Characteristics and survival of BCR/ABL negative chronic myeloid leukemia: a retrospective analysis of the Surveillance, Epidemiology and End Results database

Introduction

BCR/ABL negative or atypical chronic myeloid leukemia (CML) is a rare hematologic malignancy with an estimated incidence of 1–2% of BCR/ABL positive CML. Clinical features of BCR/ABL negative CML resembles those of BCR/ABL positive CML but does not have BCR/ABL fusion gene; rearrangement of platelet-derived growth factor receptor (PDGFR)-A, PDGFR-B or FGFR1 is also absent [Vardiman et al. 2008]. BCR/ABL negative CML, however, is associated with mutations in CSF3R (up to 40%), SETBP1 (~10%) and JAK2V617F (~5%) [Gotlib et al. 2013]. KRAS or NRAS mutations may also be common in BCR/ABL negative CML (35% in a multicenter study) [Wang et al. 2014]. In addition, a study utilizing whole-exome sequencing has revealed the presence of ETNK1 mutations in 8–13% of patients with BCR/ABL negative CML [Gambacorti-Passerini et al. 2015]. World Health Organization (WHO) criteria for the diagnosis of BCR/ABL negative CML additionally include the presence of leukocytosis ⩾13 × 10⁹/l with circulating neutrophil precursors ⩾10%, monocytes <10% and minimal basophils (usually <2%); bone marrow hypercellularity with granulocytic proliferation and dysplasia; and blood and bone marrow blast count of <20% [Vardiman et al. 2008]. The rarity of the disease has largely precluded conduction of any prospective study to optimize treatment strategy of BCR/ABL negative CML. Similarly, the mutations associated with the disease were hitherto undiscovered, which prevented identification of therapeutic targets and drug development. Consequently, BCR/ABL negative CML has been managed with palliative therapy such as hydroxyurea, low-dose cytarabine and interferon, which results in a median overall survival (OS) of approximately 2 years in small retrospective studies [Onida et al. 2002; Breccia et al. 2006]. A recent multicenter study demonstrated that BCR/ABL negative CML is distinct from unclassifiable myelodysplastic/myeloproliferative neoplasms and has a worse OS (12 versus 22 months) and acute myeloid leukemia-free survival (11 versus 19 months) [Wang et al. 2014]. A population-based study on the outcomes of BCR/ABL negative CML is lacking. In this population-based study, we utilized the Surveillance, Epidemiology and End Results (SEER) database to analyze the characteristics and outcomes of BCR/ABL negative CML. Such information is useful for patient education and may provide background information for clinical trials in the future.

Methods

We utilized the SEER 18 database to identify the study cohort of interest. The SEER 18 database includes data from 18 population-based registries in 14 states from 1973–2011 and covers about 28% of the US population. This includes 26% of African Americans, 38% of Hispanics, 44% of American Indians and Alaska Natives, 50% of Asians, and 67% of Hawaiian/Pacific Islanders [Coso et al. 2006]. It classifies cancer histology and topography information on the basis of the third edition of the International Classifications of Diseases for Oncology (ICD-O-3). Cases of BCR/ABL negative CML were identified using the ICD-O-3 code 9876. According to SEER coding, BCR/ABL-negative CML requires exclusion of BCR/ABL translocation. Our search was limited to microscopically confirmed malignancies in adults of age more than 18 years with single primary malignancy, availability of follow-up data and known race, survival and marital status. The case listing version of SEERstat version 8.1.5 (31 March 2014) was used for identification of the relevant cases. Data abstracted included age at presentation, gender, race, marital status, year of diagnosis, survival period in months and vital status (dead or alive). Institutional review board waiver was obtained from the University of Nebraska Medical Center Institutional Review Board.

Statistical analyses were conducted using Statistical Package for Social Sciences (SPSS) version 22.0 (IBM Corporation, Armonk, NY). Demographic characteristics were calculated using descriptive studies. OS was calculated using the Kaplan–Meier method. The Cox proportional hazard regression model was utilized to determine the predictors of OS. All p-values were two-sided and the level of significance was chosen at 0.05.

Results

During the study period of 1973–2011, the SEER 18 database included a total of 114 cases of BCR/ABL negative CML, which accounted for 0.59% of BCR/ABL positive CML. Cases of BCR/ABL negative CML without microscopic confirmation (n = 3) or those with more than one primary malignancy (n = 29) were excluded to avoid the influence of a second malignancy on OS. Thus, the study included 82 cases of BCR/ABL negative CML for final analysis.

BCR/ABL negative CML occurred commonly in older patients (median age 69 years), men (57%) and White (81%) (Table 1). The median and 5-year OS of BCR/ABL negative CML were 15 months and 25%, respectively. In a multivariate Cox proportional hazard regression model, OS was worse with older age at diagnosis (Table 2). Compared with patients <50 years, patients >65 years [hazard ratio (HR) 5.52, 95% confidence interval (CI) 2.17–14.03, p < 0.01] were more likely to have worse OS. Gender, race, marital status and year of diagnosis did not influence OS. Patients, who were still alive beyond 5 years of follow-up, compared with those who were not, had a significantly lower median age (46 versus 70 years; p = 0.004), but similar gender distribution (females 38% versus 50%, p = 0.50).

Table 1.

Demographic characteristics of BCR/ABL negative chronic myeloid leukemia patients.

Characteristic	Number (%)
Number of cases	82
Age in years, median (range)	69 (21–92)
Age group
- ⩽50 years	12 (14.6%)
- 51–65 years	21 (25.6%)
- >65 years	49 (59.8%)
Year of diagnosis, median (range)	2006 (2001–2011)$
Year of diagnosis
- ⩽2001	9 (11.0%)
- 2002–2005	28 (34.1%)
- 2006–2011	45 (54.9%)
Gender
- Male	50 (57.0%)
- Female	32 (43.0%)
Race
- White	67 (81.7%)
- Black	8 (9.8%)
- Others*	7 (8.5%)
Marital Status
- Married	42 (51.2%)
- Others†	40 (48.8%)

^*Other race included Native American/Alaska Native, Asian or Pacific Islander.

^†Other marital status included divorced, separated, single or widowed.

^$SEER started reporting patients with BCR-ABL negative CML using ICD-O-3 code 9876/3 from 2001 onwards only. Even though the authors did not use any year restriction for relevant cases, therefore, we were able to identify patients from 2001 only.

Table 2.

Cox proportional hazard regression model for overall survival.

Category	HR	95% CI of HR	P value
Gender
- Male (ref)	1
- Female	0.94	0.58–1.54	0.82
Race
- White (ref)	1
- Black	0.84	0.39–1.80	0.66
- Others*	0.88	0.42–1.85	0.73
Marital Status
- Married (ref)	1
- Others†	0.94	0.59–1.48	0.78
Year of Diagnosis
-⩽2005	1
->2005	1.01	0.63–1.61	0.97
Age group$
- ⩽50 years (ref)	1
-50–65 years	2.57	0.97–6.81	0.06
- >65 years	5.52	2.17–14.03	<0.01

CI, confidence interval; HR, hazard ratio.

^*Other race included Native American/Alaska Native, Asian or Pacific Islander.

^†Other marital status included divorced, separated, single or widowed.

^$When the age group of >65 years was used as a reference group in the multivariate analysis, patients between 50–65 years of age had a significantly better survival, as compared to patients >65 years (HR 0.47; 95% CI 0.24–0.93; p = 0.03).

Discussion

Our study demonstrated that BCR/ABL negative CML frequently involved older patients (median age 69 years), men (57%) and White patients (81%). Prior studies have also demonstrated a median age of sixth or seventh decade of life at the time of diagnosis [Vardiman et al. 2008; Martiat et al. 1991; Kurzrock et al. 2001; Onida et al. 2002; Breccia et al. 2006]. Although a few studies demonstrated male predominance of BCR/ABL negative CML [Martiat et al. 1991; Kurzrock et al. 2001; Onida et al. 2002], others have not [Vardiman et al. 2008]. An Italian study even demonstrated female preponderance, with females accounting for 57% of all cases [Breccia et al. 2006].

The median OS of BCR/ABL negative CML was 15 months. Prior studies have demonstrated a median OS of 14–29 months [Vardiman et al. 2008; Hernandez et al. 2000; Onida et al. 2002; Breccia et al. 2006]. A few relatively smaller series have demonstrated median OS of as high as 3 years [Martiat et al. 1991; Kurzrock et al. 2001]. Given a lack of standard of care, most of the patients have been treated with hydroxyurea, low-dose cytarabine or interferon [Kurzrock et al. 2001; Breccia et al. 2006]. With such therapies, leukemic transformation is noted in up to one-third of patients [Hernandez et al. 2000; Onida et al. 2002; Breccia et al. 2006]. Even in the absence of leukemic transformation, patients may die of progressive leukocytosis or complications of cytopenia [Kurzrock et al. 2001]. Two studies have demonstrated following adverse prognostic features: age >65 years, female, patients with hemoglobin <10 g/dl, leukocyte count >50 × 10⁹/l and immature circulating precursors [Onida et al. 2002; Breccia et al. 2006]. A recent multicenter study demonstrated that elevated leukocyte count and circulating myeloid precursors were independent negative predictors in a univariate analysis; multivariate analysis could not be performed because of lack of multiple prognostic factors [Wang et al. 2014]. It is also important to note that OS may differ by treatment modality; patients treated with allogeneic stem cell transplantation may [Koldehoff et al. 2012] but not consistently [Mittal et al. 2004] have improved OS. In one study, the use of allogeneic stem cell transplantation (n = 21) resulted in a median OS of 46 months after the transplant and a 5-year OS of 80% [Koldehoff et al. 2012]. However, poor outcome was noted with allogeneic stem cell transplantation in another small study (n = 7) because of deaths from graft-versus-host disease, sepsis or other unknown causes [Mittal et al. 2004]. In our study, OS was worse with older age at diagnosis. Although we did not have data regarding the use of different therapies, it is plausible that better OS in younger patients may be in part related to a higher utilization of stem-cell transplant in younger patients. We did not identify any racial or gender disparity. Importantly, our study indicated a lack of advancement in OS of patients with BCR/ABL negative CML in recent years.

Collaborative efforts of different investigators have recently demonstrated that BCR/ABL negative CML is associated with mutations in CSF3R (up to 40%), SETBP1 (~10%) and JAK2V617F (~5%) [Gotlib et al. 2013]. These preclinical studies have also indicated a possible therapeutic role of JAK inhibitors (ruxolitinib) and SRC kinase inhibitors (dasatinib) in select patients with BCR/ABL negative CML [Gotlib et al. 2013; Maxson et al. 2013]. In addition, a BCR/ABL negative CML patient with PDGFRB rearrangement was reported to have a durable complete response to imatinib at 1-year follow-up [Garcia et al. 2003]. Identification of actionable mutations, development of novel therapies and improvement in transplant techniques have the potential to improve the outcomes of BCR/ABL negative CML.

Limitations of this study include inability to determine the effect of therapy or differences in clinical features at presentation. We also did not have data on the molecular features of BCR/ABL negative CML. Given a retrospective database study, we could not verify the accuracy of coding; however, SEER databases undergo quality monitoring and rigorous maintenance. Nonetheless, to the best of the authors’ knowledge, this is the largest study of BCR/ABL negative CML, which provides survival outcomes at a population level. Such findings are utilizable for patient education, to inform future studies and policy making.