Ulcerogenic drugs and upper gastrointestinal bleeding

Learning predictive models of drug side-effect relationships from distributed representations of literature-derived semantic predications

Objective

The aim of this work is to leverage relational information extracted from biomedical literature using a novel synthesis of unsupervised pretraining, representational composition, and supervised machine learning for drug safety monitoring.

Methods

Using ≈80 million concept-relationship-concept triples extracted from the literature using the SemRep Natural Language Processing system, distributed vector representations (embeddings) were generated for concepts as functions of their relationships utilizing two unsupervised representational approaches. Embeddings for drugs and side effects of interest from two widely used reference standards were then composed to generate embeddings of drug/side-effect pairs, which were used as input for supervised machine learning. This methodology was developed and evaluated using cross-validation strategies and compared to contemporary approaches. To qualitatively assess generalization, models trained on the Observational Medical Outcomes Partnership (OMOP) drug/side-effect reference set were evaluated against a list of ≈1100 drugs from an online database.

Results

The employed method improved performance over previous approaches. Cross-validation results advance the state of the art (AUC 0.96; F1 0.90 and AUC 0.95; F1 0.84 across the two sets), outperforming methods utilizing literature and/or spontaneous reporting system data. Examination of predictions for unseen drug/side-effect pairs indicates the ability of these methods to generalize, with over tenfold label support enrichment in the top 100 predictions versus the bottom 100 predictions.

Discussion and Conclusion

Our methods can assist the pharmacovigilance process using information from the biomedical literature. Unsupervised pretraining generates a rich relationship-based representational foundation for machine learning techniques to classify drugs in the context of a putative side effect, given known examples.

Administration of glucocorticoids markedly increases the numbers of granulocytes and extrathymic T cells in the bone marrow

Glucocorticoids, steroid hormones, are widely used as an anti-inflammatory drug. However, clinicians have sometimes encountered adverse drug reactions such as ulcers and tissue damage. In this study, we investigated how such adverse reactions of glucocorticoids are evoked, using an experimental mice model. When hydrocortisone (0.5 or 1.0 mg/day/mouse) was administered daily for 2 weeks, severe leukocytopenia was induced in all immune system organs. However, granulocytes (Gr-1(+)Mac-1(+)) were increased in number in the bone marrow and peripheral blood. This seemed to be due to an elevated level of myelopoiesis in the bone marrow. As well as increasing in number, granulocytes were functionally activated as estimated by the Ca2+ influx and superoxide production. The proportion of primordial T cells (CD3(int)IL-2Rbeta+) in the thymus and the number of primordial T cells in the bone marrow also increased. Mice administered hydrocortisone became susceptible to stress. Thus, these mice showed gastric ulcers when they were exposed to restraint stress for 12 h. These results suggest that activated granulocytes and primordial T cells might provide a mechanism involved in steroid ulcers and tissue damage, possibly through the superoxide production of granulocytes and the autoreactivity of primordial T cells.

Which peptic ulcers bleed? Results of a case-control study. DUSUK Study Group

BACKGROUND

The incidence of ulcer bleeding has not decrease despite effective medical treatment of peptic ulcer disease. Numerous studies have been performed on risk factors for ulcer bleeding, but only a few studies have related the risk of developing ulcer bleeding to the risk of developing uncomplicated ulcer disease.

METHODS

This study was performed as a multicenter interdisciplinary case-control study and was based on diagnosis by endoscopy, prospective data collection, and multivariate analysis. To every study patient with ulcer bleeding (case) one patient with uncomplicated peptic ulcer proven at endoscopy (control) was assigned with regard to sex, ulcer localization, and age (+/-5 years). The controls were randomly taken from three sources: inpatients, outpatients, and patients treated by a private physician.

RESULTS

Owing to strict inclusion and exclusion criteria, 209 matched pairs were available for evaluation out of 401 patients with peptic ulcer bleeding. In univariate analysis the duration of ulcer pain and the number of previous ulcer treatments proved to be protective factors for ulcer bleeding. The intake of non-steroidal anti-inflammatory drugs (NSAID) as a whole, acetylsalicylic acid (aspirin), and paracetamol gave an increased risk for ulcer bleeding. Multivariate analysis identified four risk factors for peptic ulcer bleeding: lung disease, NSAID intake, no pretreatment with H2 blockers, and acetylsalicylic acid intake.

CONCLUSIONS

The following patient groups represent a risk for peptic ulcer bleeding: multimorbid patients, receiving NSAID and/or aspirin, and patients with no previous ulcer history.