The use of tyrosine kinase inhibitors (TKI) in the management of chronic myeloid leukemia (CML) has dramatically improved survival, with some 80% of patients achieving a deep and durable molecular remission (MR). The current focus for these patients is the ability to withdraw long-term treatment and a number of ‘stopping’ studies have been initiated worldwide. Many of these approaches are derived from the French STIM study which showed that 40% of patients who had been real-time quantitative PCR (RT-qPCR) negative for BCR-ABL1 for two years could cease treatment without experiencing disease relapse. However, the RT-qPCR assay used in this study was particularly stringent with a sensitivity of 10−5, compatible with a five log reduction in BCR-ABL1 transcripts (MR5), and it is not clear that the same level of success will result from studies using MR4 and MR4.5 as the indication for treatment cessation. Furthermore, because of the lack of accuracy in RT-qPCR assays when the number of BCR-ABL1 transcripts approach zero, some laboratories report as undetectable, transcript numbers <6 or even <11. In order to investigate the importance of the depth of molecular response on the risk of subsequent disease recurrence, we studied the long-term follow-up of, and RT-qPCR results from, patients who received allogeneic stem cell transplantation as treatment for CML at a time when minimal residual disease detection was performed by RT-qPCR using ABL1 as the control gene. We analysed data from 180 patients transplanted from January 1998 onwards who received an allo-SCT from an HLA-identical sibling or a matched unrelated donor and who had survived for at least 6 months post-transplant with a consistent sequence of 5 or more RT-qPCR results from the time of transplant to the end of follow-up. Patients were assessed on the depth of their MR; 9 categories of ‘complete’ MR were defined based on BCR-ABL1 transcript threshold for negativity (BCR-ABL1=0, BCR-ABL1>0 but <6, BCR-ABL1>5 but <11) and control transcript number (CTN) (CTN>104 but <104.5, CTN>104.5 but <105, but CTN>105). We ranked these categories, firstly by BCR-ABL1 transcript threshold, defining negativity at a lower threshold as a deeper response, and then sub-ranked by CTN, defining a larger CTN as a deeper response. Of the 180 patients, 49 (27%) did not achieve ‘complete’ MR by any definition and for the 131 (73%) patients who did reach some degree of ‘complete’ MR, the median time from transplant to best molecular response was 8.7 months (range, 1.0–103 months). We defined relapse as progression to an RT-qPCR level that triggered the use of donor lymphocyte infusions i.e. BCR-ABL1/ABL ratio exceeded 0.02% in 3 samples, or exceeded 0.05% in 2 samples, or showed rising levels with the last 2 samples higher than 0.02%, or worse (loss of cytogenetic or haematological remission). The 2 year relapse incidence post SCT was 94% in the group who did not achieve any degree of ‘complete’ MR, 94% in the group who achieved MR with BCR-ABL1<11 and >0, CTN>104 (n=32, 17.8%), 55% in the group BCR-ABL1=0, CTN>104 and <104.5 (n=19, 11%), 26% in the group BCR-ABL1=0, CTN>104.5 and <105 (n=47, 26%), and 6% in the group BCR-ABL1=0, CTN>105 (n=33, 18%) (p<0.0001). In multivariate analysis with adjustment for donor type, classifying the 33 patients who achieved BCR-ABL1=0, CTN>105 as the optimal molecular responders the relative risk of relapse was 90.1 in 49 patients who never achieved MR by any definition, (p<0.0001), 21.7 in the group BCR-ABL1<11 and >0, CTN>104 (n=32) (p<0.0001), 8.1 in the group BCR-ABL1=0, CTN>104 and<104.5 (n=19) (p<0.0001), and 2.11 in the group BCR-ABL1=0, CTN>104.5 and <105 (n=47) (p=0.002). In conclusion, fewer detectable BCR-ABL1 transcripts with larger numbers of control transcripts, i.e. a deeper response, predict a lower risk of relapse in post-transplant survivors and may have important implications for the ability to stop long-term TKI therapy.
No relevant conflicts of interest to declare.
Asterisk with author names denotes non-ASH members.