Juxin Liu

Bayesian analysis of a matched case-control study with expert prior information on both the misclassification of exposure and the exposure-disease association.

We propose a Bayesian adjustment for the misclassification of a binary exposure variable in a matched case-control study. The method admits a priori knowledge about both the misclassification parameters and the exposure-disease association. The standard Dirichlet prior distribution for a multinomial model is extended to allow separation of prior assertions about the exposure-disease association from assertions about other parameters. The method is applied to a study of occupational risk factors for new-onset adult asthma.

Development of a Laboratory Bioassay and Effect of Soil Properties on Sulfentrazone Phytotoxicity in Soil

Abstract Sulfentrazone is a phenyl triazolinone herbicide used for control of certain broadleaf and grass weed species. Sulfentrazone persists in soil and has residual activity beyond the season of application. A laboratory bioassay was developed for the detection of sulfentrazone in soil using root and shoot response of several crops. Shoot length inhibition of sugar beet was found to be the most sensitive and reproducible parameter for measurement of soil-incorporated sulfentrazone. The sugar beet bioassay was then used to examine the effect of soil properties on sulfentrazone phytotoxicity using 10 different Canadian prairie soils. Concentrations corresponding to 50% inhibition (I50 values) were obtained from the dose–response curves constructed for the soils. Sulfentrazone phytotoxicity was strongly correlated to the percentage organic carbon (P  =  0.01) and also to percentage clay content (P  =  0.05), whereas correlation with soil pH was nonsignificant (P  =  0.21). Because sulfentrazone phytotoxicity was found to be soil dependent, the efficacy of sulfentrazone for weed control and sulfentrazone potential carryover injury will vary with soil type in the Canadian prairies. Nomenclature: Sulfentrazone, N-[2,4-dichloro-5-[4-(difluoromethyl)-4,5-dihydro-3-methyl-5-oxo-1H-1,2,4-triazol-1-yl]phenyl]methanesulfonamide; sugar beet, Beta vulgaris L. ‘Beta 1385’.

Evaluation of the accuracy of self-reported smoking in pregnancy when the biomarker level in an active smoker is uncertain

INTRODUCTION Our main objective was to estimate smoking prevalence as well as sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of self-reported smoking among pregnant women in Edmonton, Canada, at 15-16 weeks of gestation. METHODS We used serum samples to assemble a cohort of pregnant women who underwent an optional second-trimester screening for chromosomal and developmental anomalies. We determined cotinine concentrations for 92 self-reported smokers (11% of the cohort) and for 285 self-reported nonsmoking mothers, using adapted urinary cotinine assay. Self-reports were collected at the time of delivery. In a validation study, serum cotinine was determined for known smokers and nonsmokers and used, within a Bayesian statistical framework, to define the distribution of cutoffs that differentiate true smokers from nonsmokers. This distribution of cutoffs was used to construct multiple two-by-two tables to obtain the distribution of sensitivity, specificity, PPV, NPV, and prevalence. RESULTS Sensitivity was poor (M = 47.4%, SD = 17.3%), but specificity was nearly perfect (M = 94.9%, SD = 1.1%). PPV (M = 66.6%, SD = 11.7%) was smaller than NPV (M = 84.7%, SD = 14.3%). In our sample, the prevalence of true smoking at 15-16 weeks of gestation was described by a skewed distribution with a mean of 21.6% (SD = 13.8%) and a median of 16.6%. DISCUSSION The strength of the present study includes blinding of subjects to the intention to test their sera for a biomarker of smoking. A limitation was the use of a nonrandom sample restricted to pregnancies that resulted in live births. We discuss data collection methods that would elicit more accurate smoking histories from pregnant women.

Aspergillus nidulans galactofuranose biosynthesis affects antifungal drug sensitivity

The cell wall is essential for fungal survival in natural environments. Many fungal wall carbohydrates are absent from humans, so they are a promising source of antifungal drug targets. Galactofuranose (Galf) is a sugar that decorates certain carbohydrates and lipids. It comprises about 5% of the Aspergillus fumigatus cell wall, and may play a role in systemic aspergillosis. We are studying Aspergillus wall formation in the tractable model system, A. nidulans. Previously we showed single-gene deletions of three sequential A. nidulans Galf biosynthesis proteins each caused similar hyphal morphogenesis defects and 500-fold reduced colony growth and sporulation. Here, we generated ugeA, ugmA and ugtA strains controlled by the alcA(p) or niiA(p) regulatable promoters. For repression and expression, alcA(p)-regulated strains were grown on complete medium with glucose or threonine, whereas niiA(p)-regulated strains were grown on minimal medium with ammonium or nitrate. Expression was assessed by qPCR and colony phenotype. The alcA(p) and niiA(p) strains produced similar effects: colonies resembling wild type for gene expression, and resembling deletion strains for gene repression. Galf immunolocalization using the L10 monoclonal antibody showed that ugmA deletion and repression phenotypes correlated with loss of hyphal wall Galf. None of the gene manipulations affected itraconazole sensitivity, as expected. Deletion of any of ugmA, ugeA, ugtA, their repression by alcA(p) or niiA(p), OR, ugmA overexpression by alcA(p), increased sensitivity to Caspofungin. Strains with alcA(p)-mediated overexpression of ugeA and ugtA had lower caspofungin sensitivity. Galf appears to play an important role in A. nidulans growth and vigor.

Discordance in Hormone Receptor Status Among Primary, Metastatic, and Second Primary Breast Cancers: Biological Difference or Misclassification?

INTRODUCTION Discordance in hormone receptor status has been observed between two breast tumors of the same patients; however, the degree of heterogeneity is debatable with regard to whether it reflects true biological difference or the limited accuracy of receptor assays. METHODS A Bayesian misclassification correction method was applied to data on hormone receptor status of two primary breast cancers from the Surveillance, Epidemiology, and End Results database between 1990 and 2010 and to data on primary breast cancer and paired recurrent/metastatic disease assembled from a meta-analysis of the literature published between 1979 and 2014. RESULTS The sensitivity and specificity of the estrogen receptor (ER) assay were estimated to be 0.971 and 0.920, respectively. After correcting for misclassification, the discordance in ER between two primary breast cancers was estimated to be 1.2% for synchronous ipsilateral pairs, 5.0% for synchronous contralateral pairs, 14.6% for metachronous ipsilateral pairs, and 25.0% for metachronous contralateral pairs. Technical misclassification accounted for 53%-83% of the ER discordance between synchronous primary cancers and 11%-25% of the ER discordance between metachronous cancers. The corrected discordance in ER between primary tumors and recurrent or metastatic lesions was 12.4%, and there were more positive-to-negative changes (10.1%) than negative-to-positive changes (2.3%). Similar patterns were observed for progesterone receptor (PR), although the overall discordance in PR was higher. CONCLUSION A considerable proportion of discordance in hormone receptor status can be attributed to misclassification in receptor assessment, although the accuracy of receptor assays was excellent. Biopsy of recurrent tumors for receptor retesting should be conducted after considering feasibility, cost, and previous ER/PR status.

Towards closed loop modeling: Evaluating the prospects for creating recurrently regrounded aggregate simulation models using particle filtering

Public health agencies traditionally rely heavily on epidemiological reporting for notifiable disease control, but increasingly apply simulation models for forecasting and to understand intervention tradeoffs. Unfortunately, such models traditionally lack capacity to easily incorporate information from epidemiological data feeds. Here, we introduce particle filtering and demonstrate how this approach can be used to readily incorporate recurrently available new data so as to robustly tolerate - and correct for - both model limitations and noisy data, and to aid in parameter estimation, while imposing far less onerous assumptions regarding the mathematical framework and epidemiological and measurement processes than other proposed solutions. By comparing against synthetic ground truth produced by an agent-based model, we demonstrate the benefits conferred by particle filtering parameters and state variables even in the context of an aggregate, incomplete and systematically biased compartmental model, and note important avenues for future work to make such approaches more widely accessible.

Effects of Soil Factors on Phytotoxicity and Dissipation of Sulfentrazone in Canadian Prairie Soils

Studies were conducted to examine the effects of soil properties on sulfentrazone phytotoxicity and dissipation under laboratory conditions. The pH values of five soils from Saskatchewan were altered through acidification with hydrochloric acid (HCl) and alkalization with calcium carbonate (CaCO3). The phytotoxicity of sulfentrazone to sugar beet (Beta vulgaris L. Beta 1385), determined using a shoot length bioassay, was reduced when soil pH was lowered and was greater when soil pH increased. Concentrations corresponding to 50% inhibition (I50 values) obtained from the dose–response curves were correlated with soil pH, demonstrating the relationship between soil pH and sulfentrazone phytotoxicity. Dissipation of sulfentrazone was examined in soils incubated at 25 °C and moisture content of 85% field capacity. Sulfentrazone dissipation followed a two-compartment model, and sulfentrazone half-lives estimated from the dissipation curves ranged from 21 to 111 days. Half-lives were correlated with soil pH (R = –0.857, p = 0.014) and soil organic carbon content (R = 0.790, p = 0.034) but not with clay content (R = 0.287, p = 0.533). Soil characteristics, particularly soil pH and organic carbon content, affect the bioactivity of sulfentrazone and influence both sulfentrazone efficacy in weed control and its potential for carry-over injury to subsequent crops.

Particle filtering in a seirv simulation model of H1N1 influenza

Numerous studies have been conducted using simulation models to predict the epidemiological spread of H1N1 and understand intervention trade-offs. However, existing models are generally not very accurate in H1N1 model predictions. In this report, we examine the impact of using particle filtering in a compartmental SEIRV (susceptible, exposed, infected, recovered and vaccinated) model which considers the impact of vaccination on the outbreak in the province of Manitoba. For the purpose of evaluating the performance of the particle filtering method, this work further compares the ability of particle filtering and traditional calibration to anticipate the evolution of the outbreak. Preliminary simulated results indicate that the particle filtering approach outperforms the calibration method in terms of the discrepancy between empirical data and model data.

Latent variable mixture models to test for differential item functioning: a population-based analysis

BackgroundComparisons of population health status using self-report measures such as the SF-36 rest on the assumption that the measured items have a common interpretation across sub-groups. However, self-report measures may be sensitive to differential item functioning (DIF), which occurs when sub-groups with the same underlying health status have a different probability of item response. This study tested for DIF on the SF-36 physical functioning (PF) and mental health (MH) sub-scales in population-based data using latent variable mixture models (LVMMs).MethodsData were from the Canadian Multicentre Osteoporosis Study (CaMos), a prospective national cohort study. LVMMs were applied to the ten PF and five MH SF-36 items. A standard two-parameter graded response model with one latent class was compared to multi-class LVMMs. Multivariable logistic regression models with pseudo-class random draws characterized the latent classes on demographic and health variables.ResultsThe CaMos cohort consisted of 9423 respondents. A three-class LVMM fit the PF sub-scale, with class proportions of 0.59, 0.24, and 0.17. For the MH sub-scale, a two-class model fit the data, with class proportions of 0.69 and 0.31. For PF items, the probabilities of reporting greater limitations were consistently higher in classes 2 and 3 than class 1. For MH items, respondents in class 2 reported more health problems than in class 1. Differences in item thresholds and factor loadings between one-class and multi-class models were observed for both sub-scales. Demographic and health variables were associated with class membership.ConclusionsThis study revealed DIF in population-based SF-36 data; the results suggest that PF and MH sub-scale scores may not be comparable across sub-groups defined by demographic and health status variables, although effects were frequently small to moderate in size. Evaluation of DIF should be a routine step when analysing population-based self-report data to ensure valid comparisons amongst sub-groups.

Bayesian adjustment for the misclassification in both dependent and independent variables with application to a breast cancer study.

In this paper, we propose a Bayesian method to address misclassification errors in both independent and dependent variables. Our work is motivated by a study of women who have experienced new breast cancers on two separate occasions. We call both cancers primary, because the second is usually not considered as the result of a metastasis spreading from the first. Hormone receptors (HRs) are important in breast cancer biology, and it is well recognized that the measurement of HR status is subject to errors. This discordance in HR status for two primary breast cancers is of concern and might be an important reason for treatment failure. To sort out the information on true concordance rate from the observed concordance rate, we consider a logistic regression model for the association between the HR status of the two cancers and introduce the misclassification parameters (i.e., sensitivity and specificity) accounting for the misclassification in HR status. The prior distribution for sensitivity and specificity is based on how HR status is actually assessed in laboratory procedures. To account for the nonlinear effect of one error-free covariate, we introduce the B-spline terms in the logistic regression model. Our findings indicate that the true concordance rate of HR status between two primary cancers is greater than the observed value. Copyright