The prognostic role of early tumor shrinkage in patients with hepatocellular carcinoma undergoing immunotherapy

Hepatocellular carcinoma (HCC) is the most common primary liver cancer and one of the worldwide leading causes of cancer-related deaths [1]. Systemic therapy is the mainstay treatment for patients in advanced tumor stages and for patients that experienced a failure with previous surgical or locoregional treatments [2]. In addition to the previously recommended tyrosine-kinase inhibitors, sorafenib and lenvatinib, immunotherapeutic agents have gained importance over recent years [3, 4]. Initially applied mainly for patients after tumor progression in first or second line of systemic treatment, the results of the IMbrave150 trial led to a paradigm change and the combination of the checkpoint inhibitor atezolizumab and the VEGF antibody bevacizumab is now the first line treatment for patients with advanced HCC [5‐7]. Furthermore, ongoing trials are investigating the potential of several other immunotherapeutic agents, both for the treatment of advanced HCC and for the potential treatment of earlier-stage tumors [4, 8, 9].

Despite the promising prognostic results of immunotherapy for most patients with advanced HCC, one third of patients with advanced HCC did not benefit from immunotherapy, and up to one fourth developed high-grade immune-related adverse events [10]. Thus, not all patients benefit equally. In clinical reality, one of the greatest challenges is the identification of patients most likely to benefit from immunotherapy. To date, we lack biomarkers that can predict the response to immunotherapy, and thus, provide guidance in clinical decision-making. To facilitate decision-making, novel biomarkers are needed, particularly as the treatment with those new agents distinctly differs from the treatment with tyrosine kinase inhibitors [9, 10]. Apart from tumor cell-derived and tumor microenvironment-derived biomarkers, in advanced HCC stages, specific laboratory parameters, like alpha-fetoprotein (AFP), and its changes during treatment, and inflammatory parameters, like C-reactive protein, have shown strong correlations with patient prognosis [10‐12]. However, little is known about the potential of imaging biomarkers for predicting long-term outcomes and for identifying patients that might benefit from immunotherapy.

Early tumor shrinkage (ETS) could function as one of those novel imaging biomarkers: ETS is defined as the relative reduction in tumor size between the baseline treatment and the first follow-up investigation. Originally identified as an imaging biomarker for patients with colorectal cancer undergoing chemotherapy, ETS was also shown to be predictive of outcomes in patients with other cancer entities [13‐16]. Several recent studies identified ETS as a highly promising imaging biomarker, specifically for outcomes in patients treated with immunotherapy [17‐19]. Moreover, ETS was identified as a superior parameter for assessing the treatment response, compared to the conventional response evaluation criteria in solid tumors (RECIST) and the modified RECIST (mRECIST) criteria. However, studies are scarce on ETS in patients with HCC. To date, ETS has only been investigated for HCC treatments with tyrosine-kinase inhibitors [20, 21]. Thus, the role of ETS remains unclear for patients with HCC that are treated with immunotherapeutic agents.

The ultimate goal when applying response criteria is to identify patients likely to benefit from continuation of the current treatment or patients more likely to benefit from alternative treatments, all in the light of an improved overall survival outcome. Both conventional and modified RECIST criteria establish a reduction in the maximum tumor diameter by 30% as partial response, however, new targeted therapies may be effective without showing such a decrease in imaging [22]. Although mRECIST has outperformed RECIST in prognosis prediction, correlation with the overall outcome varied tremendously in previous studies on patients with HCC and systemic treatment [21]. Thus, the cut-offs and criteria might be only partially suitable for these patients. Furthermore, the assessment of the aforementioned criteria is complex. Therefore, an additional and easy-applicable scoring system like ETS might help to complement conventional response criteria and therefore further improve patient selection and treatment decision-making.

Based on the promising preliminary work, our study hypothesis was that ETS is a highly potential and prognostic imaging biomarker for patients with HCC and immunotherapy.

The Ethics Committee of the Medical Association of Rhineland Palatinate, Mainz, Germany approved this study (permit number 837.199.10). The requirement for informed consent was waived, due to the retrospective nature of the study. This report followed the guidelines for transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD )[23].

In this study, we demonstrated that ETS could predict survival in patients with advanced-stage HCC undergoing immunotherapy. In addition, ETS could be used to further stratify patients with disease control at first follow-up, and it was associated with the AFP response during treatment. Thus, ETS showed high potential as an imaging biomarker for patients with HCC undergoing immunotherapy.

The IMbrave150 phase III-trial established that atezolizumab plus bevacizumab could serve as a new first-line regimen for systemic therapy-naïve patients with advanced-stage HCC. The objective response rate was 30% vs. 11% (atezolizumab + bevacizumab vs. sorafenib, based on RECIST1.1) [32]. Moreover, the HIMALAYA phase III-trial showed that the STRIDE regimen (tremelimumab/durvalumab) was superior to sorafenib, which further expanded immunotherapy options for unresectable HCC [33]. In addition, an interim analysis of the COSMIC-312 phase III trial found that the combination of atezolizumab/cabozantinib improved PFS over sorafenib, in the first-line setting [34]. As the treatment landscape broadens, biomarkers that predict the response to immunotherapy and prognosis are of pivotal importance, because they can guide clinical decision making in the context of personalized medicine.

Importantly, recent studies have indicated that not all patients benefit equally from treatment with immunotherapeutic agents [35]. Unfortunately, several candidate molecular and immunohistochemical biomarkers to identify patients likely to benefit, which are well-known from other cancer entities, showed no predictive capacity or have too low a prevalence in patients with HCC [9, 36‐38]. Novel biomarkers like circulating tumor cells or composition of the gut microbiota as well as newly identified gene signatures are currently investigated and showed promising initial results in first proof-of-concept studies [10, 39, 40].

Apart from biomarkers derived from tumor tissue, peripheral blood, and feces, imaging biomarkers are under consideration for identifying patients with HCC that are most likely to benefit from immunotherapy. However, it remains unknown whether imaging biomarkers that were identified for other tumor entities might be effective for patients with HCC treated with immunotherapy. One promising imaging biomarker is ETS, which was first described as an imaging biomarker in patients with colorectal liver metastasis [13]. Results from a meta-analysis that included patients from 10 different trials indicated that ETS showed enormous potential for supporting the identification of patients that would be sensitive to treatment and patients that would benefit from treatment continuation [13]. Based on those promising results, ETS was recently identified as an imaging biomarker for various other cancer entities, and it specifically showed potential as a predictive factor in patients treated with immunotherapy [14‐19].

Our results were consistent with a study that investigated ETS in patients treated with lenvatinib, by Takahashi et al., that identified a significant correlation between the AFP response and ETS [20]. Furthermore, those authors reported that ETS was highly associated with OS and PFS, calculated from treatment initiation [20]. The same results were previously observed for patients with HCC treated with sorafenib [21].

Notably, none of the laboratory parameters constituting the ALBI and MELD score did differ significantly between both ETS risk groups. However, not only the initial laboratory parameters but also their change during treatment might be relevant. For example, Granito et al. showed recently that the change of transaminases after TACE was a highly relevant predictor of treatment response [41]. One topic for future studies therefore might be the role of changes in laboratory parameters and their correlation with response and survival in patients with HCC undergoing immunotherapy.

In our study, the median ETS did not significantly differ between patients with previous treatment prior to immunotherapy and for whom the immunotherapy was the initial treatment. Furthermore, no significant difference was observed between patients who had undergone previous systemic lines of treatment and those who did not. However, our subgroups were small.

Takahashi et al. showed that ETS was highly correlated with the mRECIST criteria [20]. In their study, nearly 70% of patients with an ETS ≥10% showed an objective response, according to mRECIST. Moreover, they found that ETS showed superior predictive ability compared to the mRECIST criteria [20]. Unfortunately, the authors did not provide a subgroup analysis for the patients without progression at the initial follow-up. In the other recent study on ETS in patients with HCC treated with sorafenib, Öcal et al. suggested a cut-off of 20.0% for patient stratification [21]. In our study, patients with an ETS ≥20.0% had a prolonged survival compared to patients with an ETS < 20.0% (24.3 months vs 4.2 months). However, the optimal cut-off in our cohort was 10%, which was the same cut-off as previously published by Takahashi et al. when investigating survival of HCC under lenvatinib [20]. This suggests that under systemic treatment, a tumor reduction less than 20% or 30% already carries a significant survival advantage.

Another imaging biomarker proposed for assessing the response to treatment is the deepness of response (DpR) [13]. Similar to the ETS, the DpR is calculated based on the sum of the largest diameter of the target lesions. Originally, this parameter was identified as a potential imaging biomarker in patients with colorectal liver metastasis [13]. In a recent study, Salem et al. evaluated the DpR in patients in the IMBRAVE150 trial with HCC that received either immunotherapy (atezolizumab + bevacizumab) or tyrosine-kinase inhibition (sorafenib) [42]. In that study, patients that received immunotherapy had a higher DpR than patients that received tyrosine-kinase inhibition. The results of that study indicated that DpR showed potential as an additional, novel tool for evaluating the treatment response in patients with HCC. However, in contrast to the ETS, the DpR is defined as the maximum tumor shrinkage. Therefore, it is unclear at follow-up investigations whether maximum tumor shrinkage has yet been achieved. Thus, DpR is a retrospective measure, which is only available after progression occurs. Consequently, the DpR is not useful for assessing treatment responses at early stages, and its use is limited to clinical studies. In contrast, the ETS can be readily calculated at the first imaging evaluation; thus, it has more potential as a tool for routine clinical settings. In future studies, it could be interesting to compare the DpR with the ETS and evaluate their correlations with OS and PFS.

Since the abovementioned positive IMbrave150 results, in which the combined therapy of the immune checkpoint inhibition atezolizumab and the VEGF antibody bevacizumab showed significantly improved survival outcome, the number of trials on immune checkpoint inhibition in combination with other biological therapies for unresectable HCC has tremendously increased [5, 6, 43]. In our study, the majority of patients (n = 22) received the combination of atezolizumab and bevacizumab and therefore a combination of immune checkpoint inhibition and VEGF inhibition. Thus, we believe that our results are also indicative for combined treatment of immune checkpoint inhibitors and other biological therapies.

This study had several limitations. First and foremost, the study design was retrospective, and thus, it had inherent limitations. Second, the number of patients was limited. However, the dataset was well investigated as we only included patients with complete clinical, laboratory, and imaging data. Third, we included patients treated with various immunotherapeutic agents to validate the biomarker in terms of a “real-life” clinical setting. However, the small cohort meant that subgroups for each immunotherapy agent would have been too small for a well-powered analysis. Future studies should validate the ETS as an imaging biomarker for various immunotherapeutic agents and different treatment lines. Fourth, in our study, the ETS cut-off for optimal stratification ranged between 8.1 and 12.3%. This range can be attributed to the limited number of patients in our cohort. As 10.0% falls within this range, we followed this previously suggested cut-off in our study [20], which is lower than the 20.0% cut-off which has also been published [21]. Therefore, further studies investigating the influence of different ETS cut-off values for patients with HCC and immunotherapy with larger patient numbers are mandatory for optimal cut-off selection. Fifth, in our study most of the patients (n = 35, 89.7%) received CT imaging and only a minority of 4 patients (n = 4, 10.3%) received an MRI during diagnosis and follow-up. Although the current EASL guideline does not recommend one of the methods over the other, there is evidence that MRI is more sensitive in the detection of small liver lesions [2]. Furthermore, sensitivity of MRI for the detection of smaller nodules and therefore new lesions during follow-up could have been further increased with the use of liver-specific contrast agents [44].

Despite these limitations, this proof-of-concept study was the first to show a direct correlation between ETS and the survival outcomes in patients with HCC undergoing immunotherapy. Our results revealed that the ETS has huge potential as a novel imaging biomarker for patients with HCC. Moreover, the ETS may also serve as a useful addition to the conventional response categorization. Our positive preliminary results suggested that the ETS could function as an additional and easy to calculate parameter for the early identification of treatment responders. Thus, our results could serve as the foundation for further research on ETS as a novel imaging biomarker.

We found that the ETS was strongly associated with survival outcomes in patients with HCC undergoing immunotherapy. Its predictive capability was further improved in the subgroup of patients with tumor control according to mRECIST. Thus, ETS is a highly promising, readily assessable imaging biomarker for patients with HCC undergoing immunotherapy. In future, ETS could identify patients that are most likely to benefit from immunotherapy, and thereby, it could support clinical decision-making.