Early screening for liver cancer must be performed

Abstract

Hepatocellular carcinoma (HCC) is imposing a growing global health burden, with China accounting for nearly half of incident cases and mortality worldwide. Early screening is critical to improving survival in high-burden regions. However, the global standardized screening rate for high-risk populations is less than 24%, and HCC screening currently faces severe challenges. We synthesize recent advances in HCC screening, including optimized serum biomarkers, evolving imaging techniques, and validated models. Emerging liquid biopsy technologies and artificial intelligence further demonstrate considerable promise for enhancing noninvasive detection efficacy. Multifaceted collaboration among policymakers, healthcare systems, and communities is essential to implement effective screening programs and ultimately improve survival outcomes.

Key Words: Hepatocellular carcinoma; Early screening; Risk stratification; Serum biomarkers; Liquid biopsy; Artificial intelligence

Core Tip: Hepatocellular carcinoma imposes a significant disease burden globally, particularly in China, where early screening faces challenges of low screening rates and insufficient sensitivity of traditional methods. This review focuses on the potential of innovative strategies, including optimized combinations of serum biomarkers, advanced imaging techniques, and liquid biopsy, to enhance early diagnosis rates. We emphasize the need for multidisciplinary collaboration and risk stratification management to improve screening efficacy.

INTRODUCTION

Primary liver cancer, especially hepatocellular carcinoma (HCC), is one of the leading causes of cancer-related deaths worldwide, with a regionally uneven disease burden. In 2022, more than 860000 new cases of liver cancer were diagnosed globally[1], with China accounting for 45.3% of these cases and 39.1% of global deaths[2]. The total age-standardized death rate in China is significantly higher than the worldwide age-standardized death rate[3]. Epidemiological models predict that without effective intervention, the number of new liver cancer cases worldwide will rise to 1.4 million annually by 2040, with 1.3 million deaths[4].

The prognosis of liver cancer is highly dependent on the stage at diagnosis. Early-stage lesions can be treated with curative therapy to extend survival. Over 60% of liver cancer patients diagnosed in Japan have early-stage HCC, and the 5-year overall survival rate for early-stage HCC patients who undergo curative treatment can be as high as 43%[5], which is significantly better than in China. The survival disparity between China and Japan directly reflects the gap in early diagnosis rates. Japan’s successful experience in early liver cancer screening offers valuable insights for adoption and adaptation[6]. HCC screening is crucial for early diagnosis, improving curative rates, extending patient survival, and enhancing long-term prognosis[7]. However, current liver cancer screening strategies face challenges in clinical application, such as low implementation rates and poor sensitivity in detecting early-stage HCC. Additionally, the rapid rise in metabolic-associated fatty liver disease has posed new challenges for traditional HCC screening methods[8-10]. To address these challenges, this paper comprehensively reviews the main technological advancements and challenges in early HCC screening, aiming to promote further development of early HCC screening efforts and improve survival rates and quality of life for HCC patients.

CURRENT SCREENING METHODS

Serum biomarkers

In the early screening of liver cancer, serological marker testing has been indispensable in clinical practice due to its non-invasiveness, convenience, and low cost. Traditional markers such as alpha-fetoprotein (AFP) have played an important role in liver cancer screening, but other potential novel markers have been discovered and applied to clinical practice as the molecular mechanisms of HCC have been studied more deeply[11].

AFP

As a conventional HCC screening marker, AFP maintains high diagnostic utility[12]. Persistent AFP elevation (≥ 10% over 3-6 months) or levels ≥ 20 ng/mL in benign liver disease significantly elevate 6-month HCC risk[13]. Some researchers have developed a biosensor based on droplet evaporation, which exhibits high selectivity, stability, and reproducibility in detecting AFP, demonstrating great potential for clinical application in early screening[14]. However, as the proportion of AFP-negative liver cancers increases, studies have found that the sensitivity of AFP detection for early-stage HCC is only 56.3%, limiting its clinical application[15]. Therefore, exploring more sensitive and specific biomarkers, and their combinations, for liver cancer screening is particularly important.

Des-gamma-carboxy prothrombin

This vitamin K-deficient protein is a promising biomarker for HCC. Increasing research suggests that it outperforms traditional AFP[16-18]. Asia-Pacific consensus guidelines endorse des-gamma-carboxy prothrombin (DCP) for AFP-negative HCC detection, particularly when combined with AFP[19]. Despite superior accuracy, DCP requires complementary biomarkers for optimal screening efficacy.

Lens culinaris agglutinin-reactive fraction of AFP

Lens culinaris agglutinin-reactive fraction of AFP (AFP-L3) is a more liver cancer-specific biomarker, but it has relatively low sensitivity[20]. A phase III biomarker study showed that the sensitivity of AFP-L3% was only 35.7% in the 12 months prior to HCC diagnosis[21]. To improve the detection accuracy and clinical application value of AFP-L3%, researchers have developed an electrochemical aptamer sensor, which can precisely detect AFP-L3% and accurately diagnose liver cancer at an early stage[22].

The well-documented limitations of conventional markers in achieving high sensitivity without compromising specificity remains a significant diagnostic challenge. AFP remains the most accessible and inexpensive test, yet its utility is curtailed by suboptimal sensitivity, particularly in early-stage and non-viral HCC. DCP demonstrates superior sensitivity to AFP in many studies, though its performance can vary across etiologies. AFP-L3% offers high specificity for HCC but is hampered by low absolute sensitivity, limiting its standalone use. The insufficiency of existing biomarkers has driven extensive research into other novel biomarkers. Golgi protein 73 (GP73) participates in liver pathophysiology and may drive hepatocarcinogenesis via tumor microenvironment interactions[23]. Clinically, combining AFP, DCP, and GP73 enhances diagnostic accuracy for hepatitis B virus (HBV)-related HCC, correlating with histopathological findings[24]. Recently, researchers have developed a competitive electrochemical sensor for detecting GP73 with a detection limit of 0.15 pg/mL, demonstrating significant potential for clinical applications[25]. Due to its unique cell membrane localization, glypican-3 (GPC3) provides insights for the early screening of HCC[26]. Zhao et al[27] found that the triple detection of GP73, GPC3, and AFP showed good diagnostic value. In recent years, the development of highly selective GPC3-targeted peptides has made breakthrough progress. As a safe and effective liver cancer imaging agent, it has shown significant advantages in the early screening of HCC[28]. Heat shock protein 90α (HSP90α) is highly expressed in various malignant tumor tissues and promotes the malignant transformation of tumor cells[29]. It has important application value in early tumor screening. Han et al[30] demonstrated high diagnostic efficacy of HSP90α for HCC (sensitivity: 93.2%; specificity: 85.4%; area under the curve: 0.931). In the preliminary research, these promising potential markers (GP73, GPC3, and HSP90α) have demonstrated high diagnostic performance. However, whether they can be translated into routine clinical practice remains to be further verified by large-scale, multi-center trials and the standardization of detection protocols.

Imaging advances

Abdominal ultrasound (US), computed tomography (CT), and magnetic resonance imaging (MRI) are commonly used screening methods for early detection of liver cancer[31]. Abdominal US is recommended as the standard monitoring method for liver cancer, while CT and MRI are often used as supplementary methods for early screening. Traditional US, with its non-invasive nature, widespread availability, and low cost, is recommended by mainstream international guidelines as the standard monitoring tool for liver cancer in high-risk populations[32,33]. However, due to factors such as operator experience variability, abdominal gas in patients, and obesity, the detection rate of US is low for liver cancer. As the global etiological spectrum of HCC rapidly shifts toward non-viral fatty liver disease, the limitations of US for liver cancer screening are further exacerbated. Approximately 20% of cirrhosis patients, particularly those with obesity-related, alcohol-associated, or non-alcoholic steatotic liver disease, demonstrate reduced ultrasonographic detectability of HCC nodules[34]. Even with adjunct AFP testing, early HCC detection sensitivity remains suboptimal (74.1%)[15]. Contrast-enhanced US (CEUS) addresses these constraints through dynamic visualization of tumor microvascular perfusion following microbubble contrast administration, achieving superior diagnostic accuracy[35]. A meta-analysis of 23 studies confirmed that CEUS has a sensitivity of 77.8% and specificity of 93.8% for diagnosing liver cancer[36], making it an important tool for resolving uncertainties when small liver nodules are detected by MRI but cannot be definitively diagnosed[37].

While multiphase CT allows hemodynamic assessment of HCC, radiation concerns preclude routine screening use. Studies have shown that when the liver cancer risk index is ≥ 2.33, low-dose CT has enhanced performance vs US[38]. While conventional MRI offers high HCC detection rates, its cost and time limitations hinder routine surveillance. Abbreviated MRI (AMRI), particularly non-contrast AMRI, addresses this by reducing scan time to < 10 minutes and eliminating contrast-related risks[39,40]. Studies demonstrate non-contrast-AMRI’s superior annual sensitivity over semi-annual US, while gadoxetic acid-enhanced MRI (EOB-MRI) provides cost-effective early detection in metabolic-related cirrhosis patients[41]. Overall, MRI outperforms US in HCC surveillance in terms of earlier staging and reduced false positives[42]. Collectively, these imaging modalities present a trade-off between accessibility, cost, and accuracy.

RISK STRATIFICATION AND PREDICTION MODELS

Several liver cancer risk assessment tools have been developed (Table 1). By implementing risk stratification for patients with chronic liver disease, these tools can accurately distinguish between low-, medium-, and high-risk populations for liver cancer, thereby enabling the development of tailored screening protocols for different risk levels. This approach quantifies the probability of HCC occurrence in patients over the next few years. A Chinese research team has proposed the first liver cancer risk assessment tool that spans multiple etiologies and ethnicities, and the aMAP score has gained widespread international recognition[43].

Table 1 Validated risk stratification models for hepatocellular carcinoma development.

Model	Region	Derivation cohort	Variables included	Risk stratification cutoffs	Ref.
aMAP	Global	Treated chronic liver disease patients	Age, sex, bilirubin, albumin, platelet count	Low: 0-50. Intermediate: 50-60. High: 60-100	Fan et al[43], 2020
GAG-HCC	Hong Kong, China	CHB patients	Age, sex, HBV DNA level, core promoter mutations, cirrhosis	Low: < 100. High: ≥ 100	Yuen et al[72], 2009
CU-HCC	Taiwan, China, South Korea, Hong Kong	Non-cirrhotic CHB patients	Age, sex, platelet count, ALT level, HBeAg status	Low: < 5. Intermediate: 5-20. High: > 20	Kim et al[73], 2024
REACH-B	Taiwan, China	Untreated non-cirrhotic CHB patients	Age, albumin, bilirubin, HBV DNA, cirrhosis	Low: ≤ 5. Intermediate: 6-11. High: ≥ 12	Wong et al[74], 2010
LSM-HCC	Hong Kong, China	CHB patients	LSM, age, serum albumin, HBV DNA	Low: < 11. High: ≥ 11	Wong et al[75], 2014
mREACH-B	South Korea	CHB patients	Age, sex, HBeAg, LSM, ALT	Low: < 10. High: ≥ 10	Jung et al[76], 2015
HCC-RESCUE	South Korea	CHB patients on oral antivirals	Age, sex, cirrhosis	Low: < 5%. Intermediate: 5%-20%. High: ≥ 20%	Sohn et al[77], 2017
AGED	China	HBsAg-positive CHB patients	Age, sex, HBeAg status, HBV DNA level	Low: 0-4. Intermediate: 5-9. High: 10-12	Fan et al[78], 2019
CAMD	Hong Kong, Taiwan	CHB patients on ETV/TDF	Age, sex, cirrhosis, diabetes	Low: 0-7. Intermediate: 8-13. High: > 13	Hsu et al[79], 2018
PAGE-B	Europe	Caucasian CHB patients on antivirals (≥ 12 months)	Age, sex, platelet count	Low: ≤ 9. Intermediate: 10-17. High: ≥ 18	Papatheodoridis et al[80], 2016
mPAGE-B	South Korea	Asian CHB patients on antivirals (≥ 12 months)	Age, sex, platelet count, albumin	Low: 0-8. Intermediate: 9-12. High: ≥ 13	Kim et al[81], 2018
REAL-B	Asia-Pacific, United States	CHB patients on antivirals	Sex, age, alcohol use, diabetes, cirrhosis, platelet count, AFP	Low: 0-3. Intermediate: 4-7. High: 8-13	Yang et al[82], 2020
RWS-HCC	Singapore	CHB patients	Age, sex, cirrhosis, AFP	Low: < 4.5. Significant: ≥ 4.5	Poh et al[83], 2016
THRI	Canada	Cirrhosis patients	Age, sex, etiology, platelet count	Low: < 120. Intermediate: 120-240. High: > 240	Sharma et al[84], 2017
GES	Egypt	HCV cirrhosis or advanced fibrosis patients	Age, sex, albumin, AFP, pretreatment fibrosis stage	Low: ≤ 6. Intermediate: 6-7.5. High: > 7.5	Shiha et al[85], 2020
George et al	United States	NAFLD or ALD cirrhosis patients	Age, sex, BMI, diabetes, platelet count, albumin, AST/ALT ratio	Low: 0%-1%. Intermediate: > 1%-3%. High: > 3%	Ioannou et al[86], 2019
PAaM	United States	Cirrhosis (mixed etiology)	Prognostic liver secretome signature + AFP, age, male sex, ALBI, platelet count	Low: < 4.318. High: ≥ 5.072	Fujiwara et al[87], 2025

HCC: Hepatocellular carcinoma; CHB: Chronic hepatitis B; HBV: Hepatitis B virus; ALT: Alanine aminotransferase; HBeAg: Hepatitis B e antigen; LSM: Liver stiffness measurement; HBsAg: Hepatitis B surface antigen; ETV: Entecavir; TDF: Tenofovir disoproxil fumarate; AFP: Alpha-fetoprotein; HCV: Hepatitis C virus; NAFLD: Non-alcoholic fatty liver disease; ALD: Alcohol-associated liver disease; BMI: Body mass index; AST: Aspartate aminotransferase; ALBI: Albumin-bilirubin score.

Risk stratification optimizes the selection of screening populations and timing but cannot predict the diagnosis of HCC. To overcome the limitations of single serum markers, researchers have developed predictive diagnostic models based on serum markers combined with clinical variables (Table 2), achieving a closed-loop management system from “risk warning” to “lesion identification”. Taking the GALAD model, which includes five variables (AFP, AFP-L3%, DCP, age, and gender), as an example, this model aims to significantly enhance the early diagnostic efficacy of HCC through comprehensive assessment[44].

Table 2 Summary of hepatocellular carcinoma prediction models based on serum biomarkers.

Model name	Components/equation	Ref.
GALAD	-10.08 + 0.09 × age + 1.67 × gender + 0.04 × AFP-L3% + 2.34 × log10 AFP + 1.33 × log10 DCP	Johnson et al[88], 2014
GAAP	9.134 + 0.892 × gender + 0.073 × age + 1.057 × log10 AFP + 1.015 × log10 DCP	Liu et al[89], 2020
AALP	7.245 + 0.056 × age + 0.431 × log10 AFP + 3.112 × AFP-L3 + 1.162 × log10 PIVKA-II	Ren et al[90], 2023
ASAP	-7.577 + 0.047 × age - 0.576 × gender + 0.422 × lnAFP + 1.105 × lnDCP	Sun et al[91], 2025
GAAD	Gender, age, PIVKA-II, AFP	Piratvisuth et al[92], 2023
GAADPB	0.176 + 0.162 × gender + 0.002 × age + 0.178 × log10 AFP + 0.164 × log10 DCP - 0.007 × TP - 0.002 × TB	Chen et al[93], 2023
BALAD	Bilirubin, albumin, AFP-L3, AFP, DCP	Toyoda et al[94], 2006
BALAD-2	0.02 × (AFP-2.57) + 0.012 × (AFP-L3 - 14.19) + 0.19 × (lnDCP - 1.93) + 0.17 × (Bilirubin1/2 - 4.5) - 0.09 × (albumin - 35.11)	Fox et al[95], 2014
Doylestown	1/{1 + exp[-(-10.307 + 0.097 × age + 1.645 × gender + 2.315 × log10 AFP + 0.011 × ALP - 0.008 × ALT)]}	Wang et al[96], 2016
Doylestown Plus	Age, logAFP, PEG-precipitated IgG, fucosylated kininogen	Singal et al[97], 2022
HES V1.0	Age, current AFP level, AFP change rate, ALT, platelets	Tayob et al[98], 2021
HES V2.0	Age, current AFP level, AFP change rate, ALT, platelets, AFP-L3, DCP	El-Serag et al[99], 2025

AFP: Alpha-fetoprotein; AFP-L3: Lens culinaris agglutinin-reactive fraction of alpha-fetoprotein; DCP: Des-gamma-carboxy prothrombin; PIVKA-II: Protein induced by vitamin K absence or antagonist-II; TP: Total protein; ALP: Alkaline phosphatase; ALT: Alanine aminotransferase; IgG: Immunoglobulin G.

LIQUID BIOPSY IN EARLY DETECTION

With the advancement of liquid biopsy technology, novel biomarkers such as circulating tumor DNA (ctDNA), circulating tumor cells (CTCs), microRNA (miRNA), and long non-coding RNA (lncRNA) have provided new options for screening and early diagnosis of liver cancer patients[45].

ctDNA primarily originates from the apoptosis and necrosis of tumor cells. Luo et al[46] developed the DELFI model based on changes in cfDNA fragments, and the model demonstrated good performance in HCC screening in healthy populations and early diagnosis of HCC in high-risk populations. A research team at the University of Oxford developed a multimodal ctDNA detection method capable of simultaneously analyzing genomic and methylomic data[47]. However, the high detection cost limits its clinical application.

CTCs are tumor cells that detach from the primary or metastatic site and enter the peripheral circulatory system. Currently, various CTC detection platforms based on immunological affinity and biophysical characteristics have been developed, significantly improving the efficiency of CTC identification and detection[48]. However, most studies are still limited to small-scale, single-center case-control studies, and there are significant demographic differences, making it difficult to validate and generalize the research results[49].

miRNAs are endogenous non-coding RNAs with gene regulatory functions. In hepatitis B-related HCC, serum miRNome profiling has identified 30 upregulated and 61 downregulated species[50]. Li et al[51] used next-generation sequencing to isolate fucosylated extracellular vesicles from serum and identified five miRNAs as biomarkers for HCC detection. Nevertheless, miRNA biomarker translation faces challenges due to molecular heterogeneity and individual sample limitations, necessitating further optimization of combinatorial detection strategies.

H19 is the earliest imprinted gene identified among lncRNAs[52]. It exerts oncogenic effects through multiple signaling pathways, including regulation of epigenetic modifications and the H19/miR-675 axis, and may serve as a potential biomarker for liver cancer. Lnc-MyD88 shows diagnostic promise in HBV-HCC, achieving 80.95% sensitivity for AFP-negative cases[53]. Research on lncRNAs remains fragmented and lacks an established framework, with many gaps yet to be filled in this field of study.

Liquid biopsy technologies show promise for clinical implementation potential but lack randomized trial validation for early intervention guidance. Current protocols recommend shortened follow-up intervals for positive cases. While broader adoption requires further evidence, these high-sensitivity/specificity assays represent transformative tools for HCC diagnosis and management.

ARTIFICIAL INTELLIGENCE FOR SCREENING ENHANCEMENT

Artificial intelligence (AI) has demonstrated significant potential in the early screening of liver cancer[54]. Machine learning (ML) is a computational model inspired by the biological structure and function of the human brain. In deep learning technology, convolutional neural networks have proven to be particularly efficient in processing visual data[55]. Radiological features from CT and MRI scans, when combined with ML models, can aid in the differential diagnosis of HCC[56]. A meta-analysis indicated that models based on convolutional neural networks and those combining CEUS with AI exhibit good sensitivity[57]. Cheng et al[58] developed a model using ML methods based on personalized biological pathway analysis and regularized regression, achieving high-precision diagnosis of HCC. Wang et al[59] proposed a bioinformatics strategy named TopMarker, which calculates and screens biomarkers for HCC based on differential network topological parameters. Additionally, AI systems outperform clinicians in interpreting imaging data, minimizing diagnostic discrepancies. As technology continues to advance, the application of AI in liver cancer screening will provide more precise and efficient solutions for clinical diagnosis.

TISSUE BIOPSY: DIAGNOSTIC APPLICATIONS

Although tissue biopsy is not a first-line routine method for early screening of liver cancer, it can provide clear histological diagnosis and molecular typing evidence in patients with atypical imaging findings or the absence of cirrhosis. The American Association for the Study of Liver Diseases guidelines support the use of confirmatory liver biopsy in HCC, particularly for patients with liver nodules without cirrhosis or hepatitis B infection[60]. The advantage of liver biopsy lies in its ability to directly obtain tumor tissue structure, enabling precise pathological subtype classification and molecular target identification[61]. In recent years, tissue biopsy has been deeply integrated with molecular pathology. Based on multi-omics analysis, molecular markers for precancerous lesions and early-stage liver cancer can be identified, significantly improving diagnostic sensitivity and prognostic prediction capabilities[62].

SCREENING CHALLENGES AND FUTURE DIRECTIONS

Figure 1 outlines a proposed clinical pathway for screening and diagnosis in high-risk populations. However, despite the availability of comprehensive guidelines, various risk scoring systems, and screening technologies, the current state of liver cancer screening is not encouraging. Globally, less than a quarter (24%) of cirrhosis patients undergo HCC monitoring[63], highlighting the low screening rate and the significant gap between guideline recommendations and clinical practice[64]. Even in the United States, where medical resources are relatively abundant, only 8.78% of cirrhosis patients undergo monitoring[65]. This challenge is exacerbated by two technical issues: Metabolic dysfunction-associated steatotic liver disease (MASLD) and alcoholic liver disease have replaced viral hepatitis as the primary causes, and the high proportion of non-cirrhotic HCC further undermines the sensitivity of traditional US[66]. Lim et al[67] found that the survival benefit of HCC monitoring is more pronounced for HBV- and hepatitis C virus-related HCC patients, while for MASLD-related HCC patients, the survival benefit from monitoring is less consistent. Unfortunately, there is currently insufficient evidence supporting HCC monitoring for MASLD. Furthermore, rigid semiannual surveillance protocols lack risk adaptivity, and evidence confirms that annual monitoring is a more cost-effective strategy for patients with compensated cirrhosis or advanced fibrosis[68].

Figure 1 Clinical pathway for the integrated management of early hepatocellular carcinoma screening. AFP: Alpha-fetoprotein; DCP: Des-gamma-carboxy prothrombin; AFP-L3: Lens culinaris agglutinin-reactive fraction of alpha-fetoprotein; ctDNA: Circulating tumor DNA; CTCs: Circulating tumor cells; miRNAs: MicroRNAs; CT: Computed tomography; MRI: Magnetic resonance imaging; HCC: Hepatocellular carcinoma.

The low rate of liver cancer screening can be attributed to several factors. From the patient’s perspective, many residents lack sufficient health awareness and medical knowledge, leading to an insufficient understanding of the importance of liver cancer screening. Additionally, individuals who require follow-up after an abnormal initial screening often overlook the importance of regular follow-up visits. In a cohort study of cirrhosis patients, only 37.2% underwent routine outpatient follow-up in the year prior to liver cancer diagnosis. Even among those who did receive follow-up, 32.6% failed to undergo regular screening[69]. A study in Argentina involving 301 high-risk HCC patients found that 25% of patients were unaware of their chronic liver disease history at the time of liver cancer diagnosis[70]. Additionally, factors such as misunderstandings about HCC screening, time and financial costs, and difficulties in scheduling US examinations also affect screening compliance. Improving the screening rate for liver cancer requires not only the cooperation of patients, but also the efforts of multiple sectors, including medical institutions at all levels, the government, and communities. Methods such as lectures, online promotions, patient education, social publicity, WeChat public platforms, and telephone follow-ups can help promote public awareness of liver cancer screening. In a multicenter clinical trial, researchers mailed letters containing information and screening recommendations to high-risk patients. If there was no response, they followed up with phone calls to remind high-risk patients to schedule US examinations. The results showed that the monitoring rate increased in the intervention group (35.1% vs 21.9%)[71].

CONCLUSION

Liver cancer screening currently faces multiple challenges, including insufficient screening coverage, limited sensitivity in detecting early-stage lesions, and heterogeneous risk profiles across populations. Future efforts should actively promote screening strategies that integrate stratified, combined, and intelligent approaches. We recommend adopting risk-scoring model-based stratified management in clinical practice and promoting multimodal screening strategies combining imaging and serological markers to enhance the detection rate of early-stage liver cancer. Concurrently, governments and healthcare institutions must strengthen health education for the public and primary care providers to comprehensively improve screening adherence. Achieving early liver cancer screening and enhancing patient long-term survival ultimately requires collaborative and sustained efforts from medical institutions, research teams, public health systems, and societal forces.