WO2019115679A1 - A signature to assess prognosis and therapeutic regimen in liver cancer - Google Patents

Abstract

The present invention concerns a method or process of quantitatively and qualitatively analyze at least one of the biomarkers disclosed herein or any combination thereof in a sample previously obtained from a patient diagnosed for a liver tumor, so that to determine the prognosis of said liver tumor in a patient, thus proposing in combination or not with the clinical stratification, the most appropriate treatment and adopt the most appropriate therapeutical strategy.

Description

A SIGNATURE TO ASSESS PROGNOSIS AND THERAPEUTIC

REGIMEN IN LIVER CANCER

TECHNICAL FIELD OF THE INVENTION

The present disclosure generally relates to biomarkers and their use for the prognosis of liver cancer.

BACKGROUND OF THE INVENTION

Liver cancer, also known as hepatic cancer and primary hepatic cancer, is a cancer that starts in the liver. The most frequent is the Hepatocellular Carcinoma (HCC) which is the sixth most commonly occurring cancer in the world, the third leading cause of cancer mortality and it is mainly diagnosed in adults with an underlying liver disease. In contrast to adult liver tumors in which the predominant form is HCC, the main liver tumor in children is Hepatoblastoma (HB), accounting for two thirds of the total. Pediatric HCC (pHCC) is even rarer than HB and is usually diagnosed in older patients and adolescents. The annual incidence of HB is of 1.5 cases/million children under 15 years. HB is typically diagnosed during lactation or in young children (88% of cases occur in children under 5 years). pHCC is the second most common liver tumor in children being most of the cases diagnosed after 10 years of age and it is the main hepatic tumor in adolescents. In contrast to adult HCC, most cases in children are de novo tumors, not related to liver damage.

It is important for clinicians and physicians to establish a classification of primary liver cancers (HCC and HB) to propose the most appropriate treatment. In that regard, prognosis stratification for the above mentioned cancers and consequent treatment options are difficult to predict and they are mainly based to clinical (patient age, general status of the patient, liver function, portal pressure, serum bilirubin and alpha-fetoprotein levels), surgical (tumor resectability, potential candidate for liver transplantation) and tumor (tumor size, tumor hepatic extension, tumor rupture, blood tumor invasion or spread to other places in the body) parameters (Bruix et al, Gastroenterology, 2016; Meyers et al, Lancet Oncology, 2017). Some factors are currently used (satellite presence, differentiation degree, histological cancer subtypes, tumor markers) but have been found to not be accurate and sufficient enough to ensure a correct classification. In contrast to other cancers, no molecular markers have been included so far to improve clinical management of the patient with liver cancer.

As far as the HB is concerned, the current risk stratification system at diagnosis is based on the Children’s Hepatic Tumors International Collaboration (CHIC) classification. This system is mainly based in the PRETEXT (pre-treatment extent of disease) system and its combination with different clinical parameters such as presence of metastasis, patient age, serum AFP levels, tumor resectability and tumor spreading (i.e. intra-growth blood vessels, multifocality, tumor rupture, contiguous extrahepatic growth, caudate and lymph node involvement, metastasis). The PRETEXT (pre-treatment extent of disease) system designed by the International Childhood Liver Tumor Strategy Group (SIOPEL) is a non-invasive technique commonly used by clinicians, to assess the extent of liver cancer at diagnosis, to determine the time of surgery and to adapt the treatment protocol. This system is based on the division of the liver in four parts and the determination of the number of liver sections that are free of tumor. However, the PRETEXT system, even if reproducible and providing good prognostic value, is mainly based on imaging criteria, making this system dependent upon the technicians and clinicians. There is thus a need for a system, complementary to the current clinical CHIC stratification, based on genetic and molecular features of the liver tumors to improve the patient prognosis prediction. In addition to the latter, to our knowledge no patient stratification has been established after chemotherapeutical and surgical treatment. Usually, the children with liver tumors are treated by chemotherapy to reduce the chances of tumor recurrence but despite this measure, 20% of the patients have poor outcome due to the development of tumor metastasis resistant to chemotherapy. In case of adult HCC, the tumor recurrence rates after surgery are higher than 70%. A patient stratification based on molecular basis is thus needed to identify those patients with poor prognosis that includes those at high-risk of tumor recurrence to intensify its surveillance. At the same time, biomarkers are needed to reduce the dose of chemotherapy of children with tumors with good prognosis at molecular level. The main finding relating to molecular classification is the identification of two HB subclasses, called C1 and C2, which have been described associated to different prognosis of liver cancer (Cairo-Armengol et al, Cancer Cell, 2008) but the 16-gene signature able to assess these two tumor subclasses is difficult to implement in the clinical setting.

The present invention is thus focused to provide a new biomarker, called the 3- protein signature, easy to assess at the Pathology service of any hospital with the intention to improve the treatment, preferably at diagnosis or after surgery, of any patient diagnosed with HB, pHCC or HCC with the final aim to improve quality of life and survival rates of these patients.

BRIEF DESCRIPTION OF THE FIGURES

These and/or other aspects and advantages of the invention will become apparent and more readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings of which:

Fig. 1. Hierarchical cluster analysis of the proteomic data clearly distinguished the two HB molecular prognostic subtypes, C1 and C2. (A) Representative unsupervised hierarchical clustering (Pearson distance; Linkage method: Complete of protein expression profiles obtained from 277 Two-dimensional differential in-gel electrophoresis (DIGE) spots corresponding to the top 50% highest coefficient of variation (CV) spots of 16 HBs and 8 non-tumor samples. (B) Representative unsupervised hierarchical clustering of protein expression profiles obtained from 390 proteins (50% CV) identified by Label-Free (LF) mass spectrometry analysis of 8 HBs and 4 non-tumor samples. Abbreviations: T, tumor; NT, non-tumor (white squares); R, experimental replicate. Tumor samples were classified according to the 16-gene signature (Cairo et al., 2008) as C1 and C2 (green and red squares, respectively). Black boxes in the rows above the heat maps indicate (from top to bottom): dead of the disease, metastasis at diagnosis, and non-fetal main epithelial component.

Fig. 2. Table of the main deregulated proteins between the two C1 and C2 HB prognostic subclasses as well as C1/NL and/or C2/NL identified by DIGE and/or LF proteomic techniques (p value <0.01 , FC±2). In blue, proteins selected for Western Blot validation.

Fig. 3. Study of the protein expression of the 8 selected prognostic biomarkers by using the Western blot (WB) technique: Identification of the 3 prognostic biomarkers. (A) Protein expression of eight putative prognostic proteins assessed by Western blot in the discovery set of samples. NL, Normal liver (white boxes); HB, Hepatoblastoma (black circles). (B) Representative Western blot of 3 NL and 6 HB (3 C1 and 3 C2) of the 3 selected prognostic biomarkers. (C) Kaplan-Meier plot of Event Free Survival of training set patients classified according to the expression of the 3 prognostic biomarkers selected by Western Blot. The number of biomarkers altered in each patient (0, 1 , 2 or 3 BM means 0, 1 , 2 or 3 biomarkers altered) was strongly associated with the prognosis of HB patients and survival curves revealed the complementarity of the different biomarkers. The alteration of CKAP4 and C1 QBP was defined as overexpression of the biomarkers more than or equal than two-fold expression of the adjacent healthy liver. The alteration of CRYL1 was defined as under-expression of the protein in an amount less than the half of that determined for the adjacent healthy liver. BM, biomarker.

Fig. 4. Correlation between protein and mRNA expression of the three prognostic biomarkers, CKAP4, C1 QBP and CRYL1. Protein and gene expression were assessed by Western Blot and Human Transcriptome Array of Affymetrix, respectively. Green dots represented tumours of the C1 subtype and good prognosis; Red spots, represent tumours of the C2 subtype and poor prognosis.

Fig. 5. Definition of the 3-protein signature. (A) Method for determining the 3- protein signature. First, a calculation of a global score of the immunostaining for each marker taking into account the percentage of stained tumour cells as well as the intensity of the staining is needed. Second, the alteration of each biomarker should be determined taking into account a cut-off of the global staining score which was established by using non-tumour healthy liver staining as a reference. Thus, CKAP4 and C1 QBP immunostainings were considered as“altered” when their global staining score was >12 (2 times the NL maximum value) and CRYL1 when no expression was detected. When the biomarker is altered, a value of “1” is given whereas if the biomarker is not altered, a value f “0” is given. Finally, the 3-protein signature is obtained for a specific patient by adding the number of the 3 biomarkers altered in the tumour (3-protein score rank: 0-3). (B) Global score plots of the three different prognostic biomarkers in non-tumour (NT) and tumour (T) samples of the 144 patients of the test set. The cut-off chosen for each biomarker is represented as a red line. Taking into account the different cut-off values, 49, 48 and 11 % of the paediatric patients with liver cancer have C1 QBP, CKAP4 and CRYL1 altered in their tumours. (C) Representative images of the cytoplasmic immunostaining of the three biomarkers in Normal healthy adjacent liver (NL) and in hepatoblastoma tissues with normal and altered expression.

Fig. 6. Survival analysis of the 3-protein signature in the 144 childhood liver cancer patients. Kaplan-Meier plots of EFS (left) and OS (right) of the three biomarkers individually. Abbreviations, EFS, event-free survival; OS, overall survival.

Fig. 7. Impact of the 3-protein signature on the current clinical OH IC-HS stratification Top panel, Kaplan-Meier plots of Event Free Survival (right) and Overall survival (left) of 128 patients of the training set classified according to the expression of current clinical stratification system OH IC-HS (Meyers et al., 2017). Top middle panel, Kaplan-Meier plots of Event Free Survival (right) and Overall survival (left) of 20 patients of the training set clinically classified as low or very low risk and re-stratified according to the 3-protein signature. Bottom middle panel, Kaplan-Meier plots of Event Free Survival (right) and Overall survival (left) of 66 patients of the training set clinically classified as intermediate risk and re-stratified according to the 3-protein signature. Bottom, Kaplan-Meier plots of Event Free Survival (right) and Overall survival (left) of 32 patients of the training set clinically classified as high risk and re- stratified according to the 3-protein signature.

Fig. 8. The 3-protein signature survival analysis in non-treated diagnostic specimens. Kaplan-Meier plots of Event Free Survival (right) and Overall survival (left) of 42 patients of the training set from which non-treated specimens were studied, stratified according to a simplified 3-protein signature (“0 BM”, none of the 3 biomarkers was altered;“1 or 2 or 3”, at least 1 or 2 or 3 biomarker/s was altered). Fig. 9. Plasmatic levels of C1 QBP in paediatric control individuals (n=7,“healthy” individuals with minor health problems not related to cancer) and liver cancer patients (n=30, including 20 pediatric patients with HB (n=18) or pHCC (n=2) and 10 adult patients with HCC) assessed by ELISA. Two different groups of cancer patients were defined according to the presence of poor prognostic parameters such as increased patient age in childhood liver cancer patients, tumour multinodularity, presence of metastasis, advanced PRETEXT stage, immature histology, tumour recurrence or patient death during follow-up). “Good prognosis” means that none of the abovementioned poor prognostic parameters was present and “poor prognosis” means that at least one of the poor prognostic parameters was present.

Fig 10. Immunohistochemistry of the 3 protein markers in HCC specimens.

Fig. 11. Kaplan-Meier plots of Event free survival of 30 adult patients with HCC treated by surgical resection and stratified according to the levels of 2 out of 3 biomarkers of the 3 protein signature. CRYL1 was down-regulated in a unique case with poor outcome but no follow-up data; thus, no survival study taking into account CRYL1 could be performed. EFS, event-free survival; p, p-value obtained in the log- rank test.

Fig. 12. Kaplan-Meier plots of Event free survival of 30 adult patients with HCC treated by surgical resection and stratified according to a simplified 3 -protein signature. EFS, event-free survival; p, p-value obtained in the log-rank test.

BRIEF DESCRIPTION OF THE INVENTION

The present invention concerns a method or process of quantitatively and qualitatively analyze at least one of the biomarkers disclosed herein or any combination thereof in a sample previously obtained from a patient diagnosed for a, preferably primary, liver tumor, so that to determine the prognosis of said liver tumor in a patient, thus proposing in combination with the clinical stratification, preferably at diagnosis or alone after surgery, the most appropriate treatment. DETAILED DESCRIPTION OF THE INVENTION

DEFINITIONS

In the present disclosure, the“3-protein signature” is the combination of the three biomarkers disclosed herein that may be used in alone or may be used in combinations of two or three to further improve the prognosis of the patient with liver cancer and thus the patient's clinical evolution, treatment and outcome.

By“liver tumor”,“liver cancer” or“hepatic tumor”, it is meant a tumor originating from the liver of a patient, which is a malignant tumor (comprising cancerous cells), as opposed to a benign tumor (non-cancerous) which is explicitly excluded. Malignant liver tumors encompass mainly two kinds of tumors: hepatoblastoma (HB) or hepatocellular carcinoma (HCC). The prognosis of these two tumor types can be determined by using the presently reported biomarkers. In addition, in the context of the present invention, a patient is a child i.e., if it is a human host who is under 20 years of age. Therefore, in a particular embodiment, the liver tumor is a pediatric HB or a pediatric HCC. In another embodiment, the liver tumor is an adult HCC.

The biological sample of the present disclosure is a substance or a mixture of the substances that contain or is expected to contain/express one or more of the present biomarkers, and includes cells, tissues or bodily fluids from an organism, particularly human, for example, whole blood, urine, plasma, and serum, but is not limited thereto. Also the sample includes cells or tissues cultured in vitro as well as those derived directly from an organism. Various samples may be used for the detection of HCC or HB markers according to the present disclosure. In one embodiment, urine, whole blood, plasma and/or blood serum can be used. In other embodiment, the liver tissues/cells or in vitro cell cultures from an organism of interest, where HCC has developed or HCC is plausible or likely to be developed, may be used, but the samples are not limited thereto. Also the fractions or derivatives of the blood, cells or tissues are included. When cells or tissues are used, lysates thereof may also be used. The markers according to the present disclosure are primarily defined according to the expression level, the difference in the expression level or the changes in the expression level at the mRNA or protein level as compared to the reference sample through various quantitative, semiquantiative and/or qualitative analysis methods known in the art.

The quantitative, semiquantitative and qualitative analysis of the markers for HCC and HB prognosis according to the present disclosure may be done by various methods known in the art that can detect the proteins and mRNA quantitatively and qualitatively. For example, the methods include a western blot, an ELISA, a radioimmunoassay, an immunodiffusion, an Immunoelectrophoresis, an immunostaining, an immunoprecipitation, a complement fixation assay, a system based on tailed beads, a binding with a tailed antibody in solution/suspension and a detection by flow cytometry, or a mass spectrometry, and the like. These methods are known and documents such as chip-based capillary electrophoresis: Colyer et al. 1997. J Chromatogr A. 781 (1 -2):271 -6; mass spectroscopy: Petricoin et al. 2002. Lancet 359: 572-77; eTag systems: Chan-Hui et al. 2004. Clinical Immunology 111 :162-174; microparticle-enhanced nephelometric immunoassay: Montagne et al. 1992. Eur J Clin Chem Clin Biochem. 30:217-22 may be referred. As a way of example, an immunoassay using sandwich system like ELISA (Enzyme Linked Immuno Sorbent Assay), or RIA (Radio Immuno Assay) and the like may be used for quantitative and/or qualitative detection of the present markers. In this system, the biological samples are reacted with a first antibody fixed to a solid substrate/support such as a glass, a plastic (for example, polystyrene), polysaccharides, a bead, a nylon or nitrocellulose membrane or a microplate well to form a complex and the complex is then allowed to react with an second antibody that is usually labeled with agents that can be detected directly or indirectly such as radioactive substances like 3H or 1251, fluorescent materials, chemiluminescent substances, hapten, biotin, or digoxygenin and the like. In some cases, the labeling materials are conjugated with an enzyme such as horseradish peroxidase, alkaline phosphatase, or maleate dehydrogenase that is able to produce colors or color changes or illuminate in the presence of appropriate substrates. Other methods based on immune reaction may also be used. In this sense, an Immuno Electrophoresis such as an Ouchterlony plate, a Western blot, a Crossed IE, a Rocket IE, a Fused Rocket IE, or an Affinity IE, which can detect the markers simply by antigen-antibody reaction may be used. The agents or materials that may be used in the methods described above are known the art. For example, the markers may be detected through an antigen-antibody reaction, or a reaction with a substrate, nucleic acid or peptide aptamers, receptors or ligands that specifically recognize the present markers, or cofactors or using mass spectrometry. The agents or materials that bind or interact specifically with the markers of the present disclosure can be utilized by means of chip or with nanoparticles. The immunoassay or immunostaining methods as described above are disclosed in the following literatures : Enzyme Immunoassay, E.; Gaastra, W., Enzyme-linked immunosorbent assay(ELISA), in Methods in Molecular Biology, Vol. 1 , Walker, J.M. ed., Flumana Press, NJ, 1984 etc. The intensities of the signals generated by the immunoassay mentioned above are then analyzed, namely compared with the signals from appropriate controls for the determination.

In addition, methods for measuring the expression level of any of the biomarkers disclosed herein by using a material that detects at least one of presence, amount, and abundance pattern of mRNA transcribed by the gene and/or a protein encoded by the gene may also be used. Preferably, the material for measuring the expression level of the gene is at least one of a primer, a probe, an aptamer, and an antisense which are specifically bound to at least one selected from the group consisting of a nucleotide sequence of the gene, a complementary sequence thereof, a fragment of the nucleotide and a complementary sequence thereof. More preferably, the material for measuring the expression level of the gene is at least one selected from oligopeptides, monoclonal antibodies, polyclonal antibodies, chimeric antibodies, antibody fragments, ligands, peptide nucleic acids (PNA), aptamers, avidity multimers, and peptidomimetics which specifically bind to at least one of a polypeptide encoded by a nucleotide sequence of the gene, a polypeptide encoded by a complementary sequence thereto, and a polypeptide encoded by a fragment of the nucleotide sequence. Still more preferably, the material for measuring the expression level of the gene is a detection reagent of measuring a gene expression by at least one method of a reverse transcription polymerase chain reaction, a competitive polymerase chain reaction, a real-time polymerase chain reaction, a nuclease protection assay (RNase, S1 nuclease assay), an in situ hybridization method, a DNA microarray method, northern blotting, western blotting, an enzyme linked immuno sorbent assay (ELISA), a radioimmunoassay, an immunodiffusion method, Immunoelectrophoresis, a tissue immuno staining, an immunoprecipitation assay, a complement fixation assay, an FACS, a mass spectrometry and a protein microarray.

In the context of the present invention, the term “CKAP4” is understood as “cytoskeleton associated protein 4” (Gene ID: 10970 for human with the following Gene aliases: p63; CLIMP-63; ERGIC-63) and “63-kDa cytoskeleton-linking membrane protein” (UniProtKB ID: Q07065). In mouse, CKAP4 has Gene ID: 216197, it is also known as P63, CLIMP-63, 5630400A09Rik and has the protein UniProtKB ID: Q8BMK4). In zebrafish, UniProtKB ID: Q6NW47. In pig (Sus scrofa), CKAP4 has Gene ID: 100523493. In rat ( Rattus norvegicus), CKAP4 has Gene ID: 362859 and has the protein UniProtKB ID: D3ZH41 (D3ZH41_RAT).

In the context of the present invention, the term “C1 QBP” is understood as “complement C1 q binding protein” gene (gene ID: 708 for human with the following gene aliases: p32; HABP1 ; gC1 qR; GC1QBP; SF2p32; gC1 Q-R; COXPD33 or the“complement component 1 Q subcomponent-binding protein, mitochondrial” protein (UniProtKB ID: Q07021 for human). In mouse, C1 QBP has Gene ID: 12261 , it is also known as AA407365, AA986492, D11Wsu182e, HABP1 , P32, gCI qBP and has the protein UniProtKB ID: Q8R5L1 (Q8R5L1_MOUSE). In zebrafish, C1 QBP has Gene ID: 334619 it is also known as fa14h03, sb:cb785, si:ch1073-329n19.2, wu:fa14h03, zgc:110137. In rat, C1 QBP has Gene ID: 29681 , it is also known as Habpl , gC1 qR and has the protein UniProtKB ID: 035796 (C1 QBP RAT). In pig, C1 QBP has Gene ID: 110256103.

In the context of the present invention, the term“CRYL1” is understood as“crystallin lambda 1” gene (Gene ID: 51084 for human with the following gene aliases: GDH; HEL30; lambda-CRY) and“Lambda-crystallin homolog” protein (UniProtKB: Q9Y2S2 for human). In mouse, CRYL1 has Gene ID: 68631 it is also known as 1110025H08Rik, A230106J09Rik, C85932, gul3dh and has the protein UniProtKB ID: Q99KP3 (CRYL1 MOUSE). In zebrafish, CRYL1 has Gene ID: 751596, it is also known as im:6895749, zgc:152659. In rat, CRYL1 has Gene ID: 290277. In pig, CRYL1 has Gene ID: 396914, it is also known as CRY, GuIDH and has the protein UniProtKB ID: Q8SQ26 (CRYL1_PIG).

In the context of the present invention, the term “event-free-survival (EFS)” is understood as the length of time after primary treatment for a cancer ends that the patient remains free of certain complications or events that the treatment was intended to prevent or delay. These complications or events included cancer-related death of the patient as well as any particular kind of cancer progression and tumor spread such as tumor recurrence, metastasis in any organ, tumor growth in the blood vessels, etc.

In the context of the present invention, the term“overall-survival” is understood as the length of time from either the date of diagnosis or the start of treatment for a disease, such as cancer, that patients diagnosed with the disease are still alive.

In the context of the present invention, the term“prognosis of a hepatoblastoma (HB) or a hepatocellular carcinoma (HCC)” is understood as a medical term for predicting the likelihood of survival of a patient with malignant liver cancer. The prognosis of HB is associated with the presence of different clinical parameters at diagnosis such as patient’s age, serum alpha-fetoprotein levels, PRETEXT system (pre-treatment extent of disease), presence of metastasis and tumor spreading (i.e. intragrowth blood vessels, multifocality, tumor rupture, contiguous extrahepatic growth, caudate and lymph node involvement, metastasis). The prognosis of adult HCC is associated with the liver function (Child-Pugh scale), portal pressure, serum bilirubin, patient performance status, number of tumor nodules in the liver, presence of portal invasion, extrahepatic spread, presence of diabetes and the liver transplant consideration. Moreover, differentiation degree of tumor cells, presence of tumor satellites and microvascular invasion observed in the tumor specimen have been also associated with the prognosis of HCC patients, The main epithelial component in HB specimens has been also associated with the prognosis of HB patients. In the context of the present invention, the term “negative clinical evolution” is understood as poor prognosis or how the disease behaves over time in an unfavorable manner. Accordingly, patient during follow-up suffer certain complications or events related to cancer progression tumor spread, tumor recurrence, metastasis in any organ, tumor growth in the blood vessels among other which ultimately can derive in the death of the patient because of the disease.

In the context of the present invention, DFS, disease-free survival, is understood as the length of time after primary treatment for a cancer ends that the patient survives without any signs or symptoms of that cancer.

In the context of the present invention, PFS, progression-free survival, is understood as the length of time during and after the treatment of a disease, such as cancer, that a patient lives with the disease but it does not get worse.

In the context of the present invention, the term“control sample” is understood as the healthy biological sample such as non-tumor tissue that could be obtained from a healthy individual or from a liver cancer patient.

In the context of the present invention, the term“reference value” is understood as the value of the level of staining and/or concentration and/or the presence of the biomarker in the“control sample” or obtained thereof.

In the context of the present invention, the term “altered” is understood as a biomarker from which level of staining and/or concentration and/or the presence of the biomarker in a biological sample is different from the control sample.

DESCRIPTION

Liver cancer is a disease with an increasing worldwide incidence and a poor prognosis. The most frequent liver cancer is the Hepatocellular Carcinoma (HCC) which is the sixth most commonly occurring cancer in the world, the third leading cause of cancer mortality and it is mainly diagnosed in adults with an underlying liver disease. In childhood, the hepatoblastoma (HB) is the main liver tumor. However, it is a rare tumor with an incidence of 1 case out of 1 million children per year. A curative treatment is possible by combining chemotherapy and surgery. Nevertheless, 20% of HB patients do not survive cancer and survivors can be affected by serious side effects related to chemotherapy. CHIC clinical stratification of HB patients relies on clinical parameters including pre-treatment extension of the tumor, patient age, serum AFP levels, multifocal ity, presence of distant metastases and different degrees of tumor spread such as tumor rupture, blood tumor growth, etc. Pediatric HCC is a dismal disease, only approximately 30% of the patients survive the disease. Their prognosis is mainly linked to the tumor stage at diagnosis and its resectability. pHCC is the second most common liver tumor in children being most of the cases diagnosed after 10 years of age, it is the main hepatic tumor in adolescents and usually it is a dismal disease because usually its diagnosis is at advanced tumor stages when no chance of cure is possible.

To date, no biological markers have been incorporated into the clinical management of liver cancer to improve the prognosis of these patients suffering this kind of disease. Moreover, no markers have been proposed to stratify patients after surgery according to their prognostic risk.

As indicated in figure 7, the current clinical stratification for HB patients provides four groups of patients according to their prognosis: very low, low, intermediate and poor risk patients, being defined very low and low patients as low risk in the figure. The problem lies in the intermediate and poor prognostic groups, said problem being the uncertainty of the severity of the disease within said groups since they comprise patients with a potentially good prognosis as well as patients with a poor prognosis that are not possible to identify throughout clinical criteria. At any rate, and in absence of further reliable markers, a combined chemotherapy and surgery is administered to all of the patients pertaining to said groups giving raise to severe side effects in those patients who most probably were not in the need of the same therapy. In order to solve this problem, we herein provide a 3-protein signature as an independent predictor of prognosis optionally to be used together with the CHIC clinical stratification.

The authors of the present invention have discovered the prognostic value of three markers, which combination of them is called“The 3-protein signature”, in particular of CKAP4, C1 QBP and CRYL1 , by performing the first proteomic study on a series of 16 human HB specimens and their corresponding non-tumor tissues. Interestingly, after quantification of the biomarker expression by immunohistochemistry and defining a cut-off of biomarker alteration taking into account the healthy non-tumor liver (Figure 5), we observed an alteration of the expression of CKAP4, C1 QBP and CRYL1 in tumour tissue in 49, 52 and 11 % of the cases, respectively (Table 1 ).

Table 1. Global incidence of the 3 proteins in pediatric liver cancer. In the study, we included 139 HB and 5 pediatric HCC

Moreover, the survival study revealed that CKAP4 has a statistical significance concerning the prognosis, in terms of event free survival and overall survival, in an independent series of 144 paediatric liver cancer patients (log rank=0.0162 and 0.0479, respectively), including HB and pHCC, as shown in Figure 6. For C1 QBP, there is a trend in which cases with C1QBP altered have a worse outcome (see figure 6). Similarly, for CRYL1 , there is also a trend in terms of Event free survival but patients with CRYL1 altered showed a poor overall survival probability (log rank=0.0018, figure 6). The association of tumour CKAP4, C1 QBP and CRYL1 alteration with a more immature main epithelial component, mainly a non-foetal component (i.e. crowded foetal, embryonal, macrotrabecular, small cell undifferentiated, etc) confirmed the association of both biomarkers with the patient’s prognosis (p=0.010, p=0.008 and p=0.029, respectively; see Table 9A). Moreover, CKAP4 and C1 QBP were also associated with a tumour proliferation rate (measured by Ki67 immunostaining), another indicative of tumour aggressiveness (p<0.0001 and p=0.032, respectively). There is also a trend for which C1QBP is associated with tumour spread (VEPR variable) and multifocality (p=0.052 and p=0.051 , respectively). More interestingly, the combination of the 3 biomarkers showed a stronger impact on patient survival than the 3 biomarkers independently (Figure 6, bottom). Survival analysis in terms of EFS and OS of the 3 biomarkers individually compared with the combination of them as a 3-protein signature revealed an unexpected synergistic effect, wherein the probability of EFS at 150 months, being 96, 74, 78 or 50% depending if the tumors had 0, 1 , 2 or 3 biomarkers altered (Log-rank p=0.0041 ). The complementarity of the biomarkers probably relies on the different signalling pathways in which each of them participate.

The impact was even stronger in the OS analysis, in which OS probability at 150 months, end of follow-up, was of 100, 81 , 88 or 50% for patients with tumours with 0, 1 , 2 or 3 altered biomarkers (Figure 6, bottom). As the survival analysis showed that having 1 o 2 altered biomarkers lead to similar EFS and OS probabilities (EFS: 74 vs 78% and OS 81 vs 88%; data not shown), we defined a simplified 3-protein signature grouping as intermediate patients those with tumors with 1 or 2 altered biomarkers. The Kaplan-Meier curves showed that the group with best outcome included patients with tumours with none of the 3 biomarkers altered, that means that their expression was similar to the adjacent non-tumor liver (3-protein signature score =0 BM) with 96% probabilities of EFS at 150 months; a second group with an intermediate outcome included patients with tumours that have an alteration of 1 or 2 biomarkers having 76% probabilities of EFS at 150 months (3-protein signature score=1 or 2 BM) and finally, a third group of patients with worse prognosis characterized by having tumors with all biomarkers altered (3-protein signature score=3) showing and EFS probability of 50% (Figure 6). This stronger association of the 3-protein signature to EFS is even stronger in the analysis of OS, in which patients had 100, 85 or 50% of OS probabilities depending on having 0, 1 or 2 or 3 altered biomarkers (Log-rank p<0.0001 ). Interestingly in our cohort, only 2/6 (33%) HB tumours with HOC features had no altered biomarkers (3-protein score=0), and 4/6 (66%) had at least 1 altered biomarker (2/6 had 2-3 altered biomarkers). Regarding the pHCC, all had at least one altered biomarker (1/5 had 1 altered biomarkers whereas 4/5 had 2 or 3 altered biomarkers).

In addition and as shown in the examples, we have studied the expression of the 3 biomarkers of the 3-protein signature in a prospective series of 88 adult HCC patients. In tissue, we have seen an alteration of the expression of CKAP4, C1 QBP and CRYL1 in 17, 31 and 1 % of the cases, respectively (Table 11 , Figure 10). Similarly to pediatric liver tumors, there are also multiple combinations of the alteration of these biomarkers (Table 12) and an aberration in the expression of these biomarkers as well as their combination has been associated to clinical/pathological features of poor prognosis. Particularly, an overexpression/alteration of C1 QBP is associated with the presence of tumour vascular invasion and an advanced clinical BCLC stage (p=0.031 and p=0.037). CKAP4 is also associated with tumour features of aggressive tumours such as presence of tumor multifocality or presence of satellite nodules (p=0.002). Furthermore, the survival analysis performed with the 30 cases for which we follow- up data was available, revealed that both CKAP4 and C1 QBP have a statistical significance concerning the prognosis of HCC patients (log rank =0.005 and 0.024, respectively, see figure 11 ). We cannot perform the survival study with altered CRYL1 because of its low incidence (only 1 patient out of 88 had a loss of CRYL1 expression). Flowever, clinical and pathological data revealed that the unique tumour with CRYL1 altered had a poor differentiation degree with vascular invasion and was classified with an advanced clinical BCLC tumour stage, suggesting the potential association of CRYL1 also with aggressive tumour features.

In addition, when we studied the combination of the biomarkers the results were also significant. The alteration of at least one of the biomarkers of the 3-protein signature was associated with the presence of tumor vascular invasion and this impact on the Kaplan-Meier EFS curves (log rank=0.005).

Altogether, these results indicated that the 3-protein signature is a prognosis predictor, not only for paediatric patients with liver cancer (HB or pHCC) but also for adult HCC.

The 3-protein signature could be assessed not only at protein level but also at mRNA level (See Figure 4 and methodology of gene expression (mRNA) assessment in the last part of the examples section). In that regard, a significant correlation was observed between the quantitative protein levels of CKAP4, C1 QBP and CRYL1 assessed by Western blot and their gene expression determined by Human Transcriptome Array (Affymetrix). Thus, the 3-protein signature could be also named 3-gene signature which is also correlated with aggressive C2 tumour subtype and accordingly, patient prognosis.

Hence, detecting the level and/or concentration and/or the presence of at least one biomarker selected from the group consisting of CKAP4, C1 QBP and CRYL1 at a protein or RNA level in an isolated sample from a subject diagnosed with a hepatoblastoma (HB)) or a hepatocellular carcinoma (HCC) in children or adults, is particularly useful for the prognosis prediction of said subject. Moreover, detecting the level and/or concentration and/or the presence of at least one biomarker selected from the group consisting of CKAP4, C1 QBP and CRYL1 at a protein or RNA level in an isolated sample from a subject diagnosed with a liver cancer, is particularly useful for the prognosis of said subject and to decide treatment after surgery.

Therefore, a first aspect of the invention refers to a method for determining the prognosis of a subject diagnosed with liver cancer, comprising: a. detecting the level and/or concentration and/or the presence of at least one biomarker selected from the group consisting of CKAP4, C1 QBP and CRYL1 at a protein or RNA level in an isolated sample from said subject diagnosed with any of said cancers; and

b. comparing the detection result to that of a corresponding marker from a control sample or to that of a reference value, wherein a change or alteration in the level and/or concentration and/or the presence of at least one of the biomarkers at a protein or RNA level in the subject from the control sample or reference value, is indicative of the clinical evolution of the subject, wherein the clinical evolution refers to the event-free-survival (EFS), over-all-survival (OS), DFS, disease-free survival or PFS, progression-free survival, as well as any specific clinical data from the subject useful in the prognosis such as patient’s age, vascular invasion, any kind of tumor spreading, tumor multifocality, serum AFP levels, presence of satellites, BCLC stage, main HB epithelial component and/or HCC differentiation degree. In a preferred embodiment of the first aspect of the invention, the subject, preferably a human subject, most preferably an adult or a children, is diagnosed with a hepatoblastoma (HB) or a hepatocellular carcinoma (HCC).

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least biomarker CKAP4 in the subject compared to the control or reference value.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least biomarker C1 QBP in the subject compared to the control or reference value.

In a preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least biomarker CRYL1 in the subject compared to the control or reference value.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the biomarker CKAP4 and the change is the increased concentration and/or levels and/or presence of at least the biomarker CKAP4 in the subject compared to the control or reference value, wherein said increased concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the biomarker C1 QBP and the change is the increased concentration and/or levels and/or presence of at least the biomarker C1 QBP in the subject compared to the control or reference value, wherein said increased concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the biomarker CRYL1 and the change is the decreased or reduced concentration and/or levels and/or presence of at least the biomarker CRYL1 in the subject compared to the control or reference value, wherein said decreased or reduced concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the biomarkers CKAP4 and C1 QBP, and the change is the increased concentration and/or levels and/or presence of at least the biomarkers CKAP4 and C1 QBP in the subject compared to the control or reference value, and wherein said increased concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the biomarkers CKAP4 and CRYL1 , and the change is the increased concentration and/or levels and/or presence of at least the biomarkers CKAP4, and the reduced concentration and/or levels and/or presence of at least the biomarkers CRYL1 in the subject compared to the control or reference value, and wherein said change in the concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the biomarkers C1 QBP and CRYL1 , and the change is the increased concentration and/or levels and/or presence of at least the biomarkers C1 QBP, and the reduced concentration and/or levels and/or presence of at least the biomarkers CRYL1 in the subject compared to the control or reference value, and wherein said change in the concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject.

In another preferred embodiment of the first aspect of the invention, the method refers to a change in the level and/or concentration and/or the presence of at least the 3-protein/gene signature comprising of biomarkers: CKAP4, C1 QBP and CRYL1 , wherein the change is the increased concentration and/or levels and/or presence of at least the biomarkers CKAP4 and C1 QBP in the subject compared to the control or a reference value and the decreased concentration and/or levels and/or presence of biomarker CRYL1 in the subject compared to the control or a reference value, and wherein said increased and decreased concentration and/or levels and/or presence is indicative of a negative clinical evolution of the subject..

In yet another preferred embodiment of the first aspect of the invention or of any of its preferred embodiments, the concentration of at least the biomarkers CKAP4 and/or CRYL1 is determined in an isolated tissue liver sample and the concentration of at least the biomarkers C1 QBP is determined in an isolated tissue liver sample or in a blood, serum or plasma sample. In this sense and regarding the potential use of plasma samples, it is important to note that in order to assess the plasmatic levels of C1 QBP, as shown in the examples, an ELISA was performed using 30 plasma samples from patients with liver cancer (15 cases with an absence of parameters of poor prognosis and 15 with features of poor prognosis, including 22 paediatric and 10 adult patients with liver cancer, HB, pHCC and HCC) and 8 healthy individuals (see Fig 9).“Poor prognosis” group were samples from patients with features associated with poor prognosis such as presence of metastasis and/or multifocal tumors and/or PRETEXT IV and/or more than 5 years of age at diagnosis for HB patients The group of samples defined as“good prognosis” came from patients that did not have any abovementioned criteria associated to poor outcome. The ELISA results showed that patients with “poor” outcome features had higher concentrations of C1 QBP than those patients with “good” outcome features or healthy individuals (t-test p-value =0.0343) (Figure 9). Moreover, there was an association between higher levels of C1 QBP ([C1 QBP]>15ng/mL) and multifocal tumours (Fisher test p=0.035).

To our knowledge this is the first time, that such an association is disclosed between higher levels of C1 QBP ([C1QBP]>15ng/mL) in a serum, blood or plasma isolated sample and clinical poor prognostic features in liver cancer patients.

In yet another preferred embodiment of the first aspect of the invention or of any of its preferred embodiments, the detection is performed by at least one of an antigen- antibody reaction- or a mass spectrometry-based techniques. Preferably, the detection of the concentration or the presence of the biomarkers at the protein level is performed by at least one of a western blot, an ELISA, a radioimmunoassay, an immunodiffusion assay, and immunoelectrophoresis, an immunostaining, an immunoprecipitation, a complement fixation assay, a FACS, a mass spectrometry, or a protein microarray.

In yet another preferred embodiment of the first aspect of the invention or of any of its preferred embodiments, the material for measuring the expression level of any of the biomarkers is a material that detects at least one of presence, amount, and abundance pattern of mRNA transcribed by the gene and/or a protein encoded by the gene. Preferably, the material for measuring the expression level of the gene is at least one of a primer, a probe, an aptamer, and an antisense which are specifically bound to at least one selected from the group consisting of a nucleotide sequence of the gene, a complementary sequence thereof, a fragment of the nucleotide and a complementary sequence thereof. More preferably, the material for measuring the expression level of the gene is at least one selected from oligopeptides, monoclonal antibodies, polyclonal antibodies, chimeric antibodies, antibody fragments, ligands, peptide nucleic acids (PNA), aptamers, avidity multimers, and peptidomimetics which specifically bind to at least one of a polypeptide encoded by a nucleotide sequence of the gene, a polypeptide encoded by a complementary sequence thereto, and a polypeptide encoded by a fragment of the nucleotide sequence. Still more preferably, the material for measuring the expression level of the gene is a detection reagent of measuring a gene expression by at least one method of a reverse transcription polymerase chain reaction, a competitive polymerase chain reaction, a real-time polymerase chain reaction, a nuclease protection assay (RNase, S1 nuclease assay), an in situ hybridization method, a DNA microarray method, northern blotting, western blotting, an enzyme linked immuno sorbent assay (ELISA), a radioimmunoassay, an immunodiffusion method, immunoelectrophoresis, a tissue immuno staining, an immunoprecipitation assay, a complement fixation assay, an FACS, a mass spectrometry and a protein microarray.

Once we confirmed the impact of the 3-protein signature with the patient outcome, we were interested in evaluating its utility to be used for improving childhood liver cancer management. From our validation cohort of 144 patients, 128 patients had the clinical data to be classified with the current clinical stratification CHIC-HS (Meyers et al 2017). The survival analysis confirmed that the CHIC-HS is significantly associated with patient outcome in our cohort. Thus, patients classified as low risk have a 93% probabilities of EFS, compared to 76% of intermediate patients or 53% of high risk patients (Log-rank p<0.0001 ) (Figure 7).

In order to assess the impact or overlapping of our 3-protein signature with the clinical classification, each group of patients was in turn sub classified with the 3- protein signature. Interestingly, the results showed that our 3-protein signature is useful to improve the classification of intermediate patients, as the deregulation of the 3 proteins lead to a 50% EFS probabilities compared to patients with no altered biomarkers, who have an EFS probability of 100% (Log-rank p=0.0011 ). This effect was also seen in the OS analysis as intermediate patients with a 3-protein score=0 had 100% OS probabilities in contrast to patients with a score=1 or 2 who had 92% OS probabilities or score=3 who had 50% OS probabilities (Log-rank p=0.0002). The sub classification with the 3-protein signature for the low risk show no significant impact of the signature on patient EFS or OS, and a trend towards improving clinical classification can be observed for high risk patients (Figure 7). Moreover, the multivariate analysis confirmed these findings and identified the 3-protein signature as an independent and strong prognostic factor for liver cancer patients together with the clinical stratification (Table 10). Importantly, the same results were found in the study of 88 adult HOC samples. Particularly, we have found that the individual biomarkers of the 3-protein signature as well as their combination is also associated with the prognosis of adult HCC (Table 13; Figures 10 and 11 ). Thus, the 3-protein signature defined in the present invention is useful for predicting the prognosis of the main liver tumors not only those found in childhood but also those that arise in adulthood. This invention has particular relevance because of the high and increasing incidence of adult HCC, the sixth most common tumor in the world.

Therefore, we have identified the 3-protein/gene signature as an independent prognostic factor of liver cancer patients, preferably of childhood and adulthood liver cancer patients, diagnosed with a hepatoblastoma (HB), or a hepatocellular carcinoma (HCC) that for childhood liver cancer patients is a useful tool to improve prognostic prediction together with the CHIC clinical stratification (Table 10. Multivariate analysis). Hence, a second aspect of the invention refers to the method of the first aspect of the invention or of any of its preferred embodiments, wherein the method is used as an independent prognostic factor of liver cancer patients, preferably of liver cancer patients diagnosed with a hepatoblastoma (HB) or a hepatocellular carcinoma (HCC), or can be use together with the clinical stratification, or with any specific clinical data from the subject such as patient’s age, vascular invasion, any kind of tumor spreading, tumor multifocality, serum AFP levels, presence of satellites, BCLC stage, main HB epithelial component and/or HCC differentiation degree.

In addition, it is important to highlight the potential of the 3-protein/gene signature to predict survival in tumors at diagnosis (non-treated specimens). In this sense, as shown in the examples, the Kaplan-Meier analysis revealed that, despite not being significant, there is an association with both, EFS and OS. Interestingly, patients with zero altered biomarker (0 BM) s have 100% probabilities of EFS and OS, compared to the patients with at least 1 altered biomarkers that have lower EFS and OS probabilities of 70 and 75%, respectively (see figure 8). Besides, the association of the 3-protein signature with parameters of poor outcome in adult liver cancer (HCC), samples not treated by chemotherapy before surgery, supported the applicability to predict the prognosis risk at the time of diagnosis.

A third aspect of the invention refers to a method of treatment of a subject diagnosed with liver cancer patients, preferably of liver cancer patients diagnosed with a hepatoblastoma (HB) or a hepatocellular carcinoma (HCC), which comprises determining the clinical evolution of the subject by using the method of the first aspect of the invention or of any of its preferred embodiments, and administering a suitable treatment of liver cancer in case the method indicates a negative clinical evolution of the subject. Preferably, said suitable treatment is selected from the group consisting of surgical approaches (surgical resection, cadaveric or living donor liver transplantation), ablation, Transcatheter arterial chemoembolization (TACE) cisplatine, doxorubicin, carboplatin, sorafenib, Ifosfamide, fluorouracil, vincristine, etopside, pirarubicin, pirarubicin, or any combinations thereof. A fourth aspect of the invention refers to an in vitro use of a kit for the prognosis of a liver cancer in a subject diagnosed with liver cancer, comprising an agent to detect the level and/or concentration and/or the presence of at least one biomarker selected from the group consisting of CKAP4, C1 QBP and CRYL1 or any combination thereof, wherein prognosis is understood as the clinical evolution of the subject in terms of event-free-survival (EFS), overall-survival (OS), DFS, disease-free survival or PFS, progression-free survival, tumor recurrence and/or any of the clinical and tumor features associated to prognosis such as patient’s age, vascular invasion, any kind of tumor spreading, tumor multifocality, serum AFP levels, presence of satellites, BCLC stage, main HB epithelial component and/or FICC differentiation degree.

In a preferred embodiment of the fourth aspect of the invention, the subject is diagnosed with a hepatoblastoma (HB) or a hepatocellular carcinoma (FICC).

In another preferred embodiment of the fourth aspect of the invention, the kit comprises agents to detect the level of the 3-protein/gene signature comprising at least the three following biomarkers: CKAP4, C1 QBP and CRYL1.

As already specified in the“definitions” of the present invention, in another preferred embodiment of the fourth aspect of the invention, agents for measuring the expression level of any of the biomarkers is a material that detects at least one of presence, amount, and abundance pattern of mRNA transcribed by the gene and/or a protein encoded by the gene may be used. Preferably, the material for measuring the expression level of the gene is at least one of a primer, a probe, an aptamer, and an antisense which are specifically bound to at least one selected from the group consisting of a nucleotide sequence of the gene, a complementary sequence thereof, a fragment of the nucleotide and a complementary sequence thereof. More preferably, the material for measuring the expression level of the gene is at least one selected from oligopeptides, monoclonal antibodies, polyclonal antibodies, chimeric antibodies, antibody fragments, ligands, peptide nucleic acids (PNA), aptamers, avidity multimers, and peptidomimetics which specifically bind to at least one of a polypeptide encoded by a nucleotide sequence of the gene, a polypeptide encoded by a complementary sequence thereto, and a polypeptide encoded by a fragment of the nucleotide sequence. Still more preferably, the material for measuring the expression level of the gene is a detection reagent of measuring a gene expression by at least one method of a reverse transcription polymerase chain reaction, a competitive polymerase chain reaction, a real-time polymerase chain reaction, a nuclease protection assay (RNase, S1 nuclease assay), an in situ hybridization method, a DNA microarray method, northern blotting, western blotting, an enzyme linked immuno sorbent assay (ELISA), a radioimmunoassay, an immunodiffusion method, Immunoelectrophoresis, a tissue immuno staining, an immunoprecipitation assay, a complement fixation assay, an FACS, a mass spectrometry and a protein microarray. More preferably, the agent is to detect the concentration and/or presence or absence of the biomarkers at a protein level. Preferably, the agent to detect the concentration and/or presence or absence of the biomarkers at the protein level is an agent which is employed for a western blot, an ELISA, a radioimmunoassay, an immunodiffusion assay, and Immunoelectrophoresis, an immunostaining, an immunoprecipitation, a complement fixation assay, a FACS, a mass spectrometry, or a protein microarray. More preferably, the agent to detect the concentration and/or the presence or absence of the biomarkers at the protein level is a monoclonal antibody, a polyclonal antibody, a substrate, an aptamer, an avimer, a peptidomimetic, a receptor, a ligand or a cofactor. The following examples are merely for illustration purposes and do not limit the present invention.

EXAMPLES

Establishment of a collection of biological samples from pediatric patients with liver cancer and healthy individuals.

We have created a collection of highly annotated biological samples obtained from liver cancer patients registered at the Instituto de Salud Carlos III (ISCIII, ref C.0000226). Concerning childhood liver cancer, blood and/or tissue samples from 66 patients have been prospectively collected since 2009 (mean of 8 patients per year) including 51 HB (hepatoblastoma) and 6 pHCC (hepatocellular carcinoma). Additional retrospective samples have been obtained thanks to international collaborations. In the table 2, there is a summary of the main clinical and pathological features of the childhood liver cancer patients included in the proteomic and immunohistochemistry studies.

Concerning adult liver cancer, blood and/or tissue samples from 368 individuals with different stages of liver disease, including 175 HCC patients, 108 cirrhotic patients without primary liver cancer, 77 chronic hepatitis patients, and 8 individuals with no liver disease. Of these, 88 patients with HCC with tissue samples were included in the study to assess the 3 biomarkers of the 3-protein signature by immunohistochemistry. In the table 3, there is a summary of the main clinical and pathological features of the adult liver cancer patients included in the immunohistochemistry studies.

Table 2. Clinical, pathological and molecular features of the 160 patients included in the proteomic and immunohistochemistry (IHC) studies.

Table 3. Clinical, pathological and molecular features of the 88 adult patients with HCC included in the immunohistochemistry (IHC) study of the 3-protein signature.

Proteomic profiling of HB

With the aim to define the proteome of the HB and identify protein prognostic factors, we performed the first comprehensive proteomic study on tissue samples obtained from HB patients. To accomplish this, 16 clinical, pathological and molecular annotated tumours from 16 HB patients as well as 8 non tumor liver tissues were studied by two high-throughput techniques of quantitative proteomics: two- dimensional difference gel electrophoresis (2D-DIGE) and Label-free LC-MS (LF).

The unsupervised analysis of 2D-DIGE data revealed two main hierarchical clusters of tumor and non-tumor samples (Fisher’s Exact test, p=0.0013, Figure 1 ). Tumor samples were at the same turn classified in 2 different clusters associated to the previously 2 C1 and C2 HB prognostic subtypes (Cairo-Armengol, Cancer Cell, 2008). The experimental replicates (“R” samples in the figure) of 3 different samples (HB48, HB49 and HB59) resulted in similar protein profiles, indicating high robustness of 2D-DIGE proteomic data. Then, we selected 8 tumor and 4 non-tumor samples for LF analysis. The unsupervised study of the LF proteomic data showed again the three different groups of samples: non-tumor, C1 and C2 tumors.

In order to identify the deregulated proteins in HB, we compared the proteomic profiles of tumor and non-tumor samples. The two tumor samples that share part of their protein profiling with the non-tumor samples were excluded of this analysis. The supervised analysis comparing T vs NT revealed a total of 231 differentially expressed proteins (p<0.05 and FC+1.5); among them, 25 proteins were identified by 2D-DIGE, 178 proteins by LF and 28 were identified by both techniques. The two approaches were highly complementary since only 12% of the proteins were identified with the two proteomic techniques (fold change obtained by both techniques for common proteins is shown in Table 4.

Table 4. Proteins identified both by DIGE and LF. FC, fold change. For DIGE, if a protein have been identified in more than 1 spot, the range is shown and the number of spots in which the protein have been identified.

Three of the proteins identified by both techniques -ACTB, GRP75 and ENOA- showed discordances between both techniques, probably because of different isoforms and were excluded from further analysis. In total, 127 proteins were found to be up- and 101 proteins down-regulated in HB tumor as compared to non-tumor tissues. Top differently expressed proteins in HB are listed in Table 5. Deregulated proteins in HB. T/NT is Fold change (FC) shown; Criteria: LF p-value<0.0005; DIGE p-value<0.001 ; (*) When a protein was found by both techniques, LF FC was chosen. Abbreviations: Tech, Technique; FL, Fetal Liver; NL, Normal Liver; Ref, References related to cancer. Table 5. Top deregulated proteins in HB. T/NT, T vs. NT is fold change (FC); Criteria: LF p-value<0.0005; DIGE p-value<0.001 ; (*) When a protein was found by both techniques, LF FC was chosen. Abbreviations: Tech, Technique; Ref, References

The Ingenuity Pathway Analysis by using the list of proteins significantly deregulated in tumors (n=228; p<0.05; FC±1.5) revealed an activation of the PI3K/Akt, integrin, ILK, Rho and PAK signalling as well as an inactivation of HIPPO signaling pathway (Figure 2. Top deregulated pathways in HB vs NT according to the IPA analysis. Grey bars represent the activation z-score (z-score > ±2) and * represent the log p- value). Proteins identified by DIGE and/or LF as deregulated in HB vs NT and involved in the deregulated pathways according to IPA analysis are listed in Table 6. Table 6. Deregulated proteins in HB involved in the top deregulated pathways identified by IPA.

From the top deregulated pathways, activation of PI3K/Akt and inactivation of Hippo were selected for validation by WB. Thus, we confirmed that total Akt was under expressed in HB vs NT (t-test p=0.0106) and was even more repressed in aggressive C2 tumors (t-test p=0.0009). In contrast, phosphorylation of Akt at Ser473 was higher in HB vs NT (t-test p=0.0228). Regarding Hippo pathway, total YAP is overexpressed in HB vs NT (t-test p=0.0015). The overexpression of YAP is higher in C2 tumours than in C1 vs NT (t-test p=0.0428 and p=0.0153, respectively). Phosphorylation of YAP at Ser127 is higher in HB than in NT (t-test p=0.0256). Deregulated proteins in aggressive HB tumors

In an attempt to identify proteins deregulated in aggressive tumors, we performed a supervised analysis by comparing the proteomic profiles of the two prognostic subtypes (C1 and C2) reported above. By using the two different proteomic approaches, we were able to detect 230 differentially expressed proteins between C2 and C1 (FC+1.5, p<0.05), 108 down and 124 upregulated. Among them, 20 (9%) were identified by 2D-DIGE, 216 (94%) by LF and 6 (3%) in both techniques. The top deregulated proteins differently expressed in the 2 HB subclasses identified by the two techniques are summarized in Table 7 (see in Figure 2 a more restrictive table).

Table 7. Top deregulated proteins in aggressive C2 tumors. Criteria: p- value<0.0005 in all comparisons except for C1 or C2 vs NL DIGE data (p- value<0.001). Proteins identified by both techniques. Abbreviations: Tech =Technique; FC = Fold Change; NL= Normal Liver; Ref = References related to liver cancer; Bold indicates proteins selected for WB validation.

The Ingenuity Pathway Analysis including the top deregulated proteins in C2 vs C1 tumors (n=229; p<0.05; FC±1.5), revealed a strong over-activation of the eukaryotic initiation factor-2 (EIF2) signalling pathway (Activation z-score=2; p-value=1.22 10-6) in the aggressive C2 tumours. No other significant pathways were found with the selected criteria. Proteins from the EIF2 pathway identified as deregulated in C2 vs C1 comparison are listed in Table 8. Table 8. Deregulated proteins in C2 tumors involved in EIF2 signaling pathway.

Different phosphorylation of EIF2 between the 2 tumour prognostic subclasses was further validated by Western blot using the 19 (13 C1 and 6 C2) tumour samples revealing a decrease in the pEIF-2a (Ser51 )/total EIF in C2 HB as compared to NL (FC=-1 ,7; p=0.0010). Identification of the 3-protein signature

To select among the differently expressed proteins between the C1 and C2 tumours the candidates to be validated as prognostic biomarkers, we focused on those proteins differentially expressed between both tumour subtypes (FC± 2.5, p<0.008) but also between tumour and non-tumour tissues (FC±1.5, p<0.05). This was done with the purpose to facilitate their assessment in any pathology department by using the non-tumour liver as a control of protein expression. Therefore, we selected 8 putative prognostic markers (ALBU, DERM, GLUL, TXNL, TMPRSS13, C1 QBP, CRYL1 and CKAP4) for further validation by WB. The quantified protein expression of these markers in C1 and C2 tumours as well in non-tumour samples is shown in Figure 3A. After quantification of the specific protein bands, 3 out of the initial 8 proteins, cytoskeleton associated protein 4 (CKAP4), complement C1 q binding protein (C1 QBP) and crystallin lambda 1 (CRYL1 ) were found to be significantly deregulated in the aggressive C2 tumor subtype as compared with the non-tumor liver and were selected for further validation study by using immunohistochemistry in an independent set of samples (protein expression assessed by WB is shown in Figure 3B). In order to confirm its correlation with patient survival, WB data was used to classify patients as having 0, 1 , 2 or 3 of the biomarkers altered taking into account NL biomarker band intensity (CKAP4 and C1QBP≥ 2 NL and CRYL1 < -2 NL). The Kaplan-Meier survival analysis, showed that tumours with no altered biomarkers had an EFS probability of 100% as compared to patients with at least one altered biomarker that had a lower EFS probability (Log-rank p-value=0.0027; Figure 3C). This finding revealed the complementarity and synergistic effect of the 3 prognostic biomarkers. Moreover, high correlation between protein and mRNA quantification of the 3 biomarkers was found (Figure 4), suggesting that the 3-protein signature can be also determined at gene expression level.

Assessment of the prognostic value of the 3-protein signature in the validation cohort of childhood liver cancer patients

The immunostaining of C1 QBP, CKAP4 and CRYL1 was assessed in 11 tissue microarrays from 144 childhood liver cancer patients, including 139 with HB and 5 with pHCC in non-cirrhotic liver and additionally in 44 non tumor liver (NL) samples. CKAP4 showed a cytoplasmic, finely granular staining pattern, on both non-tumor hepatocytes and tumor cells, with different intensities (negative, weak and strong). Most cases showed a similar and diffuse intensity all over the examined tissue. No staining of nuclei or other tissue cells was seen. C1 QBP showed also a cytoplasmic, granular pattern, with a more varied shade of intensities (negative, weak, moderate and intense). Focal staining of tumor nuclei was sometimes noted. Non tumor liver was either negative or showed a faintly positive cytoplasmic staining of hepatocytes. No staining of other tissue elements was noted. CRYL stained both cytoplasm and nuclei of tumorl cells. The intensity of staining was negative, weak, moderate or intense. Non-tumor hepatocytes showed a similar pattern of staining with a tendency to increase nuclei intensity in periportal areas. No staining of other tissue elements was noted.

In order to define the value of the 3-protein signature for every tumor, a score was calculated for each protein taking into account the global score (percentage of positive cytoplamic cells multiplied by the intensity of the cytoplasmic staining). Then, we defined C1 QBP and CKAP4 biomarkers as“altered” when their staining score was two-fold higher than the staining of the adjacent non-tumor liver whereas CRYL1 tumor staining was defined as“altered” when no staining was observed as compared again with the adjacent non-tumor liver. An overview of the 3-protein signature definition is shown in Figure 5A. The global score determined by Immunohistochemistry (IHC) for the non-tumor and tumor hepatocytes (NT and T) for the different biomarkers is shown in Figure 5B. A representative staining for each marker for non-tumour as well as for an altered and non-altered tumours are shown in Figure 5C.

First, we studied the association of the individual biomarkers and the simplified 3- protein signature (grouping those patients with at least 1 altered biomarker) with clinical, pathological and molecular parameters (Table 9A and B). The results showed that none of the biomarkers were associated with any of the clinical parameters included in the clinical CH IC stratification. Interestingly, all 3 biomarkers as well as the 3-protein signature were associated with a more immature main epithelial component and moreover, CKAP4 and C1 QBP and the 3-protein signature with a high proliferation rate measured by Ki67 immunostaining. These findings revealed a high correlation of the altered biomarkers with pathological parameters associated to tumour aggressiveness and poor patient outcome including patient age (Table 9B). In that regard, patients with an altered expression of C1 QBP or with some of the biomarkers altered in the simplified 3-protein expression (3-prot sign≥1 ) had an older age with regards to those patients with C1 QBP or none of the biomarkers of the 3-protein signature altered.

Table 9A. Association of the 3 proteins with clinical, pathological and molecular features. Chi-square or Fisher Exact test p-value is shown depending on statistical convenience.

Table 9B. Association of the 3 proteins with patient age at diagnosis. P-val, unpaired t-test p-value.

Interestingly, CKAP4 has a statistical significance concerning the prognosis of pediatric liver cancer patients in terms of EFS and OS (Figure 6). For the remaining markers, there is a trend in which cases with C1 QBP or CRYL1 altered have a worse outcome, except for CRYL1 and OS (see figure 6).

More interestingly, the combination of the 3 biomarkers showed a stronger impact on patient survival than the 3 biomarkers independently (Figure 6.) and we observed an additive effect of the biomarkers and the probability of EFS at 150 months, being 96, 74, 78 or 50% depending if the tumours had 0, 1 , 2 or 3 biomarkers altered (Log-rank p-=0.0041 ).

The impact was even stronger in the OS analysis, in which OS probability at 150 months was of 100, 81 , 88 or 50% for tumours with 0, 1 , 2 or 3 altered biomarkers. As the survival analysis showed that having 1 o 2 altered biomarkers lead to similar EFS and OS probabilities (EFS: 74 vs 78% and OS 81 vs 88%), we defined a simplified 3-protein signature grouping patients with 1 or 2 altered biomarkers. The Kaplan-Meier curves showed that the group with best outcome included patients with tumors with an expression of the biomarkers similar to the adjacent non-tumor liver (global score=0) with 96% probabilities of EFS at 150 months, a second group with an intermediate outcome had patients with tumors that have an alteration of 1 or 2 biomarkers having 76% probabilities of EFS at 150 months and finally, a third group of patients with worse prognosis was characterized by having tumors with all biomarkers altered (global score=3) showing and EFS probability of 50% (Figure 6. Survival analysis of the 3 biomarkers individually and the 3-protein signature. Kaplan- Meier plots of the 3 biomarkers individually associated to EFS and OS). This stronger association of the 3-protein signature to EFS is even stronger in the analysis of OS, in which patients had 100, 85 or 50% of OS probabilities depending on having 0, 1or

2 or 3 altered biomarkers (Log-rank p<0.0001 ). Interestingly, only 2/6 (33%) analysed NOS tumours had no altered biomarkers (3-protein score=0), 4/6 (66%) had at least one altered biomarker (2/6 had 2-3 altered biomarkers). Regarding the pHCC, all had at least one altered biomarker (1/5 had one altered biomarker whereas 4/5 had 2 or 3 altered biomarkers).

The strong correlation of the 3-protein signature with patient prognosis indicated its utility to be used as a predictor of prognosis after surgical resection in HB patients. Accordingly, those patients having tumors with none of the biomarkers of the 3- protein signature altered could be benefit of no chemotherapy post-surgery and a reduction of the adverse side effects of chemotherapy whereas those with 1 , 2 or 3 biomarker altered could be treated with different cycles of chemotherapy after surgery in order to reduce the risk of tumor recurrence.

Once we confirmed the impact of the 3-protein signature with the patient outcome, we were interested in evaluating its utility to be used for improving HB management. From our validation cohort of 144 patients, 128 patients had the clinical data to be classified with the current clinical stratification CHIC-HS (Meyers et al., 2017). The survival analysis confirmed that the CHIC-HS is significantly associated with patient outcome in our cohort. Thus, patients classified as low risk have a 93% probabilities of EFS, compared to 76% of intermediate patients or 53% of high risk patients (Log- rank p<0.0001 ) (Figure 7. Clinical CHIC-HS and 3-protein signature survival analysis. One hundred and twenty-eight patients were classified according to the last clinical stratification system CHIC-HS (Meyers et al., 2017). Then low, intermediate and high risk patients were sub classified using the 3 protein-signature. Interm,, intermediate.).

In order to assess the impact or overlapping of our 3-protein signature with the clinical classification, each group of patients was in turn sub classified with the 3- protein signature. Interestingly, the results showed that our 3-protein signature is useful to improve the classification of intermediate patients, as the deregulation of the

3 proteins lead to a 50% EFS probabilities compared to patients with no altered biomarkers, who have an EFS probability of 100% (Log-rank p=0.0011 ). This effect was also seen in the OS analysis as intermediate patients with a 3-protein score=0 had 100% OS probabilities in contrast to patients with a score=1 or 2 who had 92% OS probabilities or score=3 who had 50% OS probabilities (Log-rank p=0.0002. The sub classification with the 3-protein signature for the low and high risk show no significant impact of the signature on patient EFS or OS, despite a trend towards improve classification can be observed (Figure 7. Clinical OH IO-HS and 3-protein signature survival analysis. One hundred and twenty eight patients were classified according to the last clinical stratification system OH IO-HS (Meyers et al 2017). Then low, intermediate and high risk patients were sub classified using the 3 protein- signature. Interm, intermediate.).

Interestingly, the multivariate analysis identified the 3-protein signature as an independent prognostic factor of paediatric patients with liver cancer together with the CHIC clinical stratification (Table 10. Multivariate analysis.).

Table 10. Multivariate analysis.

In addition, we wanted to assess the potential of the 3-protein signature to predict survival in non-treated specimens. The Kaplan-Meier analysis revealed that, despite not being significant, there is an association with both, EFS and OS. Interestingly, patients with zero altered biomarkers have 100% probabilities of EFS and OS, compared to patients with at least 1 altered biomarkers (see figure 8) Assessment of the prognostic value of the 3-protein signature in the cohort of adulthood liver cancer patients

In addition, we have studied the expression of the 3 biomarkers of the 3-protein signature in a prospective series of 88 adult HCC patients. The methodological approach and the criteria used to define the immunohistochemistry score was equal to the one used for the study of childhood liver tumors mentioned above. In tissue, we have seen an alteration of the expression of CKAP4, C1 QBP and CRYL1. The percentage of biomarker alteration in adult liver cancer were lower than the observed in pediatric liver tumors.

Table 11. Global incidence of the 3 proteins in adult HCC.

Table 12. Combination of the alteration of the 3 biomarkers of the protein signature in HCC (n=88). Abbreviations: BM, biomarker.

The survival analysis has been performed with only prospective cases for which we have follow-up data available (n=30). Both CKAP4 and C1 QBP have a statistical significance concerning the prognosis of HCC patients (see figure 12). We cannot perform this study with altered CRYL1 because of its low incidence (only 1 patient, 1.14% of the cases) and the survival data is not available for this patient. However, clinical and pathological data revealed a poor differentiated tumor with vascular invasion and an advanced clinical BCLC tumor stage, suggesting a potential association also with aggressive tumor features).

In addition, when we study the combination of CKAP4 and C1 QBP the results are also significant. In this case, any patient of HCC has not 2 proteins altered at the same time possibly it is due to the low number of cases (see figure 12).

There is also an association of the different biomarkers as well as the combination of them (3-prot signature) with the clinical data. Particularly, an overexpression/alteration of C1 QBP is associated with the presence of tumor vascular invasion and an advanced clinical BCLC stage and CKAP4 overexpression is associated with invasive tumors (multifocal ity or tumors with satellite nodules). The 3-protein signature was also associated with vascular invasion.

Table 13. Association of the 3 proteins with clinical and pathological features of HCC (n=88 patients, Fisher Exact test)

The mRNA study has been performed by using Human Transcriptome Array data (Affymetrix). HTA data was also correlated with qPCR data obtained by using the primers of the table 14, the RHOT2 as a reference gene and the 2^_DDa method to determine gene expression Plasma expression of C1QBP in childhood and adulthood patients with liver cancer

In order to assess the plasmatic levels of C1 QBP, ELISA was performed using 30 plasma samples from pediatric and aduthood patients with liver cancer, including cases with HB and HCC as well as 7 healthy patients. Patients with liver cancer were stratified according to the presence of prognostic features: 15 were defined as good prognosis whereas 15 with poor prognosis. Childhood patients with tumors with features of poor prognosis had metastasis and/or multifocal tumors and/or advanced PRETEXT stage and/or AFP >10⁶ ng/mL and/or more than 3 years and/or pHCC. Adult patients with tumors with features of poor prognosis had an advanced BCLC tumor stage and/or multifocal tumors and/or tumor recurrence. Childhood and adulthood patients classified as“good prognosis” did not have any criteria associated with poor outcome.

The results of the ELISA of C1 QBP revealed that patients with tumors classified as “poor outcome” had higher concentrations of C1 QBP than patients with tumors classified as “good outcome” or healthy individuals (t-test, p<0.05) (Figure 9). Interestingly, patients with levels of C1 QBP above 15 ng/mL had significatively multifocal tumors than patients with liver cancer with levels of C1 QBP lower than 15 ng/mL in plasma (Fisher test p=0.035).

Table 14. Primers used to assess the 3-protein signature at RNA level.

Claims

1. A method for assessing the prognosis of a liver cancer in a subject diagnosed with hepatoblastoma (HB) or a hepatocellular carcinoma (HCC), comprising: a. detecting the level, concentration and/or presence of at least CKAP4 (“cytoskeleton associated protein 4”), at a protein or RNA level in an isolated sample from said subject diagnosed with any of said cancers; and

b. comparing the detection result to that of a corresponding marker from a control sample or to that of a reference value, wherein an increased concentration, increased levels and/or increased presence of at least biomarker CKAP4 in the subject compared to the control or reference value is indicative of a negative clinical evolution of the subject.

2. The method of claim 1 , wherein the method detects the level, concentration and/or presence of at least biomarkers: CKAP4, C1 QBP (“complement C1q binding protein”) and CRYL1 (“crystallin lambda 1”), wherein the increased concentration, levels and/or presence of at least biomarkers CKAP4 and C1 QBP in the subject compared to the control or a reference value and the decreased concentration, levels and/or presence of biomarker CRYL1 in the subject compared to the control or a reference value is indicative of a negative clinical evolution of the subject.

3. A method of classifying a subject diagnosed with hepatoblastoma (HB) or a hepatocellular carcinoma (HCC), which comprises determining the clinical evolution of the subject by using the method of any of claims 1 to 2, and classifying or stratifying the patients on the basis of the said determination.

4. A method of classifying a subject diagnosed with hepatoblastoma (HB) or a hepatocellular carcinoma (HCC) after surgical resection, which comprises determining the clinical evolution of the subject by using the method of claim 2, wherein those patients having none of the biomarkers altered are classified as those with no risk of tumor recurrence whereas those with 1 , 2 or 3 biomarkers altered are classified as those in risk of tumor recurrence.

5. In vitro use of a kit comprising an agent to detect the level, concentration and/or the presence of at least CKAP4, for performing the method of claim 1.

6. In vitro use of a kit comprising an agent to detect the level, concentration and/or the presence of at least the biomarkers selected from the group consisting of CKAP4, C1QBP and CRYL1 , for performing the method of claim 2.

7. In vitro use of a kit comprising an agent to detect the level, concentration and/or the presence of at least the biomarkers selected from the group consisting of CKAP4, C1 QBP and CRYL1 , for performing the method of any of claims 3 to 4.

8. The use of the kit of any of claims 5 to 7, wherein the agents detect the concentration and/or presence or absence of the biomarkers at a protein level.

9. The use of the kit of claim 6 wherein the agent to detect the concentration and/or presence or absence of the biomarkers at the protein level is an agent which is employed for a western blot, an ELISA, an radioimmunoassay, an immunodiffusion assay, and Immunoelectrophoresis, an immunostaining, an immunoprecipitation, a complement fixation assay, a FACS, an mass spectrometry, or a protein microarray.

10. The use of the kit of any of claims 5 to 7, wherein the agents detect the concentration and/or presence or absence of the biomarkers at a RNA level.