Search

Machine learning and single-cell data inform personalized drug combinations that selectively co-inhibit cancer cell populations., The extensive drug resistance requires rational approaches to design personalized combinatorial treatments that exploit patient-specific therapeutic vulnerabilities to selectively target disease-driving cell subpopulations. To solve the combinatorial explosion challenge, we implemented an effective machine learning approach that prioritizes patient-customized drug combinations with a desired synergy-efficacy-toxicity balance by combining single-cell RNA sequencing with ex vivo single-agent testing in scarce patient-derived primary cells. When applied to two diagnostic and two refractory acute myeloid leukemia (AML) patient cases, each with a different genetic background, we accurately predicted patient-specific combinations that not only resulted in synergistic cancer cell co-inhibition but also were capable of targeting specific AML cell subpopulations that emerge in differing stages of disease pathogenesis or treatment regimens. Our functional precision oncology approach provides an unbiased means for systematic identification of personalized combinatorial regimens that selectively co-inhibit leukemic cells while avoiding inhibition of nonmalignant cells, thereby increasing their likelihood for clinical translation.

Liver fibrosis (LF) is a dangerous clinical condition with no available treatment. Inflammation plays a critical role in LF progression. Glucocorticoid-induced leucine zipper (GILZ, encoded in mice by the Tsc22d3 gene) mimics many of the anti-inflammatory effects of glucocorticoids, but its role in LF has not been directly addressed. Here, we found that GILZ deficiency in mice was associated with elevated CCL2 production and pro-inflammatory leukocyte infiltration at the early LF stage, resulting in enhanced LF development. RNA interference-mediated in vivo silencing of the CCL2 receptor CCR2 abolished the increased leukocyte recruitment and the associated hepatic stellate cell activation in the livers of GILZ knockout mice. To highlight the clinical relevance of these findings, we found that TSC22D3 mRNA expression was significantly downregulated and was inversely correlated with that of CCL2 in the liver samples of patients with LF. Altogether, these data demonstrate a protective role of GILZ in LF and uncover the mechanism, which can be targeted therapeutically. Therefore, modulating GILZ expression and its downstream targets represents a novel avenue for pharmacological intervention for treating LF and possibly other liver inflammatory disorders.

The extensive primary and secondary drug resistance in acute myeloid leukemia (AML) requires rational approaches to design personalized combinatorial treatments that exploit patient-specific therapeutic vulnerabilities to optimally target disease-driving AML cell subpopulations. However, the large number of AML-relevant drug combinations makes the testing impossible in scarce primary patient cells. This combinatorial problem is further exacerbated by the translational challenge of how to design such personalized and selective drug combinations that do not only show synergistic effect in overall AML cell killing but also result in minimal toxic side effects on non-malignant cells. To solve these challenges, we implemented a systematic computational-experimental approach for identifying potential drug combinations that have a desired synergy-efficacy-toxicity balance. Our mechanism-agnostic approach combines single-cell RNA-sequencing (scRNA-seq) withex vivosingle-agent viability testing in primary patient cells. The data integration and predictive modelling are carried out at a single-cell resolution by means of a machine learning model that makes use of compound-target interaction networks to narrow down the massive search space of potentially effective drug combinations. When applied to two diagnostic and two refractory AML patient cases, each having a different genetic background, our integrated approach predicted a number of patient-specific combinations that were shown to result not only in synergistic cancer cell inhibition but were also capable of targeting specific AML cell subpopulations that emerge in differing stages of disease pathogenesis or treatment regimens. Overall, 53% of the 59 predicted combinations were experimentally confirmed to show synergy, and 83% were non-antagonistic, as validated with viability assays, which is a significant improvement over the success rate of randomly guessing a synergistic drug combination (5%). Importantly, 67% of the predicted combinations showed low toxicity to non-malignant cells, as validated with flow-based population assays, suggesting their selective killing of AML cell populations. Our data-driven approach provides an unbiased means for systematic prioritization of patient-specific drug combinations that selectively inhibit AML cells and avoid co-inhibition of non-malignant cells, thereby increasing their likelihood for clinical translation. The approach uses only a limited number of patient primary cells, and it is widely applicable to hematological cancers that are accessible for scRNA-seq profiling andex vivocompound testing.

Introduction Melphalan flufenamide (melflufen), is a novel peptide-drug conjugate that targets aminopeptidases and selectively delivers alkylating agents in tumors. Melflufen was recently FDA approved for the treatment of relapsed/refractory multiple myeloma (MM) patients. Considering the challenges in treating this group of patients, and the availability of several new drugs for MM, information that can support treatment selection is urgently needed. To identify potential indicators of response and mechanism of resistance to melflufen, we applied a multiparametric drug sensitivity assay to MM patient samples ex vivo and analyzed the samples by single cell RNA sequencing (scRNAseq). Ex vivo drug testing identified MM samples that were distinctly sensitive or resistant to melflufen, while differential gene expression analysis revealed pathways associated with response. Methods Bone marrow (BM) aspirates from 24 MM patients were obtained after written informed consent following approved protocols in compliance with the Declaration of Helsinki. BM mononuclear cells from 12 newly diagnosed (ND) and 12 relapsed/refractory (RR) patients were used for multi-parametric flow cytometry-based drug sensitivity and resistance testing (DSRT) evaluation to melflufen and melphalan, and for scRNAseq. Based on the results from the DSRT tests and drug sensitivity scores (DSS), we divided the samples into three groups - high sensitivity (HS, DSS > 40 (melflufen) or DSS > 16 (melphalan)), intermediate sensitivity (IS, 31 ≤ DSS ≤ 40 (melflufen) or 10 ≤ DSS ≤ 16 (melphalan)), and low sensitivity (LS, DSS < 31 (melflufen) or DSS < 10 (melphalan)). To identify genes, responsible for the general sensitivity to melphalan-based drugs we conducted differential gene expression (DGE) analyses separately for melphalan and melflufen focusing on the plasma cell populations, comparing gene expression between HS and LS samples for both drugs ("HS vs. LS melphalan" and "HS vs. LS for melflufen", respectively). In addition, to explain the increased sensitivity of RR samples, we conducted the DGE analysis for ND vs. RR samples and searched for similarities between these three datasets. Results DSRT data indicated that samples from RRMM patients were significantly more sensitive to melflufen compared to samples from NDMM (Fig. 1A). In addition, we observed that samples with a gain of 1q (+1q) were more sensitive to melflufen while those with deletion of 13q (del13q) appeared to be less sensitive, although these results lacked significance (Fig. 1A). After separating the samples into different drug sensitivity groups (HS, IS, LS), DGE analysis showed significant downregulation of the drug efflux and multidrug resistance protein family member ABCB9 in the melflufen HS group opposed to the LS group (2.2-fold, p < 0.001). A similar pattern was detected for the melphalan HS vs. LS comparison suggesting that this alteration might be a common indicator of sensitivity to melphalan-based drugs. Furthermore, in the melflufen HS group we observed downregulation of the matrix metallopeptidase inhibitors TIMP1 and TIMP2 (3-fold and 1.6-fold, p < 0.001, respectively), and cathepsin inhibitors CST3 and CSTB (3.2-fold and 1.3-fold, p < 0.001, respectively) (Fig. 1B). This effect was observed in both "ND vs. RR" and "HS vs. LS for melflufen" comparisons, but not for melphalan, suggesting that these changes are associated with disease progression and specific indicators of sensitivity to melflufen. Moreover, gene set enrichment analysis (GSEA) showed activation of pathways related to protein synthesis, as well as amino acid starvation for malignant and normal cell populations in the HS group. Conclusion In summary, our results indicate that melflufen is more active in RRMM compared to NDMM. In addition, samples from MM patients with +1q, which is considered an indicator of high-risk disease, tended to be more sensitive to melflufen. Based on differential GSEA and pathway enrichment, several synergizing mechanisms could potentially explain the higher sensitivity to melflufen, such as decreased drug efflux and increased drug uptake. Although these results indicate potential indicators of response and mechanisms of drug efficacy, further validation of these findings is required using data from melflufen treated patients. Figure 1 Figure 1. Disclosures Slipicevic: Oncopeptides AB: Current Employment. Nupponen: Oncopeptides AB: Consultancy. Lehmann: Oncopeptides AB: Current Employment. Heckman: Orion Pharma: Research Funding; Oncopeptides: Consultancy, Research Funding; Novartis: Research Funding; Celgene/BMS: Research Funding; Kronos Bio, Inc.: Research Funding.

Introduction Here we report the results of one of the secondary endpoints of the Finnish Myeloma Group-MM02 study; composition of bone marrow (BM) immune cell subsets at treatment start and during lenalidomide (Len) maintenance focusing on 2 different response groups: good responders (GRs) and poor responders (PRs) at pre- and post-treatment stages. We evaluated the BM immune profile with CyTOF (cytometry by time-of-flight). Our hypothesis was that there would be distinct differences in immune cell profiles between patients with good and poor response, especially in the T and NK cell subsets. Patients and methods Twenty-two patients were included in this CyTOF study. Eighteen were NDMM patients from FMG-MM02 study who received 3 RVD cycles followed by ASCT. Len started 3 months after ASCT 10 mg/day in 21/28-day cycles until progression or toxicity. BM samples were collected to the Finnish Hematology Registry Clinical Biobank (FHRB Biobank) at several timepoints; from 18 patients at diagnosis, from 11 patients 1 st sample at good response during Len and 2 nd sample if this good response was maintained and from 5 patients at relapse during Len. The patients in the good response cohort (n=11) had progression-free survival (PFS) > 5 years. For comparison, we included 4 BM samples from the FHRB Biobank, taken at a good response after ASCT from MM patients without exposure to Len. Results With a median follow-up of 81 months (13-97) the median PFS was not reached in the good response (GR) cohort and was 18 months in the poor response (PR) cohort. The 1 st GR samples were collected after a median of 21 (6-46) months of Len. The 2 nd samples in GR cohort were collected after a median of 56 (45-67) months of Len. The PR samples were taken after a median of 6 (2-23) months of Len. CyTOF analysis revealed distinct good and poor responder's immune signatures at baseline level. GR, baseline group has shifted phenotype of T cells toward the CD8 T cells, expressing markers, attributed to the cytotoxicity (CD45RA, CD57), as well as having slightly higher abundance of CD8 TE and lower abundance of CD8 naïve T cells. Total T cells amounts were significantly higher in GR. Increased expression of CD56, CD57, and CD16 were also seen on NK cells in GR at the baseline, indicating both maturation and cytotoxic potential of NKs. In contrast, a significant decrease of CD56 and CD16 expression suggesting reduced cytotoxic potential and increase of CD57 were seen on NKs in PR, baseline indicating senescence status a phenotype associated with exhaustion. By using viSNE we discovered two novel populations - MAC1, and MAC2, - associated with disease pathogenesis. We further investigated the composition of these populations, and it appears that MAC1 expressed CD38+CD56+CD45-CD19- presumably malignant plasma cells and MAC2 expressed CD45- CD14+CD38+CD56+TCRgd+CCR7+CCR6+CXCR4+ CXCR3+CXCR5+CD294+, which might be the novel populations, having the features of myeloid suppressor cells. These two immune cells were higher in PR (9.2% and 8.2%, respectively) at baseline as compared to the GR (2.3% and 1.7%). NKs, T cells, and B cells showed a high expression of monocytic marker CD14 and chemokine receptors (CCR7, CCR6, CXCR4, CXCR3, and CXCR5) in PR both at baseline and relapse. This might be attributed to either biological features of MM environment, or the high level of interaction between these populations and monocytes, which left traces of their membranes on the non-monocytic populations. Conclusions Patients, responded to the treatment, have higher abundances of effector/cytotoxic cells, expressing higher levels of CD57 and/or CD45RA for T cells, and CD57, CD16, and CD56 for NK cells indicating proper differentiation and maturation of T, and NK cells to effector and cytotoxic subsets. Additionally, those patients have less degree of tumor burden as well as decreased expression of chemokine receptors. During the therapy administration, good responders show the increase in effector memory CD4 and CD8 subsets of T cells abundance, indicating even the higher cytotoxic effect of immune system. Disclosures Silvennoinen: Amgen: Consultancy, Honoraria, Research Funding; Celgene/BMS: Consultancy, Honoraria, Research Funding; Janssen: Consultancy, Honoraria; Sanofi: Consultancy, Honoraria; Takeda: Consultancy, Honoraria, Research Funding. Luoma: Amgen: Honoraria; Janssen: Honoraria; Incyte: Honoraria. Anttila: Amgen: Honoraria; Celgene: Honoraria; Janssen: Honoraria; Takeda: Honoraria. Säily: Takeda: Honoraria; Janssen: Honoraria; Sanofi: Honoraria; Celgene: Honoraria. Partanen: Takeda: Honoraria; Abbvie: Honoraria; Behring: Honoraria. Heckman: Novartis: Research Funding; Orion Pharma: Research Funding; Celgene/BMS: Research Funding; Oncopeptides: Consultancy, Research Funding; Kronos Bio, Inc.: Research Funding.

Spleen tyrosine kinase (SYK) is a non-receptor tyrosine kinase that mediates integrin and Fc receptor signaling in myeloid cells. SYK has been implicated as an oncogenic driver in acute myeloid leukemia (AML) with aberrant expression of HOXA9 and MEIS1 and cooperates with FLT3 internal tandem duplication to drive leukemogenesis. The oral SYK inhibitor entospletinib (ENTO) has demonstrated clinical activity in HOXA9/MEIS1 driven AML and is currently being investigated in a phase 3 trial of previously untreated patients with nucleophosmin1-mutated (NPM1 mut) AML. Lanraplenib (LANRA) is a next generation oral SYK inhibitor with potency and selectivity comparable to ENTO. In healthy volunteers and patients with autoimmune disease, LANRA has shown pharmacokinetic properties that compare favorably with ENTO. To support the clinical development of LANRA for the treatment of AML, ex vivo treatment of patient-derived AML cells was used to compare its activity to that of ENTO, both as a single-agent and in combination with other AML therapies. First, ENTO and LANRA single-agent activities were evaluated in peripheral blood-derived blasts from 15 AML patients, representing different mutational backgrounds including NPM1, FLT3, PTPN11, and NRAS mutations. AML cells were seeded into 96 well plates and treated with ENTO and LANRA for 6 days. Comparable effects on viability were observed across the 15 models with the 2 compounds, and in 11 of the models, the half maximal inhibitory concentration (IC 50) values were within 2-fold of each other. ENTO had a slightly lower IC 50 value than LANRA in the FLT3-mutated models possibly due to the direct FLT3 inhibitory activity of ENTO. Next, we tested the activity of ENTO and LANRA ex vivo in bone marrow-derived AML blasts from 29 AML patients representing diverse mutational backgrounds, including NPM1, IDH1, FLT3, and RAS mutations as well as MLL rearrangements. The models were treated for 9 days with either ENTO or LANRA, and viability was assessed using Annexin V and 7-aminoactinomycin D staining. Again, ENTO and LANRA showed comparable effects on cell viability with no significant differences between the compounds when compared across the different mutational backgrounds. Both studies suggest the potential for anti-leukemic activity in several different genetically defined subsets of AML. Matrix combination assays were performed by combining ENTO or LANRA with either cytarabine (NPM1 mut), gilteritinib (FLT3 mut), or trametinib (RAS mut) with cell viability and death assessed after a 3-day incubation period. Increased cell death in an additive manner was observed in all combinations tested, with results for ENTO and LANRA being similar, indicating the utility of both compounds in combinatorial treatment paradigms. These results support the clinical evaluation of LANRA in genetically defined subsets of AML. A phase 1b/2 study of LANRA in combination with the selective FLT3 inhibitor gilteritinib, in patients with relapsed or refractory FLT3 mut AML is planned for the end of this year. Disclosures Day: Cyteir Therapeutics: Current equity holder in publicly-traded company, Ended employment in the past 24 months; Kronos Bio, Inc.: Current Employment, Current equity holder in publicly-traded company. Heckman: Novartis: Research Funding; Orion Pharma: Research Funding; Celgene/BMS: Research Funding; Oncopeptides: Consultancy, Research Funding; Kronos Bio, Inc.: Research Funding. Schinzel: Kronos Bio, Inc.: Current Employment, Current equity holder in publicly-traded company. Obholzer: Kronos Bio, Inc.: Current Employment, Current equity holder in publicly-traded company, Current holder of individual stocks in a privately-held company, Current holder of stock options in a privately-held company. Lin: Kronos Bio, Inc.: Current Employment. Kumar: Kronos Bio, Inc.: Current Employment, Current equity holder in publicly-traded company. DiMartino: Kronos Bio, Inc.: Current Employment, Current equity holder in publicly-traded company. Saffran: Kronos Bio, Inc.: Current Employment, Current equity holder in publicly-traded company.