Inflammatory markers in intrahepatic cholangiocarcinoma: Effects of advanced liver disease

Abstract Background To investigate the neutrophil‐to‐lymphocyte ratio (NLR), platelet‐to‐lymphocyte ratio (PLR), and systemic immune‐inflammation index (SII) as prognostic biomarkers in intrahepatic cholangiocarcinoma (ICC) with a focus on viral hepatitis and liver status. Methods In this retrospective cohort study, patients from the institutional cancer registry with ICC from 2005 to 2016 were stratified by treatment group. Baseline inflammatory markers were dichotomized at the median. Overall survival (OS) was assessed via Kaplan‐Meier curves and Cox proportional hazard models. Multiple patient, liver, and tumor factors were included in the multivariable analysis (MVA). Results About 131 patients (median age 65 years, 52% male, 76% Caucasian) had a median OS of 13.0 months. Resection/interventional oncology with/without systemic therapy had improved survival vs systemic therapy alone in Child‐Pugh A patients (P < 0.01). In Child‐Pugh B/C patients, this survival difference became nonsignificant (P = 0.22). Increased NLR and SII were associated with decreased survival (P < 0.01), while dichotomized PLR was not (P = 0.3). On MVA, increased NLR remained an independent prognostic factor (HR 1.6, P < 0.05). In Child‐Pugh class A (n = 94), low‐NLR had higher OS vs high‐NLR (25.4 vs 12.2 months, P < 0.01). In Child‐Pugh class B/C (n = 28), NLR did not have a significant effect on median OS (low‐ vs high‐NLR: 6.7 vs 2.9 months, P = 0.2). Child‐Pugh class acted as an effect modifier on MVA for NLR (P = 0.0124). Conclusions The NLR has a stronger impact as a prognostic marker in ICC over the PLR and SII. This survival effect is decreased in advanced liver disease.


| Study population
The protocols and methods were conducted in compliance with the Health Insurance Portability and Accountability Act and were approved by the institutional review board. Patients diagnosed with ICC from 2005 to 2016 were retrospectively identified from the institutional cancer registry of a single urban academic center. Treatment allocation had been determined by a multi-disciplinary tumor board. Exclusion criteria included individuals under the age of 18, incomplete treatment or survival data, and those with a histopathologic diagnosis of hepato-cholangiocarcinoma.

| Data acquisition
The variables reported by the cancer registry included age, gender, race/ethnicity, insurance status, marital status, American Joint Committee on Cancer (AJCC) staging, and basic treatment status. Furthermore, electronic medical record review was conducted to identify multiple other tumor and liver factors at the time of diagnosis along with the temporal sequence of treatments. Along with liver factors, baseline laboratory values were used to calculate Model for End-Stage Liver Disease (MELD) and Child-Pugh scores were missing from the medical record. The Charlson Comorbidity Index 28 (CCI) was calculated to quantify the burden of non-ICC-related disease. Viral hepatitis was defined via International Classification of Diseases 9th Edition codes as well as HBsAg and HCV Ab laboratory data and included patients at various stages of treatment, including the untreated.
Treatment allocation was first stratified into systemic therapy (chemotherapy and/or radiation), IO (comprised of transarterial chemoembolization, radioembolization, and thermal ablation), resection, and supportive therapy (palliative care or no treatment). Patients who received IO or resection following or in conjunction with systemic therapy were classified as IO and resection, respectively. "Nonsurgical treatment" consisted of IO and systemic therapies. Patients who received treatment were further reclassified into three groups: those who received resection or IO first with/without subsequent systemic therapy (Group 1); patients who received systemic therapy followed by resection or IO (Group 2); and patients who received systemic therapy alone (Group 3). Overall survival (OS) was defined as time from diagnosis to time of death, regardless of etiology.
Baseline pretreatment absolute neutrophil, lymphocyte, and platelet numbers were determined from complete blood count with differential values drawn within 30 days prior to treatment or 30 days prior to diagnosis (for supportive treatment patients only) and were used to calculate the various inflammatory markers. In instances where the absolute neutrophil count (ANC) or absolute lymphocyte count (ALC) were not reported in the electronic medical record, neutrophils were calculated by multiplying the white blood cell count by the differential percentage of the neutrophils and similarly for the lymphocytes. The NLR ratio was defined as ANC/ALC. The PLR ratio was defined as platelet count/ALC, and the SII was defined as ANC*platelet count/ALC.

| Statistical analysis
Continuous variables were compared using the Wilcoxon rank sum test and categorical variables using the chi-squared test. Kaplan-Meier methods and log-rank tests were used to 5918 | SELLERS Et aL. Patients with missing variables were conservatively excluded from analyses to account for bias. All P-values reported are two-sided, with an alpha level of 0.05 considered statistically significant. Statistical analyses were completed using JMP Pro v.13.0.0 (SAS Institute Inc, Cary, NC), and GraphPad Prism v.7.0a for Mac (GraphPad Software, La Jolla, CA). Additional figure preparation was completed using R v.3.4.3 (R Core Development Team, Vienna, Austria).
Patients who received surgical treatment were more likely to be younger, Child-Pugh class A, AJCC Stage I, have solitary tumors, unilobar disease, and to not have metastases (P < 0.01). Supportive therapy patients had higher MELD and CCI scores than nonsurgical or surgical treatment (P < 0.01). Gender, race, insurance, viral hepatitis status, and the presence or absence of cirrhosis did not affect the likelihood of receiving treatment (P > 0.05). Surgical treatment was associated with lower median tumor size (4.2 cm) vs nonsurgical treatment (7.0 cm) or supportive care (7.5 cm, P = 0.0006).
Median NLR for the cohort was 3.95 (IQR 2.36-6.19), with median PLR of 156.43 (IQR 110.6-225.6) and median SII of 867.4 (455.0-1421.5). Correlation between inflammatory markers ranged from good (Spearman's ρ = 0.65) to very strong (ρ = 0.84). Decreased NLR and SII were associated with increased survival (P < 0.01) (Figure 2A,B), while dichotomized PLR was not (P = 0.3) ( Figure 2C). ANC, ALC, log-transformed PLR, and log-transformed SII were not independent prognostic factors on MVA (P > 0.05). However, an increase in log-transformed NLR continued to be associated with decreased survival (HR 1.96, P = 0.0001) on MVA. Given that the NLR was the only inflammatory marker to remain significant on MVA, the NLR was used for further subgroup analyses.

| Neutrophil-to-lymphocyte ratio
Patients with NLR greater than the median had larger median tumor size (

| Liver status and NLR
Patients with Child-Pugh A disease had higher median ALC (P < 0.05), whereas patients with Child-Pugh B/C disease had higher comorbidity indices (median 10.0 vs 6.5, P < 0.0001), MELD scores (13.0 vs 7.0, P < 0.0001), and higher rates of cirrhosis (P < 0.05  Figure 3A). In Child-Pugh class B/C (n = 28), NLR did not have a significant effect on survival (low-vs high-NLR: 6.7 vs 2.9 months, P = 0.2) ( Figure 3B).The interaction term Child-Pugh class*log(NLR) was also significant on MVA (P = 0.0124), demonstrating Child-Pugh class to act as an effect modifier for NLR.

| DISCUSSION
Intrahepatic cholangiocarcinoma is the second most common form of primary liver cancer 1 and carries a very poor prognosis. ICC risk factors, including primary sclerosing cholangitis and intrahepatic lithiasis, predispose toward cancer development through injury, immune system activation, and cellular proliferation. This study sought to clarify the significance of three circulatory inflammatory markers and to examine the effect of chronic liver disease on their prognostic value. During the study period, annual cases of newly diagnosed ICC increased from four in 2005 to 18 in 2016, a trend which is reflected on both a national 29,30 and a global scale. 4 It is unclear if this is due to a true increase in incidence or to a possible change in referral patterns or increased rates of early detection. MVA demonstrated that patients treated with IO had improved survival vs those treated with systemic therapy alone. Furthermore, patients treated with resection or IO therapies with and without systemic therapy had greater survival than patients treated with systemic therapy alone. While this survival difference was also seen on subgroup analyses in patients with Child-Pugh A disease, it became a nonsignificant trend among patients with Child-Pugh B disease. This may in part be due to the smaller numbers of Child-Pugh B patients included in this study (n = 26).
We observed strong correlation among the NLR, PLR, and SII; however, only the NLR remained an independent predictor of survival on MVA, where increased NLR was associated with decreased survival. This survival difference was also seen on subgroup analyses in patients who received systemic therapy, and there was a trend toward improved survival with decreased NLR in the surgery and IO treatment groups. Although this trend was nonsignificant, this may have been due to the smaller numbers in the surgery and IO subgroups.
Child-Pugh status was demonstrated to be an effect modifier of the NLR. For patients with none-to-mild liver disease (Child-Pugh A), the association between increased NLR and poorer survival remained significant. For those with moderate liver disease (Child-Pugh B), however, there was no significant survival difference across NLR groups. It should be noted that the Child-Pugh A and Child-Pugh B groups in our cohort had similar distribution of inflammatory markers. Perhaps, in cases of advanced chronic inflammation, the immune system becomes fatigued, and the rates of increased neutrophils may be fewer. Furthermore, in light of the low numbers of Child-Pugh class B patients present in this cohort, there may be smaller effects and nuances that are being missed due to low power.

| Interventional oncology
It is thought that both transarterial chemoembolization (TACE) 10 and Y-90 11 are safe and effective as palliative therapies for unresectable ICC. TACE has been used successfully as adjuvant therapy alongside chemotherapy 12 and surgical resection. 13 In a study comparing resection with TACE, Scheuermann et al 31 found that there was no survival benefit for resection vs TACE in patients with positive lymph nodes or positive resection margins after surgery. Radiofrequency ablation has also been shown to prolong survival in inoperable ICC, particularly in tumors <5 cm. 15,32 In our study, we found that IO was associated with a survival benefit compared to systemic therapy. Collectively, this evidence suggests that IO therapies may be an area of future study for treatment of ICC, particularly in patients with comorbidities that prevent them from being surgical candidates or who are unable to tolerate the side effects of systemic chemotherapy.

| Inflammation and ICC
Intrahepatic cholangiocarcinoma is a very heterogenous tumor, with multiple morphologies on gross, 33 cellular, 34 and molecular 35 levels. Risk factors for ICC activate the inflammatory response, leading to increased cholangiocyte proliferation and the production of free radicals and multiple other cytokines and chemicals which continue the cycle of injury and growth, 17,19 creating an ideal environment for carcinogenesis. 18,36 As available knowledge about circulating inflammatory biomarkers grows, it is important to determine which markers carry the most prognostic value. Of the markers studied here, the NLR was most useful. Similarly, Ha and colleagues 37 studied the NLR, PLR, and SII as well as soluble programmed cell death ligand-1 in 158 patients with advanced biliary tract cancers, and concluded that only NLR and soluble PD-L1 were independent prognostic factors.
Increased NLR ratios have been associated with poor prognosis following surgery 24,27,38 and chemotherapy. 25,27 The NLR has been a component of potential prognostic systems 39 and nomograms for the prediction of resection futility. 40 A recent meta-analysis comprising 26 studies and 4461 patients concluded that while increased NLR indicated a poor prognosis in primary liver cancers, subgroup analyses suggested that the predictive role of NLR in cholangiocarcinomas might be limited. 41 Of note, only 29 of these 4461 patients had confirmed ICC.
The exact mechanism of how the NLR ratio affects survival is unclear. Lin et al 26 examined 102 patients with ICC and saw that higher PD-1+CD4+ and PD-1+CD8+ T cells were found in the high-NLR group while higher amounts of IFN+CD4+ and IFN+CD8+ T cells were seen in the low-NLR group. Furthermore, the high-NLR group experienced an increased density of tumor-infiltrating CD3+T cells. However, in spite of these findings, the strengths of the NLR over PLR and SII 37 suggest that it may be the neutrophil component and not the lymphocytic component which drives the effect of the NLR.
Given the vast heterogeneity of ICCs, further work is required to better understand the differences between patients with an increased NLR and those with a lower NLR on a cellular and molecular level. The decreased effectiveness of the NLR in advanced liver disease seen in this cohort suggests that a state of chronic inflammation not directly subject to the cancer may also affect how the body responds to the tumor. Perhaps in cases of significant or advanced chronic inflammation, the immune system becomes fatigued, and the rates of increased neutrophils may be fewer.

| Limitations
Although our sample size is a limiting factor in subgroup analyses, our cohort was diverse and well-characterized among gender, race, and treatment modalities. We limited the NLR, PLR, and SII values to within 30 days prior to ICC treatment or diagnosis (for supportive therapy patients) in order to improve standardization, as our data relied on archived records. Furthermore, this study lacks information whether patients with cirrhosis and viral hepatitis also suffered from portal hypertension or hypersplenism. However, only platelet counts should be directly affected by these sequelae of liver disease. As this study has been conducted among a United States (Western) population, results may not be globally generalizable.

| CONCLUSION
The NLR ratio has the strongest impact as a prognostic marker in ICC versus the PLR and SII, and increased NLR is associated with decreased survival in ICC. This increased survival is modulated by liver status, suggesting that the interplay between acute tumor-associated inflammation and chronic liver parenchymal dysfunction may further affect survival.