Introduction
Chronic myeloid leukemia (CML), which is characterized by the presence of the Philadelphia chromosome (Ph +) and the resultant
BCR-ABL1 fusion, occurs in different phases dependent on the absence or presence of certain clinical features, including cytogenetic clonal evolution, basophilia, and elevated blast percentage [
1]. CML presents in a chronic phase (CP) in about 95% of patients [
2]. Approximately 5% of patients present with an advanced phase of disease, either in an accelerated (AP) or blast phase (BP) [
2]. The natural history of untreated or TKI-resistant CML is progression to BP (with or without an intervening AP) within 3–8 years from diagnosis of CML-CP, with a rapidly fatal disease course upon onset of BP [
3‐
5].
BCR-ABL1 TKIs have remarkably altered the prognosis of CML, with patients in CP now having a normal lifespan compared to their age-matched peers and the prospect of a cure within reach for those who enter a durable treatment-free remission (TFR) [
6,
7]. In the pivotal phase III IRIS study comparing imatinib to chemoimmunotherapy (cytarabine + IFNα), the 10-year rate of progression to an advanced phase was reduced from 12.8% with chemoimmunotherapy to 6.9% with imatinib [
8]. The 10-year cumulative incidence of BP in the large imatinib-based CML-study IV cohort was 5.8%, and this rate has reduced even further with the use of second-generation TKIs (dasatinib, nilotinib, bosutinib) in the frontline setting [
9‐
12].
Unlike CML-CP, patients who either present with de novo CML-BP or progress to BP from a previous CP/AP still have dismal outcomes, and the median OS remains less than 1 year after diagnosis of BP [
13]. Approximately 70–80% of BP cases occur in myeloid blast phase (MBP) and 20–30% occur as a lymphoid blast phase (LBP) or with blasts of mixed lineage [
1,
5]. Patients with LBP have better outcomes than those with MBP [
14]. For CML-MBP, treatment approaches vary from the use of single-agent TKI to acute myeloid leukemia (AML)-like induction therapy + TKI, and there is no consensus frontline treatment recommendation [
13,
15‐
17]. In order to clarify the optimal therapy for these patients, we retrospectively analyzed all patients receiving frontline treatment for CML-MBP over the past 20 years at our institution and assessed response rates, depth of response, and survival outcomes across different frontline therapies.
Methods
Study design and participants
This was a retrospective study including patients ≥ 18 years of age treated in the frontline setting for CML-MBP between 2000 and 2019. MBP was defined as the presence of ≥ 20% bone marrow (BM) myeloblasts with or without concurrent extramedullary (EM) disease per the World Health Organization 2017 criteria for CML-BP. Patients could have received prior therapy (including TKIs) for CML in CP/AP. Exclusion criteria were the presence of EM-only disease or any prior therapy for CML-BP aside from hydroxyurea. Patients were divided into four frontline treatment approaches: (1) intensive chemotherapy (IC) + TKI, (2) hypomethylating agent (HMA) + TKI, (3) TKI alone, or (4) IC alone. IC was defined as a regimen containing ≥ 1 g/m2 of cytarabine. All patients in the HMA + TKI group received decitabine. For certain specified analyses, groups 1 and 2 were combined into an “IC/HMA + TKI” group. TKI therapy was divided into 1st-generation (imatinib), 2nd-generation (dasatinib, nilotinib, bosutinib), or 3rd-generation (ponatinib) agents. The retrospective study was conducted at a single academic medical center (The University of Texas MD Anderson Cancer Center), approved by the Institutional Review Board, and conducted in accordance with the Declaration of Helsinki.
Response and outcome definitions
Morphologic/hematologic responses were assessed per AML response criteria, as defined according to European LeukemiaNet consensus guidelines [
18]. Patients who achieved a best response of complete remission (CR), complete remission with incomplete hematological recovery (CRi), or morphologic leukemia-free state (MLFS) were considered responders, and all others were considered nonresponders. Complete cytogenetic response (CCyR) was defined as no detectable Ph + chromosome by conventional karyotyping. Major molecular response (MMR; MR
3) was defined as a
BCR-ABL1/ABL1 ratio > 0.01 and ≤ 0.1% on the International Scale (IS). MR
4 was defined as a
BCR-ABL1/ABL1 ratio > 0.0032% and ≤ 0.01% on the International Scale (IS). MR
4.5 was defined as a
BCR-ABL1/ABL1 ratio ≤ 0.0032% on the International Scale (IS). All molecular responses were assessed by quantitative real-time PCR analysis at the molecular diagnostics laboratory at MD Anderson Cancer Center and were converted to the IS.
Relapse was defined as ≥ 5% BM blasts, new extramedullary disease, or appearance of peripheral blasts unrelated to BM recovery after an initial objective response. Cumulative incidence of relapse (CIR) was calculated from time of initial response (CR, CRi, MLFS) to relapse, censored for death in morphologic remission or if the patient was alive at last follow-up. Event-free survival (EFS) was defined as the time from diagnosis to lack of response, morphologic relapse after initial morphologic response, or death from any cause. Overall survival (OS) was defined as the time from diagnosis to death from any cause. EFS and OS were censored if the patient was alive at last follow-up.
Statistical methods
Patient characteristics were analyzed using descriptive statistics, and survival analyses were performed using Kaplan–Meier methodology. To compare two groups, Chi-Square test was performed for categorical variables, and t-test or Mann–Whitney test was performed for continuous variables. Log-rank (Mantel-Cox) test was performed to compare Kaplan–Meier survival curves. All statistical analyses were conducted using GraphPad Prism version 8.4.3.
Discussion
Outcomes for CML-CP have improved considerably in the TKI era [
13]. However, the occurrence of BP remains a challenging clinical scenario and for many patients is an end-stage event [
6]. Due to its rarity, treatment is guided primarily by retrospective series, small single-arm clinical trials, and expert opinion. In this study, we identified 104 patients with CML-MBP over a 20-year period treated with different therapeutic approaches in the frontline setting. To our knowledge, this is the largest cohort of patients with CML-MBP to include data on response rates, relapse incidence, and survival based on therapeutic approach. We found that therapy with IC/HMA + TKI in comparison to TKI alone yields higher response rates, lower risk of relapse, and improved 5-year EFS/OS, suggesting that combination therapy with IC or HMA plus a TKI should be consider the optimal upfront therapy for these patients, rather than TKI alone.
Although there are several published algorithms on how to approach CML-MBP, there is no formal category 1 recommendation in clinical guidelines. Trial enrollment, AML-like induction therapy + TKI, or TKI alone are all considered acceptable treatment options [
17]. The incorporation of a TKI is crucial, as the BCR-ABL1 fusion kinase remains a significant leukemic driver in patients with CML-BP. For historical comparison to demonstrate this point, we included patients in our study treated with IC alone. While the number of patients in our cohort who received IC alone is small, it is evident that IC alone is inadequate therapy, with a response rate of 12.5% and 5-year EFS and OS rates of 0%. The utilization of TKI monotherapy instead of IC improved outcomes for patients, however long-term outcomes remained poor with a 5-year EFS rate of 5% and OS rate of 13%. While the subsequent use of combination regimens (IC/HMA + TKI) did not significantly improve median EFS/OS compared to TKI, there was a clear benefit in the proportion of patients achieving long-term survival.
The utilization of combination therapy for CML-MBP may have yielded improved responses compared to TKI alone because progression to MBP, with or without the development of a resistant
ABL1 mutation, suggests development of leukemogenic events independent of the BCR-ABL1 kinase [
5,
20]. The biology of BP is much less understood compared to CP. Patients with BP may have mutations in genes commonly associated with AML, such as
ASXL1 and
RUNX1, as well as ACAs that confer a negative prognosis in AML, such as − 7 and inv(3) [
19,
21,
22]. These mutations and ACAs along with ongoing constitutive BCR-ABL1 kinase activity and additional cellular changes yield a complex disease biology in which both proliferative and anti-apoptotic pathways have been implicated, including but not limited to those mediated by Wnt/β-catenin, MYC, and BCL-2/BCL-xL [
5,
23]. The progression to MBP therefore appears to create a more heterogenous disease than CML-CP, and targeting a single driver pathway may be insufficient. Given the proliferation of myeloblasts and immature progenitor cells, aspects of the disease clinically appear to mimic AML more so than CML-CP. Thus, the use of an acute leukemia chemotherapeutic regimen + TKI may be efficacious by affecting multiple cell survival/death mechanisms rather than one primary target, as has been shown for patients with Ph + acute lymphoblastic leukemia (ALL) treated with TKI-based combination regimens [
24].
Patients treated with combination therapy had improved long-term survival in our cohort, and it appears that the etiology of this benefit primarily arose from a higher rate of patients proceeding to ASCT (32.5% with IC/HMA + TKI vs. 10.7% with TKI alone). Among patients who had a clinical response and underwent ASCT, 5-year OS was 58% for all patients (62% for those treated with IC/HMA + TKI). Prior to the TKI era, ASCT was the only curative option for patients with any phase of CML, and it remains the therapeutic goal for patients with CML-BP [
25]. In AML, Ph + ALL, and CML-CP, outcomes post-ASCT are influenced by the pre-transplant depth of response [
26‐
28]. TKI therapy alone resulted in a CR/CRi rate of 33.9% and CCyR rate of 10.7%. In contrast, combination therapy produced a CR/CRi in 57.5% of patients and CCyR in 45% of patients. MMR or deeper molecular response rates were relatively low with combination therapy (29.4% with IC + TKI and 18.8% with HMA + TKI) but were still higher than the 4.3% rate with TKI alone. Thus, it is possible that the improved outcomes associated with IC/HMA + TKI were due to a higher likelihood of reducing leukemia burden pre-transplant, thereby permitting more patients to proceed to the potentially curative option of ASCT.
IC + TKI and HMA + TKI have been assessed separately in small, early phase clinical trials and retrospective studies [
29‐
34]. However, to the best of our knowledge they have not been previously compared to each other within the same cohort. Though IC + TKI and HMA + TKI produced similar response rates and survival outcomes in this cohort, it is notable that patients in the IC + TKI group were younger, only received a 2nd/3rd-generation TKI, and were more likely to receive ponatinib compared to patients treated with HMA + TKI. Though 60-day mortality was similar between both groups, HMAs are typically better tolerated with fewer toxicities compared to most IC regimens for AML. With regards to which TKI to use in combination therapy, no patient treated with an imatinib-based combination regimen survived past 2 years, further supporting the use of later-generation TKIs in this context. Assessing for differences between patients treated with ponatinib or a 2nd-generation TKI is limited by the small number of patients treated with ponatinib (
n = 8). However, 5-year EFS and OS rates were similar or slightly higher with the use of ponatinib compared to 2nd-generation TKIs, which parallels findings in Ph + ALL where the use of ponatinib as the frontline TKI has demonstrated significant clinical efficacy even in the absence of a T315I mutation [
35,
36]. Ongoing studies are evaluating lower-intensity therapy (HMA + venetoclax) combined with ponatinib for patients with Ph + myeloid malignancies, including CML-MBP (NCT04188405), which may further improve outcomes for this disease. Retrospective and prospective studies also support the use of venetoclax-based combination regimens in CML-BP and advanced Ph + leukemias [
37,
38].
Our study is limited primarily due to its retrospective, nonrandomized, single-center methodology. Thus, we were unable to capture certain types of clinical information that are typically gathered for prospective clinical studies and may have influenced treatment outcomes, such as patient performance status, physician rationale for treatment choice, and adverse events. Due to the declining incidence of CML-MBP, prospective randomized trials are unlikely to occur and, if so, will likely require significant time to accrue. We aimed to minimize the potential biases inherent in a retrospective cohort analysis by having relatively strict inclusion criteria and focusing on patients receiving their first therapy for CML-MBP and excluding patients with EM-only disease. We focused on patients who received frontline treatment for CML-MBP in order to minimize the variability associated with including patients within the same analysis who had no prior BP treatment with those who had relapsed/refractory disease to one or more prior BP regimens. For patients with EM-only disease, responses as assessed by imaging rather than typical CML or AML hematologic and molecular criteria would have complicated our response assessment, particularly in a retrospective analysis where imaging modalities were not consistently used.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.