Machine learning model to predict oncologic outcomes for drugs in randomized clinical trials.

Predicting oncologic outcome is challenging due to the diversity of cancer histologies and the complex network of underlying biological factors. In this study, we determine whether machine learning (ML) can extract meaningful associations between oncologic outcome and clinical trial, drug-related biomarker and molecular profile information. We analyzed therapeutic clinical trials corresponding to 1102 oncologic outcomes from 104 758 cancer patients with advanced colorectal adenocarcinoma, pancreatic adenocarcinoma, melanoma and nonsmall-cell lung cancer. For each intervention arm, a dataset with the following attributes was curated: line of treatment, the number of cytotoxic chemotherapies, small-molecule inhibitors, or monoclonal antibody agents, drug class, molecular alteration status of the clinical arm's population, cancer type, probability of drug sensitivity (PDS) (integrating the status of genomic, transcriptomic and proteomic biomarkers in the population of interest) and outcome. A total of 467 progression-free survival (PFS) and 369 overall survival (OS) data points were used as training sets to build our ML (random forest) model. Cross-validation sets were used for PFS and OS, obtaining correlation coefficients (r) of 0.82 and 0.70, respectively (outcome vs model's parameters). A total of 156 PFS and 110 OS data points were used as test sets. The Spearman correlation (rs ) between predicted and actual outcomes was statistically significant (PFS: rs = 0.879, OS: rs = 0.878, P < .0001). The better outcome arm was predicted in 81% (PFS: N = 59/73, z = 5.24, P < .0001) and 71% (OS: N = 37/52, z = 2.91, P = .004) of randomized trials. The success of our algorithm to predict clinical outcome may be exploitable as a model to optimize clinical trial design with pharmaceutical agents.

The Crossroads of Precision Medicine and Therapeutic Decision-Making: Use of an Analytical Computational Platform to Predict Response to Cancer Treatments.

Metastatic cancer is a medical challenge that has been historically resistant to treatments. One area of leverage in cancer care is the development of molecularly-driven combination therapies, offering the possibility to overcome resistance. The selection of optimized treatments based on the complex molecular features of a patient's tumor may be rendered easier by using a computer-assisted program. We used the PreciGENE® platform that uses multi-pathway molecular analysis to identify personalized therapeutic options. These options are ranked using a predictive score reflecting the degree to which a therapy or combination of therapies matches the patient's biomarker profile. We searched PubMed from February 2010 to June 2017 for all patients described as exceptional responders who also had molecular data available. Altogether, 70 patients with cancer who had received 202 different treatment lines and who had responded (stable disease ≥12 months/partial or complete remission) to ≥1 regimen were curated. We demonstrate that an algorithm reflecting the degree to which patients were matched to the drugs administered correctly ranked the response to the regimens with a sensitivity of 84% and a specificity of 77%. The difference in matching score between successful and unsuccessful treatment lines was significant (median, 65% versus 0%, p-value <0.0001).

Structural basis for dual functionality of isoflavonoid O-methyltransferases in the evolution of plant defense responses.

In leguminous plants such as pea (Pisum sativum), alfalfa (Medicago sativa), barrel medic (Medicago truncatula), and chickpea (Cicer arietinum), 4'-O-methylation of isoflavonoid natural products occurs early in the biosynthesis of defense chemicals known as phytoalexins. However, among these four species, only pea catalyzes 3-O-methylation that converts the pterocarpanoid isoflavonoid 6a-hydroxymaackiain to pisatin. In pea, pisatin is important for chemical resistance to the pathogenic fungus Nectria hematococca. While barrel medic does not biosynthesize 6a-hydroxymaackiain, when cell suspension cultures are fed 6a-hydroxymaackiain, they accumulate pisatin. In vitro, hydroxyisoflavanone 4'-O-methyltransferase (HI4'OMT) from barrel medic exhibits nearly identical steady state kinetic parameters for the 4'-O-methylation of the isoflavonoid intermediate 2,7,4'-trihydroxyisoflavanone and for the 3-O-methylation of the 6a-hydroxymaackiain isoflavonoid-derived pterocarpanoid intermediate found in pea. Protein x-ray crystal structures of HI4'OMT substrate complexes revealed identically bound conformations for the 2S,3R-stereoisomer of 2,7,4'-trihydroxyisoflavanone and the 6aR,11aR-stereoisomer of 6a-hydroxymaackiain. These results suggest how similar conformations intrinsic to seemingly distinct chemical substrates allowed leguminous plants to use homologous enzymes for two different biosynthetic reactions. The three-dimensional similarity of natural small molecules represents one explanation for how plants may rapidly recruit enzymes for new biosynthetic reactions in response to changing physiological and ecological pressures.

Structural basis for the promiscuous biosynthetic prenylation of aromatic natural products.

The anti-oxidant naphterpin is a natural product containing a polyketide-based aromatic core with an attached 10-carbon geranyl group derived from isoprenoid (terpene) metabolism. Hybrid natural products such as naphterpin that contain 5-carbon (dimethylallyl), 10-carbon (geranyl) or 15-carbon (farnesyl) isoprenoid chains possess biological activities distinct from their non-prenylated aromatic precursors. These hybrid natural products represent new anti-microbial, anti-oxidant, anti-inflammatory, anti-viral and anti-cancer compounds. A small number of aromatic prenyltransferases (PTases) responsible for prenyl group attachment have only recently been isolated and characterized. Here we report the gene identification, biochemical characterization and high-resolution X-ray crystal structures of an architecturally novel aromatic PTase, Orf2 from Streptomyces sp. strain CL190, with substrates and substrate analogues bound. In vivo, Orf2 attaches a geranyl group to a 1,3,6,8-tetrahydroxynaphthalene-derived polyketide during naphterpin biosynthesis. In vitro, Orf2 catalyses carbon-carbon-based and carbon-oxygen-based prenylation of a diverse collection of hydroxyl-containing aromatic acceptors of synthetic, microbial and plant origin. These crystal structures, coupled with in vitro assays, provide a basis for understanding and potentially manipulating the regio-specific prenylation of aromatic small molecules using this structurally unique family of aromatic PTases.