In the present study, we have demonstrated that GPS, an inflammation-based prognostic score, is associated with tumor progression and reduced liver function and can be considered an independent marker of poor prognosis in patients with HCC notwithstanding the disease stage and/or liver functional status along with the CLIP score. Moreover, GPS proved to be more suitable than mGPS in patients with HCC with regard to its discriminating ability and the monotonicity of gradients.
Links between cancer and inflammation were first established in the nineteenth century and that were based on observations that tumors often arose at sites of chronic inflammation and that inflammatory cells are present in biopsied samples collected from tumor tissues. Today, it is well known that cancer promotes release of proinflammatory cytokines from tumor cells. The cytokines interact with immunovascular system and facilitate cancer growth, invasion, and metastasis [23, 24].
Recent studies have shown that elevated serum CRP levels may be associated with tumor size, distant metastasis, vascular invasion, lymph node metastasis and tumor recurrence, resulting in poor prognosis in patients with various cancers, including HCC [16, 25].
It has been also reported that serum albumin participate in systemic inflammatory response and that decline of its serum level is a poor prognostic factor for long-term survival in patients with various cancers [8, 19].
Based on these reports, GPS, incorporating CRP and serum albumin levels, may reflect both presence of the systemic inflammatory response (CRP), and the progressive nutritional decline (albumin) in patients with cancers, resulting in poor survival outcome .
Consistent with the Ishizuka’s report , our study demonstrates that an elevated GPS is associated with factors of tumor progression such as: maximal tumor diameter, tumor number, vascular invasion, extra hepatic metastasis, higher CLIP scores, higher JIS scores, higher BCLC scores and higher TNM classification grade. In addition, our data show that an elevated GPS is also associated with factors indicating reduced liver function such as higher T-Bil, lower albumin, higher ICG and higher Child-Pugh scores. This is partly because, unlike in the Ishizuka’s report, in our study we enrolled patients not eligible for surgical resection: with more advanced stages of the disease and reduced liver function.
Although significant differences in overall survival were found across all staging systems, such differences were not observed between mGPS 1 versus 2, CLIP score 0 versus 1, CLIP score 3 versus 4, CLIP score 5 versus 6, JIS score 0 versus 1, BCLC C versus D, and TNM grade Іversus ІІ, suggesting poorer discrimination ability in early and advanced stages of the disease. Only GPS demonstrated significant differences in overall survival between all adjacent strata, indicating that GPS could discriminate between early and advanced stages of the disease.
Linear trend χ2 test and −2 log likelihood calculated using the Cox model showed that GPS has fairly good ability of discriminating the survival of patients in different HCC stages and has greater homogeneity of survival among patients within the same stage, particularly in the curative treatment group, suggesting that it can provide better prognostic power than other scoring systems in the setting of curative treatment.
On multivariate analysis, GPS was independently associated with overall survival along with the CLIP score, which is consistent with the Ishizuka’s report . However, our results indicate that GPS is an independent marker of poor prognosis in patients with HCC in various stages of disease and different liver functional statuses. By contrast, JIS, BCLC, and TNM were not found to be independent poor prognostic factors. A scoring system should be simple and easy to apply for prognosis prediction before treatment is initiated. In this regard, GPS can be a useful tool for prognostication and stratification of patients with HCC, because being based on only 2 laboratory data, CRP and albumin, it is conventionally available without additional imaging techniques or histological examinations before commencing treatment .
The Cox model and AUC analysis showed GPS is more suitable than mGPS for patients with HCC with regard to discriminating ability and the monotonicity of gradients. Recently, in a Glasgow Inflammation Outcome Study, Proctor et al. have shown that mGPS was a powerful prognostic factor independent of tumor site in patients with cancer and was superior to GPS in the greater consistency and more general use . Their observations were based on the results that a low albumin concentration alone was uncommon (<10% of all patients) and was not significantly associated with cancer-specific survival in many cancers including hepatopancreaticobiliary cancer (P = 0.209). However, our study included 38 (25.3%) patients with low albumin concentration alone. Furthermore, serum albumin is one of the components of the Child-Pugh Classification and hypoalbuminemia has been reported as an independent poor prognostic factor in patients with HCC . Moreover, in the Glasgow Inflammation Outcome Study, hepato-pancreatico-biliary cancer included pancreatic and biliary tract cancers besides HCC. Therefore, we speculate that GPS is more suitable than mGPS for patients with HCC.
A number of scoring systems for HCC have been proposed to date. However, controversy remains about which system is best at predicting survival of HCC patients. Characteristics of tumor-related variables, the relative score weighted for each variable, preferred treatment modality in different centers, the numbers of analyzed patients, and the etiology of liver diseases could contribute to this controversy . In the current study, the CLIP system proved to be the best prognostic model for HCC with regard to discrimination ability and the monotonicity of gradients. This result is consistent with a previous study with HCC patients undergoing palliative TACE and another study with advanced HCC patients [2, 26]. The CLIP system was originally derived from unselected patient population and the majority of them had received non-surgical treatment . Therefore, it is generally accepted that the CLIP system may be more suitable for predicting the survival of HCC patients who receive non-surgical treatments than BCLC system or JIS system [26, 28]. In our study, many patients (n = 141, 94%) were treated non-surgically, which may be an important reason why the CLIP was superior to the BCLC or the JIS in predicting survival. However, a recent study from Taiwan has demonstrated that the CLIP system was the best prognostic model for HCC in terms of prognostic stratification for patients from early to advanced cancer stage irrespectively with curative or non-curative treatments . This result indicates that prediction accuracy of the CLIP system is highly stable and is independent of the treatment strategy. When GPS was combined to the CLIP system to form a new prognostic system, named inflammation-based CLIP provided even better prognostic accuracy than GPS alone, suggesting that addition of GPS could improve the discriminatory ability of the CLIP system.
The current study has some limitations. First, it was a retrospective, small sample size, single-center study. Second, the majority of patients enrolled were treated non-surgically. Our results may not be applicable to surgically oriented centers or high volume centers with frequently performed liver transplantations. Third, the therapeutic effects of each treatment method were not included into prognostic factors’ evaluation. Since many patients received multiple treatment sessions due to tumor recurrence during their follow-up periods (tumor progression and worsening liver functional reserve), it was difficult to evaluate all therapeutic effects as prognostic factors in this patient population. Fourth, there may be the potential causal relationships between liver function and inflammation. As progressive HCC patients would have poor liver function and reduced albumin, GPS in this group of patient is no longer only a marker of inflammation but also a marker of poorer liver function. However, Cervoni et al. have demonstrated that a systemic inflammation response, as evidenced by an elevated CRP concentration, is associated with poor survival in Child Pugh score > B8 cirrhotic patients independently of Model of End Stage Liver Disease . We also have shown that CRP is an independent marker of poor prognosis in patients with HCC, irrespective of tumor stage and liver function .