Introduction

Hepatocellular carcinoma (HCC) is a devastating malignancy that is the sixth most common cancer and the fourth leading cause of cancer death worldwide [1]. Although surgical resection is the treatment of choice in HCC, it is not curative for most patients because of the high recurrence rate [2]. Prediction of recurrence is important, and prevention with appropriate therapy will improve patient outcome [3].

The prognostic significance of preoperative neutrophil-lymphocyte ratio (NLR) has been suggested as an easily approachable indicator of systemic inflammation and has been highlighted as a biomarker for predicting prognosis in a variety of cancers, including HCC [4,5,6,7,8,9,10]. However, the cutoff value of NLR is not consistent, and the underlying mechanism of the prognostic effect of preoperative NLR is not yet fully understood.

In this study, we evaluated the prognostic effect of preoperative NLR in 234 patients with HCC who underwent curative resection and had long-term follow-up. We also studied the association of preoperative NLR with histologic features such as tumor necrosis or tumor-infiltrating lymphocytes (TILs).

Materials and methods

Patient population and clinical information

Initially, a total of 291 patients who underwent curative resection for primary HCCs at Samsung Medical Center, Seoul, Korea from July 2000 to May 2006 were enrolled in this study. Among them, 57 patients were excluded as follows: 8 patients with preoperative treatment such as transarterial chemoembolization, radiofrequency ablation, or radiotherapy, 16 patients with evidence of systemic inflammation (WBC > 10,580/μL or segmented neutrophil > 73.5%) and 33 patients with no record of complete blood count (CBC) from 14 days before operating day. Finally, 234 patients were included in this study. The Institutional Review Board of Samsung Medical Center approved this study and waived informed consent.

Curative resection was defined as complete resection of all tumors with negative microscopic resection margins and no visible residual tumors on a computed tomography scan 1 month after surgery. All patients were classified as Child-Pugh class A. Medical records were reviewed for clinical parameters such as age, gender, alcohol history, and results of laboratory tests, including CBC, serology for hepatitis virus A, B, C, and D, serum alpha fetoprotein (AFP), and serum albumin. American Joint Committee on Cancer (AJCC) staging system and Barcelona Clinic Liver Cancer (BCLC) staging classification were used for staging [11, 12].

Patients were followed every 3 months after surgery. To detect tumor recurrence, three phase dynamic computed tomography scans or magnetic resonance imaging was performed. The median follow-up period was 120 months (range 14–151 months) for survivors. Recurrence-free survival (RFS) was defined as the interval from the date of surgery to detection of tumor recurrence. Disease-specific survival (DSS) was defined as the interval between date of surgery and date of HCC-related death, which was defined as follows: (1) tumor occupying more than 80% of the liver, (2) portal venous tumor thrombus proximal to second bifurcation, (3) obstructive jaundice due to tumor, (4) distant metastases, and (5) variceal hemorrhage with portal venous tumor thrombus proximal to first bifurcation [13].

Neutrophil-lymphocyte ratio

Preoperative (within 14 days from surgery) and postoperative (on first follow-up day after about 1 month from surgery) NLR was calculated based on CBC values. To identify the optimal cutoff value with highest statistical significance related to patient survival, X-Tile statistics package (Yale University, New Haven, CT, USA) [14] was used. Cases were dichotomized into low and high NLR groups based on the established cutoff value.

Histopathologic evaluation: tumor necrosis and tumor-infiltrating lymphocytes

Paraffin-embedded tissues were sectioned and stained with hematoxylin and eosin. When the tumor was less than 3 cm, the entire tumor was sectioned and embedded. When the tumor was larger than 3 cm, at least four sections were obtained and the mean number of blocks was 1 per cm of tumor diameter. Two liver pathologists (SYH and CKP) reviewed hematoxylin- and eosin-stained slides and confirmed the diagnosis of HCC as well as histopathologic features of HCCs, such as differentiation, microvascular invasion, major portal vein invasion, intrahepatic metastasis, multicentric occurrence, and non-tumor liver pathology. Tumor differentiation was defined according to the criteria of Edmondson and Steiner [15]. Intrahepatic metastasis and multicentric occurrence were determined according to criteria of the Liver Cancer Study Group of Japan [16].

The presence of tumor necrosis and degree of lymphocytic infiltration in tumors were also evaluated. Tumor necrosis was defined as presence of microscopic coagulative necrosis, typified by homogeneous clusters of degenerating and dead tumor cells [17]. The percent of tumor necrosis area was recorded. The number of TILs was counted in 10 high power fields (HPFs) of hematoxylin- and eosin-stained slides. Areas with the most TILs in low power field were identified, and counting was initiated from that area. Specimens were classified into three groups according to the intensity of TILs following a modified method of Wada et al., as follows [18]: −, negative or almost no infiltration; +, mild to moderate infiltration (< 500 cells/10 HPFs); and ++, marked infiltration (> 500 cells/10 HPFs).

Statistical analysis

The association between NLR and clinicopathologic parameters was analyzed using the Chi-square test, Fisher’s exact test, or Cochran Armitage test as appropriate. Mann-Whitney or Kruskal-Wallis test was used for comparison of the mean value of NLR, absolute neutrophil, or lymphocyte count. Kaplan-Meier method was used for survival analysis, and differences in survival rates were assessed by the log rank or Breslow test. To evaluate associations between clinicopathologic parameters and survival, the Cox proportional hazard regression model was used. Significant prognostic factors by univariate analysis were entered into multivariate analysis. Proportional hazard assumption was examined graphically. All statistical analyses were performed with SPSS software (SPSS Inc., Chicago, IL, USA) or R software (version 3.03). P values less than 0.05 were considered statistically significant.

Results

Preoperative NLR in patients with HCC and its relationship with clinicopathologic characteristics

The mean preoperative NLR was 1.67 (95% confidence interval, 1.57–1.76), and the median value was 1.50 (range 0.55–4.41). The optimal cutoff value with the highest statistical significance for patient survival, determined by X-Tile package, was 2.50. Preoperative NLR was graded as low (< 2.5) or high (≥ 2.5), and high NLR was observed in 28 (12.0%) of 234 HCCs.

The associations between preoperative NLR and clinicopathologic characteristics are summarized in Table 1. High preoperative NLR was significantly associated with younger age (p = 0.027), larger tumor size (p = 0.049), high Edmonson grade (p = 0.026), microvascular invasion (p = 0.019), major portal invasion (p = 0.005), advanced AJCC T (p = 0.024) or BCLC stage (p = 0.004), and low albumin level (p < 0.001). Similar results are found in comparison between absolute value of preoperative NLR, neutrophil or lymphocyte count, and clinicopathologic variables. Detailed information is provided in Supplementary Table 1.

Table 1 Association between neutrophil to lymphocyte (NLR) groups and clinicopathologic variables
Full size table

Preoperative NLR as an independent prognostic factor for DSS

Patients with high preoperative NLR showed shorter DSS (p = 0.002) and a tendency for shorter RFS (p = 0.096) (Fig. 1). On multivariate analysis including co-variables (age, tumor size, Edmondson grade, microvascular invasion, major portal vein invasion, intrahepatic metastasis, serum albumin level, and serum AFP level) with statistical significance in univariate analysis, preoperative NLR was an independent prognostic factor for DSS (hazard ratio: 2.050 (95% confidence interval 1.139–3.691, p = 0.017) (Table 2), not for RFS.

Fig. 1
figure 1

Kaplan-Meir survival curves according to preoperative neutrophil to lymphocyte ratio. a Disease-specific survival. b Recurrence-free survival

Full size image
Table 2 Multivariate analysis for recurrence-free survival and disease-specific survival
Full size table

Preoperative NLR status was associated with tumor necrosis and tumor-infiltrating lymphocytes

Tumor necrosis was identified in 62 (26.5%) of 234 cases. TILs were observed in 112 (47.9%) cases, including 83 (35.5%) with mild to moderate infiltration and 29 (12.4%) with marked infiltration. Tumor necrosis was more frequently identified in the high NLR group than in the low group (57.1% vs 22.3%, p < 0.001), and there was a weak correlation between percent of tumor necrosis and NLR (r = 0.169, p = 0.010). However, TILs were less frequently found in the high NLR group (p = 0.009).

The mean value of preoperative NLR was higher in patients with tumor necrosis than in patients without tumor necrosis (Fig. 2a). This effect was mostly induced by differences in preoperative absolute neutrophil count between two groups according to tumor necrosis, not differences in absolute lymphocyte count (Fig. 2b, c). In contrast, the mean value of preoperative NLR was higher in patients without TILs than in patients with TILs (Fig. 2d). This effect was mostly induced by differences in preoperative absolute lymphocyte count between patient groups according to TILs, not differences in absolute lymphocyte count (Fig. 2e, f).

Fig. 2
figure 2

Association between preoperative neutrophil to lymphocyte ratio (NLR) and tumor necrosis or tumor-infiltrating lymphocytes. a–c Mean value of preoperative NLR (a), absolute neutrophil count (b), and absolute lymphocyte count (c) in two groups according to presence of tumor necrosis. e, f Mean value of preoperative NLR (d), absolute neutrophil count (e), and absolute lymphocyte count (f) in two groups according to presence of tumor-infiltrating lymphocytes

Full size image

After curative resection, NLR in patients with tumor necrosis or without TILs was decreased, and postoperative NLR was not different in different patient groups according to tumor necrosis or TILs (Fig. 3a–d).

Fig. 3
figure 3

Change of preoperative and postoperative neutrophil to lymphocyte ratio (NLR) according to presence of tumor necrosis (a, b) or tumor-infiltrating lymphocytes (c, d)

Full size image

Effects of tumor necrosis or TILs on patient survival

Patients with tumor necrosis showed shorter DSS (p < 0.001) and RFS (p < 0.001) (Fig. 4a, b). Patients with no TILs showed shorter DSS (p < 0.001) and a tendency for shorter RFS (p = 0.100) (Fig. 4c, d).

Fig. 4
figure 4

Kaplan-Meier survival curves according to tumor necrosis or tumor-infiltrating lymphocytes. a Disease-specific survival for tumor necrosis, b recurrence-free survival for tumor necrosis, c disease-specific survival for tumor-infiltrating lymphocytes, d recurrence-free survival for tumor-infiltrating lymphocytes

Full size image

On multivariate analysis including tumor necrosis and TILs as additional co-variables, preoperative NLR was not an independent prognostic factor for DSS, while tumor necrosis and TILs remained strong independent predictors for DSS (Table 3).

Table 3 Multivariate analysis for recurrence-free survival and disease-specific survival, including tumor necrosis and TILs as additional co-variables
Full size table

Discussion

In this study, we demonstrated the prognostic significance of preoperative NLR in patients undergoing curative resection for primary HCC and long-term follow-up. High preoperative NLR was significantly associated with aggressive tumor phenotype, such as large size, high Edmonson grade, microvascular invasion, major portal invasion, and advanced AJCC T or BCLC stage. High NLR was an independent adverse prognostic predictor for DSS. However, the independent prognostic effect of NLR for DSS disappeared when tumor necrosis and TILs, which were highly associated with NLR status, were added as co-variables.

The prognostic significance of preoperative NLR has been reported in many studies of a variety of cancers. In HCCs, most previously reported studies have highlighted preoperative NLR as a prognostic factor in the setting of curative hepatectomy, although the NLR cutoff value was not consistent and ranged from 2.31 to 5 (summarized in Table 4) [4,5,6,7,8,9,10]. While the underlying mechanism of the prognostic effect of preoperative NLR is not fully understood, some hypothetical explanations have been suggested. High NLR implies relative lymphocytopenia and neutrophilic leukocytosis. The former indicates a decrease in lymphocyte-mediated anti-cancer effects [19], and the latter indicates higher secretion of pro-angiogenic factors for tumor growth [20]. In a study of the largest cohort (n = 958) to date, Mano et al. showed that preoperative NLR of more than 2.81 was an independent predictor of survival after hepatectomy in patients with HCC [7]. They also demonstrated that the number of CD163-postive tumor-associated macrophages (TAM) was significantly higher in the group with high NLR than in the group with low NLR. TAMs express some cytokines, such as IL-4 and IL18, which may promote systemic neutrophilia. Montomura et al. suggested IL-17, a proinflammatory cytokine promoting neutrophil recruitment by CXC cytokines, as a key molecule involved in the relationship between NLR and HCC recurrence [21]. They showed that the number of peritumoral IL-17-producing cells and blood IL-17 levels were higher in high NLR groups than in low NLR groups.

Table 4 Summary of previous studies evaluating the prognostic role of neutrophil-lymphocyte ratio in patients with curative resection for hepatocellular carcinoma
Full size table

In this study, we hypothesized that tumor necrosis and TILs are associated with NLR based on the following facts: (1) tumor necrosis could be a stimulus for acute inflammation, which induces neutrophil recruitment, and (2) high infiltration of TILs, which are generally associated with better prognosis in a variety of cancers including HCCs, could be associated with lymphocytosis in peripheral blood. Therefore, we evaluated the presence of tumor necrosis or TILs in HCC tissue and the association with NLR status. As a result, patients with tumor necrosis showed higher preoperative NLR than patients without tumor necrosis, and this effect was mostly induced by differences in preoperative absolute neutrophil count, not absolute lymphocyte count. In contrast, the mean value of preoperative NLR was higher in patients without TILs than in patients with TILs, and this effect was mostly induced by differences in preoperative absolute lymphocyte count, not absolute lymphocyte count. These results suggest that tumor factors may influence systemic inflammation represented by NLR status, and the prognostic effect of NLR may originate from proinflammatory conditions such as tumor necrosis or absence of TILs. In a study by Han et al., high pre-treatment NLR was significantly associated with high neutrophil infiltration and low CD3(+) T cell infiltration into tumors, consistent with the results of our study [22]. After curative resection, NLR in patients with tumor necrosis or without TILs was decreased more, and postoperative NLR was not different in patient groups according to tumor necrosis or TILs. We also showed that the prognostic effect of NLR was induced by the prognostic effect of tumor necrosis or TILs. These findings suggest that high NLR may be caused by a tumor factor, represented by the presence of tumor necrosis or low TILs.

HCC can be diagnosed by radiologic examination without histologic confirmation, and there are several treatment modalities for HCC. Especially, there has been controversy in the most effective first-line treatment modality for the patients with small solitary HCC: radiofrequency ablation, surgical hepatic resection, and liver transplantation [23]. Treatment modality is usually determined by tumor stage or the underlying liver parenchymal disease with function [23]. However, tumor biology predicting tumor aggressiveness, which can be represented by histologic features, is hardly to be evaluated before surgical resection. Based on the results of this study, preoperative NLR could represent tumor biology by predicting the presence of tumor necrosis or TIL and might be used as an ancillary marker for determining the treatment modality.

Conclusion

High NLR is an independent prognostic factor in patients with HCC undergoing curative resection. The prognostic effect of high NLR might originate from the prognostic effect of tumor necrosis or TILs.