Revealing the Impact of Genomic Alterations on Cancer Cell Signaling with an Interpretable Deep Learning Model.

Cancer is a disease of aberrant cellular signaling resulting from somatic genomic alterations (SGAs). Heterogeneous SGA events in tumors lead to tumor-specific signaling system aberrations. We interpret the cancer signaling system as a causal graphical model, where SGAs affect signaling proteins, propagate their effects through signal transduction, and ultimately change gene expression. To represent such a system, we developed a deep learning model called redundant-input neural network (RINN) with a transparent redundant-input architecture. Our findings demonstrate that by utilizing SGAs as inputs, the RINN can encode their impact on the signaling system and predict gene expression accurately when measured as the area under ROC curves. Moreover, the RINN can discover the shared functional impact (similar embeddings) of SGAs that perturb a common signaling pathway (e.g., PI3K, Nrf2, and TGF). Furthermore, the RINN exhibits the ability to discover known relationships in cellular signaling systems.

An interpretable deep learning framework for genome-informed precision oncology.

Cancers result from aberrations in cellular signaling systems, typically resulting from driver somatic genome alterations (SGAs) in individual tumors. Precision oncology requires understanding the cellular state and selecting medications that induce vulnerability in cancer cells under such conditions. To this end, we developed a computational framework consisting of two components: 1) A representation-learning component, which learns a representation of the cellular signaling systems when perturbed by SGAs, using a biologically-motivated and interpretable deep learning model. 2) A drug-response-prediction component, which predicts the response to drugs by leveraging the information of the cellular state of the cancer cells derived by the first component. Our cell-state-oriented framework significantly enhances the accuracy of genome-informed prediction of drug responses in comparison to models that directly use SGAs as inputs. Importantly, our framework enables the prediction of response to chemotherapy agents based on SGAs, thus expanding genome-informed precision oncology beyond molecularly targeted drugs.

Machine learning based algorithms to impute PaO2 from SpO2 values and development of an online calculator.

We created an online calculator using machine learning (ML) algorithms to impute the partial pressure of oxygen (PaO2)/fraction of delivered oxygen (FiO2) ratio using the non-invasive peripheral saturation of oxygen (SpO2) and compared the accuracy of the ML models we developed to published equations. We generated three ML algorithms (neural network, regression, and kernel-based methods) using seven clinical variable features (N = 9900 ICU events) and subsequently three features (N = 20,198 ICU events) as input into the models. Data from mechanically ventilated ICU patients were obtained from the publicly available Medical Information Mart for Intensive Care (MIMIC III) database and used for analysis. Compared to seven features, three features (SpO2, FiO2 and PEEP) were sufficient to impute PaO2 from the SpO2. Any of the ML models enabled imputation of PaO2 from the SpO2 with lower error and showed greater accuracy in predicting PaO2/FiO2 ≤ 150 compared to the previously published log-linear and non-linear equations. To address potential hidden hypoxemia that occurs more frequently in Black patients, we conducted sensitivity analysis and show ML models outperformed published equations in both Black and White patients. Imputation using data from an independent validation cohort of ICU patients (N = 133) showed greater accuracy with ML models.

De novo Prediction of Cell-Drug Sensitivities Using Deep Learning-based Graph Regularized Matrix Factorization.

Application of artificial intelligence (AI) in precision oncology typically involves predicting whether the cancer cells of a patient (previously unseen by AI models) will respond to any of a set of existing anticancer drugs, based on responses of previous training cell samples to those drugs. To expand the repertoire of anticancer drugs, AI has also been used to repurpose drugs that have not been tested in an anticancer setting, i.e., predicting the anticancer effects of a new drug on previously unseen cancer cells de novo. Here, we report a computational model that addresses both of the above tasks in a unified AI framework. Our model, referred to as deep learning-based graph regularized matrix factorization (DeepGRMF), integrates neural networks, graph models, and matrix-factorization techniques to utilize diverse information from drug chemical structures, their impact on cellular signaling systems, and cancer cell cellular states to predict cell response to drugs. DeepGRMF learns embeddings of drugs so that drugs sharing similar structures and mechanisms of action (MOAs) are closely related in the embedding space. Similarly, DeepGRMF also learns representation embeddings of cells such that cells sharing similar cellular states and drug responses are closely related. Evaluation of DeepGRMF and competing models on Genomics of Drug Sensitivity in Cancer (GDSC) and Cancer Cell Line Encyclopedia (CCLE) datasets show its superiority in prediction performance. Finally, we show that the model is capable of predicting effectiveness of a chemotherapy regimen on patient outcomes for the lung cancer patients in The Cancer Genome Atlas (TCGA) dataset*.

Pharmacokinetics and pharmacodynamics evaluation of a thermosensitive chitosan based hydrogel containing liposomal doxorubicin.

In situ gelling thermosensitive hydrogel formulation has been reported to effectively sustain the release of macromolecules for a long time. However, the low-molecular-weight hydrophilic drugs, such as doxorubicin (DOX), are not suitable for intratumoral injection because the release will complete within one day. In this study, liposomal doxorubicin (LipDOX) was added into the hydrogel to form a novel thermosensitive formulation which prolonged the sustained release of DOX. DOX+C/GP (doxorubicin in chitosan/ß-glycerophosphate) was prepared to compare with LipDOX+C/GP (liposomal doxorubicin in chitosan/ß-glycerophosphate hydrogel). The particle size of DOX-loaded liposome was 94.2nm and the encapsulation efficiency of DOX was near 98%. In vitro release experiments, the release of DOX in both DOX+C/GP group and LipDOX+C/GP group increased along with the increasing pH of buffers. However, the LipDOX+C/GP group with lower initial burst release had a much longer releasing duration than DOX+C/GP group (21days vs. 24h). In vitro and in vivo antitumor experiments demonstrated that LipDOX+C/GP group had better antineoplastic effect and less toxicity than DOX+C/GP group. Pharmacokinetics study showed LipDOX+C/GP exhibited a higher AUC0-t and longer MRT than DOX+C/GP in blood and tumor, which indicated that LipDOX+C/GP obtained an enhanced antitumor activity compared with DOX+C/GP. In addition, the lower distribution index (the ratio of AUC of normal tissue/AUC of tumor tissue) of the LipDOX+C/GP implied it had lower toxicity to normal tissues than DOX+C/GP. Therefore, the novel thermosensitive hydrogel formulation was potential for clinical application in cancer treatment.

Biotin-Conjugated Multilayer Poly [D,L-lactide-co-glycolide]-Lecithin-Polyethylene Glycol Nanoparticles for Targeted Delivery of Doxorubicin.

Multilayer nanoparticle combining the merits of liposome and polymer nanoparticle has been designed for the targeted delivery of doxorubicin (DOX) in cancer treatment. In this study, DOX-PLGA-lecithin-PEG-biotin nanoparticles (DOX-PLPB-NPs) were fabricated and functionalized with biotin for specific tumor targeting. Under the transmission electron microscopy observation, the lipid layer was found to be coated on the polymer core. The physical characteristics of PLPB-NPs were also evaluated. The confocal laser scanning microscopy confirmed the cellular uptake of nanoparticles and targeted delivery PLPB-NPs. The in vitro release experiment demonstrated a pH-depending release of DOX from drug-loaded PLPB-NPs. Cytotoxicity studies in HepG2 cells and in vivo antitumor experiment in tumor-bearing mice both proved DOX-PLPB-NPs showed the best inhibition effect of tumor proliferation. In biodistribution studies, DOX-PLPB-NPs showed a higher DOX concentration than free DOX and DOX-PLGA-lecithin-PEG nanoparticles (DOX-PLP-NPs) in tumor site, especially in 24 h, and the lowest DOX level in normal organs. The results were coincident with the strongest antitumor ability showed among in vivo antitumor experiment. Histopathology analysis demonstrated that DOX-PLPB-NPs exhibited the strongest antitumor ability and lowest cardiotoxicity. In brief, the PLPB-NPs were proved to be an efficient delivery system for tumor-targeting treatment.

Comparative study on pharmacokinetics of a series of anticholinergics, atropine, anisodamine, anisodine, scopolamine and tiotropium in rats.

The compound series of traditional anticholinergics [atropine (Atr), anisodamine (Ani), anisodine (AT3), and scopolamine (Sco)], naturally occurring belladonna alkaloid, have been approved for numerous therapeutic uses since 1970s. Tiotropium, a novel M receptor antagonist for the treatment of chronic obstructive pulmonary disease, was structurally modified based on atropine-like drugs. Clinical phenomena suggested that the changes of substituent group were related to the pharmacokinetic and pharmacodynamic characteristics of the agents. In an attempt to compare the pharmacokinetics of the series of anticholinergics and investigate the subsets motivating selective anticholinergic potencies, a sensitive LC-MS/MS method was established to analyze the differences of pharmacokinetic parameters. In this paper, we determined the pharmacokinetics of atropine, anisodamine, anisodine, scopolamine, and tiotropium after i.v. and i.g. single dose administration. After i.v. administration, the maximum drug plasma concentrations (C max) of Atr, Ani, AT3, and Sco were 274.25 ± 53.66, 267.50 ± 33.16, 340.50 ± 44.52, and 483.75 ± 78.13 ng/mL. Tiotropium had a slightly higher area under the curve with a significant increase of C max value. Because of their partial solubility, Atr, Ani, AT3, and Sco had different bioavailability in rats of 21.62, 10.78, 80.45 and 2.52 %, respectively. Following i.g. administration of tiotropium, the C max value below 20 ng/mL revealed the very low oral absorption. The urinary excretion rates of Atr, Ani, AT3, Sco and tiotropium were 11.33, 54.86, 32.67, 8.69 and 73.91 %. This work provided relatively comprehensive preclinical data on the series of anticholinergics, which may be used to explain the clinical adverse effects and applications.

Comparison of pharmacokinetics, tissue distribution and pharmacodynamics of liposomal and free doxorubicin in tumour-bearing mice following intratumoral injection.

OBJECTIVES: The clinical application of doxorubicin (DOX) is limited by severe systemic side effects. The aim of this study was to develop a strategy that combined the liposomal DOX (LipDOX) and intratumoral injection to reduce the toxicity and enhance the antitumor efficiency. METHODS: The pharmacokinetics, tissue distribution and pharmacodynamics of LipDOX compared with free DOX were investigated by intratumoral injection in murine H22 hepatoma-bearing mice at a dose of 20 mg/kg body weight. A sensitive HPLC-tandem mass spectrometry method was used to determine the DOX levels in plasma and tissues. The tumour volume and body weight of mice were measured every 3 days. KEY FINDINGS: LipDOX administration resulted in 1.3-fold longer mean residence time (MRT) and 2.4-fold higher area under concentration (AUC)-time curve compared with free DOX administration in tumour. Free DOX caused higher peak plasma concentration (Cmax ) than LipDOX in plasma and major organs, which may result in significant mortality for acute cardiac toxicity. After successive 21 days treatment, the final volume of tumour treated by normal saline, free DOX and LipDOX was 5.0-, 1.3-fold higher and 1.6-fold lower than the initial tumour volume, respectively. CONCLUSIONS: Our results indicated that the intratumoral injection of LipDOX is a promising approach with higher therapeutic efficacy and lower systemic toxicity than free DOX.

Efficacy, pharmacokinetics, and biodistribution of thermosensitive chitosan/ß-glycerophosphate hydrogel loaded with docetaxel.

Docetaxel (DTX) is a widely used anticancer drug for various solid tumors. However, its poor solubility in water and lack of specification are two limitations for clinical use. The aim of the study was to develop a thermosensitive chitosan/ß-glycerophosphate (C/GP) hydrogel loaded with DTX for intratumoral delivery. The in vitro release profiles, in vivo antitumor efficacy, pharmacokinetics, and biodistribution of DTX-loaded C/GP hydrogel (DTX-C/GP) were evaluated. The results of in vitro release study demonstrated that DTX-C/GP had the property of controlled delivery for a reasonable time span of 3 weeks and the release period was substantially affected by initial DTX strength. The antitumor efficacy of DTX-C/GP was observed at 20 mg/kg in H22 tumor-bearing mice. It was found that the tumor volume was definitely minimized by intratumoral injection of DTX-C/GP. Compared with saline group, the tumor inhibition rate of blank gel, intravenous DTX solution, intratumoral DTX solution, and DTX-C/GP was 2.3%, 29.8%, 41.9%, and 58.1%, respectively. Further, the in vivo pharmacokinetic characteristics of DTX-C/GP correlated well with the in vitro release. DTX-C/GP significantly prolonged the DTX retention and maintained a high DTX concentration in tumor. The amount of DTX distributed to the normal tissues was minimized so that the toxicity was effectively reduced. In conclusion, DTX-C/GP demonstrated controlled release and significant efficacy and exhibited potential for further clinical development.

Contributions of intestine and plasma to the presystemic bioconversion of vicagrel, an acetate of clopidogrel.

PURPOSE: To investigate the contributions of intestine and plasma to the presystemic bioconversion of vicagrel, and track its subsequent bioconversion to 2-oxo-clopidogrel in vivo and in vitro to rationalize the design of vicagrel, an acetate analogue of clopidogrel. METHODS: The concentration-time profiles of 2-oxo-clopidogrel and active metabolite (AM) in presystem and circulation system was determined in the cannulated rats. Also, the rat intestinal S9 and human intestinal microsomes were conducted to examine the formation of 2-oxo-clopidogrel and AM. Meanwhile, the esterases in plasma and intestinal fractions responsible for the bioconversion of vicagrel to 2-oxo-clopidogrel were screened by the esterase inhibition and recombinant esterases. RESULTS: The intestine was responsible for the formation of 2-oxo-clopidogrel and AM in vivo and in vitro, where carboxylesterases 2 (CE2) contributed greatly to the vicagrel cleavage during absorption. Other related esterases in plasma were paraoxonases (PON), carboxylesterases 1 (CE1) and butyrylcholine esterases (BChE). CONCLUSION: The findings rationalized the prodrug design hypothesis that vicagrel could overcome the extensive invalid hydrolysis of clopidogrel by the hepatic CE1 but experience the extensive hydrolysis to 2-oxo-clopidogrel and subsequent oxidation to AM in the intestine. This also supported the theory of improved pharmacological activity through facilitated formation of 2-oxo-clopidogrel, thus warranting much needed future clinical pharmacokinetic studies of vicagrel.

In vitro cytotoxicity, pharmacokinetics and tissue distribution in rats of MXN-004, a novel conjugate of polyethylene glycol and SN38.

1. MXN-004 is a water-soluble PEGylated 7-ethyl-10-hydroxy-camptothecin (SN38). The aim of this study was to evaluate the in vitro cytotoxicity of MXN-004 and investigate pharmacokinetics and tissue distribution of MXN-004 and its active metabolite SN38 in rats. 2. In vitro cytotoxicity of MXN-004 was tested in A549, HepG2 and Caco-2 cancer cell lines by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay and compared with irinotecan. The pharmacokinetics and tissue distribution of MXN-004, irinotecan and their identical active metabolite SN38 were investigated after intravenous administration of MXN-004 and irinotecan at a same dose level of 16 µmol/kg in rats. 3. In vitro cytotoxicity study showed that MXN-004 was more potent in comparison with irinotecan. In rats, MXN-004 exhibited a longer half-life (sixfold) and much greater Vss as compared with irinotecan. The AUC0-∞, T1/2 and Cmax of SN38 after intravenous administration of MXN-004 were higher than those of irinotecan (3.5-, 1.92- and 10.6-fold, respectively). In addition, the concentrations of SN38 released from MXN-004 were significantly higher in all tissues than those from irinotecan, especially in the lung. 4. These results suggested that MXN-004 might be a more potential water-soluble antitumor agent with prolonged half-life of SN38 compared to irinotecan.

Pharmacokinetics study of hemin in rats by applying (58)Fe-extrinsically labeling techniques in combination with ICP-MS method.

Iron is a challenging element due to its high background in various matrixes including blood, tissues even in the air and it is urgent to develop a method for the accurate determination of iron in bio-samples. After optimization of mass spectrometric conditions using collision cell technology and compensating for interference using a mathematical correction equation, an inductively coupled plasma mass spectrometry (ICP-MS) method for the quantitative determination of (58)Fe originating from hemin extrinsically labeled avoiding endogenous interference was developed. After a single step of dilution, analysis of each sample was completed within 1.5min. The assay was linear in the concentration range from 0.005 to 1.0µg/ml. The precisions and accuracies determined within three consecutive days were in acceptable limits and there was no significant matrix effect. The optimized method was successfully applied to a pharmacokinetic study of (58)Fe originating from hemin extrinsically labeled and iron absorption measured in rats was 1.07%. Those indicated that extrinsically label techniques in combination with ICP-MS will become a new tool for the analysis of iron preparations and other endogenous substances.

Pharmacokinetics, tissue distribution, and plasma protein binding study of platinum originating from dicycloplatin, a novel antitumor supramolecule, in rats and dogs by ICP-MS.

Dicycloplatin, as a new antitumor supramolecule, was considered to have higher solubility and higher stability compared with carboplatin. The aim of the present study was to evaluate the pharmacokinetic characteristics of platinum originating from dicycloplatin. A rapid, sensitive, and specific method with inductively coupled plasma mass spectrometry (ICP-MS) has been developed for the determination of platinum in bio-samples. The study was performed in male rats and dogs at a single dose of 10 and 5 mg kg(-1) separately by intravenous injection. Pharmacokinetic parameters were calculated by non-compartmental method, and the dose of platinum was used in the calculation of these parameters. Results showed that plasma concentrations of platinum began to decrease rapidly initially but decline slowly with a long terminal phase. The mean half-life was 27.39 and 100.98 and clearance was 0.77 and 0.08 L/h/kg for rats and dogs separately. Tissue distribution showed that platinum originating from dicycloplatin had a certain distribution in testis and prostate. Plasma protein binding proportion of platinum was increased with time. In conclusion, this research investigated the pharmacokinetic characteristics including plasma kinetics, tissue distribution, and plasma protein binding of platinum originating from dicycloplatin in rats and dogs in detail for the first time by ICP-MS.