Fax: (713) 794-4297
Characteristics and outcome of patients with Philadelphia chromosome negative, bcr/abl negative chronic myelogenous leukemia
Article first published online: 3 OCT 2002
Copyright © 2002 American Cancer Society
Volume 95, Issue 8, pages 1673–1684, 15 October 2002
How to Cite
Onida, F., Ball, G., Kantarjian, H. M., Smith, T. L., Glassman, A., Albitar, M., Scappini, B., Rios, M. B., Keating, M. J. and Beran, M. (2002), Characteristics and outcome of patients with Philadelphia chromosome negative, bcr/abl negative chronic myelogenous leukemia. Cancer, 95: 1673–1684. doi: 10.1002/cncr.10832
- Issue published online: 3 OCT 2002
- Article first published online: 3 OCT 2002
- Manuscript Accepted: 6 MAY 2002
- Manuscript Received: 28 NOV 2001
- gene rearrangement;
- hematologic variables;
- myeloproliferative disorder;
- risk categories
Up to 5% of patients with chronic myelogenous leukemia (CML) do not have the Philadelphia (Ph) translocation t(9;22)(q34;q11) or a bcr/abl molecular rearrangement. Although the diagnostic criteria of this entity are still under debate, there is general agreement that patients with Ph negative, bcr/abl negative CML have a severe clinical course that is not affected significantly by current treatment options.
A population of 76 patients with bcr/abl negative CML who had received minimal or no previous therapy was characterized carefully with the intent of investigating clinical and hematologic variables and their association with survival by univariate, correlation, and multivariate analyses. A group of 73 patients with Ph negative CML who were not tested for the bcr/abl rearrangement (bcr/abl unknown) was analyzed separately and used for extension of the analysis.
In the bcr/abl negative patient population, the median overall survival was 24 months. At the time of the analysis, 38 patients (50%) had died, and blastic transformation preceded death in 31%. Chromosomal abnormalities were found in 30% of the 76 patients, with trisomy 8 the most common abnormality. Complex chromosomal abnormalities were rare, and monosomy 7 was not observed. Survival was not affected significantly by treatment. Multivariate analysis identified older age (> 65 years), anemia (hemoglobin < 10 g/dL), and severe leukocytosis (white blood cells > 50 × 109/L) as variables with independent prognostic significance for poor survival. A prognostic scoring system stratified patients into a low-risk group (53%) and a high-risk group (47%), with median survivals of 38 months and 9 months, respectively.
Bcr/abl negative CML is a distinct clinical entity associated with very poor prognosis. Two risk categories are identifiable using a simple scoring system based on age, hemoglobin level, and leukocyte number. Cancer 2002;95:1673–84. © 2002 American Cancer Society.
More than 90%1–7 of patients who are diagnosed with a morphologic picture of chronic myelogenous leukemia (CML) demonstrate the characteristic Philadelphia chromosome (Ph) karyotypic abnormality, t(9;22), by cytogenetic analysis.8 At the molecular level, t(9;22) corresponds to the fusion of parts of the c-abl gene to parts of the bcr gene (bcr/abl).9, 10 The remaining < 10% of patients are classified with Ph negative CML by conventional cytogenetics. In these patients, the bcr/abl gene rearrangement is detected by molecular studies in 25–50%.9–20 Such patients are classified with Ph negative, bcr/abl positive CML. By reliably documenting the absence of the central molecular event in CML, namely, the bcr/abl rearrangement, molecular studies brought strong support to the existence of a long-disputed category, that is Ph negative, bcr/abl negative CML.10, 12, 20–26
Although differences in clinical characteristics and disease course have been reported in patients who have bcr/abl negative CML compared with patients who have bcr/abl positive CML,15, 23, 24, 27–31 these differences were based on small numbers of patients. Such studies suggested that patients with bcr/abl negative CML had a more aggressive clinical course, a poor response to therapy, and a shorter survival. However, the study groups in these analyses were too small to allow assessment of the significance of variables associated with survival. Indeed, at the time of diagnosis, patients with this disorder usually are not distinguishable clinically from patients with bcr/abl positive CML.32 Thus, reliable information on the natural history of bcr/abl negative CML and survival-associated covariates is needed. Herein, we expand our experience22, 23, 33 with patients who have Ph negative CML, giving particular attention to the subset of patients with bcr/abl negative CML. The objectives of this study were 1) to analyze patient and disease characteristics of Ph negative, bcr/abl negative CML; 2) to identify independent covariates associated with survival; and 3) to design a simple and clinically useful scoring system capable of identifying different risk groups that would help to assign patients to risk-oriented therapeutic strategies.
MATERIALS AND METHODS
From January, 1967 through September, 1999, 295 patients were referred to The University of Texas M. D. Anderson Cancer Center with a diagnosis of Ph negative myeloproliferative disorder (MPD). Patients who met the current criteria for essential thrombocythemia,34 polycythemia vera,35 chronic myelomonocytic leukemia (CMML),33, 36 myelodysplastic syndromes (MDS),37 idiopathic myelofibrosis,38 or hypereosinophilic syndrome39 were excluded from the analysis. A diagnosis of Ph negative CML was confirmed in 189 patients.
Except for a difference in the cut-off value for the peripheral white blood cell (WBC) count (WBC > 10 × 109/L in this study compared with WBC > 20 × 109/L22, 23, 33), the criteria used to establish a diagnosis of Ph negative CML were identical to the criteria we used in our 1986 report:22 1) absence of Ph on analysis of at least 20 bone marrow mitoses; 2) hypercellular bone marrow with granulocytic hyperplasia, leftward shift in myeloid maturation, bone marrow blasts < 30%, and absence of significant dysplasia; 3) persistent, unexplained peripheral granulocytic leukocytosis (WBC count > 10 × 109/L); 4) peripheral blast cells < 30%; 5) monocyte percentage < 10% in the peripheral blood cells and absolute monocyte count < 1 × 109/L in patients with WBC counts < 20 × 109/L12, 33); and 6) absence of substantial bone marrow myelofibrosis. An absence of molecular evidence for the bcr/abl fusion gene was the final and necessary criterion used to define the bcr/abl negative subpopulation of patients. An additional criterion for inclusion in the analysis was a history of minimal therapy or no therapy prior to referral to The University of Texas M. D. Anderson Cancer Center.
Of the 189 patients who fulfilled the criteria for Ph negative CML, 116 patients were tested further for the presence of the bcr/abl rearrangement either by Southern blot analysis (performed at our institution from 1984 to 1988; n = 27 patients) or by reverse-transcriptase polymerase chain reaction (RT-PCR) analysis (performed from 1989 to the present; n = 89 patients). Forty patients who were documented with bcr/abl positive CML were excluded from this study. Thus, the main cohort of this analysis comprised patients with 76 Ph negative, bcr/abl negative CML. Molecular studies that would have been required to exclude the presence of the bcr/abl fusion gene were not performed in the remaining 73 patients. This group of patients with Ph negative, bcr/ablunknown CML was analyzed separately and was used only for an extension of the cytogenetic analysis to the whole cohort of 149 patients with Ph negative CML. The median time between the first diagnosis and referral to The University of Texas M. D. Anderson Cancer Center was 1 month (range, 0–110 months).
At the time of their referral, 40 of the 76 patients (53%) with bcr/abl negative CML had received some hydroxyurea-based or busulfan-based treatment, whereas 36 patients had not been treated. After referral, 31 patients received only supportive care, and 34 patients received single-agent therapy with hydroxyurea (n = 9 patients), α-interferon or γ-interferon (n = 17 patients), decitabine (5-aza-2′-deoxycytidine; n = 4 patients), 9-nitrocamptothecin (n = 3 patients), or 5-azacitidine (n = 1 patient). The other 11 patients received more intensive intravenous and/or combination chemotherapy. The selection of therapy for each patient at The University of Texas M. D. Anderson Cancer Center was based on the clinical presentation, disease-related symptoms, medical complications, and the period during which patients were referred.
Blood and bone marrow studies were performed upon referral to The University of Texas M. D. Anderson Cancer Center by members of the Department of Hematopathology without knowledge of previous clinical, cytogenetic, or molecular data. Differential blood counts were based on manual reading of Giemsa-stained smears and were performed on 200–400 cells. Standard morphologic criteria were used to evaluate dysplastic features.37 Iron stains were performed on bone marrow films to exclude refractory anemia with ring sideroblasts. Bone marrow biopsies were reviewed in all patients to exclude idiopathic myelofibrosis.
Except for patients who were admitted before the introduction of chromosome banding at The University of Texas M. D. Anderson Cancer Center in 1973, all patients had cytogenetic analysis using the GTG banding technique on bone marrow and peripheral blood cells, which were processed routinely after 24–48 hours in culture. A minimum of 20 G-banded metaphases were analyzed and classified according to the International System for Human Cytogenetic Nomenclature guidelines.40
Molecular analysis to detect the bcr/abl rearrangement was performed routinely using specific probes in Southern blot analysis from 1984 to 1988 and, subsequently, using specific primers with PCR-assisted methodology. High-molecular-weight DNA was prepared, digested with restriction endonucleases (International Biotechnologies Inc., New Haven, CT), and subjected to electrophoresis on 0.8% agarose gels, as described previously.41 After they were transferred to nylon membranes, the samples were hybridized to a 3′ bcr probe (1.2-kb Hind III/Bg/II genomic probe) and a larger universalbcr probe (Ph1-bcr/3; Oncogene Science, Inc., Manhasset, NY) encompassing most of the 5.8-kb bcr region. In 13 patients with bcr/abl negative status and in 3 patients with unknown bcr/abl status, N-RAS, K-RAS, and H-RAS oncogene sequences (codons 12, 13, and 61) were analyzed for the presence of mutations using PCR followed by either specific oligonucleotide hybridization42 or direct sequencing of the PCR products with the ABI Sequence Detection System 310 (PE Applied Biosystems, Foster City, CA).
Total RNA was extracted from patient samples using the reagent Trizol (Gibco BRL, Gaithersburg, MD), as recommended by the manufacturer. Total RNA was subjected to reverse transcription with random hexamers followed by PCR amplification for detection of hybrid BCR/ABL transcripts (b2-a2, b3-a2), as described elsewhere.43
Prognostic Factor Analysis and Statistical Methods
Patient-associated and disease-associated variables were investigated. The variables considered were age; gender; the presence of palpable splenomegaly; hemoglobin level; platelet and WBC, percentage of neutrophils and absolute neutrophil count; monocyte and lymphocyte counts; percentages of basophils, eosinophils, and circulating immature myeloid cells (including blast cells); percentages of bone marrow blasts, monocytes, lymphocytes, and erythroid cells; serum levels of lactate dehydrogenase (LDH) and β2-microglobulin; cytogenetic profiles; and the presence or absence of the bcr/abl rearrangement and of N-RAS, K-RAS, or H-RAS point mutations. The administration of any treatment compared with supportive therapy was also evaluated. Except for treatment that was given after referral, all variables were registered at the time of the initial presentation at The University of Texas M. D. Anderson Cancer Center.
Covariates were analyzed for their association with the duration of survival, which was calculated from the date of referral to our institution. Univariate analysis was based on Kaplan–Meier survival curves44; results were compared with the log-rank test. P values < 0.05 were considered statistically significant. Correlation coefficients were calculated to evaluate associations between pairs of continuous variables. To select significant independent factors, we applied a multivariate regression method to analyze the association of multiple patient and disease characteristics with survival. Using a Cox proportional hazards model,45 variables were selected by a backward, stepwise procedure. This technique allows the evaluation of the relative prognostic importance of each factor while correcting for the effects of other covariates.
Prognostic Scoring System
The risk variables for designing our prognostic model were selected based on their relative importance, weighing each prognostic variable according to regression coefficients that were determined by using the proportional hazards regression analysis. The numbers provided for the risk scoring values were approximated to the nearest 1.0 unit. Two risk categories were created combining patients who scored 0.0 or 1.0 into one group (low risk) and patients who scored 2.0 or 3.0 into another group (high risk).
Acute Leukemia Transformation
When possible, the disease status at the time of death was verified, and the cause of death was identified, with particular care taken to note whether blastic transformation developed before death.
Table 1 summarizes the initial characteristics of the study group, which was comprised of 42 men (55.3%) and 34 women (44.7%) with a median age of 66 years (range, 24–88 years). Thirty-eight patients (50%) presented with palpable splenomegaly.
|Continuous variables Median (range)|
|Age (yrs)||66 (24–88)|
|Hemoglobin (g/dL)||10.6 (7.3–16.1)|
|Platelets (× 109/L)||160 (8–1105)|
|White blood cells (× 109/L)||38 (11.1–296)|
|Neutrophils (%)||70 (23–89)|
|Neutrophils (× 109/L)||25.6 (5.5–156)|
|Monocytes (%)||2 (0–10)|
|Monocytes (× 109/L)||1.0 (0–9)|
|Lymphocytes (%)||9 (1–30)|
|Lymphocytes (× 109/L)||4.0 (0.4–21.5)|
|Eosinophils (%)||1 (0–19)|
|Basophils (%)||0 (0–10)|
|Peripheral blood immature myeloid cells (%)||13 (0–52)|
|Bone marrow blasts (%)||1 (0–29)|
|Bone marrow monocytes (%)||2 (0–7)|
|Bone marrow lymphocytes (%)||4 (0–23)|
|Bone marrow erythroid cells (%)||9 (0–39)|
|Lactate dehydrogenase (U/L)||1389 (210–6960)|
|β2-microglobulin (mg/L)a||3.8 (2.0–11.8)|
|Categoric variables: number of patients (%)|
|Male gender||42 (55.3)|
|Abnormal karyotype||22 (30.1)|
|RAS point mutationb||3 (23)c|
The results of cytogenetic studies are shown in Table 2. PCR analysis verified that the three patients for whom data on cytogenetics were unavailable lacked the bcr/abl molecular rearrangement. Thirty percent of patients had chromosomal abnormalities, with the most common abnormality a sole trisomy 8 (27% of abnormal karyotypes). Deletion of the long arm of chromosome 20 (20q−) was identified in four patients (18% of abnormal karyotypes). Only one patient had a complex karyotype (three or more abnormalities). No patients had monosomy 7, which has been associated with MDS and other types of MPD.46–50 A point mutation of the N-Ras or K-Ras oncogenes was detected by PCR analysis in 3 of 13 patients tested (23%) (Table 1).
|Karyotype||No. of patients (%)|
|bcr/abl Negative (n = 76 patients)||bcr/abl Unknown (n = 73 patients)|
|Total||76 (100)||73 (100)|
|Data unavailablea||3 (3.9)||0 (—)|
|1) Diploid||51 (69.9)||49 (67.1)|
|2) + 8||6 (8.2)||4 (5.5)|
|3) + 8 with one additional abnormality||1 (1.4)||1 (1.4)|
|4) + C (with or without other single abnormalities)||0 (—)||6 (8.2)|
|5) 20q- (with or without other single abnormalities)||4 (5.5)||0 (—)|
|6) t(5;12)(q33;p13)||0 (—)||2 (2.7)|
|7) Complexb||1 (1.4)||4 (5.5)|
|8) Other chromosomal abnormalitiesc||10 (13.7)||7 (9.6)|
In the group of 73 patients who were not tested for the presence of bcr/abl, the proportion of patients with normal karyotype was identical to that in the patients with bcr/abl negative CML (Table 2). Seventeen patients who were referred to our institution before 1973 had unavailable banding; of these, 6 patients had a +C abnormality (with or without other single abnormalities). Trisomy 8 was present in 21% of abnormal karyotypes, and 4 patients had complex abnormalities.
At the time of this analysis (April, 2000), 38 patients were still alive, and 38 patients had died. The median survival was 24 months; 53% of patients survived for 1 year, and only 7% of patients survived beyond 5 years. No significant difference in survival was observed between patients who received treatment or did not receive treatment before referral (42 patients and 36 patients, respectively). There was no significant difference in survival between patients who were referred to our institution at the first occurrence of hematologic disorder and patients who were referred 1 month, 3 months, 6 months, 12 months, or > 12 months later (data not shown). Finally, no significant difference in survival was detected between patients who received therapy after referral to The University of Texas M. D. Anderson Cancer Center and patients who did not, although there was a trend toward longer survival among the treated patients (Figs. 1, 2). For the 45 patients who were treated, the median survival was 26 months, compared with 19 months for the 31 patients who were not treated (P value, not significant). Thus, treatment did not appear to have a major impact on survival or to interact with the influence of other patient characteristics on survival.
Blastic transformation of disease preceded death in 8 of 26 patients for whom the cause of death was known (31%), and all such patients were documented with myeloid phenotype. The median time from referral to blast crisis from the date of referral was 11.5 months (range, 1–34 months).
Univariate Analysis of Prognostic Factors
The results of univariate analysis are shown in Table 3, which shows 75th, 50th, and 25th percentiles of survival based on Kaplan–Meier estimates. Various cut-off values, including those that demonstrated significance in our previously published reports,22, 23, 33 were analyzed individually. Older age (> 65 years) was associated with shorter survival. The hematologic variables that showed a significant adverse association with survival were hemoglobin values < 10g/dL; leukocytosis > 50 × 109/L; absolute monocytosis (monocytes > 1.0 × 109/L); the presence of > 10% peripheral blood immature myeloid cells (IMC percentage; including blasts); and LDH > 2000 U/mL. In this series of 76 patients, the monocyte count was > 5% in only 14 patients and > 8% in only 2 patients. None of the bone marrow characteristics assessed was associated significantly with duration of survival. The survival rate was shorter for the small subgroup of 12 patients with marrow blasts ≥ 5% compared with the group of patients with blasts < 5%, although the difference was not statistically significant.
|Patient characteristic||No. of patients||Percentile of survival (months)||P value (log-rank test)|
|Platelets (× 109/L)|
|White blood cells (× 109/L)|
|≤ 50||42||18||33||125||< 0.01|
|Absolute monocytes (× 109/L)|
|Peripheral blood immature myeloid cells (%)|
|≤ 10||25||18||40||NA||< 0.01|
|Lactate dehydrogenase (U/L)|
|Bone marrow blasts (%)|
|Bone marrow monocytes (%)|
|Bone marrow lymphocytes (%)|
|Bone marrow erythroid cells (%)|
We could not demonstrate a prognostic significance of abnormal karyotype or of any specific cytogenetic aberration, even when the 73 patients with Ph negative, bcr/ablunknown status were included in the univariate analysis (for a total of 149 patients with Ph negative disease), probably because of the small number of patients with abnormal karyotypes. Leukocyte alkaline phosphatase levels, β2-microglobulin levels, and the presence of splenomegaly also lacked a significant association with shorter survival. Figure 3 shows the survival for cohorts of patients with bcr/abl negative status after stratification by age, hemoglobin level, and WBC count.
To identify any associations between individual patient characteristics, we computed correlation coefficients for pairs of patient characteristics (Table 4). The strongest positive associations found were between bone marrow blasts and peripheral blood IMC percentage and between WBC count, peripheral blood IMC percentage, absolute monocyte count, and serum LDH levels. Minimal negative associations between WBC count, hemoglobin level, and platelet count were noted.
|Covariate||WBC (× 109/L)||AMC (× 109/L)||HGB (g/dL)||PLT (× 109/L)||IMC (%)||BM BI (%)|
|Age, years (yrs)||—||—||—||—||—||—|
|WBC (× 109/L)||—||—||—||—||—||—|
|AMC (× 109/L)||0.61||—||—||—||—||—|
|HGB (g/dL)||− 0.27||− 0.20||—||—||—||—|
|PLT (× 109/L)||− 0.30||− 0.21||0.27||—||—||—|
|IMC (%)||0.34||0.20||—||− 0.25||—||—|
|BM Bl (%)||—||—||—||—||0.53||—|
|BM Ery (%)||—||—||—||—||—||0.27|
|LDH (U/L)||0.54||0.38||—||− 0.20||0.56||0.37|
A multivariate analysis (Cox regression method) of characteristics of the 76 patients with bcr/abl negative status identified the following covariates, which maintained their independent association with survival: age, hemoglobin level, and WBC count (Table 5). As shown in Table 4, these covariates had little or no correlation among themselves.
|Patient characteristic||Hazard ratio||95%CI||P value|
|Age > 65 yrs||3.98||1.81–8.72||< 0.01|
|Hemoglobin ≤ 10 g/dL||2.55||1.27–5.13||< 0.01|
|WBC > 50 × 109/L||4.21||2.04–8.68||< 0.01|
Prognostic Scoring System
To stratify patients according to their expected survival, we designed a simple scoring system based on age, hemoglobin level, and WBC count. Because the three risk factors were roughly equivalent in importance (as indicated by the proportional hazard regression analysis), we assigned equal weight to each risk factor. We assigned 1 point for each of the following variables, for a maximum total score of 3 points: age > 65 years, hemoglobin ≤ 10 g/dL, and WBC count > 50 × 109/L. This scoring system stratified patients with bcr/abl negative CML into two groups based on their level of risk: low risk (scores of 0–1 point) and high risk (scores of 2–3 points). The corresponding survival curves are shown in Figure 4. Patients in the low-risk group had a median survival of 38 months, compared with only 9 months for patients in the high-risk group (P << 0.01).
This analysis identified a well-characterized study group of patients with Ph negative, bcr/abl negative CML. They represented 2.9% of all patients with CML who were seen at The University of Texas M. D. Anderson Cancer Center between 1967 and 1999 and represented 65% of 116 patients with Ph negative CML who, beginning in 1984, were tested further for the presence of the bcr/abl molecular rearrangement. This distribution is consistent with the reported incidence of bcr/abl positive disease among patients with Ph negative CML.9, 12, 15, 19, 23, 24, 33, 51
Our findings in a large and strictly defined study group confirmed the results of our previous report23 and the results of other authors,1, 3, 5–7, 52, 53 namely, that patients with Ph negative CML have a poor prognosis. The median survival was 24 months, with only 7% of patients surviving beyond 5 years. Although information from individual small series is scarce, a recent review of the literature on patients with Ph negative CML reported that patients with bcr/abl negative CML had median survival of 25 months from the time of diagnosis.12 In the same review, the median survival of patients with Ph negative, bcr/abl positive CML was 50 months, a survival similar to that reported for patients with Ph positive CML.53–61 This finding provides further support for the notion that Ph negative, bcr/abl positive CML is indistinguishable biologically and clinically from Ph positive CML.14, 20, 32, 51, 62 In patients with Ph negative CML, we22 and others16, 24 have reported features that distinguished Ph negative CML from Ph positive CML: older age, male predominance, higher incidences of anemia and thrombocytopenia, low incidence of basophilia, and lower WBC counts. A higher incidence of monocytosis also was reported,22 but that finding may have been due to the inclusion of some patients with CMML. We were first to investigate the significance of prognostic determinants of Ph negative CML and to identify older age, low platelet count, anemia, and increased percentages of blasts and promyelocytes as the features associated adversely with survival.22 Subsequently, molecular studies further refined our understanding of the Ph negative CML entity and further restricted the definition of this already rare entity. Although several studies compared patients who had bcr/abl negative CML with patients who had bcr/abl positive CML (for reviews, see Costello et al.,12 Kurzrock et al.,26 and Aurich et al.62), the two largest series of patients with bcr/abl negative CML, which included 23 patients24 and 35 patients63, could not identify significant prognostic factors in patients with bcr/abl negative CML. In this larger study group, by multivariate analysis, we identified older age, anemia, and severe leukocytosis as variables that were associated independently with poorer prognosis. On the basis of these variables, we devised a scoring system to effectively segregate patients with bcr/abl negative CML into two groups with the greatest difference in median survival: 38 months for the low-risk group and 9 months for the high-risk group. Whether such heterogeneity of survival reflects different underlying molecular mechanisms or different stages of the disease at diagnosis remains to be clarified.
Like many previously reported studies, most patients in this series were treated periodically before and/or after referral to The University of Texas M. D. Anderson Cancer Center to control leukocytosis or, occasionally, thrombocytosis. Although the impact of such treatment on survival has not been tested rigorously, most investigators seem to believe that treatment does not affect the natural course of the disease. The results of our analysis support this conclusion, because treatment was not a covariate that was associated significantly with survival.
Two thirds of our patients had a diploid karyotype. The frequency of chromosomal abnormalities was about 30% in both the bcr/abl negative group and the bcr/abl unknown group and was similar to the frequency reported in previous studies.12, 18, 20, 24, 28 Trisomy 8 was the most common abnormality. Like patients with CMML,50, 64 but unlike patients with MDS,48, 64, 65 patients with bcr/abl negative CML rarely display complex karyotypic abnormalities. An interesting finding was the absence of monosomy 7, which is the most common abnormality in two other entities included in the World Health Organization-proposed groups of MDS/MPD: CMML50, 64, 66–68 and juvenile myelomonocytic leukemia.69, 70 Perhaps due to the low frequency of abnormal karyotypes, however, we were unable to identify a correlation between specific chromosomal abnormalities and prognosis.
Recently, dysregulated specific tyrosine kinase fusion proteins have been identified as potential causative of diseases characterized by unique balanced chromosomal translocations, in analogy with the constitutive activation of bcr/abl tyrosine kinase.71 Several such tyrosine kinase fusion proteins have been implicated in MPDs in animal models.72–74 These proteins include the product of TEL/platelet-derived growth factor receptor β (PDGFβR) in t(5;12)(q33;p13),75 HIP1/PDGFβR in t(5;7)(q33;q11.2),76 and TEL/JAK2 in t(9;12).77, 78 In addition, at least five translocations involving the fibroblast growth factor receptor 1 (located on chromosome 8p11) have been identified in patients affected by Ph negative MPDs.79–81 Because Ph negative CML/MPDs and particularly bcr/abl negative CML/MPDs are rare diseases, the frequency of such karyotypic abnormalities is unknown. In our population of 149 patients with Ph negative CML, we identified 2 patients with t(5;12)(q33;p13), both in the group of patients with unknown bcr/abl status. Translocation (5;12)(q33;p13) has been described in association with CMML,75 but we did not identify any such association in our series of 213 patients with CMML.50, 64
The causes of death among patients with bcr/abl negative CML are controversial. One study26 suggested that increased tumor load and bone marrow failure is a more frequent cause of death compared with evolution into a blastic phase, which is very common in patients with bcr/abl positive CML, whereas another study19 reported frequent transformation into the blastic phase. In our cohort of patients with bcr/abl negative CML, progression to the blastic phase was documented in 31% of patients for whom the cause of death was known. This proportion is close to the 20% incidence seen in a cohort of 213 patients with CMML.50 To date, the natural history of patients with Ph negative, bcr/abl negative CML has been defined poorly, mainly because of the rarity of these patients and, particularly in the early reports, because of the heterogeneity of the Ph negative category of CML.9–20, 26 Attempts to identify objective clinicopathologic features that unmistakably would distinguish Ph negative disease from classic CML have been unsuccessful to date.,12, 24, 26 In the current study, we also were unable to identify any features unique to patients with bcr/abl negative CML. In our series, the median age of 66 years for patients with Ph negative, bcr/abl negative CML was much higher compared with patients who had Ph positive CML (46 years in the cohort of 2805 patients at The University of Texas M. D. Anderson Cancer Center); this finding is in agreement with most reports on patients with Ph negative disease.12, 22, 23, 25, 26 We noted features consistent with both the French–American–British categories of atypical CML (aCML) and CML among our patients with bcr/abl negative CML as well as within the group of patients with Ph negative, bcr/abl unknown status. This obvious heterogeneity continues to complicate the exact definition of this group of MPDs. Until these disorders are characterized further by specific molecular or functional abnormalities, the cytogenetic/molecular definition of Ph negative, bcr/abl negative MPD/CML seems to be the best alternative, because both Ph negative CML (identified by cytogenetics) and aCML (characterized by morphology) may include patients in whom the presence of the bcr/abl gene rearrangement is detectable by RT-PCR analysis.
To date, no study has attempted to evaluate the prognostic significance of disease-associated or patient-associated variables for their power to predict survival and to develop a risk-oriented scoring system that would stratify patients according to their survival expectation. We propose a simple scoring system that can be used to categorize patients with Ph negative, bcr/abl negative CML in the clinical setting. This scoring system effectively stratifies patients into subgroups according to the expected survival time and may serve as a basis for further efforts to facilitate better clinical assessment of these patients. With new therapies on the horizon, this simple, risk-oriented classification may be valuable in the evaluation of treatment responses and impact on survival.
In conclusion, our results support the hypothesis that there is an entity of bcr/abl negative MPD distinct from bcr/abl positive CML, from MDS and (to a lesser degree) from CMML. There is a considerable heterogeneity of characteristics among these patients, and only some patients in this category fulfill the criteria for aCML.36 Clearly, the major challenge for future research that aims to improve patient outcome will be the identification of alterations in genes and signaling pathways that transmit proliferation and differentiation signals similar to those of bcr/abl in CML. Indeed, recent evidence suggests that specific tyrosine kinase fusion proteins or altered receptor tyrosine kinases probably are associated with the disease process. Therefore, the subset of patients with Ph negative, bcr/abl negative CML merits further study, because this disease entity may hold clues to the understanding of leukemogenesis.
The authors thank Kate Ó. Súilleabháin for editing this article and Anne Strickland for typing assistance.
- 21Leucémie myéloïde chronique atypique: nouvelle entité individualisée au sein des myélodysplasies. Hematol Cell Ther. 1996; 38: 136a., , , , .
- 25Ph-negative chronic granulocytic leukemia: a nonentity. Am J Clin Pathol. 1985; 85: 186–191., , .
- 32Chronic myelogenous leukemia with typical clinical and morphological features can be Philadelphia chromosome negative and “bcr negative.” Hematol Pathol. 1990; 4: 76–77., , , et al.
- 33Philadelphia chromosome-negative chronic myelogenous leukemia and chronic myelomonocytic leukemia. In: CanellosG, editor. Chronic leukemia, volume 4. Hematology oncology clinics of North America. Philadelphia: WB Saunders, 1990: 389–404., , .
- 34Essential thrombocythemia: an interim report from the Polycythemia Vera Study Group. Semin Hematol. 1982; 23: 177–182., , , .
- 36The chronic myeloid leukaemias: guidelines for distinguishing chronic granulocytic, atypical chronic myeloid, and chronic myelomonocytic leukaemia. Proposals by the French–American–British Cooperative Leukaemia Group. Br J Haematol. 1994; 87: 746–754., , , et al.
- 40MittlemanF, editor. ISCN. An International System for Human Cytogenetic Nomenclature. Basel: Karger, 1995: 1–114.
- 45Regression models and life tables (with discussion). J Stat Soc. 1972; 34: 187–220..
- 64Cytogenetics of CMML and Ph-negative CML: a retrospective analysis and a comparison with MDS. Blood. 1999a; 94 (10 Suppl 1 ): 273(1219a)., , , , .
- 72Transformation of hematopoietic cell lines to growth factor independence and induction of a fatal myelo- and lymphoproliferative disease in mice by retrovirally transduced TEL/JAK2 fusion genes. EMBO J. 1998; 17: 5321–5333., , , et al.