Preoperative globulin‐to‐albumin ratio predicts outcome after curative resection in patients with gastric cancer

Abstract Aim The globulin‐to‐albumin ratio (GAR) is useful for prognostication of patients with various cancers. However, the significance of GAR in gastric cancer (GC) remains unclear. Our purpose was to investigate the relationship between the GAR and outcome after curative resection in GC patients. Methods Three‐hundred and seventy‐six patients who had undergone curative resection for GC were retrospectively reviewed. Univariate and multivariate analyses using the Cox proportional hazard model were performed to detect clinical characteristics that correlated with overall survival (OS), and their cut‐off values were identified using receiver operating characteristic (ROC) curve analyses. Kaplan–Meier analysis and log‐rank test were used for comparison of OS and relapse‐free survival (RFS). Results Multivariate analysis using 17 clinical characteristics selected by univariate analyses revealed that GAR (>0.80/≤0.80) was significantly associated with OS (hazard ratio [HR], 2.305; 95% CI, 1.122‐4.735; P = 0.023), as well as lymph node metastasis (presence/absence) (HR, 2.417; 95% CI, 1.077‐5.426; P = 0.032), neutrophil‐to‐lymphocyte ratio (>2.7/≤2.7) (HR, 2.368; 95% CI, 1.138‐4.930; P = 0.002), and serosal invasion (presence/absence) (HR, 3.443; 95% CI, 1.048‐11.31; P = 0.042). Kaplan–Meier analysis and log‐rank test demonstrated that the OS and RFS of patients with a high GAR (>0.80) were significantly poorer than those with low GAR (≤0.80). Conclusions The GAR is a useful predictor of postoperative outcome among GC patients undergoing curative resection.


| INTRODUCTION
Several studies have shown that the combination of globulin and albumin can predict the outcome of patients with various cancers such as nasopharyngeal cancer, non-small-cell lung cancer, and upper tract urothelial carcinoma. [1][2][3] The globulin-to-albumin ratio (GAR) is the ratio of the serum globulin level and the serum albumin level, both of which are routinely measured at preoperative term. Like the Glasgow Prognostic Score (GPS) and Neutrophil-to-lymphocyte ratio (NLR), the GAR reflects the systemic inflammatory response as interleukin-6 (IL-6) induces differentiation of B-lymphocytes into plasma cells, which produce immunoglobulins and reduce the level of serum albumin. 4,5 Therefore, the GAR should predict the outcome after curative resection for gastric cancer (GC), as has been shown for other cancers. [1][2][3] In fact, a recent study demonstrated that the combination of globulin and albumin was able to predict outcome in GC patients undergoing curative surgery. 6 Although the previous study mentioned that the combination of globulin and albumin stratified disease-free survival (DFS) of GC patients with lymph node invasion or serosal invasion, and this combination was a good indication for postoperative adjuvant chemotherapy in such GC patients, 6 the relationship between the GAR and tumor-node-metastasis (TNM) stage was unclear in that study. Therefore, in order to prove the significance of GAR in prognostication of GC patients, the relationship between the GAR and each TNM stage of GC patients should be investigated.
Furthermore, it was unclear whether there was an association between GAR and relapse-free survival (RFS), as only overall surgery (OS) and DFS. 6 However, OS and DFS after surgery are unable to indicate whether the GAR is significantly associated with tumor progression including recurrence after surgery, because the GAR would also reflect the nutritional status of GC patients. In order to resolve these two issues, it is necessary to investigate whether the GAR is associated with not only TNM stage but also RFS after surgery.
In the present study, therefore, we investigated not only the relationship between GAR and TNM stage but also the relationship between GAR and both OS and RFS after curative surgery in GC patients using a single-institution clinical database.

| MATERIALS AND METHODS
We retrospectively reviewed 376 patients with newly diagnosed GC who had undergone curative resection at the Second Department of Surgery, Dokkyo Medical University Hospital, between January 2003 and December 2015. We excluded patients with unresectable GC or patients who had undergone combined resection of other organs.
Because the occurrence of anastomotic leakage has been reportedly associated with outcome after total gastrectomy for GC, 7 we additionally excluded patients with postoperative anastomotic leakage.
None of the patients exhibited clinical evidence of infection or other inflammatory conditions, and none had received preoperative chemotherapy or irradiation. All procedures were performed by a single well-trained surgical team.
Routine laboratory measurements including the serum level of albumin and globulin were carried out on the day of admission. We used these preoperative data to determine the GAR in this study.

| Definition of inflammation-based prognostic systems
We calculated each patient's GPS as follows: Patient with elevated level of CRP (>1.0 mg/dL) and hypoalbuminemia (<3.5 g/dL) was assigned a score of 2, patient who exhibited one of these abnormalities was assigned a score of 1, and patient with no abnormalities was assigned a score of 0. 8 The NLR was calculated as the patient's neutrophil level (% or number of neutrophils) divided by the lymphocyte level (% or number of lymphocytes). 9 The GAR was calculated as: [the serum total protein level (g/dL)-the serum albumin level (g/ dL)] divided by the serum albumin level (g/dL). The serum total protein level included both the serum globulin level and the serum albumin level.

| Definition of operative curability
On the basis of the General Rules for Japanese classification of gastric carcinoma (Japanese Gastric Cancer Association, 3rd English Edition), residual tumors are diagnosed as: R0, no residual tumor; R1, microscopic residual tumor (positive resection margin or cancer cells evident on peritoneal cytology as CY1); R2, macroscopic residual tumor. 10 On the basis of this definition, curative resection is defined as R0.

| Definition of TNM stage
We used the TNM classification of Malignant Tumours Eighth Edition edited by the Union for International Cancer Control (UICC) for determining the TNM stage. 12

| Definition of tumor location
On the basis of the General Rules for Japanese classification of gastric carcinoma, the stomach is divided anatomically into three portions-upper (U), middle (M), and lower (L)-by lines connecting points on the lesser and greater curvatures. 10 If more than one part is involved, all involved portions should be recorded in descending order of degree of involvement, the part containing the bulk of the tumor being listed first, for example, UM, UML, or ML. 10 Tumor extension into the esophagus or duodenum is recorded as "others" in each case.

| Postoperative adjuvant chemotherapy for advanced gastric cancer
On the basis of the Japanese gastric cancer treatment guidelines 2014, patients with stage II and III GC, except for pT1 and pT3N0, are eligible for postoperative adjuvant chemotherapy for prevention of recurrence after surgery. 13,14 Oral administration of tegafur-gimestat-otastat potassium (S-1) is standard practice for postoperative adjuvant chemotherapy. 13,14 On the basis of the Japanese gastric cancer treatment guidelines 2014, administration of S-1 after curative resection has been performed at our institution. 13,14

| Statistical analysis
Data are presented as medians and interquartile ranges. Intergroup differences were analyzed using the chi-squared test or the Mann-Whitney U-test, as appropriate. Hazard ratio (HR) with 95% confidence intervals (CIs) was calculated by univariate and multivariate analyses using the Cox proportional hazards model. To identify clinical characteristics that were closely related to OS, multivariate analysis was performed using clinical characteristics shown to have a Pvalue of <0.05 in the univariate analysis. Kaplan-Meier analysis and log-rank test were used to compare the OS and RFS of the groups.
All statistical analyses were performed using the SPSS software package (version 23.0; IBM Co., New York, NY, USA), and differences with a P-value of <0.05 were considered statistically significant.
The cut-off values for the various clinical characteristics were determined using receiver operating characteristic (ROC) curve analyses. The recommended cut-off values for the characteristics were defined using the most prominent point on the ROC curve (Youden index = maximum [sensitivity-(1-specificity)]), 15 and we also calculated the area under the ROC (AUROC) curve. The optimal cut-off value for GAR was the 0.80, which provided a sensitivity of 64.8%, a specificity of 62.6%, and an AUROC curve of 0.637 ( Figure 1). All other cut-off values were defined using ROC curve analyses, including as age (year), body mass index (BMI) (kg/m 2 ), maximum tumor size (cm), NLR, platelet count (×10 4 /mm 3 ), the serum levels of carbohydrate antigen 19-9 (CA19-9) (U/mL), carcinoembryonic antigen (CEA) (ng/mL) and globulin (g/dL), PNI and white blood cell (WBC) count (×10 3 /mm 3 ), except for the serum levels of albumin (g/dL) and CRP (mg/dL). The AUROC curve for each characteristic was also calculated.

| DISCUSSION
Although our study included a small population of patients with an advanced TNM stage, the GAR was significantly associated with TNM stage. In addition, tumor markers and venous invasion were significantly associated with a high GAR. Previous studies have indicated that these two characteristics are associated increased angiogenesis, lymph node metastasis, and distant metastasis in GC patients. [16][17][18][19][20] In fact, OS in the high GAR group was significantly poorer than that in the low GAR group. These facts adequately showed that the GAR can reflect not only tumor progression but also the malignant potential of GC.
Multivariate analyses revealed that the GAR was closely associated with the postoperative outcome of GC patients as well as age and TNM stage. However, the mechanism underlying the relationship between a high GAR and poor outcome of patients with a cancer is poorly understood.
of the antitumor immune system such as cytotoxic T lymphocytes and natural killer cells in cancer patients. 23 In addition, these cytokines induce the production of various acute-phase proteins and these proteins which are contained in the serum globulin fraction. [24][25][26][27][28] Interestingly, it has been reported that these cytokines and acute-phase proteins are associated with the prognosis of cancer patients. 29 Interestingly, GAR significantly stratified OS and RFS of stage I GC patients (Figures 4 and 5). The recurrence rate in patients with early stage GC was reported to be 1.4%-3.0%. 43,44 Previous study reported that submucosal layer invasion, lymph node metastasis, and pathological differentiation (tub1 and tub2) were the risk factors for tumor recurrence after surgery for early stage GC. 43 On the other hand, another previous study revealed that high age, male, lymphovascular invasion, stage IB (proper muscle invasion or lymph node metastasis), perineural invasion, and elevated tumor marker were independent poor prognostic factors for RFS. 44 Although we did not investigate the relationship between GAR and these clinical characteristics of stage I GC which was associated with tumor recurrence, our result indicated that preoperative high GAR (>0.80) was useful for predicting poor prognosis of RFS in stage I GC patients in comparison with low GAR (<0.80). Therefore, as our new significant The bold of P-value means statistical significant value. 95% C.I., 95% confidence interval; HR, hazard ratio; BMI, body mass index; CA19-9, carbohydrate antigen 19-9; CD, Clavian-Dindo; CEA, carcinoembryonic antigen; CRP, C-reactive protein; GAR, globulin-to-albumin ratio; GPS, Glasgow prognostic score; NLR, neutrophil-to-lymphocyte ratio; TNM, Tumor-node-metastasis; WBC, white blood cell. The limitations of our study should be acknowledged. This was a retrospective study conducted at a single institution. The 11 patients among 376 patients undergoing surgery for GC had liver cirrhosis.
Because GC patients with liver cirrhosis were small in number (11/ 376, 2.9%), it was considered that liver cirrhosis would affect less influence on our results. Although cytokines reflect cancer-related inflammation, cytokine-related characteristics were not measured in the present study. However, targeting of GC patients who had undergone curative surgery would have partly reduced the degree of F I G U R E 3 Relationship between the two globulin-to-albumin ratio (GAR) groups (GAR ≤ 0.80 and GAR > 0.80 from top to bottom) and relapse-free survival in patients undergoing curative resection for gastric cancer F I G U R E 4 Relationship between the two globulin-to-albumin ratio (GAR) groups (GAR ≤ 0.80 and GAR > 0.80 from top to bottom) and overall survival in patients undergoing curative resection for each stage of gastric cancer (I/II/III) bias attributable to clinical characteristics and general status. In order to resolve these problems, further studies, both prospective and using propensity score matching, will be required.
In summary, the present study has shown that the preoperative GAR can predict the outcome after curative resection for patients with GC. On the basis of these results, the GAR can be considered an indicator of appropriate adjuvant treatments for GC patients who undergo curative resection.