Association of Systemic Diseases With Surgical Treatment for Obstructive Sleep Apnea Compared With Continuous Positive Airway Pressure.

Importance: The efficacy of surgical treatments for obstructive sleep apnea (OSA) is variable when considering only the Apnea Hypopnea Index as the treatment end point. However, only a few studies have shown an association between these procedures and improved clinically relevant outcomes, such as cardiovascular, endocrine, and neurological sequelae of OSA. Objective: To evaluate the association of surgery for OSA with clinically relevant outcomes. Design, Setting, and Participants: This retrospective cohort study used the Truven MarketScan Database from January 1, 2007, to December 31, 2015, to identify all patients diagnosed with OSA who received a prescription of continuous positive airway pressure (CPAP), were 40 to 89 years of age, and had at least 3 years of data on file. Data were analyzed September 19, 2019. Interventions: Soft tissue and skeletal surgical procedures for the treatment of OSA. Main Outcomes and Measures: The occurrence of cardiovascular, neurological, and endocrine complications was compared in patients who received CPAP alone and those who received surgery. High-dimensionality propensity score matching was used to adjust the models for confounders. Kaplan-Meier survival analysis with a log-rank test was used to compare differences in survival curves. Findings: A total of 54 224 patients were identified (33 405 men [61.6%]; mean [SD] age, 55.1 [9.2] years), including a cohort of 49 823 patients who received CPAP prescription alone (mean [SD] age, 55.5 [9.4] years) and 4269 patients who underwent soft tissue surgery (mean [SD] age, 50.3 [7.0] years). The median follow-up time was 4.47 (interquartile range, 3-8) years after the index CPAP prescription. In the unadjusted model, soft tissue surgery was associated with decreased cardiovascular (hazard ratio [HR], 0.92; 95% CI, 0.86-0.98), neurological (HR, 0.49; 95% CI, 0.39-0.61), and endocrine (HR, 0.80; 95% CI, 0.74-0.86) events. This finding was maintained in the adjusted model (HR for cardiovascular events, 0.91 [95% CI, 0.83-1.00]; HR for neurological events, 0.67 [95% CI, 0.51-0.89]; HR for endocrine events, 0.82 [95% CI, 0.74-0.91]). Skeletal surgery (n = 114) and concomitant skeletal and soft tissue surgery (n = 18) did not demonstrate significant differences in rates of development of systemic complications. Conclusions and Relevance: In this cohort study, soft tissue surgery for OSA was associated with lower rates of development of cardiovascular, neurological, and endocrine systemic complications compared with CPAP prescription in a large convenience sample of the working insured US adult population. These findings suggest that surgery should be part of the early treatment algorithm in patients at high risk of CPAP failure or nonadherence.

Predicting need for advanced illness or palliative care in a primary care population using electronic health record data.

Timely outreach to individuals in an advanced stage of illness offers opportunities to exercise decision control over health care. Predictive models built using Electronic health record (EHR) data are being explored as a way to anticipate such need with enough lead time for patient engagement. Prior studies have focused on hospitalized patients, who typically have more data available for predicting care needs. It is unclear if prediction driven outreach is feasible in the primary care setting. In this study, we apply predictive modeling to the primary care population of a large, regional health system and systematically examine the impact of technical choices, such as requiring a minimum number of health care encounters (data density requirements) and aggregating diagnosis codes using Clinical Classifications Software (CCS) groupings to reduce dimensionality, on model performance in terms of discrimination and positive predictive value. We assembled a cohort of 349,667 primary care patients between 65 and 90 years of age who sought care from Sutter Health between July 1, 2011 and June 30, 2014, of whom 2.1% died during the study period. EHR data comprising demographics, encounters, orders, and diagnoses for each patient from a 12 month observation window prior to the point when a prediction is made were extracted. L1 regularized logistic regression and gradient boosted tree models were fit to training data and tuned by cross validation. Model performance in predicting one year mortality was assessed using held-out test patients. Our experiments systematically varied three factors: model type, diagnosis coding, and data density requirements. We found substantial, consistent benefit from using gradient boosting vs logistic regression (mean AUROC over all other technical choices of 84.8% vs 80.7% respectively). There was no benefit from aggregation of ICD codes into CCS code groups (mean AUROC over all other technical choices of 82.9% vs 82.6% respectively). Likewise increasing data density requirements did not affect discrimination (mean AUROC over other technical choices ranged from 82.5% to 83%). We also examine model performance as a function of lead time, which is the interval between death and when a prediction was made. In subgroup analysis by lead time, mean AUROC over all other choices ranged from 87.9% for patients who died within 0 to 3 months to 83.6% for those who died 9 to 12 months after prediction time.

Improving palliative care with deep learning.

BACKGROUND: Access to palliative care is a key quality metric which most healthcare organizations strive to improve. The primary challenges to increasing palliative care access are a combination of physicians over-estimating patient prognoses, and a shortage of palliative staff in general. This, in combination with treatment inertia can result in a mismatch between patient wishes, and their actual care towards the end of life. METHODS: In this work, we address this problem, with Institutional Review Board approval, using machine learning and Electronic Health Record (EHR) data of patients. We train a Deep Neural Network model on the EHR data of patients from previous years, to predict mortality of patients within the next 3-12 month period. This prediction is used as a proxy decision for identifying patients who could benefit from palliative care. RESULTS: The EHR data of all admitted patients are evaluated every night by this algorithm, and the palliative care team is automatically notified of the list of patients with a positive prediction. In addition, we present a novel technique for decision interpretation, using which we provide explanations for the model's predictions. CONCLUSION: The automatic screening and notification saves the palliative care team the burden of time consuming chart reviews of all patients, and allows them to take a proactive approach in reaching out to such patients rather then relying on referrals from the treating physicians.

Kinome Screen Identifies PFKFB3 and Glucose Metabolism as Important Regulators of the Insulin/Insulin-like Growth Factor (IGF)-1 Signaling Pathway.

The insulin/insulin-like growth factor (IGF)-1 signaling pathway (ISP) plays a fundamental role in long term health in a range of organisms. Protein kinases including Akt and ERK are intimately involved in the ISP. To identify other kinases that may participate in this pathway or intersect with it in a regulatory manner, we performed a whole kinome (779 kinases) siRNA screen for positive or negative regulators of the ISP, using GLUT4 translocation to the cell surface as an output for pathway activity. We identified PFKFB3, a positive regulator of glycolysis that is highly expressed in cancer cells and adipocytes, as a positive ISP regulator. Pharmacological inhibition of PFKFB3 suppressed insulin-stimulated glucose uptake, GLUT4 translocation, and Akt signaling in 3T3-L1 adipocytes. In contrast, overexpression of PFKFB3 in HEK293 cells potentiated insulin-dependent phosphorylation of Akt and Akt substrates. Furthermore, pharmacological modulation of glycolysis in 3T3-L1 adipocytes affected Akt phosphorylation. These data add to an emerging body of evidence that metabolism plays a central role in regulating numerous biological processes including the ISP. Our findings have important implications for diseases such as type 2 diabetes and cancer that are characterized by marked disruption of both metabolism and growth factor signaling.

The STARD9/Kif16a kinesin associates with mitotic microtubules and regulates spindle pole assembly.

During cell division, cells form the microtubule-based mitotic spindle, a highly specialized and dynamic structure that mediates proper chromosome transmission to daughter cells. Cancer cells can show perturbed mitotic spindles and an approach in cancer treatment has been to trigger cell killing by targeting microtubule dynamics or spindle assembly. To identify and characterize proteins necessary for spindle assembly, and potential antimitotic targets, we performed a proteomic and genetic analysis of 592 mitotic microtubule copurifying proteins (MMCPs). Screening for regulators that affect both mitosis and apoptosis, we report the identification and characterization of STARD9, a kinesin-3 family member, which localizes to centrosomes and stabilizes the pericentriolar material (PCM). STARD9-depleted cells have fragmented PCM, form multipolar spindles, activate the spindle assembly checkpoint (SAC), arrest in mitosis, and undergo apoptosis. Interestingly, STARD9-depletion synergizes with the chemotherapeutic agent taxol to increase mitotic death, demonstrating that STARD9 is a mitotic kinesin and a potential antimitotic target.

Deep RNA sequencing analysis of readthrough gene fusions in human prostate adenocarcinoma and reference samples.

BACKGROUND: Readthrough fusions across adjacent genes in the genome, or transcription-induced chimeras (TICs), have been estimated using expressed sequence tag (EST) libraries to involve 4-6% of all genes. Deep transcriptional sequencing (RNA-Seq) now makes it possible to study the occurrence and expression levels of TICs in individual samples across the genome. METHODS: We performed single-end RNA-Seq on three human prostate adenocarcinoma samples and their corresponding normal tissues, as well as brain and universal reference samples. We developed two bioinformatics methods to specifically identify TIC events: a targeted alignment method using artificial exon-exon junctions within 200,000 bp from adjacent genes, and genomic alignment allowing splicing within individual reads. We performed further experimental verification and characterization of selected TIC and fusion events using quantitative RT-PCR and comparative genomic hybridization microarrays. RESULTS: Targeted alignment against artificial exon-exon junctions yielded 339 distinct TIC events, including 32 gene pairs with multiple isoforms. The false discovery rate was estimated to be 1.5%. Spliced alignment to the genome was less sensitive, finding only 18% of those found by targeted alignment in 33-nt reads and 59% of those in 50-nt reads. However, spliced alignment revealed 30 cases of TICs with intervening exons, in addition to distant inversions, scrambled genes, and translocations. Our findings increase the catalog of observed TIC gene pairs by 66%.We verified 6 of 6 predicted TICs in all prostate samples, and 2 of 5 predicted novel distant gene fusions, both private events among 54 prostate tumor samples tested. Expression of TICs correlates with that of the upstream gene, which can explain the prostate-specific pattern of some TIC events and the restriction of the SLC45A3-ELK4 e4-e2 TIC to ERG-negative prostate samples, as confirmed in 20 matched prostate tumor and normal samples and 9 lung cancer cell lines. CONCLUSIONS: Deep transcriptional sequencing and analysis with targeted and spliced alignment methods can effectively identify TIC events across the genome in individual tissues. Prostate and reference samples exhibit a wide range of TIC events, involving more genes than estimated previously using ESTs. Tissue specificity of TIC events is correlated with expression patterns of the upstream gene. Some TIC events, such as MSMB-NCOA4, may play functional roles in cancer.

In vivo antitumor activity of MEK and phosphatidylinositol 3-kinase inhibitors in basal-like breast cancer models.

PURPOSE: The pathways underlying basal-like breast cancer are poorly understood, and as yet, there is no approved targeted therapy for this disease. We investigated the role of mitogen-activated protein kinase kinase (MEK) and phosphatidylinositol 3-kinase (PI3K) inhibitors as targeted therapies for basal-like breast cancer. EXPERIMENTAL DESIGN: We used pharmacogenomic analysis of a large panel of breast cancer cell lines with detailed accompanying molecular information to identify molecular predictors of response to a potent and selective inhibitor of MEK and also to define molecular mechanisms underlying combined MEK and PI3K targeting in basal-like breast cancer. Hypotheses were confirmed by testing in multiple tumor xenograft models. RESULTS: We found that basal-like breast cancer models have an activated RAS-like transcriptional program and show greater sensitivity to a selective inhibitor of MEK compared with models representative of other breast cancer subtypes. We also showed that loss of PTEN is a negative predictor of response to MEK inhibition, that treatment with a selective MEK inhibitor caused up-regulation of PI3K pathway signaling, and that dual blockade of both PI3K and MEK/extracellular signal-regulated kinase signaling synergized to potently impair the growth of basal-like breast cancer models in vitro and in vivo. CONCLUSIONS: Our studies suggest that single-agent MEK inhibition is a promising therapeutic modality for basal-like breast cancers with intact PTEN, and also provide a basis for rational combination of MEK and PI3K inhibitors in basal-like cancers with both intact and deleted PTEN.

Tumor-driven paracrine platelet-derived growth factor receptor alpha signaling is a key determinant of stromal cell recruitment in a model of human lung carcinoma.

Activated fibroblasts are thought to play important roles in the progression of many solid tumors, but little is known about the mechanisms responsible for the recruitment of fibroblasts in tumors. Using several methods, we identified platelet-derived growth factor A (PDGFA) as the major fibroblast chemoattractant and mitogen from conditioned medium generated by the Calu-6 lung carcinoma cell line. In addition, we showed that Calu-6 tumors express significant levels of PDGFC, and that the levels of expression of these two PDGFRalpha ligands correlate strongly with the degree of stromal fibroblast infiltration into the tumor mass. The most intense expression of PDGFRalpha was observed in fibroblasts in the tumor outer rim. We subsequently showed that disrupting PDGFRalpha-mediated signaling results in significant inhibition of tumor growth in vivo. Furthermore, analysis of a compendium of microarray data revealed significant expression of PDGFA, PDGFC, and PDGFRalpha in human lung tumors. We propose that therapies targeting this stromal cell type may be effective in treating certain types of solid tumors.

Expression profiling the human septin gene family.

The septins are an evolutionarily conserved family of GTP-binding proteins involved in diverse processes including vesicle trafficking, apoptosis, remodelling of the cytoskeleton, infection, neurodegeneration, and neoplasia. The present paper reports a comprehensive study of septin gene expression by DNA microarray methods in 10 360 samples of normal, diseased, and tumour tissues. A novel septin, SEPT13, has been identified and is shown to be related to SEPT7. It is shown that SEPT13 and the other known human septins are expressed in all tissue types but some show high expression in lymphoid (SEPT1, 6, 9, and 12) or brain tissues (SEPT2, 3, 4, 5, 7, 8, and 11). For a given septin, some isoforms are highly expressed in the brain and others are not. For example, SEPT8_v2 and v1, 1* and 3 are highly expressed in the brain and cluster with SEPT2, 3, 4, 5, 7, and 11. However, a probe set specific for SEPT8_v1 with low brain expression clusters away from this set. Similarly, SEPT4 has lymphoid and non-lymphoid forms; SEPT2 has lymphoid and central nervous system (CNS) forms; and SEPT6 and SEPT9 are elevated in lymphoid tissues but both have forms that cluster away from the lymphoid forms. Perturbation of septin expression was widespread in disease and tumours of the various tissues examined, particularly for conditions of the CNS, where alterations in all 13 septin genes were identified. This analysis provides a comprehensive catalogue of the septin family in health and disease. It is a key step in understanding the role of septins in physiological and pathological states and provides insight into the complexity of septin biology.