The Prognostic Impact of HMGCLL1 Gene Variant on Treatment Outcomes in Chronic Myeloid Leukemia (CML) Patients: Adverse Impact on the Response, Failure, and Progression with Imatinib Which Can be Abrogated By the Use of 2nd Generation Tyrosine Kinase Inhibitor (TKI) Upfront Therapy

Background

HMGCLL1 is characterized as a lyase activity enzyme located in the extramitochondrial region. Its biological and physiological role is not fully elucidated. Our previous study (Leukemia 2019) demonstrated that: 1) The HMGCLL1 gene variant, ACGTAATG haplotype (A-haplotype) is associated with increased expression of overall levels and isoform 3 (IS3) of the HMGCLL1 gene and predicts a decreased likelihood of deep molecular response after Imatinib (IM) therapy, 2) blockade of HMGCLL1 IS3 results in G0/G1 cell cycle arrest, leading to increased eradication of CML cell lines and CML leukemic stem cells, 3) blockade of HMGCLL1 isoform 3 significantly suppressed CD34+ fractions in CML patients (pts). It was unclear whether a second-generation tyrosine kinase inhibitor (2G-TKI) can reverse the adverse impact of the HMGCLL1 gene variant in CML pts. The present study assessed the prognostic impact of the HMGCLL1 gene variant in CML pts treated with IM or 2G-TKI, and its impact on treatment-free remission (TFR) after TKI discontinuation.

Patients and methods

Samples from 387 pts were sequenced. Group 1 included 252 pts from 2 centers whose samples were archived retrospectively. These pts were treated with IM (n=190, 75.3%) or second-generation TKIs (n=62) including Dasatinib (n=15, 6.0%), Nilotinib (n=37, 14, 7%), and Bosutinib (n=10; 4.0%). Group 2 included 135 patient samples from 4 countries that attempted to discontinue TKI for TFR.

We applied our internal CML-specific smMIP panel with 332 amplicon probes including 8 HMGCLL1 SNPs (rs10948926, rs10948927, rs9370435, rs4546489, rs4275061, rs9475323, rs9475327, rs9296791). Linkage disequilibrium (LD) was assessed using Haploview software (Bioinformatics 2005).

For Group 1, treatment outcomes were evaluated concerning major molecular response (MMR), molecular response (MR) with 2 log/deeper (MR2) or 4 log/deeper (MR4), failure-free survival (FFS), progression, and overall survival (OS). Disease characteristics and treatment outcomes were evaluated and compared according to the HMGCLL1 variants using a dominant model. For Group 2, molecular relapse-free survival (RFS) was calculated from the date of TKI discontinuation to the date of confirmed loss of molecular response (either loss of major MMR or two consecutive episodes of MR loss). Kaplan-Meier survival estimate was used to evaluate RFS and Cox’s proportional hazard regression model was applied using the dominant model.

Results

1. HMGCLL1 genotype and haplotype

As shown in Fig A, the LD plot showed no recombination among the 8 SNPs, leading to one block for the 8 SNPs to construct a haplotype. Two haplotypes were constructed as CTCAGGCA (C-) or ACGTAATG (A-haplotype) with MAF of 0.674:0.279 in Group 1, and 0.703:0.216 in Group 2, respectively. Summary of outcomes in table 1.

2. Treatment outcomes following TKI therapy according to the HMGCLL1 haplotype

Homozygous A-haplotype (A/A) was associated with worse outcomes, showing a lower MR2 (65.9% vs 78.9% at 12 months, p=0.012), lower MMR (64.4% vs 86.8% at 3 years, p=0.00029), lower MR4 (34.7% vs 62.3% at 5 years, p=0.0001; worse FFS (62.8% vs 77.3% at 3 years, p=0.0014), higher risk of progression (16.2% vs 3.4% at 5 years, p=0.000678), and worse OS (56.1% vs 77.7% at 10 years, p=0.0021).

The adverse impact of A/A-haplotype was more prominent in the pts treated with Imatinib: A/A-haplotype pts showed a 30-50% lower chance of MR2, MMR, and MR4 achievement, and 2-3 times higher risk of TFR and progression in comparison to those having C-haplotype when treated with Imatinib. However, the use of 2G-TKIs could abrogate the adverse impact of the A/A-haplotype.

3. Impact of HMGCLL1 variant on treatment-free remission following TKI discontinuation

While the 12-month RFS rate was 59.9% in overall pts (n=135), an inferior 12-month RFS rate was noted in A/A haplotype (43.8% vs 62.3%) but with borderline significance (p=0.116).

Conclusion: The present study demonstrated that the HMGCLL1 gene variant is confirmed as predictive of CML outcomes. Moreover, our result suggests that the adverse impact of the HMGCLL1 genotype can be abrogated by using 2G-TKIs as upfront therapy. We aim to validate this data in a larger cohort of pts treated with 2G-TKI therapy to reach a clearer conclusion on this result. HMGCLL1 genotype information would be helpful to guide CML therapy given that pts with adverse HMGCLL1 genotypes could benefit from 2G-TKI therapy as front-line therapy.