Key Points Question Can a machine learning model provide survival risk stratification for patients with advanced oral cancer who have comprehensive clinicopathologic and genetic data? Findings In this 15-year cohort study of 334 patients, a risk stratification model using comprehensive clinicopathologic and genetic data accurately differentiated the high-risk group from the low-risk group in postoperative cancer-specific and locoregional recurrence–free survival for patients with advanced oral cancer. Meaning The proposed model demonstrated good discrimination in stratifying patients with different risks of survival by using comprehensive clinicopathologic and genetic data, which can provide additional personalized information for postoperative management of patients with advanced oral squamous cancer.

All dimensions deteriorated in the last year of life but with distinctive physical-psychological-social-spiritual/existential and overall QOL trajectories. Recognizing trajectory patterns and tipping points of accelerating deterioration in each dimension can help clinicians anticipate times of increased distress, initiate timely, effective interventions to relieve patient suffering, and facilitate high-quality end-of-life care tailored to patients' needs and preferences.

Prevalence of severe depressive symptoms increased as death approached and was associated with several modifiable factors. Healthcare professionals should become familiar with these factors to identify vulnerable patients. To decrease the likelihood of terminally ill cancer patients' severe depressive symptoms, they should receive effective interventions to manage their symptoms, appropriately foster social support to restore their fragile self-esteem due to depending on others, and lighten their SPB.

BackgroundPatient-derived xenograft (PDX) tumor model has become a new approach in identifying druggable tumor mutations, screening and evaluating personalized cancer drugs based on the mutated targets.MethodsWe established five nasopharyngeal carcinoma (NPC) PDXs in mouse model. Subsequently, whole-exome sequencing (WES) and genomic mutation analyses were performed to search for genetic alterations for new drug targets. Potential drugs were applied in two NPC PDX mice model to assess their anti-cancer activities. RNA sequencing and transcriptomic analysis were performed in one NPC PDX mice to correlate with the efficacy of the anti-cancer drugs.ResultsA relative high incident rate of copy number variations (CNVs) of cell cycle-associated genes. Among the five NPC-PDXs, three had cyclin D1 (CCND1) amplification while four had cyclin-dependent kinase inhibitor CDKN2A deletion. Furthermore, CCND1 overexpression was observed in > 90% FFPE clinical metastatic NPC tumors (87/91) and was associated with poor outcomes. CNV analysis disclosed that plasma CCND1/CDKN2A ratio is correlated with EBV DNA load in NPC patients’ plasma and could serve as a screening test to select potential CDK4/6 inhibitor treatment candidates. Based on our NPC PDX model and RNA sequencing, Palbociclib, a cyclin-dependent kinase inhibitor, proved to have anti-tumor effects by inducing G1 arrest. One NPC patient with liver metastatic was treated with Palbociclib, had stable disease response and a drop in Epstein Barr virus (EBV) EBV titer.ConclusionsOur integrated information of sequencing-based genomic studies and tumor transcriptomes with drug treatment in NPC-PDX models provided guidelines for personalized precision treatments and revealed a cyclin-dependent kinase inhibitor Palbociclib as a novel candidate drug for NPC.Electronic supplementary materialThe online version of this article (10.1186/s13046-018-0873-5) contains supplementary material, which is available to authorized users.