Prevalence of somatic mutations in peripheral blood, prior to the development of acute myeloid leukemia

Pinkal Desai, from the Weill Cornell Medical College, New York, NY, and colleagues, investigated the premalignant mutational landscape of acute myeloid leukemia (AML) and its influence on risk and time to diagnosis. On 9 July 2018, the study was published in Nature Medicine.

Specific somatic mutations at diagnosis of AML has been described. However, the impact of somatic mutations regarding AML risk and time to disease development has not yet been evaluated. Therefore, Desai and colleagues investigated the likelihood of having a myeloid-specific mutation prior to AML development.

The main objective of the study was to investigate whether specific mutations can be detected a number of years prior to the onset of AML and also to evaluate whether specific mutations will affect the risk and time-to-diagnosis of AML.  

Overall, 725 peripheral blood samples from more than 160,000 women (50–79 years of age at baseline) from the Women’s Health Initiative (WHI) were analyzed. After a median follow-up of 10.8 years, 212 women who eventually developed AML (average time to AML was 9.6 years), and 212 matched healthy non-AML controls, were identified. Targeted deep sequencing was performed on peripheral blood samples at baseline and at a follow-up of 1-, 3-, 6-, and 9-years after study entry.

Key findings:

• Participants who eventually developed AML had more mutations and had higher mutational complexity at baseline evaluation than controls: odds ratio = 4.86 (95% CI, 3.07–7.77), P = 3.8 x 10^-13
• Most frequently found mutations in AML cases and controls: 
  • DNMT3A: 36.7% vs 18.8%
  • TET2: 25.0% vs 5.5%
  • TP53: 11.2% vs 0%
  • SRSF2: 6.9% vs 0%
  • IDH2: 6.4% vs 0%
  • SF3B1: 5.9% vs 1.1%
  • JAK2: 5.3% vs 0.6%
  • ASXL1: 3.2% vs 3.3%
• Spliceosome mutations such as SF3B1, SRSF2 and U2AF1 were identified as a group: 13.8% of cases vs 1.1% of controls
• 46.8% of the cases with AML showed co-mutations compared with 5.5% of controls: odds ratio = 9.01 (95% CI, 4.1–21.4), P = 9.7 × 10⁻¹⁰
• Most common co-mutations in AML cases were:
  • DNMT3A + TET2
  • DNMT3A + SRSF2
  • TET2 + SRSF2
  • IDH2 + SRSF2
• Increased prevalence of myeloid specific mutations was observed among AML cases compared to controls: odds ratio = 2.87, P < 0.001
• A myeloid specific mutation at baseline was associated with the development of AML: OR = 2.9, P < 0.001
• Odds of developing AML were significantly higher at baseline WHI evaluation in participants with mutations in:
  • TP53: odds ratio = 47.2 (95% CI, 2.5–879.1)
  • IDH1 and IDH2: odds ratio = 28.5 (95% CI, 1.4–562.8)
  • Spliceosome genes (SF3B1, SRSF2 and U2AF1): odds ratio = 7.4 (95% CI, 1.7–32.2)
  • TET2: odds ratio = 5.8 (95% CI, 2.6–12.9)
  • DNMT3A: odds ratio = 2.6 (95% CI, 1.5–4.5)
• All participants with IDH1/2 (15/15), RUNX1 with PHF6 (3/3), or TP53 (21/21) mutations eventually developed AML
• Median time to AML was significantly shorter in participants with mutations present at baseline WHI evaluation, P = 0.00045
• Participants with TP53 mutations at baseline were more likely to develop AML within five years: odds ratio = 5.2 (95% CI, 1.9–14.5), P = 0.001
• Patients harbouring a DNMT3A mutation with spliceosome mutations were also more likely to develop AML within five years: odds ratio = 14.8 (95% CI, 1.6–136), P = 0.02
• Time to AML is inversely correlated with the rate of increasing VAF in TP53, IDH2 and DNMT3A

The authors concluded that the presence of mutations was associated with significantly greater odds of developing AML when present at baseline WHI evaluation. Furthermore, mutations in IDH, TP53, spliceosome, TET2 and DNMT3A independently associated with greater odds of AML, adjusting for age and confounding co-mutations. Additionally, time to AML is inversely correlated with increasing VAF in TP53, IDH2, and DNMT3A, which suggests there might be value in monitoring these mutations.

Taken together, this study demonstrated that subclonal stepwise acquisition of mutations may occur several years prior to AML onset, and therefore deep molecular monitoring using NGS may allow early intervention in particular patients.