TKI vs relapse after HSCT: is the jury still out?

In this issue of Blood, Saini and colleagues report that administration of BCR-ABL1 active tyrosine kinase inhibitors (TKIs) after allogeneic hematopoietic stem cell transplantation (HSCT) prevents relapse and improves relapse-free survival in patients with Philadelphia-positive acute lymphoblastic leukemia (Ph⁺ ALL).¹  Their retrospectively collected data represent experience from a large single center and reinforces the conclusion reached by a limited number of previous studies that TKI maintenance after HSCT improves patient outcome.^2-5  Surprisingly, TKIs have so far not been widely recognized to play an important role after HSCT, even though their central role in first-line therapy for Ph⁺ ALL has been recognized for nearly 2 decades. Moreover, relapse accounts for approximately half of treatment failures after HSCT and affects ∼25% of transplanted patients, making this a significant clinical problem. Ironically, the type of randomized trial that could have generated conclusive evidence supporting the use of TKIs after HSCT was prevented by the long-standing conviction that for ALL matched related donor (MRD) is a legitimate trigger for therapeutic intervention. Coupled with the knowledge that MRD after HSCT is a harbinger of relapse,⁶  and overt relapse after HSCT is nearly impossible to salvage, a nonintervention control or a control deemed ineffective such as donor-lymphocyte infusions (DLIs) was considered unethical. The only prospective randomized trial published to date compared prophylactic with preemptive (MRD-triggered) administration of imatinib and demonstrated that both treatment arms were equivalent and superior to historical experience.³  Although this trial also allayed concerns about possible detrimental effects of starting imatinib early after HSCT (eg, due to drug–drug interactions or delayed hematopoietic or immune recovery), the possibility remained that the superior outcome with TKIs compared with historical controls was merely due to patient selection and improved supportive care. This caveat persisted despite data from subsequent single-arm trials and retrospective analyses supporting the central tenant that TKIs should become part of a standard posttransplant management.

Despite the limitations inherent in retrospective analyses, the study by Saini et al adds considerable support to the concept of posttransplant TKIs by providing data on one of the largest patient cohorts studied to date. Against the backdrop of previous data based almost exclusively on imatinib, this study additionally demonstrates that second- and third-generation TKIs are likewise of benefit when given as posttransplant maintenance. Interestingly, the more advanced TKIs appeared equally effective to imatinib in reducing the relapse rate overall but were superior when TKI use was triggered by MRD. Moreover, patient outcome was significantly better with prophylactic than with MRD-triggered TKIs, a finding that differs from results of the prospective randomized trial by Pfeifer et al mentioned above.³  Likely reasons for these differences are a larger proportion of poor-risk patients in the MRD-triggered cohort and higher median MRD levels when TKIs were started. The latter point is suggested by the 4-weeks longer median time to commencing TKI administration in the MRD-triggered groups, ample time for the disease burden to increase in a disease known for rapid disease kinetics. The authors tried to address this bias by restricting their analysis to patients in CMR who were alive and in CMR 3 months after HSCT, but the number of evaluable patients meeting these criteria was small and it is doubtful whether this observed inferiority of an MRD-triggered approach can be extrapolated to all clinical situations. It is worth noting that the median time to molecular relapse in the study by Pfeifer et al was only 4 weeks, and time to MRD-triggered TKIs was 2.3 months compared with 3.2 months in the study reported here.^1,3  Physicians should therefore be acutely aware of the short time window for intervention afforded to them by the nature of this disease and base decisions on which strategy to pursue, prophylactic or MRD triggered, on logistical considerations such as turnaround time for results and probably also MRD levels at the time of HSCT. These were unfortunately not available in the study by Saini et al.

Another topic of profound practical significance for patient management is that of optimal duration of TKI administration after transplant. In their paper, the authors suggest this should be at least 2 years and up to 5 years, which is longer than the time periods chosen in other studies in which ∼ 70% of patients discontinued TKIs prematurely. TKIs are known to be subjectively less well tolerated when given after HSCT than before, even though severe toxicity is distinctly unusual. Although a 2-year treatment period appears operationally reasonable, physicians should be cognizant of the need to reassure and motivate their patients to continue taking the TKIs. Inability to ensure compliance should lead to more frequent MRD monitoring.

Has the verdict on TKI maintenance after HSCT been reached? The balance of evidence in my opinion shifts the question away from whether TKIs should be used posttransplant to how they can be employed for optimal results. With these practical considerations, the devil is in the detail, and numerous questions remain. Will results obtained for myeloablative conditioning regimens apply to reduced intensity conditioning, used more frequently in elderly transplant-eligible patients? Will prior central nervous system involvement or detection of BCR-ABL1 mutations determine efficacy or inform which TKIs should be used? Should DLIs be added in high-risk patients or MRD persistence despite TKIs? Relapse incidence is not completely abrogated by a TKI maintenance approach, and many relapses occur early after HSCT, limiting opportunities for prevention. Parameters will need to be identified that predict which patients are unlikely to benefit from posttransplant TKIs alone. These include HSCT performed beyond CR1, which may be related to disease biology and/or higher pretransplant MRD levels. If MRD thresholds can be identified that are predictive of poor outcome despite TKI maintenance, alternative strategies will need to be considered. These include combining TKIs with immunotherapeutic agents such as blinatumomab and additional pretransplant interventions to reduce MRD before commencing with HSCT. Before the jury can continue to deliberate, we need robust data obtained in prospective trials to address these issues. Although such studies are difficult to conduct, they will provide the evidence we need.