Functional evaluation of out-of-the-box text-mining tools for data-mining tasks

Objective The trade-off between the speed and simplicity of dictionary-based term recognition and the richer linguistic information provided by more advanced natural language processing (NLP) is an area of active discussion in clinical informatics. In this paper, we quantify this trade-off among text processing systems that make different trade-offs between speed and linguistic understanding. We tested both types of systems in three clinical research tasks: phase IV safety profiling of a drug, learning adverse drug–drug interactions, and learning used-to-treat relationships between drugs and indications.Materials We first benchmarked the accuracy of the NCBO Annotator and REVEAL in a manually annotated, publically available dataset from the 2008 i2b2 Obesity Challenge. We then applied the NCBO Annotator and REVEAL to 9 million clinical notes from the Stanford Translational Research Integrated Database Environment (STRIDE) and used the resulting data for three research tasks.Results There is no significant difference between using the NCBO Annotator and REVEAL in the results of the three research tasks when using large datasets. In one subtask, REVEAL achieved higher sensitivity with smaller datasets.Conclusions For a variety of tasks, employing simple term recognition methods instead of advanced NLP methods results in little or no impact on accuracy when using large datasets. Simpler dictionary-based methods have the advantage of scaling well to very large datasets. Promoting the use of simple, dictionary-based methods for population level analyses can advance adoption of NLP in practice.     

Inhibition of MT-450 rat mammary tumour growth by antibodies recognising subtypes of blood group antigen B.

Using subtractive immunization to identify cell surface epitopes expressed in a metastasis-specific fashion on cells of the rat MT-W9 mammary carcinoma model, we generated a monoclonal antibody called M-N#1. This antibody binds specifically to metastasizing cells of the MT-W9 series and also to certain other metastasizing rat mammary carcinoma cell lines. We demonstrate that the M-N#1 antibody recognizes a fucosylated N-glycosyl sugar modification, and furthermore show that the epitope specificity of the M-N#1 antibody is for blood group antigen B subtypes 2, 3 and 4 with slight cross-reactivity with blood group antigen A subtype 2. The expression of these carbohydrate epitopes on MT-450 cells is functionally important, because the M-N#1 antibody efficiently inhibits MT-450 tumour growth in spontaneous metastasis assays. These results suggest that expression of the subtypes of blood group antigen B recognized by the M-N#1 antibody does not directly participate in the metastatic cascade but rather confers a growth or survival advantage on the tumour cells.     

Characterization of a monoclonal antibody against a mucin-type glycoprotein in human sweat

We report the preparation and characterization of an IgG2 monoclonal antibody (MAb), HSMA, prepared against a human pooled sweat extract (HPSE). The major component of HPSE was a mucin-type molecule, as revealed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) stained with periodic acid-Schiff (PAS) reagent. By immunoblotting, HSMA revealed a smear in the high molecular weight range, typical of mucins. In enzyme-linked immunosorbent assay (ELISA), HSMA failed to react with HPSE fractions isolated after anionic exchange gel chromatography. Similarly, radio-immunobinding assays demonstrated no reactivity between HSMA and A, B, H, and Lewis blood group-related structures. The immunohistological labeling on normal skin showed that HSMA reacted with the cells of eccrine sweat glands, and to a lesser extent, with sebaceous glands and epidermal cells. Periodate treatment in situ abolished these reactions, thus suggesting the carbohydrate structure of the HSMA-epitope. In indirect immunofluorescence (IF) studies, HSMA also reacted with other exocrine glands, e.g. mammary glands, sublingual glands, mixed sero-mucous glands of the trachea, and in the pancreas. Sparse positive cells were also observed in the testis, kidney, thyroid and digestive tract.     

Toward personalizing treatment for depression: predicting diagnosis and severity

Objective

Depression is a prevalent disorder difficult to diagnose and treat. In particular, depressed patients exhibit largely unpredictable responses to treatment. Toward the goal of personalizing treatment for depression, we develop and evaluate computational models that use electronic health record (EHR) data for predicting the diagnosis and severity of depression, and response to treatment.

Materials and methods

We develop regression-based models for predicting depression, its severity, and response to treatment from EHR data, using structured diagnosis and medication codes as well as free-text clinical reports. We used two datasets: 35 000 patients (5000 depressed) from the Palo Alto Medical Foundation and 5651 patients treated for depression from the Group Health Research Institute.

Results

Our models are able to predict a future diagnosis of depression up to 12 months in advance (area under the receiver operating characteristic curve (AUC) 0.70–0.80). We can differentiate patients with severe baseline depression from those with minimal or mild baseline depression (AUC 0.72). Baseline depression severity was the strongest predictor of treatment response for medication and psychotherapy.

Conclusions

It is possible to use EHR data to predict a diagnosis of depression up to 12 months in advance and to differentiate between extreme baseline levels of depression. The models use commonly available data on diagnosis, medication, and clinical progress notes, making them easily portable. The ability to automatically determine severity can facilitate assembly of large patient cohorts with similar severity from multiple sites, which may enable elucidation of the moderators of treatment response in the future.     

Noroviruses bind to human ABO,Lewis, and secretor histo-blood group antigens: identification of 4 distinct strain-specific patterns

We characterized the binding of 8 Noroviruses (NORs) to histo-blood group antigens (HBGAs) in human saliva using recombinant NOR (rNOR) capsid proteins. Among the 8 rNORs tested, 6 formed viruslike particles (VLPs) when the capsid proteins were expressed in insect cells, all of which revealed variable binding activities with saliva; the remaining 2 rNORs did not form VLPs, and the proteins did not bind, or bound weakly, to saliva. Four distinct binding patterns were associated with different histo-blood types, defined by Lewis, secretor, and ABO types. Three patterns (VA387, NV, and MOH) recognized secretors, and 1 pattern (VA207) recognized Lewis-positive nonsecretors. The 3 secretor-recognizing patterns were defined as A/B (MOH), A/O (NV), and A/B/O (VA387) binders. Oligosaccharides containing the Lewis and ABH antigenic epitopes were involved in binding. Our findings suggest that different strains of NORs may recognize different human HBGAs on intestinal epithelial cells as receptors for infection.     

Enabling enrichment analysis with the Human Disease Ontology

Advanced statistical methods used to analyze high-throughput data such as gene-expression assays result in long lists of "significant genes." One way to gain insight into the significance of altered expression levels is to determine whether Gene Ontology (GO) terms associated with a particular biological process, molecular function, or cellular component are over- or under-represented in the set of genes deemed significant. This process, referred to as enrichment analysis, profiles a gene set, and is widely used to make sense of the results of high-throughput experiments. Our goal is to develop and apply general enrichment analysis methods to profile other sets of interest, such as patient cohorts from the electronic medical record, using a variety of ontologies including SNOMED CT, MedDRA, RxNorm, and others. Although it is possible to perform enrichment analysis using ontologies other than the GO, a key pre-requisite is the availability of a background set of annotations to enable the enrichment calculation. In the case of the GO, this background set is provided by the Gene Ontology Annotations. In the current work, we describe: (i) a general method that uses hand-curated GO annotations as a starting point for creating background datasets for enrichment analysis using other ontologies; and (ii) a gene-disease background annotation set - that enables disease-based enrichment - to demonstrate feasibility of our method.     

Norwalk virus infection associates with secretor status genotyped from sera

ABO histo-blood group type and secretor status are two genetically determined factors that contribute to resistance and susceptibility to Norwalk virus (NV). Archived serum samples but not saliva samples are available from NV and many other norovirus challenge studies and outbreaks. A person's ABO phenotype is easily determined from their archived sera, but the individual's secretor phenotype cannot easily be ascertained without saliva. We now report that a person's secretor genotype can also be determined from the archived serum samples. Of the 51 volunteers who participated in a NV challenge study, all eight non-secretors were resistant to NV infection, all of the 42 NV-infected volunteers were secretor positive, and a single uninfected secretor was histo-blood group type B. In agreement with a previous report, secretor status was most predictive of risk of NV infection. The methods described in this report should rapidly improve our knowledge of the associations between carbohydrate antigen expression and susceptibility to different strains of the non-cultivatable noroviruses by enabling retrospective studies from previously collected volunteer challenge and outbreak sera.     

Expression of Lea in gastric cancer cell lines depends on FUT3 expression regulated by promoter methylation

Aberrant expression of Lewis antigens has been demonstrated in gastric lesions, namely gastritis, intestinal metaplasia (IM) and gastric carcinoma (GC), and can be partly due to overexpression of the Lewis (FUT3) enzyme. Our aim was to evaluate the role of promoter methylation in FUT3 and Le(a) expression in gastric carcinoma cell lines. MKN45 cell line showed low amounts of Le(a), in the absence of FUT3; GP220 expressed high levels of Le(a) and FUT3. After 5aza-2'deoxycytidine MKN45 showed increased levels of FUT3 and Le(a), by immunohistochemistry and Real-Time PCR, whereas GP220 showed an increase in FUT3 without increase of Le(a). Enzyme activity assays confirmed an increase in alpha-1,4 fucosyltransferase activity in both cell lines by 5aza-2'deoxycytidine. Luciferase reporter gene assays, using methylated and unmethylated deletion constructs of FUT3 promoter, showed that FUT3 expression is regulated by methylation. Summing up, we showed that FUT3 overexpression in gastric cells depends upon promoter hypomethylation and that FUT3 is responsible for overexpression of Le(a) in gastric cells, in vitro. FUT3, Lea, Methylation.