Treatment with the apoptosis inhibitor Asunercept reduces clone sizes in patients with lower risk Myelodysplastic Neoplasms

For the current molecular study, n = 12 MDS patients from the clinical phase I study as performed by Boch et al. (NCT01736436) [13] were included. Detailed information about the design of the clinical trial can be found in the publication [13] and in the patient cohort (Table 1) and patient characteristics table (Table 2). In brief, it was a prospective, open-label, single-arm phase I study which tested the safety, tolerability and pharmacodynamic effects of Asunercept on erythropoiesis in transfusion-dependent MDS patients.

Serial bone marrow biopsies (BMP) were performed within the clinical trial NCT01736436 for screening (Scr), end of treatment (EoT), 12 weeks after EoT (12wFU) and 24 weeks after EoT (24wFU). Additional biopsies for the current study were obtained from initial diagnosis and further follow-up samples if available (Fig. 1a). In total, n = 58 chronological bone mononuclear marrow cells samples (MNCs) from 12 patients were subjected to whole exome sequencing: Thereby n = 1 patient had three available time points, n = 5 patients had four available time points, n = 2 patients had five available time points, n = 3 patients had 6 available time points and n = 1 patient had seven available time points. DNA from matched bone marrow derived stroma cells (MSCs) served as germline controls.

For library preparation the Illumina DNA Flex for Enrichment Kit with probes from the IDT xGen Exome Research panel was used with 150 ng DNA input.

Sequencing was performed on an Illumina NovaSeq 6000 platform in the 150 bp paired-end configuration. The mean coverage was 92.7× with a range of 44.4× to 164× per sample.

The sequencing data was mapped with bwa v0.7.5 and then deduplicated and recalibrated using Picard v1.100 and the GATK toolkit v3.8 [15‐17]. For quality control, the tools Fastqc v0.11.9 and Qualimap v2.2.1 were used [18]. Somatic point-mutations and small Indel mutations were queried with Mutect2 v4.1.3.0 [19]. Variants with a coverage less than 15 reads, a VAF less than 5% in tumor sample and a VAF in the germline control above 1% were excluded from analysis. Copy number changes and large structural chromosomal aberrations were investigated with the Sequenza v2.1.0 [20] tool. Clustering of mutations and reconstruction of individual clones was performed with SciClone v1.1.0 [21]. Graphical images were generated with the R package ggplot2 [22]. Mutational hierarchies were plotted as “Fishplots”, which were generated with the R package fishplots [23].

In the previous phase I clinical trial assessing safety and tolerability of Asunercept in low risk MDS (NCT01736436) [13] none of the patients reached erythroid response as defined by IWG criteria. Nevertheless, the aim of this current study was to gain insights into possible effects on the molecular level in the bone marrow of MDS patients treated with Asunercept. To this end, patients were separable into a group with continuous decrease of transfusion burden between the 1st and 12th week (12wFU) and the 13th and 24th week (24wFU) of follow-up and classified as “clinical responders” versus a group of patients, who displayed stable or increased transfusion requirement of pRBCs classified as “clinical non-responders”. For this study, n = 5 patients were categorized as “clinical responders” and n = 7 patients as “clinical non-responders” (Fig. 1b).

The focus of this study was to identify molecular changes in the bone marrow. A reduction of the dominant clone by a relative proportion of more than 10% between EoT and 12wFU was defined as a molecular responder. Patients not meeting this criterion were classified as molecular non-responders. A VAF (Variant Allele Frequency) reduction of over 10% was chosen to ensure the exclusion of technical fluctuations. Moreover, a threshold of 10% VAF is the minimum requirement for molecular classification within the new 2022 WHO and ICC classifications [24] and has been used as a marker of molecular response in previous clinical trials, e.g. driver mutation reduction in myelofibrosis [25].

To investigate whether Asunercept treatment induces changes of the clonal composition, we performed whole exome sequencing (WES) of sequential bone marrow samples from MDS patients and non-hematopoietic MSCs as a germ line controls to identify potential biomarkers.

In total we detected somatically acquired mutations in 787 genes in our cohort. Of these, 23 genes are known as recurrently mutated in myeloid neoplasia [26]. ASXL1 was the most frequently mutated gene in n = 4 cases, followed by SF3B1 in n = 3 cases (Fig. 2a). Mutated genes were grouped by Gene ontology (GO) pathways [26] or primary literature sources. The most affected cellular process by mutations was the spliceosome machinery with n = 9 (64%) cases in splice factor genes such as SF3B1, U2AF1, SRSF2 and ZRSR2 (Fig. 2b).

On average, the mutational burden per MDS patients in this study was 0.95 mutations per megabase (range: 0.1 to 2.7) at the screening timepoint. N = 2 patients had more mutations in terms of quantity at the time of initial diagnosis with no previous treatment with concomitant lower variant allele frequencies (VAF) than later in the time course (Fig. 2c).

In Fig. 3, the co-occurrence of the n = 30 most frequently mutated genes at any time point is shown. Of note, patients P01-01, P01-02 and P01-05 showed markedly more mutations and more C>A transitions, possibly indicating a different mode of mutational acquisition.

Overall, there was no significant molecular difference between the previously classified clinical responders and non-responders.

In order to identify distinct subclones, which are sensitive to Asunercept therapy, we reconstructed the clonal composition in the bone marrow using SciClone by VAF alterations under therapy and in the course of time (Fig. 4a). In addition to mutational data from exome sequencing, copy number variation (CNV) data was accounted for. These were available from routine cytogenetic assessment and additionally, CNVs were assessed from single nucleotide polymorphisms (SNP) allelic ratios inferred from the WES data using Sequenza (Supplemental Fig. 1). We could identify an average of 3.5 separable clones per patient (range 2–6). In addition, we were able to reconstruct the most likely hierarchical relationship between the clones in n = 3 cases (P01-02, P01-05, P01-17). The same principles as already published in [14] were applied (Supplemental methods). In the remaining cases, mutational VAF changes were not discriminative enough to unequivocally determine their relative order of historic acquisition. Based on the serial WES data, we were able to generate time course plots of mutational variant allele frequencies (VAF) and clinical parameters such as white blood cell count (WBC), hemoglobin levels (Hb), frequency of red blood concentrate (RBC) transfusions and when applicable, platelet concentrates (PC) in correlation to treatment with Asunercept. In the case of P01-02, this patient showed a putative molecular response demonstrated by a decrease in the DNMT3A-driven clone including the PRPF8-driven subclone following treatment while the SMC1A-driven clone remained stable. This coincided with the clinical findings that hemoglobin (Hb) was increased and transfused pRBCs per 30 days were reduced (Fig. 4b). Also in other cases, we observed that while the VAFs of clone-defining mutations remained stable or even increased during the actual treatment duration with Asunercept, they were considerably reduced post treatment in relation to the baseline of End of treatment (EoT) in the 12-week follow-up period (12wFU) (Fig. 5a). In total, the observation of significant VAF reduction of the dominant clone post treatment with Asunercept relative to the baseline of EoT during the 12wFU period by at least 10% was found in 9 out of 12 cases (P01-02, P01-05, P01-06, P01-11, P01-12, P01-15, P01-16, P01-17, P01-20; mean reduction 20%, range: 10.5–39.2%; Fig. 5a, Supplementary Figs. 2–12).

Additionally, we calculated the delta VAF as an absolute change between the Scr and 12wFU timepoint to detect early molecular responders (mean −3.1; range −29.2 to 22.5; Fig. 5b). Interestingly, all n = 3 patients harboring a mutation involved in methylation processes showed an early (P01-05, P01-17) and profound (P01-02) molecular response already during the 12-week treatment period (Figs. 4a, 5a, b and Supplemental Figs. 3, 10). In n = 2 of these cases (P01-02, P01-05, Fig. 4a, Supplemental Fig. 3) this reduction of the dominant clonal VAFs was associated with a reduction of transfused pRBC per 30 days by at least 4 units over 30 days.

For the remaining case (P01-17), an increase in pRBCs transfusions was observed. However, this patient developed sepsis during treatment with Asunercept, therefore the clinical response might be skewed and is hence not evaluable.

All clinical findings and molecular results are summarized in Table 3.

In the preceding clinical phase I study it was already shown that there was no evidence that Asunercept would promote progression to secondary AML on a clinical level. In line with this, none of the n = 12 patients of the current study displayed a significant increase in VAF at the molecular level (Supplemental Figs. 2–12). Even clones that carried a prognostically unfavorable mutation, such as P01-16 with a TP53 p.V173M mutation, showed no expansion in the bone marrow during treatment and follow-up with Asunercept (Supplemental Fig. 9).

As previously stated, in the original publication n = 2 patients had been described with leukemic progression, reflecting a typical course of the disease in an unfavorable risk constellation. Of these, n = 1 patient, P01-15, was further analyzed on a molecular level in this study. Again, we observed no increase of the VAF of any mutation (Supplemental Fig. 8).

In P01-17, sepsis occurred as a complication after 8 doses of Asunercept, which resulted in termination of therapy and granulocyte colony stimulating factor (G-CSF) was applied prophylactically during neutropenia. At the following study visit, an increased blast count from previously below 2% to 5% was observed, which was attributed to the application of G-CSF. At the same time point molecular genetics even showed a decrease of the dominant clone (Supplemental Fig. 10). Furthermore, in this patient, clonal evolution was observed at the 24-week follow-up. Whether this is due to the natural progression of the disease, the administration of GCS-F, or the Asunercept therapy remains unclear.