Predictive Biomarkers of Immune Checkpoint Inhibitor-Based Mono- and Combination Therapies for Hepatocellular Carcinoma

Hepatocellular carcinoma (HCC) is among the most frequent and deadly human cancers worldwide. It has been shown that interaction between immune checkpoint receptors and ligands plays a crucial role in inhibition of T cell-mediated anti-tumor immune responses, thereby assisting tumor cells to evade the host immune surveillance. Therefore, several immune checkpoint inhibitors (ICIs) that selectively block immune checkpoint receptors or ligands have been developed as clinically effective and safe immunotherapeutic agents for treating HCC, including the inhibitors targeting cytotoxic T lymphocyte-associated antigen 4, programmed death 1, and programmed death ligand 1. In addition, various combinations of ICIs and other ICIs or tyrosine kinase inhibitors or vascular endothelial growth factor inhibitors have also emerged as clinically beneficial treatments for HCC. However, the overall response rates of ICI mono-therapy and combination therapy in HCC patients remain unsatisfied, highlighting the urgent need for discovering valuable predictive biomarkers to achieve personalized therapy. This review comprehensively summarizes the literature-based evidence validating a variety of biomarkers with predictive significance for treatment responses and outcomes in HCC patients receiving various ICI-based mono- and combination therapies.


Introduction
As the predominant type of primary liver cancer, hepatocellular carcinoma (HCC) accounts for over 90% of primary liver malignancies and is the sixth most prevalent and the third most deadly human cancer worldwide, contributing to approximately 900,000 cases and 800,000 deaths per year [1][2][3].Many therapeutic options have been well established for treating HCC, including surgical therapies (such as liver transplantation and resection) [4,5], locoregional therapies (such as radiotherapy, ablation, and embolization) [6,7], and systemic therapies (such as chemotherapy and molecular targeted therapy) [8,9].
Moreover, immunotherapies such as immune checkpoint inhibitor (ICI) therapy have been developed as a promising treatment modality for HCC [10,11].However, the therapeutic efficacy of these treatment modalities varies among patients and remains to be improved.Therefore, the discovery of valuable biomarkers for predicting therapeutic responses and outcomes in HCC patients is an important goal to select the most suitable patient for the most suitable treatment (the so-called personalized therapy) to improve patient survival.
Immune checkpoint molecules include the co-inhibitory receptors expressed by effector T cells and the corresponding ligands expressed by tumor cells and stromal cells [12,13].Through the interaction between the receptors and ligands, immune checkpoint molecules play an important role in suppression of the activation and function of effector T cells, thereby facilitating tumor cell escape from T cell-mediated anti-tumor immune responses [14,15].As a result, many monoclonal antibodies that block the binding of immune checkpoint receptors to ligands (the so-called ICIs) have been generated as effective immunotherapeutic agents to restore T cell-mediated killing of tumor cells [16,17].The ICIs that have been licensed or are in clinical research for HCC treatment include the agents targeting the co-inhibitory receptors such as cytotoxic T lymphocyte-associated antigen 4 (CTLA-4) and programmed death 1 (PD-1) and the agents targeting the ligand of PD-1, programmed death ligand 1 (PD-L1) [18][19][20].In addition, the combination of ICIs targeting CTLA-4 and PD-1 or PD-L1, the combination of ICIs targeting PD-1 or PD-L1 and tyrosine kinase inhibitors (TKIs), and the combination of ICIs targeting PD-L1 and vascular endothelial growth factor (VEGF) inhibitors have been also licensed or are under clinical validation for HCC therapy [18][19][20].It has been shown that ICI-based mono-and combination therapies are active, tolerate, and clinically beneficial against HCC, although patients may concurrently receive various locoregional therapies.However, the overall response rates remain unsatisfied in HCC patients, with only about 15% and 30% for ICI-based mono-and combination therapies, respectively [18][19][20].Therefore, the development of clinically useful predictive biomarkers for identifying HCC patients who are more likely to respond to ICIs is urgently needed to advance personalized therapy for better patient outcomes.
This review provides a comprehensive summary of the hitherto published literature, which unravel various promising biomarkers at pre-treatment, on-treatment, and post-treatment time points in tissue, blood, and stool samples for predicting the therapeutic responses and clinical benefits of different categories of ICI-based therapies in HCC patients, including ICI mono-therapy and combination therapy with other ICIs or TKIs or VEGF inhibitors.
f NLR was calculated by dividing blood neutrophil count by blood lymphocyte count and was categorized as high (≥ 3) or low (< 3) ratio.g On-treatment NLR was measured at day 14 after treatment initiation.Pre-treatment, on-treatment, and post-treatment NLR were categorized as high (> 2.5, ≤ 4.1, and > 2.7) or low (≤ 2.5, < 4.1, and ≤ 2.7) ratio, respectively.h NLR was categorized as high (≥ 5) or low (< 5) ratio for pre-treatment and post-treatment.PLR was calculated by dividing blood platelet count by blood lymphocyte count and was divided into three level groups: low (≤ 118), medium (> 118 to < 224), and high (≥ 224) for pre-treatment; low (≤ 125), medium (> 125 to < 229), and high (≥ 229) for post-treatment.i NLR was categorized as high (≥ 5) or low (< 5) ratio.PLR was categorized as high (≥ 140) or low (< 140) ratio.SII was calculated by multiplying blood platelet count by blood neutrophil count and dividing by blood lymphocyte count and was categorized as high (≥ 970) or low (< 970) index.LMR was calculated by dividing blood lymphocyte count by blood monocyte count and was categorized as high (≥ 1.8) or low (< 1.8) ratio.j Monocyte index was calculated by dividing MonocyteD7/D0 by Monocyte-PDL1D7/D0, in which MonocyteD7/D0 was defined as the fold change in the frequency of classical monocytes at day 7 over day 0 after treatment initiation and Monocyte-PDL1D7/D0 was defined as the fold change in the frequency of PD-L1-expressing classical monocytes at day 7 over day 0 after treatment initiation in blood, and was categorized as high (≥ 1) or low (< 1) index.
2022 Li et al. [37] Pre-treatment PIVKA-II level and metastasis i Blood 191 HCC patients treated with PD-1 ICI combined with TKI Patients with a low-risk score had the longest OS, followed by patients with a medium-or high-risk score.
2023 Guo et al. [38] Pre-treatment ctDNA TMB or MSAF j Blood 107 HCC patients treated with PD-1 ICI combined with TKI Low ctDNA TMB predicted higher DCR and longer OS.Low ctDNA MSAF predicted higher DCR and longer OS.
2022 Xu et al. [39] Pre-treatment PD-L1 + CTC count k Blood 47 HCC patients treated with PD-1 ICI combined with locoregional therapy and TKI Low PD-L1 + CTC count predicted ORR and higher ORR and longer OS.
2022 Su et al. [40] Pre-treatment gut microbiota l Stool 74 HCC patients treated with PD-1 ICI combined with or without TKI Patients with a good signature of microbiota had the highest ORR and DCR and the longest OS and PFS, followed by patients with a fair or poor signature.
2022 Lee et al. [41] a OS was defined as the time from treatment initiation to death due to any cause.PFS was defined as the time from treatment initiation to radiological progression or death due to any cause.ORR was defined as the proportion of patients with CR or PR.DCR was defined as the proportion of patients with CR, PR, or SD.b Intratumoral CD38 + cell proportion was defined as the percentage of CD38-expressing cells in tumor tissues and was categorized as high (≥ 5%) or low (< 5%) proportion.c PNI was defined as blood albumin level plus 5 multiplies by blood lymphocyte count and was categorized as high (≥ 45) or low (< 45) index.NLR was calculated by dividing blood neutrophil count by blood lymphocyte count and was categorized as high (≥ 5) or low (< 5) ratio.PLR was calculated by dividing blood platelet count by blood lymphocyte count and was categorized as high (≥ 300) or low (< 300) ratio.d AFP decrease was defined as the percentage of decrease in serum AFP levels at 4 weeks after treatment initiation relative to pre-treatment levels and was categorized as high (> 20%) or low (≤ 20%) decrease.
e AFP decrease was defined as the percentage of decrease in serum AFP levels at 4 weeks after treatment initiation relative to pre-treatment levels and was categorized as high (> 10%) or low (≤ 10%) decrease.
f AFP decrease was defined as the percentage of decrease in serum AFP levels at 3 months after treatment initiation relative to pre-treatment levels and was categorized as high (> 15%) or low (≤ 15%) decrease.
g AFP or PIVKA-II decrease was defined as the percentage of decrease in serum AFP or PIVKA-II levels after completion of treatment relative to pre-treatment levels and was categorized as high (> 50%) or low (≤ 50%) decrease.h A total score was calculated based on the nomogram assigned ratio and was divided into three risk groups: low (≤ 182.7), medium (> 182.7 to ≤ 240.3), and high (> 240.3).
i The combination of PIVKA-II level and metastasis was categorized as high (PIVKA-II > 600 mAU/mL and with metastasis), medium (PIVKA-II > 600 mAU/mL or with metastasis), or low (PIVKA-II < 600 mAU/mL and without metastasis) risk.j TMB was defined as the number of somatic mutations per megabase of sequenced ctDNA and was categorized as high (> 4) or low (≤ 4) burden.MSAF is an indicator of the amount of ctDNA in blood and was categorized as high (> 0.027) or low (≤ 0.027) frequency.k PD-L1 + CTC count was categorized as high (≥ 2) or low (< 2) count.
d PIVKA-II level was categorized as high (≥ 186 mAU/mL) or low (< 186 mAU/mL) level.NLR was calculated by dividing blood neutrophil count by blood lymphocyte count and was categorized as high (≥ 2.5) or low (< 2.5) ratio.AFP, PIVKA-II, or NLR decrease was defined as the percentage of decrease in serum AFP, PIVKA-II, or NLR levels at the first response evaluation after treatment initiation relative to pre-treatment levels and was categorized as high (≥ 30%, ≥ 50%, or ≥ 10) or low (< 30%, < 50%, or < 10) decrease, respectively.e ALBI grade was defined as log10 blood bilirubin level multiplied by 0.66 plus blood albumin level multiplied by -0.085 and was stratified as grade 1 (≤ -2.60), 2 (> -2.60 to ≤ -1.39), or 3 (> -1.39).ALBI grade 2 and 1 were defined as high and low ALBI grade, respectively.AFP decrease was defined as the percentage of decrease in serum AFP levels at 3 weeks after treatment initiation relative to pre-treatment levels and was categorized as high (≥ 20%) or low (< 20%) decrease.
h PD-1 + granulocyte percentage was defined as the percentage of PD-1-expressing granulocytes on total granulocytes in blood and was categorized as high (≥ 13%) or low (< 13%) percentage.
i IgG increase was defined as the percentage of increase in serum IgG levels at 6 weeks after treatment initiation relative to pre-treatment levels and was categorized as high (≥  a Upward and downward arrows indicated that high and low levels of biomarkers predicted better treatment responses and outcomes, respectively.b Although classified as a pre-treatment biomarker, this biomarker was based on pre-treatment ALBI grade and on-treatment AFP decrease in prediction.Abbreviations: ICI, immune checkpoint inhibitor; HCC, hepatocellular carcinoma; TKI, tyrosine kinase inhibitor; VEGF, vascular endothelial growth factor; PD-1, programmed death 1; PD-L1, programmed death ligand 1 NLR, neutrophil to lymphocyte ratio; PLR, platelet to lymphocyte ratio; LMR, lymphocyte to monocyte ratio; SII, systemic immune-inflammation index; TGF-β, transforming growth factor-beta; CRP, C-reactive protein; AFP, alpha-fetoprotein; ALBI, albumin-bilirubin; PNI, prognostic nutritional index; ctDNA, circulating tumor DNA; TMB, tumor mutation burden; MSAF, maximum somatic allele frequency; CTC, circulating tumor cell; PIVKA-II, protein induced by vitamin K absence or antagonist-II; ECOG PS, Eastern Cooperative Oncology Group performance status; TACE, transarterial chemoembolization; EHM, extrahepatic metastasis; ALT, alanine aminotransferase; IL-6, interleukin-6; mALBI, modified albumin-bilirubin; IgG, immunoglobulin G; ORR, objective response rate; DCR, disease control rate; OS, overall survival; PFS, progression-free survival; TTP, time to progression.

Conclusions
This review comprehensively summarizes the evidence from the literature published so far which validate the predictive significance of a variety of biomarkers at different treatment time points (including pre-treatment, on-treatment, and posttreatment time points) in different sample sources (including tissue, blood, and stool samples) for the treatment responses and outcomes of HCC patients receiving different categories of ICI-based therapies (including ICI mono-therapy and combination therapy with other ICIs or TKIs or VEGF inhibitors) (Table 4).Among the current predictive biomarkers, most are derived from the blood and stool samples of HCC patients, supporting the convenience advantages of the use of noninvasive sampling methods in clinical application.Moreover, the clinical applicability varies among the predictive biomarkers.Certain biomarkers are selective for one category of ICI therapy at one specific treatment time point, such as tumoral PD-L1 expression level, intratumoral CD38 + cell proportion, ctDNA TMB or MSAF, PD-L1 + CTC count, PD-1 + granulocyte percentage, SII, PNI, and IgG change; in contrast, some biomarkers show predictive value for 2 or 3 different categories of therapies at 2 or 3 different treatment time points, such as NLR, PLR, AFP change, and PIVKA-II change.It should be carefully noted that even the same biomarker may have different cut-off values when applied for predicting HCC patients receiving different categories of therapies at different treatment time points.Besides, as a potential way to overcome the inter-patient heterogeneity among HCC patients, several biomarkers combine 2 different factors or even more factors in a nomogram to stratify the patients into more subgroups for prediction, such as CRP level and AFP level, PIVKA-II level and metastasis, ALBI grade and age, ALBI grade and AFP change, and mALBI grade and AFP level.In addition, many predictive biomarkers are based on immune cells or inflammatory cytokines, such as NLR, PLR, LMR, PD-1 + granulocyte percentage, SII, monocyte index, TGF-β, IL-6, and CRP, reflecting the clinical implication of tumor immune microenvironment in the efficacy of ICI therapy for HCC.Additionally, the predictive significance of the composition of gut microbiota, such as Erysipelotrichaceae, Veillonellaceae, Prevotella 9, and Lachnoclostridium, has been validated in HCC patients receiving ICI mono-therapy and combination therapy with TKIs.Considering the impact of gut microbiota-derived metabolites on ICI therapy for cancer [50][51][52], evaluation of the predictive significance of microbial metabolites in the blood and/or stool samples of HCC patients receiving ICI-based mono-and combination therapies may hold great promise to discover novel predictive biomarkers.Furthermore, non-coding RNAs such as microRNAs and long non-coding RNAs have been closely implicated in cancer and ICI therapy [53,54].Whether non-coding RNAs can also serve as predictive biomarkers for ICI therapy in HCC patients is worth further investigation.Last but not the least, since the HCC patient cohorts evaluated in different studies may have different clinicopathological features and receive ICI-based therapies with different drugs (even though sharing the same molecular targets) at different treatment dosages, doses, and dosing intervals, it is quite important to take this issue into consideration when applying the predictive biomarkers to select the most suitable patient for the most suitable treatment for better therapeutic responses and outcomes.

Table 2 .
Predictive biomarkers of ICI combination therapy with other ICIs or TKIs for HCC

Table 4 .
Comparative summary of predictive biomarkers of ICI-based mono-and combination therapy for HCC