Evaluation of Preoperative Inflammation-Based Prognostic Scores in Patients With Intrahepatic Cholangiocarcinoma: A Multicenter Cohort Study

Background Accumulating evidence has indicated the vital role of inflammation-based score (IBS) in predicting the prognostic outcome of cancer patients. Otherwise, their value in intrahepatic cholangiocarcinoma (iCCA) remains indistinct. The present study aimed to evaluate whether IBSs were related to survival outcomes in iCCA patients. Method Clinical characteristics were retrospectively collected in 399 patients diagnosed with iCCA from cohorts of Sun Yat-sen University Cancer Center (SYSUCC) and the First Hospital of Dalian Medical University (FHDMU). The survival curves were constructed with the Kaplan-Meier method and compared with the log-rank test. Univariate and multivariate analyses were conducted to determine the prognostic factors of overall survival (OS) and progression-free survival (PFS). The concordance index and the area under the time-dependent receiver operating characteristic (ROC) curves (AUROCs) were used to compare the predictive value of inflammation-based scores in terms of survival outcomes. Results The significant survival differences in OS and DFS were observed when patients were stratified by the modified Glasgow Prognostic Score (mGPS) (p<0.001). Multivariate analysis demonstrated that higher mGPS score was independently associated with poor OS and DFS (p<0.001). The predictive accuracy of the mGPS was superior to other IBSs (all p<0.001) in survival prediction in iCCA patients. The findings were further supported by the external validation cohort. Conclusion The mGPS is a sensitive, efficient, simple and widely applicable preoperative prognostic factor for iCCA patients. Thus, more effective therapy and frequent surveillance should be conducted after surgical resection in iCCA patients with higher mGPS scores.


BACKGROUND
Intrahepatic cholangiocarcinoma (iCCA) is the second most common malignant tumor ranking after hepatocellular carcinoma (1). Although iCCA patients in different stages can be treated with various modalities, including surgery resection, chemotherapy, and radiation therapy, the overall incidence and mortality have shown a worldwide increase in the past decades (1,2). Even though surgical resection provided the best chances to obtain prolonged survival, the median progression-free survival (PFS) time was reported to be merely 12 to 36 months in patients with resectable iCCA (3). To optimize risk-benefit assessments and stratify the patients for more individualized treatment, there is an urgent demand to seek an objective, sensitive and reliable prognostic marker for patients with iCCA. Currently, common prognostic markers, such as tumor margins, tumor differentiation, and lymph node metastases, are determined only after surgical resection (2). Therefore, there is continuing momentum in finding a practical pre-operative prognostic marker that could facilitate accurate patient stratification before surgery and improve therapeutic outcomes.
Inflammation, as a new hall marker of cancer (4), plays a vital role in the progression of tumors (5). Tumors produce inflammatory chemokines and cytokines and are locally infiltrated by leucocytes (6). Moreover, the activation of the ongoing systemic chronic inflammatory response will further lead to cachexia (6). According to these pieces of evidence, many inflammation-based scores (IBSs) were proved to be prognostic in various tumors, including Glasgow Prognostic Score (GPS) and modified Glasgow Prognostic Score (mGPS) (7), Prognostic Index (PI) (8), Prognostic Nutritional Index (PNI) (9), systemic immune-inflammation index (SII) (10), neutrophil to lymphocyte ratio (NLR) (11), platelet to lymphocyte ratio (PLR) (12), and lymphocyte to monocyte ratio (LMR) (13). Nonetheless, the research about reliable and valid inflammation-based scores in patients with iCCA after resection remains supplemented. Besides, most previous studies were conducted in a single center with a small number of patients and were mostly concentrated on a certain single IBS (14)(15)(16)(17). Thus, for evaluating the validity of the IBSs in iCCA patients, a multicenter study with a large volume of patients would be necessary and imperative. According to these findings, our study aimed to find the best combination of inflammatory factors that could predict survival outcomes for iCCA patients after surgical resection.  Figure 1. Clinical characteristics were retrospectively aggregated from the electronic medical record and were exhibited in Table 1. This study obtained the written informed consent from all the patients and was approved by the ethics committees of two participating centers.

Survival Outcomes and Follow-Up
The study's outcome variables, overall survival (OS) and PFS, were calculated from the date of surgery to the date of death and tumor progression, respectively, or the last follow-up. The first post-operative follow-up was conducted at 30 days after surgical resection, then every three months for the first year, and every six months until death or dropout. Follow-up data of two cohorts were retrieved on November 30, 2020.   ) were carried out. The preoperative imaging evaluations for iCCA included abdomen computed tomography (CT), chest CT, pelvis CT and magnetic resonance imaging. Those patients with jaundice or dilated bile ducts routinely underwent biliary drainage. Once the regional LN metastasis was implicated in the preoperative imaging evaluations or suspected during surgery, all resectable regional LNs were dissected. The postoperative pathological stage of iCCA was classified according to the eighth AJCC TNM staging system. Moreover, the adjuvant chemotherapy was routinely implemented in the patients with more advanced or aggressive tumors, particularly those with LN metastasis.

Inflammation-Based Scores
According to our previous study (18,19) related to the survival predicting performance of IBSs, NLR, PLR, LCR, LMR, PI, GPS, mGPS, PNI, and SII were included and calculated in this multicohort study to identify the IBS with highest accuracy to predict poor OS and PFS in iCCA patients. The details of IBSs were described in Table 2.

Statistical Analysis
Continuous variables were reported with median and interquartile range. Categorical variables were reported with whole numbers and proportions. Proportions were compared using the chi-square test or the Fisher Exact test. Distributions of continuous variables were compared using the Mann-Whitney U test. Maximally selected rank statistic from the R package was employed to identify the optimal cutoff points of NLR, PLR and WBC, white blood cell; HGB, hemoglobin; PLT, platelets; ALT, alanine aminotransferase; AST, glutamic-oxalacetic transaminase; GGT, gamma-glutamyl transpeptidase; ALP, alkaline phosphatase; ALB, Albumin; TBIL, total serum bilirubin; IBIL, indirect serum bilirubin; CRP, C-reaction protein; CA19-9, carbohydrate antigen 19-9; CEA, carcinoembryonic antigen; LCR, lymphocyte-C-reactive protein ratio; mGPS, modified Glasgow prognostic score; NLR, neutrophil-lymphocyte ratio; LMR, lymphocyte to monocyte ratio; PLR, platelet to lymphocyte ratio; SII, systemic immune-inflammation index; PNI, prognostic nutritional index; PI, prognostic Index; LN, lymph node.

The External Validation of Significant Prognostic Factors
According to the statistic results in the primary cohort, an external validation was conducted. The significant prognostic factors, which were defined in the primary cohort, were validated in the FHDMU cohort. The multivariate analysis based on the validation cohort indicated that only mGPS was an independent prognostic factor for both OS and PFS (  Figure 1).

Comparison of the Predictive Power of IBSs on Survival Outcomes in Two Cohorts
ROC curves and AUROC values were analyzed to contrast the prognostic capacity of eight IBSs in both primary ( Figure 4 and Table 5) and validation cohort ( Figure S2 and Table 5

DISCUSSION
Over the past decades, various progresses have been made in prophylaxis and treatment of cholangiocarcinoma (1,2).
Nevertheless, the OS and PFS of iCCA patients remained poor (3). For the prediction of prognosis, the TNM grade system has been applied as the mainstream prognostic assessment system since it was presented. However, the TNM grades can only be Abbreviations as in Table 3. calculated according to the postoperative pathological factors, and the systemic inflammation level was not included in the TNM grade system. As a result, the TNM grade system cannot make a preoperative overall assessment to guide the therapeutic strategy. To fill this gap, there has been an urgent demand to explore and validate a pre-operative potential prognostic factor for patients with iCCA. IBS, as a combination inflammation index, can objectively reflect the level of inflammation, and further indicate the prognostic and outcomes of cancer patients. The present study compared the prognostic efficacy of eight common IBSs in patients with iCCA. The univariate and multivariate analyses were further performed to verify the prognostic factors. The mGPS was identified as a significant prognostic factor for predicting OS and PFS in both SYSUCC and FHDMU cohort. Moreover, it was shown that mGPS was superior to the other IBS indexes for predicting the OS and PFS of iCCA patients. It is worth noting that classical pathological elements and TNM staging system made no significant association with the OS and PFS in this study. The probable reasons for the outcome are as follows. First, the powerful predictive performance of mGPS for OS and PFS might mask the role of pathological elements in the multivariate analysis. As a result, the pathological factors showed no significant difference in multivariate analysis. However, that was not intended to deny the predictive effect of pathological elements. As our univariate analysis presented, the TNM staging system was still a statistically significant predictor of prognosis (P = 0.003 in OS, P<0.001 in PFS). Second, the TNM system is a continuously updating and evolving standard with its graded prognostic effect remains controversial. For instance, invasive liver capsule may not adequately reflect the pathogenesis of iCCA tumor, due to the influence of tumor location and tumor size (20). Additionally, the definition of category T3 could barely indicate the biological extent of iCCA tumor (21). Moreover, the number of lymph nodes determined by preoperative imaging examinations and intraoperative findings cannot objectively indicate lymph node metastasis, which may further lead to misjudgment or underestimation of N stage (20,21). In addition, a further analysis was conducted to elucidate the relationship between mGPS and some clinical and pathological characteristics (Supplementary Table 1). The results demonstrated that there was a significant correlation between mGPS and tumor marker (CA19-9, CEA), satellite sites, microvascular invasion, tumor size, lymph node metastasis and TNM stages. These characteristics were significantly related to the poor survival outcomes. Different from these factors, mGPS could be assessed preoperatively. It was worth noting that the CRP level was the key point between mGPS 0 group and mGPS 1/2 groups. In the present study, mGPS 0 group presented a better survival outcome than mGPS1/2 group did. This could also certify the vital role which inflammation played in tumor progression. The GPS staging system was firstly established in inoperable non-small-cell lung cancer (22), with two major evaluative dimensions: serum ALB and CRP. Serum ALB may indicate the general status as well as the amount of lean tissue of cancer patients. Furthermore, hypoalbuminemia is also associated with cachexia. ALB has been shown to be a prognostic marker in gastric cancer (23) and pancreatic cancer (24), and the role of albumin as a marker of inflammation has been underscored by recent research in malignancy (25). In addition, iCCA, as a type of liver cancer, can weaken the synthesis function of the liver, further leading to the hypoalbuminemia. On the other hand, CRP is not only a sensitive indicator of the systemic inflammatory response. Accumulating evidence indicated the role of CRP in the tumor development and metastasis (26,27). Theoretically, high CRP level may be due to the production of cytokines from tumor cells (26). As an acute-phase protein, CRP together with IL-6, TNF, and other cytokines further initiates or sustains the systemic inflammatory response (27). Then, inflammation promotes the tumor proliferation, angiogenesis, invasion, and metastasis as a feedback loop (4). It has been confirmed that high level of CRP was correlated with unfavorable  Table 3.
survival in esophageal carcinoma (28), colorectal carcinoma (29), as well as multiple myeloma (30). Moreover, as a commonly used index, CRP has high sensitivity and cost-effectiveness and is easily obtained in clinical practice. With the increasing numbers of studies about GPS and survival in patients with cancers, researchers found that hypoalbuminemia regularly occurred with elevated CRP levels (31). Moreover, the survival outcomes of patients with hypoalbuminemia alone were significantly better than patients with elevated CRP levels, indicating that CRP played a more important role in survival prediction. In case of that, GPS was modified into mGPS (32). Since then, the modified GPS has been occupied in colorectal cancer (32), hepatocellular carcinoma, esophageal cancer (33), and ovarian cancer (34) and simultaneously presented robust prognostic prediction. Significantly, this is the very first study which evaluating the prognostic prediction of the common IBSs in iCCA patients. In this large, multicenter cohort study, we compared the survival curves of these eight frequently used IBSs. Surprisingly, mGPS was not only the independent prognostic factor of OS (P < 0.001) and PFS (P < 0.001) in both of our cohorts, it presented the most powerful performance of prognostic prediction in the common IBSs (all P < 0.005). Similarly, mGPS also presented prominent prognostic manifestations in perihilar cholangiocarcinoma (35) and biliary tract cancer (36) in previous studies. Furthermore, by contrast with the pathological prognostic factors, mGPS, as an inflammation-based score, could make the pre-operative prediction of cancer patients to facilitate accurate stratification and further improve the survival outcomes. Besides, assessments of serum albumin and CRP are simple and inexpensive compared to genetic assessments, which are complicated and expensive.
According to the results of the present study, the treatment strategies of patients with higher mGPS should be optimized. Clinical staff should be especially caution about the indications and contraindications of operation and take careful consideration about the overall healthy situation of these patients. The shorter follow-up intervals were conducive to earlier detection of tumor recurrence or progression. And this would further provide an opportunity for early medical intervention in recurrence. Moreover, the inclusion of routine postoperative chemotherapy in the overall treatment strategies may be beneficial.
Certain limitations of the present study merit discussion. First, the retrospective nature is a potential limitation; we enrolled two cohorts from different regions to restrain this limitation. Second, improvements in perioperative management and treatment methods may lead to the heterogeneous antitumor treatments of our patients' cohorts and further interfere with the result of the present study. Third, the underlying mechanism of mGPS and poor survival outcome has not been fully demonstrated. Finally, further extensive trans-regional studies were needed to verify the prognostic power of mGPS in iCCA patients.
In conclusion, the present study, we identified mGPS as a sensitive, efficient, simple, rapid, and widely applicable preoperative prognostic factor for iCCA patients. Elevated mGPS scores indicated poor prognosis for these patients. Thus, more effective therapy and frequent surveillance after treatment should be conducted for the iCCA patients with higher mGPS scores.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the institutional review board of Sun Yat-sen University Cancer Center and the first affiliated hospital of Dalian Medical University. The patients/participants provided their written informed consent to participate in this study.