Ismaïl Ahmed

False Discovery Rate Estimation for Frequentist Pharmacovigilance Signal Detection Methods

Pharmacovigilance systems aim at early detection of adverse effects of marketed drugs. They maintain large spontaneous reporting databases for which several automatic signaling methods have been developed. One limit of those methods is that the decision rules for the signal generation are based on arbitrary thresholds. In this article, we propose a new signal-generation procedure. The decision criterion is formulated in terms of a critical region for the P-values resulting from the reporting odds ratio method as well as from the Fishers exact test. For the latter, we also study the use of mid-P-values. The critical region is defined by the false discovery rate, which can be estimated by adapting the P-values mixture model based procedures to one-sided tests. The methodology is mainly illustrated with the location-based estimator procedure. It is studied through a large simulation study and applied to the French pharmacovigilance database.

Bayesian pharmacovigilance signal detection methods revisited in a multiple comparison setting

Pharmacovigilance spontaneous reporting systems are primarily devoted to early detection of the adverse reactions of marketed drugs. They maintain large spontaneous reporting databases (SRD) for which several automatic signalling methods have been developed. A common limitation of these methods lies in the fact that they do not provide an auto-evaluation of the generated signals so that thresholds of alerts are arbitrarily chosen. In this paper, we propose to revisit the Gamma Poisson Shrinkage (GPS) model and the Bayesian Confidence Propagation Neural Network (BCPNN) model in the Bayesian general decision framework. This results in a new signal ranking procedure based on the posterior probability of null hypothesis of interest and makes it possible to derive with a non-mixture modelling approach Bayesian estimators of the false discovery rate (FDR), false negative rate, sensitivity and specificity. An original data generation process that can be suited to the features of the SRD under scrutiny is proposed and applied to the French SRD to perform a large simulation study. Results indicate better performances according to the FDR for the proposed ranking procedure in comparison with the current ones for the GPS model. They also reveal identical performances according to the four operating characteristics for the proposed ranking procedure with the BCPNN and GPS models but better estimates when using the GPS model. Finally, the proposed procedure is applied to the French data.

Pooled analysis of iron-related genes in Parkinson's disease: association with transferrin.

Pathologic features of Parkinsons disease (PD) include death of dopaminergic neurons in the substantia nigra, presence of α-synuclein containing Lewy bodies, and iron accumulation in PD-related brain regions. The observed iron accumulation may be contributing to PD etiology but it also may be a byproduct of cell death or cellular dysfunction. To elucidate the possible role of iron accumulation in PD, we investigated genetic variation in 16 genes related to iron homeostasis in three case-control studies from the United States, Australia, and France. After screening 90 haplotype tagging single nucleotide polymorphisms (SNPs) within the genes of interest in the US study population, we investigated the five most promising gene regions in two additional independent case-control studies. For the pooled data set (1289 cases, 1391 controls) we observed a protective association (OR=0.83, 95% CI: 0.71-0.96) between PD and a haplotype composed of the A allele at rs1880669 and the T allele at rs1049296 in transferrin (TF; GeneID: 7018). Additionally, we observed a suggestive protective association (OR=0.87, 95% CI: 0.74-1.02) between PD and a haplotype composed of the G allele at rs10247962 and the A allele at rs4434553 in transferrin receptor 2 (TFR2; GeneID: 7036). We observed no associations in our pooled sample for haplotypes in SLC40A1, CYB561, or HFE. Taken together with previous findings in model systems, our results suggest that TF or a TF-TFR2 complex may have a role in the etiology of PD, possibly through iron misregulation or mitochondrial dysfunction within dopaminergic neurons.

MRI atrophy of the caudate nucleus and slower walking speed in the elderly

Cerebral white matter lesions are associated with poorer motor performances in the elderly, but the role of gray matter atrophy remains largely unknown. We investigated the cross-sectional relation between brain regional gray matter volumes and walking speed over 6m in the 3C-Dijon study, a large population-based study of community-dwelling persons aged 65 years and over (N=1623). Regional gray matter volumes were obtained using an automated anatomical labeling parcellation method. Multivariable analyses were performed using a semi-Bayes approach. After adjustment for potential confounders, persons who walked slower had a smaller volume of basal ganglia (regression coefficient [β]=0.054, standard error [SE]=0.028, p=0.05). In more detailed analyses, the volume of the caudate nucleus had a preponderant role on this association (β=0.049, SE=0.019, p=0.009), and walking speed decreased progressively with the volume of the caudate nucleus (p for linear trend<0.001). These results underline the role of gray matter subcortical structures, in particular of the caudate nucleus, in the age-related decline of motor performances among community-dwelling elderly subjects.

Lack of replication of the GRIN2A-by-coffee interaction in Parkinson disease

Overview The etiology of Parkinson disease (PD) involves both genetic susceptibility and environmental exposures. In particular, coffee consumption is inversely associated with PD but the mechanisms underlying this intriguing association are unknown. According to a recent genome-wide gene–environment interaction study, the inverse coffee–PD association was two times stronger among carriers of the T allele of SNP rs4998386 in gene GRIN2A than in homozygotes for the C allele. We attempted to replicate this result in a similarly sized pooled analysis of 2,289 cases and 2,809 controls from four independent studies (Denmark, France, Seattle-United States (US), and Rochester-US) with detailed caffeinated coffee consumption data and rs4998386 genotypes. Using a variety of definitions of coffee drinking and statistical modeling techniques , we failed to replicate this interaction. Notably, whereas in the original study there was an association between rs4998386 and coffee consumption among controls, but not among cases, none of the datasets analyzed here indicated an association between rs4998386 and coffee consumption among controls. Based on large, well-characterized datasets independent from the original study, our results are not in favor of an interaction between caffeinated coffee consumption and rs4998386 for PD risk and suggest that the original finding may have been driven by an association of coffee consumption with rs4998386 in controls. The next years will likely see an increasing number of papers examining gene–environment interactions at the genome-wide level, which poses important methodological challenges. Our findings underline the need for a careful assessment of the findings of such studies.

Selection of genes for gene-environment interaction studies: a candidate pathway-based strategy using asthma as an example.

BackgroundThe identification of gene by environment (GxE) interactions has emerged as a challenging but essential task to fully understand the complex mechanism underlying multifactorial diseases. Until now, GxE interactions have been investigated by candidate approaches examining a small number of genes, or agnostically at the genome wide level.Presentation of the hypothesisIn this paper, we propose a gene selection strategy for investigation of gene-environment interactions. This strategy integrates the information on biological processes shared by genes, the canonical pathways to which they belong and the biological knowledge related to the environment in the gene selection process. It relies on both bioinformatics resources and biological expertise.Testing the hypothesisWe illustrate our strategy by considering asthma, tobacco smoke as the environmental exposure, and genes sharing the same biological function of “response to oxidative stress”. Our filtering strategy leads to a list of 28 pathways involving 182 genes for further GxE investigation.Implications of the hypothesisBy integrating the environment into the gene selection process, we expect that our strategy will improve the ability to identify the joint effects and interactions of environmental and genetic factors in disease.

Occupational exposures and fluorescent oxidation products in 723 adults of the EGEA study

Occupational asthma can be induced by a variety of agents, including high and low molecular weight sensitisers, and respiratory irritants [1]. The role of exposure to cleaning products and disinfectants in work-related asthma is increasingly recognised, although the specific substances that increase asthma risk are not well identified [2]. Some of the numerous agents contained in these products are chemical sensitisers, but most are hypothesised to act as respiratory irritants [2]. While high molecular weight sensitisers are known to cause occupational asthma through a typical allergic response, the pathophysiological mechanisms involved in occupational asthma induced by low molecular weight (LMW) chemicals, and in irritant-induced asthma, remain poorly understood [1, 3, 4]. Associations between occupational exposures to asthmogenic chemicals and irritants and oxidative stress were found http://ow.ly/K6RSt

Comparison of two drug safety signals in a pharmacovigilance data mining framework.

Since adverse drug reactions are a major public health concern, early detection of drug safety signals has become a top priority for regulatory agencies and the pharmaceutical industry. Quantitative methods for analyzing spontaneous reporting material recorded in pharmacovigilance databases through data mining have been proposed in the last decades and are increasingly used to flag potential safety problems. While automated data mining is motivated by the usually huge size of pharmacovigilance databases, it does not systematically produce relevant alerts. Moreover, each detected signal requires appropriate assessment that may involve investigation of the whole therapeutic class. The goal of this article is to provide a methodology for comparing two detected signals. It is nested within the automated surveillance framework as (1) no extra information is required and (2) no simple inference on the actual risks can be extrapolated from spontaneous reporting data. We designed our methodology on the basis of two classical methods used for automated signal detection: the Bayesian Gamma Poisson Shrinker and the frequentist Proportional Reporting Ratio. A simulation study was conducted to assess the performances of both proposed methods. The latter were used to compare cardiovascular signals for two HIV treatments from the French pharmacovigilance database.

Smoking and Parkinson disease: Evidence for gene-by-smoking interactions

Objective To investigate whether cigarette smoking interacts with genes involved in individual susceptibility to xenobiotics for the risk of Parkinson disease (PD). Methods Two French population-based case-control studies (513 patients, 1,147 controls) were included as a discovery sample to examine gene-smoking interactions based on 3,179 single nucleotide polymorphisms (SNPs) in 289 genes involved in individual susceptibility to xenobiotics. SNP–by–cigarette smoking interactions were tested in the discovery sample through an empirical Bayes (EB) approach. Nine SNPs were selected for replication in a population-based case-control study from California (410 patients, 845 controls) with standard logistic regression and the EB approach. For SNPs that replicated, we performed pooled analyses including the discovery and replication datasets and computed pooled odds ratios and confidence intervals (CIs) using random-effects meta-analysis. Results Nine SNPs interacted with smoking in the discovery dataset and were selected for replication. Interactions of smoking with rs4240705 in the RXRA gene and rs1900586 in the SLC17A6 gene were replicated. In pooled analyses (logistic regression), the interactions between smoking and rs4240705-G and rs1900586-G were 1.66 (95% CI 1.28–2.14, p = 1.1 × 10−4, p for heterogeneity = 0.366) and 1.61 (95% CI 1.17–2.21, p = 0.003, p for heterogeneity = 0.616), respectively. For both SNPs, while smoking was significantly less frequent in patients than controls in AA homozygotes, this inverse association disappeared in G allele carriers. Conclusions We identified and replicated suggestive gene-by-smoking interactions in PD. The inverse association of smoking with PD was less pronounced in carriers of minor alleles of both RXRA-rs4240705 and SLC17A6-rs1900586. These findings may help identify biological pathways involved in the inverse association between smoking and PD.