As explained above, genetic alterations in HCC are found in a variety of cancer driver pathways with a highly heterogeneous pattern of alteration among patients [60,73,88]. The main challenge is to classify the different molecular profiles of HCC, in order to stratify the patients and more accurately define the treatment options and prognosis. However, subclassifications based only on genomic profiling do not encompass all the factors that contribute to the phenotypic heterogeneity observed in tumors, either because they do not integrate other factors, such as tumor-extrinsic factors, or due to the complexity and contribution of the possible combinations of mutations found in this cancer. There are 2 major subgroups of HCC tumors accurately defined by different biological processes: the proliferation and non-proliferation subclasses [89]. Although this classification is mainly based on the gene expression profile of the tumors, it is strongly associated with clinical and pathological features of the patients and their surrounding etiology, thus integrating different sources of information to define clinical outcomes [22]. On one hand, the proliferation subclass accounting for around 50% of patients is enriched in signals related to cell proliferation and cell cycle progression and is generally associated with a more aggressive phenotype and poor outcome. The activated signaling cascades found in this group and related to proliferation and cell survival include MET or hepatocyte growth factor receptor (HGFR) pathway [90], TGFβ signaling, insulin-like factor I (IGF) pathway [91], RAS/mitogen-activated protein kinase pathways [92], and AKT/MTOR signaling pathway [93], among others. Enrichment in the expression patterns associated with stem cell features (e.g., NOTCH) [94] and markers of progenitor cells (e.g., epithelial cell adhesion molecule) [95], also found in this subclass, indicate a moderate/poor cell differentiation state that correlates with more aggressive clinical behavior [96,97]. Another characteristic aspect of this subgroup is that tumors present higher rates of chromosomal instability due to aberrant epigenetic changes. In fact, there is a defined DNA methylation-based prognostic signature [77] that correlates with signatures of progenitor cells and poor survival. On the other hand, the non-proliferation subclass is characterized by a well-differentiated phenotype, with normal hepatocyte-like features. This subgroup of HCC is less aggressive and correlates with low levels of a-fetoprotein (AFP), the most widely used biomarker to detect HCC, contrary to the high levels found in the proliferation subclassified patients [22]. One subset of this group is characterized by the liver-specific activation of WNT signaling pathway (about 25%), mainly by mutations in CTNNB1 [98], while the other main subgroup is associated with immune response [99]. Overexpression of the epidermal growth factor receptor (EGFR) is also observed in the non-proliferation group and plays a critical role in the maintenance of the transformed phenotype of HCC cells [100]. Regarding etiologic factors, HBV-related HCC tumors are predominantly classified in the proliferation subclass, whereas HCV and alcohol-related HCC are more prevalent in the non-proliferation subclass.

3. Intra-Tumor and Intra-Patient Heterogeneity

Intra-tumor heterogeneity is a feature frequently found in many solid tumors [105,106]. HCCs are frequently detected with a nodule-in-nodule appearance, where a subclone resides embedded in another [107,108]. Many other patients present multiple hepatic tumors that can develop from different clones (presenting more than one primary tumor simultaneously) or arise from a single original tumor via intrahepatic metastasis [109,110]. Tumor cells are subjected to a natural selection process within the tumors and the microenvironment that leads to clonal evolution and the coexistence of diverse clones in a single tumor. It is commonly thought that this intratumor heterogeneity is critically associated with the development of resistance against cancer therapies. In this sense, it can be hypothesized that a therapy might have an anti-tumor effect globally while inducing a positive selection in the most resistant tumoral subclones.

About 70% of the patients with HCC under curative treatment options develop recurrence, considered early-recurrence within two years of treatment or late-recurrence when it occurs after more than two years [16,17,111,112]. Although there are some studies indicating that genetic and DNA methylation patterns of HCC can be used to identify patients at high risk for recurrence [18], the origin and evolution of the relapsed tumors are still controversial. Traditionally, it was considered that early-relapsed tumors were derived from residual or intrahepatic micrometastasis rather than de novo tumorigenesis or synchronous tumors with variable clonality. However, in a study from 1989 [14] it was already shown that there is not a clear correlation between the time of appearance of the recurrence and the clonality of the recurrent tumors, an observation that was later supported by other researchers [18,113], that there is not a clear correlation between the time of appearance of the recurrence and the clonality of the recurrent tumors. More recently, an exhaustive clonality analysis at single-cell level of early relapsed HCC cohorts demonstrated that tumoral cells in relapsed tumors derived from a minor clone from the primary tumor and were characterized by a distinct tumor microenvironment that conferred enhanced immune evasion capabilities [113]. Focusing on evolutionary trajectories in liver cancer, Ding and colleagues [18] identified TP53, CTNNB1, and TERT mutations as founder drivers and convergent mutations in key drivers including TP53, TERT, CTNNB1, TSC2, JAK1, NOTCH1, FGFR3, ATRX, and RPS6KA3. These convergent mutations can favor each ancestor clone to experience parallel series of expansions, inferring an environmental selection that allows parallel or selective seeding of ancestor clones. These findings were in line with a previous work by Torrecilla S. et al. [114] where they described TERT, TP53, and CTNNB1 mutations as trunk alterations in HCC, meaning they are ubiquitously present across different regions of the same tumor and between primary and metastatic tumors in more than 85% of the cases studied. Ding et al. [18] further defined a co-evolutionary relationship between the genome and epigenome in HCC, highlighting a close interplay between genetic and epigenetic landscapes during malignant progression and pointing towards the potential of epigenetic signatures to define novel biomarkers to trace tumor evolution and progression. Recently, Losic and colleagues characterized the interactions of different components of the HCC ecosystem during cancer evolution by integrating RNA-seq, DNA-seq, TCR-seq, and SNP information from a dataset of multiregional samples from HCC patients. Their findings indicated that clonal evolution depends on different selection pressures that are primarily immune- and treatment-mediated [115], supporting the relevant role of the tumor microenvironment in clonal selection and intra-patient heterogeneity.

From a clinical perspective, it would be expected that high intra-tumoral heterogeneity and clonal diversity in HCC correlates with worse patient prognosis and limited therapeutic options, as this enhanced heterogeneity in the tumors would trigger worse therapeutic responses in addition to more susceptibility to developing targeted therapy resistance. In fact, it has been observed that those HCC patients with higher heterogeneity correlate with a worse prognosis and poor survival [115,116]; however, additional studies are necessary to decipher this association. Most importantly, both intra-tumor and intra-patient tumor heterogeneity contribute to vastly increasing inter-patient heterogeneity.

4. Clinical Implications of HCC Heterogeneity

References to the clinical implications of liver tumor heterogeneity have been mentioned along the different aspects addressed in this revision. The devastating concerns associated with the inefficient therapeutic responses against liver cancer reaffirm the imperative need to study in detail the mechanisms driving heterogeneity, to standardize a better subclassification of the patients for clinical management. Regarding decisions about the treatment of patients with HCC, physicians commonly rely on the stage of the disease rather than its molecular singularities. However, as it has been extensively explained here, the phenotypic differences at the molecular level between patients within the same stage are abysmal. Characterization of molecular subtypes and/or oncogenic signatures have generally improved the patient outcome for several types of cancer. However, the molecular classification of HCC has not been implemented for managing patients. Due to this complexity, there are not clear inter-patient biomarkers that can predict the response to a specific therapy (with the exception of ramucirumab and high AFP levels), whereas the high incidence of intra-tumor heterogeneity can induce a positive selection in the tumor subclones that is hard to foretell. In this sense, the widespread implementation of transcriptomic profiling of tumors at single-cell resolution by performing single-cell RNA and DNA sequencing along with single-cell proteomic studies might provide a comprehensive view of the tumor microenvironment and tumor cell trajectories to more accurately provide stratification strategies for HCC patients. The implementation of this novel methodology along with the use of experimental systems to properly model and understand HCC heterogeneity will enable better therapeutic approaches, based on the identification of molecular signatures that foster the choice of one therapy over another (see Figure 1). Moreover, identification of high intra-tumor heterogeneity in patients could be decisive for the election of combinatorial therapies over single-based treatment options in order to avoid the emergence of resistance. Regarding tumor recurrence of HCC, the effects of therapeutic options can be critical as patients might have very different prognoses and/or can benefit from different therapeutic approaches; unfortunately, primary tumors are characterized in most circumstances. In fact, it might be necessary to re-biopsy recurrences, regardless of relapsed time of the patients, to establish clonal origin and make decisions about new potential therapies that might be very different from the therapies proposed in the first instance.

Figure 1. Potential implications of dissecting HCC heterogeneity. Implementation of high-throughput analysis to characterize HCC heterogeneity can provide useful biomarkers and pave the way for the development of novel models of HCC. The derived animal models and in vitro screenings that accurately recapitulate the complexity of this disease might represent powerful tools to test different drug sensitivities and potentially impact on the clinical management of patients. WGS, whole genome sequencing: WES, whole exome sequencing; CNA, copy number alteration; scRNA-seq, single-cell RNA-seq; PDX, patient-derived xenografts; PDO, patient-derived organoids. Figure prepared with Biorender.