A Study on the Application and Use of Artificial Intelligence to Support Drug Development.

PURPOSE: The Tufts Center for the Study of Drug Development (CSDD) and the Drug Information Association (DIA) in collaboration with 8 pharmaceutical and biotechnology companies conducted a study examining the adoption and effect of artificial intelligence (AI), such as machine learning, on drug development. The study was conducted to clarify and understand AI adoption across the industry and to gather detailed insights into the spectrum of activities included in the definition of AI. The study investigated and identified analytical platforms and innovations across pharmaceutical and biotechnology companies currently being used or planned for in the future. METHODS: A 2-part method was used that comprised in-depth interviews with AI industry experts and a global survey conducted across pharmaceutical and biotechnology organizations. Eleven in-depth interviews focused on use and implementation of AI across drug development. The survey assessed use of AI and included perceptions about current and future use. The survey also examined technology definitions, assessment of organizational and personal AI expertise, and use of partnerships. A total of 402 responses, including data from 217 unique organizations, were analyzed. FINDINGS: Although 7 in 10 respondents reported using AI in some capacity, a wide range of use was reported by AI type. Patient selection and recruitment for clinical studies was the most commonly reported AI activity, with 34 respondents currently using AI for this activity. In addition, identification of medicinal products data gathering was the top activity being piloted or in the planning stages, reported by 49 respondents. The study also revealed that the most significant challenges to AI implementation included staff skills (55%), data structure (52%), and budgets (49%). Nearly 60% of respondents noted planned increases in staff within 1-2 years to support AI use or implementation. IMPLICATIONS: Despite the challenges to AI implementation, the survey revealed that most organizations use AI in some capacity and that it is important to the success of an organization's workforce. Many organizations reported expectations for increasing staff as implementation of AI expands. Further research should examine the changing development landscape as the role of AI evolves.

2-Aminothiadiazole inhibitors of AKT1 as potential cancer therapeutics.

A series of 2-aminothiadiazole of inhibitors of AKT1 is described. SAR relationships are discussed, along with selectivity for protein kinase A (PKA) and cyclin-dependent kinase 2 (CDK2). Moderate selectivity observed in several compounds for AKT1 versus PKA is rationalized by X-ray crystallographic analysis. Key compounds showed activity in cellular assays measuring phosphorylation of two AKT substrates, PRAS40 and FKHRL1. Compound 30 was advanced to a mouse liver PD assay, where it showed dose-dependent inhibition of AKT activity, as measured by the inhibition of phospho-PRAS40.

Cytochrome P450-mediated epoxidation of 2-aminothiazole-based AKT inhibitors: identification of novel GSH adducts and reduction of metabolic activation through structural changes guided by in silico and in vitro screening.

A 2-aminothiazole derivative 1 was developed as a potential inhibitor of the oncology target AKT, a serine/threonine kinase. When incubated in rat and human liver microsomes in the presence of NADPH, 1 underwent significant metabolic activation on its 2-aminothiazole ring, leading to substantial covalent protein binding. Upon addition of glutathione, covalent binding was reduced significantly, and multiple glutathione adducts were detected. Novel metabolites from the in vitro incubates were characterized by LC-MS and NMR to discern the mechanism of bioactivation. An in silico model was developed based on the proposed mechanism and was employed to predict bioactivation in 23 structural analogues. The predictions were confirmed empirically for the bioactivation liability, in vitro, by LC-MS methods screening for glutathione incorporation. New compounds were identified with a low propensity for bioactivation.

Discovery and optimization of chromenotriazolopyrimidines as potent inhibitors of the mouse double minute 2-tumor protein 53 protein-protein interaction.

Tumor protein 53 (p53) is a critical regulator of cell cycle and apoptosis that is frequently disabled in human tumors. In many tumor types, p53 is deleted or mutated, but in others p53 is inactivated by overexpression or amplification of its negative regulator mouse double minute 2 (MDM2). A high-throughput screening effort identified 6,7-bis(4-bromophenyl)-7,12-dihydro-6H-chromeno[4,3-d][1,2,4]triazolo[1,5-a]pyrimidine as a potent inhibitor of the MDM2-p53 protein-protein interaction. This screening hit was found to be chemically unstable and difficult to handle due to poor DMSO solubility. Co-crystallization with the target protein helped to direct further optimization and provided a tractable lead series of novel MDM2-p53 inhibitors. In cellular assays, these compounds were shown to upregulate p53 protein levels and p53 signaling and to cause p53-dependent inhibition of proliferation and apoptosis.

Theoretical elucidation of kinetic and thermodynamic control of radical addition regioselectivity.

The cyclizations of two structurally similar 2-oxo-5-hexenyl-type radicals have been investigated by ab initio and density functional (UB3LYP/6-31+G**//UHF/6-31G* and UB3LYP/6-31G*//UB3LYP/6-31G*) calculations. The origin of apparently contradictory reports of 6-endo and 5-exo cyclizations is determined. Kinetic control favors 6-endo cyclization, while thermodynamic control gives 5-exo cyclization, and the observation of different products from different research groups arises from the difference in experimental conditions used by the two groups. The outcome of a new cyclization reaction was predicted by using these theoretical techniques. Kinetic control is predicted to yield exclusively the products of 6-endo cyclization, while thermodynamic control would lead to an approximately equal mixture of one 6-endo and one 5-exo cyclized product. Experimental studies revealed that the reaction yields only the products of 6-endo cyclization through kinetic control.