Introduction
Chronic myeloid leukemia (CML), a disease that originates from hematopoietic stem cells, is characterized by the reciprocal translocation of chromosomes 9 and 22, also known as the Philadelphia 1 (Ph1) chromosome. The Ph1 chromosome is responsible for the formation of the
BCR-ABL fusion gene [
1,
2], and the protein encoded by
BCR-ABL has constitutive tyrosine kinase activity, which provides proliferative signals and survival advantages, and finally leads to the disease [
3,
4]. Imatinib, a selective tyrosine kinase inhibitor directly targeting BCR-ABL, has been a milestone in the history of CML treatment since its introduction in 1998. Imatinib is a classic successful example of the tailored treatment of known leukemogenesis, leading to significantly improved response rates, and more importantly, the survival of CML patients [
5]. In the IRIS trial, imatinib can induce a cumulative incidence of complete cytogenetic response (CCyR) at 5 years of 87% and an overall survival (OS) of 89% [
6]. Considering its favorable outcome, it has been used as a first-line treatment for newly diagnosed CML. Despite its success in the chronic phase (CP) of CML, its applications in the accelerated (AP) and blast (BP) phases are limited due to its significantly lower efficacy in patients in these phases. In a phase II study, imatinib only induces CCyR in 17% of CML patients in the AP [
7]. Another study shows that imatinib induces an inferior CCyR of 7% in CML patients in the BP [
8]. In the present study, we enrolled patients in the three clinical phases receiving imatinib treatment to evaluate their responses (hematologic, cytogenetic, and molecular), OS, and event-free survival (EFS). The prognostic factors related with the outcome of imatinib treatment were also examined. We presented the single-institution imatinib treatment experiences of Chinese patients with CML.
Patients and methods
Patients
A total of 275 patients with Ph1-positive CML and/or
BCR-ABL fusion CML treated with imatinib from February 2001 to April 2008 were enrolled in this study. All patients were confirmed to have the disease via bone marrow morphology, cytogenetic, and molecular detection. The three phases of CML, namely, CP, AP, and BP, were diagnosed according to the WHO criteria [
9]. Among the 275 patients, 210 started imatinib in the CP, 24 in the AP, and 41 in the BP.
Response evaluation
Complete hematologic response (CHR) was defined by a leukocyte count<10 × 109/L, a platelet count<450 × 109/L,<5% myelocytes plus metamyelocytes, no blast or promyelocyte in a blood picture, no extramedullary involvement, and no sign of the AP or blast crisis of CML. The cytogenetic response in bone marrow cells was categorized as complete (no Ph-positive metaphase, CCyR) and partial (1% to 35% Ph-positive metaphases, PCyR). A major cytogenetic response (MCyR) was defined as complete plus partial responses. Response evaluation was based on G-banding in at least 20 cells in the metaphase per sample. Complete molecular response (CMoR) was defined when the BCR-ABL transcript could not be detected, and major molecular response (MMoR) was defined by a three-log reduction in the BCR-ABL transcript.
Treatment protocol
CP patients received an imatinib does of 400 mg/d. If CHR was not achieved within 3 months or major cytogenetic response (MCyR) was not achieved within 6-12 months, the dose was increased to 600-800 mg/d. AP and BP patients received imatinib doses of 600-800 mg/d. Dose reductions to 300 mg/d were considered for patients with grade 3 or 4 neutropenia according to common toxicity criteria [
10].
This study protocol was approved by the independent ethics board. All patients gave informed consent for both treatment and cryopreservation of bone marrow and peripheral blood according to the Declaration of Helsinki.
Patient monitoring
Bone marrow morphology, cytogenetic analysis of the metaphase for Ph1 chromosome, and molecular detection of BCR-ABL were assessed before imatinib initiation and every 3 months during treatment.
End points
OS was calculated from the time of imatinib initiation to the date of death. Survival was excluded at the time when treatment was discontinued in favor of other treatment methods or at the last recorded contact or evaluation for patients who were alive at the time of analysis. EFS was defined as the time of imatinib initiation to any treatment failure, such as disease progression, discontinuation of imatinib, and death.
Detection of the BCR-ABL mutation transcript
All patients were screened for kinase domain (KD) mutations before imatinib treatment and every 3-6 months during treatment. Screening was performed more frequently when there is resistance to imatinib depending on the decision of physicians. Unscheduled screening for KD mutation was performed according to the following criteria: (1) loss of CCyR or MCyR, (2) MCyR not achieved within 12 months, and (3) progression to advanced phase. In the detection procedure, the complete ABL KD of the BCR-ABL allele was analyzed by seminested reverse transcriptase-polymerase chain reaction (PCR) followed by direct sequencing. BCR-ABL was first amplified, followed by two separate PCR reactions that covered the ABL KD, and reverse directions using Dye Terminator Chemistry and a 3700 DNA Sequencer.
Statistical analyses
Data were analyzed by SPSS 16.0 statistical software. The rate difference among groups was compared by the chi-square test. Kaplan-Meier analysis was used to calculate the distributions of the OS and EFS. Log-rank test comparison was performed to compare the survival differences. The COX model was used in the multivariate analysis of associations of the potential prognostic factors with the OS and EFS. A limited backward selection procedure was used to exclude redundant variates.
Results
Patient characteristics
A cohort of 275 patients was enrolled in this study. At the start of imatinib treatment, there were 210 patients in the CP, 24 in the AP, and 41 in the BP. The median follow-up time was 48 months for CP patients, 40 months for AP patients, and 17 months for BP patients. No patient was unaccounted for during follow-up. The demographic data of the patients are presented in Table 1.
Treatment response
All CP patients achieved CHR. Among them, 178 patients (84.7%) achieved CCyR with a median time of 9 (3-42) months, and 130 (61.9%) achieved CMoR with a median time of 12 (3-65) months. In AP patients, 23 (95.8%), 12 (50%), and 7 (29.1%) patients achieved CHR, CCyR, and CMoR, respectively. In BP patients, 26 (63.4%), 10 (24.3%), and 4 (9.7%) achieved CHR, CCyR, and CMoR, respectively. The cumulative incidence of hematologic, cytogenetic, and molecular response significantly differed among the three groups (Table 2).
Survival analysis
Fig. 1A shows the Kaplan-Meier OS of patients in the three phases. At 5 years, the median OS rates of CP and AP patients were 93.2%±2.1% and 64.5%±10.2%, respectively. For BC patients, the median OS duration was 15±1.8 months. A statistically significant difference was observed among the three phases. The OS significantly differed among the three groups (P<0.001). Similarly, the EFS also significantly differed among the three groups (P<0.001). At 5 years, the median EFS rates of CP and AP patients were 86.4%±2.7% and 50.9%±10.9%, respectively. For BC patients, the median EFS duration was 12±0.8 months (Fig. 1B). We also analyzed the reasons for EFS failure in the three groups, including disease progression (n = 17, 8.0%) and therapy change due to lack of efficacy (n = 8, 3.8%) in 25 CP patients; death (n = 1, 4.2%), disease progression after achieving CHR (n = 8, 33.3%), and therapy change due to lack of efficacy (n = 2, 8.3%) in 11 AP patients; as well as death (n = 15, 36.6%), disease progression after achieving CHR (n = 18, 43.9%), and therapy change for lack of efficacy (n = 1, 2.4%) in 34 BP patients.
Landmark analysis according to the response at 12 months
Among 210 CP patients, 205 patients were still in the CP, and 175 (76.5%) of them achieved CCyR at 12 months. Fig. 2 shows a landmark analysis of the OS and EFS according to the cytogenetic response at 12 months. The patients who achieved CCyR had better 5-year OS and EFS rates than those who did not (97.4% vs. 84.5%, and 98% vs. 68.2%, respectively; P<0.001 and P = 0.005, respectively).
We further classified patients achieving CCyR at 12 months (n = 157) according to whether they also achieved MMoR at the same time, and found that 98 (62.4%) of them achieved MMoR. Fig. 3 shows the landmark analysis of OS and EFS based on the molecular response. Patients who achieved MMoR at 12 months had a better 5-year EFS than those who did not (100% vs. 87.6%, P = 0.001). However, there was no statistical difference between the OS rates (P = 0.252).
Influence of late treatment on the outcome
Given the different durations between diagnosis and imatinib initiation, we classified the CP patients into two groups, namely, early (time interval was within 6 months, n = 120) and late (time interval was more than 6 months, n = 90) treatment groups. Fig. 4 shows the cumulative frequency of the cytogenetic and molecular responses of the two groups. The probabilities of achieving MCyR, CCyR, and CMoR for the early and late treatment groups were 95.0% and 82.2% (P = 0.003), 93.3% and 73.3% (P<0.001), and 70% and 51.1% (P = 0.005), respectively. Although there was no significant difference between the OS and EFS rates of the two groups, a tendency of improved OS and EFS was observed in the early treatment group (P = 0.071 and 0.072, respectively).
Additional chromosomal abnormalities (ACAs) in CP patients
Chromosomal abnormalities in addition to Ph1 were found in 45 out of 210 CP patients in the CP, including complex Ph (n = 7), i(17q) (n = 6), +8 (n = 6), 2Ph1 (n = 4), 9q+ (n = 4), -Y (n = 2), and others (Table 3).
There were 20 patients with ACAs before treatment. Among them, 12 (60%) achieved CCyR (4 in CCyR and 8 in CMoR) and 8 patients achieved suboptimal response (2 in PCyR and 6 in CHR). Five patients who carried ACAs of i(17q) (n = 2), 9q+ (n = 1), t(1;3) (n = 1), and complex Ph of t(1; 9; 22) (n = 1) progressed to the advanced phase with a median of 7 (6-36) months.
There were 25 (11.9%) patients who acquired additional cytogenetic abnormality apart from Ph1 during treatment. Among them, 11 developed ACAs in Ph1-positive cells (five were transient and six were persistent), and one progressed to the advanced phase with a chromosomal abnormality of t(1;10). In the remaining 14 patients, ACAs were detected in Ph1-negative cells. All 14 patients maintained stable CCyR, and none developed advanced leukemia.
We classified CP patients into three groups according to the time of ACA development: (1) patients with no ACA, (2) patients with pretreatment ACAs, and (3) patients with during-treatment ACAs. Fig. 5 shows the OS of the three groups. The estimated 5-year OS was 100% for patients who did not have ACAs, 98% for patients who had pretreatment ACAs, and 75.9% for patients with during-treatment ACAs. A statistical difference was found in OS (P = 0.001).
KD mutation of the BCR-ABL transcript
In our research, 14 kinds of KD mutations were detected in 21 patients (12 CP, 3 AP, and 6 BP). Table 4 shows the distribution of all mutations and patients’ status when mutations were detected. In 8 patients (patients 1-8), mutations were detected when they were still in a relatively stable status (7 in CHR and 1 in PCyR). In the remaining 4 patients (patients 9-12), mutations were detected at disease progression. The incidence of mutations was 5.7% (12/210) in CP patients, and the median time for detecting a mutation was 16 (6-48) months. Among AP and BP patients, mutations were detected in a much shorter time (median of 6 months), and all were proven to be resistant.
Multivariate analysis
Univariate analysis of CP patients revealed the association of some clinical and biological factors with the prognosis, including cytogenetic and molecular responses at 12 months, early or late intervention of imatinib, additional chromosome abnormalities, and KD mutations. When these parameters were used in multivariate analysis, none was proven to be an independent prognostic factor for CP patients.
Discussion
In this study, we evaluated the outcome of a cohort of 275 CML patients treated with imatinib. Most of them were in the CP. We proved again the favorable response rate of imatinib on CML with an accumulative MCyR of 89.5% and CCyR of 84.7% in CP patients, similar to the results of the IRIS trial [
6]. The CP patients had a significantly higher cytogenetic response rate than AP and BP patients, concurrent with previous research and common practice [
7,
8]. We also observed significantly higher OS and EFS in CP patients than AP and BP patients with an estimated 5-year OS rate and 5-year EFS rates of 93.2% and 86.4%, respectively, for CP patients; as well as 64.5% and 50.9%, respectively, for AP patients. We observed as well median OS and EFS durations of 15±1.825 months and 12±0.775 months, respectively, for BP patients (both
P<0.001). Alternative strategies to improve the poor outcome of advanced-stage CML are warranted in this era of imatinib.
We aimed to stratify the heterogeneity of CP patients with other important biologic and clinical factors. We initially evaluated the impact of the late treatment of imatinib on the prognosis. Patients treated late with imatinib had significantly lower cytogenetic and molecular responses than those treated early. However, this adverse impact was not observed in the OS and EFS rates of the two groups. Matsuo
et al. [
11] have reported a similar result after evaluating the OS rates of patients recently diagnosed and those previously treated. A tendency of improved OS is observed in the former group (88.7% vs. 79.8%), but there is no statistical significance. Given the deficient data of the duration between diagnosis and imatinib initiation, the results are controversial. This duration in the late treatment group is not sufficiently long to produce a meaningful difference.
Baccarani
et al. [
12], on behalf of the European Leukemia Net, published in 2006 a series of empirical recommendations designed to help clinicians identify CML-CP patients responding poorly to standard-dosage imatinib. Based on these criteria, “suboptimal response” was defined as follows: (1) less than complete hematologic response at 3 months, (2) less than partial cytogenetic response at 6 months, and (3) less than CCyR at 12 months. David Marin
et al. [
13] have subsequently reported that the suboptimal response is significantly related with poor eventual outcome. In the present study, these land markers were also proven to be very important in predicting OS and EFS. Patients who did not achieve CCyR at 12 months had inferior 5-year OS and EFS than those who did (84.5% vs. 97.4%, 68.2% vs. 98%, respectively;
P<0.001 and
P = 0.005, respectively). The results suggest that for those patients who only achieve a suboptimal response, new strategies must be considered, such as increased imatinib dose, second-generation tyrosine kinase inhibitors, and other combinations.
Occurrences of ACAs in addition to Ph1 are common cytogenetic events in CML with controversial prognostic impacts in this era of imatinib. O’Dwyer
et al. [
14] have reported that the occurrence of ACAs is an adverse prognostic factor of hematological relapse in CML patients treated with imatinib. However, they do not identify the specific ACAs that can significantly predict poor prognosis. Fabarius [
15] has found the association of some important ACAs (second Philadelphia chromosome, trisomy 8, isochromosome 17q, or trisomy 19) with inferior 5-year PFS. In the current work, we discovered that ACAs were associated with the OS of CML patients, and patients without ACAs had better OS (
P<0.001). However, this finding was not proven by multivariate analysis and should be further tested in a large cohort of patients.
Mutations have been found in each clinical phase of the disease in numerous studies, especially in the advanced phase. In the present study, the cumulative incidence of mutation in CP was 5.7% (12/210), and the median time to detect a mutation was 16 (6-48) months after imatinib initiation. In AP and BP patients, mutations were observed within a much shorter time (median of 6 months) after imatinib initiation. A large number of studies indicate that KD mutations are associated with the resistance to imatinib [
16-
18]. In particular, mutations that can prevent imatinib from binding to BCR-ABL, such as T315I, Y253H, E255K, and F317L, reportedly manifest much higher resistance to imatinib than others [
19,
20]. In the current study, these mutations were also preferentially detected in the AP or BP. New targeted agents are needed to overcome the resistance of patients who carry these mutations. Notably, some patients carried two or more types of mutations, and interestingly, some patients lost their primary mutations and acquired new mutations in the course of the disease. This phenomenon has been reported by other researchers and considered to be a result of heavy treatment [
21].
Univariate analysis proved the statistical significance of the following prognostic factors: cytogenetic and molecular responses at 12 months, early or late intervention with imatinib, ACAs, and KD mutations. However, none of them was confirmed to be independent by multivariate analysis. A bigger patient pool is necessary to examine further the values of these potential prognostic factors.
In conclusion, we established again the favorable treatment outcome of imatinib in CP-CML patients in Chinese population. In this era of imatinib, late intervention with imatinib, suboptimal response at 12 months, KD mutations, and ACAs may be potential adverse predictors of OS and EFS. However, large-sample prospective trials are needed to determine their roles.
Higher Education Press and Springer-Verlag Berlin Heidelberg
PDF
(185KB)
