B cell–depleting therapies impair antibody responses in 2-dose series of mRNA vaccination against SARS-CoV-2, urging research on additional protection against COVID-19 in patients with B-cell malignancies. Greenberger et al. characterize T-cell responses after 2 doses of mRNA vaccines, and antibody responses to a 3rd vaccine dose, in a real-world cohort of over 1,000 patients with hematologic cancers. The third dose boosted antibody titers of patients seropositive after 2 doses, with 97% reaching physiologically meaningful titers. Among the 40% of patients who were seronegative after two doses, the third dose elicited antibodies in quantities sufficient for virus neutralization in approximately one in five cases. T-cell responses to the 2nd vaccine dose in 514 patients correlated with humoral immunity but were less impacted by anticancer treatments.See article, p. 481.Although BCMA-directed CAR T cell therapy (BCMA-CART) leads to high rates of response in myeloma, many of these patients experience disease relapse. There is an urgent unmet need to identify determinants of durable responses. Dhodapkar, Cohen et al. combine several high-dimensional tools to analyze tumor and immune cells in the bone marrow of myeloma patients receiving BCMA-CART. They identify distinct features of tumor and immune cells both at baseline and following therapy that correlate with durability of clinical responses. These data illustrate dynamic cross-talk between CAR T cells, endogenous immunity, and residual tumors, and suggest novel strategies to improve durability of CAR T-cell therapy in myeloma.See article, p. 490.The functional precision medicine approach screens patient cell sensitivity to hundreds of drugs in a dish in order to identify a drug that kills cancer cells but not normal cells. The results are scored based on cancer markers by clinicians, limiting scalability. In a computational screen, Heinemann, Kornauth et al. identified a neural network distinguishing cancer cells from stroma by morphology features. When compared to marker-based scoring in a clinical trial, concordant recommendations by both approaches had best outcomes. In cases with discordant recommendations, progression-free survival was 144 days in deep morphology learning–concordant cases as compared to 72 days in marker-concordant therapy cases.See article, p. 502.In the past decade, numerous revolutionary therapeutic options were made available for adults with acute myeloid leukemia (AML) but direct extrapolation to children is particularly challenging due to the marked differences in leukemia genetics between young and old. Wang, Chan, Cheng, Ng et al. performed integrated drug and genomic profiling of patient biopsies and linked to clinical parameters to build the functional genomic landscape of childhood AML, representing a high-dimensional data set to characterize the biological uniqueness of the disease. Age-specific differences in drug response, new gene-drug interactions, and predictive biomarkers were identified. Feasibility of functional precision medicine-guided treatment of children with high-risk AML was successfully demonstrated in the clinic, marking a new approach to manage difficult cases.See article, p. 516.Myeloblast expansion is a clinical hallmark of myeloid neoplasms. Ferrall-Fairbanks, Dhawan et al. resolve myeloblasts and other hematopoietic stem and progenitor cell (HSPC) populations in chronic myelomonocytic leukemia (CMML) by high-dimensional single-cell characterization of sorted CD34+ HSPCs from 39 patients at diagnosis. They identify differentiation trajectories with three biases: normal, megakaryocyte-erythroid, and monocytic. The latter associates with adverse outcomes and is phenotypically characterized by stem cell marker depletion, CD120b+, and increased diversity and activation of cytokine receptor signaling. The monocytic bias is diminished in CMML patients following hypomethylating therapy, and is instigated by inflammation in preclinical mouse models.See article, p. 536.Mutations in the splicing factor SF3B1 are commonly found in patients with myelodysplastic syndrome with ringed sideroblasts (MDS-RS), but it is unknown how the dysregulated downstream iron metabolism pathways disrupt erythroid differentiation. Adema, Ma et al. leverage single-cell transcriptomics in SF3B1-mutant MDS-RS patient samples to identify buildup of late-stage erythroid progenitor cell types, which upregulate EIF2AK1-associated heme deficiency stress response and downstream autophagy genes. Knockout of EIK2AK1 in SF3B1-mutant erythroblasts and MDS-RS cells, as well as response to hypomethylating therapy in patients, was associated with rescue of erythroid progenitor differentiation, reduced autophagy, and increased heme metabolism gene expression, suggesting targeting of the EIF2AK1 pathway as a therapeutic strategy for MDS-RS.See article, p. 554.