Is Cell-Free DNA Testing in Hepatocellular Carcinoma Ready for Prime Time?

Revamping the current biomarker landscape of hepatocellular carcinoma (HCC) with cell-free DNA (cfDNA) could improve overall outcomes. The use of commercially available cfDNA testing (also known as liquid biopsy) is limited by the low prevalence of targetable mutations and does not have any prognostic or predictive value. Thus, current cfDNA testing cannot be relied upon for perioperative risk stratification (POR), including early detection of recurrence, long-term surveillance, predicting outcomes, and treatment response. Prior evidence on cfDNA mutation profiling (non-specific detection or gene panel testing) suggests that it can be a reliable tool for POR and prognostication, but it still requires significant improvements. cfDNA methylation changes or epigenetic markers have not been explored extensively, but early studies have shown potential for it to be a prognostic biomarker tool. The predictive value of cfDNA (mutations and EM) to assist treatment selection (systemic therapy, immune-checkpoint inhibitor vs. tyrosine kinase inhibitor) and to monitor response to systemic and locoregional therapies should be a future area of focus. We highlighted the unmet needs in the HCC management and the current role of cfDNA testing in HCC in addressing them.


Introduction
Hepatocellular carcinoma (HCC) is the third leading cause of cancer-related death worldwide, with a 5-year survival rate of 18% [1,2].HCC management requires a multidisciplinary approach, and multiple societies around the world have proposed guidelines for risk-stratification and appropriate treatment selection [3,4].In early/intermediatestage tumors, curative surgical resection (SR) and locoregional therapies (LRT) ranging from image-guided ablation, transarterial chemoembolization (TACE), stereotactic body radiotherapy (SBRT), and transarterial radioembolization (TARE) are preferred in the first line.Systemic therapy (ST) with immune-checkpoint inhibitors (ICI) and tyrosine kinase inhibitors (TKI), and vascular endothelial growth factor inhibitors (VEGFi) are often reserved for unresectable/advanced tumors.In this review, we focused on the gaps in HCC management and summarized the current status of cfDNA testing in addressing them.

Unmet Need in HCC Management
ogy [5,6].The 5-year survival rate is around 75% in early-stage tumors and 35% in intermediate/advanced-stage tumors (median overall survival (OS) of 37 m) [5,7].Median progression-free survival (PFS) after surgery is around 9 months, and 38% recur in less than 2 years [8].Post-recurrence 5-year survival rate is around 43% and current evidence does not justify the use of adjuvant (AT) or neoadjuvant (NAT) as standard-of-care (SOC).However, interim results of IMbrave 050 (atezolizumab and bevacizumab vs. surveillance in high-risk HCC) presented at the American Association for Cancer Research (AACR) this year are encouraging, with a 28% improvement in progression-free survival (PFS) (HR= 0.72, 95% CI, 0.56, 0.93; p = 0.0120) with adjuvant therapy.The risk stratification is based on tumor burden, MVI, and differentiation.
LEGACY study, TRACE, and RASER trials have shown that TARE is a favorable option in uncomplicated solitary HCC in patients with Child-Pugh score (CP) A and good performance status (PS), but the failure of other prominent trials (SORAMIC, SARAH, SIRvsNIB) indicates that it may not be an ideal option over ST in the treatment of advanced tumors (CP B, high tumor burden, MVI) [9][10][11][12][13][14]. Milan's prognostic scale (based on tumor location, MVI, total bilirubin, and tumor burden) has prognostic value based on retrospective review, but is not typically used in clinical practice [15,16].SBRT is another local therapy for patients not eligible for resection.Recent data from RTOG 1112 showed an improvement in overall survival (OS) and PFS when SBRT was combined with sorafenib over sorafenib alone.Multiple other studies have demonstrated the utility of SBRT for the treatment of HCC [17][18][19][20][21][22][23][24][25].
Treatment selection (SR vs. TACE vs. SBRT vs. TARE vs. ST) is often based on the patient and tumor-related characteristics such as size, number, and location of the lesions, baseline liver functional status (by CP or Albumin-Bilirubin (ALBI) score), PS, and feasibility of the procedure intended.One of the main reasons for poor outcomes is the lack of biomarkers with a reliable prognostic or predictive value that guides HCC management.Alpha-fetoprotein (AFP) has poor sensitivity and specificity to detect recurrence or assess treatment response [26].Similarly, imaging modalities (CT or computerized tomography and MRI or magnetic resonance imaging) lack the necessary accuracy to detect early recurrences, particularly after LRTs such as TARE or SBRT.Oftentimes, identification of residual disease or progression is delayed for months due to post-treatment changes seen early following treatment [27].
Genomic testing and tumor or cell-free DNA (cfDNA) do not have roles in HCC management in its current form except in identifying rare targetable mutations in advanced cancers.Circulating tumor DNA (ctDNA), genomic material from the tumor cells, is often used synonymously with cfDNA, but there is a considerable difference between them.cfDNA refers to DNA floating in the bloodstream, including ctDNA, circulating tumor cells (CTC), exosomal DNA, and DNA from normal cells [28].The access to tissues is limited in HCC as it is generally not acquired for diagnosis, relying on distinctive imaging features.Hence, cfDNA testing is the main source of genomic testing for HCC.The use of cfDNA has the potential to address three areas of unmet need biomarker testing in HCC: (i) perioperative risk stratification (POR) to guide surgeons in identifying highrisk populations for recurrence and poor survival; (ii) prognostic biomarkers to estimate poor outcomes in intermediate/unresectable HCC; (iii) predictive biomarkers to select appropriate ST (ICI vs. TKI) or LT (SBRT vs. TACE vs. TARE) and identify the population who benefit from post-procedural ST.
Etiology-specific cfDNA markers were reported in previous studies that can help in certain situations.Hepatitis B (HBV) carriers with HCC tend to have higher circulating ERBB2 and TERT mutations, higher methylation rates in RASSF1, TFPI2, TRG5 (along with AFP), and XPO4, and low methylation rates in CDKN2A than those without HCC [43,59,63,[70][71][72].Higher RASSF1 methylation rates are frequent in hepatitis C (HCV) patients with HCC (compared to HCC-negative) [73].Some of them can help in detecting the recurrence (e.g., virus-host chimera DNA (vh-DNA), generated from junctions of HBV integration in the HCC chromosome in HBV-HCC patients, or forecasting the outcomes (e.g., higher cfDNA levels in HCV-carriers) [74,75].The current evidence on cfDNA testing in HCC is summarized below based on the areas of unmet need discussed above.

Perioperative Risk Stratification (POR) with cfDNA Testing in HCC
The prior cfDNA studies for POR can be broadly divided into three main categories (Table 1): (a) the detection and quantification of cfDNA levels (high vs. low); (b) detection and quantification by mutation allelic fraction (MAF) of specific mutations using small panels (hot-spots) or larger panels with next-generation sequencing (NGS) or whole exosome sequencing (WES); (c) EM.Evidence suggests that mutation profiling with larger panels (by NGS or WES) are not valuable over carefully selected hot-spot panels, and epigenetic testing needs a considerable amount of work before we can use it in clinical practice.
cfDNA testing preoperatively (preop) helps select a population of patients who can benefit from NAT, and by serial testing, we can determine the ideal time for resection.Similarly, postoperative (post-op) serial testing can complement current SOC (AFP and imaging).The mutation profiling was shown to be more robust than imaging and other protein markers such as AFP, AFP-L3%, and des-gama-carboxy prothrombin (DCP) [76].The ctDNA testing detected recurrence on an average of 4.6 months earlier than imaging, which is the current gold standard [77].Patients with postoperative ctDNA-positivity were identified as significant adverse risk factors for PFS and OS.DCP-positivity combined with ctDNA was shown to increase sensitivity for detection of minimal residual disease.
A tissue agnostic serial ctDNA mutation test proved useful to predict recurrence and microvascular invasion by serial monitoring [40].The ctDNA detection was identified by PCR on preoperative samples, and whole exosome sequencing (WES) was performed on postoperative cfDNA samples.Serial ctDNA testing was more sensitive for poor recurrence rates than AFP or DCP.Interestingly, the ctDNA detection rate was lower (15%) than reported in other studies and there were non-synonymous mutations both in cfDNA and tumor tissue.DNMTs mRNA levels, malondialdehyde (MDA), xanthine oxidase (XOD), glutathione hormone (GSH), and glutathione-S-transferases (GST) levels were also tested MDA and XOD levels were significantly higher in the IGFBP7 methylated group than the unmethylated group, while GSH level was lower in the methylated group than in unmethylated group (DA p = 0.001) (XOD p < 0.001).

Prognostic Value of cfDNA Testing in HCC
Reliable blood-based (cfDNA, proteins, or CTC) markers that can foretell poor clinicopathological features beyond standard imaging modalities (CT or MRI) may be helpful in treatment planning for patients with HCC (summarized in Table 2).Most prior studies correlated detection of mutations in particular genes such as TERT or BCL9 and RPS6KB1 with outcomes (OS and PFS) or advanced pathological features such as tumor size, PVTT, or TNM stage [34,70,85,86].The quantitative analysis by mutation allelic fraction (MAF) in cfDNA testing also proved to be useful in some studies [31,44].
EM were less explored than mutations in HCC.EM were usually single-gene-based, such as LINE1, TFPI2, IGFB7, or small panels (APC, GSTP1, RASSF1A, and SFRP1) [59,87,88].Interestingly, one study by Xu at al. constructed two major models based on methylation markers (CpG sites): one for diagnosis (n = 10 genes) and another for prognosis (n = 8).The former was also valuable in predicting tumor burden and poor outcomes [55].

Predictive Biomarkers
Genomic biomarkers have not been studied to predict the response to common LRTs performed for HCC such as TACE, TARE, SBRT, or systemic therapy (ICI and TKI).Very few studies have explored this area of HCC management and are summarized in Table 3.These studies follow the same trend as prognostic or POR studies.Most of these genomic biomarkers are non-specific (cfDNA levels), while some studies have used NGS or hotspot panels on pre-treatment samples.
Sefriouri et.al studied the change in MAF of TERT mutations and cfDNA levels from baseline, post-TACE (day 2), and one month later [89].They reported that patients who do not show any drop/change in either (TERT or cfDNA) are at an increased risk of having residual disease or progression.In another study, patients with TERT mutations who received TACE (44/130, rest had TKIs) had poorer OS (and large size) [90].Higher cfDNA levels post-radiation (SBRT and external beam radiation) and TARE (from cfDNA testing on patients in SORAMIC trial before systemic therapy) were also indicators of poor response (Table 2) [91,92].The latter study also gives us insight into biomarkers for sorafenib response as well.There were no studies that used cfDNA EM for this purpose.
Nakatsuka et al. reported that higher baseline cfDNA levels were an independent negative risk factor among HCC patients receiving LRTs (TACE and RFA) and systemic therapy (TKIs and ramucirumab) [93].Interestingly, a greater rise in cfDNA level post-therapy (day 2 of LRT and average of day 3 after systemic therapy) was an indicator of a favorable response.Higher baseline cfDNA levels and AFP (>400 ng/mL), along with TERT positivity, were associated with poor outcomes in patients treated with atezolizumab/bevacizumab combination.Patients with mutations in PIK3CA/mTOR pathway genes (PIK3CA, PTEN, TSC1, TSC2, and RPS6KA3) have poor response rates to TKIs, while mutations in WNT pathways had no effect on ICI response [29].EM were not explored enough as predictive biomarkers for HCC.In particular, the detection rate increased significantly from 31 to 54% in the systemic-therapy-treated cases (p = 0.045).
A higher baseline cfDNA (>70.7 ng/mL) was associated with poor survival (5.5 m vs. 13.7 m, p < 0.001) and is an independent factor for OS.
In Lenvatinib responders, AMER1, MLL3, and NOTCH2 were mutated.In some patients' primary resistance, there is no change in mutations profiles, but there is increase in VA.
In some patients with partial response, mutations have disappeared-not clear if they help in monitoring response TACE-Transarterial chemo embolization; PFS-progression-free survival; OS-overall survival; XRT-radiation; SBRT-stereotactic radiation therapy; HR-hazards ratio; MVI-macrovascular invasion; MAF-mutation allelic fraction; TKI-tyrosine kinase inhibitor; ICI-immune-checkpoint inhibitor; * post treatment samples were collected on day 2 for TACE and RFA (radiofrequency ablation), and average of day 3 for systemic therapy.

Conclusions
HCC is rising in incidence and overall outcomes remain poor.Despite having robust screening procedures in place, patients are often diagnosed in advanced stages with limited effective treatment options.A lack of reliable biomarkers that could improve current POR and predict or monitor the response to LRT and ST presents a clinical challenge for patients and treating physicians alike.One promising area of interest involves the use of cfDNA testing; however, serial improvements are needed before widespread use.Early reports of EM are promising and should be further explored on how to best incorporate it in current practice (summarized in Figure 1 below).Non-specific cfDNA level testing can help in monitoring treatment response or assessing the tumor burden.Its value without specific mutation or EM testing is questionable in POR.Developing models with specific mutations and EM in blood with high detectable rates will provide a non-invasive biomarker tool for risk-stratification perioperatively, estimate the survival, be vigilant of advanced pathological features (tumor size, PVTT), and help with treatment selection resistance monitoring.cfDNA testing to assist treatment selection (ST vs. LRT) in nonmetastatic advanced and intermediate stage HCC and monitoring the treatment response could significantly impact outcomes in HCC patients.

19 Figure 1 .
Figure 1.Current status of cell-free DNA testing in hepatocellular carcinoma.

Author Contributions:
Conceptualization by A.M.; writing-original draft preparation, A.M. and S.J.; writing-review and editing, A.E.; writing-review, E.M.; writing-review, A.S. All authors have read and agreed to the published version of the manuscript.

Table 1 .
cfDNA testing for perioperative risk stratification in hepatocellular carcinoma.

Table 2 .
cfDNA mutations and epigenetic markers with prognostic value in hepatocellular carcinoma (HCC).

Table 3 .
cfDNA testing for predicting response to locoregional therapy and systemic therapy.