Ex vivo methods for predicting the response to immunotherapy in hepatocellular carcinoma patients.

Abstract

The present invention relates to an ex vivo method for predicting the response of a subject to immunotherapy for hepatocellular carcinoma, comprising: 1) determining the energy metabolic state of a liver tissue sample from the subject, wherein the sample optionally includes pathological tissue and non-pathological tissue; and 2) detecting the presence or absence of an energy shift consisting of reduced mitochondrial respiration in the pathological tissue compared to non-pathological tissue from the subject's liver tissue sample or a standard non-pathological liver sample, wherein if reduced mitochondrial respiration is present in the pathological tissue, the subject is less likely to respond to immunotherapy, and if not reduced mitochondrial respiration is present in the pathological tissue, the subject is more likely to respond to immunotherapy.

Description

TECHNICAL FIELD

[0001] The present invention relates to an ex vivo method for predicting the response of a subject to immunotherapeutic treatment of hepatocellular carcinoma.

[0002] Accordingly, the present invention finds utility in the medical field, more particularly in the fields of diagnosis, prognosis and prediction of treatment response.

[0003] In the following description, numbers in square brackets ([]) refer to the list of references placed at the end of the present text. BACKGROUND ART

[0004] Therapeutic options for cancer are expanding day by day. However, in order to work towards the goal of precision medicine, it is essential for oncologists to identify which treatment will actually be beneficial for each patient. Several targeted therapies already benefit from companion testing. However, detecting mutations in targeted oncogenes or labeling targets for immunotherapy by immunohistochemistry does not guarantee an objective response. Today, several cancers, including liver cancer, cannot benefit from any current tools for treatment selection decision-making.

[0005] Liver cancer is a major health problem, ranking as the second leading cause of cancer death worldwide, with more than 850,000 new cases each year. The number of cases has been increasing over the past 20 years in Western countries. Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer. HCC occurs frequently in cases of cirrhosis, but is rare in healthy livers. Although cirrhosis surveillance programs can detect the majority of early-stage HCC for which curative treatments (surgical resection, percutaneous destruction, liver transplantation) are effective, 50 to 70% of HCC cases are actually diagnosed at a late stage. Advanced HCC has a very poor prognosis and can only be treated with local therapies (transarterial chemoembolization) or palliative systemic therapies.

[0006] The role of liver biopsy in the diagnosis and management of HCC has always been questioned. Historically, the use of liver biopsy in HCC management has been limited due to the risk of complications such as tumor dissemination and bleeding, as well as the availability of effective non-invasive diagnostic imaging. However, it is now widely accepted that these risks are rare, manageable, and do not affect the course of the disease. Therefore, they should not be considered reasons to avoid liver biopsy. Conversely, the risks of insufficient sampling have also been highlighted by numerous studies. This is primarily due to the fact that HCC-related gene mutations or transcriptome profiles have shown no predictive value regarding currently proposed treatments. Indeed, liver tumor biopsy is undoubtedly undervalued and underutilized, especially considering that fixed biopsies contain valuable information that could reverse the current risk / benefit ratio.

[0007] Sorafenib was approved in most countries until recently as the standard first-line treatment for advanced HCC. While sorafenib initially demonstrated a survival benefit despite very low response rates and significant side effects, subsequent Phase III clinical trials failed to show a superior survival benefit in advanced HCC. Therefore, the market introduction of a new drug with proven efficacy was a revolution in palliative care for HCC. Different Phase III trials demonstrated the non-inferiority of lenvatinib compared to sorafenib in overall survival for patients with advanced HCC, along with an equivalent safety and tolerability profile, justifying its use in first-line treatment. Following clinical recommendations, patients may also receive second-line treatment including cabozantinib, ramucirumab, nivolumab, or pembrolizumab. In addition, following the initial promising results of immunotherapy (nivolumab, pembrolizumab), the combination of atezolizumab (anti-programmed death ligand 1 (PDL1) antibody) and bevacizumab (anti-vascular endothelial growth factor (VEGF) antibody) has recently shown dramatic benefits as a first-line treatment compared to standard care with sorafenib. This combination has indeed become the new standard first-line treatment (Gordan JD et al. ([1])). This expansion of treatment options represents a considerable improvement and has already shown an increase in median overall survival from 6 months to over 26 months in patients with better preserved liver function and advanced HCC. However, a treatment strategy to guide the selection of first-line treatment and / or treatment sequences remains undefined. Therefore, predicting responses to these different treatments and improving their efficiency is crucial.

[0008] Laser capture and mass spectral (MS) analysis have been previously combined to thoroughly analyze the proteomic profile of tumors (Henriet E et al. [2], Dourthe C et al. [3]). This methodology, suitable for both formalin-fixed paraffin-embedded tissue (FFPET) and very small amounts of material (diagnostic biopsy), has already enabled the identification of the biomarker ASS1, which is now used to diagnose sonic hedgehog hepatocellular adenomas associated with a high bleeding risk (Henriet E et al. [2], Sala M et al. [4]). Furthermore, it is now possible to perform proteomic profile matching in clinical routines using recently developed machine learning tools. In fact, by comparing the proteomic profile of a tumor with that of a reference database, it is possible to make a diagnosis or predict malignant transformation (Dourthe C et al. [3]).

[0009] However, no precision medical tools exist that enable prognosis prediction for each type of cancer. For example, aside from contraindications to treatment, there is no biological evidence that can be used to guide treatment selection in hepatocellular carcinoma (HCC). Atezolizumab / bevacizumab immunotherapy is presented as the first-line treatment due to its best objective response rate (20-30%). However, there are no prognostic tools for treating advanced hepatocellular carcinoma, and only 25% of patients have time to try secondary treatment if primary treatment fails. [Prior art documents] [Non-patent literature]

[0010] [Non-Patent Document 1] Gordan JD, Kennedy EB, Abou-Alfa GK, Beg MS, Brower ST, Gade TP, et al. Systemic Therapy for Advanced Hepatocellular Carcinoma: ASCO Guideline. JCO.2020;38:4317-4345 [Non-Patent Document 2] Henriet E, Abou Hammoud A, Dupuy JW, Dartigues B, Ezzoukry Z, Dugot-Senant N, et al. Argininosuccinate synthase 1 (ASS1): A marker of unclassified hepatocellular adenoma and high bleeding risk. Hepatology. 2017;66:2016-2028 [Non-Patent Document 3] Dourthe C, Julien C, Di Tommaso S, Dupuy J, Dugot-Senant N, Brochard A, et al. Proteomic profiling of hepatocellular adenomas paves the way to new diagnostic and prognostic approaches. Hepatology. 2021; hep. 31826 [Non-Patent Document 4] Sala M, Gonzales D, Leste-Lasserre T, Dugot-Senant N, Paradis V, Di Tommaso S, et al. ASS1 Overexpression: A Hallmark of Sonic Hedgehog Hepatocellular Adenomas; Recommendations for Clinical Practice. Hepatol Commun. 2020;4:809-824 [Overview of the project]

[0011] Therefore, there is a need for new tools to predict the response to treatment for hepatocellular carcinoma. This invention satisfies these and other needs.

[0012] After extensive research, the applicant has surprisingly demonstrated that enrichment of specific biological pathways, particularly energy metabolism, is a highly discriminative and robust tool for predicting the response of HCC to immunotherapy, especially atezolizumab / bevacizumab immunotherapy.

[0013] Therefore, the present invention is based on using the tumor's energy metabolic state as a novel tool for predicting the response to immunotherapy in HCC.

[0014] One of the main advantages of the present invention is that, when the tumor exhibits an objective response profile, it is possible to guide the patient to another potentially more appropriate treatment, such as a re-evaluation of a multi-kinase inhibitor (lenvatinib or sorafenib).

[0015] Another advantage of the present invention is that it can be performed using a diagnostic biopsy prepared before the procedure. Therefore, only a small amount of tissue is required for this analysis, and thus the sample can be reused without depleting it.

[0016] Another significant advantage of this process is that a complete analysis can be achieved within a week, thus still meeting the deadlines imposed by medical practice. Furthermore, once proteomics data is generated, it remains a resource that can be reused for further applications (e.g., prognosis prediction, prediction of responses to other treatments, and discovery of new pharmacological targets).

[0017] In addition, the present applicant has developed a response prediction approach, in particular a proteome profiling-based approach, for carrying out the prediction process of the present invention. Another advantage is that proteome profiling, combined with predictive signatures for response to treatment, allows the most appropriate treatment to be directly prescribed for each patient without loss of opportunity, thus extending the life expectancy of HCC.

[0018] Accordingly, in a first aspect, the present invention provides an ex vivo method for predicting the response to immunotherapy treatment of hepatocellular carcinoma in a subject, comprising: 1) a step of determining the energy metabolic state of a liver tissue sample from the subject, wherein the sample optionally comprises pathological tissue and non-pathological tissue; 2) a step of detecting the presence or absence of an energy shift consisting of reduced mitochondrial respiration in pathological tissue, as compared to non-pathological tissue from the subject's liver tissue sample or a standard non-pathological liver sample; and - when mitochondrial respiration is reduced in the pathological tissue, the subject has a low probability of responding to the immunotherapy treatment, - when mitochondrial respiration is not reduced in the pathological tissue, the subject has a high probability of responding to the immunotherapy treatment, the ex vivo method is provided. [Mode for Carrying Out the Invention]

[0019] "Hepatocellular carcinoma" or HCC, as used herein, refers to any HCC, regardless of its developmental stage. By way of example, it may be early-stage HCC, intermediate-stage HCC or advanced HCC, and resectable HCC or unresectable HCC. HCC may or may not have already been treated with primary therapy.

[0020] As used herein, a "subject" is a patient, animal, mammal, or human suffering from HCC.

[0021] According to the present invention, the immunotherapeutic treatment of HCC may be any treatment for HCC known per se in the art by activating the immune system of a subject. It may be a treatment that has already been approved in at least one country by a drug regulatory authority, or it may be a treatment that is unapproved and in the experimental stage. The treatment may be a single molecule or a combination of at least two molecules. The immunotherapeutic treatment may be a treatment with a monoclonal antibody. The monoclonal antibody for use in immunotherapy may be naked, i.e., non-conjugated, or conjugated, i.e., bound to a chemotherapeutic agent or radioactive particle. Furthermore, monoclonal antibodies for cancer immunotherapy can be designed to be bispecific, i.e., to recognize and bind to two different proteins. For example, the immunotherapeutic treatment may be selected from anti-PDL1 monotherapy such as nivolumab, atezolizumab and bevacizumab combination therapy, anti-PDL1 and anti-VEGF combination therapy, anti-PD1 and CTLA4 combination therapy, PDL1 and CTLA4 combination therapy, anti-VEGFR2 therapy such as ramucirumab therapy, as well as anti-PD1 therapy such as pembrolizumab therapy.

[0022] The liver tissue sample may be a physical tissue sample or an image of liver tissue from medical imaging. Accordingly, the liver tissue sample may be any sample obtainable by at least one technique selected from known techniques, for example macrodissection, laser microdissection of formalin-fixed paraffin-embedded tissue or frozen tissue, biopsy, in particular fresh biopsy, and positron emission tomography. The tissue sample may be of any surface area, as long as it contains pathological tissue and optionally non-pathological tissue.

[0023] As used herein, "pathological tissue" refers to liver tissue having HCC cancer cells.

[0024] As used herein, "non-pathological tissue" refers to liver tissue that does not have HCC cancer cells.

[0025] Advantageously, liver tissue samples from subjects contain both pathological and non-pathological tissue.

[0026] "Energy metabolism," as used herein, refers to all or part of the biological pathways involved in the process of producing energy (i.e., ATP) from nutrients in a liver tissue sample of interest. Energy metabolism generally involves the following pathways: mitochondrial respiratory pathways, including oxidative phosphorylation, the TCA cycle, and fatty acid beta-oxidation. The state of energy metabolism refers to the level of activation of all or part of these metabolic pathways. Therefore, the state of energy metabolism determined in the method of the present invention may include determining the mitochondrial respiratory pathway, including oxidative phosphorylation, the TCA cycle, and fatty acid beta-oxidation, or determining at least one pathway selected from the oxidative phosphorylation pathway, the TCA cycle pathway, and the fatty acid oxidation pathway. Advantageously, the state of energy metabolism includes the biological pathways of oxidative phosphorylation and the TCA cycle.

[0027] In step 1) of this method, the determination of the energy metabolic state may be performed by any method known to those skilled in the art that enables the measurement of metabolic activity within the tissue. This may be, for example, at least one method selected from proteomic profiling matching, immunohistochemistry, Western blotting, comprehensive or targeted mass spectrometry, reverse transcriptase PCR, and positron emission tomography (PET). In this regard, if the liver tissue sample is a physical tissue sample, proteomic profiling matching, immunohistochemistry, Western blotting, comprehensive or targeted mass spectrometry, and reverse transcriptase PCR may be used. If the liver tissue is an image of liver tissue from medical imaging, PET may be used. This may be a direct measure of the expression level of metabolic enzymes, or an indirect measure of the expression level of metabolic enzymes via a transcript level measure, or a medical imaging for measuring mitochondrial volume using histological techniques such as immunohistochemistry or immunofluorescence, or for measuring the energy metabolic state using labeled tracers.

[0028] In a preferred embodiment of the present invention, the energy metabolic state may be determined by examining the expression of proteins involved in energy metabolism. For example, the determination of the energy metabolic state may be carried out by determining the abundance of proteins involved in energy metabolism in the liver tissue sample from the subject. This may be done directly, i.e., by quantifying the proteins in the tissue, or indirectly, i.e., by measuring the mRNA of the proteins.

[0029] The determination of protein abundance can be carried out by any method known to the extent of the technology. This may be relative protein abundance or quantified protein abundance.

[0030] When determining relative protein abundances, any method known to the extent of the art may be used. For example, this could be a mass spectrometry-based analysis of the relative protein abundances of proteins involved in energy metabolism between pathological and non-pathological tissues. More precisely, this could be the relative abundance of each protein in non-pathological tissue to the corresponding protein in pathological tissue, or vice versa.

[0031] When quantifying protein abundance, this can be done by any method known to those skilled in the art, such as targeted quantification techniques. This may be, for example, liquid chromatography-tandem mass spectrometry (LC-MS / MS) in data-dependent acquisition (DDA) or data-independent acquisition (DIA) in exploratory mode or targeted approaches (e.g., multiple reaction monitoring (MRM), parallel reaction monitoring (PRM)).

[0032] The determination of the energy metabolic state is made in the histopathology of the liver tissue sample in question. If the liver tissue is also intended to include non-pathological tissue, the determination can also be made in the non-pathological tissue. Advantageously, in this case, the detection of energy shifts is made by comparing the pathological and non-pathological tissues in question.

[0033] In another embodiment, the comparison may be made between the pathological tissue in question and a standard non-pathological tissue sample of the liver specimen used as a reference, for example, an extract of normal liver tissue from one or a mixture of several patients. Alternatively, the comparison may be made with a predetermined reference energy metabolic state, which may be obtained from a non-pathological liver specimen from a third party, or from several non-pathological liver specimens to have statistical representativeness.

[0034] "Reduction of mitochondrial respiration" as used herein refers to a decrease in the activation of biological pathways involved in mitochondrial respiration, particularly those favoring the glycolytic pathway. While not intended to be bound by any theory, this corresponds to an energy shift, also known as the Warburg effect, in which pyruvate is converted to lactate by glycolysis at the expense of mitochondrial activity. The hypothesis is that even if antibody targets are well-expressed, if an energy shift occurs, tumor hypoxia, low pH, inhibitory metabolites, and poor nutrient availability can all prevent T cells from achieving optimal effector function, which would explain non-response to immunotherapy. The reduction can be any statistically significant reduction. Statistical analysis may be performed by any method known in the level of the art for similar purposes, depending on the method used and the dataset being tested. This could be, for example, the Wilcoxon-Mann-Whitney U test or a t-test. Those skilled in the art can, considering their knowledge of this art, select one or more of the statistical analyses.

[0035] In one embodiment of the present invention, the energy metabolic state refers to determining the mitochondrial respiratory pathway as defined above, and in addition, determining the glycolytic pathway, where the glycolytic pathway is assumed to be unmodified, i.e., stable, or increased in the case of a Warburg-type metabolic shift.

[0036] "Likely to be effective" to an immunotherapy treatment means, as used herein, a statistically significant level of tumor regression and / or reduction in alpha-fetoprotein (AFP) levels. This may be tumor regression according to RECIST criteria, for example, where tumor regression is considered significant at 30% tumor volume and 50% AFP. Alternatively, this may be a reduction in tumor regression and / or AFP levels of at least 5%, or at least 10%, or at least 20%, or at least 30%, or at least 40%, or at least 50%, or at least 60%, or at least 70%, or at least 80%, or at least 90%, or even 100%, or any number or range in between, in the first two months after initiation of treatment, compared to values ​​without immunotherapy treatment.

[0037] In this specification, "low likelihood of response" to immunotherapy treatment refers to other cases. This may be tumor progression according to the RECIST criteria, where 20% tumor progression is considered significant. Alternatively, it may be tumor progression comparable to that without treatment, for example, at least 5%, at least 10%, at least 20%, at least 30%, at least 40%, or more tumor progression within two months.

[0038] In another aspect, the present invention provides a method for selecting an immunotherapy treatment for HCC patients, comprising the steps of performing the ex vivo prediction method defined above and selecting patients in whom no energy shift is detected.

[0039] In another embodiment, the present invention is a method for treating an HCC patient, (i) A step of selecting patients for treatment in whom no energy shift is detected, (ii) the process of administering an effective dose of immunotherapy treatment to the patient This provides a method that includes [something].

[0040] The present invention is further illustrated by the following embodiments in connection with the accompanying drawings, but should not be construed as limiting. [Brief explanation of the drawing]

[0041] [Figure 1] Figure 1: This figure shows principal component analysis of proteome profiles indicating dysregulation of protein expression between tumor and non-tumor tissues (T / NT) in patients who showed objective response (ORuAB n=6) or disease progression (PuAB n=12) under the immunotherapy atezolizumab / bevacizumab. The dataset was reduced to 113 protein signatures that showed significantly different expression between ORuAB patients (gray) and PuAB patients (black). [Figure 2] Figure 2: This figure shows the bootstrap analysis of signatures that distinguish PuAB patients from ORuAB patients. 54 / 113 (48%) of the proteins were significant in over 80% of the resampling configurations. [Figure 3] Figure 3: This figure shows a gene set enrichment analysis obtained from a gene ontology database (cellular components) using 113 proteins that showed significant changes in expression between patients who showed objective response (ORuAB) and patients who experienced disease progression (PuAB) with atezolizumab / bevacizumab combination therapy. [Figure 4] Figure 4: This figure shows a gene set enrichment analysis obtained from the Ingenuity Pathways database (IPA Qiagen, canonical pathway) using 113 proteins that showed significant changes in expression between patients who showed objective response (ORuAB) and patients who showed disease progression (PuAB) under the combined treatment of atezolizumab / bevacizumab. [Figure 5] Figure 5: This figure shows the tumor-to-nontumor expression ratio (T / NT) of proteins involved in oxidative phosphorylation for patients who progressed to disease (PuAB) or showed an objective response (ORuAB) to atezolizumab / bevacizumab immunotherapy (the ratio is expressed on a logarithmic scale). [Figure 6]Figure 6: This figure shows the tumor-to-nontumor expression ratio (T / NT) of proteins involved in the TCA cycle for patients who progressed to disease (PuAB) or showed an objective response (ORuAB) to atezolizumab / bevacizumab immunotherapy (the ratio is expressed logarithmically). [Figure 7] Figure 7: This figure shows the tumor-to-nontumor expression ratio (T / NT) of glycolysis-related proteins for patients who progressed to disease (PuAB) or showed an objective response (ORuAB) to atezolizumab / bevacizumab immunotherapy (the ratio is expressed logarithmically). [Figure 8-1] Figure 8: Study on the effect of inhibiting mitochondrial respiration on immune cell infiltration into spheroids in hepatocellular carcinoma cells cultured in 3D in vitro. (Figure 8A) Relative basal OCR of Huh7 spheroids after 48 hours of IACS-010759 treatment, normalized to protein content and mean vehicle (mean basal OCR in vehicle: 219.58 pmol / min). (Figure 8B) Relative lactate production in Huh7 spheroid supernatant after 48 hours of IACS-010759 treatment, normalized to protein content and mean vehicle (mean lactate concentration in vehicle: 1.60 mM). (Figure 8C) Immune infiltration of PBMCs in Huh7 spheroids, assessed by flow cytometry, normalized to mean vehicle (mean percentage of live PBMCs relative to total live cells in vehicle: 32.13%). (Figure 8D) Immunoinfiltration of PBMCs in Huh7 spheroids, normalized to the mean vehicle and evaluated by CD45 staining (mean percentage of CD45-positive cells in the vehicle: 26.04%). [Figure 8-2] Same as above. [Figure 9]Figure 9: This figure shows the proteome profiles of pre-treatment biopsies that differentiate between cases that responded to sorafenib and cases that progressed. (Figure 9A) Principal component analysis of proteome profiles showing dysregulation of protein expression between tumor and non-tumor tissue (T / NT) in patients who showed objective response (n=9) or progression (n=15) with sorafenib. The dataset was reduced to signatures of 87 proteins whose expression levels differed significantly between patients with progression (orange) and patients with response (blue). (Figure 9B) This figure shows the proteome profiles of pre-treatment biopsies that differentiate between cases that responded to sorafenib and cases that progressed. Comparison of proteins whose expression differentiates the response (response or progression) to two treatments (atezolizumab / bevacizumab or sorafenib). [Examples]

[0042] Example 1: Prediction of response to immunotherapy in hepatocellular carcinoma using tumor energy metabolic state material and method patient Patients with advanced HCC treated with atezolizumab / bevacizumab were selected and divided into two groups based on their response to treatment. Response was determined according to radiological tumor assessment and / or alpha-fetoprotein (AFP) levels evaluated during the first two months after the initiation of treatment. In this example, objective response corresponded to >30% tumor regression and / or a >50% reduction in AFP levels. >20% tumor progression corresponded to no response. Prior to the initiation of treatment, each patient was provided with sufficient tumor and non-tumor tissue (>1 mm) for subsequent proteomic analysis by MS for inclusion in the study. 2 A biopsy was taken, including the following:

[0043] Laser capture To determine the histopathological examination and the area of ​​interest to be dissected, liver tissue sections from each patient were stained with hematoxylin-eosin-saffron. Next, under the supervision of a pathologist, tumor (T) livers and non-tumor (NT) livers with the same surface area and uniform tissue composition (no fibrosis) were selected. Pre-selected tumor and non-tumor tissues (1 mm) were then taken from adjacent serial hematoxylin-eosin stained sections. 2 The tissue was microdisected from 5 μm thick FFPE sections using a PALM Type 4 (Zeiss) laser microdisecter. The same tissue region was cut into serial sections and processed in the same manner to form three technical replicas.

[0044] Sample preparation for proteomic analysis Defixation and protein extraction from tissue sections, in-gel digestion of proteins, peptide extraction from the gel, and peptide sample preparation for LC-MS / MS analysis were performed as previously described (Sala M et al. ([4])).

[0045] Mass spectral analysis Online nanoLC-MS / MS analysis was performed using an Ultimate 3000 RSLC Nano-UPHLC system (Thermo Scientific, USA) connected to a Nanospray Q-Exactive hybrid quadrupole-orbitrap mass spectrometer or a Nanospray Orbitrap Fusion® Lumos® Tribrid® mass spectrometer (Thermo Scientific, USA), as previously described (Henriet E et al. ([2]), Dourthe C et al. ([3]), Sala M et al. ([4])).

[0046] Database search, MS result processing and quantification. For protein identification, the Mascot 2.5 algorithm of Proteome Discoverer 2.5 software (Thermo Fisher Scientific Inc.) was used in batch mode, by searching the UniProt Homo sapiens database (71793 entries, Reference Proteome Set, published December 2017) from the website: http: / / www.uniprot.org / . Up to two enzymatic cleavage errors were tolerated. Mass tolerances in MS and MS / MS were set to 10 ppm and 0.02 Da, respectively. Methionine oxidation and lysine acetylation were searched as dynamic modifications. Cysteine ​​carbamidamide methylation was searched as a fixed modification. Raw LC-MS / MS data were imported into Proline Web for feature detection, alignment, and quantification. Protein identification was accepted only if at least two specific peptides had a pretty rank of 1 and a false detection rate of <1.0% calculated using the Mascot "decoy" option. Label-free quantification of proteomics data using MS1 extracted ion chromatograms was performed using the parameters previously described (Henriet E et al. ([2]), Dourthe C et al. ([3]), Sala M et al. ([4])). Normalization was performed on the median ratio. Missing value estimation was applied to 5% background noise.

[0047] Bioinformatics analysis Protein expression changes between tumor and non-tumor tissue samples from each patient were analyzed by calculating the logarithmic ratio of the normalized median protein abundances for each pair of tumor / non-tumor tissue samples.

[0048] To identify potential biomarkers that distinguish objective response (ORuAB) and progression (PuAB) under sorafenib, Welch's t-test (without assuming equal population variances) was performed. The inventors obtained predictive signatures for 113 proteins by setting the adjusted p-value significance threshold to 0.05.

[0049] Using this dataset, principal component analysis was performed using FactoMineR and Factoextra, available in R. Bootstrap analysis was performed using R database packages (base package and stats package). To identify pathways that are significantly dysregulated between ORuAB and PuAB, the inventors performed gene set enrichment analysis on 113 protein signatures against the Gene Ontology (GO) database (cellular components) and the Ingenuity Pathways (IPA, Qiagen) database (canonical pathways).

[0050] result Using a tissue proteome profiling methodology, the inventors first compared relative protein expression levels between tumor and non-tumor tissues in each patient, thereby creating a proteome profile of HCC (Dourthe C et al. [3]). The analysis was performed on FFPE liver biopsies taken before initiating treatment with atezolizumab / bevacizumab combination therapy. The analytical strategy was designed to limit bias due to unavoidable variability. On the one hand, three technical replicates were designed to limit technical variability in quantification. In addition, by comparing the ratio of tumor and non-tumor protein expression for each patient, background noise caused by inter-individual variability was reduced, and thus greater statistical power was obtained. This methodology allowed for robust data even with relatively small cohorts (Henriet E et al. [2], Dourthe C et al. [3]).

[0051] The tumor's distinctly different proteomic profile depends on the response to the atezolizumab / bevacizumab combination. To identify proteomic profiles that can predict ORuAB and PuAB, the inventors applied an adjusted Wilcoxon t-test comparing T / NT ratios from patients in the ORuS and PuS groups. The inventors created signatures for 113 proteins, setting the significance threshold for the adjusted p-value to 0.05. Patients in the ORuAB and PuAB groups were sufficiently different in principal component analysis of protein expression differential profiles (Figure 1).

[0052] The robustness of this signature was verified using a bootstrap method involving 10,000 resamplings of the sample. In fact, the inventors demonstrated that 54 / 113 (48%) of the proteins were significantly accurate, >80% (Figure 2). This means that more than half of the identified response predictor signatures showed high reproducibility. However, this does not mean that the remaining proteins have no predictive value; rather, their dysregulated expression showed greater variability within the ORuAB and PuAB groups.

[0053] From a functional perspective, at least 70% of the proteins in the signature were related to mitochondrial function, as shown by gene set enrichment analysis performed on the gene ontology database (GO, cellular components) (Figure 3). Consistent with this, as shown by gene set enrichment analysis performed on the Ingenuity Pathways database (IPA, Qiagen, canonical pathway), the most enriched biological pathway in this signature was related to mitochondrial energy metabolism function (oxidative phosphorylation, TCA cycle, fatty acid beta-oxidation) (Figure 4).

[0054] Focusing on the T / NT ratio of each protein, the inventors found that patients whose disease progressed after immunotherapy (PuAB) had a significant and overall decrease in the expression of mitochondrial respiratory proteins in the tumor (oxidative phosphorylation (Figure 5), TCA cycle (Figure 6)), while patients who showed objective response (ORuAB) had a T / NT expression ratio > 1 and sometimes showed significant overexpression in the tumor. However, the expression of glycolytic proteins did not change significantly in PuAB or ORuAB, and there was no difference between the two patient groups (Figure 7).

[0055] Consideration The lack of predictive markers for treatment response in advanced HCC remains a significant challenge, despite numerous studies already conducted. These predictive markers could potentially alter how patients are treated, especially as new treatments become available. Currently, three distinct treatment options are available for first-line treatment of advanced HCC: sorafenib, lenvatinib, and more recently, the combination of atezolizumab / bevacizumab. It is clear that patients respond differently to these treatments. Other considerations include the adverse effects of these drugs and their substantial medical costs. Consequently, a focus on precision medicine is necessary to determine the optimal treatment for each individual patient. All the information needed for this response prediction strategy can realistically be found in fixed diagnostic biopsies.

[0056] By combining laser capture and MS, it has become possible to access the expression profiles of dysregulated proteins from fixed biological tissue samples that can be used in routine medical practice. In fact, one of the main advantages of the approach according to the present invention is its broad applicability, given that needle biopsies are used in routine medical practice, and the potential of this approach in predicting the response to treatment of HCC.

[0057] The results obtained by the inventors show a decrease in oxidative metabolism in tumors that did not respond to immunotherapy, which likely contributes to an increase in glycolytic flux without alteration of glycolytic protein expression.

[0058] To supply energy to cells, glucose enters the cell via transporters. It is then metabolized to pyruvate through a series of enzymatic reactions.

[0059] Under conditions where oxidative phosphorylation is effective, pyruvate enters the mitochondria, where it is converted to acetyl-CoA, broken down in the Krebs cycle, and then produces energy via the electron transport chain in the inner mitochondrial membrane.

[0060] When an energy shift (known as the Warburg effect) occurs, pyruvate is converted to lactate at the expense of mitochondrial activity and expelled from the cell by monocarboxylic acid transporters (MCTs). The secreted lactate then contributes to extracellular acidification (DePeaux K et al. [5]).

[0061] As already mentioned in reference (21), high glycolysis rates can lead to localized depletion of oxygen and glucose levels (hypoxic regions accompanied by high lactate levels).

[0062] In addition, increased tumor growth is often associated with dysregulation of the vascular system (damage and leakage of blood vessels), which reduces blood flow. Immunosuppression is established, and cells are unable to leave the tumor.

[0063] The strategy of checkpoint inhibitor-based anti-cancer immunotherapy requires a sustained effector T cell response to induce a lasting clinical response. T cells need to enter the tumor microenvironment and functionally survive. However, even if antibody targets are sufficiently expressed, if metabolic shifts occur, tumor hypoxia, low pH, inhibitory metabolites, and poor nutrient availability can all prevent T cells from achieving optimal effector function. This is likely what occurred in HCC patients who progressed under immunotherapy, as analyzed by the inventors. Differences in energy metabolic status differentiate between HCC patients who responded to immunotherapy and those who did not, making it possible to guide treatment solely by detecting metabolic shifts.

[0064] Example 2: In vitro functional verification in a 3D hepatocellular carcinoma model: Oxidative phosphorylation inhibition modulates immune cell tumor invasion. Exploratory proteomics data show downregulation of oxidatively phosphorylated proteins in the mitochondrial respiratory chain, which under physiological conditions produces energy via the TCA cycle. In tumor cells, a metabolic shift known as the Warburg effect occurs at the expense of the Oxphos pathway (oxidative phosphorylation), rapidly providing high energy for cancer growth and converting pyruvate to lactate via aerobic glycolysis, even in the presence of oxygen (Vander Heiden MG et al. [6]). Lactate is then effluxed from the cell by monocarboxylic acid transporters (MCTs), contributing to extracellular acidification and immunosuppression (Kao KC et al. [7]).

[0065] Here, the decrease in mitochondrial respiration is likely compensated for by an increase in glycolysis, which may be reflected in an increase in flux without a significant increase in glycolytic enzyme expression. We hypothesized that this metabolic reprogramming, which we propose as a predictor of response to atezolizumab / bevacizumab treatment, is the mechanism of resistance in HCC patients (DePeaux K et al. ([8])).

[0066] To test this hypothesis, the inventors developed a 3D model that mimics the metabolic shift (from oxidative metabolism to glycolysis) in tumor cells associated with increased lactate levels in the microenvironment that hinders immune cell infiltration into tumors. The metabolic shift identified in vivo was mimicked in a spheroid model of HCC cell lines (Huh7, Hep3b) using a specific inhibitor of mitochondrial respiratory chain complex I named IACS-010759. This highly potent inhibitor was selected (instead of other classical complex I inhibitors such as rotenone) because it has shown promising efficacy in animal models and is already used in clinical trials (Yap TA et al. ([9])). A detailed plan of the experiment is shown in Figure 8. IACS-010759 activity was evaluated using oxygen consumption rate (OCR) measurements in seahorses (Figure 8A). A dose of 5 nM reduced basal OCR by 50%, and this dose was selected for the following experiments.

[0067] Lactic acid production in the Huh7 spheroid supernatant was evaluated by high-speed ion chromatography and conductivity measurements. Treatment with IACS-010759 resulted in a dose-dependent increase in lactic acid production, with a 2x increase at 5 nM (Figure 8B).

[0068] Next, the inventors evaluated the effect of IACS-010759 treatment of HCC spheroids on immunoinfiltration by co-culturing Huh7 spheroids and PBMCs. Quantitative analysis of immunoinfiltration was performed by flow cytometry using the ratio of live PBMCs to whole live cells. IACS-010759 treatment resulted in a 60% reduction in immunoinfiltration in spheroids (Figure 8C). A control experiment was conducted to evaluate the effect of IACS-010759 on PBMC invasiveness, showing that pretreatment of PBMCs with IACS-010759 did not affect PBMC invasion in Huh7 spheroids.

[0069] The inventors confirmed these results by immunohistochemistry, which visualizes and quantifies immune cell infiltration using the immune marker CD45. Immune infiltration quantification was performed using the percentage of CD45-positive cells. Treatment with IACS-010759 resulted in a 50% reduction in immune infiltration in Huh7 spheroids (Figure 8D) and a 25% reduction in Hep3b spheroids. Notably, the inventors observed necrotic cores in Hep3b spheroids after IACS-010759 treatment. This suggests that the Hep3b cell line is more sensitive to IACS-010759 than the Huh7 cell line.

[0070] Example 3: The proteomics signatures predicting the response to treatment are completely different for atezolizumab / bevacizumab and sorafenib. The inventors wanted to evaluate the specificity of energy metabolism protein dysregulation in atezolizumab / bevacizumab-resistant patients treated with sorafenib. By identifying and quantifying 3153 proteins per patient (>2 specific peptides per protein), the inventors were able to extract 87 proteins that showed significant dysregulation between response and progression cases. By performing PCA on this significant dataset, the inventors were also able to differentiate patients based on their response to TKIs (tyrosine kinase inhibitors) (Figure 9A). The heatmap of this dataset clearly shows distinctly different proteomic profiles between patients with objective response and those with progression. Only four proteins were found common to these signatures, demonstrating the specificity of proteomic profiling (Figure 9B).

[0071] The inventors performed a biological pathway enrichment analysis using GSEA, but found no enrichment of energy metabolic pathways among the proteins that were significantly enriched by sorafenib treatment.

[0072] List of References 1. Gordan JD, Kennedy EB, Abou-Alfa GK, Beg MS, Brower ST, Gade TP, et al. Systemic Therapy for Advanced Hepatocellular Carcinoma: ASCO Guideline. JCO.2020;38:4317-4345. 2. Henriet E, Abou Hammoud A, Dupuy JW, Dartigues B, Ezzoukry Z, Dugot-Senant N, et al. Argininosuccinate synthase 1 (ASS1): A marker of unclassified hepatocellular adenoma and high bleeding risk. Hepatology. 2017;66:2016-2028. 3. Dourthe C, Julien C, Di Tommaso S, Dupuy J, Dugot-Senant N, Brochard A, et al. Proteomic profiling of hepatocellular adenomas paves the way to new diagnostic and prognostic approaches. Hepatology. 2021; hep. 31826. 4. Sala M, Gonzales D, Leste-Lasserre T, Dugot-Senant N, Paradis V, Di Tommaso S, et al. ASS1 Overexpression: A Hallmark of Sonic Hedgehog Hepatocellular Adenomas; Recommendations for Clinical Practice. Hepatol Commun. 2020;4:809-824. 5. DePeaux K, Delgoffe G. Metabolic barriers to cancer immunotherapy. Nat Rev Immunol. 2021 Dec;21(12):785-797. 6. Vander Heiden MG, Cantley LC, Thompson CB. Understanding the Warburg Effect: The Metabolic Requirements of Cell Proliferation. Science. 2009;324:1029-1033. 7. Kao KC, Vilbois S, Tsai CH, Ho PC. Metabolic communication in the tumor-immune microenvironment. Nat Cell Biol. 2022;24:1574-1583. 8. DePeaux K, Delgoffe GM. Metabolic barriers to cancer immunotherapy. Nat Rev Immunol. 2021;21:785-797. 9. Yap TA, Daver N, Mahendra M, Zhang J, Kamiya-Matsuoka C, Meric-Bernstam F, et al. Complex I inhibitor of oxidative phosphorylation in advanced solid tumors and acute myeloid leukemia: phase I trials. Nat Med. 2023;29:115-126.

Claims

1. An ex vivo method for predicting the response to immunotherapy treatment for hepatocellular carcinoma in a subject, 1) A step of determining the energy metabolic state of a liver tissue sample from the subject, wherein the sample includes pathological tissue and optionally non-pathological tissue. 2) A step of detecting the presence or absence of an energy shift consisting of a reduction in mitochondrial respiration in the pathological tissue by comparing it with the target liver tissue sample or a non-pathological tissue from a standard non-pathological liver sample, Includes, If mitochondrial respiration is reduced in the pathological tissue, the subject is less likely to respond to the immunotherapy treatment. If mitochondrial respiration is not reduced in the pathological tissue, the subject is likely to respond to the immunotherapy treatment. Ex vivo method.

2. The ex vivo method according to claim 1, wherein the energy metabolic state includes at least one pathway selected from the mitochondrial respiratory pathway, the oxidative phosphorylation pathway, the TCA cycle pathway, and the fatty acid oxidation pathway.

3. The ex vivo method according to any one of claims 1 to 2, wherein the energy metabolic state is determined by at least one method selected from proteome profiling matching, immunohistochemistry, Western blotting, comprehensive or targeted mass spectrometry, reverse transcriptase PCR, and positron emission tomography.

4. The ex vivo method according to any one of claims 1 to 3, wherein the step of determining the energy metabolic state is performed by directly determining the amount of protein involved in the energy metabolism in a liver tissue sample from the subject, or by indirectly determining it, for example by measuring the mRNA of the protein, or by measuring the amount of mitochondria.

5. The ex vivo method according to claim 4, wherein the step of determining the abundance of protein is performed by label-free proteome analysis of the relative abundance of the protein involved in energy metabolism between pathological tissue and non-pathological tissue.

6. The ex vivo method according to claim 5, wherein the protein abundance is the relative abundance of each protein in the non-pathological tissue to the corresponding protein in the pathological tissue.

7. The ex vivo method according to claim 4, wherein the step of determining the amount of protein present is performed by a targeted quantification technique.

8. The ex vivo method according to any one of claims 1 to 7, wherein the immunotherapy treatment is selected from anti-PDL1 monotherapy, combination therapy with atezolizumab and bevacizumab, combination therapy with anti-PDL1 and anti-VEGF, combination therapy with anti-PD1 and CTLA4, combination therapy with PDL1 and CTLA4, anti-VEGFR2 therapy such as ramucirumab therapy, and anti-PD1 therapy such as pembrolizumab therapy.

9. The ex vivo method according to any one of claims 1 to 8, wherein the immunotherapy treatment is as defined in claim 8, and the protein involved in energy metabolism is at least one of the proteins involved in mitochondrial proteins, the tricarboxylic acid cycle, oxidative metabolism, and fatty acid beta-oxidation pathways.

10. An ex vivo method according to any one of claims 1 to 9, comprising measuring proteins involved in glycolysis.

11. The ex vivo method according to any one of claims 1 to 10, wherein the hepatocellular carcinoma is resectable or unresectable.

12. The ex vivo method according to any one of claims 1 to 11, wherein the sample can be obtained by at least one technique selected from macrodissection, laser microdissection of formalin-fixed paraffin-embedded tissue or frozen tissue, fresh biopsy, and positron emission tomography.

Images