Section: New Results

Translational vaccinology

HIV vaccine development

We have finalized the data science analyses of two HIV vaccine clinical trials: 1) ANRS VRI01, a randomized phase I/II trial evaluating for different prime boost vaccine strategies in healthy volunteers; 2) ANRS 149 LIGHT, a randomized phase II trial comparing a prime-boost therapeutic HIV vaccine strategy to placebo in HIV-infected patients undergoing antiretroviral treatment interruption. This included integrative statistical analyses using sPLS methods (as developed by the team) to relate markers from different high-dimensional immunogenicity or gene expression assays or virological assays to each other. In the ANRS VRI01 data set this allowed to disentangle the immune responses induced by the different vaccines used in the prime-boost strategies, showing specific effects of one of the vaccines (MVA HIV-B). We further identified a gene expression signature that correlates with later functional T-cell responses across the three different prime-boost association in which the MVA HIV-B vaccine was used. The corresponding manuscripts are currently in preparation.

Other HIV vaccine trials are currently being set-up by the French VRI (Vaccine Research Institute) and the European consortium EHVA, with strong contributions of SISTM team members to the trial designs.

Ebola vaccine development

The main results of the two randomized phase II Ebola vaccine trials conducted by the IMI-2 EBOVAC2 consortium (coordinated by Rodophe Thiébaut from the SISTM team) were finalized in 2019. and presented at international conferences. The results showed that the tested vaccine strategy (two-dose heterologous Ad26.ZEBOV and MVA-BN^® -Filo Ebola vaccine regimen, developed by Janssen) was safe and immunogenic in both European and African volunteers (EBL2001 and EBL2002 trials). Deeper analyses of the induced immune responses are currently ongoing, and systems vaccinology analyses will soon start in the SISTM team.

The SISTM team is also a partner in the related IMI-2 EBOVAC1 and EBOVAC3 consortia (assesing the same vaccine regimen in other trial populations and/or trial phases), in which the main contributions of the team are related to mechanistic modeling of the immune responses (ongoing).

Vaccine development against other pathogens

Two other phase I vaccine trials (one testing a placental malaria vaccine, Primalvac trial; and one testing a nasal Pertussis vaccine, BPZE-1 trial), in which members of the SISTM team were strongly involved, have shown promising results. Results of the Primalvac trial have been accepted for publication in the Lancet Infectious Disease journal, and results of the BPZE-1 trial have been submitted for publication.

Methodological developments for vaccine trials

At the interface between the axis on "Mechanistic learning" and the axis "Translational vaccinology", modelling done within a PhD project (M. Alexandre, supervised by R. Thiébaut and M. Prague) has informed the definition of the primary endpoint and statistical analysis method to be used in two therapeutic HIV vaccine trials with antiretroviral treatment interruption (EHVA T02 and ANRS DALIA-2). The methodological choices and their rationale have been presented to the governance bodies of the research consortia, patient associations and have been submitted for ethics and regulatory approvals. This will be subject to a specific methodology publication.

Edouard Lhomme (PhD student in the axis "Translational Vaccinology, supervised by L. Richert) has developed a statistical method for functional T-cell assay data from vaccine trials (in particular the intracellular cytokine staining assay) that takes into account non-specific immune responses. We propose using a bivariate linear model for the analysis of the cellular immune responses to obtain accurate estimations of the vaccine effect. We benchmarked the performance of the model in terms of both bias and control of type-I and -II errors, and applied it to simulated data as well as real pre- and post-vaccination data from two recent HIV vaccine trials (ANRS VRI01 and ANRS 149 LIGHT in HIV-infected participants). This method has been published in the Journal of Immunological Methods and is now used in the SISTM team as the standard method for analyses of functional cellular data with non-stimulated control conditions, for instance in the currently ongoing analysis of cellular proliferation data from the EBOVAC2 EBL2001 trial. We have also established an online interface based on R Shiny to make this analysis method available for use by immunologists without specific training in statistical modelling (https://shiny-vici.apps.math.cnrs.fr/).

Prediction of the survival of patients based on RNA-seq data

In collaboration with the Inria MONC team, with the Inserm Angiogenesis and Tumor micro-environment team, and with clinicians from Milan and Bergen, the project GLIOMA-PRD aims to improve the prediction of the evolution of the lower grade glioma, a primary brain tumor, based on clinical, imaging and genomic data. In the SISTM team, we first had to determine the sufficient sample size for determining a predictive signature based on RNA-seq data for the survival of the patients. We concluded that 50 patients were a good enough sample size for this aim. We then explored the potential methods to analyze these data, with a particular focus on the methods grouping the genes by pathways, as the pilot data (The Cancer Genome Atlas Research Network, NEJM, 2015) showed a high correlation structure. We particularly compared two methods, Generalized Berk-Jones (GBJ), proposed by [54], and tcgsaseq, proposed by [1]. The first method could be applied to the survival context and thus be appropriate for our data.

Once the RNA-seq data were available, we could observe a high batch effect since the data were sequenced in two different lanes. One of these two batch included only patients that exhibited a particular tumor at the PET-scan, called "COLD", while the second batch included both patient labelled as "COLD", but also the other type of tumor, called "DIFFUSE". As this experiment structure might leads to confusing the difference due to the batch effect with the one due to the biological different, we had to explore methodologies that could remove this batch effect.

Once this batch effect removed, we will then analyze the RNA-seq data in order to identify the genes, or the group of genes, that could be predictive of the survival of the patients.