Introduction

Morphologic complete remission (CR) or composite CR (cCR) requiring bone marrow evaluations (BMEs) remain the gold standard for response assessment in acute myeloid leukemia (AML), chronic myelomonocytic leukemia (CMML), and myelodysplastic neoplasms (MDS). Response criteria still vary by disease subgroup, despite overlapping treatment approaches. Outside randomized clinical trials (RCTs), BMEs are performed in only ~50% of patients (pts) during follow-up [Mukherjee S 2023], underscoring the need for response criteria that do not rely on invasive procedures. Peripheral blood complete remission (PB-CR)—defined as hb ≥11 g/dL, PLT 100–<450 G/L, ANC ≥1 G/L, WBC 1.5–<10 G/L, monocytes ≤1 G/L, 0% PB blasts, and no recent transfusions or growth factors—has shown prognostic equivalence to CR/CRi in hypomethylating agent (HMA) treated cohorts [Pleyer 2023; Bewersdorf 2024]. An IWG23-defined PB-CR with hb≥10 g/dL showed similar results [Pleyer 2024].

Methods

We analyzed pts from the Austrian Myeloid Registry (AMR, n=2210; NCT04438889) and the phase 3 AML-001 RCT (n=488; NCT01074047). The latter evaluated azacitidine vs intensive chemotherapy (IC), low-dose AraC or best supportive care (BSC) in pts with newly diagnosed AML with >30% BM blasts [Dombret, 2015].

Pts were stratified by WHO22 diagnosis [Arber 2022], response type [Zeidan 2023; Pleyer 2023/2024], and 1st line therapy. Overall survival (OS) was estimated via Kaplan-Meier. Multivariable Cox modeling (baseline covariates missing in <150 pts, univariate p≤0.1) was used to adjust (adj) all OS analyses using SAS®9.4.

Results

Pts with complete data for parameters retained in the final Cox model (BM blasts, PB blasts, ECOG-PS, IPSS, age, PLT, hb, PLT transfusions, MCH, bilirubin, secondary disease) formed the primary analysis cohort (n=2042).

Diagnoses included AML (63%, including 11% with secondary AML), CMML (14%), and MDS (23%). 1st line therapy was IC (288), venetoclax +/- HMA (VEN; 6%), HMA (n=58%), others (13%), and BSC (9%). Death at 30 (p=0.64) and 60 days (p=0.81) was similar for all treatment groups. BMEs during 1st line therapy were performed in 74% (RCT) vs 55% (AMR). Cytogenetic (90%) and molecular testing (50%) were performed, with 47% abnormal karyotypes, 77% HR-IPSS and 2.4 mean mutation counts. IC recipients were younger (61 vs 74 yrs), had higher mean PB (24 vs 12%) and BM (62 vs 38%) blasts and more often ECOG-PS 0-1 (92 vs 81%) (all p<0.0001). A greater proportion of BSC pts had <5% BM blasts vs treated pts (54% vs 22%, p<0.0001). Adj OS for BSC was 14.3 [11.7–17.2] months (mo).

Adj OS from 1st line start was 21.5 [95% CI 18.8–24.5], 15.7 [12.6–19.2], 13.2 [12.4–14.1], and 12.7 [11.2–14.4] mo for IC, VEN, HMA, or other regimens, respectively. IWG23-defined PB-CR, ORR and cCR were reached in pts receiving 1st line IC (8, 35, 24%), VEN (28, 61, 34%), HMA (27, 71, 19%) and others (29, 57, 19%). No BSC pts achieved ORR, cCR, or PB-CR. Adj OS for pts achieving IWG23-defined PB-CR, ORR or cCR for pts receiving IC (28.7, 23.7, 24.8 mo), VEN (26.5, 19.4, 27.8 mo), HMA (24.5, 15.6, 23.5 mo), and others (29.9, 16.5, 23.2 mo).

Most IC or VEN recipients had AML. Adj OS for AML pts achieving IWG23-defined PB-CR, ORR or cCR for pts receiving IC (22.3, 19.2, 22.8 mo), VEN (24.8, 17.5, 24.9 mo), HMA (20.6, 13.3, 22.3 mo) and others (26.5, 13.6, 22.1 mo).

In AML pts on HMA, adj OS was 10.7 [8.9–12.2] vs 11.6 [10.3–12.8] mo in RCT (n=216 [published numbers differ due to eligibility criteria defined above]) vs AMR (n=464) (p=0.32), supporting AMR data quality. In HMA-treated AML pts achieving IWG23-defined PB-CR, ORR or cCR, adj OS was 21.3 [17.7–24.8] vs 21.3 [18.5–24.4] (p=1.0), 12.8 [11.4–15.0] vs 14.3 [12.7–16.3] (p=0.26), and 22.8 [19.4–25.8] vs 22.3 [17.7–24.9] (p=0.69) for RCT vs AMR cohorts.

Conclusions

IWG23-defined PB-CR and IWG23 response criteria—originally developed for HR-MDS—are predictive of survival across diagnoses (AML, MDS, CMML) and 1st line therapies (IC, VEN, HMA, others). Their applicability supports harmonizing response definitions across myeloid malignancies. Given that BMEs for response assessment are performed in only ~50% of pts outside trials and ~75% in RCTs, PB-CR offers a clinically relevant, non-invasive alternative. Ongoing work includes extended follow-up to death, incorporation of MDS-001 RCT data (n=358; NCT00071799 [Fenaux 2009]), and development of time-dependent neural network models to address guarantee-time bias.

This content is only available as a PDF.
2025
Sign in via your Institution