The potential that telemedicine interventions may have in effectively delivering remote specialized cardiovascular care to large numbers of patients with heart diseases has recently come under question. In the first phase of this thesis, a systematic review and meta-analysis was conducted to compare the impact of a basic form of telemedicine that is regular patient follow-up by telephone, with usual care for individuals with coronary artery disease following their discharge. In the second phase of this thesis, a network meta-analysis, using Bayesian methods for multiple treatment comparisons, was conducted to compare the more complex forms of telemedicine for patients with heart failure. In the third and final phase of this thesis, a randomized controlled trial was designed to compare the impact of two forms of telemedicine, identified in the earlier two phases as being the most promising, on clinical outcomes, cardiac risk factors and patient reported outcomes.

Cell cycle (CC) and TP53 regulatory networks are frequently deregulated in cancer. While numerous genome-wide studies of TP53 and CC-regulated genes have been performed, significant variation between studies has made it difficult to assess regulation of any given gene of interest. To overcome the limitation of individual studies, we developed a meta-analysis approach to identify high confidence target genes that reflect their frequency of identification in independent datasets. Gene regulatory networks were generated by comparing differential expression of TP53 and CC-regulated genes with chromatin immunoprecipitation studies for TP53, RB1, E2F, DREAM, B-MYB, FOXM1 and MuvB. RNA-seq data from p21-null cells revealed that gene downregulation by TP53 generally requires p21 (CDKN1A). Genes downregulated by TP53 were also identified as CC genes bound by the DREAM complex. The transcription factors RB, E2F1 and E2F7 bind to a subset of DREAM target genes that function in G1/S of the CC while B-MYB, FOXM1 and MuvB control G2/M gene expression. Our approach yields high confidence ranked target gene maps for TP53, DREAM, MMB-FOXM1 and RB-E2F and enables prediction and distinction of CC regulation. A web-based atlas at www.targetgenereg.org enables assessing the regulation of any human gene of interest.

Most statistical methods to design and analyze randomized controlled trials with time-to-event data, and synthesize their results in meta-analyses, use the hazard ratio (HR) as the measure of treatment effect. However, the HR relies on the proportional hazard assumption which is often violated, especially in cancer trials. In addition, the HR might be challenging to interpret and is frequently misinterpreted as a risk ratio (RR). In meta-analysis, conventional methods ignore that HRs are estimated over different time supports when the component trials have different follow-up durations. These issues also pertain to advanced statistical methods, such as network meta-analysis and surrogate endpoints validation. Novel methods that rely on the difference in restricted mean survival times (RMST) would help addressing these issues. In this dissertation, I first developed a Bayesian network meta-analysis model using the difference in RMST. This model synthesizes all the available evidence from multiple time points and treatment comparisons simultaneously through within-study covariance and between-study covariance for the differences in RMST. I proposed an estimator of the within-study covariance and estimated the model under the Bayesian framework. The simulation studies showed adequate performance in terms of mean bias and mean squared error. I illustrated the model on a network of randomized trials of second-line treatments of advanced non-small-cell lung cancer. Second, I introduced a novel two-stage meta-analytical model to evaluate trial-level surrogacy. I measured trial-level surrogacy by the coefficient of determination at multiple time points based on the differences in RMST. The model borrows strength across data available at multiple time points and enables assessing how the strength of surrogacy changes over time. Simulation studies showed that the estimates of coefficients of determination are unbiased and have high precision in almost all of the scenarios we examined. I demonstrated my model in two individual patient data meta-analyses in gastric cancer. Both methods, for network meta-analysis and surrogacy evaluation, have the advantage of not involving extrapolation beyond the observed time support in component trials and of not relying on the proportional hazard assumption. Finally, motivated by the common misinterpretation of the HR as a RR, I investigated the theoretical relationship between the HR and the RR and compared empirically the treatment effects measured by the HR and the RR in a large sample of oncology RCTs. When there is evidence of superiority for experimental group, misinterpreting the HR as the RR leads to overestimating the benefits by about 20%. / 2026-02-08T00:00:00Z

Les revues systématiques sont des outils indispensables à la synthèse des connaissances en évaluation thérapeutique. Il est désormais fréquent que plusieurs traitements soient disponibles pour une même indication. L’objectif des patients et des cliniciens est alors de savoir quels sont, parmi l’ensemble des traitements disponibles, le(s) meilleur(s). Compte tenu de la nécessité de synthétiser les données disponibles pour tous les traitements et de maintenir cette synthèse à jour, notre objectif était d’évaluer les limites du système actuel de synthèse et de développer une méthodologie alternative. Nous avons d’abord évalué la capacité des revues systématiques à prendre en compte l’ensemble des preuves disponibles sur l’effet des multiples traitements. Nous avons utilisé l’exemple des traitements de deuxième ligne du cancer bronchique non à petites cellules métastatique non muté pour EGFR ou de statut inconnu. Nous avons montré que les 29 revues systématiques publiées jusque 2015 sur cette question, considérées collectivement, fournissaient une synthèse fragmentée et non à jour de la preuve disponible. Au moins 40% des 77 essais, des 45 traitements, des 54 comparaisons de traitements et des 28 636 patients n’étaient constamment pas pris en compte dans les revues systématiques. Nous avons discuté les raisons pour lesquelles le système de synthèse des données actuel ne permettait pas de couvrir l’ensemble des données disponibles. Nous avons ensuite développé une nouvelle forme de synthèse de la preuve disponible au cours du temps, la méta-analyse en réseau cumulative et dynamique. Elle consiste à passer d’une série de méta-analyses à une méta-analyse en réseau unique, incluant l’ensemble des traitements disponibles pour une indication donnée, avec une mise à jour du réseau d’essais et de la synthèse des données dès que les résultats d’un nouvel essai deviennent disponibles. Elle débute par une méta-analyse en réseau initiale suivie d’une succession de mises à jour répétées à intervalles réguliers. Nous avons décrit les étapes méthodologiques, et développé le protocole d’une étude de preuve de concept, appliquée aux traitements de deuxième ligne du cancer bronchique non à petites cellules. Enfin, nous avons réalisé la méta-analyse en réseau initiale sur ce même exemple. Nous avons inclus 98 essais randomisés évaluant 60 traitements chez 34 179 patients. Nous avons montré que les traitements par immunothérapie (nivolumab et pembrolizumab) avaient un effet sur la survie globale supérieur aux chimiothérapies et thérapeutiques ciblées actuellement recommandées (nivolumab versus docetaxel HR=0,68 (IC95% 0,55-0,83) ; versus pemetrexed HR=0,65 (0,5-0,83) ; versus erlotinib HR=0,66 (0,51-0,84) and versus gefitinib HR=0,65 (0,51-0,82)). Les résultats étaient similaires pour le pembrolizumab. Pour la survie sans progression, le nivolumab avait aussi un effet supérieur aux quatre traitements recommandés. La méta-analyse en réseau cumulative et dynamique pourrait devenir l’outil permettant de changer de paradigme dans la synthèse des connaissances afin d’améliorer la prise de décision médicale. / Systematic reviews are essential tools to synthesize available evidence for therapeutic evaluation. Multiple treatments are now frequently available for a given condition. Patients and physicians want to know which one is the best among all treatments. Thus we need to retrieve and synthesize all available evidence across all treatments and furthermore to maintain it updated when new evidence and new treatments become available. Our objective was to evaluate the limits of the current ecosystem of evidence synthesis and to develop an alternative methodology. We have first assessed the capacity of systematic reviews to cover all available evidence of multiple treatments. We took the example of second-line treatments of advanced non-small cell lung cancer with EGFR wild-type or unknown status. We have shown that the 29 systematic reviews published in this condition up to 2015, considered collectively, failed to provide a complete and updated synthesis of all available evidence. Almost 40% of the 77 trials, of the 45 treatments, of the 54 treatment comparisons and of the 28,636 patients were always missing from systematic reviews. We have discussed the reasons why the ecosystem of evidence synthesis fails to encompass all available evidence. We then developed a new paradigm to synthesize evidence over time called live cumulative network meta-analysis. This new concept consists in switching from a series of standard meta-analyses to a single network meta-analysis covering all treatments and systematically updated as soon as the results of a new trial become available. Live cumulative network meta-analysis is initiated with a network meta-analysis which is iteratively updated. We have described the methodological steps, developed the protocol of a proof-of-concept study applied to second-line treatments of advanced non-small cell lung cancer. Finally, we have performed the initial network meta-analysis in this condition. We have included 98 trials including 34,179 patients and assessing 60 treatments. We have shown that nivolumab was more effective in term of overall survival compared to docetaxel HR=0.68 (IC95% 0.55-0.83), to pemetrexed HR=0.65 (0.5-0.83), to erlotinib HR=0.66 (0.51-0.84) and to gefitinib HR=0.65 (0.51-0.82). Similar results were found with pembrolizumab. In progression free survival, nivolumab had a more important treatment effect compared to the four recommended treatments. Live cumulative network meta-analysis should become a paradigmatic shift for systematic reviews and meta-analysis in order to improve medical decision making.

The relative risk of major gastrointestinal bleeding (GIB) among different direct oral anticoagulants (DOACs) is debatable. Randomized controlled trials (RCTs) comparing DOACs with each other are lacking. We performed network meta-analysis to assess whether the risk of major GIB differs based on type and dose of DOAC. Literature search of PubMed, EMBASE and Cochrane databases from inception to August 2019, limited to English publications, was conducted to identify RCTs comparing DOACs with warfarin or enoxaparin for any indication. Primary outcome of interest was major GIB risk. We used frequentist network meta-analysis through the random-effects model to compare DOACs with each other and DOACs by dose to isolate the impact on major GIB. Twenty-eight RCTs, including 139 587 patients receiving six anticoagulants, were selected. The risk of major GIB for DOACs was equal to warfarin. Comparison of DOACs with each other did not show risk differences. After accounting for dose, rivaroxaban 20 mg, dabigatran 300 mg and edoxaban 60 mg daily had 47, 40 and 22% higher rates of major GIB versus warfarin, respectively. Apixaban 5 mg twice daily had lower major GIB compared to dabigatran 300 mg (OR, 0.63; 95% CI, 0.44-0.88) and rivaroxaban 20 mg (OR, 0.60; 95% CI, 0.43-0.83) daily. Heterogeneity was low, and the model was consistent without publication bias (Egger's test: P = 0.079). All RCTs were high-quality with low risk of bias. DOACs at standard dose, except apixaban, had a higher risk of major GIB compared to warfarin. Apixaban had a lower rate of major GIB compared to dabigatran and rivaroxaban.

京都大学 / 新制・課程博士 / 博士(医学) / 甲第23756号 / 医博第4802号 / 新制||医||1056(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 中山 健夫, 教授 村井 俊哉, 教授 小杉 眞司 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Les cancers des voies aérodigestives supérieures (VADS) représentent la 5e cause de cancer en France. Ils sont fréquemment découverts à un stade avancé, et leur mauvais pronostic a conduit à l’élaboration de traitements intensifiés. De nombreux essais randomisés ont évalué l’apport de la chimiothérapie et de modifications du fractionnement de la radiothérapie. Leurs résultats ont été synthétisés dans deux méta-analyses sur données individuelles coordonnées par l’Institut Gustave Roussy. Cependant ces méta-analyses génèrent des questions cliniques et méthodologiques, qui constituent le socle de cette thèse. Ainsi nous avons exploré par différents moyens l’étude de l’interaction entre des covariables de niveau individuel, le site tumoral, et l’effet du traitement. Nous avons adapté la méthodologie des méta-analyses en réseau pour les données de survie afin réaliser une analyse globale de l’ensemble de ces essais randomisés et classer les traitements selon leur efficacité sur la survie. Certains de ces traitements n’avaient pas fait l’objet de comparaison directe, et nos résultats se sont vérifiés dans des essais publiés ultérieurement. Nous avons passé en revue les avantages et les limites de la méta-analyse en réseau. Nous avons enfin engagé la mise à jour de ce corpus de méta-analyses pour produire des résultats en accord avec les pratiques actuelles, avec un suivi long, et en explorant des problématiques variées, telles que l’efficacité, la toxicité et l’adhérence au protocole thérapeutique. Les résultats finaux de la méta-analyse sur la chimiothérapie d’induction avec taxanes sont présentés dans cette thèse. / Head and neck cancers represent the fifth cause of death from cancer in France. They are often diagnosed at an advanced stage. The poor prognosis of these diseases has led to the introduction of intensified treatments. Numerous randomized trials have evaluated the benefits of the addition of chemotherapy to locoregional treatment and of the modification of radiotherapy fractionation. The results of these trials have been synthesized in two individual patient data meta-analyses coordinated by the Meta-Analysis Unit of Gustave Roussy Cancer Center. However these meta-analyses bring up clinical and methodological questions, some of which are dealt with in this thesis. First we have studied by different means the interaction between patient level covariate, tumor site and treatment effect. We have also adapted the methodology of network meta-analyses to survival data to perform a global analysis of the entire meta-analysis database, and to rank treatments according to their efficacy, including some treatments that had not been directly compared. Some of these results were eventually confirmed by subsequently published randomized trials. We have reviewed the advantages and limits of network meta-analysis. We have also launched the update of all these meta-analyses in order to produce results consistent with actual clinical practice, update patient follow-up, and collect additional data regarding treatment efficacy, toxicity and compliance. The final results of the taxane induction meta-analysis are presented in this manuscript.

La méta-analyse (MA) en réseau, en généralisant la MA conventionnelle, permet d'évaluer toutes les comparaisons deux à deux possibles entre interventions. Les biais de publication ont reçu peu d’attention dans ce contexte. Nous avons évalué l’impact des biais de publication en utilisant un réseau de 74 essais randomisés évaluant 12 antidépresseurs contre placebo enregistrés à la FDA et un réseau de 51 essais parmi les 74 dont les résultats étaient publiés. Nous avons montré comment les biais de publication biaisaient les quantités d'effet estimées et le classement des traitements. L'effet du biais de publication peut différer entre MA en réseau et MA conventionnelle en ce que les biais affectant un traitement peuvent affecter le classement de tous les traitements. Nous avons ensuite généralisé un test de détection des biais à la MA en réseau. Il est basé sur la comparaison entre les nombres attendu et observé d’essais avec résultats statistiquement significatifs sur l’ensemble du réseau. Nous avons montré par des études de simulation que le test proposé avait une puissance correcte après ajustement sur l’erreur de type I, excepté lorsque la variance inter-essais était élevée. Par ailleurs, le test indiquait un signal significatif de biais sur le réseau d’essais d’antidépresseurs publiés. Enfin, nous avons introduit deux modèles d’analyse de sensibilité des résultats d'une MA en réseau aux biais de publication: un modèle de méta-régression qui relie la quantité d’effet estimée à son erreur standard, et un modèle de sélection dans lequel on estime la propension d’un essai à être publié puis l’on redresse le poids des essais en fonction de cette propension. Nous les avons appliqués aux réseaux d’essais d’antidépresseurs. Ce test et ces modèles d'ajustement tirent leur force de tous les essais du réseau, sous l’hypothèse qu'un biais moyen commun opère sur toutes les branches du réseau. / Network meta-analysis (NMA), a generalization of conventional MA, allows for assessing all possible pairwise comparisons between multiple treatments. Reporting bias, a major threat to the validity of MA, has received little attention in the context of NMA. We assessed the impact of reporting bias empirically using data from 74 FDA-registered placebo-controlled trials of 12 antidepressants and their 51 matching publications. We showed how reporting bias biased NMA-based estimates of treatments efficacy and modified ranking. The effect of reporting bias in NMAs may differ from that in classical meta-analyses in that reporting bias affecting only one drug may affect the ranking of all drugs. Then, we extended a test to detect reporting bias in network of trials. It compares the number of expected trials with statistically significant results to the observed number of trials with significant p-values across the network. We showed through simulation studies that the test was fairly powerful after adjustment for size, except when between-trial variance was substantial. Besides, it showed evidence of bias in the network of published antidepressant trials. Finally, we introduced two methods of sensitivity analysis for reporting bias in NMA: a meta-regression model that allows the effect size to depend on its standard error and a selection model that estimates the propensity of trial results being published and in which trials with lower propensity are weighted up in the NMA model. We illustrated their use on the antidepressant datasets. The proposed test and adjustment models borrow strength from all trials across the network, under the assumption that conventional MAs in the network share a common mean bias mechanism.