Transcriptomic-Based Microenvironment Classification Reveals Precision Medicine Strategies for Pancreatic Ductal Adenocarcinoma.

BACKGROUND & AIMS: The complex tumor microenvironment (TME) of pancreatic ductal adenocarcinoma (PDAC) has hindered the development of reliable predictive biomarkers for targeted therapy and immunomodulatory strategies. A comprehensive characterization of the TME is necessary to advance precision therapeutics in PDAC. METHODS: A transcriptomic profiling platform for TME classification based on functional gene signatures was applied to 14 publicly available PDAC datasets (n = 1657) and validated in a clinically annotated independent cohort of patients with PDAC (n = 79). Four distinct subtypes were identified using unsupervised clustering and assessed to evaluate predictive and prognostic utility. RESULTS: TME classification using transcriptomic profiling identified 4 biologically distinct subtypes based on their TME immune composition: immune enriched (IE); immune enriched, fibrotic (IE/F); fibrotic (F); and immune depleted (D). The IE and IE/F subtypes demonstrated a more favorable prognosis and potential for response to immunotherapy compared with the F and D subtypes. Most lung metastases and liver metastases were subtypes IE and D, respectively, indicating the role of clonal phenotype and immune milieu in developing personalized therapeutic strategies. In addition, distinct TMEs with potential therapeutic implications were identified in treatment-naive primary tumors compared with tumors that underwent neoadjuvant therapy. CONCLUSIONS: This novel approach defines a distinct subgroup of PADC patients that may benefit from immunotherapeutic strategies based on their TME subtype and provides a framework to select patients for prospective clinical trials investigating precision immunotherapy in PDAC. Further, the predictive utility and real-world clinical applicability espoused by this transcriptomic-based TME classification approach will accelerate the advancement of precision medicine in PDAC.

Acute Kidney Injury in Burn Patients: Clinically Significant Over the Initial Hospitalization and 1 Year After Injury: An Original Retrospective Cohort Study.

OBJECTIVE: To examine the development of acute kidney injury (AKI) after burn injury as an independent risk factor for increased morbidity and mortality over initial hospitalization and 1-year follow-up. BACKGROUND: Variability in fluid resuscitation and difficulty recognizing early sepsis are major barriers to preventing AKI after burn injury. Expanding our understanding of the burden AKI has on the clinical course of burn patients would highlight the need for standardized protocols. METHODS: We queried the Healthcare Cost and Utilization Project State Inpatient Databases in the states of Florida and New York during the years 2009 to 2013 for patients over age 18 hospitalized with a primary diagnosis of burn injury using ICD-9 codes. We identified and grouped 18,155 patients, including 1476 with burns >20% total body surface area, by presence of AKI. Outcomes were compared in these cohorts via univariate analysis and multivariate logistic regression models. RESULTS: During initial hospitalization, AKI was associated with increased pulmonary failure, mechanical ventilation, pneumonia, myocardial infarction, length of stay, cost, and mortality, and also a lower likelihood of being discharged home. One year after injury, AKI was associated with development of chronic kidney disease, conversion to chronic dialysis, hospital readmission, and long-term mortality. CONCLUSIONS: AKI is associated with a profound and severe increase in morbidity and mortality in burn patients during initial hospitalization and up to 1 year after injury. Consensus protocols for initial burn resuscitation and early sepsis recognition and treatment are crucial to avoid the consequences of AKI after burn injury.

Automated machine learning (AutoML) can predict 90-day mortality after gastrectomy for cancer.

Early postoperative mortality risk prediction is crucial for clinical management of gastric cancer. This study aims to predict 90-day mortality in gastric cancer patients undergoing gastrectomy using automated machine learning (AutoML), optimize models for preoperative prediction, and identify factors influential in prediction. National Cancer Database was used to identify stage I-III gastric cancer patients undergoing gastrectomy between 2004 and 2016. 26 features were used to train predictive models using H2O.ai AutoML. Performance on validation cohort was measured. In 39,108 patients, 90-day mortality rate was 8.8%. The highest performing model was an ensemble (AUC = 0.77); older age, nodal ratio, and length of inpatient stay (LOS) following surgery were most influential for prediction. Removing the latter two parameters decreased model performance (AUC 0.71). For optimizing models for preoperative use, models were developed to first predict node ratio or LOS, and these predicted values were inputted for 90-day mortality prediction (AUC of 0.73-0.74). AutoML performed well in predicting 90-day mortality in a larger cohort of gastric cancer patients that underwent gastrectomy. These models can be implemented preoperatively to inform prognostication and patient selection for surgery. Our study supports broader evaluation and application of AutoML to guide surgical oncologic care.