A White-Box Machine Learning Approach for Revealing Antibiotic Mechanisms of Action.

Current machine learning techniques enable robust association of biological signals with measured phenotypes, but these approaches are incapable of identifying causal relationships. Here, we develop an integrated "white-box" biochemical screening, network modeling, and machine learning approach for revealing causal mechanisms and apply this approach to understanding antibiotic efficacy. We counter-screen diverse metabolites against bactericidal antibiotics in Escherichia coli and simulate their corresponding metabolic states using a genome-scale metabolic network model. Regression of the measured screening data on model simulations reveals that purine biosynthesis participates in antibiotic lethality, which we validate experimentally. We show that antibiotic-induced adenine limitation increases ATP demand, which elevates central carbon metabolism activity and oxygen consumption, enhancing the killing effects of antibiotics. This work demonstrates how prospective network modeling can couple with machine learning to identify complex causal mechanisms underlying drug efficacy.

Clinician and patient attitudes toward use of organs from hepatitis C viremic donors and their impact on acceptance: A contemporary review.

BACKGROUND: The use of Hepatitis C (HCV) NAT positive allografts remains unusual and is clustered at few centers. We conducted a contemporary literature review to assess whether patient and clinician attitudes toward viremic organs impact acceptance. METHODS: Databases including PubMed, MEDLINE, and SCOPUS databases were reviewed to identify studies focused on evaluating patient and provider perceptions of HCV NAT positive organ use within the DAA era (January 2015-April 2021). Search included MeSH terms related to Hepatitis C, transplantation, and patient and clinician attitudes. Two investigators extracted study characteristics including information on willingness to accept viremic organs, HCV-specific outcomes knowledge, HCV-specific concerns, and factors that contributed to acceptance or non-acceptance. RESULTS: Eight studies met all inclusion criteria. These included three pretransplant patient-directed studies, two post-transplant patient-directed studies, one pre- and post-transplant patient-directed study, and two clinician-directed studies. Common themes identified were concerns regarding HCV cure rates, viremic organ quality, DAA cost, stigma, and the possibility of HCV transmission to household members. The perception of decreased waitlist time was associated with viremic organ acceptance. Physician trust played a mixed role in acceptance patterns. CONCLUSIONS: Knowledge of high cure rates, shorter waitlist times, and higher organ quality appear to have the highest impact on organ acceptance.

Risk assessment in liver transplantation for hepatocellular carcinoma: long-term follow-up of a two-centre experience.

BACKGROUND: Liver transplantation (LT) is a well-established treatment for hepatocellular carcinoma (HCC), but there are ongoing debates regarding outcomes and selection. This study examines the experience of LT for HCC at a high-volume centre. METHODS: A prospectively maintained database was used to identify HCC patients undergoing LT from 2000 to 2020 with more than or equal to 3-years follow-up. Data were obtained from the centre database and electronic medical records. The Metroticket 2.0 HCC-specific 5-year survival scale was calculated for each patient. Kaplan-Meier and Cox-regression analyses were employed assessing survival between groups based on Metroticket score and individual donor and recipient risk factors. RESULTS: Five hundred sixty-nine patients met criteria. Median follow-up was 96.2 months (8.12 years; interquartile range 59.9-147.8). Three-year recurrence-free (RFS) and overall survival (OS) were 88.6% ( n =504) and 86.6% ( n =493). Five-year RFS and OS were 78.9% ( n =449) and 79.1% ( n =450). Median Metroticket 2.0 score was 0.9 (interquartile range 0.9-0.95). Tumour size greater than 3 cm ( P =0.012), increasing tumour number on imaging ( P =0.001) and explant pathology ( P <0.001) was associated with recurrence. Transplant within Milan ( P <0.001) or UCSF criteria ( P <0.001) had lower recurrence rates. Increasing alpha-fetoprotein (AFP)-values were associated with more HCC recurrence ( P <0.001) and reduced OS ( P =0.008). Chemoembolization was predictive of recurrence in the overall population ( P =0.043) and in those outside-Milan criteria ( P =0.038). A receiver-operator curve using Metroticket 2.0 identified an optimal cut-off of projected survival greater than or equal to 87.5% for predicting recurrence. This cut-off was able to predict RFS ( P <0.001) in the total cohort and predict both, RFS ( P =0.007) and OS ( P =0.016) outside Milan. Receipt of donation after brain death (DBD) grafts (55/478, 13%) or living-donor grafts (3/22, 13.6%) experienced better survival rates compared to donation after cardiac death (DCD) grafts ( n =15/58, 25.6%, P =0.009). Donor age was associated with a higher HCC recurrence ( P =0.006). Both total ischaemia time (TIT) greater than 6hours ( P =0.016) and increasing TIT correlated with higher HCC recurrence ( P =0.027). The use of DCD grafts for outside-Milan candidates was associated with increased recurrence ( P =0.039) and reduced survival ( P =0.033). CONCLUSION: This large two-centre analysis confirms favourable outcomes after LT for HCC. Tumour size and number, pre-transplant AFP, and Milan criteria remain important recipient HCC-risk factors. A higher donor risk (i.e. donor age, DCD grafts, ischaemia time) was associated with poorer outcomes.

Urine drug screen positive for cocaine and amphetamine is not an adverse risk factor for cardiovascular morbidity or mortality in trauma.

BACKGROUND: Urine drug screening (UDS) is a component of trauma workup and of perioperative risk evaluation. Illicit stimulant use has been associated with cardiovascular complications. This study investigates the impact of stimulant use and its interaction with surgery on cardiovascular complications in trauma patients. METHODS: Patients were identified from the 2017 National Trauma Data Bank. Univariate and multivariate analyses were used to evaluate the effect of amphetamine and cocaine on mortality, myocardial infarction (MI), and stroke. We evaluated three subsets: all screened patients, those who underwent surgery, and those whose surgery was immediate. Significance was tested with χ2 test for categorical variables, Student's t-test for continuous variables, and logistic regression for multivariate analysis. RESULTS: 317 688 (32.1%) patients underwent UDS. Multivariate analysis showed protective association between cocaine and mortality OR 0.9 (p=0.028). Cocaine was a non-significant predictor of MI and stroke: OR 0.63 (p=0.065) and 0.91 (p=0.502), respectively. Amphetamine was a non-significant predictor of mortality, MI, and stroke: OR 0.97 (p=0.405), 0.80 (p=0.283), and 1.02 (p=0.857), respectively.On univariate analysis, amphetamine showed a protective association with MI for all screened patients: relative risk (RR) 0.58 (p=0.005), and for surgical patients: RR 0.58 (p=0.019). Amphetamine showed a protective association with mortality for all three subsets: RR 0.83 (p<0.001), 0.78 (p<0.001), and 0.71 (p<0.001), respectively. Cocaine showed a protective association with MI for all screened patients: RR 0.45 (p=0.001), and for surgical patients: RR 0.44 (p=0.005). Cocaine showed a protective association with mortality for all three subsets: RR 0.76 (p<0.001), 0.71 (p<0.001), and 0.63 (p<0.001), respectively. DISCUSSION: UDS positive for cocaine or amphetamine is not an adverse risk factor in trauma, including trauma patients who underwent surgery. The apparent protective effects of illicit drugs warrant further investigation. LEVEL OF EVIDENCE: Therapeutic/care management, level IV.

Pharmaceutical Induction of PGC-1α Promotes Retinal Pigment Epithelial Cell Metabolism and Protects against Oxidative Damage.

Retinal pigment epithelium (RPE) dysfunction due to accumulation of reactive oxygen species and oxidative damage is a key event in the development of age-related macular degeneration (AMD). Here, we examine the therapeutic potential of ZLN005, a selective PGC-1α transcriptional regulator, in protecting RPE from cytotoxic oxidative damage. Gene expression analysis on ARPE-19 cells treated with ZLN005 shows robust upregulation of PGC-1α and its associated transcription factors, antioxidant enzymes, and mitochondrial genes. Energetic profiling shows that ZLN005 treatment enhances RPE mitochondrial function by increasing basal and maximal respiration rates, and spare respiratory capacity. In addition, ZLN005 robustly protects ARPE-19 cells from cell death caused by H2O2, ox-LDL, and NaIO3 without exhibiting any cytotoxicity under basal conditions. ZLN005 protection against H2O2-mediated cell death was lost in PGC-1α-silenced cells. Our data indicates that ZLN005 efficiently protects RPE cells from oxidative damage through selective induction of PGC-1α and its target antioxidant enzymes. ZLN005 may serve as a novel therapeutic agent for retinal diseases associated with RPE dystrophies.