Battle of the biopsies: Role of tissue and liquid biopsy in hepatocellular carcinoma

Summary

The diagnosis and management of hepatocellular carcinoma (HCC) have improved significantly in recent years. With the introduction of immunotherapy-based combination therapy, there has been a notable expansion in treatment options for patients with unresectable HCC. Simultaneously, innovative molecular tests for early detection and management of HCC are emerging. This progress prompts a key question: as liquid biopsy techniques rise in prominence, will they replace traditional tissue biopsies, or will both techniques remain relevant? Given the ongoing challenges of early HCC detection, including issues with ultrasound sensitivity, accessibility, and patient adherence to surveillance, the evolution of diagnostic techniques is more relevant than ever. Furthermore, the accurate stratification of HCC is limited by the absence of reliable biomarkers which can predict response to therapies. While the advantages of molecular diagnostics are evident, their potential has not yet been fully harnessed, largely because tissue biopsies are not routinely performed for HCC. Liquid biopsies, analysing components such as circulating tumour cells, DNA, and extracellular vesicles, provide a promising alternative, though they are still associated with challenges related to sensitivity, cost, and accessibility. The early results from multi-analyte liquid biopsy panels are promising and suggest they could play a transformative role in HCC detection and management; however, comprehensive clinical validation is still ongoing. In this review, we explore the challenges and potential of both tissue and liquid biopsy, highlighting that these diagnostic methods, while distinct in their approaches, are set to jointly reshape the future of HCC management.

Introduction

Background

Liver cancer, mainly hepatocellular carcinoma (HCC), is a growing cause of worldwide cancer-related fatalities and is predicted to cause more than 1.3 million annual deaths by 2040.¹ HCC typically arises on cirrhotic livers, most commonly in patients with chronic liver diseases including chronic hepatitis B or C (HBV or HCV) infection, alcohol-related liver disease, or the increasingly prevalent metabolic dysfunction-associated steatotic liver disease.^1,2 Several new tyrosine kinase inhibitors and immunotherapies have recently gained approval for the treatment of advanced HCC, signalling a promising shift in therapeutic options. Additionally, there are ongoing developments in molecular stratification of HCC. However, HCC continues to be diagnosed only at advanced stages, complicating treatment efforts. Moreover, we lack robust, predictive and prognostic biomarkers to tailor therapeutic strategies, making it difficult to identify which patients would derive the most benefit from specific therapies. This highlights the critical need for access to molecular data on HCC to refine biomarker development for improved detection and management.

The promise of precision medicine in HCC

The advent of molecular diagnostics in oncology has ushered in a new era of precision medicine. By detecting specific genetic alterations within cancer cells, these tests can reveal the molecular underpinnings of the disease, facilitate early diagnosis, and guide the choice of targeted treatments. These molecular tests can enable us to tailor therapy to the patient’s unique genetic profile, thus increasing the likelihood of a positive response. Furthermore, molecular diagnostics can be used to monitor disease progression and detect minimal residual disease, enhancing our ability to prevent recurrence. They may also aid in predicting patient prognosis, providing valuable information for patient counselling and management. In essence, the application of molecular diagnosis can streamline the path from diagnosis to treatment, improving patient outcomes in cancer care. Despite these promising advances, the full potential of molecular stratification and prognostication in HCC is yet to be realised, underscoring the need for continued innovation in this crucial area of research.

To create high-precision tests for HCC, comprehensive tumour molecular data is vital. Yet, such data remains elusive in HCC due to the infrequency of tissue biopsies. Liquid biopsies have emerged as an innovative solution, providing an alternative and minimally invasive means of obtaining critical genetic data on tumours. By enabling analysis of circulating tumour cells (CTCs) and tumour products like DNA in the blood, liquid biopsies serve as a genomic looking glass into the dynamic tumour landscape. They can offer a real-time view of tumour evolution, tracing disease progression and response to treatment. Yet, the use of tissue or liquid biopsies has not yet been integrated into standard clinical practice for HCC. Uncertainties concerning their sensitivity, specificity, and cost-effectiveness, as well as the relative merits of tissue vs. liquid biopsies, are under intense investigation. It remains to be seen whether liquid biopsies can fully obviate the need for tissue biopsies or if each retains a distinct role. In this review, we discuss recent advances in the field of precision medicine for HCC, addressing the role of both tissue and liquid biopsy.

Molecular landscape of HCC

HCC is a complex and heterogeneous disease involving both cancer cell-intrinsic genetic aberrations and changes within the tumour immune microenvironment.³ Comprehensive multiomic profiling has been used to uncover this diversity and complexity, and could identify potential opportunities for individualised therapeutic strategies and biomarkers. Certain common genomic occurrences, such as a mutation in the promoter region of the telomerase reverse transcriptase (TERT) gene, Wnt pathway disruptions, TP53 and AT-rich interactive domain-containing protein (ARID) gene mutations, MYC oncogene activation, and transforming growth factor-β (TGFβ) pathway dysregulation, show promise as potential therapeutic targets or predictive biomarkers in HCC.^4–7 For instance, pTERT mutations, which are the most frequent genetic alterations observed in HCC, are evident in early-stage HCC and are associated with a poorer recurrence-free survival rate.^8,9 This suggests their potential use as a tool for early detection and risk stratification. The Wnt pathway, crucial in liver development and regeneration,¹⁰ is often disrupted in HCC.¹¹ Activation of this pathway has been suspected to be associated with a reduced response to immunotherapy,^12,13 however, post hoc analyses of the CheckMate 459 and IMbrave150 trials did not confirm this association.^14,15 Further data are awaited. Additionally, recurrent TP53 gene mutations, particularly common in advanced-stage HCC, correlate with an aggressive disease phenotype, poor overall survival, and tumour recurrence.^16–19 Mutations in ARID genes, present in 3–5% of HCC cases, are associated with advanced-stage tumours characterised by high vascularity.²⁰ The MYC oncogene, regularly overexpressed in HCC, is a key promoter of tumorigenesis, although targeting it directly remains a challenge.²¹

Other promising targets include NRF2/KEAP1, a cytoprotective system, and TGFβ signalling, which is activated in a subset of HCC tumours.^22,23 Mutations in mechanistic target of rapamycin (mTOR) pathway genes or downstream activation of mTOR signalling in subsets of patients often leads to metabolic dysregulation, and may eventually be targeted using mTOR inhibitors in select patients with mTOR addiction, presenting yet another possible therapeutic approach for HCC, potentially in combination with immunotherapy.^24,25 Thus, detection of these genetic events via tissue or liquid biopsy has the potential to change management. Lastly, histological features may also provide insights into genetic alterations and oncogenic pathways, and thus guide patient prognosis. For example, poorly differentiated, proliferative tumours tend to harbour TP53 mutations, FGF19 (fibroblast growth factor 19) amplifications, or activation of TGFβ, RAS/MAPK, PI3K/AKT pathways. On the other hand, well-differentiated, non-proliferative tumours tend to harbour CTNNB1 (encoding β-catenin) mutations, show activation of Wnt and JAK/STAT pathways, and have microtrabecular and pseudoglandular patterns on histology, with less immune infiltration.^26,27

Overview of current management of HCC

Treatment options for HCC depend on the stage of the disease.^28–30 In early-stage HCC, potential curative approaches include surgical resection, transplantation, or local ablation. Organ scarcity limits the availability of liver transplantation, which offers high survival rates with low recurrence risk by addressing both HCC and underlying cirrhosis. For intermediate-stage HCC, intra-arterial therapies such as transarterial embolisation, transarterial chemoembolisation, and transarterial radioembolisation serve either as first-line treatments or as bridging therapies before transplantation.³¹ Stereotactic body radiotherapy, with its high local control rates, has emerged as another potential bridging therapy option for patients with HCC awaiting transplantation.³² In advanced-stage HCC, systemic therapies involving targeted therapies and immune checkpoint inhibitors are commonly employed. The groundbreaking IMbrave150 study demonstrated improved overall survival, along with significant enhancements in overall response rate and progression-free survival, by combining the PD1 (programmed death 1) inhibitor atezolizumab and the VEGF (vascular endothelial growth factor) inhibitor bevacizumab, leading to its adoption as the new frontline standard of care for advanced HCC.³³ Subsequently, the combination of the anti-CTLA-4 (cytotoxic T lymphocyte antigen-4) antibody tremelimumab and the anti-programmed death ligand 1 (PD-L1) agent durvalumab has also been approved as a first-line treatment option for advanced HCC.³⁴ This consensus is reinforced by international guidelines, consolidating the significance of immune checkpoint inhibitor-based regimens in HCC treatment.³⁰ Thus, recent advances in treatment have vastly extended treatment options for patients with HCC, both for early and advanced stages.

Tissue biopsy for diagnosis and management of HCC

Current non-invasive imaging diagnostic tests for HCC

Screening for early detection of HCC is crucial since the therapeutic window for curative surgical intervention for HCC is often limited to its earliest stages. The American Association for the Study of Liver Diseases and the European Association for the Study of the Liver currently recommend biannual ultrasound surveillance, with or without serum alpha-fetoprotein (AFP) measurement, for high-risk populations.^35,36 This includes individuals with cirrhosis of both viral and non-viral aetiology, as well as non-cirrhotic patients with chronic HBV infection. Ultrasound, while central to routine surveillance, is not without its challenges. Despite demonstrating a sensitivity of 58–89% and a specificity of over 90%, adherence to regular ultrasound examinations is often undermined by various factors.³⁷ These factors include knowledge gaps, social disparities impacting healthcare access, and interobserver variability, which can lead to high false-positive rates. Furthermore, ultrasound often falls short when it comes to visualising and differentiating early HCC lesions and dysplastic nodules, although contrast enhancement can potentially boost its sensitivity.^38,39

Non-invasive imaging tools such as triphasic CT or MRI have become invaluable in diagnosing and staging HCC. These methods leverage the distinctive vascular changes during hepatic carcinogenesis and the high likelihood of HCC in cirrhotic livers. Diagnostically, HCC often exhibits a radiographic signature consisting of arterial phase hyperenhancement (wash-in) and venous phase hypo-enhancement (wash-out) of contrast agents, owing to their arterial hypervascularity.^40,41 For lesions exceeding 2 cm, this radiographic pattern presents over 90% sensitivity and specificity.⁴² Moreover, tumour necrosis visualised on CT/MRI can define a response to treatment according to RECIST criteria for HCC.⁴³ In summary, while the importance of ultrasound surveillance in early HCC detection cannot be overstated, it is challenged by variability and adherence issues. On the other hand, cross-sectional imaging with contrast-enhanced CT or MRI, the primary methods for diagnosing HCC and assessing treatment responses, also face obstacles, particularly false positivity, and cost. Thus, it is essential to consider these limitations and strive to develop more accessible, cost-effective, and accurate screening and diagnostic tests for HCC.

Current role of tissue biopsy in diagnosis and management of HCC

Even though tissue biopsy is not routinely employed for the diagnosis of HCC, there are specific instances where a biopsy is necessary. For instance, lesions between 1–2 cm that remain indeterminate on dynamic contrast-enhanced CT or MRI can be biopsied to arrive at a definitive diagnosis.³⁶ A biopsy is generally necessary in the non-cirrhotic liver setting, where the specificity of imaging in diagnosing HCC declines.³⁶ In addition, a liver tumour biopsy is indicated in patients with suspected intrahepatic cholangiocarcinoma, mixed HCC-cholangiocarcinoma, or secondary malignancies, as histological markers can differentiate HCC from other tumours.⁴⁴ Moreover, evaluation of the non-tumoural liver in these biopsies may also be useful to characterise the underlying liver disease and provide insight into HCC prognosis.⁴⁴ Another major advantage of tissue biopsy is the ability to classify HCC into histological subtypes that correlate with molecular classifications and have prognostic implications, such as the highly differentiated, CTNNB1-mutated and the poorly differentiated, TP53-mutated subtypes.^26,45,46 While pathology-based biomarkers have yet to demonstrate clinical efficacy in terms of guiding treatment selection, tissue biopsy may become a more routine diagnostic procedure in the era of precision medicine.

We acknowledge that tissue biopsy has a few limitations (Fig. 1). Due to sampling bias and tumour heterogeneity, the small tissue cores obtained during biopsy may not be entirely representative. Additionally, liver biopsy is associated with increased risks of pain, bleeding, and very rarely, hypotension, tachycardia, pneumothorax, or biliary peritonitis.^47,48 Another possible complication is the potential for “needle tract seeding,” where tumour cells can detach from the primary site and travel along the needle pathway. Although older studies reported a risk of needle tract seeding of up to 3%,⁴⁹ technical improvements and the performance of a high volume of biopsies are associated with a much lower rate of needle tract seeding of less than 1%.^50,51 Based on our own experience, the fear of needle tract seeding appears to be overstated. Thus, despite its limitations, biopsy remains a vital diagnostic tool, and with advances in techniques, particularly in high-volume centres, risks such as needle tract seeding are very low.

Fig. 1. Overview of advantages and disadvantages of tissue vs. liquid biopsy in hepatocellular carcinoma.

cfDNA, cell-free DNA; CTCs, circulating tumour cells; EVs, extracellular vesicles; TME, tumour microenvironment.

Potential future role of tissue biopsy in management of HCC

As more targeted therapies become available, liver biopsy may play a larger role in guiding target identification and HCC management (Fig. 1). This is based on the premise that, in numerous other cancers, biomarkers derived from biopsy have proven invaluable in identifying cancer subtypes and determining the optimal treatment strategy. Among the 206 distinct oncology therapeutic products approved by the US FDA between 2000 and 2022, an overwhelming majority (97%) are aimed at specific molecular targets, such as PD1, EGFR, BCR-ABL1 fusion, CD19 or HER2.⁵² In fact, tissue biomarkers have paved the way for site-agnostic approvals, with notable examples being BRAFV600E mutations, NTRK fusions, RET fusions, high tumour mutation burden, microsatellite instability-high, and mismatch repair-deficient status.⁵³ However, it is important to note that these biomarkers may only be applicable to a small proportion (5–10%) of patients with HCC. While biomarker-defined populations accounted for 49% of oncology approvals for targeted therapies by the FDA, liver cancer currently has no biomarker-driven therapy, compared to breast cancer (88% of approvals) or lung cancer (61% of approvals).⁵² As our understanding of the molecular landscape of HCC continues to improve, the development of targeted therapies and their integration into the treatment armamentarium for HCC becomes more crucial.

Currently, biomarker-driven management is limited, given the lack of evidence supporting the efficacy of targeted therapy in molecular subsets of HCC. However, based on results from the REACH-2 trial, serum AFP has found a place in HCC management. Patients with HCC and AFP >400 are more likely to demonstrate benefit with ramucirumab, based on improved overall survival (7.8 months vs. 4.2 months with placebo) noted in the second-line treatment setting.⁵⁴ In the future, biomarkers identified from liver biopsy may guide targeted treatment stratification. This is supported by a study which performed targeted exome sequencing of advanced HCC, and found that nearly a quarter of patients had potentially clinically actionable driver mutations involving TSC1/2 (8.5%), FGF19 (6.3%), MET (1.5%), and IDH1 (<1%).^13,55 Moreover, the landscape is evolving with ongoing clinical trials for HCC incorporating both all-comer patient enrolment and biomarker-guided approaches. Several tissue-based biomarkers, such as glypican-3 (GPC3: NCT05003895, NCT05103631), epithelial cell adhesion molecule (NCT05028933, NCT03013712), MET (NCT01755767), mucin 1 (NCT02587689), MHC1 (NCT05195294), and TERT (NCT05595473), are being incorporated into these clinical trials. Thus, as our understanding of the biological complexity of HCC deepens, the potential role of biomarkers, and by extension liver biopsy, in guiding treatment choices and improving patient outcomes will become increasingly evident.

Another area of active research is the identification of biomarkers to predict response to immune checkpoint inhibitors. We highlight three such putative predictive biomarkers. First, PD-L1 expression on tissue biopsy may be predictive of response to immunotherapy in patients with HCC. However, the dynamic nature of PD-L1 expression and lack of standardised methods to evaluate its status adds complexity to these assessments. For instance, trials like CheckMate 459 and KEYNOTE-224 observed higher response rates in patients with advanced HCC and PD-L1 positivity when treated with immunotherapy.^56,57 In contrast, trials like the CheckMate 040 study reported no significant difference in response between PD-L1-positive and -negative patients treated with nivolumab.⁵⁸ This discrepancy across trials is not unexpected given the studies were not specifically designed to differentiate responders based on PD-L1 expression. Additionally, PD-L1 expression is generally not a sensitive biomarker across many cancer types, with sensitivities differing based on expression thresholds and regional sampling bias.^59,60 Second, high tumour mutational burden or microsatellite instability have been used as biomarkers for response to immunotherapy in other cancers, but less than 1% of HCCs appear to have these features.⁶¹ Third, multiple gene signatures which can predict response to immunotherapy have also been reported but require further validation.^62–65 These findings underscore the current complexities and highlight the potential for tissue-based biomarkers in the future, which may become instrumental in stratifying patients for immunotherapy in HCC.

Thus, tissue biopsy remains an invaluable tool in cancer diagnosis. By collecting and analysing tumour samples, we can expand our understanding of cancer biology, identify novel biomarkers, and guide the development of new therapies. A retrospective cohort of stored clinical specimens is usually critical to verify biomarkers developed in a preclinical setting, before conducting large, expensive, prospective, randomised-controlled trials. Without biological tissue specimens from a significant number of patients with HCC, especially from those with advanced or metastatic HCC, this pivotal phase in the development of biomarkers for HCC is significantly impeded. Tissue biopsies, when available, can be retrospectively queried in research settings where integrated genomics, digital pathology, and AI can be utilised to derive clinically meaningful biomarkers. The main caveat of existing studies is that tissue biopsies tend to be representative only of early-stage, resected lesions.^4–6,66 Nevertheless, investigators have successfully used these tissues to define novel molecular subclasses of patients with HCC for treatment stratification.⁶⁷ For example, Zeng et al. developed an AI algorithm based on 336 whole-slide histological images integrated with RNA sequencing data to identify immune-related gene signatures potentially correlating with responses to immunotherapy.⁶⁸ Other studies employed deep learning systems to diagnose HCC, perform histological classification, and provide prognostic indicators through whole-slide imaging analysis.^69–71 Thus, it is plausible that deep learning algorithms, fine-tuned on integrated histopathology and spatial transcriptomic data, will enable the diagnosis, tumour classification, and prediction of treatment outcomes for patients with HCC. While the current lack of biomarkers and molecular subtypes may draw into question the ethics of routine biopsies for all patients with HCC, a counter perspective emphasises the urgency of escalating the frequency of tumour biopsies, especially within clinical trial frameworks. Such an approach may hasten the discovery of novel biomarkers and fast-track the evolution of personalised targeted therapies for HCC.

Liquid biopsy for diagnosis and management of HCC

Current serum biomarkers for HCC

With the push towards less invasive techniques in medicine, liquid biopsy provides a supplementary approach to imaging and promises to augment current strategies for the early detection of HCC, the identification of minimal residual disease, treatment selection, and the monitoring of therapeutic responses (Fig. 1). Apart from being able to provide serial samples, liquid biopsy provides a plethora of genomic and transcriptomic information that would otherwise only be accessible via tissue biopsy, as well as enabling the characterisation of intra- and inter-tumoural heterogeneity.⁷² Intra-tumoural heterogeneity can arise from spatial and temporal clonal differences, while inter-tumoural heterogeneity can arise from differences in the tumour microenvironment (consisting of cancer cells, stroma, endothelial cells, fibroblasts, and immune cells). Through liquid biopsy, serial measurements of blood-based analytes over a treatment window may provide novel insights into biological processes contributing to proteogenomic dynamics, matrix remodelling, metabolic reprogramming, and clonal diversity (Fig. 1).

Over the last several decades, the most widely utilised serum biomarker for HCC screening and diagnosis is AFP. AFP is the foetal analogue of serum albumin, which rises in production during foetal liver development, and diminishes quickly after birth.⁷³ However, various complex mechanisms deregulate this epigenetic silencing and turn on AFP expression in the development of HCC, and thus it has become a widely used screening tool for patients with HCC. Additionally, other prognostic serum proteins identified include fucosylated fraction of AFP (AFP-L3), des-gamma carboxyprothrombin (DCP), and GPC-3.^74,75 There are multiple studies investigating AFP as a biomarker for early-stage screening as part of complex panels (e.g., GALAD; gender + age + AFP + DCP + AFP-L3), alongside various methylation markers (e.g., HOXA1, TSPYL5, and B3GALT6),⁷⁶ or combined with clinicopathologic information.^77,78 Moreover, a few groups have assessed AFP as a biomarker for post-resection recurrence or as a biomarker to predict response to treatment with cabozantinib or the combination of atezolizumab + bevacizumab.^79,80 Despite the wide acceptance of AFP as a screening tool, limitations include that it is influenced by race/ethnicity, aetiology of HCC (viral vs non-viral), molecular subclasses, and tumour burden, and that it is abundant in other benign liver conditions.⁸¹ The rest of this section will detail novel cell-free analytes being tested in the clinic for HCC detection, prognostication, and monitoring of therapeutic responses (Table 1).

Table 1.

Recent Investigations of Liquid Biopsy to Detect and Monitor Treatment Response in HCC

Study Author, Year	Patient Population	AUC (Sensitivity/ Specificity)	Analyte Detected	Comments
Circulating Tumor Cells
Wang et al., 2020	N=344 HCC patients preoperatively	HR=0.46 (TTR) HR=0.32 (OS)	CTC identified as DAPI+/CD45−/CK+ cells or DAPI+/CD45−/EpCAM+ cells	HCC prognostication of recurrence following TACE
Winograd et al., 2020	N=87 HCC patients	0.81 (71.1/91.8) early vs metastatic disease HR=3.22 (PD-L1+ CTCs predict shorter OS) 5/5 ICI responders had PD-L1+ CTCs at baseline, while 1/5 ICI nonresponders had PD-L1+ CTCs at baseline	CTCs identified as DAPI+/CD45−/CK+ cells And PD-L1+ CTCs identified as DAPI+/CD45−/CK+/PD-L1+ cells	HCC detection and prognostication, prospective study
Zhao et al., 2023	N=270 preoperative HCC patients (N=52 validation set)	0.94 (NA/NA) CTC number: HR=11.89 CTC clusters: HR=13.67	CTCs identified as DAPI+/CD45−/CK+ cells	HCC recurrence post-op prediction
Wei et al., 2023	N=227 HCC patients	0.84 (NA/NA) CTC number: HR=1.98	CanPatrol™ CTC-enrichment technique	HCC extrahepatic recurrence prediction
cfDNA Mutations
Kaseb et al., 2019	N=206 HCC patients	N/A 87.9% confirmed alterations	Blood-derived ctDNA Actionable drivers identified: EGFR, MET, ARID1A, MYC, NF1, BRAF, and ERBB2 Nonactionable drivers identified: TP53, CTNNB1, APC	Cohort was mix of different stages
Qu et al., 2019	N=331 at risk patients	0.93 (100.0/94.0)	Blood-derived ctDNA profiling of TERT, TP53, CTNNB1, AXIN1 + HBV integration breakpoint + AFP + DCP + age + gender	Early HCC detection HCCscreen
Zhang et al., 2021	N=571 HCC patients	N/A 77.1% confirmed alterations	Blood-derived ctDNA Drivers identified: TP53, TERT, CTNNB1, TSC2, RB1, ARID1A, DNMT3A, MLL2, AXIN1, APC	68% sensitivity early stage detection of ctDNA 86.3% sensitivity late stage detection of ctDNA
Nguyen et al., 2023	N=55 HCC patients	0.86 (81.0/81.0)	Blood-derived ctDNA genes profiled: APC, ARID1A, AXIN1, BRAF, CDKN2A, CTNNB1, EGFR, KRAS, PIK3CA, PTEN, STK11, TP53, TERT + fragment length profiles	HCC classification compared to healthy participants
cfDNA Methylation Signatures
Xu et al., 2017	N=1,098 HCC patients	0.94 (83.3/90.5)	401 HCC-specific cfDNA methylation markers	Early HCC detection; prognostic survival
Kiesel et al., 2019	Phase I: N=21 HCC patients Phase II: N=95 HCC patients	Phase I: 0.91 (86.0/87.0) Phase II: 0.96 (95.0/92.0)	6 methylated cfDNA markers (HOXA1, EMX1, ECE1, AK055957, PFKP, CLEC11A) normalized by B3GALT6	Cohort was mix of different stages; phase I tested markers alone or in combination (reported EMX1 and CLEC11A combination here)
Luo et al., 2022	N=120 HCC patients (training set) N=67 HCC patients (validation set)	0.93 (84.0/96.0)	2321 methylated cfDNA markers	Early HCC detection
Chalasani et al., 2022	N=156 HCC patients	0.86 (88.0/82.0)	3 methylated cfDNA markers (HOXA1, TSPYL5, B3GALT6) + AFP + sex	Early HCC detection
Lin et al., 2022	N=122 HCC patients	0.94 (75.7/91.2)	28 methylated cfDNA markers + protein markers (AFP, AFLP-L3, DCP) + clinical variables (age, sex)	Early HCC detection; HelioLiver Test
Cheishvilli et al., 2023	N=504 HCC patients	0.94 (84.5/95.0)	Differential methylation of CpGs: CHFR, VASH2, CCNJ, GRID2IP, F12	Early HCC detection; EpiLiver Test
ncRNAs Signatures
Lin et al., 2015	N=27 patients in validation set	0.83 (85.7/91.1)	Seven-miRNA panel (miR-29a, miR-29c, miR-133a, miR-143, miR-145, miR-192, and miR-505)	Early HCC detection from controls at-risk
Teufel et al., 2019	N=349 patients	NA	Nine-miRNA panel (miR-30a, miR-122, miR-125b, miR-200a, miR-374b, miR-15b, miR-107, miR-320, and miR-645)	Prognosticate overall survival after regorafenib (RESORCE trial)
Yamamoto et al., 2020	N=173 patients in validation set	0.99 (97.7/94.7)	Eight-miRNA panel (miR-320b, miR-6724–5p, miR-6877–5p, miR-4448, miR-4749–5p, miR-663a, miR-4651, and miR-6885–5p)	Early HCC detection from controls at-risk
Pratama et al., 2020	N=86 HCC patients under treatment	0.84 (72.0/75.0)	Four-miRNA panel (miR-4443, miR-4454, miR-4492, miR-4530)	HCC therapy response and disease-free survival
Fu et al., 2022	N=76 HCC patients in validation set	0.91	Three lncRNA panel of AC005332.5, ELF3-AS1 and LINC00665	HCC detection from healthy controls
Ning et al., 2023	N=171 patients in validation set	0.94 (84.0/86.0)	6 cfRNA markers: 1 lncRNA (CYTOR), 1 miRNA (miR-21–5p), 3 cfRNA fragments (WDR74, SNORD89, RN7SL1), and 1 alternative splicing candidate (GGA2)	Early HCC detection HCCMDP Panel
Extracellular Vesicles
Sun et al., 2020	N=158 HCC patients	0.93 (94.4/88.5)	EV-derived 10 mRNA signature (AFP, GPC3, ALB, APOH, FABP1, FGB, FGG, AHSG, RBP4, TF)	Early HCC detection from controls at risk
von Felden et al., 2021	N=209 HCC patients	0.87 (86.0/91.0)	3-small RNA cluster	Early HCC detection from controls at risk
Rui et al., 2022	N=124 HCC patients	0.95 (89.0/92.0)	EV-derived 3 miRNA signature (miR-122–5p, let-7d-5p, and miR-425–5p)	Early HCC detection, classify HCC from non-tumor patients
Yang et al., 2022	N=50 HCC patients	0.97 (92.0/90.0)	EV-derived 3 miRNA signature (miR-26a, miR-29c, miR-199a)	Early HCC detection from controls at risk
Li et al., 2022	N=60 HCC patients	0.76 (48.2/93.9)	EV-cfDNA to detect c.747 G > T mutation in TP53 gene	Prediction of HCC and prognosticate microvascular invasion
Ye et al., 2022	N=7 HCC patients	0.86 (88.0/86.0)	EV-associated proteins (CO9, LBP, SVEP1, VWF)	HCC detection from controls at risk
Sun et al., 2023	N=72 HCC patients (validation cohort)	0.93 (94.0/81.0)	EpCAM+CD63+, CD147+CD63+, and GPC3+CD63+ EVs	Early HCC detection from controls at-risk

Circulating tumour cells

Profiling CTCs was one of the initial liquid biopsy approaches in oncology. CTCs migrate from the primary tumour or metastatic site following tumour invasion into nearby vasculature, allowing for entry into systemic circulation where their quantity correlates with tumour burden. Also, profiling CTCs provides a unique window into active cellular processes directly from the tumour. Thus, CTCs may be restricted in their ability to detect early HCC but may be advantageous for monitoring recurrence or treatment response.⁸² In fact, a recent meta-analysis of 20 studies (1,191 total patients) demonstrated a 95% sensitivity and 60% specificity for CTC-based HCC diagnosis.⁸² There are many investigations utilising different technologies to isolate, enrich, and profile CTCs for the purpose of monitoring HCC recurrence and treatment response (Table 1). For example, Zhao et al. and Wei et al. profiled CTCs from over 200 patients with HCC preoperatively and were able to predict HCC recurrence with an AUC of 0.95 and 0.84, respectively.^83,84 Additionally, Wang et al. demonstrated that transarterial chemoembolisation increased the time to recurrence and death in patients positive for CTCs, but showed no such effect in patients negative for CTCs.⁸⁵ Additionally, Winograd et al. observed that PD-L1+ CTCs may be useful to predict ICI response.⁸⁶ Thus, CTCs may hold promise for predicting minimal residual disease, treatment response, and recurrence.

The main challenge associated with using CTCs has been their heterogeneity within patients with HCC. Different investigations have described single-cell RNA sequencing of CTCs and found intra-CTC heterogeneity, likely representing the different origins and various biological factors influencing release under different microenvironmental pressures.^87–89 Specifically, Sun et al. utilised single-cell RNA sequencing on CTCs to demonstrate that CCL5 from CTCs recruits regulatory T cells to enable immune evasion and metastasis.⁸⁹ Thus, improved understanding of CTC biology and their spatial relationships could help to determine which subpopulations provide the most prognostic value.

Cell-free DNA

Cell-free DNA (cfDNA) refers to the large pool of circulating double-stranded DNA fragments associated with nucleosomes, while circulating tumour DNA (ctDNA) refers to smaller (150 bp) fragments transiently present in bodily fluids (<2 h) released by apoptotic, necrotic, or actively proliferating tumour cells. Traditionally, cfDNA has been probed for total quantity, integrity, and copy number alterations,⁹⁰ while more novel approaches probe for methylation patterns, particularly at CpG islands of tumour suppressor genes, or mutational signatures reflecting the patient’s current “oncologic footprint”. These two cellular processes are closely intertwined in HCC as mutations in chromatin remodelling genes – ARID1A (13%) and ARID2 (7%), which are key constituents of SWI/SNF (SWItch/sucrose non-fermentable) chromatin remodelling components – are frequently observed, along with hypermethylation of CpG islands and hypomethylation in open sea regions.⁹¹

Genomic profiling has identified mutations in TERT, CTNNB1, and TP53 as the major drivers in HCC, with various groups attempting to profile these mutations in cfDNA for HCC detection and prognosis (Table 1). Therefore, cfDNA from patients at risk is likely to contain these mutations and could be used to identify early lesions given their truncal role. Seminal investigations have indicated that TERT promoter and TP53 mutations are present in cfDNA samples in >75% of patients with early-stage HCC.^92,93 Specifically, for diagnostic detection, Qu et al. developed a panel which profiles the major trunk mutations, HBV integration breakpoints, clinical variables, and protein markers to diagnose early HCC with 100% sensitivity and 94% specificity.⁹⁴ Additionally, analysing the cfDNA levels of the TERT promoter, NRAS, NFE2L2, and MET mutations may predict long-term treatment outcomes.^95,96 However, despite the importance of detecting molecular pathways implicated in disease progression, no precision medicines have been approved for first-line HCC treatment yet.

Various groups have defined different epigenetic changes in cfDNA of patients with HCC as a diagnostic tool (Table 1). Detecting methylation signatures may provide improved diagnostic specificity due to the inherent tissue specificity of methylation signatures, along with convenience in quantitating aberrant DNA methylation compared to mutation profiling. Different approaches have been utilised to detect early HCC, including profiling known HCC-associated methylated DNA markers (e.g., HOXA1, EMX1, TSPYL5, SEPT9, ECE1, PFKP, CLEC11A),^77,97 alongside clinicodemographic factors (GALAD score),⁹⁸ or profiling bisulfite-converted cfDNA in an unbiased fashion.^99,100 These early studies were limited due to a lack of tissue specificity and high cost. Thus, Cheishvilli et al. utilised publicly available DNA methylation datasets of tumour and normal tissues to define an HCC-specific methylation signature (“epiLiver”) which achieved 84.5% sensitivity and 95% specificity.¹⁰¹

In addition to profiling the mutation and methylation patterns of cfDNA, various groups have begun to integrate cfDNA fragmentation with methylation and mutation signatures.¹⁰² cfDNA “fragmentomes” refer to the entire map of circulating cfDNA fragments. Circulating cfDNA is highly fragmented with base pair sizes depending on nucleosome packing. This “fingerprint” provides insights into tissue-specific origins, e.g. patients with HCC show more 4-mer end motifs.¹⁰³ Early work demonstrated that cfDNA fragmentomes can be used to detect early HCC in at-risk patients (AUC 0.995; 96.8% sensitivity and 98.8% specificity).^103,104 This technology has also been shown to detect HBV-related HCC with 87.1% sensitivity and 88.4% specificity.¹⁰⁵ More intriguingly, the cfDNA fragmentome can also be probed in an unbiased manner to detect novel CpG methylation patterns which could be used to classify patients with HCC.¹⁰⁶ Recently, Wang et al. demonstrated the utility of a multiplex cfDNA mutation + methylation profiling technology, which was validated in a prospective cohort of 311 asymptomatic patients with HBV, demonstrating 80% sensitivity and 94% specificity for the detection of early HCC.¹⁰⁷ Lastly, Foda et al. developed an approach to infer genomic alterations driving HCC tumorigenesis from cfDNA and were able to detect HCC with 85% sensitivity and 80% specificity in a large-population of patients at high risk.¹⁰⁸

Overall, the main limitation of using cfDNA is that there is a strong correlation between quantity of ctDNA and lesion size, suggesting early-stage lesions are more difficult to detect than advanced disease.¹⁰⁹ Therefore, cfDNA may be a better tool for detecting recurrence and/or treatment response and resistance.

Non-coding RNA species

Non-coding RNAs (ncRNAs), such as microRNAs (miRNAs) (25–30 bp) and long non-coding RNAs (>200 bp) are not translated into protein but instead act as important regulatory elements involved in chromatin alterations, DNA transcription, and chromosomal looping. ncRNAs are also involved in diverse cellular processes implicated in cancer pathogenesis, including cellular proliferation, migration, invasion, and cell death. Their stability in the bloodstream has made them attractive candidates for liquid biopsy, along with their tissue-specific expression and ability to distinguish cancer from premalignancy.¹¹⁰ Various groups have defined panels of miRNAs (e.g., miR-21, miR-26a, miR-27a, miR-29a, miR-29c, miR-122, miR-133a, miR-143, miR-145, miR-192, miR-223, miR-505, and miR-801) which have demonstrated the ability to detect early-stage HCC,^111–113 predict overall¹¹⁴ and disease-free survival,¹¹⁵ and prognosticate following systemic therapy¹¹⁶ (Table 1). Other groups have utilised long non-coding RNA panels (e.g., AC005332.5, ELF3-AS1 and LINC00665) for diagnosis and prognostication in HBV-related HCC.¹¹⁷ However, the main challenges associated with translating an HCC ncRNA-based test have been the lack of standardised reporting methods, variation in RNA detection techniques, and incomplete data on their contribution to HCC.

Extracellular vesicles

Extracellular vesicles (EVs) are lipid bilayer nanovesicles (50 nm to >1,000 nm in size) spontaneously produced by nearly all mammalian cells and released into the extracellular milieu where they can travel systemically to other organs. The main subclasses of EVs include exosomes, microvesicles, and apoptotic vesicles, with each containing their own distinct nucleic acid species (cfDNA, cfRNA), proteins, and lipids. These cargo molecules participate in diverse cell-to-cell signalling circuits, thereby mediating pathophysiological states.¹¹⁸ EVs have become attractive liquid biopsy analytes given their cargo reflects the genomic and transcriptomic states of their parental cells of origin.¹¹⁹ Various groups have defined EV-associated nucleic acid and proteomic signatures either alone or in combination to detect HCC and monitor treatment responses^120,121 (Table 1). For example, EV-derived miRNA panels can classify HCC with >90% sensitivity and specificity.^120,122 Also, a meta-analysis of 16 studies determined that exosomal ncRNAs, particularly downregulated miRNAs, can prognosticate disease-free survival in HCC.¹²³ EVs also contain unannotated small RNA clusters which can classify HCC from controls at risk with 86% sensitivity and 91% specificity.¹²⁴ Additionally, Sun et al. demonstrated that HCC-specific EV subpopulations can be probed to detect early-stage HCC with 94% sensitivity and 84% specificity through profiling epithelial cell adhesion molecule+, CD147+, and GPC3+ EVs.¹²⁵ Moreover, EV-specific proteins have been shown to detect HCC and predict response to sorafenib +/− radiotherapy.^126,127 Lastly, TP53 mutations detected in EV-derived cfDNA have been used to predict microvascular invasion.¹⁹

Despite being able to probe for multiple molecular targets in one isolate, there are several limitations associated with utilising circulating EVs for liquid biopsy, including tissue specificity and purity. First, without a method to capture tumour-specific EVs, the cargo molecules isolated from EVs may be derived from other tissue sources, including non-malignant tissues, which spontaneously release EVs into the systemic circulation. Second, best practices for isolating EVs from biofluids are not well defined and can be labour intensive, costly, and result in impurities. However, emerging nanotechnologies may circumvent issues in EV isolation, detection, enumeration, and analyte profiling.¹²⁸

Update on translation of liquid biopsy to the clinic

Overview of multi-panel analyte-based tests for HCC

In this section, we discuss advanced technologies that employ multi-analyte panels for diagnosing and managing HCC, highlighting their potential and discussing their efficacy. Although these liquid biopsy systems, which identify DNA alterations or hypermethylation and often pair genomic data with protein markers or patient demographics, have not yet become a standard part of clinical practice, a few such platforms have received FDA approval as breakthrough diagnostic tests in recent years (Table 2).

Table 2.

Commercial Liquid Biopsy Platforms for Detection and Treatment Stratification in HCC

Assay	Analyte(s) Assessed	Intended Purpose	AUC (Sensitivity/Specificity)	Comments and Clinical Studies
Elecsys®GALAD	Protein markers: AFP, AFP-L3, DCP; gender; age	Early HCC detection	Overall: 0.947 (85.8/90.8) Early-stage: 0.913 (73.8/90.8)	FDA Breakthrough Device designation in 2022
Oncoguard® Liver	Methylation markers: HOXA1, TSPYLS, B3GALT6; AFP; sex	Early HCC detection	Overall: 0.94 (88/87) Early-stage: 0.92 (82/87)	NCT05064553 (ongoing)
HelioLiver™	Methylation markers: 77 CpG sites in 28 genes; protein markers: AFP, AFP-L3, DCP; patient demographics	Early HCC detection	Overall: 0.944 (85.2/91.2) Early-stage: 0.924 (75.7/91.2)	NCT05059665 (completed); NCT05053412, NCT03694600, NCT04539717 (ongoing)
HCCscreen™	Methylation markers; genomic alterations (SNVs, indels, translocations, viral integration sites)	Early HCC detection and mutation profiling	Overall: 0.93 (90/94) Early-stage: 0.95 (91/95)	FDA Breakthrough Device designation in 2020
HCCBloodTest	Methylated SEPT9	Early HCC detection	Overall: 0.93 (85.1/91.07) Early-stage: 0.863 (72.73/86.39)
Guardant360®	SNVs, indels, CNV, fusion events in 68 genes	Treatment stratification in advanced HCC and mutation profiling	N/A; 79% of patients had potentially actionable alterations	Point mutations: TP53, CTNNB1, PTEN, CDKN2A, ARID1A, MET; amplifications: CDK6, MYC, BRAF, RAF1, FGFR1, CCNE1, PIK3CA, HER2

Multi-panel scores based on protein marker analysis

The Elecsys-GALAD score is derived by combining gender and age with an immunoassay that measures serum markers AFP, AFP-L3, and DCP (also known as protein-induced by vitamin K absence-II, or PIVKA-II). A multicentre study comparing the performance of the Elecsys PIVKA-II and Elecsys AFP assays in diagnosing HCC determined that the PIVKA-II assay had an overall sensitivity of 86.9% (95% CI 80.8–91.6) and specificity of 83.7% (77.9–88.4), while the AFP assay achieved an overall sensitivity of 51.8% (44.0–59.5) and specificity of 98.1% (95.1–99.5). When combining both assays, the overall sensitivity for detecting HCC increased to 92% with a specificity of 82%, outperforming the two individual assays.¹²⁹ A scoring algorithm based on GALAD received a Breakthrough Device designation from the FDA in 2020 for use as an in vitro diagnostic. A large meta-analysis of 15 original studies of 19,021 patients showed that the GALAD score had a pooled sensitivity of 0.82 (95% CI 0.78–0.85) for any-stage HCC and 0.73 (0.66–0.79) for early-stage HCC, easily surpassing the 38% detection rate of AFP alone for early-stage HCC.¹³⁰ Thus, the robust sensitivity and efficiency of the Elecsys-GALAD score positions it as a promising surveillance tool for HCC. Lastly, the Elecsys-GAAD is another multi-panel diagnostic test for early-stage HCC which incorporates gender and age with immunoassays for just two biomarkers instead of three, AFP and PIVKA-II.¹³¹ The performances of Elecsys-GAAD and Elecsys-GALAD are similar, suggesting that the AFP-L3 assay is only a minor contributor towards HCC detection. We await further validation of these results.

Combining insights from cfDNA methylation alterations and protein markers

Other technologies for HCC diagnosis integrate cfDNA methylation state, DNA alterations, and serum marker levels. One such platform, Oncoguard Liver, detects early-stage HCC through a multi-target blood test that measures a panel of methylation sites and serum markers to produce a qualitative result. In a 2021 study of this platform, the panel included four methylation markers (HOXA1, TSPYLS, EMX1, and the reference marker B3GALT6), AFP, and AFP-L3. With these values, the test had an overall sensitivity of 80% (72–86) and early-stage sensitivity of 71% (60–81) at a specificity of 90%, showing greater sensitivity than AFP, AFP-L3, DCP, and GALAD.⁷⁷ A later study modified the Oncoguard Liver panel by removing EMX1 and AFP-L3 and considering patient sex. In a cohort of 156 patients with HCC and 245 control patients, the overall sensitivity was 88% (82–92) and early-stage sensitivity was 82% (72–89) at a specificity of 87% (82–91). Furthermore, the test maintained its performance across multiple liver aetiologies.^77,132 Following the results of this study, the ALTernative to Ultrasound (ALTUS) study (NCT05064553) was initiated to measure the performance of Oncoguard Liver as a diagnostic tool in patients undergoing HCC surveillance with ultrasound or CT/MRI. This study aims to enrol 3,000 patients across 60 sites in the US and will allow for large-scale, longitudinal assessment of the multi-target blood test.¹³³

Another diagnostic platform is HelioLiver, a multi-analyte test designed for monitoring patients at risk of HCC, especially those with cirrhosis, chronic viral hepatitis, and metabolic dysfunction-associated steatotic liver disease.^133,134 The test targets 77 methylation sites in 28 target genes using next-generation sequencing libraries and measures AFP, AFP-L3, and DCP from immunoassays. Along with patient demographics, these data are used in a diagnostic algorithm to produce a qualitative test result. The Performance Evaluation of HelioLiver^™ Test for Detection of HCC (ENCORE) (NCT05059665), a prospective phase II biomarker study, evaluated HelioLiver in a cohort of 122 patients with HCC and 125 control patients with benign liver disease. The test achieved an overall sensitivity of 85.2% (77.8–90.4) and specificity of 91.2% (84.9–95.0), with an early-stage sensitivity of 76% (59.9–86.7), demonstrating greater sensitivity than AFP, AFP-L3, DCP, and GALAD for both all-stage HCC and early-stage disease.⁹⁸ Other than ENCORE, three other clinical studies of HelioLiver have been initiated, two of which aim to compare the test results with ultrasound or MRI (NCT05053412, NCT03694600, NCT04539717). Insights from these studies will help decipher the potential role of HelioLiver in clinical applications.

HCCscreen analyses both cfDNA methylation state and genomic alterations, such as single nucleotide variants, indels (insertions and deletions), translocation, and viral genome integration.¹³⁵ It uses mutation capsule technology, which simultaneously identifies alterations in both DNA sequence and methylation state through highly sensitive multiplex reactions, resulting in a molecular profile of cancer.⁹⁴ In 2020, the FDA designated HCCscreen as a Breakthrough Device to accelerate its approval process. A later study by Wang et al. assessed HCCscreen in a cohort of 436 patients with HBV, 148 of whom had HCC. Using a panel of methylation markers and alterations prevalent in HCC tumours, the group constructed an HCC detection algorithm, which demonstrated an overall sensitivity of 90% (0.79–0.96) and specificity of 94% (0.90–0.97). The combined methylation marker and alteration panel performed better than AFP and either the methylation or alteration panels alone in detecting HCC.¹⁰⁷ Expanding the clinical application of HCCscreen, an early HCC detection programme has been ongoing in China, where 150,000 tests will be administered in local communities over 3 years.

Some platforms are harnessing the methylation state of cfDNA alone to determine the presence of HCC. One such technology is HCCBloodTest, which assesses hypermethylation of the SEPT9 (Septin 9) gene, a driver of HCC carcinogenesis, in bisulfite-converted DNA derived from cfDNA in plasma. As bisulfite treatment of DNA converts unmethylated cytosine to uracil sulfonate and does not alter 5-methylcytosine, bisulfite-converted DNA can be used to ascertain the degree of methylated SEPT9 after PCR amplification, which is converted to a qualitative result regarding HCC status.¹³⁶ In an initial observational study, HCCBloodTest achieved a sensitivity of 94.1% (83.8–98.8) and specificity of 94.4 (77.2–90.1), while a subsequent case-control study with age- and gender-matched patients reported similar results, with a sensitivity of 85.1% (71.1–93.8) and specificity of 91.07% (80.4–97.0).¹³⁷ In both cases, HCCBloodTest had significantly higher accuracy than AFP for detecting HCC. Notably, there was a difference in the specificity of HCCBloodTest between cirrhosis aetiologies, as the specificity ranged from 86.4% for viral hepatitis-associated cirrhosis to 39.4% for metabolic dysfunction-associated steatohepatitis-associated cirrhosis.¹³⁷ Though the cause for the varying specificity was uncertain, HCCBloodTest should continue to be validated in different patient populations.

Liquid biopsy based on cfDNA-targeted exome sequencing

Although most plasma- or serum-based diagnostics for HCC are intended to aid early detection, this technology can also be used for treatment stratification. Guardant360 CDx is the first FDA-approved comprehensive liquid biopsy for all advanced solid tumours.¹³⁸ The approved assay utilises next-generation sequencing and profiles single nucleotide variants and indels in 55 genes, copy number variants in two genes, and fusion events in four genes, though alterations in other genes have been identified with the assay. In addition, Guardant360 CDx has been approved as a companion diagnostic for breast cancer and non-small cell lung cancer (NSCLC) to identify whether patients’ mutation profiles are suitable for targeted treatments. In a study by Bauml et al., Guardant360 CDx was used to determine if patients with NSCLC had a KRAS driver mutation that could be targeted with sotorasib, a small molecule inhibitor. Out of 109 patients who were successfully tested, 78 patients were positive for the mutation and began sotorasib therapy, achieving a 36.4% (25.7–48.1) objective response rate.¹³⁹ The study also assessed agreement between plasma testing with Guardant360 CDx and tissue analysis with PCR, determining that the overall percent agreement was 0.82 (0.76–0.87).¹³⁹

While targeted therapies for driver mutations in HCC have yet to become the standard of care, this study in patients with NSCLC provides a framework for the potential role of liquid biopsies in stratifying HCC treatment as more precision therapies are developed. In fact, a study by Ikeda et al. evaluated 14 patients with advanced HCC using Guardant360, which features an expanded panel of 68 oncogenes and tumour suppressor genes, to identify actionable alterations. The assay identified somatic alterations in all patients, with 79% of patients having a potentially actionable alteration. Five patients received treatment based on the assay results. One patient with an early-stage HCC began treatment with cabozantinib, a kinase inhibitor, after ctDNA analysis revealed a mutation in the proto-oncogene MET. After 8 weeks, a second ctDNA analysis showed this mutation had disappeared, suggesting a response to therapy.¹⁴⁰ As more precision therapies are developed and applied to actionable mutations in HCC, liquid biopsies are expected to play a greater role not only in aiding early diagnosis, but also in selecting treatment, monitoring response, and examining tumour evolution.

There are numerous other multi-cancer detection tests under rigorous evaluation that promise to revolutionise cancer diagnostics. Notable among them are the methylation based MCED test,¹⁴¹ CancerSEEK,¹⁴² and FoundationOneCDx (F1CDx),¹⁴³ to name a few. It is worth noting that, just in 2022, over 200 clinical trials examining a diverse range of liquid biopsy biomarkers were initiated. These continuous advances in liquid biopsy technology, coupled with the sheer volume of ongoing research, signify a promising future for non-invasive cancer detection and management.

Challenges in translating liquid biopsy to the clinic

The rising use of liquid biopsy in liver cancer care showcases its potential, yet several challenges remain, of which we would like to highlight three. First, the majority of liquid biopsy tests still lack robust evidence supporting their clinical validity and utility, relegating their use to research settings. This challenge is likely to be answered by the various ongoing multicentre clinical studies. Second, many of these assays either focus on a single analyte or use a multiparametric approach, but do not have easy scoring criteria. The application of advanced statistical tools employing high-dimensional machine learning techniques can potentially overcome this challenge. The third challenge is the substantial variations in assay results, which complicate clinical interpretations. Homogenising analyte preparatory procedures and standardising reporting can allow for the comparison or integration of results from diverse studies. Collaborative data sharing is equally vital to further investigate integration of liquid biopsy assays into clinical workflows.

Conclusions and future perspectives

Molecular diagnostics, whether through tissue or liquid biopsy, are likely to play a larger role in the management of patients with HCC in the near future. We highlight four specific areas where tissue and/or liquid biopsy can potentially make an impact (Fig. 2). The first is in HCC surveillance. The limitations of imaging surveillance and the challenges with patient adherence to surveillance intervals underscore the appeal of a non-invasive blood-based test that can potentially be employed by primary care physicians and community practices. While liquid biopsy must address current sensitivity, accessibility, and cost issues, it has the potential to complement, if not replace, imaging-based surveillance. Moreover, it can aid in establishing a diagnosis for indeterminate nodules and obviate the need for invasive tissue biopsies. Second, in early-stage HCC where patients undergo surgery or resection, tissue-informed liquid biopsy tests can play a crucial role in detecting minimal residual disease and monitoring for recurrence, thus identifying individuals who may benefit from adjuvant therapies. Third, in patients with intermediate- or advanced-stage HCC receiving targeted or immune-based therapies, biomarker expression analysis from both tissue and liquid biopsy can facilitate the selection of appropriate candidates for specific treatments, leading to improved response rates and reduced side effects. Lasty, the utility of liquid biopsy extends to monitoring the response to therapy at all stages of HCC, making biopsy-based molecular tests a valuable tool in the comprehensive management of this complex disease. In conclusion, the advent of molecular diagnostics through tissue or liquid biopsy heralds a promising era in HCC diagnosis and management, offering sensitive and versatile tools that can be employed across all stages of the disease.

Fig. 2. Potential future applications of liquid or tissue biopsy-based molecular diagnostic tests in HCC.

cfDNA, cell-free DNA; CTCs, circulatory tumour cells; EVs, extracellular vesicles; HCC, hepatocellular carcinoma; MRD, minimal residual disease.

Keypoints.

• HCC is often diagnosed at advanced stages and is associated with poor clinical outcomes, highlighting the critical need for access to molecular insights through either tissue or liquid biopsy.

• Utilising this molecular data can pave the way for reliable biomarker development, thereby improving early detection and refining treatment strategies for HCC.

• While questions regarding the need for routine HCC biopsies persist, increasing the frequency of tissue biopsies, especially in clinical trials, could expedite the discovery of novel biomarkers and personalised treatments.

• Liquid biopsy is transforming cancer care by enabling non-invasive, real-time tumour monitoring and early detection of circulating tumour cells, cell-free DNA, non-coding RNAs, and extracellular vesicles.

• Multi-analyte panels for HCC, which combine genomic data with protein markers or patient demographics, are gaining traction, with some recently receiving FDA breakthrough diagnostic test approval.

• With the increase in biomarker-stratified trials, tissue biopsies may become pivotal in matching patients to appropriate treatments, and liquid biopsies are likely to play complementary roles in early HCC detection and monitoring of treatment response.

Abbreviations

AFP

alpha-fetoprotein

AFP-L3

fucosylated fraction of AFP

ARID

AT-rich interactive domain-containing protein

cfDNA

cell-free DNA

CTCs

circulating tumour cells

ctDNA

circulating tumour DNA

DCP

des-gamma carboxyprothrombin

EVs

extracellular vesicles

GPC-3

glypican-3

HCC

hepatocellular carcinoma

miRNA/miR

microRNA

ncRNA

non-coding RNA

NSCLC

non-small cell lung cancer

PD-L1

programmed death ligand 1

TERT

telomerase reverse transcriptase

TGFβ

transforming growth factor-β