Background The recent Coronavirus Disease 2019 (COVID-19) pandemic has placed severe stress on healthcare systems worldwide, which is amplified by the critical shortage of COVID-19 tests. Methods In this study, we propose to generate a more accurate diagnosis model of COVID-19 based on patient symptoms and routine test results by applying machine learning to reanalyzing COVID-19 data from 151 published studies. We aim to investigate correlations between clinical variables, cluster COVID-19 patients into subtypes, and generate a computational classification model for discriminating between COVID-19 patients and influenza patients based on clinical variables alone. Results We discovered several novel associations between clinical variables, including correlations between being male and having higher levels of serum lymphocytes and neutrophils. We found that COVID-19 patients could be clustered into subtypes based on serum levels of immune cells, gender, and reported symptoms. Finally, we trained an XGBoost model to achieve a sensitivity of 92.5% and a specificity of 97.9% in discriminating COVID-19 patients from influenza patients. Conclusions We demonstrated that computational methods trained on large clinical datasets could yield ever more accurate COVID-19 diagnostic models to mitigate the impact of lack of testing. We also presented previously unknown COVID-19 clinical variable correlations and clinical subgroups.

Head and neck squamous cell carcinoma persists as one of the most common and deadly malignancies, with early detection and effective treatment still posing formidable challenges. To expand our currently sparse knowledge of the noncoding alterations involved in the disease and identify potential biomarkers and therapeutic targets, we globally profiled the dysregulation of small nucleolar and long noncoding RNAs in head and neck tumors. Using next-generation RNA-sequencing data from 40 pairs of tumor and matched normal tissues, we found 2808 long noncoding RNA (lncRNA) transcripts significantly differentially expressed by a fold change magnitude ≥2. Meanwhile, RNA-sequencing analysis of 31 tumor-normal pairs yielded 33 significantly dysregulated small nucleolar RNAs (snoRNA). In particular, we identified two dramatically downregulated lncRNAs and one down-regulated snoRNA whose expression levels correlated significantly with overall patient survival, suggesting their functional significance and clinical relevance in head and neck cancer pathogenesis. We confirmed the dysregulation of these noncoding RNAs in head and neck cancer cell lines derived from different anatomic sites, and determined that ectopic expression of the two lncRNAs inhibited key EMT and stem cell genes and reduced cellular proliferation and migration. As a whole, noncoding RNAs are pervasively dysregulated in head and squamous cell carcinoma. The precise molecular roles of the three transcripts identified warrants further characterization, but our data suggest that they are likely to play substantial roles in head and neck cancer pathogenesis and are significantly associated with patient survival.

Head and neck squamous cell carcinoma (HNSCC) is an aggressive disease marked by frequent recurrence and metastasis and stagnant survival rates. To enhance molecular knowledge of HNSCC and define a non-coding RNA (ncRNA) landscape of the disease, we profiled the transcriptome-wide dysregulation of long non-coding RNA (lncRNA), microRNA (miRNA), and PIWI-interacting RNA (piRNA) using RNA-sequencing data from 422 HNSCC patients in The Cancer Genome Atlas (TCGA). 307 non-coding transcripts differentially expressed in HNSCC were significantly correlated with patient survival, and associated with mutations in TP53, CDKN2A, CASP8, PRDM9, and FBXW7 and copy number variations in chromosomes 3, 5, 7, and 18. We also observed widespread ncRNA correlation to concurrent TP53 and chromosome 3p loss, a compelling predictor of poor prognosis in HNSCCs. Three selected ncRNAs were additionally associated with tumor stage, HPV status, and other clinical characteristics, and modulation of their expression in vitro reveals differential regulation of genes involved in epithelial-mesenchymal transition and apoptotic response. This comprehensive characterization of the HNSCC non-coding transcriptome introduces new layers of understanding for the disease, and nominates a novel panel of transcripts with potential utility as prognostic markers or therapeutic targets.

Cancer stem cells (CSCs) have been shown as a distinct population of cancer cells strongly implicated with resistance to conventional chemotherapy. Metformin, the most widely prescribed drug for diabetes, was reported to target cancer stem cells in various cancers. In this study, we sought to determine the effects of metformin on head and neck squamous cell carcinoma (HNSCC). CSCs and non-stem HNSCC cells were treated with metformin and cisplatin alone, and in combination, and cell proliferation levels were measured through MTS assays. Next, potential targets of metformin were explored through computational small molecule binding analysis. In contrast to the reported effects of metformin on CSCs in other cancers, our data suggests that metformin protects HNSCC CSCs against cisplatin in vitro. Treatment with metformin resulted in a dose-dependent induction of the stem cell genes CD44, BMI-1, OCT-4, and NANOG. On the other hand, we observed that metformin successfully decreased the proliferation of non-stem HNSCC cells. Computational drug–protein interaction analysis revealed mitochondrial complex III to be a likely target of metformin. Based on our results, we present the novel hypothesis that metformin targets complex III to reduce reactive oxygen species (ROS) levels, leading to the differential effects observed on non-stem cancer cells and CSCs.