Context. AML patients (pts) with a high number of drivers have a poor prognosis (Papaemmanuil, NEJM 2016). Whether this unfavorable outcome is caused by clonal heterogeneity or by a high mutational load in the dominant clone remains undetermined. So far, the prognostic impact of clonal heterogeneity in AML has only been studied in pts with complex karyotypes, where it worsens prognosis (Bochtler, JCO 2013). We addressed this question in a two-center cohort of AML pts treated with IC.

Methods. We retrospectively evaluated AML pts treated with IC from 2 centers with targeted sequencing of a 43-gene myeloid panel (mean depth 1193X). Variant allele frequencies were adjusted for copy number variation to estimate Cancer Cell Fractions (CCFs). Clone size for each mutation was derived from differences in CCF, assuming a linear accumulation of mutations. Relative clone sizes were then used to derive the Shannon Index (SI), a standard metric of genetic diversity (Maley, Nat Rev Cancer 2017, Figure A). Pts with no gene mutation were not evaluable for SI. Pts with a single mutation have a SI of 0, with increasing SI corresponding to greater dispersal of clone sizes, and thus indirectly of CCFs.

Results. We included 292 pts (median age 57y, 37% ELN-2017 adverse risk), 269 (92.1%) with ≥ 2 mutations. Median number of drivers, including both mutations and cytogenetic alterations (the latter defined as in [Papaemmanuil, NEJM 2016]) was 4. SI increased with the number of drivers (Figure B, p<10-4). In bivariate analyses adjusted on the number of drivers, SI was higher in adverse risk (p=0.0007) but was not associated with age or type of AML (de novo vs secondary).

Median follow-up and EFS were 26.9 and 18.9 months, respectively. In univariate analyses, an increasing number of driver lesions significantly worsened EFS (HR=1.11, p=0.011), while SI as a continuous variable had no impact (HR=0.89, p=0.3). In a multivariate Cox model, the number of drivers (HR=1.74, p=0.002) and SI (HR=0.58, p=0.001), both as continuous variables, had significant but opposite prognostic value independently of age >60y (HR=1.48, p=0.022), WBC > 50x109/L (HR=1.74, p=0.005), and ELN 2017 adverse-risk AML (HR=1.74, p=0.001). There was no significant statistical interaction between number of drivers and SI. OS was also independently poorer in patients with higher number of drivers (HR=1.19, p=0.021) and with lower SI (HR=0.65, p=0.03) in a similar Cox model.

Because of the positive correlation between the number of drivers and SI, but inverse prognostic role, we defined pts with high clonal diversity (132 [49.0%] of the 269 pts with ≥2 driver lesions) as those with a SI higher than the median value of SI of all AMLs with the same number of drivers (Figure B). In those 269 pts, median EFS was 11.3 months in pts (n=73) with ≥5 drivers vs 22.2 months in those with <5 drivers (n=196, p=0.04). EFS was significantly shorter in pts with low clonal diversity (median 10.1 vs. 33.4 months in pts with high diversity, p=0.002). Combining these criteria defined 4 groups with significantly different outcome (log-rank p=0.004, Figure C).

We then compared our estimation of SI with a 'true' SI accounting for further clonal variegation caused by branching and/or LOH in 29 pts with resolved clonal architecture (Hirsch, Nat Comm 2016). Estimated and 'true' SI were highly correlated (Pearson's correlation 0.73, p<10-5), showing that our estimated SI captures most of the genetic heterogeneity of AML.

We finally performed a validation analysis in a publicly available cohort of 1540 AML pts (Papaemmanuil, NEJM 2016). Median OS of pts with low clonal diversity was 20.7 months vs 33.7 months in those with high clonal diversity (p=0.004). In a multivariate Cox model of OS accounting for age >60y, type of AML (de novo vs else) and ELN 2010 risk (FLT3 allele ratio not available for stratification per ELN 2017), the number of drivers (HR=1.17, p<10-5) and SI (HR=0.82, p=0.012), as continuous variables, significantly affected OS in opposite ways.

Conclusion. Our results confirm that a higher number of drivers worsens AML prognosis, but also demonstrate that conversely, higher clonal diversity, estimated with Shannon's Index, confers a better outcome. These findings are consistent with a simple model whereby both a higher driver load and a higher expansion of the most mutated clone independently worsen prognosis in AML. Measuring both clonal diversity and dynamics could improve AML risk stratification.

Figure
Figure
View largeDownload slide
Disclosures

Thomas:ABBVIE: Honoraria; INCYTE: Honoraria; PFIZER: Honoraria; DAICHI: Honoraria. Ades:Abbvie: Membership on an entity's Board of Directors or advisory committees; Astellas: Membership on an entity's Board of Directors or advisory committees; Celgene: Membership on an entity's Board of Directors or advisory committees; Amgen: Research Funding; Silence Therapeutics: Membership on an entity's Board of Directors or advisory committees; Jazz: Membership on an entity's Board of Directors or advisory committees; Agios: Membership on an entity's Board of Directors or advisory committees; Helsinn Healthcare: Membership on an entity's Board of Directors or advisory committees; Takeda: Membership on an entity's Board of Directors or advisory committees; Novartis: Membership on an entity's Board of Directors or advisory committees. Fenaux:Aprea: Research Funding; Jazz: Honoraria, Research Funding; Astex: Honoraria, Research Funding; Celgene Corporation: Honoraria, Research Funding. Boissel:NOVARTIS: Consultancy. Dombret:CELGENE: Consultancy, Honoraria; AGIOS: Honoraria; Institut de Recherches Internationales Servier (IRIS): Research Funding.

Topics:
brachial plexus neuritis, cancer, chemotherapy regimen, clone cells, follow-up, heterogeneity, immune reconstitution inflammatory syndrome, iris, karyotype determination procedure, leukemia, myelocytic, acute

Author notes

*

Asterisk with author names denotes non-ASH members.

© 2019 by The American Society of Hematology
2019
Sign in via your Institution