Managing chronic myeloid leukemia for treatment-free remission: a proposal from the GIMEMA CML WP

Tyrosine kinase inhibitors (TKIs) have been introduced in the therapy of chronic myeloid leukemia (CML) beginning with imatinib in 2001.^1,2  Four other TKIs have been approved between 2001 and 2013,^3-6  and are now available almost worldwide. The first attempt at establishing an internationally shared policy of treatment with TKIs was made by a panel of experts under the heading of the European Leukemia Network (ELN) in 2006,⁷  revised in 2009⁸  and 2013.⁹  After 2013, other data have become available. The results of the IRIS trial, comparing imatinib with interferon-α (IFNα), and of the trials comparing imatinib with the second-generation TKIs (2GTKIs) dasatinib and nilotinib, have been updated.^10-12  Moreover, the results of 2 trials comparing imatinib with another 2GTKI (bosutinib), several retrospective or prospective studies of different TKIs, and many comprehensive reviews^13-33  have been published. The ELN recommendations have not yet been updated. The National Comprehensive Cancer Network (NCCN)³⁴  regularly updates another set of treatment recommendations. The European Society for Medical Oncology has updated its recommendations in 2017.³⁵  The definitions of “optimal response” and “failure” (suggesting no treatment change or a switch to a different TKI, respectively) varied, but the primary goal of all recommendations thus far was to ensure the best survival. However, today, the expectation of patients and doctors has changed, and there is a trend to move the primary goal of CML treatment from survival to the achievement of a condition of treatment-free remission (TFR). Therefore, a debate on which treatment policies should be adopted for achieving TFR is open.

Over the last 20 years, the CML scenario has changed. Initially, due to the limited availability of TKIs and the rarity of the disease, only some referral centers could acquire a robust clinical experience. Today, the management of CML is extended to many centers, also small centers with limited experience, or even to individual practitioners. It is uncertain whether treating for TFR may be allowed in all centers and to every doctors, or if it is still investigational and should be limited to referral centers, but the possibility of achieving TFR can no longer be denied to any patient. The results of several phase 2 single-arm TFR studies,^36-44  individual experience, patient will, as well as commercial pressure, may bias the choice of treatment.^{13,24,26,45-48}  For these reasons, the CML Working Party (WP) of Gruppo Italiano Malattie EMatologiche dell’Adulto (GIMEMA; Italian Group for Hematologic Diseases of the Adult) has developed a project that has involved the representatives of 50 hematologic centers, with the purpose of suggesting a treatment policy aiming to the achievement of TFR.

Fifty Italian centers, responsible for the care of >50% of the Italian CML patients, appointed a representative who participated actively in all steps of this project. A questionnaire addressing 326 key questions was circulated between July and September 2018. All 50 questionnaires were filled in and returned. The responses were analyzed, and the results were discussed in a first meeting held in October 2018. A second questionnaire with 70 questions was elaborated after the first meeting and a second meeting was held on April 2019. After that, the responses to a third questionnaire with 45 questions were analyzed for conclusions. During the meetings, the items receiving the consensus of >70% of panel members were reviewed and finally approved by all members, sometimes with minor modifications. The items that had not received the consensus of >70% of panel members were discussed, and the results of the discussion were included in the second and the third questionnaire. After the analysis of the third questionnaire, the items that had reached a consensus of 70% or more were approved by all members and included in the list of the suggestions. When this consensus was not reached, it was acknowledged that neither evidence nor experience were sufficient to make proposals.

The project was not discussed or negotiated with the ELN panel, and, differently from the ELN recommendations, was specifically and entirely dedicated to the development of treatment policies finalized to TFR. M. Baccarani was no longer the coordinator of the ELN panel. Some senior panel members (M. Baccarani, F. Pane, G.R., G.S., and S. Soverini) who had been previously members of the ELN panel for CML recommendations were involved in the discussion, but they did not fill in the questionnaires. The methodology was quite similar to the Delphi technique.⁴⁹  All items were reviewed, discussed, and approved based on the interpretation of the available data by panel members and on their practice. The final proposals were not graded. No statistical procedures were used.

In the era of conventional chemotherapy, almost all patients died of leukemia. In 1984, a score predicting survival was elaborated by Sokal et al⁵⁰ ; it became familiar worldwide, and was used in the majority of CML trials, until today because it was also able to predict response and survival in patients treated with IFNα and TKIs. Later on, 2 additional scores were proposed, based on patients treated with IFNα (EURO score)⁵¹  and with imatinib (EUTOS score).⁵²  Considering that today many patients no longer die of leukemia,^9,34  a fourth risk score has been elaborated, to predict the probability of leukemia-related death (LRD)⁵³  in patients treated with imatinib, the new EUTOS Long Term Survival (ELTS) score. The ELTS score is based on simple hematologic data, spleen size, and age, just like the Sokal one. It predicts LRD better than Sokal, and the main difference is due to the fact that the negative prognostic weight of age is lower in TKI-treated patients (ELTS) than in patients treated with conventional chemotherapy (Sokal). Therefore, Sokal score classifies inappropriately more elderly patients than ELTS in the intermediate- and high-risk groups, and this is not useful for planning the intensity of treatment. A comparison of Sokal and ELTS scores is shown in Table 1. The panel suggests use of the new ELTS score to assess the baseline CML risk, and suggests that intermediate- and high-risk patients be grouped together.

Several other risk factors^9,34  have been identified, but so far none have entered into clinical use. An important exception is the presence of additional clonal chromosome abnormalities in Ph⁺ cells (additional clonal chromosome abnormalities [ACA]/Ph⁺), including so far +8, +Ph, +19, +21, +17/i17, 11q23, abnormalities of chromosome 3 and 7, and complex karyotypes.^56-62  ACA/Ph⁺ are associated with poor response to TKIs and higher rate of progression.^56-62  They should be enrolled in specifically designed trials. Presently, we suggest that they be classified and treated as high-risk patients.

The age distribution of newly diagnosed patients with CML varies, particularly from western countries, where the median age is close to 60 years, to Asia and Africa, where the median age is lower than 50 years.^63-65  In Europe, <2% of patients are children or adolescents, ∼15% are 18 to 40 years old, ∼50% are 41 to 65 years old, ∼25% are 66 to 80 years old, and ∼6% are >80 years old. Age is an obvious determinant of life expectancy, is associated with comorbidities, and is also a prognostic factor of LRD.^63-65  However, and more importantly, it is the need for TFR that depends on age. Obviously, there are individual variations, but different ages are associated with different life-styles, different parental and family planning problems, different work conditions and career expectations, and different financial issues. The life of a young person is different from the life of an adult, of an elderly, and of a very elderly patient. Having settled that every patient may wish, and has the right, to achieve TFR, the need for TFR is related to age. Therefore, we suggest identifying 5 age groups: children and adolescents (<18 years old), young adults (18-40 years old), adults (41-65 years old), elderly (66-80 years old), and very elderly (>80 years old). It is important to notice that these age boundaries have no prognostic value and do not predict the probability of achieving TFR. Age is important because it is the best objective variable for the evaluation of the individual need for TFR, helping to choice the more convenient treatment policy for TFR. Defining age boundaries rises easy criticisms, but boundaries are always necessary: we all agree to use boundaries when defining responses, molecular, cytogenetic, and hematologic, and the risk score.

General health conditions and several specific comorbidities, irrespective of the relationship with age, are important variables that may influence the treatment plan, particularly in the first-line.⁶⁶  For each TKI, the respective prescribing information alert that some comorbidities may represent a problem requiring careful monitoring, but do not mention specific contraindications. Many comorbidities have been already identified.⁶⁷  The panel was asked to specify which comorbidities can represent an obstacle to the use of any particular TKI in first-line. A panel consensus was reached for nilotinib, dasatinib, and ponatinib. Strong contraindications to nilotinib and ponatinib include a history or a condition of ischemic heart disease, ranging from myocardial infarction to angina (symptomatic, or under medical treatment, or having required a coronary bypass or stents), of peripheral arterial thrombosis, of cerebrovascular events, and of diabetes mellitus. Strong contraindications to dasatinib include a history or a condition of a pleuropulmonary disease, including pulmonary fibrosis, respiratory failure, and pulmonary arterial hypertension. No strong contraindications were suggested for imatinib and bosutinib. If a change of the TKI is planned, in a second or a subsequent line, for resistance to the previous TKI, these comorbidities have only the value of an alert because the risk of leukemia progression may overcome the risk of complications.

Four TKIs, imatinib, dasatinib, nilotinib, and bosutinib, have been approved and are available in first-line, without limitations, although the criteria for the reimbursability of these TKIs differ from country to country. The approval of 2 GTKIs was based on the 1-year results of company-sponsored registration trials^3,4,6  (Table 2). A 5-year update of these trials is available only for dasatinib and nilotinib.^11,12  The results can be summarized as follows: the patients treated with any of the 3 2GTKIs achieved faster and deeper molecular responses, as compared with those treated with imatinib. Progression-free survival (PFS) and overall survival (OS) at 5 years, and the rate of patients switching from the first-line TKI to another TKI, were almost the same with imatinib, dasatinib, and nilotinib. Such data are not yet available for bosutinib. Median age was 46 to 49 years in the DASISION⁴  and ENESTnd³  trials, and 52 to 53 years in the BFORE trial,⁶  far less than the median age of unselected newly diagnosed patients. This point is important because these results may be erroneously interpreted as a general superiority of 2GTKIs, whereas superiority was found in an age-selected population of young and adult patients, so that the difference cannot be extended as such to all patients. A fifth TKI, ponatinib, has been approved (PACE study)⁵  only in second-line (in some countries after failure of 2GTKIs, in some countries also after imatinib failure), or in patients with the T315I mutation. There are no studies comparing a 2GTKI with another 2GTKI, either in first-line or in second-line.

Toxicity and tolerability have been the object of a recent, comprehensive, critical review.⁶⁷  It is relevant to consider that all the comparative trials were designed, sponsored, performed, and analyzed by companies that had an obvious interest in the development of 2GTKIs. Moreover, in all registration studies, the criteria for patient selection, particularly age and comorbidities, and the rigid criteria of treatment adaptation, limit the applicability of the results to “real-life” patients. Toxicity and tolerability were compared only between imatinib and 2GTKIs because there are no studies comparing the 2GTKIs. Based on the interpretation of the available data and on the experience matured with practice, the panel has concluded that nilotinib and ponatinib may expose the patient to metabolic and cardiovascular toxicity, and dasatinib to pleuropulmonary toxicity, whereas the toxicity of imatinib and bosutinib is generally weak, although it may affect considerably the quality of life. Concerning tolerability, the side effects are different according to the TKI, but no major differences were identified, and no consensus was reached.

All TKIs were approved and registered at a fixed dose. A change of the dose is allowed only in case of toxicity or tolerability. Some studies have stressed the importance of taking always the full registered dose and have highlighted the problem of compliance.^68,69  Compliance is important and all patients should take the standard dose, whenever tolerated. However, it has not been confirmed that a full dose is necessary in all patients, when the dose is adapted to tolerance.^70-72  Because the relationship between the dose and the response can easily be monitored, the panel agreed that greater flexibility should be allowed for the use of TKIs at a reduced dose,⁷³  so that in a patient who cannot tolerate the standard full dose, but has achieved an optimal response, the TKI dose should not be increased.

The costs of the TKIs are very different, but are always a heavy financial burden with the exception of the generic formulations of imatinib.^74-76  In Italy, the yearly cost ranges from a few hundreds euro for generic imatinib, to ∼10 000 euro for branded imatinib (Glivec), ∼35 000 euro for nilotinib (Tasigna), ∼45 000 euro for dasatinib (Sprycel) and bosutinib (Bosulif), and ∼75 000 euro for ponatinib (Iclusig). Generic, cheaper formulations of nilotinib and dasatinib will soon be available.

The response is evaluated by cytogenetic and/or molecular tests.⁷⁷  Cytogenetic monitoring requires at least 2 or 3 marrow biopsies, starting from the third month until the achievement of a complete cytogenetic response (CCyR), to obtain the cells for chromosome banding analysis (CBA) of at least 20 marrow cell metaphases.⁷⁷  The panel acknowledged the historic importance of cytogenetic monitoring, but reached a full consensus that cytogenetics alone is no longer sufficient for disease monitoring. However, cytogenetics is always mandatory at baseline, as well as in case of treatment failure and in some selected cases, like in cases with ACA/Ph⁺ at baseline, and cases with atypical transcripts that cannot be quantified. Instead, molecular monitoring requires only a blood sample. It is based on a standardized assay, real-time quantitative polymerase chain reaction (qPCR), which has been internationally recognized, used and validated for >10 years.^78-81  A qPCR should be performed at least every 3 months. Longer intervals would compromise the control of leukemia in the patients who have not achieved a major molecular response (MMR), and would not be sufficient to assess the stability of deep molecular response (DMR) in the patients who may become eligible for TFR.

The 2013 ELN recommendations⁹  proposed defining the response as optimal and failure. In the NCCN guidelines,³⁴  the ELN definition of optimal corresponds to the recommendation to continue the same TKI, whereas the ELN definition of failure corresponds to the suggestion to change the TKI. There are data showing that failure on imatinib can be successfully rescued with 2GTKIs, whereas there are less data in cases of failure on 2GTKIs. However, also in this case, the switch to another 2GTKI is necessary. The ELN suggested a third type of response,” warning,” formerly “suboptimal,” as an alert suggesting a closer monitoring of the response. This definition may be useful for the early response, but not for later responses. We propose substituting “warning” with “nonoptimal,” where “nonoptimal” is not an alert, but opens an option for a switch, the decision of switching depending on the characteristics of the patients (age, health conditions, comorbidities, quality of life, etc). The proposed classification of the response to first-line treatment is shown in Table 3. A comparison of this classification with the ELN 2013 and the NCCN 1.2019 classifications is shown and commented on in Table 4.

The definition of the response to second- and subsequent lines of treatment is useful because up to 50% of newly diagnosed patients may change the first TKI, sooner or later, for several reasons. Because TFR may similarly be the main goal of second-line treatment, the panel agreed to suggest that the response to second-line should be monitored and classified like the response to first-line (Table 5). In third and fourth line, and in case of a late switching in search of a deeper molecular response, the panel agreed that neither evidence nor experience were sufficient to make specific recommendations, so that a personalized evaluation of any single patient is more appropriate.

The choice of the first-line TKI has not yet been settled, and is a matter of debate.^9,34,82,83  Panel members were required to indicate which TKI they are currently prescribing in the first-line setting. The responses are shown in Figure 1, as a percentage of the answers of panel members. Because several members indicated >1 TKI, the total of the responses may exceed 100%. A consensus was reached for the choice of a 2GTKIs, in all young patients (18-40 years) and in intermediate- and high-risk adult patients (41-65 years), and for the choice of imatinib, in low-risk elderly patients (66-80 years) and in all very elderly patients (>80 years). However, in low-risk adults (41-65 years) and in intermediate- and high-risk elderly patients (66-80 years), a consensus was not reached. It should be noticed that the 2013 ELN recommendations did not assign a priority to any TKI. The 1.2019 NCCN guidelines did not assign a priority for low-risk patients, irrespective of age, but suggested some priority to 2GTKIs, including bosutinib, in high- and intermediate-risk ones, not elderly.

All TKIs were approved for second-line treatment based on company-sponsored, phase 2, single-arm trials.⁸⁴  No academic, investigator-initiated studies have addressed a comparison among the available TKIs. Therefore, the data are not sufficient to make suggestions regarding the choice of the TKI in the second- or subsequent-line setting, apart from the differential sensitivity in patients with the presence of BCR-ABL1 point mutations (Table 6). Because the detection of a mutation is the only factor that can specifically guide the choice of another TKI,⁸⁵  when a switch is planned for resistance/failure, a mutational analysis should always be performed using at least Sanger sequencing (SS), and whenever possible using next-generation sequencing (NGS), which is not yet widely available, but being more sensitive than SS^86-90  can avoid a wrong choice in up to 25% of cases. Whether mutational analysis should be performed in the case of nonoptimal response has been debated, but a consensus was not reached. Apart from mutations, the characteristics of the patient (age and comorbidities) and the toxicity profile of the TKIs count.

In the case of failure/resistance, at any time point, including 3 months, the change of the TKI is appropriate, also if the first-line TKI was a 2GTKI. In the case of nonoptimal response, the change from a TKI to another TKI is optional, if it is believed that it may help to achieve a deeper molecular response and to bring the patient to TFR. There are some data,^20,26,32,43  and there is some experience, of switching from imatinib to a 2GTKI, not from a 2GTKI to another TKI. All studies suggest that such a switch may increase the rate of deep molecular response (DMR), hence the number of patients eligible for treatment discontinuation and TFR, but, regrettably, all of these studies are single-arm and do not allow for calculation of the benefit of a switching policy against the risk of increased toxicity. The panel did not reach a consensus, either pro or con.

Many studies, some retrospective and some prospective, have shown beyond any doubt that a consistent proportion, ranging between 30% and 70%, of the patients who discontinue treatment after having achieved a DMR (MR 4.0, BCR-ABL1 ≤0.01%^IS, or MR 4.5, BCR-ABL1 ≤0.0032%^IS) may remain treatment-free for an as-yet undefined period of time.^36-48  It is important to underline that the patients with molecular relapse after discontinuation do not progress, and are almost all able to achieve DMR again upon retreatment, and that some of them can become eligible for a second attempt of treatment discontinuation. The probability of achieving TFR may depend on several factors: leukemia or patient characteristics, TKI type, treatment duration, DMR duration, etc.^36-48  Based on the interpretation of the available data and on their experience, the panel members agreed that neither age nor risk are relevant. To ensure safety and avoid disease progression, the minimum duration of treatment should be 3 years with a 2GTKI and 5 years with imatinib. The minimum duration of a stable DMR should be 2 years, with any TKI. There was no consensus on the “optimal” duration of treatment and of DMR, as well as on the depth of the molecular response, MR 4.0 or MR 4.5. After discontinuation, patients must be monitored monthly by qPCR for at least 6 months,^26,37-44,47  but there was no consensus on the frequency of monitoring after the first 6 months, either monthly, or bimonthly, or every 3 months.

Safe parenting, an important goal of a treatment policy aiming at TFR, has been the subject of recent reviews.^91,92  The panel agreed that TKIs should not be used during pregnancy due to their potential teratogenicity, and must be discontinued as soon as the pregnancy starts. Pregnancy can also be planned in the case of MMR, provided that the MMR has been stable for a minimum of 2 years, and that treatment has been stopped. For male patients, the panel agreed that imatinib, dasatinib, and nilotinib should not be stopped. However, male patients should be advised that data on bosutinib and ponatinib are not yet sufficient.

The first pivotal report on TFR was published in 2007.³⁶  Since then, the interest in TFR has increasingly grown. However, regrettably, no trials were performed to compare the efficacy of different treatment policies and different TKIs for TFR, with the exception of an ongoing GIMEMA trial (SUSTRENIM), investigating the 5-year TFR rate in newly diagnosed CML patients treated first-line with nilotinib or with imatinib followed by switch to nilotinib in the case of nonoptimal response.⁹³  SUSTRENIM results will be available not earlier than 3 years from now, so the debate is still open. Strong pressure exists toward privileging TFR over privileging survival, resulting in a kind of equilibrium that on one hand may be wise, but on the other hand is ambiguous, implying subjective decisions. The GIMEMA CML WP has privileged TFR over survival, as the main goal, for virtually all patients, and has undertaken a project ending with specific treatment suggestions for safe achievement of TFR, preserving patients from exposure to unnecessary potentially more toxic drugs, but maximizing the probabilities of achieving TFR.

Age boundaries have been proposed not for the prognostic value of age, but considering age as a measure of the need of achieving TFR. Strong comorbidities were identified because they are a factor as important as the potency of the TKI, for the choice of the TKI in the first-line setting. For the first time, a specific suggestion has been made on the choice of the first-line TKI. The definitions of the response to first-line treatment, especially at 3 and 24 months, were made more stringent, and were also adopted for the definition of the response to second-line treatment. It is not yet clear whether the same definitions should also be adopted when the primary goal is survival, in patients where TFR cannot be achieved. Sometimes, the data are still incomplete or missing, and the opinions are different, so that it has been acknowledged that sometimes neither evidence nor experience were sufficient to make a suggestion. The debate on the cost-to-benefit ratio of using 2GTKIs is still open, and is hot.^76,94  Our suggestions are not guidelines, and have no professional or legal implications, but may help optimize the treatment of CML for TFR, not only increasing the TFR rate but also limiting toxicity and controlling the cost.

The skilled technical support of Michela Apolinari and Chiara Ferri is kindly acknowledged. This project did not receive any support, even indirect, from pharma companies.

Contribution: M. Baccarani, S. Soverini, E.A., M. Bonifacio, M. Breccia, F. Castagnetti, F. Pane, D. Russo, G.S., P.V., and G.R. designed the project and formulated the questions to GIMEMA panel members; M. Baccarani, S. Soverini, F. Castagnetti, and G.R. collected and ordered the answers to the questions; M. Baccarani, F. Castagnetti, and G.R. calculated the consensus and formulated the proposals; M. Baccarani, S. Soverini, F. Castagnetti, and G.R. wrote the manuscript; and all coauthors filled in the questionnaires, participated personally in the meetings, revised the manuscript, and approved the final version of the manuscript.

Conflict-of-interest disclosure: E.A., S.G., G.G., F.S., M.T., M. Breccia, F. Castagnetti, F. Pane, G.S., and G.R. declare honoraria for consultancy, speaker fees, and logistic (travel/hotel) support for international or educational meetings, from Bristol-Myers Squibb, Incyte, Novartis, and Pfizer. M. Baccarani, A.G., A.I., A.L., P.P., M. Bocchia, and P.V. declare honoraria for consultancy and speaker fees, and logistic (travel/hotel) support for international or educational meetings, from Incyte, Novartis, and Pfizer. V.G. declares honoraria for consultancy and speaker fees, and logistic (travel/hotel) support for international or educational meetings, from Novartis and Pfizer. M.A., G. Beltrami, G. Binotto, I.C., and C. Fava declare logistic support for international or educational meetings from Incyte, Novartis, and Pfizer. C.E. declares logistic support for international or educational meetings from Novartis and Pfizer. G.L.B. declares logistic support for international or educational meetings from Incyte. The remaining authors declare no competing financial interests.

Correspondence: Michele Baccarani, Institute of Hematology “L. and A. Seràgnoli,” S. Orsola-Malpighi University Hospital, Via Massarenti 9, 40138 Bologna, Italy; e-mail: michele.baccarani@unibo.it.