Bayesian Method for Causal Discovery of Latent-Variable Models from a Mixture of Experimental and Observational Data

Inhibitor of DNA Binding Protein 3 (ID3) and Nuclear Respiratory Factor 1 (NRF1) Mediated Transcriptional Gene Signatures are Associated with the Severity of Cerebral Amyloid Angiopathy

Cerebral amyloid angiopathy (CAA) is a degenerative vasculopathy. We have previously shown that transcription regulating proteins- inhibitor of DNA binding protein 3 (ID3) and the nuclear respiratory factor 1 (NRF1) contribute to vascular dysregulation. In this study, we have identified sex specific ID3 and NRF1-mediated gene networks in CAA patients diagnosed with Alzheimer's Disease (AD). High expression of ID3 mRNA coupled with low NRF1 mRNA levels was observed in the temporal cortex of men and women CAA patients. Low NRF1 mRNA expression in the temporal cortex was found in men with severe CAA. High ID3 expression was found in women with the genetic risk factor APOE4. Low NRF1 expression was also associated with APOE4 in women with CAA. Genome wide transcriptional activity of both ID3 and NRF1 paralleled their mRNA expression levels. Sex specific differences in transcriptional gene signatures of both ID3 and NRF1 were observed. These findings were further corroborated by Bayesian machine learning and the GeNIe simulation models. Dynamic machine learning using a Monte Carlo Markov Chain (MCMC) gene ordering approach revealed that ID3 was associated with disease severity in women. NRF1 was associated with CAA and severity of this disease in men. These findings suggest that aberrant ID3 and NRF1 activity presumably plays a major role in the pathogenesis and severity of CAA. Further analyses of ID3- and NRF1-regulated molecular drivers of CAA may provide new targets for personalized medicine and/or prevention strategies against CAA.

Changes in Older Adult Trauma Quality When Evaluated Using Longer-Term Outcomes vs In-Hospital Mortality

Importance

Lack of knowledge about longer-term outcomes remains a critical blind spot for trauma systems. Recent efforts have expanded trauma quality evaluation to include a broader array of postdischarge quality metrics. It remains unknown how such quality metrics should be used.

Objective

To examine the utility of implementing recommended postdischarge quality metrics as a composite score and ascertain how composite score performance compares with that of in-hospital mortality for evaluating associations with hospital-level factors.

Design, Setting, and Participants

This national hospital-level quality assessment evaluated hospital-level care quality using 100% Medicare fee-for-service claims of older adults (aged ≥65 years) hospitalized with primary diagnoses of trauma, hip fracture, and severe traumatic brain injury (TBI) between January 1, 2014, and December 31, 2015. Hospitals with annual volumes encompassing 10 or more of each diagnosis were included. The data analysis was performed between January 1, 2021, and December 31, 2022.

Exposures

Reliability-adjusted quality metrics used to calculate composite scores included hospital-specific performance on mortality, readmission, and patients' average number of healthy days at home (HDAH) within 30, 90, and 365 days among older adults hospitalized with all forms of trauma, hip fracture, and severe TBI.

Main Outcomes and Measures

Associations with hospital-level factors were compared using volume-weighted multivariable logistic regression.

Results

A total of 573 554 older adults (mean [SD] age, 83.1 [8.3] years; 64.8% female; 35.2% male) from 1234 hospitals were included. All 27 reliability-adjusted postdischarge quality metrics significantly contributed to the composite score. The most important drivers were 30- and 90-day readmission, patients' average number of HDAH within 365 days, and 365-day mortality among all trauma patients. Associations with hospital-level factors revealed predominantly anticipated trends when older adult trauma quality was evaluated using composite scores (eg, worst performance was associated with decreased older adult trauma volume [odds ratio, 0.89; 95% CI, 0.88-0.90]). Results for in-hospital mortality showed inverted associations for each considered hospital-level factor and suggested that compared with nontrauma centers, level 1 trauma centers had a 17 times higher risk-adjusted odds of worst (highest quantile) vs best (lowest quintile) performance (odds ratio, 17.08; 95% CI, 16.17-18.05).

Conclusions and Relevance

The study results challenge historical notions about the adequacy of in-hospital mortality as the single measure of older adult trauma quality and suggest that, when it comes to older adults, decisions about how quality is evaluated can profoundly alter understandings of what constitutes best practices for care. Composite scores appear to offer a promising means by which postdischarge quality metrics could be used.

A Better Mechanistic Understanding of Big Data through an Order Search Using Causal Bayesian Networks

Every year, biomedical data is increasing at an alarming rate and is being collected from many different sources, such as hospitals (clinical Big Data), laboratories (genomic and proteomic Big Data), and the internet (online Big Data). This article presents and evaluates a practical causal discovery algorithm that uses modern statistical, machine learning, and informatics approaches that have been used in the learning of causal relationships from biomedical Big Data, which in turn integrates clinical, omics (genomic and proteomic), and environmental aspects. The learning of causal relationships from data using graphical models does not address the hidden (unknown or not measured) mechanisms that are inherent to most measurements and analyses. Also, many algorithms lack a practical usage since they do not incorporate current mechanistic knowledge. This paper proposes a practical causal discovery algorithm using causal Bayesian networks to gain a better understanding of the underlying mechanistic process that generated the data. The algorithm utilizes model averaging techniques such as searching through a relative order (e.g., if gene A is regulating gene B, then we can say that gene A is of a higher order than gene B) and incorporates relevant prior mechanistic knowledge to guide the Markov chain Monte Carlo search through the order. The algorithm was evaluated by testing its performance on datasets generated from the ALARM causal Bayesian network. Out of the 37 variables in the ALARM causal Bayesian network, two sets of nine were chosen and the observations for those variables were provided to the algorithm. The performance of the algorithm was evaluated by comparing its prediction with the generating causal mechanism. The 28 variables that were not in use are referred to as hidden variables and they allowed for the evaluation of the algorithm’s ability to predict hidden confounded causal relationships. The algorithm’s predicted performance was also compared with other causal discovery algorithms. The results show that incorporating order information provides a better mechanistic understanding even when hidden confounded causes are present. The prior mechanistic knowledge incorporated in the Markov chain Monte Carlo search led to the better discovery of causal relationships when hidden variables were involved in generating the simulated data.

Nuclear respiratory factor 1 transcriptomic signatures as prognostic indicators of recurring aggressive mesenchymal glioblastoma and resistance to therapy in White American females

PURPOSE

The mechanisms contributing to recurrence of glioblastoma (GBM), an aggressive neuroepithelial brain tumor, remain unknown. We have recently shown that nuclear respiratory factor 1 (NRF1) is an oncogenic transcription factor and its transcriptional activity is associated with the progression and prognosis of GBM. Herein, we extend our efforts to (1) identify influential NRF1-driven gene and microRNA (miRNA) expression for the aggressiveness of mesenchymal GBM; and (2) understand the molecular basis for its poor response to therapy.

METHODS

Clinical data and RNA-Seq from four independent GBM cohorts were analyzed by Bayesian Network Inference with Java Objects (BANJO) and Markov chain Monte Carlo (MCMC)-based gene order to identify molecular drivers of mesenchymal GBM as well as prognostic indicators of poor response to radiation and chemotherapy.

RESULTS

We are the first to report sex-specific NRF1 motif enriched gene signatures showing increased susceptibility to GBM. Risk estimates for GBM were increased by greater than 100-fold with the joint effect of NRF1-driven gene signatures-CDK4, DUSP6, MSH2, NRF1, and PARK7 in female GBM patients and CDK4, CASP2, H6PD, and NRF1 in male GBM patients. NRF1-driven causal Bayesian network genes were predictive of poor survival and resistance to chemoradiation in IDH1 wild-type mesenchymal GBM patients. NRF1-regulatable miRNAs were also associated with poor response to chemoradiation therapy in female IDH1 wild-type mesenchymal GBM. Stable overexpression of NRF1 reprogramed human astrocytes into neural stem cell-like cells expressing SOX2 and nestin. These cells differentiated into neurons and form tumorospheroids.

CONCLUSIONS

In summary, our novel discovery shows that NRF1-driven causal genes and miRNAs involved in cancer cell stemness and mesenchymal features contribute to cancer aggressiveness and recurrence of aggressive therapy-resistant glioblastoma.