AUTHORS
Yingwan Luo, MD, PhD, Wei Wang, PhD2, Mixue Xie3, Hongyan Tong, MD, PhD4
AFFILIATIONS
1 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China
2 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China
3 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China
4 The First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China
ABSTRACT
Background: Myelodysplastic neoplasms (MDS) are clinically and molecularly heterogeneous, and outcomes remain poor in high-risk disease. CR rates with hypomethylating agent (HMA)-based regimens are generally ~15-25%, underscoring the need for more effective and tolerable approaches. In real-world practice, older patients or those with prior malignancies, infections, or other comorbidities may be ineligible for transplant, unable to tolerate prolonged BCL-2 inhibitor exposure, or excluded from conventional trials. Lisaftoclax (APG-2575) is a novel BCL-2 inhibitor with a short half-life (~4-5 h), which may enable a controllable short-course strategy. We evaluated 7-day APG-2575 plus azacitidine (AZA) in high-risk MDS and used paired single-cell multi-omics to explore biomarkers and mechanisms of benefit.
Methods: Patients with MDS treated with APG-2575 plus AZA were included. In cycle 1, APG-2575 was given orally with ramp-up (200 mg day 1, 400 mg day 2, 600 mg days 3-7) plus AZA 75 mg/m2 subcutaneously days 1-7. From cycle 2, APG-2575 600 mg days 1-7 plus AZA days 1-7 was repeated. Patients without response after 2 cycles discontinued therapy; responders continued until loss of response or unacceptable toxicity. Responses were assessed by IWG 2023 criteria. CRc was defined as CR + CRh + CRL + CR equivalent, and ORR as CRc + HI + PR. For single-cell analyses, bone marrow mononuclear cells from 22 longitudinal samples from 8 patients and 9 healthy donor (HD) samples were profiled by scFAST-seq (31 samples total), integrating whole-transcriptome single-cell RNA sequencing with targeted mutation profiling. Cell populations were annotated by canonical markers, and mutation calls were mapped across HSC/MPP subclusters.
Results
As of July 29, 2026, 20 patients had received short-course APG-2575 plus AZA. Median follow-up was 6.5 months (range, 1-11), and maximum exposure was 9 cycles. The cohort was clinically challenging: 40.0% (8/20) were >70 years, 10.0% (2/20) had prior malignancies, and 20.0% (4/20) had concurrent infection or severe comorbidities. IPSS-R high/very high risk was present in 80.0% (16/20), and IPSS-M high/very high risk in 95.0% (19/20).
The ORR was 70.0% (14/20), including CRc in 55.0% (11/20) and CR in 30.0% (6/20). Molecular subgroups showed distinct outcomes. Patients with TP53 mutation had an ORR and CRc rate of 28.6% (2/7). In contrast, non-TP53-mutant patients had an ORR of 92.3% (12/13), CRc rate of 69.2% (9/13), and CR rate of 38.5% (5/13). All RUNX1-mutant patients responded (ORR 100%, 6/6), including CR in 66.7% (4/6) and HI in 33.3% (2/6), suggesting that RUNX1 mutation may mark a subgroup more likely to benefit. Safety with hematologic recovery data are being updated.
Given the encouraging clinical activity of APG-2575 plus AZA observed in this pilot study, we performed single-cell analyses. After quality control, 378,168 cells were retained and classified into 16 major hematopoietic/immune populations. Treatment was associated with reductions in HSC/MPP and pDC populations and increases in erythroid cells and megakaryocyte-erythroid progenitors, with the largest shift in HSC/MPP. Among 20,776 HSC/MPP cells, 14 subclusters were identified, including several restricted to MDS. HSC/MPP composition was relatively consistent across HDs but heterogeneous across MDS samples. In responders, dominant HSC/MPP subclusters contracted after treatment and shifted toward a less dominated, more HD-like composition; this pattern was not seen in the non-responder.
Single-cell-derived VAF estimates correlated strongly with adjusted target-NGS VAF (UMI-based Pearson r=0.82; cell-rate-based r=0.83; both P<0.001; n=55), supporting mutation assignment at single-cell resolution. RUNX1- and STAG2-mutant cells preferentially localized to HSC_C01 and HSC_C11, whereas TP53-mutant cells were enriched in MDS-specific subclusters (C03, C08, C14). Clinical response was accompanied by depletion of mutation-defined populations. In RUNX1-mutant patients, RUNX1-mutant HSC/MPP cells decreased in number and proportion after treatment and became undetectable at CR.
Conclusions
Short-course APG-2575 plus AZA showed encouraging activity in this pilot MDS cohort, with especially promising responses in RUNX1-mutant patients. Paired scFAST-seq linked response to HSC/MPP remodeling and clearance of mutation-defined populations, supporting further evaluation of this regimen and single-cell biomarkers to identify patients most likely to benefit.
DISCLOSURES
Yingwan Luo, MD, PhD: No relevant conflicts of interest to declare.