Clinical significance of the globulin‐to‐albumin ratio for prediction of postoperative survival in patients with colorectal cancer

Abstract Aim A previous study has revealed that the albumin/globulin ratio (GAR) before treatment is a predictor of cancer‐specific survival in patients with colorectal cancer (CRC). The aim of the present study was to investigate the clinical significance of GAR for prediction of postoperative survival in patients with CRC. Methods Nine hundred and forty‐one patients who had undergone elective CRC surgery were enrolled. Uni‐ and multivariate analysis models were performed to detect the clinical characteristics that were most closely associated with overall survival (OS). All recommended cutoff values were defined using receiver operating characteristic curve (ROC) analyses. Kaplan–Meier analysis was used to compare the OS curves between the high GAR (GAR > 0.83) and low GAR (GAR ≤ 0.83) groups. Results Multivariate analysis using eight clinical characteristics selected by univariate analyses showed that the GAR was associated with OS (>0.83/≤0.83) (hazard ratio [HR], 1.979; 95% CI, 1.321‐2.966; P = 0.001) along with carcinoembryonic antigen (CEA; >8.7/≤8.7, ng/mL; HR, 2.319; 95% CI, 1.569‐3.428; P < 0.001), carbohydrate antigen 19‐9 (CA19‐9; >18.5/≤18.5, U/mL; HR, 1.727; 95% CI, 1.178‐2.532; P = 0.005), and the neutrophil‐to‐lymphocyte ratio (NLR; >2.9/≤2.9; HR, 2.132; 95% CI, 1.454‐3.126; P < 0.001), and the area under the ROC (AUROC) curve revealed that the GAR had the largest AUROC among these four clinical characteristics (GAR 0.711 > CEA 0.698 > CA19‐9 0.676 > NLR 0.635). A significant difference in OS was observed between patients with low GAR and those with high GAR (P < 0.001). Conclusion Globulin‐to‐albumin ratio is a useful predictor of postoperative survival in patients with CRC.


| INTRODUCTION
In Japan, colorectal cancer (CRC) is the second leading cause of cancer-related death, accounting for approximately 50 000 deaths annually. 1 CRC is also the third most common cancer and the fourth leading cause of cancer-related death worldwide. 2 The term "biomarker" commonly means a measurable serum protein substance whose concentration reflects the presence or degree of progression of a neoplasm. In the clinical field, prognostic biomarkers are often used for treatment decision-making, because they are known to be useful predictors of outcome in cancer patients. 3 In fact, the neutrophil-to-lymphocyte ratio (NLR), 4 carcinoembryonic antigen (CEA), 5 and carbohydrate antigen 19-9 (CA19-9) 5 have been broadly used as prognostic biomarkers for patients with CRC. In particular, we have previously reported that the C-reactive protein (CRP)-to-albumin ratio (CAR) 6  Interleukin-6 (IL-6) is a typical inflammatory cytokine, and its level is known to be increased in patients with cancer. 8 IL-6 not only induces differentiation of B-lymphocytes into plasma cells to produce immunoglobulins, but also decreases albumin production in the liver. 9 Recently, this mechanism has been the focus of attention in patients with CRC, and the albumin/globulin ratio is reported to be useful for prognostication in such patients. 10 We have also reported that the globulin-to-albumin ratio (GAR) is a useful predictor of postoperative survival in patients with hepatocellular carcinoma (HCC) undergoing potentially curative liver resection. 11 Because the GAR is the ratio of albumin relative to globulin, it is less affected by measurement variability and can be determined as easily as the NLR.
However, the first study of GAR in CRC patients did not determine the ideal cutoff value, and therefore, further refinement of this value would be of potential clinical utility. 10 Using a clinical database at a single institution, this study was conducted to determine whether the preoperative GAR would be useful for prognostication in CRC patients. The cutoff values for various clinical characteristics were determined using receiver operating characteristic (ROC) curve analyses.

| ME TH ODS
The recommended cutoff values were based on the most prominent point on the ROC curve for "sensitivity" and "1specificity," respectively. Then, the ideal cutoff values were defined using the Youden index maximum [sensitivity − (1 − specificity)]. 13 The area under the ROC (AUROC) curve was also calculated. Inflammation-based prognostic system consists of not only protein-based system but also blood cell-based system such as NLR and thrombocytosis. Therefore, WBC and platelet are also added to the univariate analysis in the study. Multivariate analysis was performed using the clinical characteristics selected by univariate analysis with a P-value of <0.05.

| Definition of inflammation-based prognostic systems
The GPS was calculated as follows: Patients with both an elevated level of CRP (>1.0 mg/dL) and hypoalbuminemia (albumin <3.5 g/dL) were allocated a score of 2, and patients showing one or neither of these blood chemistry abnormalities were allocated a score of 1 or 0, respectively. 14 HACHIYA ET AL.

| Definition of operative curability
On the basis of the Japanese Classification of Colorectal Carcinoma (Japanese Society for Cancer of the Colon and Rectum, Second English Edition), 15  In addition, most stage II patients undergoing surgery were not considered for postoperative chemotherapy. 24

| Statistical analysis
Data are presented as mean ± SD (standard deviation). Differences among the groups were analyzed using the χ 2 test and Mann-Whitney U-test. Hazard ratio (HR) with 95% CI was calculated by univariate or multivariate analysis using the Cox proportional hazards model. Multivariate analysis was performed using clinical characteristics with P < 0.05 selected in the univariate analysis to assess those most closely related to OS.
Kaplan-Meier analysis and log rank test were used to compare the survival curves for the two groups. Deaths prior to 31 December 2014 were included in this analysis.
Statistical analyses were performed using the IBM SPSS statistics version 22.0 software package for Windows (IBM Co., New York, NY, USA) at a significance level of P < 0.05.

| RESULTS
A total of 941 patients were enrolled (male:female = 581:360). There were 473 patients with low GAR (GAR ≤ 0.83) and 468 with high GAR (GAR > 0.83). Table 1 shows the distribution of the categorical clinical background characteristics of the studied patients in the two GAR groups.
There were significant differences between the two groups in age   Patients with high GAR (>0.83) showed poorer OS than those with low GAR (≤0.83; Figure 2).  It has been reported that persistent chronic inflammation is involved in cancer development, progression, and metastasis. 25 Therefore, such inflammation is associated with hypermetabolism, weight loss, and anorexia in cancer patients. 26 In fact, acute-phase proteins such as CRP, which is an important marker of systemic inflammatory response (SIR), is induced by inflammatory cytokines. 27 Tumor necrosis factor-α (TNF-α), interleukin-10 (IL-10), and IL-6 are also inflammatory cytokines that are well known to be involved in various types of cancer. 25 Among them, IL-6 is produced by a variety of cells including T cells, B cells, fibroblasts, monocytes, endothelial cells, and mesangial cells. 28 IL-6 is associated with the development and progression of CRC. 29 Recent studies have shown that the IL-6 level is significantly increased in the tumor microenvironment and that tumor proliferation is suppressed by anti-IL-6 receptor antibody. 30 In fact, the transmembrane IL-6 receptor is expressed on the cell surface of hepatocytes. 31 When IL-6 acts on hepatocytes, there is an increase in gene transcription rate of acute-phase proteins such as CRP, and the rate and synthesis of albumin mRNA transcription compensates decreases. 32 Similarly, when IL-6 acts on B cells, it induces their differentiation into plasma cells, which produce antibody and upregulate immunoglobulin production. 7 Previous studies have also shown that the preoperative albumin/ globulin ratio is an independent prognostic factor in CRC patients. 10 Serum proteins are divided into two groups: albumin and globulins.

| DISCUSSION
Serum albumin constitutes approximately 60% of the total protein, 7 being produced in the liver and functioning to maintain osmotic pressure and the transport of various metabolites. Albumin also has an antioxidant effect against carcinogens (e.g., nitrosamines and aflatoxins) by stabilizing cell proliferation and deoxyribonucleic acid (DNA) replication. 33 Laursen et al. 34 36 It is known that the globulin level increases with chronic inflammation due to the activity of inflammatory cytokines. 37 Because the serum levels of albumin and globulin are affected by many factors such as stress, liver failure, dehydration, and edema, they also show measurement variability. However, because the GAR is a ratio rather than an absolute value, it is less affected by measurement variability than the serum levels of albumin and globulin separately. In fact, we have reported that even if patients have liver dysfunction, the GAR would be rarely affected by such condition and could predict the postoperative survival of such patients. 11 The present multivariate analysis revealed four prognostic biomarkers: GAR, NLR, CEA, and CA19-9. However, CEA and CA19-9 may show normal levels even in patients with advanced cancer. In fact, it has been reported that approximately 73% of CRC patients who underwent surgery had normal preoperative serum CEA levels 38 and 84% had normal CA19-9 levels. 39  for HCC and 0.83 for CRC. 11 Because it is difficult to define the ideal cutoff value for GAR because of variation of organs, future studies will need to examine the utility of GAR as a prognostic biomarker. The present study showed superiority over the previous one 10 as its sample size was almost 2-fold larger, and the data were examined in terms of cutoff values. In addition, GAR is less expensive to determine than commonly used tumor markers, and can be repeatedly measured, making it not only versatile but also universally applicable.
In conclusion, the GAR can be used as an easy, cheap, objective, and noninvasive biomarker for prognostication of CRC patients undergoing surgery and is useful for prediction of postoperative survival.