Hepatocellular carcinoma (HCC) is the fifth most common cancer entity worldwide and is responsible for the second-highest number of cancer-related deaths (1, 2). Following the current guidelines from the European Association for the Study of the Liver (EASL) and the American Association for the Study of Liver Diseases (AASLD), the Barcelona Clinic Liver Cancer (BCLC) classification system is the preferred framework for treatment allocations and prognosis predictions (3, 4). According to the BCLC classification, transarterial chemoembolization (TACE) is the standard of care for patients with intermediate-stage HCC (5, 6). However, there is substantial heterogeneity among patients with intermediate-stage HCC, due to considerable differences in tumor burden and liver function (7). This heterogeneity hampers prognosis prediction, and consequently, it remains difficult to make treatment decisions for these patients. To support clinicians in the decision-making process, several scoring systems have been proposed (8–10). However, all have failed in external validation (11, 12).

Recently, a novel scoring system, based only on tumor burden, was promoted by Wang et al. (13). They called this system the Six-and-Twelve (SAT) score. It sums the number of tumor nodes and tumor size to obtain a readily applicable stratification system. In the original study, the score showed promising performance, but in an external validation, contrary results were found for the prognosis prediction (14–18).

Additionally, the Tumor Burden Score (TBS) is a novel scoring system, based on a combination of tumor number and tumor size. The TBS was originally developed for patients undergoing resections of colorectal liver metastasis. However, very recently, the TBS was successfully applied to stratify patients with HCC undergoing TACE (19).

A third novel scoring system, the Seven-Eleven criteria (SEC), was recently developed to estimate tumor burden (20). These criteria also include tumor size and the number of tumor nodes in the calculation. The introduction of the SEC aimed at improving the already existing up-to-7 criteria, which have been originally developed for patients with HCC undergoing liver transplantation, and the up-to-11 criteria, which are part of the BCLC stage B subclassification, through a combination of both (21–23).

Overall, it remains unclear which of these tools might be optimal for estimating tumor burden in HCC. A direct head-to-head comparison of all scoring systems is lacking. Furthermore, it is questionable whether tumor burden alone is sufficient to support clear-cut treatment decisions, because HCC development is associated with liver cirrhosis in more than 80% of Western patients (3). Thus, most patients have two diseases: HCC and liver cirrhosis, and both contribute to an impaired prognosis. The complex interactions between cirrhosis and tumor development and progression might be underrepresented in a scoring system based only on estimates of the actual tumor burden.

This study aimed to perform the first head-to-head comparison of SAT, TBS, and SEC performances in predicting HCC prognosis. Additionally, we aimed to compare these scores to several established scoring systems in terms of prognostic power.

This multicentric retrospective study was approved by the Ethics committee of the Medical Association of Rhineland Palatinate, Mainz, Germany (permit number 2021-15913). The other responsible Ethics committees followed this approval. The requirement for informed consent was waived, due to the retrospective nature of the analysis. Patient records and information were anonymized and de-identified prior to analysis. TRIPOD guidelines were applied to the organization of this manuscript (24).

To the best of our knowledge, this study was the first to perform a head-to-head-comparison of the SAT, the TBS, and the SEC for predicting the survival of patients with HCC undergoing TACE. In summary, all three scoring systems could successfully stratify patients in univariate analyses. However, only the SAT was identified as an independent prognostic factor in the multivariate analysis. Furthermore, all three scores showed only modest predictive ability.

In a recent study, Ho et al. promoted the TBS as a novel tumor burden-related scoring system and a new prognostic marker for patients with HCC undergoing TACE (19). The TBS was originally developed in patients with colorectal liver metastasis undergoing resections (34). Based on its promising results for those patients, several authors evaluated whether the TBS could be applied to patients with HCC undergoing curative treatments (35–37). However, among patients with HCC, the suitability for resection, according to the remnant liver function, differs tremendously between intermediate and advanced stages. For patients undergoing TACE, Ho et al. showed that the TBS could accurately discriminate risk stratifications for OS (19). However, subgroup analyses on patients within different BCLC stages and ALBI grades showed considerable overlap in the survival curves. Additionally, a comparison with stratification systems incorporating remnant liver function is missing.

Notably, in the Ho et al. study, the TBS showed a disparate distribution of patients in different risk categories. The TBS predicted that 72% of patient were at medium risk of mortality, compared to only 20% at low risk and 8% at high risk. Those findings were consistent with our results, where the TBS predicted that 76% were at medium risk, only 21% were at low risk, and only 3% were at high risk. This unequal distribution might have contributed to the lower predictive ability we found for the TBS compared to the SAT and the SEC, which showed a more balanced distribution (13, 20).

In contrast to our study cohort, viral liver disease was the most common etiology of HCC in the Ho et al. cohort (19). Furthermore, compared to that study, our cohort had considerably lower initial albumin levels and higher bilirubin levels, which indicated poor liver function. Similarly, in the initial reports on SAT and SEC, viral liver disease was the most common etiology, and liver function was considerably better in those cohorts than in our study cohort (13, 20). In the original study on the SAT, the score outperformed the BCLC classification and the ALBI score in predictive ability. In external validation studies, the C-Indexes for the SAT ranged between 0.60 and 0.70 (14–18). However, in all the validation studies, the predictive performance of the SAT was within the same range or somewhat below the performance of the BCLC and ALBI scoring systems (14). In the original study on the SEC, the authors did not compare the SEC with scoring systems that incorporated risk factors other than tumor burden (20). In our first external validation of the SEC score’s predictive performance, the C-Index was 0.58, which fell between the C-indexes of the SAT (0.59) and the TBS (0.54).

Our study showed that a low albumin level and a high bilirubin level were both independent predictors of impaired survival. Consequently, for our cohort, systems that took the remnant liver function into account performed best. These results confirmed our initial hypothesis that the complex interplay between cirrhosis and tumor development and progression might be underrepresented in scoring systems that are only based on tumor burden. Nevertheless, in daily clinical routine, a rapid estimation of tumor burden may provide a convenient bedside tool (particularly the SAT). Additionally, our results suggested the SAT performed better than the SEC and the TBS; thus, the SAT might be useful for characterizing study populations according to tumor burden in clinical trials. However, our results also indicated that none of the scores alone could support treatment decision making in daily clinical routines.

Apart from tumor burden and liver function, several other biomarkers play an important role in the evaluation of patients with HCC. Macrovascular tumor invasion has been identified as an independent factor in these patients (38, 39). Although macrovascular invasion appears more often in patients with higher tumor burden, it is not necessarily associated to multifocal tumor growth (40). Therefore, apart from the tumor burden, macrovascular invasion has to be considered as an additional and independent factor when evaluating biological tumor aggressiveness in these patients. Consequently, even in cases where tumor burden might be low, patients with macrovascular invasion are at higher risk for a poor post-TACE prognosis and immediate switch to systemic treatment should be discussed.

Following our results, we could not substantiate the benefit of continually developing novel scoring systems for ongoing scientific discussion. Instead, we recommend focusing on continually improving and updating the existing systems. Growing knowledge on novel risk factors, including the immune system and the nutritional status of patients, could be further integrated with growing knowledge of automated artificial intelligence-based risk predictions (25, 41). The novel developments in this field might allow integrating data pipelines in clinical and radiology information systems to facilitate the automation of highly individual risk predictions.

Clearly, the present study had several limitations. First, it had the limitations inherent to a retrospective study design. Thus, our results should be validated in a prospective trial. Second, we did not perform subgroup analyses of patients treated with different TACE techniques. However, in multiple previous studies, no significant differences in OS have been reported between conventional TACE and drug-eluting bead delivery TACE techniques (42–44). Moreover, recent study results have indicated that the applicability of various scoring and staging systems was similar for the different types of TACE (45).

In this direct head-to-head comparison of three scoring systems for HCC prognosis, we found that the SAT was superior to the TBS and the SEC in survival stratification and in predictive ability. Thus, the SAT should be preferred when stratifying patients according to tumor burden; e.g., in the context of clinical trials. However, none of these scores was superior to established scoring systems in predicting survival. Therefore, they offered no added benefit for daily clinical routines, and treatment decisions should not be made based on tumor burden alone. Instead, a thorough interdisciplinary discussion that considers all types of risk factors is mandatory for improving patient stratification, and ultimately, patient survival.

The datasets presented in this article are not readily available because data cannot be shared publicly because of institutional and national data policy restrictions im-posed by the Ethics committee of the Medical Association of Rhineland Palatinate, Mainz, Germany since the data contain potentially identifying patient information. Data are available upon request for researchers who meet the criteria for access to confidential data. Requests to access the datasets should be directed to roman.kloeckner@unimedizin-mainz.de.

The studies involving human participants were reviewed and approved by Ethics committee of the Medical Association of Rhineland Palatinate, Mainz, Germany. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

LM, FH, TA, UF, BG, JH, SZ, M-SK, ME, TD, DB, VS, D-HC, DZ, DP, and RK devised the study, assisted in data collection, participated in the interpretation of the data, and helped draft the manuscript LM, TA, UF, JH, SZ, M-SK, ME, TD, DB, VS, DZ, and RK carried out the data collection. BG, D-HC, and DP supported the data collection efforts. LM, FH, DP, and RK created all of the figures and participated in the interpretation of data. LM, FH, DP, and RK performed the statistical analysis. All authors contributed to the article and approved the submitted version.

LM is supported by the Clinician Scientist Fellowship “Else Kröner Research College: 2018_Kolleg.05”.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.