For patients for whom locoregional therapy of hepatocellular carcinoma (HCC) is inappropriate, effective systemic therapy became available after sorafenib was approved. In the decade following its approval, many other targeted agents emerged as options for first-line or salvage therapy for advanced HCC [1]. Subsequently, immune checkpoint inhibitors (ICIs) were introduced for HCC treatment alone or in combination with targeted therapy. At present, more than a dozen effective treatment regimens exist for advanced HCC. The problem for clinicians is obvious: how to choose between all these regimens?

Predictive markers can guide clinical decision-making. They are routinely used for targeted therapy in other malignancies, such as EGFR mutations for non-small-cell lung cancer and KRAS mutations for colorectal cancer [2, 3]. Unfortunately, no gene alterations of HCC cells have been proven to guide targeted therapy for advanced HCC, probably because HCC cells lack oncogene addiction and the drug targets are not exclusively tumor cells. Tumors with mismatch repair deficiency respond well to ICIs [4], but such deficiency is rare in HCC [5]. The expression of programmed death-ligand 1 (PD-L1) in tumor tissue can also be used to determine ICI use in certain cancers, but the treatment efficacy in patients with HCC with high and low PD-L1 expression is not different enough to delineate a suitable subgroup for ICI use [6, 7]. In brief, even a decade and half after the successful development of systemic therapy for advanced HCC, no useful predictive markers are available.

Yet, many prognostic markers have been proposed for HCC, including advanced disease. Clinical parameters and scoring systems based on them, such as the CLIP score, have been used to estimate patient survival [8]. However, prognostic and predictive markers should not be confused. Even if a patient is expected to exhibit poor survival under a treatment according to a prognostic marker, this does not mean that the patient should not receive the treatment. That determination is based exclusively on predictive markers.

Without a useful predictive marker, however, the management of patients with HCC relies on prognostic markers. Although these cannot be used to determine whether a treatment should be provided, they can still guide the care of patients under treatment. For example, patients expected to have poor prognosis might require more frequent hospital visits and tumor assessments.

The reported prognostic markers for HCC include clinical parameters, tissue biomarkers, imaging parameters, and blood biomarkers (primarily, cytokines, angiogenic factors, and cell counts) [9]. In this issue, Saeki et al. reported that the levels of methylated septin 9 (m-SEPT9), a frequent target of liquid biopsy, may serve as a prognostic marker for patients who receive sorafenib or lenvatinib as first-line therapy for advanced HCC [10].

Although this study is adequately designed and employed both training and validation cohorts, the results may not be immediately clinically useful. Targeted therapy alone is no longer the standard first-line therapy for most patients with advanced HCC. Whether the m-SEPT9 level is also prognostic for patients who receive immunotherapy remains unclear. If it is, this implies that the blood m-SEPT9 level represents tumor aggressiveness and extent. If not, m-SEPT9 may be associated with the mechanisms of targeted therapy.

Unfortunately, the study by Saeki et al. lacks the power to identify whether the m-SEPT9 level can differentiate between the efficacy of sorafenib and lenvatinib. A predictive marker remains the dream of all physicians taking care of patients with advanced HCC. The current study results encourage the authors to continue their work on m-SEPT9, especially with patients who receive various systemic therapies. Hopefully, this line of research might reveal the first useful predictive marker for advanced HCC.