Search

Despite the potential of CAR-T therapies for hematological malignancies, their efficacy in patients with relapse and refractory Acute Myeloid Leukemia has been limited. The aim of our study has been to develop and manufacture a CAR-T cell product that addresses some of the current limitations. We initially compared the phenotype of T cells from AML patients and healthy young and elderly controls. This analysis showed that T cells from AML patients displayed a predominantly effector phenotype, with increased expression of activation (CD69 and HLA-DR) and exhaustion markers (PD1 and LAG3), in contrast to the enriched memory phenotype observed in healthy donors. This differentiated and more exhausted phenotype was also observed, and corroborated by transcriptomic analyses, in CAR-T cells from AML patients engineered with an optimized CAR construct targeting CD33, resulting in a decreased in vivo antitumoral efficacy evaluated in xenograft AML models. To overcome some of these limitations we have combined CRISPR-based genome editing technologies with virus-free gene-transfer strategies using Sleeping Beauty transposons, to generate CAR-T cells depleted of HLA-I and TCR complexes (HLA-IKO/TCRKO CAR-T cells) for allogeneic approaches. Our optimized protocol allows one-step generation of edited CAR-T cells that show a similar phenotypic profile to non-edited CAR-T cells, with equivalent in vitro and in vivo antitumoral efficacy. Moreover, genomic analysis of edited CAR-T cells revealed a safe integration profile of the vector, with no preferences for specific genomic regions, with highly specific editing of the HLA-I and TCR, without significant off-target sites. Finally, the production of edited CAR-T cells at a larger scale allowed the generation and selection of enough HLA-IKO/TCRKO CAR-T cells that would be compatible with clinical applications. In summary, our results demonstrate that CAR-T cells from AML patients, although functional, present phenotypic and functional features that could compromise their antitumoral efficacy, compared to CAR-T cells from healthy donors. The combination of CRISPR technologies with transposon-based delivery strategies allows the generation of HLA-IKO/TCRKO CAR-T cells, compatible with allogeneic approaches, that would represent a promising option for AML treatment.

Artificial intelligence (AI) can unveil novel personalized treatments based on drug screening and whole-exome sequencing experiments (WES). However, the concept of “black box” in AI limits the potential of this approach to be translated into the clinical practice. In contrast, explainable AI (XAI) focuses on making AI results understandable to humans. Here, we present a novel XAI method -called multi-dimensional module optimization (MOM)- that associates drug screening with genetic events, while guaranteeing that predictions are interpretable and robust. We applied MOM to an acute myeloid leukemia (AML) cohort of 319 ex-vivo tumor samples with 122 screened drugs and WES. MOM returned a therapeutic strategy based on the FLT3, CBFβ-MYH11, and NRAS status, which predicted AML patient response to Quizartinib, Trametinib, Selumetinib, and Crizotinib. We successfully validated the results in three different large-scale screening experiments. We believe that XAI will help healthcare providers and drug regulators better understand AI medical decisions.