53 gene configuration by Southern blot analysis. The p53 gene was neither deleted nor rearranged at either stage (Fig 2B), ruling out the possibility that the mutated p53 gene was deleted at relapse. Furthermore, in this patient, FLT3 gene was mutated at relapse, whereas the same karyotype 46XX, inv(11) was observed. These results indicated that gene mutations of N-ras and FLT3 were unstable, and that not only acquisition but also deletion of mutations occurred during the course of leukaemia.

    Figure 2.

    . SSCP and Southern blot analyses of the p53 gene in patient 7. (A) PCR products of exon 7 of the p53 gene were electrophoresed as described in Materials and Methods. A mobility shift found at diagnosis was not detected at relapse. (B) Southern blots showed that the p53 gene was neither deleted nor rearranged. Genomic DNA was digested by Bgl II endonuclease. Full-sized cDNA of the p53 gene was used as a probe. Lane 1, control DNA; lane 2, at relapse; lane 3, at diagnosis.

    Karyotype analysis

    Karyotypes were compared between initial diagnosis and relapse in 24 patients (Table II). 13 patients exhibited the same karyotype at both stages. Chromosomal deletions, i.e. 3p−, 9q−, 20q−, appeared at relapse in patients 3, 26 and 11, respectively. Translocations were detected at relapse in patients 5 and 9. In patients 1, 13, 14, 20 and 28, cytogenetically unrelated clones emerged at relapse. In patient 6, loss of the abnormalities was observed.

    Table 2. Table II. Karyotype abnormalities at diagnosis and relapse.Thumbnail image of

    Clinical significance

    Since the above results indicated that gene mutations and karyotypes were unstable in relapsed AML, we investigated which factor influenced the instability. Whereas the DNA alterations in either N-ras, FLT3 or p53 genes at relapse were associated with the M4/5-subtype (P= 0.09, by the 2 × 2 table), the karyotype changes at relapse were not significantly associated with any parameters.

    We analysed the clinical significance of the N-ras, FLT3 and p53 gene mutations and karyotype changes at relapse using Kaplan-Meier curves (Fig 3). The N-ras gene mutation and karyotype change were unrelated to the prognosis (P= 0.43 and P= 0.72, respectively). On the other hand, FLT3/ITD and p53 mutations were related to short survival after the first relapse (P= 0.029, and P= 0.002, respectively). Both groups whose FLT3/ITD was +/+ and −/+ had a similarly poor prognosis com-pared with those without it (−/− and +/−) (data not shown).

    Figure 3.

    . Kaplan-Meier curves of overall survival in patients with relapsed AML. Patients with relapsed AML received either the same therapy as the induction chemotherapy or a combination therapy including mitoxantrone, VP-16 and Ara-C. Whereas N-ras gene mutation and karyotype change were unrelated to the prognosis, FLT3/ITD and p53 mutations were associated with short survival after the first relapse.


    In this study we showed that: (1) the N-ras and FLT3 gene configuration frequently changed in relapse of AML, (2) not only acquisition, but also loss of mutations occurred at relapse, and (3) FLT3/ITD and mutant p53 were associated with short survival after relapse.

    Sequential analyses of the mutated N-ras gene in the same patients with AML and myelodysplastic syndrome (MDS) have provided puzzling data regarding the association with leukaemia progression/relapse. Although N-ras mutation often appears during progression of MDS (Hirai et al, 1987), some AML patients lost the mutations at relapse (Farr et al, 1988). In this study, half (3/6) of the patients lost the N-ras mutation at relapse, whereas two patients acquired the mutation at relapse. The prognostic significance of the N-ras gene mutation also remains unclear (Radich et al, 1990; Neubauer et al, 1994). These results suggest that the N-ras gene mutation is unstable during the disease, and contributes little to the clonal advantage against treatment.

    This is the first report in which FLT3/ITD emerged at relapse. So far, FLT3/ITD has been found in 20% of de novo AML and 3% of MDS (Yokota et al, 1997). Sequential study of the MDS patients showed that FLT3/ITD appeared during leukaemia progression (Horiike et al, 1997). In this study we showed that 6/22 patients without FLT3/ITD at initial diagnosis carried it at relapse, whereas one of the six patients with the mutation at initial diagnosis lost it. Although the molecular significance of FLT3/ITD is still under investigation, the mutant FLT3 products are ligand-independently activated (Kiyoi et al, 1998). IL-3-dependent myeloid progenitor cell lines, FDC/P1 and 32D, exhibit IL-3-independent growth when transfected with mutant FLT3 genes (unpublished observations). Since FLT3/ITD is thus associated with leukaemia cell growth and/or inhibition of apoptosis, it is plausible that FLT3/ITD is closely associated with relapse of AML in addition to the leukaemia progression of MDS.

    p53 gene mutations are most frequently observed, occurring in approximately 50% of human tumours. In leukaemia, although the incidence is far lower, p53 gene mutations are significantly associated with poor prognosis (Wattel et al, 1994), resistance to chemotherapy, progression and relapse (Hu et al, 1992; Fenaux et al, 1992). In this study p53 gene mutations were found in 3/28 patients at initial diagnosis, and one of these three patients lost the mutation at relapse. All three patients failed to achieve second CR and thereafter lived for short periods: 15, 84 and 84 d.

    Unexpectedly, a significant percentage of the mutations found at initial diagnosis were undetectable at relapse. Typical examples were patients 7 and 14. In both patients, although the p53 gene or N-ras gene mutations were deleted, respectively, FLT3/ITD was newly identified at relapse. We postulate that relapse is best understood in the context of clonal shifts or collateral succession rather than simple acquisition of gene mutations. In this sense, relapse might be different from progression of MDS and similar to the adaptation that results from negative selection. Chemotherapy itself might increase the incidence of DNA-replication error, which leads to a molecular heterogeneity of leukaemia clones.

    Finally, we believe that sequential and molecular analysis of leukaemia provides a new perspective on genetic instability and resistance to chemotherapy, and propose a larger-scale study to elucidate the clinical significance of FLT3/ITD in relapse.


    We thank Chisato Kamiya and Yoko Kudo for technical assistance. This work was supported in part by grants-in-aid from the Japanese Ministry of Health and Welfare (No. 9-3), the Public Trust Haraguchi Memorial Cancer Research Fund, Kane Foundation for Life and Socio-Medical Science, Uehara Memorial Foundation and Osaka Cancer Research Foundation.