Genomic Profiling of a Phase III Clinical Trial of Interferon Versus Hydroxyurea in MPN Patients Reveals Mutation-Specific and Treatment-Specific Patterns of Response

Background: To investigate the role of genomics in determining response and resistance to front line treatment in MPN, we performed somatic mutational profiling of the DALIAH trial, a randomized controlled phase III trial of interferon versus hydroxyurea in newly diagnosed MPN patients.

Methods: We performed genomic analyses on 202 pre-treatment primary MPN samples obtained from patients enrolled on the DALIAH trial (NCT01387763) and 135 samples obtained after 24 months of treatment. Genomic profiling comprised targeted next generation sequencing (NGS) of 100 genes, selected on the basis of their known or suspected involvement in the pathogenesis of myeloid malignancies, and 1609 informative single nucleotide polymorphisms (SNPs) on chromosome 9p. Clinicohematological response was determined by central review using ELN 2009 (ET, PV and pre-MF) and EUMNET 2005 (PMF) criteria. We evaluated the association of somatic mutations with clinical parameters and with attainment of clinicohematological complete response (CR) at 24 months.

Results: Prior to treatment, 191 of 202 (95%) patients had somatic mutations, including 93% with canonical MPN phenotypic drivers: JAK2 (74%), CALR (14%), and MPL (5%). Among those with JAK2 mutations 37% had JAK2 copy-neutral loss of heterozygosity (JAK2 CN-LOH). Patients with PV were more likely to have JAK2 CN-LOH (p = 0.0001) as compared with patients with other MPN subtypes. At baseline, patients with JAK2 CN-LOH had significantly higher hemoglobin (p = 0.0001), higher white blood cell count (WBC, p = 0.002) and lower platelet count (p=0.0001) than patients without JAK2 CN-LOH.

Mutations in TET2 (24%), DNMT3A (16%), and ASXL1 (10%) were the most frequent co-occurring non-MPN phenotypic driver mutations and they occurred across all MPN subtypes. In addition, 5% of patients had spliceosome gene mutations, and 6% had mutations in genes involved in RAS/MAPK signaling. Patients with TET2, DNMT3A or ASXL1 mutations were significantly older than patients without these mutations (p= 0.0001) and there was a significant association between the presence of a TET2, DNMT3A or ASXL1 mutation and prior stroke (p = 0.004). There were no other significant associations between somatic mutation status and baseline clinical characteristics.

The probability of attaining clinicohematological CR at 24 months was independent of baseline somatic mutations.

Among patients with JAK2 mutations who remained on interferon treatment at 24 months, those with CR had a greater reduction in mean variant allele fraction (VAF) (28% to 8%, p<0.0001) than those who did not achieve CR (34% to 23%, p=0.03). In contrast, the mutant CALR VAF did not significantly decline in either those achieving CR with interferon treatment (16% to 12%, p=0.38) or those not achieving CR (20% to 16%, p=0.35). In patients with JAK2 mutations treated with hydroxyurea, CR at 24 months was associated with a change in VAF (mean 24% to 13%, p=0.04).

Among patients who remained on treatment for 2 years, 44 mutations in 35 patients were newly detected or expanded on serial sampling. DNMT3A mutations were the most frequently acquired, accounting for 41% of new mutations. ASXL1, TET2, PPM1D, TP53, IDH2, and CBL mutations were also recurrently acquired. 97% of patients who acquired new mutations were JAK2-mutant. Among those with acquired DNMT3A mutations, 85% were treated with interferon, and 23% had CR at 24 months. Among those that acquired non-DNMT3A mutations, 38% were treated with interferon and 47% had CR at 24 months. The VAF of newly detected mutations was low (median 1.5%), and half of the patients with newly acquired mutations had at least 50% reduction in JAK2V617F VAF, suggesting that new mutations could either have arisen independently or be subclonal to the dominant JAK2-mutant clone.

Conclusions: Using sequential genomic analyses of a phase III clinical trial of interferon versus hydroxyurea in MPN patients, we found mutation-specific and treatment-specific patterns of response. We uncovered distinct patterns of response to interferon in JAK2-mutant MPN as compared with CALR-mutant MPN. We found that DNMT3A mutations were the most frequent acquired mutations at 24 months and that these were enriched in patients treated with interferon. In aggregate, these results provide insights into molecular response and resistance to interferon and inform the clinical use of interferon in MPN patients.