Standard Risk-Assessment of AML Doesn't Predict the Outcome in Older AML Patients Undergoing Non-Intensive Chemotherapy

Background: Acute myeloid leukemia (AML) still has a poor prognosis for elderly patients, both because of specific characteristics of the disease in this population, and especially because of their frailty that limits the use of intensive treatments (IT). Low dose Aracytine (LDAC) and 5-Azacitidine (5-AZA) are the 2 options currently recommended for these patients, but real-life data are lacking about response rates, survival, and superiority of one treatment over the other.

Methods: In our multi-center study based on the French Hauts-de-France AML observatory, we enrolled all patients with newly diagnosed AML treated in first line with LDAC and 5-AZA.

The primary endpoint was the overall survival (OS). Our secondary objectives were: evaluating the response rates, identifying predicting factors of early mortality and the profile of long-term survivors (i.e alive 6 months after the diagnosis).

Results: Between January 2009 and September 2019, 279 patients who received LDAC (n=87) or 5-AZA (n=192) were included. Median age was 76 years (IQR 70 - 81) (LDAC 79 years (IQR 75 - 83) and 5-AZA 74 years (IQR 68 - 79), p<0.001)) and the Charlson comorbidity index was significantly higher in the LDAC arm: 7 (IQR 6 - 8) versus 6 (IQR 5 - 7) in 5-AZA, p=0.007). Leukocytosis was significantly lower in the 5-AZA group: 3.2G/L (IQR 1.8- 7.7) versus 8.5G/L in LDAC (IQR 2.3-35), p<0.001. Patients treated with 5-AZA predominantly had AMLs associated with myelodysplasia-related changes and adverse cytogenetics according to ELN 2017, whereas those treated with LDAC predominantly had AML NOS with intermediate karyotype. After a median follow-up of 6.9 months (IQR 1.9-11.6), median OS (Figure 1A) was 7.8 months (IQR 2.5 - 13.6): 4.8 months (IQR 2.13 - 14.41) for the LDAC group versus 8.9 months (IQR 3.2 - 13.5) for the 5-AZA group (p=0.046), but there was no significant difference in OS at the end of the study (Figure1B). Complete remission rate was 27.3% without any difference between the 2 groups. At day 30, on the basis of Kaplan-Meier estimates, 25.3% of patients in the LDAC group had died compared to 9.4% in the 5-AZA group (OR 0.28; 95% CI (0.09 - 0.84), p<0.023). After six months of treatment, there was no difference in OS between LDAC and 5-AZA groups (HR 1.37; 95% CI (0.92-2.04), p= 0.116). The predicting factors of mortality at day 30, at 6 months and after 6 months are summarized in the Table 1 (multivariate analysis, odds-ratio). Interestingly, the predictive factors of early mortality were albuminemia less than 30g/L (adjusted OR 6.25; 95% CI (2.08 - 20.00); p<0.001) and treatment with LDAC (adjusted OR 3.57; 95%CI (1.19 - 11.11); p=0.023). At 6 months, ECOG 2-3, high bone marrow blast count at diagnosis and albuminemia less than 30g/L significantly impaired the OS. After 6 months, the only predictive factor of survival was albuminemia greater than 30g/L (adjusted HR 0.65, 95% CI (0.44- 0.96); p=0.030). Usual molecular markers and prognosis classifications (ELN and Lindsley) were not helpful to predict mortality (both early and late) of this specific population.

Discussion: Standard prognostic classifications of AML, such as the ELN 2017 and Lindsley classifications, which have been evaluated in patients eligible for IT, do not appear to be accurate in predicting outcome in older unfit patients treated with LDAC or 5-AZA. Albumin, a simple biomarker, was the strongest predictor of early mortality and prolonged survival beyond 6 months, highlighting that nutritional status and global fitness are the key determinants of survival in older patients non eligible for IT. Overall, the low survival emphasizes the poor prognosis of AML in these patients, and the urgent need for innovative therapies.

No relevant conflicts of interest to declare.