-    Problem, Context and Scientific Objectives / Societal Issues

This project targets organ deficiency, by proposing to go back upstream of the donor-organ-recipient route by developing quantitative and reproducible solutions for donor selection.

Indeed, currently donors and their organs are evaluated qualitatively by team members, and the resulting variability means that many organs are not used, although it is established that at least 2/3 of them could have been successfully transplanted . At the societal level, this implies that each year more than 200 patients waiting for a kidney transplant are not transplanted, and for patients waiting for hearts, lungs and livers, for which no replacement technology exists, there are 100, 400 and 200 organs which are respectively not used, with a dramatic impact for the patient.

Thus, our objectives are todetermination and validation of new biomarkers as well as the formulation of an algorithm for determining the quality of grafts using machine learning.

-    Positioning in relation to the state of the art in the field and justification of relevance

Kidney transplantation presents major challenges, particularly regarding the shortage of organs. Demand for transplants far exceeds supply (4,354 transplants for 10,990 patients waiting in 2022) .Faced with this shortage, donations must be optimized, but this will remain impossible as long as the capacity to quantify the quality of an organ, to deploy optimized care for the organ, is not established.

Indeed, at present the only donor qualification score is the Kidney Donor Profile Index (KDPI), whose performance is moderate in its initial cohort (c statistics of 0.6) and unreliable in a European cohort. , making the discovery of new biomarkers essential. In order to validate these in an easy-to-use algorithm, the use of omics tools, such as transcriptomics and metabolomics, coupled with data analysis in machine learning, is essential.

-    Relevance of the methodological approach in the scientific context

The biological complexity of brain death, coupled with the multifactorial aspect of ischemia-reperfusion that the organ undergoes between harvest and transplantation, requires the use of wide-field (open ended) omics analyses. We propose the use of Metabolomics, the large-scale study of small molecules, commonly called metabolites. This approach is powerful because the metabolites and their concentrations directly reflect the biochemical activity and therefore the molecular phenotype. These molecules are easily measured with the mass spectrometers fitted to hospital laboratories, so their use in the clinic is easily implemented.

    Methodology:

- Patient cohort: The collection of biological samples (blood and urine) from organ donor patients who died of brain death began in 2017 and now includes 110 sets of samples. This cohort is supplemented here by a biobank from the Nantes University Hospital, comprising 70 donors. To date, it will be the largest organ donor biobank available for biomarker research.

-Metabolomics: a plasma extract is injected into the liquid chromatography - high resolution mass spectroscopy (LC-HRMS) system, according to 4 successive programs, combining two column technologies (C18 and HILIC) as well as two types of ionization ( negative or positive). The acquisition is carried out on an Orbitrap Exploris 120 (Thermo Fisher). This protocol is the result of extensive optimization and has been shown to produce the best quality and quantity of data.

- Analysis of the signal by Artificial Intelligence:

The goal will be to produce an algorithm that can estimate the quality of the organ, which we equate to the function of the organ once transplanted in the short and long term. We will build a machine learning model that will predict the quality of the organ based on data from omics tools.

The need to use multiple criteria for evaluating organ quality comes from its multifactorial aspect. Exploratory Factor Analysis makes it possible to measure latent variables such as the quality of the organ, which would jointly vary all of the evaluation criteria. We will be able to integrate this type of analysis into our machine learning models in order to improve the quality of its prediction. Finally, we will validate these new biomarkers on the basis of their predictive quality.

- Preliminary data:We carried out a preliminary metabolomics study on a sub-cohort of donors generating hundreds of signals that we explored using machine learning. Our results show the good performance (aitre under the ROC curve of 0.75) of an algorithm predicting organ function 3 months after transplantation, demonstrating that there is real potential for omics.

-    Share of innovation and socio-economic impact:

Innovation: As our preliminary data show, metabolomics and transcriptomics can identify molecules that provide information on organ quality. These targets therefore offer new opportunities for the development of therapeutics and biomarkers. The open nature of our analyzes will allow us to isolate previously unidentified disease factors, which will be analyzed with ontology tools in order to unveil new pathophysiological pathways and networks, as we have shown previously. .

Socio-economic impact: an algorithm for determining the quality of the organ will allow the healthcare team to implement more advanced preservation procedures, or even ex-vivo repair, to better match the organ and the recipient, to anticipate post-transplant complications... all of these elements substantially increase the patient's quality of healthy life, in total alignment with the goals of health and aging well. These improvements will also reduce healthcare costs.

-    Thesis co-financing information:

We requestes a ½ thesis grant from the Nouvelle Aquitaine region, under the co-direction of Raphael Thuillier (metabolomics, machine learning), Thomas Kerforne (resuscitation, transplantation) and Carlos Prieto (machine learning), at the U1313 IRMETIST laboratory (director: Luc Pellerin) and the Universidad de Salamanca, Servicio de Bioinformática. We have already secured the first half of the grant with the university of Poitiers.

-    Integration of the “open science” component into the project

The results of this project will initially remain confidential in order to allow evaluation, in collaboration with the University of Poitiers, of their patentability. Once this security has been achieved, our approach will be entirely in accordance with the 'open science' policy: on the one hand, the publications resulting from this work will, once published in a journal of the specialty, be put online on HAL, and on the other hand the databases as well as the source codes of the algorithm for determining the quality of the organs will be put online on public platforms (under the control of the EPST). In addition, the doctoral student will be encouraged to participate in popularization initiatives aimed at the general public, such as the 'my thesis in 180 seconds' event to which U1313 students are accustomed.

Funding category: Financement public/privé

En cours d'acquisition

PHD title: Doctorat de Biologie

PHD Country: France