Mad-Hélénie Elsensohn

Familial vs. sporadic sarcoidosis: BTNL2 polymorphisms, clinical presentations, and outcomes in a French cohort

BackgroundThe occurrence of familial forms of sarcoidosis (OMIM 181100) suggests a genetic predisposition. The involvement of butyrophilin-like 2 (BTNL2) gene (rs2076530 variant) has to be investigated.ResultsThe study performed independent analyses of BTNL2 polymorphism, clinical phenotypes, and outcomes in familial vs. sporadic presentations in 256 sporadic and 207 familial cases from 140 families. The logistic multivariate model showed that a young age at diagnosis and the combination of lung and skin involvement at diagnosis may distinguish sporadic from familial sarcoidosis (p = 0.016 and p = 0.041). We observed also that Sarcoid Clinical Activity Classification (SCAC) profiles were significantly different between familial and sporadic cases (p = 0.0497).Variant rs2076530 was more frequent in patients than in controls (OR = 2.02; 95% CI: [1.32–3.09]) but showed no difference between sporadic and familial cases and no difference according to the clinical phenotype or the outcome.ConclusionDespite a significant difference in BTNL2 polymorphism between sarcoid patients and controls, there was no such difference between familial and sporadic sarcoidosis cases and no correlation between BTNL2 polymorphism and disease severity or outcome. Thus, BTNL2 difference cannot be considered as a key marker for disease classification or patient management.

A graphical method to assess distribution assumption in group-based trajectory models

Group-based trajectory models had a rapid development in the analysis of longitudinal data in clinical research. In these models, the assumption of homoscedasticity of the residuals is frequently made but this assumption is not always met. We developed here an easy-to-perform graphical method to assess the assumption of homoscedasticity of the residuals to apply especially in group-based trajectory models. The method is based on drawing an envelope to visualize the local dispersion of the residuals around each typical trajectory. Its efficiency is demonstrated using data on CD4 lymphocyte counts in patients with human immunodeficiency virus put on antiretroviral therapy. Four distinct distributions that take into account increasing parts of the variability of the observed data are presented. Significant differences in group structures and trajectory patterns were found according to the chosen distribution. These differences might have large impacts on the final trajectories and their characteristics; thus on potential medical decisions. With a single glance, the graphical criteria allow the choice of the distribution that best capture data variability and help dealing with a potential heteroscedasticity problem.

Comparison of two next-generation sequencing kits for diagnosis of epileptic disorders with a user-friendly tool for displaying gene coverage, DeCovA

In recent years, molecular genetics has been playing an increasing role in the diagnostic process of monogenic epilepsies. Knowing the genetic basis of one patients epilepsy provides accurate genetic counseling and may guide therapeutic options. Genetic diagnosis of epilepsy syndromes has long been based on Sanger sequencing and search for large rearrangements using MLPA or DNA arrays (array-CGH or SNP-array). Recently, next-generation sequencing (NGS) was demonstrated to be a powerful approach to overcome the wide clinical and genetic heterogeneity of epileptic disorders. Coverage is critical for assessing the quality and accuracy of results from NGS. However, it is often a difficult parameter to display in practice. The aim of the study was to compare two library-building methods (Haloplex, Agilent and SeqCap EZ, Roche) for a targeted panel of 41 genes causing monogenic epileptic disorders. We included 24 patients, 20 of whom had known disease-causing mutations. For each patient both libraries were built in parallel and sequenced on an Ion Torrent Personal Genome Machine (PGM). To compare coverage and depth, we developed a simple homemade tool, named DeCovA (Depth and Coverage Analysis). DeCovA displays the sequencing depth of each base and the coverage of target genes for each genomic position. The fraction of each gene covered at different thresholds could be easily estimated. None of the two methods used, namely NextGene and Ion Reporter, were able to identify all the known mutations/CNVs displayed by the 20 patients. Variant detection rate was globally similar for the two techniques and DeCovA showed that failure to detect a mutation was mainly related to insufficient coverage.

Statistical method to compare massive parallel sequencing pipelines

BackgroundToday, sequencing is frequently carried out by Massive Parallel Sequencing (MPS) that cuts drastically sequencing time and expenses. Nevertheless, Sanger sequencing remains the main validation method to confirm the presence of variants. The analysis of MPS data involves the development of several bioinformatic tools, academic or commercial. We present here a statistical method to compare MPS pipelines and test it in a comparison between an academic (BWA-GATK) and a commercial pipeline (TMAP-NextGENe®), with and without reference to a gold standard (here, Sanger sequencing), on a panel of 41 genes in 43 epileptic patients. This method used the number of variants to fit log-linear models for pairwise agreements between pipelines. To assess the heterogeneity of the margins and the odds ratios of agreement, four log-linear models were used: a full model, a homogeneous-margin model, a model with single odds ratio for all patients, and a model with single intercept. Then a log-linear mixed model was fitted considering the biological variability as a random effect.ResultsAmong the 390,339 base-pairs sequenced, TMAP-NextGENe® and BWA-GATK found, on average, 2253.49 and 1857.14 variants (single nucleotide variants and indels), respectively. Against the gold standard, the pipelines had similar sensitivities (63.47% vs. 63.42%) and close but significantly different specificities (99.57% vs. 99.65%; p < 0.001). Same-trend results were obtained when only single nucleotide variants were considered (99.98% specificity and 76.81% sensitivity for both pipelines).ConclusionsThe method allows thus pipeline comparison and selection. It is generalizable to all types of MPS data and all pipelines.

Estimating the parameters of multi-state models with time-dependent covariates through likelihood decomposition

BACKGROUND Multi-state models become complex when the number of states is large, when back and forth transitions between states are allowed, and when time-dependent covariates are inevitable. However, these conditions are sometimes necessary in the context of medical issues. For instance, they were needed for modelling the future treatments of patients with end-stage renal disease according to age and to various treatments. METHODS The available modelling tools do not allow an easy handling of all issues; we designed thus a specific multi-state model that takes into account the complexity of the research question. Parameter estimation relied on decomposition of the likelihood and separate maximisations of the resulting likelihoods. This was possible because there were no interactions between patient treatment courses and because all exact times of transition from any state to another were known. Poisson likelihoods were calculated using the time spent at risk in each state and the observed transitions between each state and all others. The likelihoods were calculated on short time intervals during which age was considered as constant. RESULTS The method was not limited by the number of parameters to estimate; it could be applied to a multi-state model with 10 renal replacement therapies. Supposing the parameters of the model constant over each of seven time intervals, this method was able to estimate one hundred age-dependent transitions. CONCLUSIONS The method is easy to adapt to any disease with numerous states or grades as long as the disease does not imply interactions between patient courses.

Centralization errors in comparative genomic hybridization array analysis of pituitary tumor samples

Reliable interpretation of comparative genomic hybridization array (aCGH) results requires centralization and normalization of the data. We evaluated the reliability of aCGH centralization by comparing aCGH results (with classical centralization‐normalization steps) to fluorescence in situ hybridization (FISH) results. In addition, we propose a method to correct centralization bias. Sixty‐six pituitary tumors were analyzed (Agilent aCGH + SNP 4 × 180K microarray). For each tumor, the FISH‐based log2(ratios) of a subset of chromosomes were compared with the corresponding aCGH raw log2(ratios). With our new normalization‐centralization process, this difference was added to all log2(ratios), before performing loess regression on non‐altered probes only. Finally, the mean log2(ratio) and the percentage of normal probes were compared between CGHnormaliter and our new FISH‐based method. For 11 tumors, FISH results and raw CGH log2(ratios) differed significantly. In addition, nine tumors showed discrepancies between results generated by CGHnormaliter and our new‐method. Such discrepancies seemed to occur with tumours with many abnormalities (0%‐40% normal probes), rather than in those tumours with fewer abnormalities (31%‐100% normal probes). Five tumors had too few normal probes to allow normalization. In these tumors, which can exhibit many changes in DNA copy number, we found that centralization bias was frequent and uncorrected by current normalization methods. Therefore, an external control for centralization, such as FISH analysis, is required to insure reliable interpretation of aCGH data.

Restricted mean survival time over 15 years for patients starting renal replacement therapy

Background. The restricted mean survival time (RMST) estimates life expectancy up to a given time horizon and can thus express the impact of a disease. The aim of this study was to estimate the 15‐year RMST of a hypothetical cohort of incident patients starting renal replacement therapy (RRT), according to their age, gender and diabetes status, and to compare it with the expected RMST of the general population. Methods. Using data from 67 258 adult patients in the French Renal Epidemiology and Information Network (REIN) registry, we estimated the RMST of a hypothetical patient cohort (and its subgroups) for the first 15 years after starting RRT (cRMST) and used the general population mortality tables to estimate the expected RMST (pRMST). Results were expressed in three different ways: the cRMST, which calculates the years of life gained under the hypothesis of 100% death without RRT treatment, the difference between the pRMST and the cRMST (the years lost), and a ratio expressing the percentage reduction of the expected RMST: (pRMST − cRMST)/pRMST. Results. Over their first 15 years of RRT, the RMST of end‐stage renal disease (ESRD) patients decreased with age, ranging from 14.3 years in patients without diabetes aged 18 years at ESRD to 1.8 years for those aged 90 years, and from 12.7 to 1.6 years, respectively, for those with diabetes; expected RMST varied from 15.0 to 4.1 years between 18 and 90 years. The number of years lost in all subgroups followed a bell curve that was highest for patients aged 70 years. After the age of 55 years in patients with and 70 years in patients without diabetes, the reduction of the expected RMST was >50%. Conclusion. While neither a clinician nor a survival curve can predict with absolute certainty how long a patient will live, providing estimates on years gained or lost, or percentage reduction of expected RMST, may improve the accuracy of the prognostic estimates that influence clinical decisions and information given to patients.

Using repeated-prevalence data in multi-state modeling of renal replacement therapy

Multi-state models help predict future numbers of patients requiring specific treatments but these models require exhaustive incidence data. Deriving reliable predictions from repeated-prevalence data would be helpful. A new method to model the number of patients that switch between therapeutic modalities using repeated-prevalence data is presented and illustrated. The parameters and goodness of fit obtained with the new method and repeated-prevalence data were compared to those obtained with the classical method and incidence data. The multi-state model parameters’ confidence intervals obtained with annually collected repeated-prevalence data were wider than those obtained with incidence data and six out of nine pairs of confidence intervals did not overlap. However, most parameters were of the same order of magnitude and the predicted patient distributions among various renal replacement therapies were similar regardless of the type of data used. In the absence of incidence data, a multi-state model can still be successfully built with annually collected repeated-prevalence data to predict the numbers of patients requiring specific treatments. This modeling technique can be extended to other chronic diseases.