Current Hematologic Malignancy Reports

, Volume 7, Issue 2, pp 109–115

Clinical Trials in Chronic Myeloid Leukemia


    • Universitätsmedizin Mannheim, Universität Heidelberg
  • Markus Pfirrmann
    • Institut für Medizinische Informationsverarbeitung, Biometrie und Epidemiologie (IBE)Ludwig-Maximilians-Universität München
Chronic Myeloid Leukemia (J Goldman, Section Editor)

DOI: 10.1007/s11899-012-0118-1

Cite this article as:
Saussele, S. & Pfirrmann, M. Curr Hematol Malig Rep (2012) 7: 109. doi:10.1007/s11899-012-0118-1


The introduction of the tyrosine kinase inhibitor (TKI) imatinib in the treatment of Philadelphia chromosome-positive chronic myeloid leukemia (CML) has substantially improved the outcome of CML patients. Despite the positive results, problems and questions remained. This was the rationale to setup trials for treatment optimization, where imatinib was administered in higher dose and/or in combination with other therapy but where also new and potentially more efficacious second-generation TKI, nilotinib and dasatinib, were investigated. This review summarizes data of recently published first-line studies with the standard treatment imatinib 400 mg as one study arm. Results of randomized comparisons to higher-dose imatinib treatment, nilotinib or dasatinib are discussed. With regard to outcome interpretation, general aspects on statistical issues and endpoint definitions are put into focus. Considering decidedly increased longevity thanks to TKI treatment, future research should include the evaluation of the quality of life (QoL). Relating also to QoL, safe ways of drug discontinuation need to be investigated.


Chronic myeloid leukemiaClinical trialsTKIImatinibDasatinibNilotinibMolecular remission (response)Cytogenetic remission (response)Competing riskCumulative incidenceComposite endpointsProgression-free survivalOverall survivalStoppingCure


With the introduction of the BCR-ABL tyrosine kinase inhibitor (TKI) imatinib in the treatment of Philadelphia chromosome-positive (Ph-positive) chronic myeloid leukemia (CML), the outcome of CML patients has been substantially improved. The 5-year survival probability increased from 50%–70% with previous, usually interferon alpha (IFN)-based standard therapy [1] to about 90% in the imatinib era [2•]. In newly diagnosed, untreated CML patients in chronic phase (CP), the superiority of imatinib to interferon was first established in the International Randomized Study of Interferon Alpha Versus STI571 (imatinib), the IRIS trial [3]. Compared to the control arm with interferon and cytarabine, the trial succeeded in showing significantly higher rates of complete cytogenetic response (CCR) and major molecular response (MMR) as well as decreased probabilities of progression to accelerated phase (AP) or blast crisis (BC) with imatinib treatment. As a consequence, imatinib with a daily dosage of 400 mg became the recommended initial therapy for CML patients [4••].

Despite these positive results, problems and questions remained. In the study with the largest follow-up, the IRIS trial, 45% of the 553 patients randomized to the imatinib arm were off imatinib treatment after 8 years [5]. Besides lacking renewal of consent and miscellaneous reasons (17%) or stem cell transplantation (SCT, 3%), reasons for discontinuation were intolerance (6%), unsatisfactory therapeutic outcome (16%), and death (3%). Unsatisfactory outcome comprises the failure to achieve response by a specific time point (e.g., complete hematological response [CHR] at 3 months; primary resistance) or the loss of initial response (e.g., loss of CCR, secondary resistance). Early after imatinib introduction, it became obvious that resistance and hence unsatisfied outcome is frequently associated with mutations in the kinase domain of BCR-ABL [6, 7]. So far, over 100 different point mutations in the BCR-ABL kinase domain have been isolated from CML patients resistant to imatinib [8]. The frequency of BCR-ABL mutations in resistant patients was reported in 42% to 90%, depending on the methodology of detection, the definition of resistance, and the phase of the disease [9] as mutations are found more frequently in AP or BC. Other causes of primary and secondary resistance in CML patients without mutations include clonal evolution through additional chromosomal aberrations or duplication of the Philadelphia chromosome, pharmacokinetic alterations in molecules like hOCT-1 (human organic cation transporter-1) or MDR-1 (multidrug resistance), variability in the plasma level of TKIs, activation of alternative signaling cascades leading to BCR-ABL independent growth, and alterations in the epigenetic regulation of the expression of the BCR-ABL sequence [10].

This was the rationale to setup trials for treatment optimization, where imatinib was administered in higher dose and/or in combination with other therapy [2•, 11, 12•] but where also new and potentially more efficacious drugs were investigated. These new drugs, named second-generation TKIs, have already been tested and approved for first-line and second-line treatment of CML [13••, 14••].

The goal of this review is to summarize the exciting results of the new first-line studies published in 2010 and 2011, dealing with treatment optimization but also with stopping TKI due to the hope for potential cure. Furthermore, some reflections on the analysis of clinical endpoints and the interpretation of their results are provided.

Finally, the questions and challenges in CML therapy which should be addressed by future studies are discussed.

General Aspects on Clinical Trials

Due to the treatment success of TKIs, the mortality with CML has become remarkably low. In trials comparing TKIs, possible improvement in overall survival (OS) will not be recognized for a long time. However, to optimize treatment for as many patients as possible, there is a need to identify advantages of a certain treatment approach earlier on. Accordingly, alternative primary efficacy endpoints were introduced in clinical trials in CML [15•].

Different remission parameters are regarded as surrogate markers for OS. Investigators perform time-to-first-remission analysis, for example, with “remission” defined by CCR or MMR.

It is important to note that the studies for first-line treatment which are cited in this review used different primary endpoints: CCR at 12 months (BELA trial [16]), confirmed CCR at 12 months (DASISION [14••]), or MMR at 12 months (ENESTnd [13••], CML IV [2•]). Together with the most meaningful time for judgment of treatment success, the impact of these surrogate markers on OS is still a matter of debate [17].

Due to the presence of the inevitable competing risk “death before achieving a remission at a specific time point,” the appropriate approach to estimate the probabilities of a time-to-remission endpoint is the calculation of the cumulative incidence function (CIF) [15•, 18]. Whether probability estimation was performed by calculation of the CIF or by application of the, in this case, rather biased reverse Kaplan-Meier method is not always obvious. To support understanding and interpretation of results, it is essential that in all publications the applied statistical methods are clearly described and not only shifted to the online appendix or even entirely omitted.

Nearly all investigators use composite endpoints for trial results presentation. The term “composite endpoints” means that at least two different events are summarized as a result of the same kind [15•]. The most common composite endpoint is progression-free survival (PFS). Whichever of the three events “AP,” “BC,” or “death” is observed first ends the time of the observation [19]. A definition of failure-free survival (FFS) might comprise all events of PFS plus “loss of partial cytogenetic response (PCR)” [4••] as a further event. To avoid the bias that patients who have never achieved a CCR will be not at risk for a loss, FFS should also consider the failure to obtain remission until a certain time. As there was no standard procedure up to now, an approach to determine the remission status at a certain time point has been suggested [15•]. Compared with OS with death as the only event, composite endpoints have the advantage of a possibly earlier discovery of treatment differences. More events usually demand lower sample sizes. The motivation to choose a composite endpoint like FFS is the notion to cover all events leading to treatment discontinuation in clinical reality. However, in particular with FFS, the gravity of the events counted as “failures” differs considerably which leads to difficulties with respect to interpretation [15•].

Investigators or readers tend to compare composite endpoints between different studies. However, as the MD Anderson group demonstrated with regard to event-free survival (EFS), endpoint definitions are considerably different between studies [20]. The MD Anderson group evaluated the outcome of 435 CP-CML patients treated with imatinib (n = 281), nilotinib (n = 78), and dasatinib (n = 76) using definitions of PFS and EFS used in the IRIS [21], the ENESTnd [13••], the DASISION [14••], and MD Anderson Cancer Center (MDACC) trials. Of the 435 patients, 123 (28%) were taken off TKI therapy, 33 patients (7.6%) have died; 8 patients on TKI therapy, 2 patients within 60 days of being off TKIs, and 23 patients after being off TKIs for more than 60 days. Of the 33 deaths, 19 deaths would be counted as progression/events on the IRIS/ENESTnd/DASISION studies, whereas 14 deaths would be censored at time off TKI. On the basis of the four definitions used by IRIS, ENESTnd, DASISION, and MDACC trials, the corresponding 5-year PFS/EFS rates were 96%, 90%, 89%, and 81% [20].

Established prognostic scores in CML, the Sokal score [22] and the New CML score [23], support the assessment of disease burden at diagnosis in a certain patient group. Also in the TKI era, the prognostic scores still allow the discrimination of groups with different clinical outcome. Consequently, stratified analyses in accordance to one of the prognostic scores were performed in the ENESTnd and DASISION trials [13••, 14••]. Recently, the EUTOS score for prediction of CCR and subsequent progression-free survival was published [24]. Knowledge of the risk group distribution of a prognostic score may help for a raw understanding of different outcomes between trials. However, in any case, results between two treatment arms of different trials are not directly comparable. The application of advanced statistical methods provides a chance to reduce some bias [15•].

In summary, the use of appropriate, well-defined statistical methods and clinical endpoints is inevitable for a meaningful interpretation of results and supports comparability between trials.

Imatinib Treatment Optimization Trials

The optimal dose of imatinib for first-line therapy has not been finally determined. Dose-escalation has shown efficacy in CP-CML patients with suboptimal response or relapse to imatinib 400 mg/d [25, 26]. A number of prospective, nonrandomized trials showed faster achievement and higher rates of CCR and MMR in CP-CML patients receiving 600 mg or 800 mg of imatinib daily [2730]. Two randomized trials, the TOPS trial and the ELN trial, were conducted. Both trials randomized patients with newly diagnosed CML in CP to either imatinib 800 mg/d (400 mg twice daily, IM 800) or 400 mg/d (IM 400) [31, 32]. In the TOPS trial, 476 patients were randomized. At 12 months, differences in MMR and CCR rates were not statistically significant (MMR, IM 800: 46% vs. IM 400: 40%; CCR, 70% vs. 66%). The PFS probabilities at 18 months were 97% and 95%, respectively, while the OS probabilities were estimated to be 99% and 98% [32].

In the ELN trial, 216 patients with high risk according to the Sokal score were randomized. In accordance with intention-to-treat analysis, no significant difference was found for the primary endpoint CCR rate at 1 year. The estimated rates were 64% for the high-dose arm and 58% for the standard-dose arm [31]. No significant differences were detectable in the CCR and MMR rates at the investigated time points, 3, 6, and 12 months, or in progression-free and overall survival. In a subgroup analysis, 24 of 25 patients who could tolerate the full 800-mg dose achieved a CCR, but only 4 of 17 patients who could tolerate less than 350 mg achieved a CCR.

Because of different modes of action of imatinib and IFN [33, 34] and, in case of cytogenetic response, 8-year survival rates with IFN above 60% [35], a combination of imatinib and IFN appeared to be a rational option.

Three randomized trials aiming at treatment optimization were published in 2010 and 2011, using at least one of the above-mentioned concepts with either high-dose imatinib and/or standard-dose imatinib plus IFN as investigational arm: The SPIRIT study by the FI-LMC in France [12•], the CML study IV in Germany and Switzerland [2•], and the phase II study by the Nordic CML Study Group [36•].

The investigators of the SPIRIT study [12•] randomly assigned 636 patients to receive IM 400 alone, IM 400 plus cytarabine, IM 400 plus pegylated IFN alpha-2a (pegIFN), or imatinib alone at a dose of 600 mg daily (IM 600). At 12 months, the rates of CCR were similar among all four treatment groups. Compared with the other treatments, the addition of pegIFN to IM 400 demonstrated the most promising results with regard to molecular response. The rate of MMR as well as the rate of superior molecular response (defined as a molecular response rate <0.01% BCR-ABL on IS) at 12 months was significantly higher for patients receiving IM 400 and pegIFN (57% and 30%) than for patients receiving IM 400 alone (38% and 14%; P < 0.001 and P = 0.001). With 38% and 21% at 24 months, respectively, the difference in the rate of superior response remained significant (P = 0.001). During the first year of the trial, 45% of the patients discontinued pegIFN. Grade 3 or 4 neutropenia as well as thrombocytopenia were significantly more frequent with IM 400 plus pegIFN than with IM 400 alone. The number of patients who progressed to AP or BC within the first 2 years was with 4 or 5 patients similar between the four treatments [12•]. A reduced dose of pegIFN (45 ug/week; adjustment of the initial dose of 90 μg/week) resulted in improved tolerability and increased duration of the drug delivery. In addition, the reduced dose resulted in similar molecular response rates [37].

In the main phase of CML study IV [2•], IM 400 was compared with IM 800 and with IM 400 plus IFN in newly diagnosed chronic-phase CML patients. In contrast to the French trial, basically no pegIFN was used. Early high-dose therapy followed by rapid adaptation to good tolerability (tolerability-adapted imatinib 800 mg/d) increased the rate of MMR at 12 months significantly (IM 800 vs. IM 400: 59% vs. 44%; P < 0.001; IM 800 vs. IM 400 plus IFN: 59% vs. 46%; P = 0.002). Median dose in the 800 mg/d arm was 628 mg/d with a maximum dose of 737 mg/d during months 4 to 6 and a maintenance dose of 600 mg/d. All three treatment approaches were well tolerated with similar grade 3 and 4 adverse events. Independent of treatment approach, MMR at 12 months showed better progression-free survival (P = 0.002; 3-year survival probability: 99% vs. 94%) and overall survival (P = 0.001; 3-year survival probability: 99% vs. 93%) when compared with >1% but showed no difference to the group with remission rates from 0.1% to ≤1% on the IS [2•].

In the study of the Nordic group, 112 newly diagnosed chronic-phase CML patients with low or intermediate risk according to the Sokal score and with imatinib-induced CHR were randomized to receive either a combination of pegylated-interferon-α2b (pegIFN; 50 μg/w) and IM 400 mg or IM 400 only. In comparison to IM 400, the MMR rate at 12 months was significantly higher in the combination arm (P = 0.002; 54% vs. 82%). The CCR rate at 12 months was not significantly different (84% vs. 91%) [36•].

Recently, a meta-analysis of the CML IV, Spirit, TOPS, and ELN trials was published [38]. Compared with standard-dose IM 400, the relative chance to be in CCR or in MMR at 12 months was 17% (P < 0.001) and 26% (P < 0.001) higher with an imatinib dose >600 mg/d. Significant differences with respect to progression-free or overall survival could not be shown. Higher doses resulted in a significantly higher relative risk of developing a grade III/IV neutropenia (56% increase) and thrombocytopenia (86% increase) [38].

Second-Generation TKIs in First Line

Nilotinib and dasatinib have shown greater efficacy than imatinib in patients with newly diagnosed Ph-positive CML [13••, 14••]. In 2011, both studies to these second-generation TKIs were updated. In addition, data on bosutinib were presented [16]. This TKI has not yet been approved for treatment in CML, neither for first-line nor for second-line treatment.


In 2011, data from the ENESTnd study were presented after a minimum follow-up of 24 months [39•]. In this phase III trial, 846 patients with newly diagnosed CP CML within the previous 6 months were randomized (1:1:1) to receive nilotinib 600 mg (300 mg twice a day, NIL 600, n = 282), nilotinib 800 mg (400 mg twice a day, NIL 800, n = 281), or imatinib 400 mg once a day (n = 283) [13••]. The rates of the primary endpoint MMR at 12 months were significantly higher with NIL 800 and NIL 600 when each was compared with imatinib (43%, 44%, and 22%, respectively; P < 0.001 for both comparisons) [13••]. At 24 months, with 67% vs. 71% vs. 44% (P < 0.001 for both comparisons), the significant differences remained [39•]. The 2-year overall survival probability was 97.8% with NIL 800, 97.4% with NIL 600, and 96.3% with imatinib. Two-year progression-free survival probabilities resulted in 97.7, 98.0, and 95.2, respectively. Grade 3 or 4 neutropenia was more frequent with imatinib (21%) than with either dose of nilotinib (NIL 400: 11%, NIL 300: 12%).


In the phase III DASISION trial, 519 patients with newly diagnosed CP CML were randomly assigned to receive dasatinib 100 mg or imatinib 400 mg once daily [14••]. The primary endpoint of this study was confirmed CCR at 12 months. At 12 months, the rate of confirmed CCR in the dasatinib group was significantly higher than in the imatinib group (77% vs. 66%, P = 0.007). Also regarding the rate of MMR at 12 months, dasatinib showed a significant advantage over imatinib (46% vs. 28%, P < 0.001) [14••]. At the update by 24 months [40•], the cumulative response rates for CCR were 86% vs. 82%, for MMR 64% vs. 46%, and for a BCR-ABL reduction to ≤0.0032% IS 17% vs. 8%. The cumulative incidence analyses of the three response criteria over the whole observation period demonstrated a significantly faster response with dasatinib. The 2-year overall survival probabilities were 95.3 with dasatinib and 95.2 with imatinib. With respect to safety analyses, the dasatinib group had a significantly more favorable outcome than imatinib considering less frequent fluid retention, superficial edema, myalgia, vomiting, and rash whereas imatinib had a significant advantage regarding less frequent observations of pleural effusion and grade 3/4 thrombocytopenia [14••].

A comparative assessment of nilotinib versus dasatinib was performed by a matching adjusted indirect comparison of the ENESTnd and the DASISION studies [41]. In comparison to dasatinib, the investigators associated nilotinib with significantly higher rates of MMR and overall survival at 12 months. Lacking availability of individual patient data of the DASISION trial was resolved by an elaborated matching procedure adjusting for different patient profiles between the trials. However, data on spleen size could not be considered for the matching, a prognostic factor which is eminent in all three prognostic scores, the Sokal, the EURO, and the EUTOS score [2224]. To ensure comparability with regard to the frequencies of molecular assessments, overcome confounding due to unobserved patient characteristics, and to be able to analyze overall survival over the whole observation time, a future head-to-head randomized trial would provide valuable answers.


Bosutinib, a dual Src/Abl kinase inhibitor, has shown potent activity in CML in a phase I/II trial [42]. In the phase III BELA trial, patients with newly diagnosed CP CML were randomized 1:1 to receive oral bosutinib 500 mg/d (n = 250) or IM 400 (n = 252). The results of the primary efficacy endpoint CCR at 12 months were not significantly different between the treatments (bosutinib: 70%, imatinib: 68%). At the same time, the MMR rates differed significantly (41% vs. 27%, P < 0.05). In the most recent update [16] of the study, the overall survival probability at 18 months was 99% with bosutinib and 95% with imatinib.

Regarding adverse events, in comparison with imatinib, bosutinib was associated with higher incidences of diarrhea, vomiting, pyrexia, and abdominal pain and with lower incidences of edema and musculoskeletal events. Patients with bosutinib reported less frequently grade 3/4 laboratory abnormalities of neutropenia while grade 3/4 liver function test abnormalities occurred more frequently with bosutinib. Adverse events with bosutinib were typically transient, managed with dose modifications, and not life-threatening.


Ponatinib is a potent, oral, pan-BCR-ABL inhibitor active against the native enzyme and all tested resistant mutants, including the uniformly resistant T315I mutation [43]. Promising results were presented for ponatinib from the PACE trial, a phase II study in patients with CML and Ph + ALL resistant or intolerant to dasatinib or nilotinib, or with the T315I mutation [43]. So far no data in first line are available.

Stopping Trials

In comparison to the times prior to the TKI era, the advances in CML therapy led to an extension of survival of many years for most of the patients. However, TKI treatment is associated with side effects which impair patients’ quality of life (QoL). Considering the rapidly increasing prevalence of CML, this is of individual but also of socioeconomic importance.

Discontinuation of treatment was already suggested in the IFN era. In two studies [44, 45], patients who stopped treatment had similar probabilities of remaining in CCR and of overall survival as patients in CCR who continued IFN treatment. RTQ-PCR techniques were not sufficiently advanced; a criterion like MMR status of these patients was not discussed. In the TKI era, 12 patients with undetectable residual disease for at least 2 years while under imatinib treatment were registered for a pilot study on stopping therapy [46]. Six patients relapsed during the first 6 months whereas after a median follow-up of 18 months (range: 9–24 months), 6 patients remained in persistent molecular remission. On this basis, the prospective multicenter Stop Imatinib (STIM) study was started and recruited patients under imatinib for at least 3 years and with stable CMR for at least 2 years [47•]. After a median follow-up of 30 months, a molecular relapse occurred in 61 out of 100 patients. Fifty-eight relapses occurred during the first 7 months and 3 late relapses after 18 months. The probability of maintenance of CMR at 36 months was 39%. All patients were sensitive to an imatinib restart [38]. In a multivariate analysis, Sokal score and imatinib therapy duration were independent prognostic factors for prediction of molecular relapse. Recently, it has been demonstrated that second-generation TKI (dasatinib and nilotinib) may be safely discontinued in CML patients with a stable CMR. In a preliminary study, 4 of 12 patients lost MMR by 6 months (median follow-up 12 months). Upon early TKI re-introduction, MMR was rapidly regained [48].


Compared to standard monotherapy treatment with imatinib 400 mg/d, regarding complete cytogenetic or major molecular remission at 12 months, studies demonstrated superiority with IM 400 plus pegIFN [12•], a higher dose of imatinib (>600 mg/d) [2•, 38], nilotinib [13••], and dasatinib [14••]. However, so far, an advantage with regard to significantly better overall survival probabilities could not be shown for any of the alternative treatment approaches. Both CCR and MMR are understood as surrogate markers for overall survival. Nevertheless, the strength of the association between these surrogate markers and overall survival is variable and not deterministic.

Whether and when an overall survival advantage over standard IM 400 treatment will be shown in a randomized trial for any of the alternative treatments remains open. On the one hand, in case of unsatisfactory response, patients with IM 400 treatment will receive higher dose, on the other hand, patients with adverse events due to high dosing will have a dose reduction. Alternatively, patients not responding to imatinib-based treatment will change to one of the second-line TKI, nilotinib or dasatinib. Furthermore, in contrast to the time right after imatinib approval in the beginning of the last decade, the introduction of the first version of the ELN recommendations in 2006 supported the decision for an earlier change of treatment if deemed necessary [19]. Thus, in the long run, high proportions of treatment changes might prevent the discovery of clinically relevant overall survival differences between randomized treatment arms with intention-to-treat analysis. The future will tell whether the two new upcoming TKIs—bosutinib and ponatinib—yield more prominent survival differences in comparison to IM 400. However, in general, the detection of differences in drugs and the interpretation of study results became more complicated. Korn et al. discuss problems and interpretations when overall survival differences between randomized therapies are affected through effective subsequent therapies [49].

Due to the treatment success of TKIs, the traditional primary endpoint overall survival was substituted by various alternatives. The clinical markers regarded as (preliminary) primary endpoints varied between the studies, as did the definition and the analysis of endpoints. Even the established endpoint progression-free survival was subjected to different definitions hampering comparability between trials. Thus, investigators are challenged to agree upon the common selection and definition of endpoints and to acknowledge proper statistical methods for their analysis.

With the extension of survival time, the QoL of CML patients moved into focus and should be part of any new study. If superiority between treatment strategies cannot be decided through overall survival, a more favorable outcome with regard to side effects could result in a clinically relevant difference in the perception of QoL and thus, identify a more suitable treatment.

If medically justifiable, the best results in QoL could be provided through no treatment. The results of the above-mentioned stopping TKI studies are neither directly comparable nor transferable into general recommendations. The main reason is that the definition of CMR was not homogenous. Nowadays, results for BCR-ABL transcripts are defined according to the International Scale. However, thus far, validation has only been achieved up to a level of 1/1,000 which corresponds to MMR [50, 51]. A consistent CMR definition is a prerequisite to start new stopping trials and to answer many open questions: which level of MR is necessary to acquire a stable remission of disease, how long should treatment before discontinuation last, and has gender, the type of TKI, or the combination of treatments any influence on stability of remission after treatment stop? So far, only limited experience is available with nilotinib or dasatinib. The identification of patients who would benefit most from discontinuation of TKI treatment is one of the key issues of future research.


The authors would like to thank Arthur Gil (IBE München) and Gabriele Bartsch (Medizinische Fakultät Mannheim, Universität Heidelberg) for their contribution.


S. Saussele: honoraria from Novartis, BMS, and Pfizer; M. Pfirrmann: none.

Copyright information

© Springer Science+Business Media, LLC 2012