David J. Price

Conformational risk factors of brachycephalic obstructive airway syndrome (BOAS) in pugs, French bulldogs, and bulldogs.

Extremely brachycephalic, or short-muzzled, dog breeds such as pugs, French bulldogs, and bulldogs are prone to the conformation-related respiratory disorder—brachycephalic obstructive airway syndrome (BOAS). Affected dogs present with a wide range of clinical signs from snoring and exercise intolerance, to life-threatening events such as syncope. In this study, conformational risk factors for BOAS that could potentially aid in breeding away from BOAS were sought. Six hundred and four pugs, French bulldogs, and bulldogs were included in the study. Soft tape measurements of the head and body were used and the inter-observer reproducibility was evaluated. Breed-specific models were developed to assess the associations between the conformational factors and BOAS status based on functional grading. The models were further validated by means of a BOAS index, which is an objective measurement of respiratory function using whole-body barometric plethysmography. The final models have good predictive power for discriminating BOAS (-) and BOAS (+) phenotypes indicated by the area under the curve values of >80% on the receiver operating curves. When other factors were controlled, stenotic nostrils were associated with BOAS in all three breeds; pugs and bulldogs with higher body condition scores (BCS) had a higher risk of developing BOAS. Among the standardized conformational measurements (i.e. craniofacial ratio (CFR), eye width ratio (EWR), skull index (SI), neck girth ratio (NGR), and neck length ratio (NLR)), for pugs EWR and SI, for French bulldogs NGR and NLR, and for bulldogs SI and NGR showed significant associations with BOAS status. However, the NGR in bulldogs was the only significant predictor that also had satisfactory inter-observer reproducibility. A NGR higher than 0.71 in male bulldogs was predictive of BOAS with approximately 70% sensitivity and specificity. In conclusion, stenotic nostrils, BCS, and NGR were found to be valid, easily applicable predictors for BOAS (+).

Sample size calculations for randomised trials including both independent and paired data.

Randomised trials including a mixture of independent and paired data arise in many areas of health research, yet methods for determining the sample size for such trials are lacking. We derive design effects algebraically assuming clustering because of paired data will be taken into account in the analysis using generalised estimating equations with either an independence or exchangeable working correlation structure. Continuous and binary outcomes are considered, along with three different methods of randomisation: cluster randomisation, individual randomisation and randomisation to opposite treatment groups. The design effect is shown to depend on the intracluster correlation coefficient, proportion of observations belonging to a pair, working correlation structure, type of outcome and method of randomisation. The derived design effects are validated through simulation and example calculations are presented to illustrate their use in sample size planning. These design effects will enable appropriate sample size calculations to be performed for future randomised trials including both independent and paired data. Copyright

An induced natural selection heuristic for finding optimal Bayesian experimental designs

Abstract Bayesian optimal experimental design has immense potential to inform the collection of data so as to subsequently enhance our understanding of a variety of processes. However, a major impediment is the difficulty in evaluating optimal designs for problems with large, or high-dimensional, design spaces. An efficient search heuristic suitable for general optimisation problems, with a particular focus on optimal Bayesian experimental design problems, is proposed. The heuristic evaluates the objective (utility) function at an initial, randomly generated set of input values. At each generation of the algorithm, input values are “accepted” if their corresponding objective (utility) function satisfies some acceptance criteria, and new inputs are sampled about these accepted points. The new algorithm is demonstrated by evaluating the optimal Bayesian experimental designs for the previously considered death, pharmacokinetic and logistic regression models. Comparisons to the current “gold-standard” method are given to demonstrate the proposed algorithm as a computationally-efficient alternative for moderately-large design problems (i.e., up to approximately 40-dimensions).

Accounting for twin births in sample size calculations for randomised trials

BACKGROUND Including twins in randomised trials leads to non-independence or clustering in the data. Clustering has important implications for sample size calculations, yet few trials take this into account. Estimates of the intracluster correlation coefficient (ICC), or the correlation between outcomes of twins, are needed to assist with sample size planning. Our aims were to provide ICC estimates for infant outcomes, describe the information that must be specified in order to account for clustering due to twins in sample size calculations, and develop a simple tool for performing sample size calculations for trials including twins. METHODS ICCs were estimated for infant outcomes collected in four randomised trials that included twins. The information required to account for clustering due to twins in sample size calculations is described. A tool that calculates the sample size based on this information was developed in Microsoft Excel and in R as a Shiny web app. RESULTS ICC estimates ranged between -0.12, indicating a weak negative relationship, and 0.98, indicating a strong positive relationship between outcomes of twins. Example calculations illustrate how the ICC estimates and sample size calculator can be used to determine the target sample size for trials including twins. CONCLUSIONS Clustering among outcomes measured on twins should be taken into account in sample size calculations to obtain the desired power. Our ICC estimates and sample size calculator will be useful for designing future trials that include twins. Publication of additional ICCs is needed to further assist with sample size planning for future trials.

ASXL1 and BIM germ line variants predict response and identify CML patients with the greatest risk of imatinib failure

Scoring systems used at diagnosis of chronic myeloid leukemia (CML), such as Sokal risk, provide important response prediction for patients treated with imatinib. However, the sensitivity and specificity of scoring systems could be enhanced for improved identification of patients with the highest risk. We aimed to identify genomic predictive biomarkers of imatinib response at diagnosis to aid selection of first-line therapy. Targeted amplicon sequencing was performed to determine the germ line variant profile in 517 and 79 patients treated with first-line imatinib and nilotinib, respectively. The Sokal score and ASXL1 rs4911231 and BIM rs686952 variants were independent predictors of early molecular response (MR), major MR, deep MRs (MR4 and MR4.5), and failure-free survival (FFS) with imatinib treatment. In contrast, the ASXL1 and BIM variants did not consistently predict MR or FFS with nilotinib treatment. In the imatinib-treated cohort, neither Sokal or the ASXL1 and BIM variants predicted overall survival (OS) or progression to accelerated phase or blast crisis (AP/BC). The Sokal risk score was combined with the ASXL1 and BIM variants in a classification tree model to predict imatinib response. The model distinguished an ultra-high-risk group, representing 10% of patients, that predicted inferior OS (88% vs 97%; P = .041), progression to AP/BC (12% vs 1%; P = .034), FFS (P < .001), and MRs (P < .001). The ultra-high-risk patients may be candidates for more potent or combination first-line therapy. These data suggest that germ line genetic variation contributes to the heterogeneity of response to imatinib and may contribute to a prognostic risk score that allows early optimization of therapy.

Endotracheal tube placement during computed tomography of brachycephalic dogs alters upper airway dimensional measurements

Computed tomography (CT) is used to document upper airway lesions in dogs with brachycephalic obstructive airway syndrome. The presence of an endotracheal tube during CT scanning is often required for general anesthesia. We hypothesized that the endotracheal tube placement would change the soft tissue dimensions of the upper airway. The aims of this prospective, method comparison study were to evaluate the reliability of the previously reported upper airway CT measurements with endotracheal tube placement, and to propose measurements that are minimally affected by the endotracheal tube. Twenty brachycephalic dogs were included in this study. Each dog underwent head/neck CT with an endotracheal tube, followed by a second scan without the endotracheal tube. Ten measurements of the soft palate, nasopharynx, and trachea were performed. Tracheal dimension was significantly larger with the endotracheal tube compared to without, whereas the soft palate cross-sectional area was significantly smaller with the endotracheal tube than without the endotracheal tube. The influence of the endotracheal tube on the caudal nasopharynx cross-sectional (transverse-sectional) area varied with a mean proportional absolute difference of 35%. Rostral soft palate thickness, tracheal perimeter, and cross-sectional area of the rostral nasopharynx were the measurements least affected by the endotracheal tube (intraclass correlation coefficient = 0.964, 0.967, and 0.951, respectively). Therefore, we proposed that these three measurements may be most useful for future brachycephalic obstructive airway syndrome studies that require CT scanning of intubated animals. However, with endotracheal tube placement, measurements of soft palate length, caudal nasopharyngeal cross-sectional area, and trachea height and width may not be reliable.

EFFECTS OF IMAGE PLANE, PATIENT POSITIONING, AND FORAMINAL ZONE ON MAGNETIC RESONANCE IMAGING MEASUREMENTS OF CANINE LUMBOSACRAL INTERVERTEBRAL FORAMINA

Degenerative lumbosacral stenosis has been suspected to have a dynamic component, especially regarding encroachment of the L7 nerve roots exiting the lumbosacral foramina. Angled cross-sectional imaging of the neuroforamina has been found improve the accuracy of the diagnosis of stenosis in humans. In this anatomic study, foraminal apertures were evaluated by MRI at the entry, middle, and exit zones of the nerve roots in 30 dogs that were clinically affected by lumbosacral disease. Standard vs. oblique planar orientation and neutral vs. hyperextended positioning of the lumbosacral area were compared by measuring the median values for entry, middle, and exit zones. The neuroforaminal area acquired using oblique plane acquisition was significantly smaller than standard parasagittal measurements. Furthermore, standard parasagittal neuroforaminal dimensions in the hyperextended position were significantly smaller than standard parasagittal measurements in the neutral position. This statistical difference was even more pronounced for neuroforaminal dimension evaluated in the oblique plane and hyperextended position. Positioning of the dog during imaging has a significant effect on neuroforaminal dimension, corroborating the notion that spinal position may influence neural claudication in clinically affected patients. Reductions in neuroforaminal dimension are more evident on oblique planar image acquisition, suggesting that this approach may be more useful than parasagittal imaging as a tool for identifying subtle changes in L7 neuroforaminal dimensions in cases of canine lumbosacral stenosis.

An efficient moments-based inference method for within-host bacterial infection dynamics.

Over the last ten years, isogenic tagging (IT) has revolutionised the study of bacterial infection dynamics in laboratory animal models. However, quantitative analysis of IT data has been hindered by the piecemeal development of relevant statistical models. The most promising approach relies on stochastic Markovian models of bacterial population dynamics within and among organs. Here we present an efficient numerical method to fit such stochastic dynamic models to in vivo experimental IT data. A common approach to statistical inference with stochastic dynamic models relies on producing large numbers of simulations, but this remains a slow and inefficient method for all but simple problems, especially when tracking bacteria in multiple locations simultaneously. Instead, we derive and solve the systems of ordinary differential equations for the two lower-order moments of the stochastic variables (mean, variance and covariance). For any given model structure, and assuming linear dynamic rates, we demonstrate how the model parameters can be efficiently and accurately estimated by divergence minimisation. We then apply our method to an experimental dataset and compare the estimates and goodness-of-fit to those obtained by maximum likelihood estimation. While both sets of parameter estimates had overlapping confidence regions, the new method produced lower values for the division and death rates of bacteria: these improved the goodness-of-fit at the second time point at the expense of that of the first time point. This flexible framework can easily be applied to a range of experimental systems. Its computational efficiency paves the way for model comparison and optimal experimental design.

Designing group dose-response studies in the presence of transmission

Dose-response studies are used throughout pharmacology, toxicology and in clinical research to determine safe, effective, or hazardous doses of a substance. When involving animals, the subjects are often housed in groups; this is in fact mandatory in many countries for social animals, on ethical grounds. An issue that may consequently arise is that of unregulated between-subject dosing (transmission), where a subject may transmit the substance to another subject. Transmission will obviously impact the assessment of the dose-response relationship, and will lead to biases if not properly modelled. Here we present a method for determining the optimal design - pertaining to the size of groups, the doses, and the killing times - for such group dose-response experiments, in a Bayesian framework. Our results are of importance to minimising the number of animals required in order to accurately determine dose-response relationships. Furthermore, we additionally consider scenarios in which the estimation of the amount of transmission is also of interest. A particular motivating example is that of Campylobacter jejuni in chickens. Code is provided so that practitioners may determine the optimal design for their own studies.

Inferring within-host bottleneck size: A Bayesian approach

Recent technical developments in microbiology have led to new discoveries on the within-host dynamics of bacterial infections in laboratory animals. In particular, they have highlighted the importance of stochastic bottlenecks at the onset of invasive disease. A number of approaches exist for bottleneck-size estimation with respect to within-host bacterial infections; however, some are more appropriate than others under certain circumstances. A Bayesian comparison of several approaches is made in terms of the availability of isogenic multitype bacteria (e.g., WITS), knowledge of post-bottleneck dynamics, and the suitability of dilution with monotype bacteria. A sampling approach to bottleneck-size estimation is also introduced. The results are summarised by a guiding flowchart, which we hope will promote the use of quantitative models in microbiology to refine the analysis of animal experiment data.