The 2017 European Leukemianet Genetic Risk Classification Performs Poorly in Older Patients with Acute Myeloid Leukemia (AML) and Should be Refined to Identify Patients Requiring Additional or Alternative Treatment

AML is a disease of the elderly. Patients aged ≥60 years (y) account for ~65% of all AML patients. However, while 40% of younger AML patients achieve long term survival, only 5-15% of older patients do. Especially given increasing availability of alternative treatment strategies the identification of patients who benefit from standard induction chemotherapy and those who do not is of crucial importance. The 2017 European LeukemiaNet Genetic Risk Classification (ELN GRC) provides guidance on genetic risk stratification of AML patients and is currently routinely used in clinical practice for both younger and older AML patients.

The goal of our study was to test the performance of the 2017 ELN GRC in our cohort of 399 de novo AML patients aged ≥60 y (median: 69 y) which were similarly treated with cytarabine/daunorubicin-based chemotherapy on Cancer and Leukemia Group B/Alliance for Clinical Trials in Oncology protocols. Our cohort did not contain any patients with core-binding factor AML [i.e., with inv(16)/t(16;16) or t(8;21)]. We also aimed to identify additional molecular prognosticators in older AML patients.

Utilizing the 2017 ELN GRC, 31%, 25% and 44% of patients were classified as Favorable (Fav), Intermediate (Int) and Adverse (Adv) risk, respectively. The median CR rates of 81% for Fav risk, 55% for Int risk and 39% for Adv risk separated patients in accordance with the ELN GRC groups (P<.001). However, there was no difference in disease-free (DFS) and overall survival (OS) rates for Int and Adv risk patients (3-year [3y] DFS, Int risk, 6%, Adv risk, 3%, P=.91; 3y OS; Int risk, 11%, Adv risk, 6%, P=.46). Fav risk patients had superior DFS and OS compared to both Int and Adv risk patients, but with 3y rates for DFS of 25% and OS of 29%, outcomes were still relatively poor.

We next tested the prognostic impact of each 2017 ELN GRC marker in our patient cohort. Fav risk group patients with NPM1 mutations (NPM1mut) and either no FLT3-ITD or low FLT3-ITD allelic ratio (FLT3-ITD^no/low) had a CR rate of 84%, while patients with biallelic CEBPAmut had a CR rate of only 54% (P=.02, Table 1). While 3y OS of NPM1mut/FLT3-ITD^no/low was 31%, it was only 15% for patients with biallelic CEBPAmut (P=.13). Within Int risk patients the outcomes of NPM1mut patients with high FLT3-ITD allelic ratio (FLT3-ITD^high) and NPM1 wildtype (NPM1wt)/FLT3-ITD^no/low patients were equally dismal, and actually resembled those of NPM1wt/FLT3-ITD^high patients (Table 1). All other patients with gene mutations classified as ELN Adv risk had poor outcomes (Table 1). Thus, the only genetic classification in 2017 ELN GRC that associated with favorable outcome in AML patients ≥60 years was NPM1mut/FLT3-ITD^no/low, while biallelic CEBPAmut patients resembled more Int risk patients. NPM1mut/FLT3-ITD^high, NPM1wt/FLT3-ITD^no/low andNPM1wt/FLT3-ITD^highpatients all had CR rates in line with Int risk, but very poor DFS and OS.

To potentially refine criteria used to classify patients, we performed outcome analyses of mutations co-occurring with NPM1mut on CR, DFS and OS. In addition to the known adverse impact of FLT3-ITD^high, presence of DNMT3Amut was also associated with lower CR and OS in NPM1mut (CR, 85 vs 32%, P<.001; 3-yr OS rates, 15 vs 7%, P<.001) vs DNMT3Awt patients, resembling the outcomes of Adv risk patients. In contrast, NPM1mut patients that also harbored mutations in SRSF2 (n=18) had longer DFS and OS compared to NPM1mut/SRSF2wt patients (3-yr DFS rates, 44 vs 16%, P=.01; 3-yr OS rates, 50 vs 21%, P=.03).

Thus, with the exception of the small subsets of NPM1mut/FLT3-ITD^no/low and patients harboring both NPM1mut/SRSF2mut all evaluated subsets of AML patients had 3-y DFS and OS rates of <20%, indicating the need for early additional interventions in patients that achieve a CR.

In summary, our data demonstrate the very poor DFS and OS of older AML patients treated with standard chemotherapy, despite the relatively high probability of achieving a CR. We show that the 2017 ELN GRC performs suboptimally to risk-stratify AML patients aged ≥60 years. NPM1mut/FLT3-ITD^no/low status was the only current 2017 ELN GRC prognostic factor associated with improved outcomes in older AML patients. We identified NPM1/DNMT3A (which associated with shorter DFS+OS) and NPM1/SRSF2 (which associated with longer DFS+OS) as additional mutation combinations that might be useful for further group refinement.

View largeDownload slide

View largeDownload slide

Close modal