Machine learning (ML), artificial intelligence (AI) and other modern statistical methods are providing new opportunities to operationalize previously untapped and rapidly growing sources of data for patient benefit. Whilst there is a lot of promising research currently being undertaken, the literature as a whole lacks: transparency; clear reporting to facilitate replicability; exploration for potential ethical concerns; and, clear demonstrations of effectiveness. There are many reasons for why these issues exist, but one of the most important that we provide a preliminary solution for here is the current lack of ML/AIspecific best practice guidance. Although there is no consensus on what best practice looks in this field, we believe that interdisciplinary groups pursuing research and impact projects in the ML/AI for health domain would benefit from answering a series of questions based on the important issues that exist when undertaking work of this nature. Here we present 20 questions that span the entire project life cycle, from inception, data analysis, and model evaluation, to implementation, as a means to facilitate project planning and post-hoc (structured) independent evaluation. By beginning to answer these questions in different settings, we can start to understand what constitutes a good answer, and we expect that the resulting discussion will be central to developing an international consensus framework for transparent, replicable, ethical and effective research in artificial intelligence (AI-TREE) for health.
Studies of the human microbiome, and microbial community ecology in general, have blossomed of late and are now a burgeoning source of exciting research findings. Along with the advent of next-generation sequencing platforms, which have dramatically increased the scope of microbiome-related projects, several high-performance sequence analysis pipelines (e.g. QIIME, MOTHUR, VAMPS) are now available to investigators for microbiome analysis. The subject of our manuscript, the graphical user interface-based Explicet software package, fills a previously unmet need for a robust, yet intuitive means of integrating the outputs of the software pipelines with user-specified metadata and then visualizing the combined data.
Background
Chimeric antigen receptor (CAR)-T cells can induce powerful immune responses in patients with hematological malignancies but have had limited success against solid tumors. This is in part due to the immunosuppressive tumor microenvironment (TME) which limits the activity of tumor-infiltrating lymphocytes (TILs) including CAR-T cells. We have developed a next-generation armored CAR (F i-CAR) targeting receptor tyrosine kinase-like orphan receptor 1 (ROR1), which is expressed at high levels in a range of aggressive tumors including poorly prognostic triple-negative breast cancer (TNBC). The F i-CAR-T is designed to release an anti-PD-1 checkpoint inhibitor upon CAR-T cell activation within the TME, facilitating activation of CAR-T cells and TILs while limiting toxicity.
Methods
To bolster potency, we developed a F i-CAR construct capable of IL-2-mediated, NFAT-induced secretion of anti-PD-1 single-chain variable fragments (scFv) within the tumor microenvironment, following ROR1-mediated activation. Cytotoxic responses against TNBC cell lines as well as levels and binding functionality of released payload were analyzed in vitro by ELISA and flow cytometry. In vivo assessment of potency of F i-CAR-T cells was performed in a TNBC NSG mouse model.
Results
F i-CAR-T cells released measurable levels of anti-PD-1 payload with 5 h of binding to ROR1 on tumor and enhanced the cytotoxic effects at challenging 1:10 E:T ratios. Treatment of established PDL1 + TNBC xenograft model with F i-CAR-T cells resulted in significant abrogation in tumor growth and improved survival of mice (71 days), compared to non-armored CAR cells targeting ROR1 (F CAR-T) alone (49 days) or in combination with systemically administered anti-PD-1 antibody (57 days). Crucially, a threefold increase in tumor-infiltrating T cells was observed with F i-CAR-T cells and was associated with increased expression of genes related to cytotoxicity, migration and proliferation.
Conclusions
Our next-generation of ROR1-targeting inducible armored CAR platform enables the release of an immune stimulating payload only in the presence of target tumor cells, enhancing the therapeutic activity of the CAR-T cells. This technology provided a significant survival advantage in TNBC xenograft models. This coupled with its potential safety attributes merits further clinical evaluation of this approach in TNBC patients.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.