Long-Term Survival after Intensive Chemotherapy or Hypomethylating Agents in AML Patients Aged 70 Years and Older: A Large Patient Data Set Study from Dataml, SAL and Pethema European Registries

The outcome of AML patients (pts) ≥ 70 years is poor. Defining the best treatment option remains controversial especially when choosing between intensive chemotherapy (IC) and hypomethylating agents (HMAs). We set up a multicentric European database collecting data of AML pts ≥ 70 y. The primary objective was to compare overall survival in pts selected for IC or HMAs.

Individual pt data were collected from 3 European AML registries (DATAML, SAL and PETHEMA). All pts ≥70 y newly diagnosed between 01/01/2007 and 06/30/2018 were included. Variables were age, sex, diagnosis date, AML status, WBC, BM blasts %, cytogenetic risk, NPM1, FLT3-ITD mutations, first-line therapy, response, allo-SCT in first complete remission (CR), date of relapse and/or death. First-line treatments included IC, a semi-intensive regimen (fludarabine, cytarabine, filgrastim), HMAs, low-dose cytarabine (LDAC) or supportive care (SC).

3 700 AML pts ≥ 70y were identified. Pts treated with semi-intensive chemotherapy (n=464), LDAC (n=127) or SC (n=837) were not included in this analysis. Thus, the study population included 1 199 IC pts and 1 073 HMA pts. The median follow-up was 49.5 months. In the HMA group, pts were older, had lower WBC count and BM blast %, and they more frequently had ECOG > 1, secondary AML (sAML) and adverse-risk cytogenetics (CG) compared to the IC group. NPM1 and FLT3-ITD mutations were more frequent in the IC group. IC regimens were daunorubicin-AraC (n=432, 36.0%), idarubicin-AraC (n=381, 31.8%) or ida-AraC-CCNU (n=214, 17.8%). AlloSCT was performed in 70 IC pts (5.8%) and only in 7 HMA pts (0.7%) (P<0.001).

CR/CRi was achieved in 673 (56.1%) and 211 (19.7%) pts in the IC and HMA groups (P<0.0001). Multivariate (MV) analysis showed that age ≥ 75 y, ECOG > 1, adverse-risk CG and WBC >30 giga/L were significantly associated with a lower response rate whereas NPM1 mutation was significantly associated with a higher response rate. HMA treatment was associated with a lower response rate than IC (OR, 0.25; 95%CI : 0.20-0.31 ; P<0.001).

Day-60 death occurred in 247 (20.6%) and 194 (18.1%) pts in the IC and HMA groups (P=0.129). MV analysis showed that age ≥ 75 years, ECOG > 1, adverse-risk CG and WBC > 30 giga/L were significantly associated with a higher d60 death rate. HMA treatment was associated with a lower d60 death rate than IC (OR, 0.69; 95%CI : 0.54-0.88 ; P=0.003).

The median OS was 10.9 (95%CI: 9.7-11.6) and 9.2 months (95%CI: 8.3-10.2) in the IC and HMA groups. OS at 1, 3 and 5 y was 46.0 (95%CI: 43.0-48.9) vs. 40.6% (95%CI: 37.6-43.7), 20.8 (95%CI: 18.3-23.4) vs. 8.3% (95%CI: 6.5-10.4) and 12.4 (95%CI: 10.2-14.9) vs 2.8% (95%CI: 1.7-4.4) in the IC and HMA groups. In MV analysis, ECOG > 1, adverse-risk CG, WBC > 30 giga/liter and sAML were significantly associated with a poorer OS.

The treatment effect on OS was time-dependent (Fig 1A). To account for the non-proportionality of risks, we used a Royston and Parmar model, which took into account the interactions between time and treatment effect and allowed graphical representation of the adjusted risk of death at all times during follow-up. This model showed that HMA pts had a significantly lower risk of death before 1.5 months of follow-up ; there was no significant difference between both groups between 1.5 and 4.0 months, and OS was significantly better with IC from 4.0 months of follow-up (Fig 1B). There was no significant interaction between treatment (HMAs vs. IC) and all confounding factors (in particular age, performance status or CG risk).

We also used the propensity score method. A MV logistic regression model was generated to estimate for each pt a propensity score to receive HMAs or IC. The performance of the model was estimated with the c ²-Hosmer-Lemeshow statistic (P-value= 0.169) and the C-statistic (0.82, 95%CI: 0.81-0.84). The mean propensity score was 0.320 (±0.232) in IC (N=1199) and 0.642 (±0.234) in HMA (N=1073). Based on propensity score, 532 subjects with IC were matched with 532 subjects with HMAs. The mean propensity score was the same in IC and HMA (0.491 ± 0.219) in the matched sample. The results of HMAs vs. IC comparisons on response, early mortality and overall survival (Fig 1C-D) in this subgroup of propensity score-matched pts were similar to those of the MV analysis.

With a fairly long median follow-up and a large number of pts, this study shows that IC remains the treatment strategy that offers better chances for prolonged survival compared with HMAs even in AML pts ≥ 70y.

Figure 1

View largeDownload slide

Figure 1

View largeDownload slide

Close modal