• chronic myeloid leukemia;
  • population survival;
  • tyrosine kinase inhibitor;
  • age groups; bcr-abl


  1. Top of page
  2. Abstract


Outcomes for patients with chronic myeloid leukemia (CML) have improved after the advent of tyrosine kinase inhibitors (TKIs), which target the BCR/ABL fusion gene product. Nonetheless, differences in survival persist between age groups. The authors performed a retrospective cohort study using the Surveillance, Epidemiology, and End Results (SEER) database to assess 5-year overall survival (OS) in various patient age groups.


Patients who had a diagnosis of CML were identified using the SEER 19 registries database. Patients who were included had SEER diagnosis codes for CML not otherwise specified (code 9863) and BCR/ABL-positive CML (code 9875) diagnosed between January 2000 and December 2005. Patients were divided into cohorts based on age at diagnosis: ages 15 to 44 years, 45 to 64 years, 65 to 74 years, and 75 to 84 years. OS was estimated using the Kaplan-Meier method, and Cox regression was used to estimate predictors of patient survival.


In total, 5138 patients with a new CML diagnosis were identified. Five-year OS improved for all patients between the years 2000 and 2005. Compared with patients who were diagnosed in 2000, 5-year survival improved among patients ages 15 to 44 years (hazard ratio [HR] for mortality, 0.424; P < .0001), ages 45 to 64 years (HR, 0.716; P = .0315), and ages 65 to 74 years (HR, 0.692; P = .0126); and patients ages 75 to 84 years had an increased 5-year OS rate from 19.2% in 2000 to 36.4% in 2005 (HR, 0.568; P < .0001).


OS at 5 years improved among all patients, including those ages 75 to 84 years, a group with historically poor outcomes. However, older age retained an association with worse survival, suggesting opportunities for further progress. Cancer 2013;119:2620-2629. © 2013 American Cancer Society.


  1. Top of page
  2. Abstract

Chronic myeloid leukemia (CML) is distinguished by the presence of a novel fusion gene, BCR/ABL, which typically arises from a reciprocal translocation between chromosomes 9 and 22, resulting in a constitutively activated ABL tyrosine kinase.[1-3] Before the development of targeted tyrosine kinase inhibitor (TKI) therapy, survival was poor; only 40% of patients with CML ages 20 to 44 years remained alive at 5 years after diagnosis. Outcomes were significantly worse for patients aged >65 years, who had a 5-year overall survival (OS) rate of approximately only 20%.[4] These trends exhibited no significant change between 1973 and 1993.[4] Imatinib mesylate, a TKI with activity against the ABL kinase, revolutionized both the care of patients with CML and the approach to molecular targets in cancer therapies. Imatinib and other second-generation TKIs now constitute the backbone of CML treatment, and clinical outcomes for patients with CML have dramatically improved over the past 10 years.[5, 6] Nonetheless, differences in survival persist between various age groups, particularly among older patients, who have historically had poorer outcomes.[4, 7, 8]

Imatinib was approved by the US Food and Drug Administration (FDA) for use in the United States in 2001[9]; however, few studies have evaluated changes in the survival of patients with CML on the population level during this period. One such analysis examined differences in relative survival (RS) in the United States between 1980 and 2004 using the Surveillance, Epidemiology, and End Results (SEER) database.[10] During that period, there were improvements in 5-year and 10-year RS among patients up to age 65 years. However, patients aged >65 years did not realize significant improvements over this time, although they comprised the largest segment of this CML population. Another population-based study described similar improvements in RS among 3173 Swedish patients between 1973 and 2008.[5] RS improved significantly between the period from 1994 to 2000 and the period from 2001 to 2008 for patients up to age 79 years. That analysis suggested that age differences in patient outcomes occur in part because of treatment variation: patients aged >79 years had poor RS and more typically received treatment with hydroxyurea rather than a TKI. Both analyses evaluated relative, rather than overall, survival.

Data published from clinical trials have demonstrated long-term efficacy and survival benefits among elderly patients who receive imatinib mesylate.[7, 11-13] However, patients enrolled in trials may not be representative of the general population,[14-16] and significant disparities remain among the number of elderly patients who receive TKI therapy, limiting the interpretation of survival trends in clinical trials.[17, 18] Among older patients in the United States outside of clinical trials, long-term OS is less well described.

There are now mature data available within the SEER database to assess 5-year outcomes of patients who receive modern TKI therapy at the population level. The current study represents an update on CML survival outcomes using 1 of the largest databases available to assess patient populations. Rather than evaluating RS, we chose to evaluate OS to account both for mortality related to CML as well as other potentially related causes, such as toxicities of treatment regimens. We performed a retrospective cohort study of patients registered in the SEER database to estimate the 5-year OS of patients who received treatment for CML in the era of TKI therapy and to assess differences in survival outcomes, focusing particularly on changes within different age groups between 2000 and 2005.


  1. Top of page
  2. Abstract

Patients with a diagnosis of CML were identified using the SEER database (1973-2009; November 2011 submission) issued on April 16, 2012.[19] The population captured by SEER represents 28% of the US population (86.4 million individuals) based on the 2010 US Census and includes incident malignancies occurring in 19 cancer registries across the United States: Alaska, Atlanta (Georgia), California, Connecticut, Detroit (Michigan), Hawaii, Iowa, New Mexico, Rural Georgia, San Francisco-Oakland (California), San Jose-Monterey (California), Seattle (Washington), Utah, Kentucky, Los Angeles, Louisiana (including areas impacted by Hurricane Katrina), New Jersey, and Greater Georgia. Patient vital status, date of last contact, and recurrences are updated to maintain accurate surveillance information. The SEER database has a case ascertainment standard of 98% and is regarded as an accepted standard for population study in the United States.[20]

Patient Selection

CML diagnoses were identified in SEER using version 3 of the International Classification of Diseases for Oncology.[21] We included all patients who were diagnosed during the study period who carried a SEER diagnosis code of CML not otherwise specified (code 9863) or BCR/ABL-positive CML (code 9875). The SEER data set includes a recode for CML (code 35022) to account for variations in diagnostic criteria over time; however, to reflect modern classification of patients with CML, we excluded patients with a diagnosis code of BCR/ABL-negative CML (code 9876) as well as patients with a diagnosis of chronic myelomonocytic leukemia (CMML) not otherwise specified or juvenile CMML (codes 9945 and 9946), who would otherwise be included within the recoded data. We included patients diagnosed between January 2000 and December 2005. We selected this time interval because it brackets the FDA approval of imatinib in 2001[9] and the incorporation of imatinib into the National Comprehensive Cancer Network (NCCN) guidelines in 2002.[22] The SEER database records various disease characteristics, patient sex, race and ethnicity, state and county of residence, method of diagnostic confirmation, age at diagnosis, the month and year of diagnosis, and the month and year of last follow-up or of death, which is linked to National Death Index data from the National Center for Health Statistics. Patient race and ethnicity are recorded according to federal guidelines and those of the North American Association of Central Cancer Registries. Cause of death was recorded according to the 10th edition of the International Classification of Diseases.[21] In this manner, survival and cause of death can be assessed. Because we were interested in survival trends among various age groups, we chose to investigate all-cause mortality. RS was determined using population survival estimates from the year 2000 US census data.[23] On the basis of the state and county, each patient was designated as an urban (metropolitan) or rural (nonmetropolitan) dweller according to the 2003 rural-urban continuity codes within the SEER data set. These, in turn, are based on Office of Management and Budget (OMB) metropolitan area delineations for the 2000 Census.[24, 25]

Statistical Analysis

Descriptive statistics were used to describe patient baseline characteristics. Patients were divided into cohorts based on their recorded age at the time of diagnosis: ages 15 to 44 years, 45 to 64 years, 65 to 74 years, and 75 to 84 years. The Kruskal-Wallis test was used to compare the median age at diagnosis among the 6 years of observation. After confirming the proportional hazards assumption, univariate Cox proportional hazards regression was used within each age cohort to estimate the hazard ratio (HR) for mortality based on the year of diagnosis. We then performed multivariable Cox proportional hazards regression within each age cohort using patient year of diagnosis, sex, race/ethnicity, and rural or urban residence. OS was calculated from the date of diagnosis to the date of death (from any cause). The censored follow-up time for patients without recorded death was the date of last contact/follow-up. Patients were divided into subgroups based on age and year of diagnosis, and OS for each subgroup was estimated using the method of Kaplan and Meier. OS was also assessed by creating era cohorts that encompassed patients diagnosed between 2000 to 2002 and between 2003 to 2005, the latter after the NCCN incorporation of TKI therapy. RS was defined as observed survival in our patient cohort (in which all deaths were considered events) divided by the expected survival of a comparable cohort from the general population, which was assumed to be free from CML. Expected survival was estimated using the Ederer II method from the US population life tables stratified by age, sex, race and ethnicity, and calendar period. All analyses were performed using SAS 9.2 statistical software (SAS Institute, Inc., Cary, NC); SEER*Stat software (version 8.0.1; Surveillance Research Program, National Cancer Institute, Bethesda, Md) was used for the RS analysis.


  1. Top of page
  2. Abstract

We identified 5138 patients registered in the SEER database with a new diagnosis of CML between January 2000 and December 2005.

Patient Characteristics

Fifty-eight percent of the identified patients were male (Table 1). The median age at diagnosis of the selected subset (ages 15-84 years) was 58 years (interquartile range, 44-72 years), and 39% of patients were aged ≥65 years. There was no difference in median age by year of diagnosis (range, 57-60 years; P = .18) (Tables 2 and 3). This was the first recorded primary malignancy for 89% of the cohort. A diagnosis of CML was made most commonly by histology (91%) (Table 1). Although the SEER database draws from across the United States, there is a large proportion of patients with CML drawn from registries within the state of California (39% of all patients included). Patients were mostly urban dwellers (87% of patients). The majority of patients were white non-Hispanic (69%); black and Hispanic patients represented 11% each, and 9% of patients were other races or ethnicities.

Table 1. Patient Demographic Informationa
CharacteristicNo. of Patients (%)
  1. Abbreviations: CML, chronic myeloid leukemia; SMSA, standard metropolitan statistical area.

  2. a

    Classifications shown are from Surveillance, Epidemiology, and End Results population data.

Age at diagnosis, y 
15-441356 (27)
45-641766 (34)
65-74976 (19)
75-841040 (20)
Male2959 (58)
Female2179 (42)
White non-Hispanic3541 (69)
Black569 (11)
White Hispanic568 (11)
Other: American Indian, Asian/Pacific Islander, other460 (9)
San Francisco-Oakland SMSA239 (5)
Connecticut193 (4)
Metropolitan Detroit349 (7)
Hawaii75 (1)
Iowa235 (5)
New Mexico113 (2)
Seattle, Puget Sound280 (5)
Utah117 (2)
Metropolitan Atlanta132 (3)
Alaska9 (0.2)
San Jose-Monterey124 (2)
Los Angeles524 (10)
Rural Georgia11 (0.2)
Greater California, excluding San Francisco, Los Angeles, and San Jose1,126 (22)
Kentucky284 (6)
Louisiana, including Katrina-affected areas320 (6)
New Jersey654 (13)
Greater Georgia, excluding Atlanta and rural Georgia353 (7)
Rural/urban residence 
Urban, metropolitan counties4486 (87)
Rural, nonmetropolitan counties652 (13)
Method of diagnostic confirmation 
Positive histology4701 (91)
Positive cytology81 (2)
Positive microscopic confirmation, method not specified14 (0.3)
Positive laboratory test/marker study134 (3)
Radiology and other imaging techniques without microscopic confirmation1 (<0.1)
Clinical diagnosis only, other than above47 (1)
Unknown whether microscopically confirmed; death certificate only160 (3)
No. of prior malignancies 
CML is the only identified malignancy4194 (82)
CML is first of 2 or more primaries348 (7)
Table 2. Patient Age at Diagnosis
 Year of Diagnosis 
Age at Diagnosisa200020012002200320042005Total
  1. a

    The median age at diagnosis between years was compared using the Kruskal-Wallis test for nonparametric comparison; there was no significant difference in median age of diagnosis between years (P = .1829).

No. of patients9318917588198768635138
Median age, y58596059565758
Mean age, y57.4357.2957.5857.2456.1456.0856.95
Age range, y15-8415-8416-8416-8415-8415-8415-84
Table 3. Survival by Age Group: All Years
Age at DiagnosisNo. of Patients (%)HR for MortalityaHR Lower LimitHR Upper LimitP
  1. Abbreviations: HR, hazard ratio.

  2. a

    Cox regression was used to determine the HRs for mortality based on patient age group. HRs are relative to the group ages 15 to 44 years; a lower value indicates a survival benefit.

Ages 15-44 y1356 (26.4)1.00   
Ages 45-64 y1766 (34.4)1.561.3521.8< .0001
Ages 65-74 y976 (19)3.7563.2574.332< .0001
Ages 75-84 y1040 (20.2)6.9226.0447.927< .0001

Patient Survival

Five-year OS improved among patients in every age group between the years 2000 and 2005 (Table 4, Fig. 1, top). Compared with patients who were diagnosed in the year 2000, patients diagnosed in 2005 between ages 15 and 44 years had the greatest improvement in 5-year OS. This age group had an improvement from 71.6% to 86.4% alive (Fig. 2) (HR for mortality, 0.424; 95% confidence interval (CI), 0.275-0.654; P = .0001). It is noteworthy that patients ages 75 to 84 years also had significant gains in absolute 5-year survival during this period. The percentage of patients ages 75 to 84 years at CML diagnosis who remained alive at 5 years increased from 19.2% in 2000 to 36.4% in 2005 (Fig. 2) (HR for mortality, 0.568; 95% CI, 0.441-0.734; P < .0001). There were similar, although relatively smaller, improvements in absolute OS rates among patients ages 45 to 64 years (from 67.5% to 76.3% remaining alive; HR for mortality, 0.716; 95% CI, 0.528-0.971; P = .0315) and ages 65 to 74 years (from 38.1% to 51.2% remaining alive; HR for mortality, 0.692; 95% CI, 0.518-0.924; P = .0126) (Tables 4 and 5). To assess the potential impact of incorporating TKIs into CML therapy, we also determined patient survival using era cohorts; these results also reflected an improvement in OS for the cohort diagnosed between 2003 and 2005 compared with those diagnosed between 2000 and 2002, the latter era being before the NCCN incorporation of recommendations for TKI therapy (Table 4).


Figure 1. Estimated rates of (Top) overall survival (OS) (the proportion of patients who remained alive) at 5 years and (Bottom) relative survival (RS) at 5 years are illustrated according to year of diagnosis. There was an improvement in 5-year OS and 5-year RS between the years 2000 and 2005. RS is based on a comparison with the general population, which is represented as “1.” [Color figure can be viewed in the online issue, which is available at]

Download figure to PowerPoint


Figure 2. These charts illustrate Kaplan-Meier (KM) estimates of overall survival (OS) by year of diagnosis and patient age at the time of diagnosis. Patients were divided into age groups (ages 15-44 years, 45-64 years, 65-74 years, and 75-84 years). Survival curves were calculated based on the year of patient diagnosis. There was an improvement in survival by year of diagnosis in each age group, although survival was shorter among older patients. [Color figure can be viewed in the online issue, which is available at]

Download figure to PowerPoint

Table 4. Kaplan-Meier Estimates of 5-Year Overall Survival and 5-Year Relative Survival by Year of Diagnosisa
  Year of Diagnosis
Age Group, yTotal No.2000200120022003200420052000-20022003-2005
  1. Abbreviations: OS, overall survival; RS, relative survival.

  2. a

    Survival estimates are provided by age group and year of diagnosis. In addition, Kaplan-Meier survival estimates at 5-years are indicated for era cohorts composed of patients diagnosed between 2000 and 2002 and between 2003 and 2005.

  3. b

    This statistic could not be calculated.

5-Year OS estimate: All patients         
Median follow-up, y  
5-Year RS estimates: All patients         
65-749760.4390.440.5870.5740.665 0.4820.614
Table 5. Multivariable Cox Proportional Hazards Regression for Overall Survival by Year of Diagnosisa
 Ages 15-44 YearsAges 45-64 YearsAges 65-74 YearsAges 75-84 Years
Year of DiagnosisNo.HR [CI]PNo.HR [CI]PNo.HR [CI]PNo.HR [CI]P
  1. Abbreviations: CI, confidence interval; HR, hazard ratio.

  2. a

    Cox regressions were performed within each age group to determine the HRs for mortality based on year of diagnosis for patients. HRs are relative to the year 2000; a lower value indicates a survival benefit.

20002381.00 3241.00 1761.00 1931.00 
20012310.677[0.48-0.953].02563000.895 [0.696-1.15].3863174 1.002 [0.782-1.285].98641860.86 [0.696-1.063].1635
20022010.689 [0.476-0.996].04762410.871 [0.661-1.15].33031370.706 [0.53-0.94].01711790.857 [0.689-1.066].1648
20032110.604 [0.411-0.888].01032790.772[0.584-1.021].06991500.661 [0.497-0.878].00431790.617 [0.491-0.775]<.0001
20042320.433 [0.285-0.657]<.00013200.629[0.469-0.844].0021750.578 [0.434-0.769].00021490.77 [0.605-0.979].0329
20052430.424 [0.275-0.654].00013020.716 [0.528-0.971].03151640.692 [0.518-0.924].01261540.568 [0.44-0.734]<.0001

Because population-based cancer survival is often described in terms of RS rather than absolute survival,[5, 10, 26] we also performed an RS analysis (Fig. 1, bottom). Improvements in RS were consistent with previously reported data;[5, 10, 11] however, RS estimates were higher than OS estimates among older patients. In 2004, the 5-year RS percentage among patients ages 15 to 44 years was 88.4% compared with an OS rate of 87.6%. That same year, for patients ages 75 to 84 years, the RS was 39.5% compared with an OS rate of 28% (Table 4). For the entire CML population ages 15 to 84 years, 56.7% of patients were alive at 5 years, 15.7% of all patients died because of CML (Tables 6 and 7), and 11.6% of patients died of nonmalignant causes, most commonly from cardiac disease (4%) and chronic obstructive pulmonary disease (1.2%). CML was the most common cause of death for patients ages 15 to 64 years but not for those ages 65 to 84 years (Tables 6 and 7).

Table 6. Cause of Death by Patient Age Cohort
 Age Cohort: No. of Patients (%)a
CODAges 15-44 YearsAges 45-64 YearsAges 65-74 YearsAges 75-84 YearsAll Age Groups
  1. Abbreviations: CML, chronic myeloid leukemia; COD, cause of death.

  2. a

    Percentages indicate percent of total deaths, except in the case of total no. alive, which indicates percent of all patients in that age group.

CML173 (60.7)234 (42.6)160 (28.4)237 (28.6)804 (36.1)
Other malignancy67 (23.5)190 (34.6)218 (38.7)303 (36.6)778 (34.9)
Nonmalignant COD45 (15.8)125 (22.8)186 (33)289 (34.9)645 (29)
Total no. alive1071 (79)1217 (68.9)412 (42.2)211 (20.3)2911 (56.7)
Total no. of patients1356176697610405138
Table 7. Patient Disposition According Year of Chronic Myeloid Leukemia Diagnosis
 Year of CML Diagnosis: No. of Patients
Patient Disposition200020012002200320042005All Years
  1. Abbreviations: CML, chronic myeloid leukemia; COD, cause of death; COPD, chronic obstructive pulmonary disease.

  2. a

    Deaths in these patients were attributed to cardiac disease. This is a subgroup of the nonmalignant COD total (above).

  3. b

    Deaths in these patients were attributed to COPD. This is a subgroup of the nonmalignant COD total (above).

Died from CML2211721271039487804
Died from other malignancy162149142116105104778
Died from nonmalignant COD137144871149172645
Cardiac subgroupa485123283221203
COPD subgroupb91213125859
Total no. of patients9318917588198768635138

Predictors of Survival

Other variables associated with patient survival included patient sex, race, and ethnicity (Table 8). Female sex was associated with a decreased HR for mortality among patients ages 45 to 84 years (HR, 0.767-0.865; P < .0389). This was not significant when combining all age groups (P = .0653). Black race was associated with an increased HR for mortality compared with white race among younger patients ages 15 to 64 years (HR, 1.573-1.723; P < .0004), but not across all ages combined (P = .8591). Hispanic ethnicity, however, as well as other races or ethnicities (primarily American Indian and Asian/Pacific Islander), were associated with improved OS at 5-years compared with white race (HR for mortality, 0.612 [P < .0001] and 0.58 [P < .0001], respectively). There was no effect of rural versus urban residence on OS (P = .1027). In the SEER patient population, age at diagnosis remained a strong predictor of OS; compared with patients ages 15 to 44 years, there were increasing HRs for mortality among patients ages 45 to 64 years (HR, 1.56; P < .0001), ages 65 to 74 years (HR, 3.76; P < .0001), and ages 75 to 84 years (HR, 6.92; P < .0001) (Table 3).

Table 8. Predictors of Survival by Multivariable Cox Proportional Hazards Regression Within Each Age Group
VariableNo. of Patients (%)HRaHR Lower LimitHR Upper LimitP
  1. Abbreviations: AI, American Indian; API, Asian/Pacific Islander; HR, hazard ratio.

  2. a

    HRs are relative to the index variable and indicate the relative hazard for mortality; a lower value indicates a survival benefit.

Ages 15-44 y     
Male834 (61.5)1.00   
Female522 (38.5)0.8020.6281.025.0776
White non-Hispanic706 (52.1)1.00   
Black209 (15.4)1.7231.2722.333.0004
White Hispanic268 (19.8)1.2620.9241.722.143
Other: AI, API, other173 (12.8)0.9370.6291.395.748
Area of residence     
Metropolitan county1221 (90)1.00   
Nonmetropolitan county135 (10)1.2530.8561.833.2461
Ages 45-64 y     
Male1020 (57.8)    
Female746 (42.2)0.7670.6450.913.0028
White non-Hispanic1203 (68.1)    
Black205 (11.6)1.5731.2461.985.0001
White Hispanic190 (10.8)1.0020.7511.336.9901
Other: AI, API, other168 (9.5)0.8040.5821.112.1872
Area of residence     
Metropolitan county1526 (86.4)    
Nonmetropolitan county240 (13.6)0.9170.7111.184.5074
Ages 65-74 y     
Male568 (58.2)    
Female408 (41.8)0.8260.6970.979.0275
White non-Hispanic741 (75.9)    
Black93 (9.5)1.2590.9671.639.0867
White Hispanic68 (7)0.6220.420.92.0173
Other: AI, API, other74 (7.6)0.7440.5241.056.0978
Area of residence     
Metropolitan county839 (86)    
Nonmetropolitan county137 (14)1.0670.8451.347.5856
Ages 75-84 y     
Male537 (51.6)    
Female503 (48.4)0.8650.7540.993.0389
White non-Hispanic891 (85.7)    
Black62 (6)1.0940.8221.455.5379
White Hispanic42 (4)1.1350.8061.599.4685
Other (AI, API, other)45 (4.3)0.8180.5691.175.2768
Area of residence     
Metropolitan county900 (86.5)    
Nonmetropolitan county140 (13.5)1.1160.9151.36.2786
All Years: 2000-2005     
Male2959 (57.6)1.00   
Female2179 (42.4)0.9240.8491.005.0653
White non-Hispanic3541 (68.9)1.00   
African American569 (11.1)0.9880.8681.125.8591
White Hispanic568 (11.1)0.6120.5230.715<.0001
Other: AI, API, other460 (8.9)0.580.4870.691<.0001
Area of residence     
Metropolitan county4486 (87.3)1.00   
Nonmetropolitan county652 (12.7)1.1060.981.249.1036


  1. Top of page
  2. Abstract

TKIs that target BCR/ABL have been approved by the FDA for the treatment of CML since 2001 and are highly effective therapies for this disease. They have drastically altered the approach to CML; and, since their advent, patient outcomes, as reported through clinical trials, have improved markedly.[7, 11-13] Nonetheless, there are few studies evaluating patient survival in the modern era at a population level. Two of the largest such studies—1 from the SEER database with data up to the year 2004 and another evaluating outcomes in Sweden until 2008—documented improved patient survival after the year 2001, although not among elderly patients aged >65 years and ≥79 years, respectively.[5, 10] In the current study, we evaluated CML OS rates using a large population cohort, the SEER database, and a modern definition of CML. We observed significant 5-year survival gains for all patients with CML, most markedly among patients ages 15 to 44 years, but also a near doubling in the 5-year survival rate for those ages 75 to 84 years: a finding not previously described in data from years preceding the widespread use of TKI therapies.

Previously, Brenner et al used the SEER database to evaluate RS in CML between 1980 and 2004, with a period analysis to calculate RS estimates.[10] SEER definitions of CML have varied over time, and earlier recodes of CML classifications include patients with CMML and juvenile CMML. The current classification of CML would exclude these definitions, as we chose to do in the current study. This is an important distinction, because CMML and juvenile CMML do not carry the characteristic BCR/ABL mutation, and the management and outcomes of these patients are quite distinct from those for patients with CML. In addition, our current study provides an update of SEER outcomes with longer patient follow-up since the introduction of TKIs.

The other recent, population-based study from Sweden reported improved RS during the period from 2001 to 2008 compared with the period from 1994 to 2000, although patients aged >79 years continued to do poorly. A strength of that study was its linkage to prescription information, allowing some delineation of the treatment prescribed. From these data, the authors suggested that TKI treatment variability was responsible for the poor patient outcomes: only 18% of patients aged ≥80 years in their study were prescribed a TKI.[5] In contrast to both of those studies, our analysis focused on OS rates—an indicator of all-cause mortality that we believe provides a more clinically relevant measure of patient outcomes. OS captures not only death from CML but also death that may be related to effects of treatment or other causes of death that are observed at increased frequency among older patients. In this study, we calculated RS rates as a comparison; particularly among older patients, RS yielded higher survival estimates than the OS rates.

When performing our analysis, we specifically examined survival rates among different patient age groups. Historically, prognostic models in CML have included advanced patient age as a poor prognostic indicator;[27, 28] however, after the introduction of imatinib, the significance of age as a negative factor is less clear.[12, 17, 29, 30] Imatinib and second-generation TKIs—dasatinib and nilotinib—have demonstrated long-term efficacy with reasonable side-effect profiles in several clinical trials.[8, 11] A subgroup analysis of the Italian Group for Hematologic Diseases in Adults (GIMEMA) study of CML patients in late chronic phase who failed interferon therapy examined the outcomes of elderly patients (aged >65 years) and observed no difference in survival compared with younger patients (91% remained alive in each group at 4 years), although elderly patients had more adverse events.[7] In addition, several single-institution series have demonstrated success when treating elderly patients with imatinib mesylate. A series of 351 patients with Philadelphia chromosome-positive CML who received imatinib at The University of Texas M. D. Anderson Cancer Center, including 49 patients who were aged >60 years, did not report any significant relation between age and OS.[30] A larger series of 3548 patients who were treated between 1965 and 2010 at the same institution corroborated this finding, demonstrating that the correlation between age and survival diminished significantly in the period after 2001, when TKIs became the predominant first-line therapy for chronic phase CML.[29] Another series from Italy described 117 patients who received imatinib as first-line therapy in early chronic phase, including 40 patients aged >65 years. They reported an OS rate of 82% at 5 years for these patients, which was not significantly different from the rate among younger patients.[31]

Our current analysis demonstrates improved OS for all age groups over time; however, unlike the studies described above, the effect of age remained a highly significant predictor of overall patient survival within the general population, suggesting that some of the success documented in these trials has not been fully translated at the population level. Discrepancies have been reported between outcomes reported in clinical trials based predominantly at referral centers and the outcomes observed in the general population, and some have reported up to a 50% difference in survival rates.[14] There are several possible explanations for these differences, including those of younger age and better performance status, as well as fewer comorbidities among trial participants.[15]

In addition to age, there were several other predictors of OS. Black race was associated with poorer survival among patients ages 15 to 64 years. Although a survival difference based on patient race or ethnicity was not the primary outcome of our study, this is an area for potential future study. In particular, there are data suggesting disparities between patients of different race or ethnicity with chronic leukemia undergoing transplantation,[32, 33] a treatment strategy that has changed in light of TKI therapy. Patients who were Hispanic, American Indian, or Asian/Pacific Islander appeared to have better outcomes compared with both white non-Hispanic and black patients; this was a somewhat unexpected finding and merits future investigation. We also noted a survival benefit for women compared with men between ages 45 and 84 years, a finding that has been noted previously,[34] although the published data are limited. Overall, CML was the most common cause of death among patients ages 15 to 64 years, whereas other causes of death were relatively more prevalent among patients ages 65 to 84 years.

There are limitations to studies performed using the SEER database. In particular, the SEER database did not contain prescription information during our analysis period; therefore, we cannot make direct links between patient outcomes and treatment regimens. There is a linked SEER-Medicare database, which contains oral medication prescription information from Medicare Part D, but this benefit was not in existence before 2006. A previous analysis performed a record review of 423 patients within the SEER database from 2003 and, from this small selection determined that, particularly among the elderly, age disparities in receipt of imatinib likely contributed heavily to survival differences.[17] It is reasonable to conclude that lower use of TKIs contributed to worsened RS and OS in the elderly in our study compared with other age groups, particularly because the data set spans the year 2003. Ideally, this analysis would assess survival after initiation of therapy, and not at diagnosis, but these data are not available through SEER. This would have the advantage of making any findings more specific to an intervention and would demonstrate effects from either the initiation or the delay of treatment on outcomes; without this information, we conclude that our findings relate to treatment with TKIs.

The population captured by SEER is representative of the US population with regard to measures of poverty and education but tends to be more urban and has a higher percentage of foreign-born individuals (17.9% vs 12.8%).[35] It also includes a larger proportion of the total US American Indian/Alaska Native (43.8%), Asian (50.4%), and Hawaiian/Pacific Islander (66.5%) populations compared with the total US white (24.9%) and black (25.6%) populations, which may diminish the generalizability of our findings. Furthermore, our data reflect trends in CML survival only in those geographic regions contained within the SEER database and may not reflect other areas of the United States. Nonetheless, given the large number of patients included in SEER and the high level of case ascertainment, the current study encompasses 1 of the largest and most comprehensive analyses of current CML outcomes.

Our data demonstrate a marked improvement in survival among all patients with CML over the last decade, corresponding with the period after the advent of BCR/ABL-targeted TKIs. In particular, although younger patients had the greatest survival gains, we observed marked improvements in OS at 5 years among patients ages 75 to 84 years, a group that historically has had very poor outcomes. Our data, in the largest such analysis to date, are consistent with other work suggesting the beneficial impact of BCR/ABL TKIs on the outcomes of older patients with CML, likely by providing a tolerable and effective treatment option not previously available. Nonetheless, elderly patients continue to have worse outcomes on the population level compared with outcomes from institutional series or clinical trials; suggesting that TKI therapy may continue to be underused in the elderly. Our results may also suggest an effect of the increasing prevalence of comorbidities with older age in the general population, which is underrepresented in clinical trial participants. Further study is needed to determine specific factors that have contributed to the significant improvements in survival demonstrated here. In the future, all patients, including those in older age groups, are likely to experience ongoing benefit from novel and effective therapeutics with tolerable side effect profiles.


  1. Top of page
  2. Abstract
  • 1
    De Klein A, Van Kessel AG, Grosveld G, et al. A cellular oncogene is translocated to the Philadelphia chromosome in chronic myelocytic leukaemia. Nature. 1982;300:765-767.
  • 2
    Schaefer-Rego K, Dudek H, Popenoe D, et al. CML patients in blast crisis have breakpoints localized to a specific region of the BCR. Blood. 1987;70:448-455.
  • 3
    Daley GQ, Van Etten RA, Baltimore D. Induction of chronic myelogenous leukemia in mice by the P210bcr/abl gene of the Philadelphia chromosome. Science. 1990;247:824-830.
  • 4
    Xie Y, Davies SM, Xiang Y, Robison LL, Ross JA. Trends in leukemia incidence and survival in the United States (1973-1998). Cancer. 2003;97:2229-2235.
  • 5
    Bjorkholm M, Ohm L, Eloranta S, et al. Success story of targeted therapy in chronic myeloid leukemia: a population-based study of patients diagnosed in Sweden from 1973 to 2008. J Clin Oncol. 2011;29:2514-2520.
  • 6
    Druker BJ, Sawyers CL, Kantarjian H, et al. Activity of a specific inhibitor of the BCR-ABL tyrosine kinase in the blast crisis of chronic myeloid leukemia and acute lymphoblastic leukemia with the Philadelphia chromosome. N Engl J Med. 2001;344:1038-1042.
  • 7
    Rosti G, Iacobucci I, Bassi S, et al. Impact of age on the outcome of patients with chronic myeloid leukemia in late chronic phase: results of a phase II study of the GIMEMA CML Working Party. Haematologica. 2007;92:101-105.
  • 8
    Breccia M, Tiribelli M, Alimena G. Tyrosine kinase inhibitors for elderly chronic myeloid leukemia patients: a systematic review of efficacy and safety data. Crit Rev Oncol Hematol. 2012;84:93-100.
  • 9
    Cohen MH, Williams G, Johnson JR, et al. Approval summary for imatinib mesylate capsules in the treatment of chronic myelogenous leukemia. Clin Cancer Res. 2002;8:935-942.
  • 10
    Brenner H, Gondos A, Pulte D. Recent trends in long-term survival of patients with chronic myelocytic leukemia: disclosing the impact of advances in therapy on the population level. Haematologica. 2008;93:1544-1549.
  • 11
    Druker BJ, Guilhot F, O'Brien SG, et al. Five-year follow-up of patients receiving imatinib for chronic myeloid leukemia. N Engl J Med. 2006;355:2408-2417.
  • 12
    Gugliotta G, Castagnetti F, Palandri F, et al. Frontline imatinib treatment of chronic myeloid leukemia: no impact of age on outcome, a survey by the GIMEMA CML Working Party. Blood. 2011;117:5591-5599.
  • 13
    Rousselot P, Cony-Makhoul P, Nicolini F, et al. French Intergroup for Chronic Myelogenous Leukemia (Fi-LMC). Long-term safety and efficacy of imatinib mesylate (Gleevec) in elderly patients with chronic phase chronic myelogenous leukemia: results of the AFR04 study. Am J Hematol. 2013;88:1-4.
  • 14
    Pulte D, Gondos A, Redaniel MT, Brenner H. Survival of patients with chronic myelocytic leukemia: comparisons of estimates from clinical trial settings and population-based cancer registries. Oncologist. 2011;16:663-671.
  • 15
    Layde PM, Broste SK, Desbiens N, et al. Generalizability of clinical studies conducted at tertiary care medical centers: a population-based analysis. J Clin Epidemiol. 1996;49:835-841.
  • 16
    Begg CB, Engstrom PF. Eligibility and extrapolation in cancer clinical trials. J Clin Oncol. 1987;5:962-968.
  • 17
    Wiggins CL, Harlan LC, Nelson HE, et al. Age disparity in the dissemination of imatinib for treating chronic myeloid leukemia. Am J Med. 2010;123:764.e1-764.e9.
  • 18
    Seiter K. Considerations in the management of elderly patients with chronic myeloid leukemia. Clin Lymphoma Myeloma Leuk. 2012;12:12-19.
  • 19
    National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch. Surveillance, Epidemiology, and End Results (SEER) Program Research Data (1973-2009), released April 2012, based on the November 2011 submission. 2012. Available at: Accessed October 16, 2012.
  • 20
    Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin. 2012;62:10-29.
  • 21
    Percy C, Fritz A, Jack A, et al. eds. International Classification of Diseases for Oncology. 3rd ed. Geneva, Switzerland: World Health Organization; 2000.
  • 22
    Susman E. New NCCN CML guidelines incorporate use of imatinib. Oncol Times. 2002;24:54-55.
  • 23
    Arias E. United States life tables, 2000. Natl Vital Stat Rep. 2002;51:1-39.
  • 24
    United States Census Bureau. Census 2000 Gateway. Available at: Accessed October 21, 2012.
  • 25
    United States Department of Agriculture Economic Research Service. 2003 Rural-Urban Continuum Codes. Available at: Accessed November 16, 2012.
  • 26
    Henson DE, Ries LA. The relative survival rate. Cancer. 1995;76:1687-1688.
  • 27
    Kantarjian HM, Smith TL, McCredie KB, et al. Chronic myelogenous leukemia: a multivariate analysis of the associations of patient characteristics and therapy with survival. Blood. 1985;66:1326-1335.
  • 28
    Hasford J, Pfirrmann M, Hehlmann R, et al. A new prognostic score for survival of patients with chronic myeloid leukemia treated with interferon alfa. Writing Committee for the Collaborative CML Prognostic Factors Project Group. J Natl Cancer Inst. 1998;90:850-859.
  • 29
    Kantarjian H, O'Brien S, Jabbour E, et al. Improved survival in chronic myeloid leukemia since the introduction of imatinib therapy: a single-institution historical experience. Blood. 2012;119:1981-1987.
  • 30
    Cortes J, Talpaz M, O'Brien S, et al. Effects of age on prognosis with imatinib mesylate therapy for patients with Philadelphia chromosome-positive chronic myelogenous leukemia. Cancer. 2003;98:1105-1113.
  • 31
    Latagliata R, Breccia M, Carmosino I, et al. “Real-life” results of front-line treatment with Imatinib in older patients (≥65 years) with newly diagnosed chronic myelogenous leukemia. Leuk Res. 2010;34:1472-1475.
  • 32
    Baker KS, Davies SM, Majhail NS, et al. Race and socioeconomic status influence outcomes of unrelated donor hematopoietic cell transplantation. Biol Blood Marrow Transplant. 2009;15:1543-1554.
  • 33
    Majhail NS, Nayyar S, Santibanez MEB, Murphy EA, Denzen EM. Racial disparities in hematopoietic cell transplantation in the United States. Bone Marrow Transplant. 2012;47:1385-1390.
  • 34
    Berger U, Maywald O, Pfirrmann M, et al. Gender aspects in chronic myeloid leukemia: long-term results from randomized studies. Leukemia. 2005;19:984-989.
  • 35
    Surveillance, Epidemiology, and End Results (SEER) Program. Population Characteristics: Characteristics of the SEER Population Compared with the Total United States Population. Available at: Accessed October 21, 2012.