L'ADN tumoral circulant (ADNtumc) est apparu depuis plusieurs années comme un biomarqueur prometteur susceptible d'apporter des informations permettant l'optimisation de la prise en charge du patient en oncologie. L'objectif de cette thèse était double et s'articule autour de deux axes : i) évaluer différentes conditions préanalytiques et analytiques (digitale PCR (dPCR) principalement) pour la détection de ce biomarqueur ii) évaluer l'intérêt clinique potentiel de ce biomarqueur en oncologie digestive. La première partie rapporte 3 travaux (3 articles originaux dont une collaboration nationale (équipe parisienne dirigée par J. Tost)). Dans le travail n°1, nous avons montré la faisabilité de détecter l'ADNtumc par dPCR directement à partir du plasma de 43 prélèvements de patients avec cancer colorectal métastatique (CCRm). Il n'y avait pas de différence significative pour le taux de détection des mutations KRAS circulantes entre les groupes avec et sans extraction d'ADN (93 % (40/43) versus 88 °A) (38/43), respectivement). Dans le travail n°2, nous avons mis au point une méthode basée sur l'apport d'héparinase pour la détection d'ADNtumc à partir de 194 prélèvements héparinés de patients suivis en oncologie. Ce traitement des échantillons par l'héparinase permettait l'analyse de l'ADNtumc pour 117/194 (60 %) patients avec inhibition Préalable de la dPCR par l'héparine. Enfin, dans le travail n°3, nous avons comparé plusieurs plate-formes de détection d'ADNtumc et montré que la dPCR affichait des résultats de détection comparables sur le plan qualitatif et quantitatif avec une plateforme ultrasensible d'Enhanced-ice-COLD-PCR (E-ice-COLD-PCR) pour les échantillons avec une fréquence allélique d'ADNtumc >0,4 °A La deuxième partie rapporte 3 travaux (3 articles originaux) sur l'intérêt clinique de la détection d'ADNtumc par dPCR en oncologie digestive. Nous avons ainsi montré que ce biomarqueur conférait un intérêt diagnostique (travail n°4), Pronostique (travail n 4 à 6) et prédictif de la réponse aux traitements (travail n°6) chez les Patients avec adénocarcinome pancréatique (AP) (travail n°4) et CCRm (travail n°5 à 6). / For several years, circulating tumor DNA (ctDNA) has emerged as a promising biomarker providing relevant information to optimize patient care in oncology. The aim of this thesis was both: (i) to evaluate different preanalytical and analytical conditions (digital PCR (dPCR) mainly) for the detection of this biomarker; (ii) to evaluate the potential clinical interest of this biomarker in digestive oncology. The first part reports 3 works (3 original articles including a national collaboration (Parisian team led by J. lost)). In work no. 1, we have shown the feasibility of ctDNA detection by dPCR directly from the plasma of 43 samples from patients with metastatic colorectal cancer (mCRC). There was no significant difference in the detection rate of circulating KRAS mutations between groups with and without DNA extraction (93% (40/43) versus 88% (38/43), respectively). In work no. 2, we developed a method based on the heparinase addition for the ctDNA detection from 194 heparinized samples of patients followed in oncology. This treatment of samples by heparinase allowed the ctDNA analysis of 117/194 (60%) patients with prior inhibition of dPCR by heparin. Finally, in work no. 3, we compared several ctDNA detection platforms and snowed that dPCR displayed qualitatively and quantitatively comparable detection results with an ultrasensitive platform of E-ice-COLD-PCR for the samples with ctDNA allelic fraction ?.0 4%. The second part reports 3 works (3 original articles) on the clinical interest of the ctDNA detection by dPCR in digestive oncology. We have thus shown that this biomarker had a diagnostic (work no. 4). prognostic (works no. 4 to 6) and predictive response to treatments (work no. 6) interest in patients with pancreatic adenocarcinoma (work no. 4) and mCRC (works no. 5 to 6).

Indiana University-Purdue University Indianapolis (IUPUI) / Triple-negative breast cancers (TNBCs) are pathologically defined by the absence of estrogen, progesterone, and HER2 receptors. Compared to other breast cancers, TNBC has a relatively high mortality. In addition, TNBC patients are more likely to relapse in the first few years after treatment, and experiencing a shorter median time from recurrence to death. Detecting the presence of tumor in patients who are technically “disease-free” after neoadjuvant chemotherapy and surgery as early as possible might be able to predict recurrence of patients, and then provide timely intervention for additional therapy. To this end, I applied the analysis of “liquid biopsies” for early detection of minimal residual disease (MRD) on early-stage TNBC patients using next-generation sequencing. For the first part of this study, I focused on detecting circulating tumor DNA (ctDNA) from TNBC patients after neoadjuvant chemotherapy and surgery. First, patient-specific somatic mutations were identified by sequencing primary tumors. From these data, 82% of the patients had at least one TP53 mutation, followed by 16% of the patients having at least one PIK3CA mutation. Next, I sequenced matched plasma samples collected after surgery to identify ctDNA with the same mutations. I observed that by detecting corresponding ctDNA I was able to predict rapid recurrence, but not distant recurrence. To increase the sensitivity of MRD detection, in the second part I developed a strategy to co-detect ctDNA along with circulating tumor RNA (ctRNA). An advantage of ctRNA is its active release into the circulation from living cancer cells. Preliminary data showed that more mutant molecules were identified after incorporating ctRNA with ctDNA detection in a metastatic breast cancer setting. A validation study in early-stage TNBC is in progress. In summary, my study suggests that co-detection of ctDNA and ctRNA could be a potential solution for the early detection of disease recurrence. / 2021-08-05

La caractérisation exaustive des variations de l'ADN peut aider à progresser dans de nombreux champs liés à la génomique du cancer. Le séquençage nouvelle génération (NGS en anglais pour Next Generation Sequencing) est actuellement la technique la plus efficace pour déterminer une séquence ADN, du aux faibles coûts et durées des expériences comparé à la méthode de séquençage traditionnelle de Sanger. Cependant, la détection de mutations à partir de données NGS reste encore un problème difficile, en particulier pour les mutations somatiques présentes en très faible abondance comme lorsque l'on essaye d'identifier des mutations sous-clonales d'une tumeur, des mutations dérivées de la tumeur dans l'ADN circulant libre, ou des mutations somatiques dans des tissus normaux. La difficulté principale est de précisement distinguer les vraies mutations des artefacts de séquençage du au fait qu'ils atteignent des niveaux similaires. Dans cette thèse nous avons étudié la nature systématique des erreurs dans les données NGS afin de proposer des méthodologies efficaces capables d'identifier des mutations potentiellement en faible abondance. Dans un premier chapitre, nous decrivons needlestack, un nouvel outil d'appel de variants basé sur la modélisation des erreurs systématiques sur plusieurs échantillons pour extraire des mutations candidates. Dans un deuxième chapitre, nous proposons deux méthodes de filtrage des variants basées sur des résumés statistiques et sur de l'apprentissage automatique, dans le but de d'améliorer la précision de la détection des mutations par l'identification des erreurs non-systématiques. Finalement, dans un dernier chapitre nous appliquons ces approches pour développer des biomarqueurs de détection précoce du cancer en utilisant l'ADN circulant tumoral / Comprehensive characterization of DNA variations can help to progress in multiple cancer genomics fields. Next Generation Sequencing (NGS) is currently the most efficient technique to determine a DNA sequence, due to low experiment cost and time compared to the traditional Sanger sequencing. Nevertheless, detection of mutations from NGS data is still a difficult problem, in particular for somatic mutations present in very low abundance like when trying to identify tumor subclonal mutations, tumor-derived mutations in cell free DNA, or somatic mutations from histological normal tissue. The main difficulty is to precisely distinguish between true mutations from sequencing artifacts as they reach similar levels. In this thesis we have studied the systematic nature of errors in NGS data to propose efficient methodologies in order to accurately identify mutations potentially in low proportion. In a first chapter, we describe needlestack, a new variant caller based on the modelling of systematic errors across multiple samples to extract candidate mutations. In a second chapter, we propose two post-calling variant filtering methods based on new summary statistics and on machine learning, with the aim of boosting the precision of mutation detection through the identification of non-systematic errors. Finally, in a last chapter we apply these approaches to develop cancer early detection biomarkers using circulating tumor DNA

L’ADN tumoral circulant (ADNtc) porte des altérations spécifiques de la tumeur des patients, qui sont détectables par un acte minimalement invasif. L’ADNtc représente donc un biomarqueur d’intérêt pour le suivi de l’évolution du cancer. Sa détection requière une technique hautement sensible et quantitative. Dans ce contexte, ce travail de thèse a porté sur la quantification et le suivi de l’ADNtc par PCR digitale en gouttelettes (PCRdg). Cet outil permet la détection d’altérations à l’échelle d’un ADN unique, offrant ainsi une sensibilité allant jusqu’à 0.001%. La détection de cet ADNtc a été réalisée par l’évaluation des biomarqueurs tels qu’une mutation spécifique de la tumeur, la fragmentation de l’ADNtc et l’hyperméthylation de séquences cibles. D’une part, nous avons observé que chez les patients atteints de cancer, l’ADN muté circulant est plus fragmenté que l’ADN non muté, et que cet ADN circulant de patients est globalement plus fragmenté que chez les sujets sains. D’autre part, une corrélation entre les pourcentages d’ADN muté et d’ADN hyperméthylé circulants a été observée au cours du suivi de patients. Ceci suggère la possibilité d’un suivi précis et quantitatif de l’ADNtc par l’évaluation de l’hyperméthylation en alternative à la détermination du statut mutationnel. Nous avons ensuite appliqué nos tests de détection de l’ADNtc dans le cadre de deux études cliniques. L’étude PLACOL, incluant 82 patients atteints de cancer colorectal métastatique, a permis de mettre en évidence deux facteurs pronostiques : un seuil de 0.1 ng/mL et la mesure de la pente de décroissance de la concentration en ADN muté ou hyperméthylé circulant. Dans la seconde étude, portant sur le mélanome métastatique dans le contexte d’une thérapie ciblée (vémurafenib), une corrélation inverse entre les concentrations d’ADNtc et de vémurafenib a été observée. Ces résultats suggèrent le potentiel clinique de l’ADNtc pour l’orientation thérapeutique des patients atteints de cancer avancé. / Circulating tumor DNA (ctDNA) carries tumor-specific alterations that are detectable by minimally invasive sampling. It represents a highly pertinent marker for cancer monitoring during patients’ follow-up. CtDNA detection requires a highly sensitive and quantitative technique. In this context, this project focused on ctDNA quantification and monitoring by picoliter-droplet digital PCR. Thanks to the compartmentalization in millions of picoliter droplets, this tool allowed the detection of single DNA molecule with a sensitivity reaching 0.001%. Testing of ctDNA was performed through the evaluation of different potential biomarkers: specific mutations, ctDNA fragmentation, and hypermethylation of target sequences. On one hand, we observed in cancer patients that ctDNA is more fragmented than wild-type DNA, and, globally more fragmented than circulating DNA in healthy individuals. On the other hand, a strong correlation between percentages of hypermethylated and mutated DNA was observed during the follow-up of patients. Such results suggest the feasibility to precisely and quantitatively monitor ctDNA by the evaluation of hypermethylation as an alternative to the determination of mutational status. We have applied such ctDNA detection strategies in the context of two clinical studies. The PLACOL study, enrolling 82 metastatic colorectal cancer patients, allowed to highlight two prognostic factors: a ctDNA concentration threshold of 0.1 ng / mL, and the evaluation of ctDNA decreasing slope. In the second study, ctDNA was monitored in 11 melanoma patients in the context of a targeted therapy (vemurafenib). An inverse correlation between the concentrations of vemurafenib and ctDNA was demonstrated. These results suggest the clinical relevancy of ctDNA in advanced cancer patients, for the optimization of therapeutic management.

Circulating tumor DNA (ctDNA) in peripheral blood of patients with metastatic colorectal cancer appears to be a promising molecular marker that provides various applications. ctDNA levels vary depending on the presence, alternatively on the volume of tumor mass within patient's body, which can be used primarily for early detection of disease progression or recurrence and moreover for evaluating radicality of surgical treatment, all within long-term postoperative follow-up of the patient. Due to minimal invasivity of ctDNA analysis from peripheral blood (so-called liquid biopsy), it is possible to perform it repeatedly at relatively short time intervals. On account of very low fraction of ctDNA in total cell-free DNA (cfDNA) ranging between units and hundreds of percent, the key factor is optimal methodology covering all steps from the isolation process to a sufficiently sensitive detection technology. In this thesis I focus on an optimization of isolation process and analysis of ctDNA obtained from tumor tissue and plasma of selected patients with metastatic colorectal cancer in connection with surgical radicality and correlation with a clinical status of the patients.

No description available.

Glioblastomas are the most common malignant brain tumors in adults. Despite resection of the tumor, combined radio- and chemotherapy and new-targeted therapy approaches, the average life expectancy is only 15 months with a 3-year survival rate of less than 5%. Invasive growth of tumor cells into the surrounding brain tissue complicates treatment and causes recurrence. Imaging techniques such as magnetic resonance imaging (MRI) are conventionally used for diagnosis and for monitoring after surgery in order to detect tumor lesions. However, next to tumor progression, contrast enhancement could also result from pseudoprogression or radiation necrosis. Similarly, next to therapy response, a reduction of contrast enhancement could also mimic a success in therapy (pseudoresponse). Furthermore, the use of targeted therapies requires the molecular detection of biomarkers, which is often performed on tissue biopsies. During therapy, resistance mechanisms to therapy may develop, which have a significant impact on the success of the therapy. Repeated biopsies are needed to distinguish tumor progress from therapy-associated tissue changes and simultaneously detect therapy resistance. Often they are not feasible due to the high risk for the patient. Currently, there are methods to improve monitoring in glioblastoma patients. Besides improved imaging techniques, liquid biopsies are a promising method to detect tumor progression. At the same time, liquid biopsies also allow molecular characterization of the lesions. A key advantage over tissue biopsies is that they are minimally invasive, making them well suited for longitudinal tumor monitoring. For clinical application of liquid biopsies in the form of blood and CSF, components such as circulating tumor cells, circulating tumor DNA, extracellular vesicles, or tumor-educated platelets are used, with the former two probably being the most common tumor markers at present. In this study, a highly sensitive next generation sequencing-based method was established to detect tumor mutations in plasma and CSF of glioblastoma patients. To this end, crucial sample preparation steps were initially optimized. Therefore, four isolation kits extracting cell-free DNA from plasma were compared and the best-performing kit was chosen for the analysis of patient material. Furthermore, a multi-gene next generation sequencing panel (cfDNA-GBM panel) was designed specifically for use in liquid biopsies from glioblastoma patients and tested on a reference standard for cell-free DNA with mutations of different allele frequencies. The newly designed cfDNA-GBM panel, which uses hybrid-capture based enrichment, was compared to a primer-extension based panel. Due to the more consistent coverage of target regions, superior sensitivity, and better clinical applicability due to its smaller size, the cfDNA-GBM panel was used for subsequent clinical investigations. In both enrichments, unique molecular barcodes were used to label all original fragments and thus identify and eliminate errors in subsequent data analysis that occur during amplification in library preparation This increases the sensitivity of the method. During the establishment of the methodology, the use of molecular barcodes clarified the limitation of sequence information due to the small input amount of cell-free DNA. For tumor tissue analysis, an already established larger panel was adapted and extended based on the results of a proof of concept study. After the establishment of the sensitive method, clinical samples of glioblastoma patients were analyzed. Isolation and analysis of cell-free DNA from plasma and CSF resulted in highly variable amounts of cell-free DNA with characteristic oligonucleosomal fragment lengths. Furthermore, no difference in the amount or fragmentation of cell-free DNA from pre- or post-surgery plasma was determined. To increase the amount of cell-free DNA for sequencing, the amount of blood samples was increased from 10 ml (5 patients) to 50 ml (9 patients). Detection of circulating tumor DNA in plasma and CSF from glioblastoma patients was performed using two strategies. In a first approach, somatic tumor mutations were detected in genomic DNA from tumor tissue, which were searched for in the sequence data of liquid biopsies in a second step. In this approach, an increased detection rate (in 56% of patients compared to 25% in 10 ml blood samples) of circulating tumor DNA was detected in blood samples with a volume of 50 ml. Almost all tumor mutations lying in the target region of the smaller cfDNA-GBM panel could be detected in CSF (92%) with similar allele frequencies to those detected in the tumor. In a second strategy, somatic variants were detected in plasma and CSF without prior knowledge of variants in tumor tissue. Interestingly, three variants were detected in the CSF of one patient, which occurred with an allele frequency over 5% in the CSF, while they were not detectable in the analyzed tumor tissue. These variants may be indicators for tumor heterogeneity that was not accessible by analyzing sections of tumor tissue. Finally, mutation-specific probes were designed for three tumor-somatic variants detected in the patients' plasma to validate them by digital PCR. A pair of probes for a mutation in PTEN was successfully established to robustly detect minimal allele frequencies down to 0.1% but could not validate the mutation in cell-free DNA from plasma of the respective patient. Overall, it was shown that analysis of tumor somatic mutations using the cfDNA-GBM sequencing panel designed and established in this study is limited in the analysis of cell-free DNA from plasma of patients with glioblastomas, whereas the detection of circulating tumor DNA from CSF is promising.