the glycosylation of proteins or newly synthesized glycoproteins may play important roles in the myeloid differentiation process. The difference spectrum showed ...
Correspondence
1674 the glycosylation of proteins or newly synthesized glycoproteins may play important roles in the myeloid differentiation process. The difference spectrum showed an obvious increase for the band centered at 1154 cm1 (Figure 2c). This band assumed a special meaning because it changed from 1170 cm1 consecutively (Figure 4). They are due to the hydrogen-bonded and nonhydrogen-bonded C–O stretching mode of the C–OH groups of serine, threonine and tyrosine in proteins, respectively as well as the C–O groups in carbohydrates,8 indicating the protein glycosylation and phosphorylation processes more readily take place in the differentiation process. In contrast with the 968 cm1 band, the band at 1086 cm1 shifted apparently to lower frequency after ATRA treatment, indicating that the hydrogen bonding interactions within nucleic acids were not as strong as the consistent untreated ones. This, in turn accounted for structural alterations in the chromosomes of differentiated cells, as manifested by morphological observation (Figure 1a). Two parameters (D1053/D1085 and D1153/D1085) were introduced to describe these changes, representing the optical density ratios between 1053, 1153 and 1085 cm1. Both of them increased gradually in a time- and dose-dependent manner, resulting in good linear correlations with the differentiation index (the correlation coefficient 40.9, Po0.001). The spectral changes in HL-60 cells were similar with NB4, indicating the shared development process in myeloid differentiation. However, the spectrum of NB4-MR2 cells did not change much with ATRA treatment. In summary, it is possible to use FT-IR spectroscopy to monitor the biochemical events underlying the different phases of ATRAinduced NB4 cell differentiation. The clearcut spectral changes in lipids, nucleic acids and carbohydrates could lead to several testable hypotheses, providing important insights into the mechanisms of leukemogenesis and normal granulocytic differentiation. One or more spectroscopic parameters may have potential uses as diagnostic indicators to assist treatment decisions for APL patients. FT-IR spectroscopy also has the advantages of good reproducibility, rapid determination, and the possibility of examining untreated microsamples. It may have potential application in clinical study for fast and reagent-free diagnosis. However, such a possibility still requires further exploration.
Acknowledgements This work was supported by grants from Tsinghua University 985 Project. The authors would like to thank Dr Zhu CHEN (Rui-jin Hospital, Shanghai Second Medical University) for kindly providing NB4 and NB4-derived MR2 subclone cells.
M-J Liu1 Z Wang R-C Wu1 S-Q Sun2 Q-Y Wu1
1 Department of Biological Sciences and Biotechnology, Tsinghua University, Beijing, People’s Republic of China; 2 Department of Chemistry, Tsinghua University, Beijing, People’s Republic of China
References 1 Chomienne C, Ballarini P, Baltitrand N, Amar M, Benard JF, Boivin P et al. Retinoic acid therapy for acute promyelocytic leukemia. Lancet 1989; 23: 746–747. 2 Liu TX, Zhang JW, Tao J, Zhang RB, Zhang QH, Zhao CJ et al. Gene expression networks underlying retinoic acid-induced differentiation of acute promyelocytic leukemia cells. Blood 2000; 96: 1496–1504. 3 Jackson M, Sowa MG, Mantsch HH. Infrared spectroscopy: a new frontier in medicine. Biophys Chem 1997; 68: 109–125. 4 Gasparri F, Muzio M. Monitoring of apoptosis of HL60 cells by Fouriertransform infrared spectroscopy. Biochem J 2003; 369: 239–248. 5 Liu KZ, Skultz CP, Johnston JB, Beck FWJ, Al-Katib AM, Mohammad RM et al. Infrared spectroscopic study of bryostatin-1-induced membrane alterations in a CLL cell line. Leukemia 1999; 13: 1273–1280. 6 Spremolla G, Benedetti E, Vergamini P, Andreucci MC, Macchia P. An investigation of acute lymphoblastic leukemia (ALL) in children by means of infrared spectroscopy. Part IV. Haematologica 1988; 73: 21–24. 7 Smolenska-Sym G, Zdebska E, Golaszewska E, Wozniak J, Durzynski T, Maj S et al. Contents of total and protein-bound carbohydrates were low in leukemic leukocytes from patients with acute myelogenous leukemia. Acta Biochim Pol 1998; 45: 361–371. 8 Rigas B, Wong PT. Human colon adenocarcinoma cell lines display infrared spectroscopic features of malignant colon tissues. Cancer Res 1992; 52: 84–88.
Minimal residual disease in acute myeloid leukemia is predicted by P-glycoprotein activity but not by multidrug resistance protein activity at diagnosis Leukemia (2003) 17, 1674–1677. doi:10.1038/sj.leu.2403025
TO THE EDITOR Although chemotherapy induces morphologic complete remission (CR) in most acute myeloid leukemia (AML) patients, many will eventually relapse due to the persistence and subsequent outgrowth of minimal residual disease (MRD). It has been reported that high MRD levels at follow-up and high P-glycoprotein (Pgp, ABCB1) activity at diagnosis correlate with shorter survival.1–3 This is less clear for other ABC-transporters such as multidrug resistance protein (MRP1, ABCC1) and breast cancer resistance protein (BCRP, ABCG2).3 Based on our previous findings of homogeneous distribution of drug efflux capacity over the total blast population at diagnosis and the constancy of both Pgp, MRP1 and BCRP function during the course of disease, including MRD and relapse,4
Correspondence: GJ Schuurhuis, Department of Hematology, VU University Medical Center, PO Box 7057, 1007 MB Amsterdam, The Netherlands; Fax: +31 20 444 2601 Received 5 February 2003; accepted 18 April 2003 Leukemia
we hypothesized that the emergence of MRD and its frequency at follow-up might be a direct resultant of the presence and level of activity of ABC-transporters at diagnosis. No study has been designed thus far to address this question directly, although preliminary evidence favoring this idea has been provided for Pgp:1,2 patients with high MRD levels after chemotherapy had higher Pgp expression/activity at diagnosis. No such studies have been performed for MRP1 and BCRP. Since we previously found that the majority of AML patients display Pgp and MRP1 activity at diagnosis, whereas BCRP activity was present in 3/26 patients only,4 we focused this study primarily on the ABC transporters Pgp and MRP1. For these purposes, we collected after informed consent, bone marrow (BM) samples from AML patients at the time of diagnosis at the VU University Medical Center. A total of 100 patients were included with a median age of 56 years, ranging from 16 to 88, with the following French–American–British classification: M0 (9), M1 (9), M2 (17), M3 (4), M4 (17), M5 (26), M6 (4), MDS (9), not specified (5). Patients were treated with chemotherapeutic schemes according to the HOVON protocols of the Dutch–Belgian Hemato– Oncology Cooperative Group of AML. From 54 of these patients, we
Correspondence
1675 obtained in total 117 follow-up samples in situations of MRD, when the patient was in morphologic CR. Blasts at diagnosis were immunophenotyped in order to identify leukemia-associated phenotypes (LAPs), that is, combinations of cell surface markers
that are not, or only in low frequencies, expressed on normal hematopoietic cells as previously described.4–6 In the present patient cohort, 76% of the patients displayed one or more LAPs, enabling the detection of MRD at follow-up. High MRD frequencies
Figure 1 Correlation between Pgp activity at diagnosis and MRD frequency. Pgp activity at diagnosis plotted against MRD levels detected in BM after the first (a) and second (c) course of induction chemotherapy, in autologous stem cell grafts (e) and in BM after consolidation therapy (g). A ratio 41.1 indicates Pgp activity. Box-and-Whisker plots in (b), (d), (f) and (h) show in the box the 25th and 75th percentile, while the whiskers show the 2.5th and 97.5th percentile and the central lines the median. In the left boxes in (f) and (h), the central line is not visible due to colocalization with the lower line of the boxes. Patients were grouped according to a Pgp activity level at diagnosis of o1.5 and X1.5, for which the corresponding MRD% is depicted. In the lower left side of (a), (c), (e) and (g) are 10, 10, 12 and 12 patients, respectively, with a Pgp activity o1.5 and an MRD percentage of 0.01%. Leukemia
Correspondence
1676 at follow-up, that is, in BM after the first and second induction course, autologous peripheral blood stem cell grafts and BM after a consolidation course, were associated with shorter disease-free survival (Spearman’s r: 0.434, P ¼ 0.034 with n ¼ 24; 0.613, P ¼ 0.001 with n ¼ 27; 0.585, P ¼ 0.003 with n ¼ 23; and 0.656, P ¼ 0.001 with n ¼ 24, respectively). These results show that MRD frequency assessment at follow-up is a strong predictor of survival, which is in concordance with literature.1,2 Then we studied the relation between activities of Pgp and MRP1 at diagnosis and MRD frequency at follow-up. For this, we measured Pgp and MRP1 function at diagnosis using a green fluorescent probe assay with Syto16/PSC833 (for Pgp) and calceinAM/probenecid (for MRP1) as a specific substrate/modulator combination as previously described.4,5 Pgp activity (median ratio 1.9, range 1.1–32) was found in 76% (76/100) and MRP1 activity (median ratio 1.3, range: 1.1–2.1) in 56% (40/72) of de novo AML patients studied, defined by a cutoff of fluorescent shift X1.1.4 We observed a correlation between Pgp activity levels at diagnosis and MRD frequency at follow-up, when Pgp was analyzed as continuous variable (Figure 1a, c, e and g): the best correlations were found after the second induction course and after consolidation therapy, with a trend for autologous stem cell grafts (Figure 1c, g and e, respectively). In contrast, in BM after the first induction course, Pgp activity did not correlate at all with MRD frequency (Figure 1a). Patients with Pgp activity X1.5 at diagnosis showed higher MRD levels at follow-up (Figure 1b, d, f and h), again with the best correlations after the second induction course and after consolidation treatment (Figure 1d and h, respectively). Again no such differences in MRD frequencies could be observed in BM obtained after the first induction course (Figure 1b). These findings were confirmed by using an alternative approach: when defining patients with low (p0.03%) and high (40.03%) levels of MRD at follow-up and by comparing the Pgp activity at diagnosis in the resulting groups, it was found that patients with high MRD levels after the second induction course and after consolidation therapy had significantly higher Pgp activity at diagnosis (median ratio of 1.41 vs 1.09: Mann–Whitney U-test P ¼ 0.044; and 2.13 vs 1.14: P ¼ 0.018, respectively), when compared with patients showing low MRD levels, similar to the data of San Miguel et al.1 In contrast, for MRP1 function, no significant correlations were found with MRD levels at follow-up, neither when MRP1 activity was analyzed as a continuous variable nor when using different cutoff levels (data not shown). Considering the established strong relation between MRD frequency and survival parameters on the one hand and between Pgp but not MRP1 activity and MRD frequency on the other hand, correlations of Pgp and MRP1 activity with outcome were made: when using the same cutoff level for Pgp activity as used to correlate with MRD frequency at follow-up, that is, X1.5, it was found that patients with Pgp activity X1.5 showed significantly shorter event-free survival (median 8 vs 12 months, Kaplan–Meier logrank test: P ¼ 0.042) when compared with patients showing lower Pgp activities. In line with the complete absence of a direct correlation between MRP1 activity and MRD levels, there was no correlation between MRP1 activity and event-free survival, neither when MRP1 activity was analyzed as a continuous variable (r: 0.161, P ¼ 0.193 with n ¼ 67) nor with different cutoff levels (not shown). Although BCRP activity was not present in the majority of AML patients, all patients did show low levels of BCRP expression, with the monoclonal antibodies BXP-21 and BXP-34.4 However, these levels were all lower than found for BCRP-negative cell lines.4 We found no correlation between BCRP expression and event-free survival (r: 0.008, P ¼ 0.969 and 0.058, P ¼ 0.770 for BXP-34 and BXP-21, respectively). Similarly, in a subgroup of 13 patients, we could not detect a correlation with MRD frequency determined after the second induction course (r: 0.157, P ¼ 0.609 and 0.399, P ¼ 0.177 for BXP-34 and BXP-21, respectively). Although tested in Leukemia
a small patient group, based on these findings we suppose that it is not likely that BCRP plays a major role in treatment failure in adult AML. Altogether our results show that of the ABC-transporters studied, Pgp activity directly correlates with MRD frequency measured at several time points during treatment of disease. This in turn explains poor outcome and thus offers an explanation for the adverse prognostic impact of Pgp in AML. The observed correlation between Pgp activity and MRD frequency may either be causally related to the decreased ability of blasts to retain cytostatic drugs in the presence of active Pgp or result from an alternative function of Pgp, that is, its role as an inhibitor of apoptosis.7 Inhibitors of Pgp function, although in some studies showing improvement in therapy response, thus far have not been very successful due to considerable increases of toxicity.8 Further intensification of the currently applied chemotherapeutic treatment protocols in general will not be applicable for the whole group of patients. However, recently, by combining detection of MRD frequency, normal CD34+ cell frequency and length of hypoplasia period after the induction courses, we identified a subgroup of patients that responded poorly to antitumor treatment and, in addition, experienced low BM toxicity and might thus be eligible for dose intensification.6 The level of MRD is likely directly related to the level of resistance of leukemic cells toward chemotherapy-induced cell death. It is presumed that apart from ABC-transporters other factors might contribute; these include apoptosis resistance or more indirect effects resulting from cellular interactions with the BM microenvironment. This is compatible with the following finding: although we observed that the majority of patients with high MRD frequency at follow-up showed high Pgp activity at diagnosis, a considerable part of the patients with high MRD frequencies had low Pgp activity at diagnosis (see Figure 1). At present, this complexity makes it difficult to predict upfront, at the time of diagnosis with high accuracy, which patients will end up with high MRD and in turn a high probability of relapse. On the other hand, the finding that after effective debulking of the tumor, that is, after the second course of induction chemotherapy, the resulting MRD levels are highly predictive for relapse opens the intriguing possibility to apply alternative therapies in the MRD situation that might circumvent transport-related resistance.
Acknowledgements We thank G Evertse, C Dekker-van Roessel, M Leidekker, C Eeltink, M Leissink and Y den Hartog for retrieving clinical patient’s characteristics. J Pater and M van der Maas are acknowledged for technical assistance with the drug efflux experiments. 1 MA van der Pol1 Department of Hematology, VU University N Feller1 Medical Center, Amsterdam, The Netherlands; 2 GJ Ossenkoppele1 Department of Medical Oncology, VU University Medical Center, Amsterdam, The GWD Weijers1 Netherlands AH Westra1 A van Stijn1 HJ Broxterman2 GJ Schuurhuis1
References 1 San Miguel JF, Martinez A, Macedo A, Vidriales MB, Lopez-Berges C, Gonzalez M et al. Immunophenotyping investigation of minimal residual disease is a useful approach for predicting relapse in acute myeloid leukemia patients. Blood 1997; 90: 2465–2470. 2 Venditti A, Buccisano F, Del Poeta G, Maurillo L, Tamburini A, Cox C et al. Level of minimal residual disease after consolidation therapy
Correspondence
1677 predicts outcome in acute myeloid leukemia. Blood 2000; 96: 3948– 3952. 3 van der Kolk DM, de Vries EG, Muller M, Vellenga E. The role of drug efflux pumps in acute myeloid leukemia. Leuk Lymphoma 2002; 43: 685–701. 4 van der Pol MA, Broxterman HJ, Pater JM, Feller N, van der Maas M, Weijers GWD et al. Function of the ABC transporters, P-glycoprotein, multidrug resistance protein and breast cancer resistance protein in minimal residual disease in acute myeloid leukemia. Haematologica 2003; 88: 134–147. 5 van der Pol MA, Pater JM, Feller N, Westra AH, van Stijn A, Ossenkoppele GJ et al. Functional characterization of minimal residual disease for P-glycoprotein and multidrug resistance protein activity in acute myeloid leukemia. Leukemia 2001; 15: 1554–1563.
6 Feller N, Schuurhuis GJ, van der Pol MA, Westra G, Weijers GW, van Stijn A et al. High percentage of CD34-positive cells in autologous AML peripheral blood stem cell products reflects inadequate in vivo purging and low chemotherapeutic toxicity in a subgroup of patients with poor clinical outcome. Leukemia 2003; 17: 68–75. 7 Johnstone RW, Cretney E, Smyth MJ. P-glycoprotein protects leukemia cells against caspase-dependent, but not caspase-independent, cell death. Blood 1999; 93: 1075–1085. 8 Baer MR, George SL, Dodge RK, O’Loughlin KL, Minderman H, Caligiuri MA et al. Phase 3 study of the multidrug resistance modulator PSC-833 in previously untreated patients 60 years of age and older with acute myeloid leukemia: Cancer and Leukemia Group B Study 9720. Blood 2002; 100: 1224–1232.
Absence of somatic mutations within the Runt domain of AML2/RUNX3 in acute myeloid leukaemia Leukemia (2003) 17, 1677–1678. doi:10.1038/sj.leu.2403007
TO THE EDITOR, The three RUNX/AML proteins constitute a family of DNA-binding alpha subunits of the heterodimeric transcription factor complex PEBP2/CBF. The Runt domain, named after its sequence homology with the Drosophila pair-rule gene, runt, is a protein motif of 128 amino acids that has been highly conserved during evolution. This domain is responsible for binding of the transcription factor to the consensus sequence 50 -TG(T/C)GGT-30 as well as for heterodimerisation with the non-DNA-binding b subunit. While the expression of RUNX2/AML3 is restricted to osteochondrogenic tissue, RUNX1/AML1 and RUNX3/AML2 are expressed at high levels in cells of haematopoietic origin. AML1 was first identified as the gene on chromosome 21 that is disrupted in the (8;21)(q22;q22) translocation. This translocation leads to expression of the chimeric transcription factor AML-ETO and is one of the most frequent chromosomal abnormalities identified in acute myeloid leukaemia (AML). Subsequent work has shown that the AML1 gene is also disrupted in t(3;21)(q26;q22) in blast crisis of chronic myelogenous leukaemia and therapy-related AML as well as in t(12;21)(p12;q22), a translocation frequently found in precursor Bcell acute lymphoblastic leukaemia of children. The human CBFB gene that encodes the b subunit is disrupted in inv(16)(p13;q22), a chromosomal abnormality associated with AML, usually of the FAB subtype M4eo. Several recent reports have described somatic mutations involving one or both alleles of the AML1 gene coding region in AML and myelodysplastic syndrome (MDS).1–3 Constitutional mutations of one AML1 allele result in the syndrome of familial platelet disorder with predisposition to acute myeloid leukaemia (FPD/AML; OMIM #601399).4 Two other haematopoietic transcription factors that have been shown to act in concert with AML1 are C/EBPalpha and PU.1. While AML-ETO downregulates C/EBPalpha, PU.1 synergises with AML1 at the level of target gene promoters. For both factors, somatic mutations have been identified in AML.5,6 Thus, a genetic event disturbing AML1 or its molecular context seems to contribute to leukaemogenesis in a sizeable fraction of AMLs. Recently, the question of leukaemogenic mutations was also addressed for the AML1 heterodimerisation partner CBFb. However, in contrast to the genes encoding AML1, C/EBPalpha
Correspondence: Dr F Otto, Division of Haematology/Oncology, Department of Internal Medicine I, Hugstetterstr. 55, Freiburg D-79106, Germany; Fax: +49 761 270 7177 Received 12 February 2003; accepted 3 April 2003
and PU.1, no evidence could be found for a dysregulation by point mutations of the CBFB gene in AML or MDS.7 The AML2 protein binds to the same consensus sequence in the promoters of target genes as AML1 and is expressed in an overlapping fraction of haematopoietic cells. Therefore, it is reasonable to hypothesise that AML2 may have an overlapping functional role with AML1 in these cells and might even act as a substitute. This is further strengthened by the observation that AML2 expression is regulated by AML1.8 Genetic events inactivating AML2 might therefore be expected in AML leukaemic blasts. Based on these findings we wished to determine whether somatic mutations affecting the Runt domain of AML2 are present in AML cells. Exons 2–4 of the AML2 gene were amplified and sequenced from genomic DNA samples extracted from peripheral blood or bone marrow of 46 AML patients (for primer sequences see Table 1). These exons encompass the entire Runt domain of AML2 including the nuclear localisation signal (see Figure 1). The samples correspond to the FAB subtypes as follows: M0, 10; M1, 6; M2, 9; M3, 4; M4, 6; M5, 6; unknown, 5. The choice of leukaemia samples was to cover all subtypes according to FAB, particularly subtype M0, because previous studies have shown a high frequency of Table 1 cing
Sequences of primers used for amplification and sequen-
Exon
Primer
Sequence
Exon 2
Ex2 for Ex2 rev
50 -CTT CTG CTT TCC CGC TTC TC -30 50 -TCC CGC ACT CAC CTT GAA G -30
Exon 3
Ex3 for Ex3 rev
50 -GTC TTC AGG TGG TGG CAT TGG -30 50 -TTC CAC TTA CCT CGC CCA CTG -30
Exon 4
Ex4 for Ex4 rev
50 -CAA CCG CCT GCC TCT ATT C-30 50 -TCG GTG GCA CTT ACG TCT G -30
Figure 1 Relative location of primers in the AML2/RUNX3 gene. Forward and reverse primers were designed to bind to intronic sequence located 50 and 30 of exons 2–4, respectively. Exons are represented by boxes. Dark boxes indicate the Runt domain. Primers are shown as arrows. Intron sizes are indicated. Leukemia