Final Report: Somatic Mutations and HMGCLL1 Haplotype Are Not Associated with Molecular Relapse-Free Survival in Patients with Chronic Myeloid Leukemia Who Attempt Treatment-Free Remission

Background

Chronic myeloid leukemia (CML) is a clonal hematopoietic stem cell disorder resulting from the BCR::ABL1 rearrangement. Patients with CML have a comparable life expectancy to that of age-matched individuals in the general population when they achieve an optimal response with tyrosine kinase inhibitor (TKI) therapy. With the aim of improving the quality of life of patients who are significantly affected by drug-related adverse events and financial burden, the practical goal of CML treatment has shifted to treatment-free remission (TFR), defined as the successful discontinuation (DISC) of TKI therapy without losing leukemia control. TKI treatment DISC is a safe procedure but requires appropriate candidate selection and frequent molecular monitoring. Several biomarkers were investigated, but no definite biomarker has yet been successfully identified and validated. We have investigated somatic mutations before TKI discontinuation and HMGCLL1 haplotype as a potential biomarker for TFR, given that they were previously described as predictors of CML outcomes.

Patients and methods

A total of 155 patients who attempted TKI discontinuation for TFR from 5 countries were included: Canada (n=71), Brazil (n=30), Argentina (n=28), Italy (n=23), and the Czech Republic (n=3). We applied our in-house single molecule-molecular inversion probe (smMIP) panel which is a barcoded error-corrected DNA-based sequencing with a limit of detection up to 0.2%. The panel encompasses 40 genes with 332 amplicon probes: epigenetic modifiers (n=7), activation signaling (n=12), myeloid transcription factor (TF; n=5), spliceosome (n=3), tumor suppressor (n=3), cohesion (n=4), miscellaneous (n=3) and 8 HMGCLL1 single nucleotide polymorphisms (SNPs). Two groups were constructed based on these eight SNPs: group 1: CTCAGGCA (C- haplotype) or group 2: ACGTAATG (A-haplotype). Samples were collected prior to TKI DISC when patients met the criteria for TFR attempt. Molecular relapse-free survival (mRFS) was calculated from the date of TKI DISC to the date of confirmed loss of deep molecular response (DMR). Loss of DMR includes single-time loss of major molecular response (MMR) or two consecutive episodes of loss of molecular response 4 log or deeper response (MR4). Kaplan-Meier survival estimate was used to evaluate mRFS and Cox's proportional hazard regression model was applied.

Results

The median MR4 duration prior to TKI DISC was 7.2 years (range 4.7-9.3), while overall TKI treatment duration was 8.5 years (range 6.0-11.6). With a median follow-up of 831 days (range: 460-1,415), out of the 155 patients, 68 (43.9%) lost response, among whom 6 lost MMR only, 12 lost MR4 only, and 50 lost MMR and MR4, both. Median time to MMR loss was 95 days (range 83.5-137). A total of 71 mutations were detected in 51 patients (32.9%). TET2 mutation was the most frequent mutation, detected in 17 cases (24%) followed by ASXL1 (n=14; 20%) and DNMT3A (n=12; 17%). When grouped by relevant biological pathways involved, mutations in epigenetic genes (n=42, 61%) were the most frequently involved. mRFS rate was 64% (95% CI [55.8-71.1%]) and 56% [48.2-64.0%] at 6 and 12 months, respectively. During the univariate analysis, no significant difference was observed in mRFS according to the presence, number, or type of somatic mutations, or the HMGCLL1 haplotype. Additionally, there were no significant differences in mRFS based on age, sex, Sokal risk group, line of TKI therapy, or TKI drug type prior to TKI DISC. However, MR4 duration prior to TKI DISC was associated with mRFS rate: patients with≥7.3 years (n=71) showed higher mRFS: 74.09% [62.05-82.82%] compared to those <7.3 years (n=74) with mRFS rate of 44.36% [32.54 -55.51] (p <0.01, HR 0.36 [0.21-0.63]). In addition, a longer duration of TKI treatment before TKI DISC was associated with higher mRFS: patients with ≥ 8.3 years of treatment (n=85) showed 71.8% mRFS [60.4-80.4], compared to 42.5% mRFS [30.5-54.0%] in those with <8.3 years (n=71) (p<0.01, HR 0.48 [0.31-0.74]). Multivariate analyses for mRFS confirmed that MR4 duration and the total duration of TKI treatment before TKI DISC were associated with TFR outcomes, but not the presence of epigenetic mutations, any mutation, or HMGCLL1 haplotype (Table1).

Conclusion

In CML, somatic mutations before attempting TFR are frequently observed. However, our current study does not support their prognostic significance as TFR-predictive biomarkers.