Extracting diagnoses and investigation results from unstructured text in electronic health records by semi-supervised machine learning

Background

Electronic health records are invaluable for medical research, but much of the information is recorded as unstructured free text which is time-consuming to review manually.

Aim

To develop an algorithm to identify relevant free texts automatically based on labelled examples.

Methods

We developed a novel machine learning algorithm, the ‘Semi-supervised Set Covering Machine’ (S3CM), and tested its ability to detect the presence of coronary angiogram results and ovarian cancer diagnoses in free text in the General Practice Research Database. For training the algorithm, we used texts classified as positive and negative according to their associated Read diagnostic codes, rather than by manual annotation. We evaluated the precision (positive predictive value) and recall (sensitivity) of S3CM in classifying unlabelled texts against the gold standard of manual review. We compared the performance of S3CM with the Transductive Vector Support Machine (TVSM), the original fully-supervised Set Covering Machine (SCM) and our ‘Freetext Matching Algorithm’ natural language processor.

Results

Only 60% of texts with Read codes for angiogram actually contained angiogram results. However, the S3CM algorithm achieved 87% recall with 64% precision on detecting coronary angiogram results, outperforming the fully-supervised SCM (recall 78%, precision 60%) and TSVM (recall 2%, precision 3%). For ovarian cancer diagnoses, S3CM had higher recall than the other algorithms tested (86%). The Freetext Matching Algorithm had better precision than S3CM (85% versus 74%) but lower recall (62%).

Conclusions

Our novel S3CM machine learning algorithm effectively detected free texts in primary care records associated with angiogram results and ovarian cancer diagnoses, after training on pre-classified test sets. It should be easy to adapt to other disease areas as it does not rely on linguistic rules, but needs further testing in other electronic health record datasets.     

Predicting Drug-Target Interaction Networks Based on Functional Groups and Biological Features

Background

Study of drug-target interaction networks is an important topic for drug development. It is both time-consuming and costly to determine compound-protein interactions or potential drug-target interactions by experiments alone. As a complement, the in silico prediction methods can provide us with very useful information in a timely manner.

Methods/Principal Findings

To realize this, drug compounds are encoded with functional groups and proteins encoded by biological features including biochemical and physicochemical properties. The optimal feature selection procedures are adopted by means of the mRMR (Maximum Relevance Minimum Redundancy) method. Instead of classifying the proteins as a whole family, target proteins are divided into four groups: enzymes, ion channels, G-protein- coupled receptors and nuclear receptors. Thus, four independent predictors are established using the Nearest Neighbor algorithm as their operation engine, with each to predict the interactions between drugs and one of the four protein groups. As a result, the overall success rates by the jackknife cross-validation tests achieved with the four predictors are 85.48%, 80.78%, 78.49%, and 85.66%, respectively.

Conclusion/Significance

Our results indicate that the network prediction system thus established is quite promising and encouraging.     

A Novel Compound NSC745885 Exerts an Anti-Tumor Effect on Tongue Cancer SAS Cells In Vitro and In Vivo

Objective

Oral squamous cell carcinoma (OSCC) is a prevalent cancer, especially in developing countries. Anthracyclines and their anthraquinone derivatives, such as doxorubicin, exhibit a cell growth inhibitory effect and have been used as anti-cancer drugs for many years. However, the cardiotoxicity of anthracycline antibiotics is a major concern in their clinical application. NSC745885 is a novel compound synthesized from 1,2-diaminoanthraquinone, which subsequently reacts with thionyl chloride and triethylamine. The present study aimed to investigate the anti-oral cancer potential and the safety of NSC745885.

Methods

We investigated the anti-cancer potential of NSC745885 in oral squamous carcinoma cell lines and in an in vivo oral cancer xenograft mouse model. The expression of apoptotic related genes were evaluated by real-time RT-PCR and western bloting, and the in vivo assessment of apoptotic marker were measured by immunohistochemical staining. The anti-tumor efficiency and safety between doxorubicin and NSC745885 were also compared.

Results

Our results demonstrated that NSC745885 exhibits anti-oral cancer activity through the induction of apoptosis in cancer cells and in tumor-bearing mice, and this treatment did not induce marked toxicity in experimental mice. This compound also exhibits a comparable anti-tumor efficiency and a higher safety in experimental mice when compared to doxorubicin.

Conclusions

The data of this study provide evidence for NSC745885 as a potential novel therapeutic drug for the treatment of human OSCC.     

Efficacy and safety of nitazoxanide, albendazole, and nitazoxanide-albendazole against Trichuris trichiura infection: a randomized controlled trial

Background

The currently used anthelmintic drugs, in single oral application, have low efficacy against Trichuris trichiura infection, and hence novel anthelmintic drugs are needed. Nitazoxanide has been suggested as potential drug candidate.

Methodology

The efficacy and safety of a single oral dose of nitazoxanide (1,000 mg), or albendazole (400 mg), and a nitazoxanide-albendazole combination (1,000 mg–400 mg), with each drug administered separately on two consecutive days, were assessed in a double-blind, randomized, placebo-controlled trial in two schools on Pemba, Tanzania. Cure and egg reduction rates were calculated by per-protocol analysis and by available case analysis. Adverse events were assessed and graded before treatment and four times after treatment.

Principal Findings

Complete data for the per-protocol analysis were available from 533 T. trichiura-positive children. Cure rates against T. trichiura were low regardless of the treatment (nitazoxanide-albendazole, 16.0%; albendazole, 14.5%; and nitazoxanide, 6.6%). Egg reduction rates were 54.9% for the nitazoxanide-albendazole combination, 45.6% for single albendazole, and 13.4% for single nitazoxanide. Similar cure and egg reduction rates were calculated using the available case analysis. Children receiving nitazoxanide had significantly more adverse events compared to placebo recipients. Most of the adverse events were mild and had resolved within 24 hours posttreatment.

Conclusions/Significance

Nitazoxanide shows no effect on T. trichiura infection. The low efficacy of albendazole against T. trichiura in the current setting characterized by high anthelmintic drug pressure is confirmed. There is a pressing need to develop new anthelmintics against trichuriasis.

Trial Registration

Controlled-Trials.com ISRCTN08336605     

Identification of common biological pathways and drug targets across multiple respiratory viruses based on human host gene expression analysis

Background

Pandemic and seasonal respiratory viruses are a major global health concern. Given the genetic diversity of respiratory viruses and the emergence of drug resistant strains, the targeted disruption of human host-virus interactions is a potential therapeutic strategy for treating multi-viral infections. The availability of large-scale genomic datasets focused on host-pathogen interactions can be used to discover novel drug targets as well as potential opportunities for drug repositioning.

Methods/Results

In this study, we performed a large-scale analysis of microarray datasets involving host response to infections by influenza A virus, respiratory syncytial virus, rhinovirus, SARS-coronavirus, metapneumonia virus, coxsackievirus and cytomegalovirus. Common genes and pathways were found through a rigorous, iterative analysis pipeline where relevant host mRNA expression datasets were identified, analyzed for quality and gene differential expression, then mapped to pathways for enrichment analysis. Possible repurposed drugs targets were found through database and literature searches. A total of 67 common biological pathways were identified among the seven different respiratory viruses analyzed, representing fifteen laboratories, nine different cell types, and seven different array platforms. A large overlap in the general immune response was observed among the top twenty of these 67 pathways, adding validation to our analysis strategy. Of the top five pathways, we found 53 differentially expressed genes affected by at least five of the seven viruses. We suggest five new therapeutic indications for existing small molecules or biological agents targeting proteins encoded by the genes F3, IL1B, TNF, CASP1 and MMP9. Pathway enrichment analysis also identified a potential novel host response, the Parkin-Ubiquitin Proteasomal System (Parkin-UPS) pathway, which is known to be involved in the progression of neurodegenerative Parkinson''s disease.

Conclusions

Our study suggests that multiple and diverse respiratory viruses invoke several common host response pathways. Further analysis of these pathways suggests potential opportunities for therapeutic intervention.     

Dopaminergic and serotonergic drug use: a nationwide register-based study of over 1,300,000 older people

Objective

To investigate the use of dopaminergic and serotonergic drugs in elderly people.

Methods

We analyzed data on age, sex and dispensed drugs for individuals aged ≥65 years registered in the Swedish Prescribed Drug Register from July to September 2008 (n = 1 347 564; 81% of the total population aged ≥65 years in Sweden). Main outcome measures were dopaminergic (enhancing and/or lowering) and serotonergic (enhancing and/or lowering) drugs and combinations of these.

Results

Dopaminergic and serotonergic drugs were used by 5.6% and 13.2% the participants, respectively. Female gender was related to use of both dopaminergic and, particularly, serotonergic drugs. Higher age was associated with use of dopamine lowering drugs and serotonergic drugs, whereas the association with use of dopamine enhancing drugs declined in the oldest old. The occurrence of combinations of dopaminergic and serotonergic drugs was generally low, with dopamine lowering + serotonin lowering drug the most common combination (1.6%). Female gender was associated with all of the combinations of dopaminergic and serotonergic drugs, whereas age showed a mixed pattern.

Conclusion

Approximately one out of ten older patients uses serotonergic drugs and one out of twenty dopaminergic drugs. The frequent use of dopaminergic and serotonergic drugs in the elderly patients is a potential problem due to the fact that aging is associated with a down-regulation of both these monoaminergic systems. Future studies are needed for evaluation of the impact of these drugs on different cognitive and emotional functions in old age.     

A phenome-guided drug repositioning through a latent variable model

Background

The phenome represents a distinct set of information in the human population. It has been explored particularly in its relationship with the genome to identify correlations for diseases. The phenome has been also explored for drug repositioning with efforts focusing on the search space for the most similar candidate drugs. For a comprehensive analysis of the phenome, we assumed that all phenotypes (indications and side effects) were inter-connected with a probabilistic distribution and this characteristic may offer an opportunity to identify new therapeutic indications for a given drug. Correspondingly, we employed Latent Dirichlet Allocation (LDA), which introduces latent variables (topics) to govern the phenome distribution.

Results

We developed our model on the phenome information in Side Effect Resource (SIDER). We first developed a LDA model optimized based on its recovery potential through perturbing the drug-phenotype matrix for each of the drug-indication pairs where each drug-indication relationship was switched to “unknown” one at the time and then recovered based on the remaining drug-phenotype pairs. Of the probabilistically significant pairs, 70% was successfully recovered. Next, we applied the model on the whole phenome to narrow down repositioning candidates and suggest alternative indications. We were able to retrieve approved indications of 6 drugs whose indications were not listed in SIDER. For 908 drugs that were present with their indication information, our model suggested alternative treatment options for further investigations. Several of the suggested new uses can be supported with information from the scientific literature.

Conclusions

The results demonstrated that the phenome can be further analyzed by a generative model, which can discover probabilistic associations between drugs and therapeutic uses. In this regard, LDA serves as an enrichment tool to explore new uses of existing drugs by narrowing down the search space.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2105-15-267) contains supplementary material, which is available to authorized users.     

Tamoxifen Enhances the Hsp90 Molecular Chaperone ATPase Activity

Background

Hsp90 is an essential molecular chaperone that is also a novel anti-cancer drug target. There is growing interest in developing new drugs that modulate Hsp90 activity.

Methodology/Principal Findings

Using a virtual screening approach, 4-hydroxytamoxifen, the active metabolite of the anti-estrogen drug tamoxifen, was identified as a putative Hsp90 ligand. Surprisingly, while all drugs targeting Hsp90 inhibit the chaperone ATPase activity, it was found experimentally that 4-hydroxytamoxifen and tamoxifen enhance rather than inhibit Hsp90 ATPase.

Conclusions/Significance

Hence, tamoxifen and its metabolite are the first members of a new pharmacological class of Hsp90 activators.     

Anticancer properties and enhancement of therapeutic potential of cisplatin by leaf extract of Zanthoxylum armatum DC.

Background

Clinical use of chemotherapeutic drug, cisplatin is limited by its toxicity and drug resistance. Therefore, efforts continue for the discovery of novel combination therapies with cisplatin, to increase efficacy and reduce its toxicity. Here, we screened 16 medicinal plant extracts from Northeast part of India and found that leaf extract of Zanthoxylumarmatum DC. (ZALE) induced cytotoxicity as well as an effect on the increasing of the efficiency of chemotherapeutic drugs (cisplatin, mitomycin C and camptothecin). This work shows detail molecular mechanism of anti-cancer activity of ZALE and its potential for combined treatment regimens to enhance the apoptotic response of chemotherapeutic drugs.

Results

ZALE induced cytotoxicity, nuclear blebbing and DNA fragmentation in HeLA cells suggesting apoptosis induction in human cervical cell line. However, the apoptosis induced was independent of caspase 3 activation and poly ADP ribose polymerase (PARP) cleavage. Further, ZALE activated Mitogen-activated protein kinases (MAPK) pathway as revealed by increased phosphorylation of extracellular-signal-regulated kinases (ERK), p38 and c-Jun N-terminal kinase (JNK). Inhibition of ERK activation but not p38 or JNK completely blocked the ZALE induced apoptosis suggesting an ERK dependent apoptosis. Moreover, ZALE generated DNA double strand breaks as suggested by the induction γH2AX foci formation. Interestingly, pretreatment of certain cancer cell lines with ZALE, sensitized the cancer cells to cisplatin and other chemotherapeutic drugs. Enhanced caspase activation was observed in the synergistic interaction among chemotherapeutic drugs and ZALE.

Conclusion

Purification and identification of the bio-active molecules from the ZALE or as a complementary treatment for a sequential treatment of ZALE with chemotherapeutic drugs might be a new challenger to open a new therapeutic window for the novel anti-cancer treatment.

Electronic supplementary material

The online version of this article (doi:10.1186/s40659-015-0037-4) contains supplementary material, which is available to authorized users.     

Trends in breast cancer mortality in Sweden before and after implementation of mammography screening

Background

Incidence-based mortality modelling comparing the risk of breast cancer death in screened and unscreened women in nine Swedish counties has suggested a 39% risk reduction in women 40 to 69 years old after introduction of mammography screening in the 1980s and 1990s.

Objective

We evaluated changes in breast cancer mortality in the same nine Swedish counties using a model approach based on official Swedish breast cancer mortality statistics, robust to effects of over-diagnosis and treatment changes. Using mortality data from the NordCan database from 1974 until 2003, we estimated the change in breast cancer mortality before and after introduction of mammography screening in at least the 13 years that followed screening start.

Results

Breast mortality decreased by 16% (95% CI: 9 to 22%) in women 40 to 69, and by 11% (95% CI: 2 to 20%) in women 40 to 79 years of age.

Discussion

Without individual data it is impossible to completely separate the effects of improved treatment and health service organisation from that of screening, which would bias our results in favour of screening. There will also be some contamination of post-screening mortality from breast cancer diagnosed prior to screening, beyond our attempts to adjust for delayed benefit. This would bias against screening. However, our estimates from publicly available data suggest considerably lower benefits than estimates based on comparison of screened versus non-screened women.     

Variation in Drug Sensitivity of Malignant Mesothelioma Cell Lines with Substantial Effects of Selenite and Bortezomib,Highlights Need for Individualized Therapy

Background

Malignant mesothelioma cells have an epithelioid or sarcomatoid morphology, both of which may be present in the same tumor. The sarcomatoid phenotype is associated with worse prognosis and heterogeneity of mesothelioma cells may contribute to therapy resistance, which is often seen in mesothelioma. This study aimed to investigate differences in sensitivity between mesothelioma cell lines to anti-cancer drugs. We studied two novel drugs, selenite and bortezomib and compared their effect to four conventional drugs. We also investigated the immunoreactivity of potential predictive markers for drug sensitivity; Pgp, MRP-1, ERCC1, RRM1, TS, xCT and proteasome 20S subunit.

Materials and methods

We treated six mesothelioma cell lines with selenite, bortezomib, carboplatin, pemetrexed, doxorubicin or gemcitabine as single agents and in combinations. Viability was measured after 24 and 48 hours. Immunocytochemistry was used to detect predictive markers.

Results

As a single agent, selenite was effective on four out of six cell lines, and in combination with bortezomib yielded the greatest response in the studied mesothelioma cell lines. Cells with an epithelioid phenotype were generally more sensitive to the different drugs than the sarcomatoid cells. Extensive S-phase arrest was seen in pemetrexed-sensitive cell lines. MRP-1 predicted sensitivity of cell lines to treatment with carboplatin and xCT predicted pemetrexed effect.

Conclusions

The observed heterogeneity in sensitivity of mesothelioma cell lines with different morphology highlights the need for more individualized therapy, requiring development of methods to predict drug sensitivity of individual tumors. Selenite and bortezomib showed a superior effect compared to conventional drugs, motivating clinical testing of these agents as future treatment regime components for patients with malignant mesothelioma.     

Stability of gene expression and epigenetic profiles highlights the utility of patient-derived paediatric acute lymphoblastic leukaemia xenografts for investigating molecular mechanisms of drug resistance

Background

Patient-derived tumour xenografts are an attractive model for preclinical testing of anti-cancer drugs. Insights into tumour biology and biomarkers predictive of responses to chemotherapeutic drugs can also be gained from investigating xenograft models. As a first step towards examining the equivalence of epigenetic profiles between xenografts and primary tumours in paediatric leukaemia, we performed genome-scale DNA methylation and gene expression profiling on a panel of 10 paediatric B-cell precursor acute lymphoblastic leukaemia (BCP-ALL) tumours that were stratified by prednisolone response.

Results

We found high correlations in DNA methylation and gene expression profiles between matching primary and xenograft tumour samples with Pearson’s correlation coefficients ranging between 0.85 and 0.98. In order to demonstrate the potential utility of epigenetic analyses in BCP-ALL xenografts, we identified DNA methylation biomarkers that correlated with prednisolone responsiveness of the original tumour samples. Differential methylation of CAPS2, ARHGAP21, ARX and HOXB6 were confirmed by locus specific analysis. We identified 20 genes showing an inverse relationship between DNA methylation and gene expression in association with prednisolone response. Pathway analysis of these genes implicated apoptosis, cell signalling and cell structure networks in prednisolone responsiveness.

Conclusions

The findings of this study confirm the stability of epigenetic and gene expression profiles of paediatric BCP-ALL propagated in mouse xenograft models. Further, our preliminary investigation of prednisolone sensitivity highlights the utility of mouse xenograft models for preclinical development of novel drug regimens with parallel investigation of underlying gene expression and epigenetic responses associated with novel drug responses.

Electronic supplementary material

The online version of this article (doi:10.1186/1471-2164-15-416) contains supplementary material, which is available to authorized users.     

Telomere Shortening Sensitizes Cancer Cells to Selected Cytotoxic Agents: In Vitro and In Vivo Studies and Putative Mechanisms

Background

Telomere/telomerase system has been recently recognized as an attractive target for anticancer therapy. Telomerase inhibition results in tumor regression and increased sensitivity to various cytotoxic drugs. However, it has not been fully established yet whether the mediator of these effects is telomerase inhibition per se or telomere shortening resulting from inhibition of telomerase activity. In addition, the characteristics and mechanisms of sensitization to cytotoxic drugs caused by telomerase inhibition has not been elucidated in a systematic manner.

Methodology/Principal Findings

In this study we characterized the relative importance of telomerase inhibition versus telomere shortening in cancer cells. Sensitization of cancer cells to cytotoxic drugs was achieved by telomere shortening in a length dependent manner and not by telomerase inhibition per se. In our system this sensitization was related to the mechanism of action of the cytotoxic drug. In addition, telomere shortening affected also other cancer cell functions such as migration. Telomere shortening induced DNA damage whose repair was impaired after administration of cisplatinum while doxorubicin or vincristine did not affect the DNA repair. These findings were verified also in in vivo mouse model. The putative explanation underlying the phenotype induced by telomere shortening may be related to changes in expression of various microRNAs triggered by telomere shortening.

Conclusions/Significance

To our best knowledge this is the first study characterizing the relative impact of telomerase inhibition and telomere shortening on several aspects of cancer cell phenotype, especially related to sensitivity to cytotoxic drugs and its putative mechanisms. The microRNA changes in cancer cells upon telomere shortening are novel information. These findings may facilitate the development of telomere based approaches in treatment of cancer.     

Factors influencing the emergence and spread of HIV drug resistance arising from rollout of antiretroviral pre-exposure prophylaxis (PrEP)

Background

The potential for emergence and spread of HIV drug resistance from rollout of antiretroviral (ARV) pre-exposure prophylaxis (PrEP) is an important public health concern. We investigated determinants of HIV drug resistance prevalence after PrEP implementation through mathematical modeling.

Methodology

A model incorporating heterogeneity in age, gender, sexual activity, HIV infection status, stage of disease, PrEP coverage/discontinuation, and HIV drug susceptibility, was designed to simulate the impact of PrEP on HIV prevention and drug resistance in a sub-Saharan epidemic.

Principal Findings

Analyses suggest that the prevalence of HIV drug resistance is influenced most by the extent and duration of inadvertent PrEP use in individuals already infected with HIV. Other key factors affecting drug resistance prevalence include the persistence time of transmitted resistance and the duration of inadvertent PrEP use in individuals who become infected on PrEP. From uncertainty analysis, the median overall prevalence of drug resistance at 10 years was predicted to be 9.2% (interquartile range 6.9%–12.2%). An optimistic scenario of 75% PrEP efficacy, 60% coverage of the susceptible population, and 5% inadvertent PrEP use predicts a rise in HIV drug resistance prevalence to only 2.5% after 10 years. By contrast, in a pessimistic scenario of 25% PrEP efficacy, 15% population coverage, and 25% inadvertent PrEP use, resistance prevalence increased to over 40%.

Conclusions

Inadvertent PrEP use in previously-infected individuals is the major determinant of HIV drug resistance prevalence arising from PrEP. Both the rate and duration of inadvertent PrEP use are key factors. PrEP rollout programs should include routine monitoring of HIV infection status to limit the spread of drug resistance.     

Quality of reporting of bioequivalence trials comparing generic to brand name drugs: a methodological systematic review

Background

Generic drugs are used by millions of patients for economic reasons, so their evaluation must be highly transparent.

Objective

To assess the quality of reporting of bioequivalence trials comparing generic to brand-name drugs.

Methodology/Principal Findings

PubMed was searched for reports of bioequivalence trials comparing generic to brand-name drugs between January 2005 and December 2008. Articles were included if the aim of the study was to assess the bioequivalency of generic and brand-name drugs. We excluded case studies, pharmaco-economic evaluations, and validation dosage assays of drugs. We evaluated whether important information about funding, methodology, location of trials, and participants were reported. We also assessed whether the criteria required by the Food and Drug Administration (FDA) and the European Medicine Agency (EMA) to conclude bioequivalence were reported and that the conclusions were in agreement with the results. We identified 134 potentially relevant articles but eliminated 55 because the brand-name or generic drug status of the reference drug was unknown. Thus, we evaluated 79 articles. The funding source and location of the trial were reported in 41% and 56% of articles, respectively. The type of statistical analysis was reported in 94% of articles, but the methods to generate the randomization sequence and to conceal allocation were reported in only 15% and 5%, respectively. In total, 65 articles of single-dose trials (89%) concluded bioequivalence. Of these, 20 (31%) did not report the 3 criteria within the limits required by the FDA and 11 (17%) did not report the 2 criteria within the limits required by the EMA.

Conclusions/Significance

Important information to judge the validity and relevance of results are frequently missing in published reports of trials assessing generic drugs. The quality of reporting of such trials is in need of improvement.