Adding ex vivo chemosensitivity assay to mutational profiling improves the ability to predict the outcome of patients with AML

Refractory and relapsed acute myeloid leukemia (AML) remain big clinical challenges contributing to patient mortality. Therefore, early identification of patients at high risk of relapse is important. The presence of cytogenetic and molecular aberrations at diagnosis and poor response to therapy are known to be associated with increased risk of AML relapse. They are particularly useful when deciding whether chemotherapy or hematopoietic stem cell transplantation (HSCT) should be used.¹ Based on the prognostic relevance of many recurrent somatic mutations, mutational profiling became incorporated as a prognostic factor in AML European LeukaemiaNet (ELN) guidelines.²

Tailoring novel drugs targeting genetic mutations to patients with a particular sensitivity could improve clinical responses. The limited number of biomarkers associated with drug sensitivity makes predicting patient response before treatment difficult.³ Individual in vitro and ex vivo sensitivity assays have been used in the past but are mainly restricted to clinical trials.^4,5

Previously reported data demonstrate that mutations in FLT3, IDH1, KIT, and RAS have an impact on drug sensitivity.³ However, it remains unclear whether combining molecular profiling with ex vivo chemosensitivity profiling could improve the potential to predict patients that will respond. The clinical relevance of a such combined strategy was explored in patients with AML in a study by Esther Onecha and colleagues, recently published in the British Journal of Haematology.⁶

Study design

Patients with AML (non-M3 type) from a multicentre, non-interventional cohort study of the Programa espaňol de tratamientos hematológicos (PETHEMA) group were profiled using next-gene sequencing (NGS) and ex vivo chemosensitivity flow cytometry assay. In total, 57 (77%) patients were treated with a 3 + 7 schedule and 17 (23%) with fludarabine plus low-dose cytarabine (FLUGA scheme).

Mutational profiling

Patients’ DNA samples (n = 190) from bone marrow or peripheral blood were used in the analysis. All samples were genetically characterized at the time of diagnosis by G-band karyotyping and fluorescence in situ hybridization. Additionally, samples were examined for 32 most commonly mutated genes in myeloid diseases, NMP1 mutations, and internal duplications.

Chemosensitivity profiling

Assessment of live pathological cells was performed on cells extracted from bone marrow samples after incubation with varied concentrations of therapeutic agents:

• Cytarabine (n = 74 patients)
• Idarubicin (n = 74)
• Fludarabine (n = 73)
• Clofarabine (n = 54)
• Daunorubicin (n = 48)
• Mitoxantrone (n = 49)
• Etoposide (n = 42)
• Amsacrine (n = 29)
• 6-Thioguanine (n = 24)
• Decitabine (n = 23)

Main findings

• In total, samples from 190 patients were analyzed by NGS and 74 by ex vivo  assay. Patients characteristics are presented in Table 1

Table 1. Selected patient characteristics⁶

AML, acute myeloid leukemia; CR, complete remission; ELN, European leukaemiaNet (2010); FLUGA, fludarabine + cytarabine; HSCT, hematopoietic stem cell transplantation; MDS, myelodysplastic syndromes; NGS, next generation sequencing; PR, partial remission; tAML, treatment-related AML
	NGS assay (n = 190)	Ex vivo assay (n = 74)
Gender, % Male Female	53 47	60 40
Median age at diagnosis (range), years	57 (18–91)	58 (19–91)
Median blasts at diagnosis (range), %	63 (4–99)	67 (20–99)
AML origin De novo AML-MDS tAML	80 11 9	84 11 5
Cytogenetics Risk Group ELN 2010 Low Intermediate High	7 69 24	15 63 22
HSCT, % Autologous Allogenic None	24 17 59	20 20 60
Induction treatment, % (3 + 7) scheme Azacitidine Decitabine FLUGA scheme Support	81 1 0.5 14 3.5	77 — — 23 —
Response to induction, % CR PR Resistance Death	58 16 10 16	57 26 17 —
Median follow-up (range), months	26 (1–150)	20 (0.5–70)

• Overall, 264 non-recurrent somatic variants were detected in 86% of patients, with single nucleotide variants (SNVs) contributing 82.2% and insertion/deletions 17.8%
  • 19.7% of variants were predicted by algorithms to be pathogenic
  • 57.2% were known to be associated with cancer
• Most frequently mutated genes included
  • NPM1 (in 28.6% of patients)
  • DNMT3A (26.5%)
  • TET2 (21.1%)
  • NRAS (19.5%)
  • FLT3 (25%)
• There was a significant association between the presence of specific mutations and OS (overall survival; based on data for n = 185), with an increased risk of death (HR 3.29; 95% CI, 1.78–6.08; p < 0.0001) in patients harboring at least one mutation in any of the following genes
  • EZH2 (HR 2.44; 95% CI, 1.27–4.86; p = 0.011)
  • KMT2A (HR 2.21; 95% CI, 1.20–4.05; p = 0.011)
  • U2AF1 (HR 3.19; 95% CI, 1.47–6.88; p = 0.003)
  • TP53 (HR 2.92, 95% CI, 1.78–4.79; p < 0.001)
• The cytogenetic status (by ELN 2010 criteria) did not show an impact on OS in the complete or ex vivo study cohorts (p = 0.08 and p = 0.88, respectively)
• There was a significant correlation between the ex vivo classification and clinical response to induction therapy
  • Classification as resistant by the assay agreed with lack of clinical response in 63% of patients
  • Classification as sensitive by the assay aligned with clinical response in 74% of patients
  • Overall correlation was correct in 69% of patients (p = 0.001)
• Presence of certain mutations was associated with ex vivo drug sensitivity while others were associated with resistance. The summary is presented in Table 2

Table 2. Mutations associated with increased and decreased drug sensitivity⁶

Mutated gene	Drug	p value (n)
Mutations associated with increased sensitivity
KMT2A	Idarubicin Fludarabine	0.001 (74) 0.044 (73)
FLT3 FLT3-ITD FLT3-SNV	Daunorubicin 6-Thioguanine	0.007 (48) 0.044 (24)
NPM1	Mitoxantrone Amsacrine	0.029 (49) 0.031 (29)
Mutations associated with resistance
TP53	Fludarabine Mitoxantrone	0.044 (73) 0.045 (49)
U2AF1	Amsacrine 6-Thioguanine	0.032 (29) 0.047 (24)
IDH2	Cytarabine	0.049 (74)
EPOR	Cytarabine	0.043 (74)

• In total, 23 patients were identified as multi-drug resistant, with increased risk of death (HR 2.09; 95% CI, 1.14–3.82; p = 0.017)
• None of the analyzed genes, AML types, prognoses, or cytogenetic risks were found to associate with the multi-drug resistance
• Mutated KMT2A was unique to a multi-drug sensitive group
• Multivariable analysis revealed a higher risk of mortality with
  • the multi-drug resistant profile (HR 2.09; 95% CI, 1.14–3.82; p = 0.017)
  • KMT2A mutations (HR 3.49; 95% CI, 1.16–10.45; p = 0.026)
  • TP53 mutations (HR 3.23; 95% CI, 1.36–7.70; p = 0.008)
• A prognostic index was derived that is able to predict an overall risk custom score based on the integrated information from mutational and chemosensitivity profiles
  • Patients with an overall adverse profile had a greater risk of death than patients with a favorable profile (HR 3.40; 95% CI, 1.69–6.84; p < 0.001)
    • Patients with the adverse mutational and multi-drug resistant profiles (HR 4.82; 95% CI, 1.76–13.25; p = 0.002)
    • Patients with the adverse mutational profile only compared to patients with a favorable mutational profile (HR 4.19; 95% CI, 1.89–9.27; p < 0.001)
    • Patients with the multi-drug resistance profile only compared to patients without the multi-drug resistance profile (HR 2.58; 95% CI, 1.24–5.34; p = 0.011)

Conclusion

The authors developed a new score, combining information about an innate resistance to chemotherapy with mutational analysis. As the impact of somatic mutation present in malignant cells, on drug resistance is still unknown, the ex vivo chemosensitivity assay could help to address that. If validated in a prospective study, the findings could improve the risk stratification of patients with AML and aid in more accurately predicting outcomes of chemotherapy at an early stage in clinical care.