Hepatocellular carcinoma (HCC) is one of the most prevalent malignancies and one of the leading causes of cancer-related deaths worldwide (1). In accordance with the World Health Organization (WHO), primary liver cancer (75–85% of all HCC cases) has the sixth-highest incidence globally and accounts for the third largest number of cancer deaths (1, 2). Due to its insidious onset, high invasiveness, rapid progression, and difficulties in early diagnosis, most HCC cases in China are diagnosed at locally advanced stages or with distant metastases, ruling out opportunities for surgical resection (3).

For patients with unresectable HCC (uHCC), transarterial chemoembolisation (TACE) is recommended as an effective palliative therapy that has shown encouraging survival outcomes (4). In particular, the LAUNCH study investigated TACE combined with systemic lenvatinib (LEN) treatment in uHCC and found improved efficacy and patient benefit compared with LEN monotherapy (5). Additionally, immunotherapy with immune checkpoint inhibitors (ICIs) has also been demonstrated to be effective in the treatment of HCC. In accordance with the phase III IMbrave150 study, the US Food and Drug Administration (FDA) approved atezolizumab (an anti-PD-L1 antibody) combined with bevacizumab for patients with systemic treatment-naïve, unresectable, or metastatic HCC (6).

There are a few other combination immunotherapies that have recently been developed as first-line options for HCC, including LEN with pembrolizumab (KEYNOTE-524) (7), rivoceranib with camrelizumab (CARES-310) (8), and sintilimab plus a bevacizumab biosimilar (ORIENT-32) (9). A notable example is the CHANCE001 study which found that TACE combined with PD-(L)1 inhibitors and tyrosine kinase inhibitors(TKI) showed considerable benefit in progression free survival (PFS), overall survival (OS), and objective response rate (ORR) compared with TACE monotherapy, as well as having an acceptable safety profile (10, 11). The combination of TACE with anti-PD-(L)1 antibodies and molecular targeted therapies (MTT) has been widely adopted especially in China, and its effectiveness has been demonstrated by a number of subsequent studies (10, 12–14). However, while it is crucial to identify patient populations that will benefit before undergoing this combination therapy, how to accomplish this remains uncertain. Several biomarkers have been explored to predict the immunotherapy efficacy. But to date, there is no validated biomarker that can be used in clinical decision-making. C-reactive protein (CRP) is an acute-phase reactant that indicates systemic inflammation (15). Recently, Fukuda et al. applied a novel CRP kinetics approach to nivolumab immunotherapy in metastatic renal cell carcinoma (mRCC), finding great tumour shrinkage and outcomes in ‘flare-responders’ whose CRP levels initially increased but then decreased within three months to levels below baseline (16). Several other studies have demonstrated a correlation between early CRP kinetics and improved response in metastatic urothelial carcinoma (mUC) and non-small cell lung cancer (NSCLC). In addition, preoperative CRP has been identified as a potential biomarker for HCC patients who have received surgery, TKIs, and ICIs (17). However, the impact of early CRP kinetics on TACE combined with LEN and ICIs remains to be fully demonstrated in HCC patients.

In this study, we explored the association between early CRP kinetics and prognosis in HCC patients treated with TACE combined with LEN and ICIs (TACE–LEN–ICIs), in order to offer valuable insights for oncologists in their clinical practice.

In this retrospective study, we validated that early CRP kinetics is a promising predictive biomarker for HCC treated with the TACE–LEN–ICIs combination regimen. To our knowledge, this is the first report assessing the prognostic utility of a CRP kinetics model in HCC patients. This is significant because the majority of HCC cases in China are diagnosed at advanced stages, where high tumour burden and portal vein tumour thrombus (PVTT) often manifest as prevalent features due to the insidious onset of the disease. Unfortunately, these cases have a remarkably unfavourable prognosis, meaning that Chinese doctors are more inclined to use more aggressive treatment approaches with higher intensity such as the TACE–LEN–ICIs regimen investigated here. In fact, the combination of TACE with ICIs plus TKIs is a commonly used treatment strategy for patients with advanced HCC in China. As indicated by the CHANCE001 study, although TACE combined with PD-(L)1 inhibitors and MTT showed better PFS and OS than TACE monotherapy, around 40% of patients still fail to achieve an objective response (10). So, there is still an urgent need to find a concise and effective biomarker to predict the effectiveness of triple therapy.

The acute-phase protein CRP is well-recognised as an indicator of cancer-induced systemic inflammation, which is often evident in clinical symptoms (20). A multi-institutional study revealed a significant correlation between baseline CRP levels and prognosis in patients with HCC undergoing LEN treatment. Specifically, a baseline CRP level above 0.5 mg/dL was determined to be an unfavourable prognostic factor in HCC patients treated with LEN (21). Another study found that elevated CRP levels may indicate aggressive cancer progression and potential resistance to therapy with the ICI atezolizumab in combination with bevacizumab (17). Hence, it has been demonstrated that an elevated baseline concentration of CRP before oncological treatment is associated with worse clinical outcomes in HCC patients. Unfortunately, however, the dynamics of early changes in systemic inflammation after therapeutic intervention have mostly been neglected.

Recently, Fukuda et al. (16) described the CRP ‘flare-response’ phenomenon, characterised by an early increase in CRP after initial treatment of ICIs, followed by a subsequent reduction below baseline. It is noteworthy that this novel concept can correctly predict therapeutic success in 42 mRCC patients who received a PD-1 inhibitor. To date, here is substantial evidence to support the ability of CRP kinetics to rationalise the therapeutic effect of ICIs has been widely validated in mRCC (22), NSCLC (23), and mUC (24). However, there has been demonstrated a correlation between CRP kinetics and therapeutic response in HCC patients, so this retrospective study aimed to address the knowledge gap on CRP kinetics in the HCC field.

In total, 143 patients treated with TACE–LEN–ICIs were enrolled into the study. Based on early CRP kinetics, 19 patients were assigned as CRP flare-responders, 60 patients as CRP responders, and 64 patients as CRP non-responders. After initiating triple therapy, the median PFS of CRP flare-responders, CRP responders, and CRP non-responders was 40.3, 11.4, and 5.5 months, respectively (p < 0.001). The median OS of the same three groups was NE (not estimable), 18.9 months, and 9.9 months, respectively (p < 0.001). Thus, in this study, OS and PFS varied significantly across the three groups according to CRP kinetics. Because of its easy accessibility in clinical practice, CRP kinetics could be easily adopted in routine practice and serve as a valuable tool for oncologists in clinical decision-making.

Furthermore, recently published results from LEAP-012 demonstrated that for intermediate-stage HCC, TACE combined with lenvatinib and pembrolizumab significantly prolonged PFS (14·6 months) compared with TACE alone (25). Notably, 86.7% of patients in our study were classified as BCLC stage C. The median PFS and OS in the CRP flare-responders group were significantly superior to the results of LEAP-012. However, both the median PFS and OS in CRP responders and CRP non-responders were shorter than those reported in LEAP-012. These findings indicate that through effective biomarker screening, even patients with more advanced stages can achieve favourable treatment outcomes. This further underscores the predictive value of CRP kinetics, which can assist in our clinical decision-making.

Consistently, HCC patients characterised as CRP flare-responders or CRP responders experienced higher ORR (78.9% or 78.4% versus 48.4% for non-responders) and DCR (94.8% or 96.7% versus 78.1% for non-responders) than those with CRP non-responders in the whole cohort. The fact that only 48.4% of CRP non-responders responded to TACE–LEN–ICIs, which means CRP kinetics could identify such patients early. Alternative locoregional treatments such as salvage radiotherapy, hepatic arterial infusion chemotherapy (HAIC), drug-eluting beads transarterial chemoembolisation (D-TACE), or selective intra-arterial radiotherapy (SIRT) are also effective.

There has been a growing body of research on inflammation-based prognostic scores have been developed in recent years, including the NLR (26, 27), PLR (28), SII (29), CAR (30), prognostic nutritional index (PNI) (20), and CRAFITY score (31, 32). These scores were proven to perform acceptably in predicting cancer prognosis. To further assess the early CRP kinetics model, the present study performed direct comparisons of the prognostic scores above with CRP kinetics. The AUROC values of the CRP kinetics model for PFS prediction at 12, 18, and 24 months were 0.654, 0.721, and 0.872, while for OS prediction were 0.669, 0.702, and 0.745, respectively. Our results suggested that the CRP kinetics model may predict OS and PFS more accurately than other inflammation markers for HCC patients treated with the TACE–LEN–ICIs regimen. Although CRP kinetics has exhibited superiority in predicting therapeutic efficacy in the present study, the explanation for its correlation with the immunological underpinnings of HCC remains insufficient. To further facilitate the utility of CRP kinetics in patient stratification and the prediction of treatment responses in future clinical practice, additional studies are required to investigate the potential associations among baseline CRP levels, distinct post-immunotherapy CRP kinetic patterns, and the composition of the tumour microenvironment.

Since CRP is regularly used to monitor systemic infections and inflammatory status, it is essential to expand oncologists’ awareness of CRP kinetics in triple therapy. While acute local inflammation may occur in the liver after a TACE procedure, leading to an increase in white blood cells and CRP, a previous investigation found no remarkable destructive change along with minimal to absent inflammatory cell infiltration in the surrounding non-tumoural liver parenchyma at 14 days post-surgery (33). Therefore, we reasonably speculate that TACE will not affect the serum CRP level 1 month after the procedure. Our clinical experience also indicates that the inflammatory response in the body generally disappears by this time, so we consider it reasonable to believe that TACE has little effect on CRP level after 1 month. In addition, a rapid CRP increase during initiation of ICIs should not be misinterpreted as indicative of systemic infection. In our centre, we regularly monitor procalcitonin and clinical symptoms to determine whether infection is present. Nevertheless, further studies are necessary to certificate whether complementing CRP with other markers, including procalcitonin, might facilitate differentiation between immunotherapy responses, systemic infections, or TACE-induced systemic infections.

This study has four main limitations. First, selection bias cannot be avoided in a single-centre retrospective cohort study. Second, the total sample size was limited, especially in the CRP flare-responder group, which may restrict the statistical power of subgroup analyses. Third, the follow-up period was relatively short, which may compromise the reliability of survival analyses. Fourth, as the majority of our patients were positive for HBV, our model may be more applicable to Chinese populations. Thus, the model requires testing with large, multicentre cohorts based on different populations to further validate its performance.

Early CRP kinetics may have predictive value for prognosis in HCC patients undergoing TACE–LEN–ICIs regimen. A large scale prospective randomised controlled clinical trial is still needed to confirm this conclusion in the future.