Introduction: Acute myeloid leukemia (AML) is a heterogeneous hematological malignancy with variable clinical outcomes depending on baseline patient and disease characteristics. After achieving complete remission (CR) with induction chemotherapy, eligible patients are treated with consolidation chemotherapy or allogeneic hematopoietic stem cell transplantation (HSCT) to prevent relapse. In patients with intermediate-risk cytogenetics (IRC) without adverse-risk mutations, the decision for post-remission therapy is challenging because of the significant risks of morbidity and mortality associated with HSCT which can offset its benefit in preventing relapse. Therefore, novel biomarkers are needed to support the decision to recommend HSCT in first CR (CR1) in patients with IRC AML. Methods: We used the Leucegene AML prognostic cohort composed of patients diagnosed with de novo IRC AML and treated with intensive induction chemotherapy (N = 316) to identify biomarkers that are predictive for the benefit of HSCT in CR1. Whole-transcriptome sequencing was performed on diagnostic patients' samples. Gene expression data was normalized in transcripts per million (TPM) then log transformed and standardized into Z-scores. We used univariable and multivariable Cox proportional hazards (CPH) regression models for overall survival (OS) and relapse-free survival (RFS) to evaluate the association between gene expression and clinical outcomes. For multivariable analyses (MVA), we included age and white blood cell (WBC) count at diagnosis and NPM1, FLT3-ITD, DNMT3A, biallelic CEBPA, ASXL1 and RUNX1 mutations as covariables. To identify genes predictive of the benefit of HSCT in CR1, we evaluated interaction terms between gene expression and HSCT in CR1 as a time-dependent variable (HSCT-TD) in extended CPH regression models for OS and RFS. We dichotomized the expression of genes based on optimal cutoffs for p value and hazard ratio (HR) for OS and RFS while optimizing their clinical utility to predict the benefit of HSCT. Results: We identified IL1R1 as the top gene associated with RFS in MVA and with a significant interaction with HSCT-TD (p < 0.05). Based on an optimal cutoff of 2.0 TPM, 193 (61%) and 123 (39%) patients had low (< 2.0 TPM, IL1R1low) and high (≥ 2.0 TPM, IL1R1high) expression of IL1R1, respectively. Patients with IL1R1high were older (median age 59 years with IL1R1high vs 54 years with IL1R1low, p < 0.01), had a higher WBC count (44 vs 31 x 109/L, p = 0.06), and a higher frequency of myelomonocytic or monocytic differentiation (43% vs 25%, p < 0.01). FLT3-ITD (54% vs 33%, p < 0.01) and RUNX1 (18% vs 9%, p = 0.04) mutations were more frequent in patients with IL1R1high whereas biallelic CEBPA mutations (1% vs 6%, p=0.04) were more frequent in patients with IL1R1low. High expression of IL1R1 was associated with worse clinical outcomes. The CR rate was 74% and 85% in patients with IL1R1high and IL1R1low, respectively (p = 0.02). The 5-year OS rate was 10% in patients with IL1R1high vs 38% in patients with IL1R1low (HR 2.27, p < 0.01). The 5-year cumulative incidence of relapse (CIR) was 17% higher in patients with IL1R1high (76% vs 59%, p < 0.01). In MVA, high expression of IL1R1 was independently associated with OS (HR 1.80, p < 0.01) and RFS (HR 1.81, p < 0.01). HSCT in CR1 significantly improved OS in patients with IL1R1high (HR for HSCT-TD 0.26, p < 0.01), but not in patients with IL1R1low (HR 0.70, p = 0.17). With a 6-month landmark analysis, the 5-year OS rates were 67% vs 27% among patients with IL1R1high and 62% vs 54% among patients with IL1R1low in patients who underwent HSCT in CR1 or not, respectively (Figure 1). Specifically among patients without FLT3-ITD mutations, patients with IL1R1high benefited from HSCT in CR1 (HR 0.41, p = 0.04) whereas patients with IL1R1low did not (HR 1.03, p = 0.92). Among patients who have undergone HSCT in CR1, the 5-year post-HSCT OS was 60% vs 56% in patients with IL1R1high and IL1R1low, respectively (HR 0.85, p = 0.68), confirming that HSCT may abrogate the adverse impact of high IL1R1 expression. Conclusion: In patients with IRC AML, high expression of IL1R1 is independently associated with worse OS and RFS which may be overcome by HSCT in CR1. Among patients without FLT3-ITD mutations who account for ~60% of IRC AML, those with IL1R1high benefit from HSCT in CR1, as opposed to those with IL1R1low. IL1R1 expression may guide the decision to proceed with HSCT in CR1 in these patients. Figure 1View largeDownload PPTFigure 1View largeDownload PPT Close modal
Read full abstract