Table 3.

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

Publication	Methods	Tested Markers	Findings	Additional Information-Frequency
Sefrioui et al., 2022 (28) [89] TACE	ctDNA for TERT ddPCR cfDNA was quantified by the fluorometric method.	TERT hotspot cfDNA levels Samples collected prospectively: baseline (D − 1), day 2 (D + 2), and one month later (DM)	High-risk patients progressed in one month (80.0% vs. 4.3%, p = 0.001) and worse PFS (1.3 vs. 10.3 months, p = 0.002)	Score based on changes in cfDNA and ctDNA. High-risk or non-responders = no change in both; low-risk = change in at least one marker. cfDNA change from D-1 to D M+ 1 of > 31.4% (44.4% vs. 3.8%) and ctDNA change >0% as high risk (50.0% vs. 5.0%)
Park et al., 2018 (55, 34 fractionated XRT and 21 SBRT) [91]	Ultraviolet-visible spectrophotometer	cfDNA levels before and after XRT	High pre-XRT cfDNA levels are associated with advanced disease (p = 0.049) and larger tumors (p = 0.017). High post-XRT cfDNA level is associated with poor response rates (p = 0.017), local control (p = 0.006), or intrahepatic recurrence rates (p = 0.035)	Cut-off values for cfDNA levels: 33.65 ng/mL before XRT is and 37.25 ng/mL after CRT)
Hirai et al., 2021 (130, 44 TACE and 86 with lenvatinib or sorafenib) [90]	Digital droplet- polymerase chain reaction (dd-PCR)	TERT	Detection of TERT mutation and high fractional abundances (≥1%) associated with worse OS (p< 0.01).	TERT promoter mutations correlated with large intrahepatic tumor size (p = 0.05) and high des-gamma carboxyprothrombin (p = 0.005)
Alunni-Fabbroni et al., 2019 (13, 10, TARE + sorafenib, 2 RFA + sorafenib, 1placebo) [92]	Next-generation sequencing (NGS)	597 gene panel. Samples were collected between local therapy and beginning of sorafenib-based systemic therapy (T1), then at 3 time points—8 weeks apart from the date of starting sorafenib (T2, T3, T4)	High cfDNA concentrations at T1 associated with metastasis (p = 0.012) CYP2B6 variant detection at T1–T2 associated with worse OS (p = 0.013)	High T2 has a trend towards significance for metastasis (p = 0.07) T1 and T2 concentration do not affect OS; however, high levels at T3 (p = 0.057) and T4 (p = 0.095) have trends. Early detection of BAX (for MVI, p = 0.014) and HFN1A (for liver cirrhosis, p = 0.032) confer worse clinicopathological features.
Oh et al., 2019 (151) [94] Sorafenib	Whole genome sequencing Baseline	cfDNA level, VEGFA-to-EIF2C1 ratios I score for genomic instability.	Patients who failed sorafenib had significantly higher cfDNA levels (0.82 vs. 0.63 ng/μL; p = 0.006) and I-scores (3405 vs. 1024; p = 0.0017) than those achieving disease control. VEGFA: EIF2C1 ratio was not significantly associated with treatment outcomes.	94% had previous therapy cfDNA-high group had a worse time to progression compared to the low group (2.2 vs. 4.1 months; HR = 1.71; p = 0.002) and OS (4.1 vs. 14.8 months; HR = 3.50; p < 0.0001) Genome instability-high group progressed earlier (2.2 vs. 4.1 months; HR = 2.09; p < 0.0001) and had worse OS (4.6 vs. 14.8 months; HR = 3.35; p < 0.0001).
Fuji et al., 2021 (24) [30] Lenvatinib	NGS cfDNA at the baseline and 4 weeks later	Guardant 360 panel	Reduction MAFmean 4 weeks after treatment was useful in predicting PR and CR, but not for SD or PD	Baseline mutations and MAF change were not indicative of OS or PFS MAFmean change ≥ 0 vs. <0 HR 8.4, 95% CI = 2.3–31.2, p = 0.002) as an independent factor was associated with poor PFS. MAFmean reduction was more sensitive than AFP
Fu et al., 2022 [79] (n = 258, lenvatinib + ICI)	PCR	High-risk genes APC, ARID1A, CDKN2A, FAT1, LRP1B, MAP3K1, PREX2, TERT, and TP53	Preserved FAT1 or LRP1B variants without TP53 associated with poor PFS (HR = 17.1, p < 0.001).	Samples were collected post-surgery
Nakatsuka et al., 2021 (100 pre and 87 post treatment) [93] Treatments received include TACE, RFA (day 2), TKIs (lenvatinib, 27; sorafenib, 6; regorafenib, 1), and 1 ramucirumab *	Targeted ultra-deep sequencing	22,0009 coverage in a panel of 275 cancer-related genes	Greater increase in the post-treatment cfDNA levels (>66.5 ng/mL) associated with increased response (AUC of 0.807, sensitivity 60.0%, specificity 92.9%)	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.
Matsumae et al., 2022 (85) [95] (Atezolizumab/bevacizumab)	.	Custom-made panel for detecting mutations in 25 HCC-related cancer genes	Higher cfDNA levels have poor objective response rates (PR and CR, p = 0.03) and PFS (p = 0.021) TERT-positive patients have poor OS (p = 0.001), but no difference in PFS.	Higher plasma cfDNA (≥2.23 ng/µL), TERT positivity, high AFP (>400 ng/mL) were independent prognostic factors for OS OS–TERT+ and AFP− high < TERT+ or AFP− high < TERT− and AFP low (p < 0.001)
Von Feleden et al., 2021 (51, TKI = 23; ICI = 38) [29]	Ultra-deep sequencing for 25 genes ddPCR for TERT	PI3K/mTOR pathway genes for patients on TKI WNT pathway genes for patients on ICI	Mutation-positive patients have lower PFS (2.1 vs. 3.7 months, p < 0.001) No effect of mutations in WNT pathway on ICI response	Serial testing–not clear if it helps in predicting response 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.