HAL-ENS-LYON
Not a member yet
104280 research outputs found
Sort by
Induction chemotherapy with a single anthracycline-containing cycle in younger adults with newly diagnosed AML – the french backbone intergroup (BIG)-1 study on behalf of the filo, ALFA, and SFGM-TC study groups
No full textInternational audienceIntroduction The BIG-1 multicenter study was a prospective trial designed for younger AML patients with multiple randomizations at each stage of treatment, including induction, post-induction and hematopoietic stem-cell transplantation. Results of the post-induction randomization (intermediate vs high dose cytarabine) have been already reported (Hunault M et al, NEJM Evid 2025). Here, we report the results of the first randomization that compared high dose idarubicin (IDA, 45 mg/m2 total dose) to high dose daunorubicin (DNR, 270 mg/m2 total dose) during the first and unique induction cycle containing anthracyclines. Methods Patients (pts) aged 18-60 years with newly diagnosed AML, untreated post-MDS AML, or t-AML were eligible if they had an ECOG performance status ≤3, normal cardiac, liver and renal functions, no active infection or neoplasia. Pts with APL, Ph+ AML, CBF AML, or post-MPN AML were not eligible. Pts were randomly assigned to receive either daunorubicin (90 mg/m², d1-3) or idarubicin (9 mg/m², d1-5), combined with cytarabine 200 mg/m² (d1-7). The protocol planned for a single cycle of anthracycline. All patients alive after this first cycle, including those not achieving CR/CRi, underwent a second randomization to compare post-induction IDAC vs HDAC. The primary endpoint was overall survival (OS). Evaluations of treatment effects were adjusted on ELN-2022 risk, post-induction IDAC vs HDAC randomization, and time-dependent HSCT in first CR/CRi. According to French's regulation, no racial or ethnic data were collected. Results 1,159 pts were included from 01/2015 to 05/2018, 578 in the DNR arm and 581 in the IDA arm. Baseline characteristics were well balanced between the two arms. Median age was 50y, 563 pts were female. According to ELN-2022, 308 (27%), 330 (28%), 464 (40%) and favorable, intermediate or adverse risk; 57 pts (5%) were non-classified. With a median follow-up of 5.6y, 5-year OS was 51.8% (95% CI, 47.5-55.9) in the DNR arm vs 50.3% (95% CI, 46.0-54.5) in the IDA arm (adjusted HR, 1.03 [95% CI, 0.87-1.22], p= 0.75). When evaluated in patient subgroups including ELN-2022 risk groups, post-induction IDAC vs HDAC, and allo-HSCT in first remission, no significant interactions with the DNR vs IDA treatment effect were observed for OS. 5y-estimates of EFS (38 vs 38%), RFS (41 vs 44%), and cumulative incidence of relapse (43 vs 41%) did not differ between the two arms. Following induction, there was no difference in remission rate (CR+ CRi, 72.0 vs 73.7%) or early death (2.8 vs 2.8%) in the DNR and IDA arms, respectively. After salvage chemotherapy, the rates of CR/CRi, persistent AML, and early death were 85.6 vs 81.6%, 10.6 vs 13.9%, and 3.8 vs 4.5% in the DNR and IDA arms, respectively. Post-induction CR/CRi (DNR vs IDA) by ELN-2022 groups were: 96 vs 95%, 73 vs 78%, 55 vs 56% in favorable, intermediate and adverse groups respectively, whereas post-salvage CR/CRi were 97 vs 98%, 87 vs 84% and 76 vs 68%. 5y-OS by ELN-2022 groups was 73 vs 74%, 60 vs 58% and 38 vs 38% in favorable, intermediate and adverse groups, respectively. The severity of chemotherapy-induced myelosuppression and the incidences of adverse events were lower after DNR with shorter durations of thrombocytopenia and neutropenia, lower needs for RBC transfusions, number of days on antibiotics, and frequency of fungal infections. The rate of allo-HSCT in first CR/CRi was 43% in the DNR arm and 42% and the IDA arm. The BIG-1 trial enrolled more pts in further phase 2 studies planned in the protocol. Since there was no difference in efficacy and treatment-related mortality between DNR and IDA, we pooled the two arms to build a cohort of 2023 pts included between 01/2015 and 06/2021. To determine the crude contribution of first-line chemotherapy to OS in selected subgroups, we computed the salvage-free, transplantation-free survival (STFS) as the time between the date of inclusion until treatment failure, salvage treatment, morphological or molecular relapse, allo-HSCT or death. 5y-STFS was 52%, 31%, 15%, 11% and 2% in pts with CEBPA-bZIP, NPM1, IDH2, IDH1 mutations or KMT2A rearrangement (except KMT2A-MLLT3), respectively. Conclusions The first two randomizations of the BIG-1 study allowed us to establish a simplified treatment regimen consisting of a single cycle of daunorubicin 90 for induction and IDAC for consolidation. We now consider this regimen as a standard backbone on which new molecules can be added to improve results
Les galets et les graviers sous surveillance : comprendre le transport des sédiments grossiers pour améliorer l'état écologique des fleuves
No full textNational audienc
Searching for substellar companion candidates with Gaia. III. Search for companions to members of young associations
No full textContext. Absolute astrometry with Gaia is expected to detect and characterize the orbits of thousands of exoplanets in the coming years. A tool, GaiaPMEX, was recently developed to characterize multiple systems based on two binarity indicators derived from the DR3 astrometric solution: the astrometric signature α ruwe in linear motion residuals in Gaia-only data, and, when the sources were also observed by Hipparcos, the astrometric signature α PMa in the Gaia-Hipparcos proper motion anomaly (PMa). Aims. Our aim is to identify close (less than typ. 20 au) (sub)stellar companion candidates to star members of close-by young associations previously monitored in radial velocity (RV) surveys. We wish to compare the detection capabilities of absolute astrometry and spectroscopy, and to characterize planetary-mass companions, combining the astrometric data with direct imaging and RV data. Methods. We use GaiaPMEX to identify binary stars members of close young associations and constrain the mass and semi-major axes (sma) of possible companions. For companion masses possibly in the planetary range, we use direct imaging and when possible, RV data as well, to further constrain their nature and orbital properties. Results. For each of our targets, we provide a diagnosis on its binarity based on absolute astrometry. When no binary is detected, GaiaPMEX provides detection limits in the (sma, mass) space. We identify several companions with possible masses down to the brown dwarfs (BD; 50+) or planetary masses (13). Around the M-type star G80-21, we detected a new companion orbiting at less than 1-2 au. Adding RV and high contrast imaging data shows this companion is a giant planet. In other two cases, AB Pic and HD 14082 B, we confirm the presence of substellar companions, and determine the first robust solutions for their mass and orbital properties. We further identified 9 potentially interesting candidates for planetary mass companions, which remain to be studied. Finally, a detailed treatment of noises in Gaia astrometric measurements shows that there are no evidence at a 2-σ level of two exoplanet detections that were previously announced based on the same set of data. Conclusions. Our approach allows to detect all stellar mass companions with sma in the range 0.1-10 au. For separations below 0.1 au, however, spectroscopy outperforms absolute astrometry. Combining GaiaPMEX and RV data is therefore perfectly adapted for a full exploration of the 0.01-10 au sma range when searching for stellar companions, and increases the expected rate of detections derived from RV surveys. Moreover, in the 0.5 to 5 au domain, GaiaPMEX has an excellent sensitivity to BDs, and a good sensitivity to planetary mass planers as well for this sample
Equivariant Deep Equilibrium Models for Imaging Inverse Problems
No full textEquivariant imaging (EI) enables training signal reconstruction models without requiring ground truth data by leveraging signal symmetries. Deep equilibrium models (DEQs) are a powerful class of neural networks where the output is a fixed point of a learned operator. However, training DEQs with complex EI losses requires implicit differentiation through fixed-point computations, whose implementation can be challenging. We show that backpropagation can be implemented modularly, simplifying training. Experiments demonstrate that DEQs trained with implicit differentiation outperform those trained with Jacobian-free backpropagation and other baseline methods. Additionally, we find evidence that EI-trained DEQs approximate the proximal map of an invariant prior
Efficient Algorithms for Maximal Matroid Degenerations and Irreducible Decompositions of Circuit Varieties
No full textInternational audienceMatroid theory provides a unifying framework for studying dependence across combinatorics, geometry, and applications ranging from rigidity to statistics. In this work, we study circuit varieties of matroids, defined by their minimal dependencies, which play a central role in modeling determinantal varieties, rigidity problems, and conditional independence relations. We introduce an efficient computational strategy for decomposing the circuit variety of a given matroid , based on an algorithm that identifies its maximal degenerations. These degenerations correspond to the largest matroids lying below in the weak order. Our framework yields explicit and computable decompositions of circuit varieties that were previously out of reach for symbolic or numerical algebra systems. We apply our strategy to several classical configurations, including the Vámos matroid, the unique Steiner quadruple system , projective and affine planes, the dual of the Fano matroid, and the dual of the graphic matroid of . In each case, we successfully compute the minimal irreducible decomposition of their circuit varieties
Probabilistic well-posedness of dispersive PDEs beyond variance blowup I: Benjamin-Bona-Mahony equation
No full text45 page
Molecular switches based on boron complexes for the modulation of chiroptical properties
No full textNational audienc
Gérard de Nerval, Aurélia
No full textInternational audienceOuvrage de préparation aux concours des khâgnes. Programme : Jean de Léry, *Histoire d'un voyage faict en la terre du Brésil*, Madame d'Aulnoy, *Contes des Fées*, Gérard de Nerval, *Aurélia*, Louis Aragon, *Le Paysan de Paris*. Axes : "Le merveilleux", "L'oeuvre littéraire et le lecteur", "La représentation littéraire"
Variable Elimination as Rewriting in a Linear Lambda Calculus
No full textInternational audienceAbstract Variable Elimination (VE ) is a classical exact inference algorithm for probabilistic graphical models such as Bayesian Networks, computing the marginal distribution of a subset of the random variables in the model. Our goal is to understand Variable Elimination as an algorithm acting on programs in an idealized probabilistic functional language—a linear simply-typed λ-calculus suffices for our purpose. Precisely, we express VE as a term rewriting process , which transforms a global definition of a variable into a local definition, by swapping and nesting let-in expressions. We exploit in an essential way linear types