Transplantation in CML in the TKI era: who, when, and how?

Chronic myeloid leukemia (CML), a disease of predominantly older adult (age >60 years) and male patients, is in many aspects a model for malignant diseases. Described by Virchow and Bennett in 1845 as leucocythemia, it was the first malignancy with a common chromosomal alteration.^1,2  After molecular identification of the breakpoints, the main molecular pathomechanisms were unraveled.^3,4 

Hematopoietic stem cell transplantation (HSCT) was shown to be the only curative option.⁵  Almost at the same time, treatment with interferon alfa was shown to induce cytogenetic remissions (CyR) in a small proportion of patients.⁶  The use of BCR-ABL1 transcripts for disease monitoring and donor lymphocyte infusion (DLI) to treat relapse after HSCT were the next hallmarks in the treatment of this disease.^7,8  Finally, in 2001 a specific inhibitor of BCR-ABL tyrosine kinase revolutionized the treatment of CML.⁹  In 2010, the discontinuation of TKIs in patients with optimal response was described and the term “operational cure” used.¹⁰ 

After decades of development, HSCT was increasingly used worldwide, and more than 1.5 million transplants (now 90 000 annually) were reported up to 2019.¹¹  Patients with early chronic-phase (chP) CML were considered an ideal indication for HSCT.¹²  Interval-diagnosis HSCT (<12 months and >12 months), disease stage (chP, accelerated phase [AP], and blast crisis [BC]), donor (related and unrelated), and gender differences were recognized as major prognostic factors influencing transplant-related mortality (TRM; nonrelapse mortality caused by graft-versus-host disease [GVHD], infections, or toxicities), and relapse incidence (European Society for Blood and Marrow Transplantation [EBMT] risk score).¹²  Alternative donors became available in 1989 using cord blood and haploidentical donors by ex vivo or in vivo T-cell depletion.^13,14  In 1998 reduced- intensity preparative regimens for older adults and patients with comorbidities were developed.¹⁵  Today, TRM of 5% to 10% in chP CML and 20% to 25% in advanced phase (AdP) CML remain, and GVHD represents a major complication after HSCT. After the availability of TKIs, HSCT activity in the EBMT registry continuously decreased for chP but remained stable for AP/BC (Figure 1). Nowadays, more patients in non-chP than in chP are transplanted worldwide,¹¹  which underlines the importance of optimizing outcome by improving the timing of HSCT.

Oral TKI therapy in patients with chP CML resulted in an overall survival (OS) of 80% at 10 years, but only approximately 50% remained on first-generation (1G) treatment.^9,16  After first, second, and third G-TKIs became available, major molecular remissions (MMR) and even treatment-free remissions were observed.¹⁷  Despite this success, a significant proportion of patients become refractory or progress, become intolerant, or develop cytopenia. Approximately 31% of patients switch from first-line to second-line therapy after 11 months.¹⁸  Reasons for switching include failure in 32%, intolerance in 57%, and other issues in 12% of patients. Recently, asciminib has demonstrated a higher MMR rate in combination with a more favorable safety profile as bosutinib.¹⁹  Despite being well tolerated overall, side effects on TKIs, especially in higher G-TKIs, occur (eg, vasculopathy up to 53% of patients on nilotinib, vascular events in 37% on ponatinib, and pleural effusion in approximately 20% on dasatinib).^20,21  Other studies showed an intolerance of 16% or 7% on treatment with dasatinib (5 years) and/or imatinib (5 and 10 years).^16,22 

Despite HSCT being the only curative treatment, timing assumes an essential role in this process. Decisions should not be taken too early and incur unnecessary risks or too late, which could jeopardize the outcome after HSCT. Results in AdP CML (especially in BC) are poor with single TKI therapy and single HSCT (see below). Recommendations from experts in the field and published results are of fundamental importance in this process.

A 38-year-old construction worker presented with sudden fatigue, nosebleeds, and pneumonia. He had leukocytosis (112 × 10⁹/L), thrombocytopenia (15 × 10⁹/L), and anemia (hemoglobin level of 8.4  g/dL). Initially, an acute myeloid leukemia (AML) was suspected after the bone marrow aspiration revealed 97% blasts with a myeloid phenotype. In addition, a Philadelphia chromosome (Ph) in 20 out of 20 metaphases and a complex aberrant karyotype (including monosomy 7) were detected. A T315I mutation was described in the molecular analyses and a blast crisis (BC) of CML diagnosed.

At a routine checkup, a 60-year-old woman had 30 × 10⁹/L leukocytes, 400 × 10⁹/L platelets, and a hemoglobin level within a normal range. She currently experienced shortness of breath but otherwise felt healthy. After referral to a hematologist, a left shift was found in the differential but no enlargement of the lymph nodes or spleen. Since the Ph chromosome was detected in all metaphases upon bone marrow aspiration, chP CML and a low-risk EUTOS long-term-survival-score was diagnosed.

Outcomes of HSCT have improved substantially during the last decades. In a real-world analysis, the Swedish registry reported an OS of 96% at 5 years in chP1 in patients with a median age of 43 years using matched donors.²³  In another prospective study, the German CML study group reported 3-year OS rates of 88% and 94% after elective HSCT in high-risk chP (but a 0-1 EBMT score) and in imatinib-failure chP CML patients, respectively. TRM of only 8% was described in this multicenter study, which compares favorably to TRM of 26% in an earlier randomized study after IFN-based treatment in chP CML.²⁴  TKI therapy with only imatinib led approximately 96% OS at 3 years in the same study.²⁴  The European LeukemiaNet (ELN) criteria and the US National Comprehensive Cancer Network (NCCN) guidelines are updated on a regular basis and guide the treatment not only in chP CML. The difference between the 2 involves the definition of response to therapy. Transcripts higher than 10% at 3 months define patients with possible TKI-resistant disease and after 3 months with definite TKI-resistant disease (NCCN guidelines). In the ELN guidelines, transcripts higher than 10% are defined as a warning at 3 months and if confirmed within the next 3 months as failure. Failure means changing to another TKI if therapy started with 1G-TKI but to assessment for HSCT only if therapy started with 2G-TKI. In NCCN guidelines, possible TKI resistance automatically leads to evaluation for HSCT.^25,26  On the other hand, the ELN recommendations specify resistance to 2G-TKI as an indication for HSCT.

HSCT results in patients with AP CML are clearly inferior to those in chP1. Before the TKI era, outcomes of HSCT in AP CML were 35% at 2 years and in the imatinib era, 59% at 2 years (compared to 88% and 94% in early chP CML).^27,28  In a prospective study, Jiang et al demonstrated an advantage for HSCT over TKI in AP CML (6-year OS, 83.3% vs 51.4%).²⁹  Similar results of 50% to 60% OS at 5 years after imatinib, nilotinib, and dasatinib (60%-70% OS and 10% MMR at 2 years) and bosutinib (60% OS and 11% MMR at 4 years) were reported with TKI monotherapy.³⁰  Ponatinib had slightly higher response rates (84% OS and 34% MMR at 1 year), but randomized comparisons are lacking. Patients with de novo AP CML treated with nilotinib or dasatinib (70% MMR and 90% OS at 3 years, respectively) have been reported to have superior results than AP CML that develops while on therapy.³⁰  Patients with AP CML under treatment should immediately be considered for HSCT, while de novo AP CML might become eligible for HSCT if the response to TKIs is not optimal.

Recommendations according to the ELN and to the NCCN regarding allogeneic stem cell transplantation are outlined in Table 1.

Although AdP CML occurs in a minority of patients (de novo 10%; 5% on dasatinib and 7% on imatinib develop AdP after 5 years), outcomes are inferior to chP CML after HSCT and TKIs alone. Outcomes of BC HSCT in the pre-imatinib era were reported to be only 21% OS at 2 years.²⁸  1G-TKIs in patients with BC resulted in a median OS of 7 to 10 months, while treatment with 2G-TKIs (nilotinib or dasatinib) resulted in an OS of 32% and 30% at 2 years, respectively. In a retrospective study with 104 patients, 1-3G-TKIs plus intensive chemotherapy (IC) and TKIs plus hypomethylating agents led to a higher rate of CRi (57.5% vs 33%), a higher complete CyR rate (45% vs 10.7%), and more patients proceeding to HSCT (32.5% vs 10.7%) than TKIs or IC alone. Long-term results were similar in the combinations and clearly inferior to TKIs or IC alone (OS, 30%-28% vs 13%-0% at 5 years).³¹  HSCT resulted in long-term OS in patients with advanced CML (34% CI, 23-46, at 15 years). OS was improved in non-BC patients at HSCT with donors 36 years of age or younger and with a higher CD34+ cell dose in the graft.³²  The ELN and NCCN provide information on induction chemotherapy according to AML-based morphology and acute lymphocytic leukemia treatment. The ELN emphasizes the need to attempt to return to chP CML with subsequent HSCT without delay and that patients with untreated BC should not undergo HSCT. A study found that in patients in remission for BC, conventional risk factors such as advanced age, poor performance status, a longer interval from diagnosis to HSCT, myeloablative conditioning (MAC), and unrelated donors remained the major determinants of outcome, whereas in those with active BC at transplant, unrelated donor transplantation was associated with prolonged leukemia-free survival (LFS).³³  Similar results for advanced CML in the AP or BC or pretreated with TKIs beyond frontline therapy have been reported by other investigators and are listed in Table 2.^34-41  The emergence of high-risk additional chromosomal aberrations (ACAs; high-risk ACAs include +8, a second Ph chromosome, i(17q), +19, −7/7q-, 11q23, 3q26.2, and complex aberrant karyotypes) predict a poorer response to TKIs and a higher risk of progression. According to ELN criteria, patients with high-risk ACAs are considered high-risk patients,²⁵  and they should be observed and considered for intensification of treatment, including early HSCT. Gene mutations (RUNX1, ASXL1, IKZF1, WT1, TET2, IDH1/2, CBFB/MYH11, TP53) are found in CML and might be associated with progression to BC.^42,43  Such mutations may lead to a genetically based risk classification in the future with the potential for combination with non–BCR-ABL1 targeted therapy. The molecular landscape of mutations, especially concerning epigenetics, has been featured in recent publications.⁴⁴ 

Optimal preconditions for HSCT are a low CI Sorror score and good performance status, in addition to the requirement for a suitable donor, preferably a human leukocyte antigen (HLA)–compatible sibling followed by an unrelated donor (10 out of 10 matches).⁴⁵  If no matched donor is available, alternative donors, such as haploidentical HSCT with post cyclophosphamide or mismatched donors or cord blood, should be taken into consideration.^46-49 

In addition, HSCT is feasible in patients previously treated with 2G-TKIs with a posttransplant complication rate comparable to that of TKI-naive or imatinib-treated patients.⁵⁰ 

The patient was treated with an AML-like induction chemotherapy in combination with ponatinib, a 3G-TKI to overcome the resistance to the T315I mutation. After reaching chP, the patient was referred to an HSCT program.

After therapy with hydroxyurea, the patient was started on the 1G-TKI imatinib. Her response according to ELN criteria was optimal after 3 months. After she discussed the possibility of HSCT with her hematologist, it was decided to continue TKI therapy under continuous BCR-ABL1 monitoring.

It is generally acceptable to start first-line therapy with TKIs (first or second generation) in chP CML without affecting the outcome of HSCT.²⁵  The response and tolerance to the TKI determine further treatment. The response to the TKI depends on the EUTOS long-term survival score and the presence of high-risk cytogenetics and TKI mutations.⁵¹  The presence of the T315I mutation, which is responsive to ponatinib, is a trigger for HSCT.

The ELN provides regular recommendations on how to manage TKI treatment according to BCR-ABL1 transcript reduction at defined time points.⁵² BCR-ABL1 transcripts higher than 10% at 3 and 6 months of treatment and higher than 1% at 12 months are considered treatment failures and should lead to second-line TKI treatment (either 2G- or 3G-TKI or a TKI responsive to the detected mutation). A donor search should be initiated upon resistance to 2G-TKI treatment.²⁵  Approximately 65% respond to second-line treatment with BCR-ABL1 transcripts of less than or equal to 10%. Nonresponding patients (>10% transcripts at 3 months) have durable CyR of only 50% at 4 years.⁵¹ 

HSCT in early chP CML has been shown to have the best outcome, even after first-line TKI resistance.¹²  In addition, TKI therapy before HSCT has been associated with better posttransplant outcomes.³²  A Center for International Blood and marrow Transplant Research (CIBMTR) study confirmed the beneficial effect of pretransplant TKIs in chP CML for posttransplant survival.⁵³  Delaying HSCT to late chP CML incurs the risk of progression to AdP disease, with unfavorable results. Patients not eligible for HSCT (>80 years of age) should continue on 3G-TKIs or new inhibitors in the testing phase. In pediatric patients, lifelong TKI therapy, including side effects, needs to be balanced with an increased HSCT TRM at the beginning and a possible increased morbidity with chronic GVHD.⁵⁴ 

The goal in advanced CML is a return to a chP followed by HSCT. The choice of pretransplant TKIs for AdP CML is not well standardized, but dasatinib and nilotinib have at least safely been administered before HSCT without increased TRM.^50,55  Thus, it is appropriate to use TKIs to reduce the disease burden before HSCT for AP CML. The search for a donor should be initiated at diagnosis and HSCT planned after a response (Table 2). In BC, a downgrade to chP2 should be considered a clinical condition requiring, whenever possible, treatment with a combination of TKIs with IC or TKIs with hypomethylating agents followed by HSCT.³¹  The combination of TKIs plus IC with HSCT has been shown to result in 54% OS at 2 years in AdP CML.²⁷  HSCT remains a unique therapeutic option for patients in chP CML after the failure of 2 TKIs or in those potentially harboring the T315I mutation (after a trial of ponatinib therapy).⁵⁶ 

After weighing the pros and cons, the patient decided to undergo HSCT from the HLA-matched brother at a JACIE-certified transplant center. Three months after diagnosis, he underwent peripheral blood HSCT.

After 2 years of optimal response, the patient showed increased BCR-ABL1 transcripts. Molecular analyses detected V299L resistant to dasatinib and Y253H resistant to nilotinib, but both were sensitive only to ponatinib. Treatment with ponatinib was started. Since her 70-year-old brother, her only sibling, had CLL, an unrelated donor search was initiated. Following intolerance to ponatinib (causing headaches and vomiting) but after achieving BCR-ABL1 negativity and lacking access to asciminib, an HSCT was considered.

In a CIBMTR study, the prognostic favorable factors for LFS and OS after HSCT in advanced CML were the time from diagnosis to transplantation, good performance status, and access to an HLA fully matched donor.⁵⁷ 

In an EBMT study, risk factors for poorer survival were active BC at transplant, advanced age (>45 years), low performance status, a longer interval from diagnosis to transplant, MAC, and the use of an unrelated donor.³³ 

Other studies reported that a low CD34+ cell count in the graft and a higher donor age were adverse risk factors for OS.²⁷  Although no prospective randomized study exists, reduced intensity conditioning (RIC) seems to be a feasible approach that increases accessibility to the transplant procedure.⁵⁸  Whether younger patients and patients without comorbidities should be transplanted after RIC instead of MAC is still a matter of debate. Similar OS but lower chronic GVHD rates were observed in RIC when compared to MAC.⁵⁹  RIC may be a reasonable alternative to MAC in the TKI era, but results should be confirmed in a randomized study. However, due to the high risk of relapse in AdP CML, patients who can tolerate MAC should receive it.

Ex vivo T-cell depletion is associated with a high relapse risk, but it has been observed that using anti–T-lymphocyte globulin within the conditioning regimen reduces the incidence of GVHD without increasing the risk of relapse.⁶⁰ 

The detection of BCR-ABL1 transcripts early after HSCT has no adverse prognostic significance. However, BCR-ABL1 transcripts (persistently negative, fluctuating, or persistently positive) 6 months post transplant predict a risk of relapse.⁶¹  Several reports suggest that early posttransplant TKIs (including 2G- TKIs) are safe to administer effectively in chP CML but are less effective in AdP CML.^62,63  Maintenance therapy with TKIs with or without DLIs appears to be safe and has been associated with a lower incidence of extensive chronic GVHD.^27,61  Continued regular long-term monitoring of BCR-ABL1 transcripts is required to anticipate the occasional late relapsing patient; however, both the optimal frequency of monitoring and the threshold of BCR-ABL1 transcripts for preemptive therapy with TKIs or DLIs need to be established in the contexts of conditioning regimen and graft manipulation. Although the use of posttransplant TKI therapy is widespread, prospective studies are needed to explore the best TKI dose, the treatment duration, and possible coadministration with DLIs.

A relapse of CML can occur as late as the second decade after HSCT and involves molecular, cytogenetic, and hematological relapse.⁵  Relapsed CML has been treated with DLIs, TKIs, chemotherapy, and second HSCTs.^27,64-67  Combinations of TKI therapy and DLIs can achieve a 5-year postrelapse survival rate of 62%, and some patients became TKI-free, suggesting a persistent graft-versus-leukemia effect.⁶⁷  Molecular remission without GVHD can be achieved, especially when DLIs were given beyond 1 year from HSCT for molecular or cytogenetic relapse.⁶⁸ 

The patient underwent a transplant, after conditioning with fludarabine and myeloablative busulfan, from an HLA-identical brother and, except for neutropenic fever, had an uneventful posttransplant period. During immunosuppression tapering, he developed mild chronic skin GVHD. However, BCR-ABL1 transcripts were still detected 6 months after HSCT. Because of a high risk of relapse, ponatinib was restarted, and the BCR-ABL1 transcripts became negative in the blood. After consultation with his physician following continuous negative BCR-ABL1 transcripts, the patient opted to terminate TKI treatment 5 years post transplant.

A matched unrelated donor was found, and after RIC with busulfan/fludarabine, HSCT was performed without complications. BCR-ABL1 transcripts remained undetectable post HSCT. No TKI prophylaxis was given, and the patient remained BCR-ABL1-negative at 5 years after HSCT.

Rapid developments in the treatment of CML require continual adaptation regarding indications for curative stem cell transplantation. Patients with chP CML should be started with 1G- or 2G-TKIs and monitored according to ELN or NCCN guidelines. The emergence of high-risk ACAs predict a poorer response to TKIs and a higher risk of progression and are triggers for HSCT. Patients with resistant disease (NCCN) or failure (ELN) should be regarded as candidates for HSCT. Patients with resistance to G2-TKIs should be considered for HSCT if they are not responding to G3 after 3 months. An HLA-identical related or fully matched unrelated donor should be selected, and the intensity of the conditioning regimen should be adjusted according to age and comorbidities. Patients in AP and BC CML should be started on IC with the possible addition of TKIs and are candidates for HSCT. Posttransplant monitoring with quantitative polymerase chain reaction testing is indicated to guide prophylactic or preemptive TKI treatment with or without DLIs.

Christian Niederwieser: no competing financial interests to declare.

Nicolaus Kröger: research funding: Bristol Myers Squibb, Jazz, Neovii, Novartis, Riemser; honoraria: Bristol Myers Squibb, Gilead, Jazz, Neovii, Novartis, Riemser, Sanofi.

Christian Niederwieser: nothing to disclose.

Nicolaus Kröger: nothing to disclose.