Quantification of spatial pharmacogene expression heterogeneity in breast tumors.

BACKGROUND: Chemotherapeutic drug concentrations vary across different regions of tumors and this is thought to be involved in development of chemotherapy resistance. Insufficient drug delivery to some regions of the tumor may be due to spatial differences in expression of genes involved in the disposition, transport, and detoxification of drugs (pharmacogenes). Therefore, in this study, we analyzed the spatial expression of 286 pharmacogenes in six breast cancer tissues using the recently developed Visium spatial transcriptomics platform to (1) determine if these pharmacogenes are expressed heterogeneously across tumor tissue and (2) to determine which pharmacogenes have the most spatial expression heterogeneity. METHODS AND RESULTS: The spatial transcriptomics technology sequences the transcriptome of 55 um diameter barcoded sections (spots) across a tissue sample. We analyzed spatial gene expression profiles of four biobank-sourced breast tumor samples in addition to two breast tumor sample datasets from 10× Genomics. We define heterogeneity as the interquartile range of read counts. Collectively, we identified 8887 spots in tumor regions, 3814 in stroma, 44 in lymphocytes, and 116 in normal regions based on pathologist annotation of the tissues. We showed statistically significant differences in expression of pharmacogenes in tumor regions compared to surrounding non-tumor regions. We also observed that the most heterogeneously expressed genes within tumor regions were involved in reactive oxygen species (ROS) handling and detoxification mechanisms. GPX4, GSTP1, MGST3, SOD1, CYP4Z1, CYB5R3, GSTK1, and NAT1 showed the most heterogeneous expression within tumor regions. CONCLUSIONS: The heterogeneous expression of these pharmacogenes may have important implications for cancer therapy due to their ability to impact drug distribution and efficacy throughout the tumor. Our results suggest that chemoresistance caused by expression of GPX4, GSTP1, MGST3, and SOD1 may be intrinsic, not acquired, since the heterogeneity is not specific to chemotherapy-treated samples or cell type. Additionally, we identified candidate chemoresistance pharmacogenes that can be further tested through focused follow-up studies.

Does lack of exposure to individual antidepressants at different points during pregnancy associate with reduced risk of adverse newborn outcomes?

BACKGROUND: The objective of this study was to determine if the lack of exposure to individual antidepressants at certain times in pregnancy improved maternal and infant outcomes. METHODS: This was a retrospective cohort study of 2741 pregnant women prescribed antidepressant(s) before or during pregnancy. Data were obtained from electronic medical records. Analysis was limited to women prescribed one of five antidepressants (bupropion, citalopram, escitalopram, fluoxetine, sertraline). Period of exposure was determined using prescription order dates. Primary outcomes were neonatal intensive care unit (NICU) admission and adaptation syndrome in the newborn. Logistic regression, adjusted for maternal age, race, and insurance, compared consistent exposure throughout pregnancy versus (A) no exposure in the third trimester, (B) no exposure early in pregnancy, and (C) exposure in the midtrimester alone. RESULTS: Compared to women prescribed an antidepressant continually throughout pregnancy, NICU admission was less likely for women lacking exposure in the third trimester if they had been taking bupropion (aOR 0.43, 95% CI 0.21-0.90) or escitalopram (aOR 0.49, 95% CI 0.28-0.85). Women previously taking escitalopram but lacking third trimester exposure also had lower odds of adaptation syndrome (aOR 0.19, 95% CI 0.07-0.48). No differences were found in other outcomes for women taking other antidepressants or for any outcomes for women who lacked early pregnancy drug exposure compared to exposure throughout pregnancy. CONCLUSION: For the five antidepressants included in this study, lack of exposure early or late in pregnancy compared to consistent exposure throughout pregnancy generally did not change newborn outcomes. The exceptions were bupropion and escitalopram, where lack of exposure in the third trimester associated with lower rates of adaptation syndrome or NICU admission. These data may help pregnant women with depression in need of drug therapy to have informed discussions with providers about the potential risks and benefits to continuing or stopping drugs at different times during pregnancy.

Physiologically based pharmacokinetic modelling in pregnancy: Model reproducibility and external validation.

AIMS: Physiologically based pharmacokinetic (PBPK) models have been previously developed for betamethasone and buprenorphine for pregnant women. The goal of this work was to replicate and reassess these models using data from recently completed studies. METHODS: Betamethasone and buprenorphine PBPK models were developed in Simcyp V19 based on prior publications using V17 and V15. Ability to replicate models was verified by comparing predictions in V19 to those previously published. Once replication was verified, models were reassessed by comparing predictions to observed data from additional studies in pregnant women. Model performance was based upon visual inspection of concentration vs. time profiles, and comparison of pharmacokinetic parameters. Models were deemed reproducible if parameter estimates were within 10% of previously reported values. External validations were considered acceptable if the predicted area under the concentration-time curve (AUC) and peak plasma concentration fell within 2-fold of the observed. RESULTS: The betamethasone model was successfully replicated using Simcyp V19, with ratios of reported (V17) to reproduced (V19) peak plasma concentration of 0.98-1.04 and AUC of 0.95-1.07. The model-predicted AUC ratios ranged from 0.98-1.79 compared to external data. The previously published buprenorphine PBPK model was not reproducible, as we predicted intravenous clearance of 70% that reported previously (both in Simcyp V15). CONCLUSION: While high interstudy variability was observed in the newly available clinical data, the PBPK model sufficiently predicted changes in betamethasone exposure across gestation. Model reproducibility and reassessment with external data are important for the advancement of the discipline. PBPK modelling publications should contain sufficient detail and clarity to enable reproducibility.

Application of machine learning to predict reduction in total PANSS score and enrich enrollment in schizophrenia clinical trials.

Clinical trial efficiency, defined as facilitating patient enrollment, and reducing the time to reach safety and efficacy decision points, is a critical driving factor for making improvements in therapeutic development. The present work evaluated a machine learning (ML) approach to improve phase II or proof-of-concept trials designed to address unmet medical needs in treating schizophrenia. Diagnostic data from the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) trial were used to develop a binary classification ML model predicting individual patient response as either "improvement," defined as greater than 20% reduction in total Positive and Negative Syndrome Scale (PANSS) score, or "no improvement," defined as an inadequate treatment response (<20% reduction in total PANSS). A random forest algorithm performed best relative to other tree-based approaches in model ability to classify patients after 6 months of treatment. Although model ability to identify true positives, a measure of model sensitivity, was poor (<0.2), its specificity, true negative rate, was high (0.948). A second model, adapted from the first, was subsequently applied as a proof-of-concept for the ML approach to supplement trial enrollment by identifying patients not expected to improve based on their baseline diagnostic scores. In three virtual trials applying this screening approach, the percentage of patients predicted to improve ranged from 46% to 48%, consistently approximately double the CATIE response rate of 22%. These results show the promising application of ML to improve clinical trial efficiency and, as such, ML models merit further consideration and development.