Development of a Novel In Silico Classification Model to Assess Reactive Metabolite Formation in the Cysteine Trapping Assay and Investigation of Important Substructures.

Predicting whether a compound can cause drug-induced liver injury (DILI) is difficult due to the complexity of drug mechanism. The cysteine trapping assay is a method for detecting reactive metabolites that bind to microsomes covalently. However, it is cumbersome to use 35S isotope-labeled cysteine for this assay. Therefore, we constructed an in silico classification model for predicting a positive/negative outcome in the cysteine trapping assay. We collected 475 compounds (436 in-house compounds and 39 publicly available drugs) based on experimental data performed in this study, and the composition of the results showed 248 positives and 227 negatives. Using a Message Passing Neural Network (MPNN) and Random Forest (RF) with extended connectivity fingerprint (ECFP) 4, we built machine learning models to predict the covalent binding risk of compounds. In the time-split dataset, AUC-ROC of MPNN and RF were 0.625 and 0.559 in the hold-out test, restrictively. This result suggests that the MPNN model has a higher predictivity than RF in the time-split dataset. Hence, we conclude that the in silico MPNN classification model for the cysteine trapping assay has a better predictive power. Furthermore, most of the substructures that contributed positively to the cysteine trapping assay were consistent with previous results.

Development of Novel Methods for QSAR Modeling by Machine Learning Repeatedly: A Case Study on Drug Distribution to Each Tissue.

Artificial intelligence is expected to help identify excellent candidates in drug discovery. However, we face a lack of data, as it is time-consuming and expensive to acquire raw data perfectly for many compounds. Hence, we tried to develop a novel quantitative structure-activity relationship (QSAR) method to predict a parameter more precisely from an incomplete data set via optimizing data handling by making use of predicted explanatory variables. As a case study we focused on the tissue-to-plasma partition coefficient (Kp), which is an important parameter for understanding drug distribution in tissues and building the physiologically based pharmacokinetic model and is a representative of small and sparse data sets. In this study, we predicted the Kp values of 119 compounds in nine tissues (adipose, brain, gut, heart, kidney, liver, lung, muscle, and skin), although some of these were not available. To fill the missing values in Kp for each tissue, first we predicted those Kp values by the nonmissing data set using a random forest (RF) model with in vitro parameters (log P, fu, Drug Class, and fi) like a classical prediction by a QSAR model. Next, to predict the tissue-specific Kp values in a test data set, we constructed a second RF model with not only in vitro parameters but also the Kp values of other tissues (i.e., other than target tissues) predicted by the first RF model as explanatory variables. Furthermore, we tested all possible combinations of explanatory variables and selected the model with the highest predictability from the test data set as the final model. The evaluation of Kp prediction accuracy based on the root-mean-square error and R2 value revealed that the proposed models outperformed other machine learning methods such as the conventional RF and message-passing neural networks. Significant improvements were observed in the Kp values of adipose tissue, brain, kidney, liver, and skin. These improvements indicated that the Kp information on other tissues can be used to predict the same for a specific tissue. Additionally, we found a novel relationship between each tissue by evaluating all combinations of explanatory variables. In conclusion, we developed a novel RF model to predict Kp values. We hope that this method will be applied to various problems in the field of experimental biology which often contains missing values in the near future.

CDK4/6 inhibitor-induced bone marrow micronuclei might be caused by cell cycle arrest during erythropoiesis.

BACKGROUND: A micronucleus test is generally used to evaluate the genotoxic potential of chemicals. Exaggerated erythropoiesis, as occurs following bleeding, may induce an unexpected increase in micronucleus frequency. This false positive result would be typical in a genotoxicity study due to the enhanced progression of the cell cycle that restores decreased blood cells. The cyclin-dependent kinase (CDK) family is known to play an essential role in preventing genomic instability. Conversely, a selective CDK4/6 inhibitor PD0332991, clinically named Palbociclib, is reported to have genotoxic potential, shown by positive results in both in vitro and in vivo micronucleus studies. To clarify the mechanism by which cell cycle arrest induced by a CDK4/6 inhibitor increases micronucleus frequency, we investigated the positive results of the bone marrow micronucleus test conducted with PD0332991. RESULTS: Rats treated with PD0332991 exhibited increased micronucleus frequency in an in vivo bone marrow micronucleus test whereas it was not increased by treatment in human lymphoblastoid TK6 cells. In addition, all other genotoxicity tests including the Ames test and the comet assay showed negative results with PD0332991. Interestingly, PD0332991 treatment led to an increase in erythrocyte size in rats and affected the size distribution of erythrocytes, including the micronucleus. The mean corpuscular volume of reticulocytes (MCVr) in the PD0332991 treatment group was significantly increased compared to that of the vehicle control (83.8 fL in the PD0332991, and 71.6 fL in the vehicle control.). Further, the average micronucleated erythrocytes (MNE) size of the PD0332991 group and vehicle control was 8.2 and 7.3 µm, respectively. In the histogram, the vehicle control showed a monomodal distribution with a peak near 7.3 µm. In contrast, the PD0332991 group showed a bimodal distribution with peaks around 7.5 and 8.5 µm. Micronucleated erythrocytes in the PD0332991 group were significantly larger than those in the vehicle control. These results suggest that the increase in micronucleus frequency induced by the CDK4/6 inhibitor is not due to genotoxicity, but is attributable to disturbance of the cell cycle, differentiation, and enucleation of erythroblasts. CONCLUSIONS: It was suggested that the positive outcome of the in vivo bone marrow micronucleus test resulting from treatment with PD0332991 could not be attributed to its genotoxicity. Further studies to clarify the mechanism of action can contribute to the development of drug candidate compounds lacking intrinsic genotoxic effects.

On the difficulty of validating molecular generative models realistically: a case study on public and proprietary data.

While a multitude of deep generative models have recently emerged there exists no best practice for their practically relevant validation. On the one hand, novel de novo-generated molecules cannot be refuted by retrospective validation (so that this type of validation is biased); but on the other hand prospective validation is expensive and then often biased by the human selection process. In this case study, we frame retrospective validation as the ability to mimic human drug design, by answering the following question: Can a generative model trained on early-stage project compounds generate middle/late-stage compounds de novo? To this end, we used experimental data that contains the elapsed time of a synthetic expansion following hit identification from five public (where the time series was pre-processed to better reflect realistic synthetic expansions) and six in-house project datasets, and used REINVENT as a widely adopted RNN-based generative model. After splitting the dataset and training REINVENT on early-stage compounds, we found that rediscovery of middle/late-stage compounds was much higher in public projects (at 1.60%, 0.64%, and 0.21% of the top 100, 500, and 5000 scored generated compounds) than in in-house projects (where the values were 0.00%, 0.03%, and 0.04%, respectively). Similarly, average single nearest neighbour similarity between early- and middle/late-stage compounds in public projects was higher between active compounds than inactive compounds; however, for in-house projects the converse was true, which makes rediscovery (if so desired) more difficult. We hence show that the generative model recovers very few middle/late-stage compounds from real-world drug discovery projects, highlighting the fundamental difference between purely algorithmic design and drug discovery as a real-world process. Evaluating de novo compound design approaches appears, based on the current study, difficult or even impossible to do retrospectively.Scientific Contribution This contribution hence illustrates aspects of evaluating the performance of generative models in a real-world setting which have not been extensively described previously and which hopefully contribute to their further future development.

Development of a 2D-QSAR Model for Tissue-to-Plasma Partition Coefficient Value with High Accuracy Using Machine Learning Method, Minimum Required Experimental Values, and Physicochemical Descriptors.

BACKGROUND: The demand for physiologically based pharmacokinetic (PBPK) model is increasing currently. New drug application (NDA) of many compounds is submitted with PBPK models for efficient drug development. Tissue-to-plasma partition coefficient (Kp) is a key parameter for the PBPK model to describe differential equations. However, it is difficult to obtain the Kp value experimentally because the measurement of drug concentration in the tissue is much harder than that in plasma. OBJECTIVE: Instead of experiments, many researchers have sought in silico methods. Today, most of the models for Kp prediction are using in vitro and in vivo parameters as explanatory variables. We thought of physicochemical descriptors that could improve the predictability. Therefore, we aimed to develop the two-dimensional quantitative structure-activity relationship (2D-QSAR) model for Kp using physicochemical descriptors instead of in vivo experimental data as explanatory variables. METHODS: We compared our model with the conventional models using 20-fold cross-validation according to the published method (Yun et al. J Pharmacokinet Pharmacodyn 41:1-14, 2014). We used random forest algorithm, which is known to be one of the best predictors for the 2D-QSAR model. Finally, we combined minimum in vitro experimental values and physiochemical descriptors. Thus, the prediction method for Kp value using a few in vitro parameters and physicochemical descriptors was developed; this is a multimodal model. RESULTS: Its accuracy was found to be superior to that of the conventional models. Results of this research suggest that multimodality is useful for the 2D-QSAR model [RMSE and % of two-fold error: 0.66 and 42.2% (Berezohkovsky), 0.52 and 52.2% (Rodgers), 0.65 and 34.6% (Schmitt), 0.44 and 61.1% (published model), 0.41 and 62.1% (traditional model), 0.39 and 64.5% (multimodal model)]. CONCLUSION: We could develop a 2D-QSAR model for Kp value with the highest accuracy using a few in vitro experimental data and physicochemical descriptors.

Prediction of Compound Plasma Concentration-Time Profiles in Mice Using Random Forest.

Pharmacokinetic (PK) parameters such as clearance (CL) and volume of distribution (Vd) have been the subject of previous in silico predictive models. However, having information of the concentration over time profile explicitly can provide additional value like time above MIC or AUC, etc., to understand both the efficacy and safety-related aspects of a compound. In this work, we developed machine learning models for plasma concentration-time profiles after both i.v. and p.o. dosing for a series of 17 in-house projects. For explanatory variables, MACCS Keys chemical descriptors as well as in silico and experimental in vitro PK parameters were used. The predictive accuracy of random forest (RF), message passing neural network, 2-compartment models using estimated CL and Vdss, and an average model (as a control experiment) was investigated using 5-fold cross-validation (5-fold CV) and leave-one-project-out validation (LOPO-V). The predictive accuracy of RF in 5-fold CV for i.v. and p.o. plasma concentration-time profiles was the best among the models studied, with an RMSE for i.v. dosing at 0.08, 1, and 8 h of 0.245, 0.474, and 0.462, respectively, and an RMSE for p.o. dosing at 0.25, 1, and 8 h of 0.500, 0.612, and 0.509, respectively. Furthermore, by investigating the importance of the in vitro PK parameters using the Gini index, we observed that the general prior knowledge in ADME research was reflected well in the respective feature importance of in vitro parameters such as predicted human Vd (hVd) for the initial distribution, mouse intrinsic CL and unbound fraction of mouse plasma for the elimination process, and Caco2 permeability for the absorption process. Also, this model is the first model that can predict twin peaks in the concentration-time profile much better than a baseline compartment model. Because of its combination of sufficient accuracy and speed of prediction, we found the model to be fit-for-purpose for practical lead optimization.

Significance of Basal Membrane Permeability of Epithelial Cells in Predicting Intestinal Drug Absorption.

Drug absorption from the gastrointestinal tract is often restricted by efflux transport by P-glycoprotein (P-gp) and metabolism by CYP3A4. Both localize in the epithelial cells, and thus, their activities are directly affected by the intracellular drug concentration, which should be regulated by the ratio of permeability between apical (A) and basal (B) membranes. In this study, using Caco-2 cells with forced expression of CYP3A4, we assessed the transcellular permeation of A-to-B and B-to-A directions and the efflux from the preloaded cells to both sides of 12 representative P-gp or CYP3A4 substrate drugs and obtained the parameters for permeabilities, transport, metabolism, and unbound fraction in the enterocytes (fent) using simultaneous and dynamic model analysis. The membrane permeability ratios for B to A (RBA) and fent varied by 8.8-fold and by more than 3000-fold, respectively, among the drugs. The RBA values for digoxin, repaglinide, fexofenadine, and atorvastatin were greater than 1.0 (3.44, 2.39, 2.27, and 1.90, respectively) in the presence of a P-gp inhibitor, thus suggesting the potential involvement of transporters in the B membrane. The Michaelis constant for quinidine for P-gp transport was 0.077 µM for the intracellular unbound concentration. These parameters were used to predict overall intestinal availability (FAFG) by applying an intestinal pharmacokinetic model, advanced translocation model (ATOM), in which permeability of A and B membranes accounted separately. The model predicted changes in the absorption location for P-gp substrates according to its inhibition, and FAFG values of 10 of 12 drugs, including quinidine at varying doses, were explained appropriately. SIGNIFICANCE STATEMENT: Pharmacokinetics has improved predictability by identifying the molecular entities of metabolism and transport and by using mathematical models to appropriately describe drug concentrations at the locations where they act. However, analyses of intestinal absorption so far have not been able to accurately consider the concentrations in the epithelial cells where P-glycoprotein and CYP3A4 exert effects. In this study, the limitation was removed by measuring the apical and basal membrane permeability separately and then analyzing these values using new appropriate models.

Urate-lowering therapy for gout and asymptomatic hyperuricemia in the pediatric population: a cross-sectional study of a Japanese health insurance database.

BACKGROUND: Our previous research showed that uric acid lowering therapy (ULT) for gout and hyperuricemia is being prescribed for pediatric patients even though these drugs have not been approved for use in children. However, the actual clinical situation has not been clearly elucidated. In this paper, we provide an in-depth look at the details of actual clinical practice. METHODS: This retrospective cross-sectional study accessed health insurance data for 696,277 children from April 2016 through March 2017 to identify pediatric patients with gout or asymptomatic hyperuricemia, calculate the proportion of patients prescribed ULTs, and analyze population characteristics. Adherence and mean dose for febuxostat and allopurinol, the most commonly prescribed drugs, were also analyzed. RESULTS: Among children with gout or asymptomatic hyperuricemia, we found that 35.1% (97/276) were prescribed ULT. This proportion increased with age, especially among males. By comorbidity, ULT was prescribed to 47.9% (46/96) of patients with kidney disease, 41.3% (26/63) for cardiovascular disease, 40.0% (6/15) for Down syndrome, and 27.1% (32/118) for metabolic syndrome. In patients with kidney disease, febuxostat was prescribed more than twice as frequently as allopurinol (28 vs. 12). Median values for the medication possession ratio (MPR) of febuxostat and allopurinol were 70.1 and 76.7%, respectively, and prescriptions were continued for a relatively long period for both drugs. Both drugs were prescribed at about half the adult dose for patients 6-11 years old and about the same as the adult dose for patients 12-18 years old. CONCLUSIONS: This study showed that the continuous management of serum uric acid is being explored using off-label use of ULT in pediatric patients with gout or asymptomatic hyperuricemia in Japan. Drug selection is based on patient characteristics such as sex, age, and comorbidities, and pediatric dosage is based on usage experience in adults. To develop appropriate pediatric ULT, clinical trials are needed on the efficacy and safety of ULT in the pediatric population. TRIAL REGISTRATION: UMIN000036029 .

A novel glucokinase activator TMG-123 causes long-lasting hypoglycemia and impairs spermatogenesis irreversibly in rats.

The importance of glucose is well known as an energy source in testes. In order to evaluate the effects of long-lasting hypoglycemia on testes, a novel glucokinase activator, TMG-123, was dosed to rats at 5, 20 and 100 mg/kg for 13 weeks. As a result, plasma glucose levels decreased for several hours with increasing doses over the dose range of 5 to 100 mg/kg. No toxicological findings attributable to the test article were observed in clinical observation, measurements of body weight and food consumption, necropsy, and organ weight measurement. Histopathology showed scattered degeneration of seminiferous tubules in testes, and exfoliation of germ cells related to the degeneration of seminiferous tubules was observed in the lumen of both epididymides in the same animals at the end of the dosing period. Similar histopathological findings were noted at the end of the recovery period. In addition, a fertility study was conducted at the same doses for 13 weeks for males and 5 weeks for females. Sperm analysis showed decreases in the sperm concentration and the motility index and an increase in the incidences of sperm malformations. However, there were no abnormalities in the copulation or fertility rate. These results suggest that long-lasting hypoglycemia in rats is harmful to spermatogenesis and the testicular damage does not recover.

Prevalence of gout and asymptomatic hyperuricemia in the pediatric population: a cross-sectional study of a Japanese health insurance database.

BACKGROUND: Although gout is rare in children, chronic sustained hyperuricemia can lead to monosodium urate deposits progressing to gout, just as in adults. This study assessed prevalence and characteristics of gout and asymptomatic hyperuricemia, and incidence of gouty arthritis in the pediatric population, using data from Japanese health insurance claims. The diagnosis and treatment of pediatric gout and hyperuricemia were analyzed, and specific characteristics of those patients were assessed. Since Japanese guidelines recommend treatment with uric acid lowering drugs for asymptomatic hyperuricemia as well as for gout, these data were also used to investigate the real-world use of uric acid lowering drugs in a pediatric population. METHODS: This cross-sectional study was based on a 2016-2017 Japanese health insurance claims database, one of the largest epidemiology claims databases available in Japan, which included 356,790 males and 339,487 females 0-18 years of age. Outcomes were measured for prevalence, patient characteristics, treatment with uric acid lowering drugs for gout and asymptomatic hyperuricemia, and prevalence and incidence of gouty arthritis. Because uric acid can be elevated by some forms of chemotherapy, data from patients under treatment for malignancies were excluded from consideration. RESULTS: Total prevalence of gout and asymptomatic hyperuricemia in 0-18 year-olds was 0.040% (276/696,277 patients), with gout prevalence at 0.007% (48/696,277) and asymptomatic hyperuricemia at 0.033% (228/696,277). Prevalence of gout and asymptomatic hyperuricemia was highest in adolescent males, at 0.135% (176/130,823). The most common comorbidities for gout and asymptomatic hyperuricemia were metabolic syndrome at 42.8% (118/276) and kidney disease at 34.8% (96/276). Of the patients diagnosed with gout or asymptomatic hyperuricemia, 35.1% (97/276) were treated with uric acid lowering drugs. Gouty arthritis developed in 43.8% (21/48) of gout patients during the study, at an incidence of 0.65 flares/person-year. CONCLUSIONS: Even the pediatric population could be affected by asymptomatic hyperuricemia, gout, and gouty arthritis, and uric acid lowering drugs are being used in this population even though those drugs have not been approved for pediatric indications. Such off-label use may indicate a potential need for therapeutic agents in this population. TRIAL REGISTRATION: UMIN000036029 .

The effects of long-lasting hypoglycemia on male reproductive organs in rats.

Glucose has an important role in spermatogenesis. Nevertheless there are few reports in which the effects of long-lasting hypoglycemia on male reproductive organs have been evaluated. Therefore, insulin was administered subcutaneously at 100, 200, and 400 IU/kg to male rats twice a day for one month. This treatment regimen produced plasma glucose levels that rapidly decreased after treatment, with decreased glucose levels lasting for several hours after each administration on the first and final treatment days. During the treatment period, no abnormalities in clinical signs or body weight were observed. No statistically significant differences were noted in the weights of testes, epididymides, prostates and seminal vesicles, or pituitary glands. Histopathological examination revealed that the insulin-treated animals exhibited degeneration of seminiferous tubules in the testes and exfoliation of germ cells in the lumens of epididymides as a secondary change related to the testicular lesions. The incidences of the histopathological findings were found to be proportional to insulin dose. Sperm analysis of the group receiving the highest dosage indicated that the sperm concentration tended to decrease and the incidences of sperm malformations tended to increase. Our results suggest that long-lasting hypoglycemia affects male reproductive organs in rats.