Randomized study of imatinib for chronic myeloid leukemia: comparing standard dose escalation with aggressive escalation

The introduction of imatinib mesylate, a selective BCR-ABL tyrosine kinase inhibitor (TKI), in the early 2000s was a revolutionary change in the management of chronic myeloid leukemia (CML) and a shift in cancer management paradigms, in general.¹  Imatinib, given orally at a daily dose of 400 mg, had been a standard initial therapy for patients with chronic-phase CML (CML-CP) before the introduction of second-generation TKIs in ∼2010.^2-4 

The Japan Adult Leukemia Study Group (JALSG) previously conducted a prospective multicenter phase 2 study (CML202) to examine the efficacy and safety of imatinib therapy in newly diagnosed Japanese patients with CML-CP.⁵  In 481 evaluable patients, the estimated 7-year overall survival (OS) and event-free survival after a median follow-up of 65 months were 93% and 87%, respectively. However, dose reduction or interruption was required in 223 patients (46%), and dose-escalation was performed in only 10 patients (2%) during the first 24 months.

Despite the excellent survival reported in previous studies, including ours, CML was resistant to 400-mg imatinib in 5% to 20% of patients.^4,6,7  The 3-year survival rates for patients who failed imatinib in the chronic phase were reported to be 72%.⁸  The treatment guidelines of the European LeukemiaNet (ELN2006) proposed a “dose escalation of Imatinib based on clinical assessments of disease response.”⁹  Preclinical data and results from single-arm trials suggested that better results could be achieved with a higher dose.¹⁰  The European group conducted a randomized study to compare high-dose imatinib (800 mg) with the standard dose (400 mg) as a front-line therapy for all high-risk CML patients, according to the Sokal score¹¹ ; however, the results from this study did not support the extensive use of high-dose imatinib.

Another approach is to start administering imatinib at 400 mg and increase the dose in a response-oriented manner. This method has the advantage of avoiding the enhanced risk of adverse effects from the higher dose and the increased expense in patients who show an adequate response with 400 mg.

Hughes et al conducted a trial of dose escalation to 800 mg/d for suboptimal responses, and Jabbour et al assessed the efficacy of imatinib dose escalation in patients who met the criteria for failing standard-dose imatinib.^12,13  Both studies found some merits for response-oriented dose escalation; however, these studies did not include controls.

An alternative strategy is aggressive dose escalation based on more sensitive assessments of the clinical response than those suggested in the standard escalation ELN2006. We designed a phase 3 randomized study of imatinib therapy for CML-CP that compared aggressive dose escalation with standard dose escalation to assess the rate of a major molecular response (MR3) at 12 months, which serves as a surrogate marker for long-term progression-free survival (PFS).^6,14,15  Furthermore, MR3 at 12 months is reported to be a surrogate marker for deep molecular response at 5 years,¹⁶  which opens the door to treatment-free survival.¹⁷ 

The present study was a prospective multicenter phase 3 study of previously untreated newly diagnosed patients with Philadelphia chromosome (Ph)⁺ CML-CP. Patients were consecutively registered at 61 participating institutes belonging to the JALSG. Patients were randomly assigned to group A or group B, stratified by Sokal risk (Figure 1). In group A, dose escalation was performed according to ELN2006 criteria: if a complete hematological response (CHR) was not achieved at 3 months, the dose was increased to 600 mg, and if a partial cytogenetic response (PCyR) was not achieved at 6 months, the dose was increased to 600 mg. In group B, the dose escalation was more aggressive: if a complete cytogenetic response (CCyR) was not achieved at 3 months, the dose was increased to 600 mg, and if a major molecular response (MR3) was not achieved at 6 months, the dose was increased to 600 mg. The dose escalation was performed once during the study. The primary end point was the MR3 rate at 12 months, which serves as a surrogate marker for PFS. The secondary end points were OS and PFS at 3 years, the MR3 rate at 2 years, and the incidence of grade 3 and 4 adverse effects. This study was approved by the institutional review boards at each participating institution. Written informed consent was obtained from all patients before registration in accordance with the Declaration of Helsinki. This study was registered at http://www.umin.ac.jp/ctr/ as #R000000965.

Patients were eligible for inclusion in this study if they were 15 to 69 years old, had de novo Ph⁺ CML, and had not received interferon-α treatment. Further eligibility criteria were adequate liver function (serum bilirubin level ≤2.0 mg/dL and serum liver aminotransferase levels less than threefold above the normal upper limits), adequate kidney function (serum creatinine ≤2.0 mg/dL), normal heart and lung function, an Eastern Cooperative Oncology Group performance status of 0 to 3, and no prior or concurrent malignancy.

The phases of CML and CHR were defined as described in the IRIS study.²  Cytogenetic responses were evaluated by G banding of ≥20 marrow cells in metaphase and were categorized as CCyR (no cells positive for the Ph chromosome) or PCyR (1-35% of cells positive for the Ph chromosome). The molecular response was measured by reverse-transcription real-time quantitative polymerase chain reaction or transcription-mediated amplification and hybridization protection assay. MR3 was achieved when <100 BCR-ABL mRNA copies were detected per 1 μg of RNA, corresponding to a 3-log reduction from the standardized baseline. MR4 represents a >4-log reduction. We previously reported that a transcription-mediated amplification and hybridization protection assay shows a significant correlation with a FusionQuant M-BCR kit (R > 0.971, P < .01), a standard quantitative nucleic acid method used in Europe, and with reverse-transcription real-time quantitative polymerase chain reaction.^5,18 

PFS was defined as the time between registration and the earliest occurrence of any of the following events: death due to any cause, progression to accelerated phase (AP) or blast phase (BP), and/or loss of major cytogenetic response or CHR. OS was defined as the time between the date of registration and death due to any cause. Patients, with the exception of those who voluntarily withdrew from the study, continued to be monitored for progression and survival for 3 years. Additionally, these patients’ prognoses were studied in July of 2016. Adverse events were assessed according to National Cancer Institute–Common Toxicity Criteria version 2.0 (http://ctep.cancer.gov/protocolDevelopment/electronic_applications/ctc.htm). The mean daily dose of imatinib over a designated period was defined as the total of the doses administered divided by the total number of days on which it was administered.

The primary end point was the proportion of MR3 at 1 year. From previous studies, the expected MR3 rate was 35% for patients in group A at 1 year of treatment (400 mg daily). To detect an absolute difference (an MR3 rate of 55% in experimental arm group B) with a 1-β power of 80% and a 2-sided α of 0.05, the number of patients to accrue was established as 100 for each arm. The secondary end point was the proportion of MR3 at 2 years. The analysis was performed on the intent-to-treat (ITT) population, and patients who left the study early or those without sufficient data for any other reason were included in the ITT analysis as <MR3. Efficacies were tested at a significance level of 0.05 using the Cochran–Mantel–Haenszel test, with adjustment for stratification according to the Sokal score. The Kaplan-Meier method was used to estimate the time to the first MR3 and to analyze OS, PFS, and all times to event. Differences between the subgroups of patients were evaluated using the log-rank test. The cumulative rate of incidence of each reason for withdrawal was estimated according to competing risk methods, and the Gray test was used for comparison.¹⁹  Comparisons of baseline characteristics in the subgroups were made using the χ² test or Fisher's exact test for categorical variables. All statistical analyses were performed using STATA software release 14 (College Station, TX). Two-sided P < .05 was considered significant.

A total of 248 patients entered this study between 16 June 2007 and 15 June 2011. Their median age was 49 years (range 15-69); 86 were female, and 162 were male. Three patients were excluded for violation of consent, Ph⁻ CML, or AP. Thus, a total of 245 patients was randomly assigned to group A (n = 122) and group B (n = 123). According to the Sokal scores, 45 patients were high risk, 77 patients were intermediate risk, and the remaining 123 patients were low risk. The median white blood cell count at diagnosis was 43 × 10⁹/L (range, 10-881). Age, sex, Sokal scores, peripheral blood cell counts, and additional cytogenetic abnormalities were equivalent between the groups (Table 1).

All patients started with a dose of 400 mg of imatinib; however, during the first year, the dose was reduced to <400 mg in 31 patients in group A (reduced in 22, discontinued in 9) and in 19 patients in group B (reduced in 5, discontinued in 14). The median dose intensities of imatinib in the first year were 375.1 mg/d in group A and 424.2 mg/d in group B (P < .01, Student t test). Interruption of imatinib was observed at similar proportions in group A (28.8%) and group B (24.1%).

According to the cytogenetic response results by ITT analysis, the aggressive dose escalation group (group B) showed a better early cytogenetic response at 6 months (78.7%) compared with the standard dose escalation group (group A; 69.4%), with borderline significance (P = .09, Fisher's exact test). However, the rate of CCyR at 12 months was similar in both groups (81.8% in group A and 84.3% in group B).

The rate of MR3 by ITT analysis at 12 and 24 months, which were the primary and secondary end points of this study, is shown in Figure 2. MR3 was achieved in 52.1% (95% confidence interval [CI], 42.8-61.2) and 58.7% (95% CI, 49.4-67.0) of patients in groups A and B, respectively, at 12 months (P = NS). At 24 months, the rate of MR3 was similar in group A (70.0%) and group B (68.3%). The cumulative incidence for the first MR3 is shown in Figure 3. MR3 rates at 12 months and 24 months were 62.8% and 86.3%, respectively, in group A and 67.4% and 80.9%, respectively, in group B.

Twenty-two patients switched to new-generation TKIs in the first year because of adverse effects (n = 16), treatment failure (n = 3), or other reasons (n = 4). There were no differences between the 2 groups (12 in group A and 10 in group B). To eliminate the positive effects of the new TKIs, the MR3 rate at 12 months (the primary end point) was calculated excluding these patients. The actual rates of MR3 at 12 months were 54.1% and 62.2% in groups A and B, respectively (P = NS).

The results of the dose escalation at 3 and 6 months are shown in Figure 4A. At 3 months in group B, the protocol called for 56 patients to increase their dose to 600 mg because they did not achieve CCyR. Twenty-eight patients (50.0%) increased their dose, but 28 patients (50.0%) failed to increase their dose. The remaining 65 patients maintained their dose of imatinib because they achieved CCyR. At 6 months in group B, 34 of 65 patients who maintained the drug dose at 3 months needed a dose escalation to 600 mg because they did not achieve MR3. Seventeen patients (50.0%) increased their dose, but 17 patients (50.0%) failed to increase the dose. The remaining 29 patients achieved MR3, and 2 patients did not receive a molecular response test. The primary reason why patients failed to increase the dose was adverse events. Overall, the protocol called for a dose increase of imatinib at 3 or 6 months in a total of 90 patients in group B; 45 patients (50.0%) increased the dose according to protocol, and the remaining 45 patients failed to increase. In contrast, in group A, only 4 and 6 patients failed to increase the dose at 3 and 6 months, respectively.

Figure 4B shows the MR3 rates at 12 months for group B. Eight patients who did not receive molecular response tests were excluded from this analysis. Twenty-nine patients who reached CCyR at 3 months and MR3 at 6 months maintained the MR3 status at 12 months. The proportion of MR3 at 12 months was significantly higher among patients who were able to increase the dose of imatinib according to the protocol compared with those who were not (60.0% vs 35%, P < .05).

Another secondary end point was adverse events. Table 2 shows that the incidence of nonhematologic events was equivalent in both groups. With regard to hematologic events, a trend toward a higher incidence of lymphocytopenia and anemia was observed in group B. The aggressive dose escalation group (group B) did not show a higher frequency of withdrawal due to drug toxicity compared with group A (7.4% vs 8.1% at 12 months, 11.6% vs 9.8% at 24 months).

The OS and PFS rates at 3 years, which were secondary end points, are shown in Figure 5. OS was 99.3% in group A and 97.4% in group B, with no significant difference. PFS, with AP/BP and “failure” defined per the ELN2006 guidelines as events, was 89.1% in group A and 94.2% in group B (P = .17). An insufficient response was observed in 20 patients: 6 with BP (1 with preceding AP) and 14 with failure without AP/BP. In group B, 5 patients developed BC 6 months, 6 months, 9 months, 16 months, and 17 months after the start of treatment. However, in group A, only 1 patient developed AP/BC at 3 months. On the other hand, treatment failure, determined by cytogenetic analysis, primarily occurred in group A (12 in group A, 2 in group B). The cumulative incidence of withdrawal by failure without AP/BP was higher in group A than in group B (9.2% vs 2.5% at 24 months, P = .15). Five patients died during the 3-year period: 3 died of BP, 1 died of a cardiovascular event during chemotherapy for BP, and 1 died as a result of drug toxicity. Among 6 patients with BP, 3 received hematopoietic stem cell transplantation, and 2 survived through the observation period.

The OS rates at 8 years were 93.0% for group A and 93.9% for group B (Figure 5A). The causes of death after 3 years were unrelated to leukemia.

This is the first prospective randomized study to compare 2 dose-escalation programs among newly diagnosed patients with CML-CP. With the aggressive dose escalation program (group B), we expected a better MR3 rate at 12 months, which is a surrogate marker of PFS. However, 52.1% and 58.7% of patients in groups A and B, respectively, achieved MR3 at 12 months (P = .3). Similarly, although the CCyR rate at 6 months after treatment was slightly better in group B (89%) than in group A (79%), with borderline significance (P = .1), the overall CCyR rate at 12 months was ∼83% in both groups. Additionally, the rates of OS and PFS at 3 years were equivalent in groups A and B. Furthermore, the OS probability at 8 years was 93.0% and 93.9% for groups A and B, respectively. Long-term results within the IRIS study strongly suggested that early molecular response is related to 7-year and 10-year PFS.^14,15  This study is in accordance with the long-term results found in IRIS, because our molecular response at 1 year was equivalent in groups A and B.

There are several explanations for these results, which were contrary to our expectations. First, the protocol called for 96 patients in group B to increase their dose of imatinib to 600 mg, but only 59 patients (53%) followed the protocol. Forty-three patients (47%) did not increase the dose. In reality, the median dose intensity of imatinib in the first year in group B was only 424.2 mg/d. An Australian group conducted a similar type of study, and a dose escalation from 600 mg to 800 mg was achieved in only 47% of their patients. For Japanese patients, the maximal tolerated dose of imatinib in practice might be 600 mg/d.⁵  Inadequate dose escalation in group B may explain the failure to demonstrate the efficacy of aggressive dose escalation.

Second, the excellent MR3 rate (52.1%) at 12 months in group A (European LeukemiaNet guideline) in comparison with other major studies may reduce the difference between groups A and B.^3,11,20,21  In our previous JALSG CML202 study, which is comparable to the group A protocol, only 27% achieved MR3 at 12 months.⁵  In other previous studies, <30% of patients achieved MR3 at 12 months. Thus, we designed this protocol expecting the MR3 rate at 12 months to be 35% in group A. Furthermore, our result demonstrated similar results with the studies of nilotinib and dasatinib.^22,23  Although second-generation TKIs were introduced during the timeframe of this study, this effect was minimal, as described previously. The excellent results in the standard arm may be due to an improvement in the clinical ability to manage imatinib-based treatment in Japan. Indeed, the median daily dose in the JALSG CML202 study was 349.1 mg/d, whereas that of group A in the current study was 375.1 mg/d.^24,25  The discrepancy between western countries may also be explained by lower body weight and lower Sokal risk in the current study.^5,26 

It is important to note that BP developed primarily in group B (1 in group A, 5 in group B) during the first 3 years of this study. On the other hand, treatment failure, determined by cytogenetic analysis, primarily occurred in group A (12 in group A, 2 in group B). These results suggest the following hypotheses. First, BP could not be overcome by a higher dose of imatinib. Resistant clones might have already developed in the patient’s body at initial diagnosis and then manifested as AP/BP during the first 5 years of imatinib treatment.^14,27  Because additional oncogenes occur during cell division, theoretically, transformation should seldom occur during imatinib treatment.^28-30  Second, higher doses of imatinib may affect the early response; however, it is not related to better long-term survival in this study. In patients who do not have resistant clones, even if leukemic cells are slowing down or reaching failure criteria, their numbers gradually decrease, so the disease is not considered to be progressing. In fact, in the 3-year follow-up in the present study, 8 of the 14 patients who initially met failure criteria achieved a CCyR, and 7 patients achieved MR3 at the last follow-up (data not shown). Most of these patients were in group A, and the number of leukemic cells in these patients appeared to decrease slowly, even with a small amount of imatinib.

Among the patients for whom the protocol called for a dose escalation, a better MR3 rate was observed at 12 months in those who complied with the dose increase in comparison with those who did not, suggesting that aggressive dose escalation may be associated with a better outcome in patients who tolerate the drug well. Interestingly, groups A and B showed similar rates of withdrawal as a result of adverse events, which may be due to the small difference in daily imatinib doses (375.1 mg vs 424.2 mg). Cortes et al reported similar rates of patient withdrawal in the 400-mg arm and the 800-mg arm.³¹  In the current study, severe adverse events, including anemia and lymphocytopenia, were observed more frequently in group B. This is compatible with the results of a previous study, which showed that high doses of imatinib can produce myelosuppression.

From this study, we conclude that aggressive dose escalation based on more aggressive criteria, such as a CCyR at 3 months and an MR3 at 6 months, is not recommended for all patients. However, aggressive dose escalation may be associated with a better outcome for some patients if the drug dose is managed carefully based on the patient’s condition (eg, residual leukemia, adverse effects), especially for those who desire treatment-free remission. These conclusions are useful for hematologic malignancies, as well as for cancer treatment, in general.

The authors thank the participating doctors and other medical staff of the 61 hospitals who enrolled patients in the present study and provided the necessary data to make the study possible.

This work was supported by a grant from the Japanese Ministry of Health, Labor and Welfare and by a Japanese Red Cross Nagoya First Hospital Research Grant.

For a complete list of the members of the JALSG, see the supplemental Appendix.

Contribution: K.M., K.O., and S.O. conceived and designed the study; K.M., Y.M., and T.N. conducted the clinical trial; K.M. wrote the manuscript; N.U., C.N., H.F., S.F., T.S., H.O., N.I., and N.E. coordinated and participated in patient care; S.H. designed and performed the statistical analyses; and K.O., N.U., and K.F. analyzed and interpreted data; and all authors reviewed the manuscript.

Conflict-of-interest disclosure: K.M. received honoraria from Bristol-Myers Squibb, Novartis, and Pfizer. N.U. received honoraria from Astellas Pharma, Bristol-Myers Squibb, CIMIC, Otsuka Pharmaceutical, Pfizer, and Takeda Bio Development. C.N. received honoraria from Bristol-Myers Squibb, Novartis, and Pfizer and research funding from Pfizer. S.F. received honoraria from Bristol-Myers Squibb, Novartis, Otsuka Pharmaceuticals, and Pfizer and research funding from Pfizer and Novartis. H.O. received honoraria from Novartis. N.I. received honoraria from Bristol-Myers Squibb, Novartis, Otsuka Pharmaceuticals, and Pfizer. N.E. received honoraria from Bristol-Myers Squibb, Novartis, and Pfizer. Y.M. received honoraria from Celgene, Dainippon-Sumitomo, Kyowa-Hakko Kirin, Novartis, and Shinbio Technologies; received research funding from Takeda and Chugai; and served as a consultant for Agios and Novartis. T.N. received research funding from Astellas Pharmaceuticals, Otsuka Pharmaceuticals, Pfizer, Toyama Chemical, Nippon Shinyaku, and FUJIFILM. The remaining authors declare no competing financial interests.

Correspondence: Koichi Miyamura, Department of Hematology and Hematopoietic Cell Transplantation Center, Japanese Red Cross Nagoya First Hospital, 3-35 Michishita-cho, Nakamura-ku, Nagoya 453-8511, Japan; e-mail: miyamu@nagoya-1st.jrc.or.jp.