Identify diabetic retinopathy-related clinical concepts and their attributes using transformer-based natural language processing methods.

BACKGROUND: Diabetic retinopathy (DR) is a leading cause of blindness in American adults. If detected, DR can be treated to prevent further damage causing blindness. There is an increasing interest in developing artificial intelligence (AI) technologies to help detect DR using electronic health records. The lesion-related information documented in fundus image reports is a valuable resource that could help diagnoses of DR in clinical decision support systems. However, most studies for AI-based DR diagnoses are mainly based on medical images; there is limited studies to explore the lesion-related information captured in the free text image reports. METHODS: In this study, we examined two state-of-the-art transformer-based natural language processing (NLP) models, including BERT and RoBERTa, compared them with a recurrent neural network implemented using Long short-term memory (LSTM) to extract DR-related concepts from clinical narratives. We identified four different categories of DR-related clinical concepts including lesions, eye parts, laterality, and severity, developed annotation guidelines, annotated a DR-corpus of 536 image reports, and developed transformer-based NLP models for clinical concept extraction and relation extraction. We also examined the relation extraction under two settings including 'gold-standard' setting-where gold-standard concepts were used-and end-to-end setting. RESULTS: For concept extraction, the BERT model pretrained with the MIMIC III dataset achieve the best performance (0.9503 and 0.9645 for strict/lenient evaluation). For relation extraction, BERT model pretrained using general English text achieved the best strict/lenient F1-score of 0.9316. The end-to-end system, BERT_general_e2e, achieved the best strict/lenient F1-score of 0.8578 and 0.8881, respectively. Another end-to-end system based on the RoBERTa architecture, RoBERTa_general_e2e, also achieved the same performance as BERT_general_e2e in strict scores. CONCLUSIONS: This study demonstrated the efficiency of transformer-based NLP models for clinical concept extraction and relation extraction. Our results show that it's necessary to pretrain transformer models using clinical text to optimize the performance for clinical concept extraction. Whereas, for relation extraction, transformers pretrained using general English text perform better.

miR-335-5p Targets SDC1 to Regulate the Progression of Breast Cancer.

The objective of the study was to explore the role of SDC1 in breast cancer cells. Our study also investigated the regulatory relationship between SDC1 and the microRNA (miRNA) miR-335-5p as well as the impact of these two genes on the progression of breast cancer. Bioinformatic approaches were employed to analyze the differentially expressed messenger RNAs (mRNAs) and miRNAs (DE-mRNAs and DE-miRNAs) in breast cancer tissue. Then mRNA SC1 was obtained. Differentially downregulated mRNAs were intersected with target miRNAs predicted by databases, and miR-335-5p was determined as the study object. Quantitative reverse transcription polymerase chain reaction was applied to assess the expressions of SDC1 and miR-335-5p in each cell line. Next, Western blot assay was conducted to detect the protein level of SDC1 and dual-luciferase assay was performed to verify the binding relationship between miR-335-5p and SDC1. Finally, we conducted methyl thiazolyl tetrazolium (MTT), colony formation, and Transwell assays and flow cytometry to further investigate the impacts of SDC1 and miR-335-5p on the progression of breast cancer. SDC1 was significantly highly expressed while miR-335-5p was remarkably lowly expressed in human breast cancer. Silencing SDC1 in breast cancer blocked the proliferation, migration and invasion of the cells. In breast cancer, SDC1 was a target gene of miR-335-5p and silencing miR-335-5p notably increased SDC1 expression. Compared with the silence of miR-335-5p, simultaneous silences of miR-335-5p and SDC1 significantly reduced the proliferative, migratory and invasive abilities of breast cancer cells. The result revealed the interaction between miR-335-5p and SDC1 in the progression of breast cancer, which may contribute to the treatments for this cancer.

Measurement of Semantic Textual Similarity in Clinical Texts: Comparison of Transformer-Based Models.

BACKGROUND: Semantic textual similarity (STS) is one of the fundamental tasks in natural language processing (NLP). Many shared tasks and corpora for STS have been organized and curated in the general English domain; however, such resources are limited in the biomedical domain. In 2019, the National NLP Clinical Challenges (n2c2) challenge developed a comprehensive clinical STS dataset and organized a community effort to solicit state-of-the-art solutions for clinical STS. OBJECTIVE: This study presents our transformer-based clinical STS models developed during this challenge as well as new models we explored after the challenge. This project is part of the 2019 n2c2/Open Health NLP shared task on clinical STS. METHODS: In this study, we explored 3 transformer-based models for clinical STS: Bidirectional Encoder Representations from Transformers (BERT), XLNet, and Robustly optimized BERT approach (RoBERTa). We examined transformer models pretrained using both general English text and clinical text. We also explored using a general English STS dataset as a supplementary corpus in addition to the clinical training set developed in this challenge. Furthermore, we investigated various ensemble methods to combine different transformer models. RESULTS: Our best submission based on the XLNet model achieved the third-best performance (Pearson correlation of 0.8864) in this challenge. After the challenge, we further explored other transformer models and improved the performance to 0.9065 using a RoBERTa model, which outperformed the best-performing system developed in this challenge (Pearson correlation of 0.9010). CONCLUSIONS: This study demonstrated the efficiency of utilizing transformer-based models to measure semantic similarity for clinical text. Our models can be applied to clinical applications such as clinical text deduplication and summarization.

Interfering cellular lactate homeostasis overcomes Taxol resistance of breast cancer cells through the microRNA-124-mediated lactate transporter (MCT1) inhibition.

BACKGROUND: Breast cancer, the most common invasive cancer of women, is a malignant neoplasm and the second main cause of cancer death. Resistance to paclitaxel (Taxol), one of the frequently used chemotherapy agents for breast cancer, presents a major clinical challenge. Recent studies revealed that metabolic alterations of cancer cells play important roles in chemo-resistance. MATERIALS AND METHODS: In this study, Human breast cancer cells, BT474, SKBR3 and MCF7 were used to study the causal relationship between the lactate exporter, MCT1 (SLC16A1)-modulated glucose metabolism and Taxol resistance of breast cancer cells. Taxol resistant breast cancer cells were established. The intracellular lactate and extracellular lactate levels as well glucose uptake and oxygen consumption were measured. MicroRNA-124 expressions were detected by qRT-PCR from both breast cancer patient samples and breast cancer cells. Target of miR-124 was predicted and verified by Western blot and luciferase assay. An xenograft mice model was established and evaluated for the in vivo tumor therapeutic effects of MCT1 inhibitor plus microRNA-124 treatments. RESULTS: Low toxic Taxol treatments promoted cellular glucose metabolism and intracellular lactate accumulation with upregulated lactate dehydrogenase-A (LDHA) and MCT1 expressions. By establishing Taxol resistant breast cancer cell line, we found Taxol resistant cells exhibit upregulated LDHA and MCT1 expressions. Furthermore, glucose consumption, lactate production and intracellular ATP were elevated in Taxol resistant MCF7 cells compared with their parental cells. The miR-124, a tumor suppressive miRNA, was significantly downregulated in Taxol resistant cells. Luciferase assay and q-RT-PCR showed MCT1 is a direct target of miR-124 in both breast cancer cell lines and patient specimens. Moreover, co-treatment of breast cancer cells with either MCT1 inhibitor or miR-124 plus Taxol led to synergistically cytotoxic effects. Importantly, based on in vitro and in vivo results, inhibition of MCT1 significantly sensitized Taxol resistant cells. Finally, rescue experiments showed restoration of MCT1 in miR-124 overexpressing cells promoted Taxol resistance. CONCLUSIONS: This study reveals a possible role of miRNA-214-mediated Taxol resistance, contributing to identify novel therapeutic targets against chemoresistant breast cancers.

Down-regulated MAC30 expression inhibits breast cancer cell invasion and EMT by suppressing Wnt/ß-catenin and PI3K/Akt signaling pathways.

Breast cancer (BC) is a leading cause of cancer mortality in women worldwide. MAC30/Transmembrane protein 97 (TMEM97) is aberrantly up-regulated in many human carcinoma cells. However, the function of MAC30 in invasion and EMT of BC cells is uncertain. qRT-PCR was used to determine the level of MAC30 in BC tissues and cell lines. si-MAC30 was transfected into BC cells, and the effects of MAC30 silencing on the invasion and EMT were explored by qRT-PCR as well as transwell and western blot assays. Also, we determined the effects of MAC30 silencing on Wnt/ß-catenin and PI3K/Akt signaling pathways by western blot. We found that MAC30 is significantly up-regulated in BC tissues and cell lines. Down-regulation of MAC30 expression efficiently inhibited the invasion of BC cells. Furthermore, the EMT of BC cells was also inhibited by down-regulation of MAC30. Finally, we found that MAC30 knockdown inhibited Akt phosphorylation, ß-catenin, survivin, and cyclin D1 expressions. To our knowledge, this is the first report investigating the effect of MAC30 on invasion and EMT in BC cells by suppressing Wnt/ß-catenin and PI3K/Akt signaling pathways. MAC30 may be a potential therapeutic target for BC.