Is there still a role for allo-HSCT in CML?

Tyrosine kinase inhibitor (TKI) therapy in CML has dramatically reduced the number of patients requiring allogeneic hematopoietic stem cell transplantation (allo-HSCT). This procedure remains a very effective therapeutic option in chronic phase CML resistant or intolerant to TKIs, and in accelerated phase CML. Advances in transplant procedure and donor availability have increased transplant feasibility and improved outcome. The ability to identify patients whose outcome if treated with TKIs will be poor and continuous improvements in transplant outcome will likely increase the role of allo-HSCT in CML.

Chronic myeloid leukemia: anoverview

Chronic myeloid leukemia (CML) is a clonal myeloproliferative disorder (MPN) of the hematopoietic stem cell (HSC) characterized by the presence of the 22q- or «Philadelphia» (Ph) chromosome, formed by the reciprocal translocation, t(9;22) (q34;q11), encoding the BCR-ABL1 fusion gene. It accounts for 15% to 20% of adult leukaemias, being the most common of the MPN.¹ The worldwide annual incidence of CML is 0.4–2 cases/100000 persons with inter-country variations; it increases with age and has a slight male predominance. The median age at presentation is between 57 and 60 years in Europe.² The majority of patients (90–95%) is diagnosed in chronic phase (CP), a relatively indolent condition easily controlled with treatment. The natural history continues with a bi- or triphasic stage, becoming more aggressive through accelerated phase (AP) and then blast crisis (BC) or directly from CP to BC.^3,4 Several multivariate-derived prognostic models and staging systems have been proposed to help define individual prognosis and allow assigning patients to different strategies of therapy based on risks. The most commonly used are the Sokal and the Hasford scores.^5,6 The ELN have attempted to improve upon these scores using large cohorts of patients treated with tyrosine kinase inhibitors (TKIs) from diagnosis, proposing the EUTOS score⁷ to predict complete cytogenetic response (CCyR) at 18 months. EUTOS has not proved to be a consistent predictor of overall survival (OS) or progression-free survival (PFS), suggesting that CML itself is now a rare cause of death and patients are more likely to die for other medical conditions. The latest EUTOS long-term survival score (ELTS) has been developed to try to predict death from disease.⁸ Although these scoring systems are useful, particularly for choosing first-line therapy, the most important prognostic indicators for CML remain phase of disease at diagnosis and the speed and depth of response to TKI therapy.

Treatment of chronic myeloid leukemia

Before the development of TKIs, CML patients who did not undergo allo-HSCT had a median survival of 3–5 years in the busulfan (BU) and hydroxycarbamide era and 6–8 years in the interferon-alpha (IFN-α) era. Only 30–60% of patients survived beyond 10 years. Early CML therapy included radiotherapy, at the beginning of the 20^th century, and later oral chemotherapy, with haematological responses in 50–80% of thecases. In the early 80s IFN-α showed a survival advantage inducing complete hematologic responses of 40–80%, with cytogenetic response of 15–58% and CCyR of 5–25%, and was recommended as first-line therapy until 2001. Later association of IFN-α with cytarabine produced higher cytogenetic response rate at 12 months (61%), and significantly better survival (5-year rate of 70%).⁹ Nevertheless, these approaches could control the signs and symptoms of CML but could not prevent transformation into a rapidly fatal chemoresistant blastic disease. The first curative treatment that eradicated the Ph+ cells was bone marrow transplant, initially described in syngeneic twins and soon followed by procedures involving HLA-matched siblings and later unrelated donors (URD).^10–12 Autologous HSCT started about at the same time, in the late 70s, aiming to set up the clock to early phase in transformed disease and in patients refractory to IFN-α.¹³ Following the introduction of TKIs, the number of auto-HSCT has decreased rapidly, and it is currently not a recommended strategy in CML. The benefit of allo-HSCT is to provide definitive cure, but the clear disadvantage is its association with considerable morbidity and mortality.¹⁴ Outcome can be improved by better selection of those most likely to benefit. Gratwohl et al. developed the EBMT risk score for patients with CML based on five variables: donor type, disease phase, recipient age, donor/recipient gender combination, and interval from diagnosis to transplant, which together result in a score of 0–7.¹⁵ The EBMT score was highly predictive for survival and transplant related mortality (TRM) in a big validating cohort of 5018 patients transplanted from 1989–1997.¹⁶ During this period, allo-HSCT was the treatment of choice for all patients. Since 2000 allo-HSCT has been replaced by TKIs as frontline therapy, and hence the reasons for patients undergoing transplant are not always clear from registry data. Interpretation of more recent results should be made with some caution. In CML it is worth focusing specifically on the impact of disease phase, because one of the few problems of TKIs therapy is that within the cohort of patients receiving transplants, the proportion transplanted in or after BC has increased over time. In fact, allografts were initially restricted to AP CML, and improvements in survival came only when transplant was performed in CP.¹⁷ The effect of disease phase on the outcome of transplantation has not changed over the years. Nevertheless, an analysis recently repeated by the CMWP of the EBMT for 3497 patients transplanted from 2007 to 2017 confirmed improved outcome of 5-year OS across all-risk scores by 11–26% (Fig.1).¹⁸

Tyrosine kinase inhibor therapy in CML

Development of TKIs has radically changed the standard therapeutic approach for all phases of CML. Nowadays, in western countries, the life expectancy of a newly diagnosed Ph+ CML in CP is very close to that of age-matched individuals in the general population.^19,20 The phase III International Randomized Study comparing imatinib to the combination of IFNα and cytarabine (IRIS study), which leaded to regulatory approval of imatinib in 2001, has been recently updated showing a dramatic change in the prognosis of CML in TKIs era. With a median follow-up of 10.9 years, the OS at 10 years of patients treated with imatinib was 83.3%, and 97.8% when the analysis was limited to CML-related deaths, in patients achieving major molecular response (MMR: <0.1% BCR-ABL1 IS). Moreover, there were low rates of progression to AP or BP in the imatinib group (6.9%, 0%–0.9% per year from year 4 to 8).^21–23 Since 2018, 4 other TKIs have been approved for 1^st and 2^nd line treatment of CML, all representing excellent choices, with different toxicity profiles. The efficacy of 2^nd and 3^d generation TKIs (2GTKI and 3GTKI) has led to their use as 1^st line therapy, and recently completed phase III studies suggest that approximately 80% of patients will achieve CCyR within the 1^st year, compared to 65% on imatinib, with a more rapid achievement of MMR.^24,25 Based on these results, dasatinib, nilotinib and bosutinib have been licensed for use in newly diagnosed patients. The 3GTKI ponatinib, the only one that is effective against T315I mutation, was then approved to treat T315I-mutated CML.²⁶ Other benefits have been reported for 1^st-line use of 2GTKI as compared to imatinib, including the development of fewer mutations conferring TKI resistance and decreased rates of progression to AP and BP.¹ Still, imatinib remains the most cost-effective choice and also the TKI with the longest safety track record. Nevertheless, 2- and 3GTKI can be preferred in specific situations. The response to TKIs has a significant impact on prognosis and response criteria are well defined. Guidelines for milestones for response and recommendations for monitoring and intervention in TKI therapy have been created by ELN and NCCN and are based upon IS molecular response (Fig. 2a,b).^27,28

Importantly, the main goals in CML treatment are:

1. to reach response milestones, in order to ensure normal life span

2. to optimize quality of life while taking daily medication

3. to minimize longer-term potentially irreversible toxicities

4. to achieve such deep and durable molecular responses that TKI discontinuation can be considered, resulting in «treatment free remission» (TFR).

Consequently, even if achievement of response milestones represents the main factor guiding therapeutic decisions, both specific TKIs toxicities and patient’s comorbidities, together with their personal therapeutic goals, have to be taken in account for an optimal CML treatment decision–making, as well as adherence assessment and mutational screening results.^{29, 30}

Allo-HSCT in CML nowadays

Despite the dramatic changes in CML treatment in the last 2 decades, allo-HSCT still has a place in patients cure, ensuring long-term PFS.³¹ This is particularly true as the leukemic quiescent stem cells are not dependent on BCR-ABL signaling for survival, and are not targeted by TKIs, leading to a proportion of patients who will relapse or will have resistant disease despite TKI therapy. A potent graft-versus-leukemia (GVL) effect was early demonstrated in CML by registry studies that showed a decreased risk of relapse in patients developing chronic graft-versus-host-disease (GvHD) and increased relapse rates in those receiving in vivo T-cell depletion (TCD).³² The importance of the GVL effect was subsequently supported by the observation that donor lymphocyte infusion (DLI) could restore durable molecular remissions in CML patients relapsing after allo-HSCT.³³ These findings led to the development of reduced intensity conditioning regimens (RIC), which extended transplant to older and unfit patients with TKI resistant or intolerant CML in whom allograft might previously be associated with unacceptable toxicity.³⁴

Allo-HSCT, once the treatment of choice for CP CML patients, has been relegated to 2^nd-, 3^rd- or 4^th-line treatment after the development of TKIs. With extended follow-up, it appears that 60% of patients can achieve excellent long-term disease control on imatinib, and a small proportion may even be able to stop treatment without experiencing disease recurrence.²² Approximately half of this group will achieve or regain remission on one of the 2GTKI or 3GTKI.^35–37 In addition, 70–80% of patients receiving 2GTKI as 1^st-line therapy will achieve durable responses, of these 20–30% will require ponatinib, since patients who are resistant to 2GTKI have a low probability of achieving a durable response to an alternative 2GTKI.^26,38,39 Allo-HSCT is the therapy of choice for CP CML who failed to respond, develop TKI resistant mutations, and lose an established response and/or are intolerant to the drug. Retrospective registry and single-center studies report 3-year survival rates ranging from 70% to 90% in patients in CP transplanted with a matched sibling donor after a myeloablative conditioning (MAC) regimen.^40,41 The time to proceed to transplant remains controversial, particularly for the substantial number of patients being started on 2GTKI as 1^st-line therapy, who, in case of resistance, progression, or relapse, may be rescued with either another 2GTKI or 3GTKI, and for whom the question to proceed to transplant immediately or wait for another progression and 3^rd-line therapy rescue before allo-HSCT is a matter of debate. This is less true for those who are failing 3^rd-line therapy or have T315I mutation, for whom allo-HSCT is recommended.⁴² Failure to respond to ponatinib after 3 months therapy indicates a patient at high risk of progression, and early transplant is indicated.^43,44 A number of national and international study groups are now reporting that long-term response to imatinib and 2GTKI can be predicted by the rate of fall of BCR-ABL transcript levels (as measured by RQ-PCR at 3 and 6 months).⁴⁵ It is therefore possible to identify patients destined for transplant within the 1^st year of diagnosis while still in CP and return to a more measured approach to it. Moreover, as more potent TKIs move to 1^st-line therapy, patients destined to respond poorly to these drugs are identified earlier, and transplant will return to use as an earlier-line strategy. A small number of patients develop serious cytopenias after exposure to TKIs, with insufficient residual haematopoiesis to repopulate the bone marrow. For these, allo-HSCT is the only effective therapy.

Allo-HSCT remains the therapy of choice for AP CML. For de novo AP patients, search for a donor and referral to a transplant centre should be considered at diagnosis and TKIs serve as a bridge to transplant. Outcome in AP CML treated with TKIs alone compares unfavourably with those achieved in CP (20–40% achieve CCyR and responses are often not durable). Nevertheless, for AP CML without additional cytogenetic aberrations nor elevated blast percentage, HSCT may be deferred depending on response to 1^st-line TKI therapy.^46,47 For patients progressing from CP to AP during treatment, transplant should be considered immediately after obtaining a new response to TKI as their outcome is not good without. Five-years PFS rates of 50%–80% have been reported in patients with AP CML allografted using a sibling or URD.

For patients presenting in, or progressing to BP, the long-term outcome with any of the currently available TKI is poor,^48,49 although a subset of patients with blasts percentage <30% may have outcomes similar to AP patients. Allo-HSCT should be offered after initial control of the disease, since inducing a second CP in BP CML yields outcomes comparable to AP CML.⁵⁰ The addition of a TKI to chemotherapy-based AML or ALL regimens improves the chance of achieving CP.^51,52 Thereafter, patient should proceed to transplant without delay since PFS in BP is low and time to allo-HSCT plays a crucial role.⁵³ Long-term survival rates for patients who achieve a second CP range from 30% to 40% after a transplant that used a MAC regimen. Transplantation in BP is not recommended.

Recently the EBMT-CMWP analysed data of patients transplanted for CML in the 3GTKI era showing that the number of TKIs given prior to allo-HSCT seems not to impact on outcome. However, the stage of the disease, as also reported by the CIBMTR, as also the performance status did have an important impact.^54,55 It is therefore very important to try to keep patients in 1^st CP and avoid progression, even for those rescued to ≥CP2 after having progressed to advanced phase, as the results after allograft are worse in this group.

Source of HSC

About 2/3 of the transplants done nowadays for CML use peripheral blood stem cells (PBSC) as source of HSC, similarly to what is seen in other haematological malignancies.¹² There is no difference in general outcome depending on the HSC source, although bone marrow may have a decrease incidence of chronic GvHD and its severity.⁵⁶ The source of HSC is therefore left open, but PBSC may potentially be preferred to decrease the risk of graft failure and relapse in more advanced disease, particularly with the use of RIC.

Over time, HSCT from all type of donors were performed in CML. Some reports suggest that relapse rate with matchedURDs could be lower than after matched-related transplantation and minor antigen disparity could enhance GVL effect.⁵⁷

Conditioning regimen and GvHD prophylaxis

For CML patients, the best conditioning regimen as well as the best GvHD prophylaxis remain to be determined. Regarding MAC, CY combined either with BU or TBI is still the one that has shown the best overall long-term survival.⁵⁸ RIC has been introduced later and has dramatically increased the number of patients who could potentially benefit from it, resulting in encouraging results over time.^59,60 Still, it did not show improved outcome over MAC, particularly in relation with a higher incidence of relapse.^61,62 Therefore, for elderly or patients with comorbidities, RIC (FLU with BU or MEL) will be the choice, and for the others, particularly with advanced phases, MAC should be proposed for a better control of the disease. For GvHD prophylaxis, CsA with short course MTX remains the standard.⁵⁸ In order to reduce incidence and severity of GvHD, TCD was introduced in the 80s, showing increased relapse rates,³² which led to many groups abandoning TCD in sibling andalso in URD allografts for CML. Others continued to use TCD and have reported good outcomes in sibling transplants, particularly following the introduction of DLI. In a small series of 23 CML patients with a median age of 36 years transplanted with sibling donors and MAC between 1998 and 2016 at the University Hospital of Geneva using partialTCD with alemtuzumab, the 15-year OS and LFS were 95% using the strategy of escalating doses DLI for early molecular relapses, with a low incidence of acute and chronic GvHD (Chalandon, unpublished data).

Strategies after allo-HSCT

Thirty to 70% of patients transplanted for CML will relapse and several factors were identified to determine risk of disease recurrence as AP CML, use of RIC and TCD.⁵⁰ After allo-HSCT, rising or persistently high levels of BCR-ABL1 mRNA can be detected prior to cytogenetic or haematological relapse. Low or falling BCR-ABL1 transcript levels are associated with continuous remission, while high or rising transcript levels predict relapse. Therefore monitoring BCR-ABL1 post-allo-HSCT for CML is of outmost importance to identify patients at higher risk of relapse and permit early therapeutic interventions, even in the long term, due to relapses occurring >15 years post-HSCT.^63,64 Many CML patients will remain RQ-PCRpositive during the first 3 months after allo-HSCT, especially in the RIC era or using TCD. In patients who are at least 4 months post-allo-HSCT, a working definition of molecular relapse is one of the following:

• BCR-ABL/ABL1 ratio higher than 0.02% in 3 samples a minimum of 4 weeks apart

• Clearly rising BCR-ABL/ABL1 ratio in 3 samples a minimum of 4 weeks apart with the last 2 higher than 0.02%

• BCR-ABL/ABL1 ratio higher than 0.05% in 2 samples a minimum of 4 weeks apart⁶⁵

DLI remains the most effective salvage therapy in patients relapsing after allo-HSCT, allowing remission re-induction in 60–90% of patients with CML transplanted and relapsing in CP. The use of escalating doses in case of persistent disease reduces the risk of GvHD.^66–68 An EBMT study showed 69% 5-year survival in 328 patients who received DLI for relapsed CML. DLI-related mortality was 11%, and disease-related mortality was 20%. Some form of GvHD was observed in 38% of patients. Risk factors for developing GvHD after DLI were T-cell dose at first DLI, time interval from transplant to DLI and donor type. In multivariate analysis, GvHD after DLI was associated with an increased risk of death.⁶⁹

With the advent of TKI, CML post-transplant interventions are more complex but give more opportunities to rescue patients. TKI therapy is often effective in the setting of post-transplant relapse and can be used when GvHD is present and DLI is not an option.⁷⁰ It is possible to combine DLI and TKI for relapsing patients; however, the best order (TKI first, DLI first, or both combined) has not yet been defined.⁷¹ The EBMT-CMWP reported 431 patients with CML relapses post-allo-HSCT who received TKI either alone (55%) or in combination with DLI (14.5% before, 4.4% at the same time, and 26% after TKI). Only 42% of the patient obtained either a complete molecular (17.7%), cytogenetic (4.4%), or haematological (20.2%) remission with a 5-year OS of 60% and of 47% for RFS. This rather low response rate may be in relation with the fact that 235 patients were transplanted for advanced phases (AP, BC or > CP1).⁷² Outcomes for patients with AP CML at relapse are not as good. A recent study of 14 AP patients reported CCyR rates of 71% and undetectable BCR-ABL1 transcripts in 57%, either with imatinib or dasatinib treatment alone or in combination with DLI and achievement of undetectable transcripts was very strongly associated with OS.⁷³Therefore, in high risk patients maintenance with TKI is often recommended for at least 1 year after transplant.^74,75

Literature: