Gastric cancer (GC) is the fifth most frequently diagnosed cancer and the third leading cause of cancer death worldwide (1). The prognosis in a considerable number of patients with GC is still poor even after curative surgery. Patients with GC often have insufficient oral intake due to various cancer-related symptoms such as generalized fatigue, pain, nausea, and anorexia (2). Cancer-associated obstruction directly reduces oral intake and induces malnutrition (2), which is recognized as a risk factor of perioperative complications (3-7). Furthermore, immunosuppression is induced by malnutrition (8) and is associated with poor prognosis after curative surgery (9). Therefore, it is important to evaluate nutritional status before surgery.

The Geriatric Nutritional Risk Index (GNRI) was defined by Bouillanne et al. in 2005 (10). The GNRI was a status of nutrition first used to predict the risk of morbidity and mortality in hospitalized elderly patients. It is calculated using the serum albumin and body mass index (BMI). Previous studies reported the association between the GNRI and prognosis for various types of cancer, including esophageal cancer, pancreatic ductal carcinoma, lung cancer, and renal cell carcinoma (11-17). Recently, studies reported a relationship between the GNRI and prognosis of patients with GC (18-20). Syuhei et al. reported that the GNRI was associated with postoperative complications in elderly patients with GC (18). Furthermore, the GNRI was reported as a poor prognostic factor in elderly patients with GC (19), and in elderly patients who underwent laparoscopic gastrectomy (20). However, the patients with gastric cancer that were analyzed in these studies were limited by age or operative procedure.

The present study aimed to investigate the utility of the GNRI to predict the outcome after curative surgery for GC and morbidity of postoperative complications in patients of all ages. Furthermore, we compared the utility of the GNRI between patients aged ≥75 years and those aged <75 years.

Study population. Seven hundred and ninety-five patients who underwent curative surgery for GC between 2008 and 2016 at our Institution were reviewed retrospectively. Tumor staging and pathological findings were classified according to the 15^th Japanese Classification of Gastric Carcinoma issued by the Japanese Gastric Cancer Association (21). Histopathological type was classified into two groups: Differentiated and undifferentiated. We performed surgery according to the Japanese Gastric Cancer Association guidelines for the treatment of gastric cancer (22). After curative surgery, patients were followed-up every 3-6 months for the first 2 years after surgery, and the follow-ups continued for 5 years. This study was approved by the institutional review boards of Kyoto Prefectural University of Medicine (approval number ERB-C-1327). The present study was performed in accordance with the principles of the Declaration of Helsinki. Before surgery, written informed consent was acquired from all patients.

Definition of the GNRI. The GNRI was calculated using BMI and serum albumin level measured before curative surgery as: GNRI=1.489 × albumin (g/l) + 41.7 × BMI/22.

Patients with BMI >22 kg/m² were included as having a BMI of 22 kg/m² according to a previous report (19). Four grades of nutrition-related risk were defined according to the GNRI (10): Major risk: GNRI <82; moderate risk: GNRI 82 to <92; low risk: GNRI 92 to ≤98; and no risk: GNRI >98. Furthermore, patients were classified into two groups: high GNRI: low and no risk; low GNRI: major and moderate risk.

Definition of postoperative complications. We analyzed the complications that occurred after surgery during hospitalization. Complications of grade II or more as assessed by the Clavien–Dindo classification were included (23). Infectious complications included surgical site infection and pneumonia.

Statistical analysis. Categorical variables were compared using the chi-squared test. Overall survival (OS) and cancer-specific survival (CSS) analyses were performed using the Kaplan–Meier method and data were compared using the log-rank test. A multivariate analysis for prognosis was performed using Cox’s proportional hazard model to identify independent prognostic factors for OS, and the hazard ratio (HR) and 95% confidence interval (CI) were subsequently calculated. All statistical tests used in this study were two-sided, and a p-value of less than 0.05 was considered statistically significant. All statistical analyses were performed using JMP 10 software (SAS institute, Cary, NC, USA).

GNRI and clinicopathological factors. Clinicopathological factors of all patients are shown in Table I. Table II shows the association between the GNRI and clinicopathological factors. Low GNRI was correlated to age ≥75 years (p<0.001), tumor size ≥45 mm (p<0.001), pathological T factor 3-4 (p<0.001), the presence of lymph node metastasis (p<0.001), the presence of lymphatic invasion (p<0.001), venous invasion (p=0.012), carcinoembryonic antigen level before surgery ≥5 ng/ml (p=0.004), and pathological stage II-III (p<0.001).

Association between GNRI and prognosis of patients after curative surgery for GC. Figure 1 shows the association between the GNRI and prognosis of patients with GC. The 5-year OS and CSS in patients with stage I-III GC were stratified according to GNRI risk (Figure 1, left panel) and were found to differ signs, p<0.001. The 5-year OS and CSS were significantly worse in the group with low GNRI than those with high GNRI (Figure 1, right panel). The 5-year OS rate of the group with low GNRI was 46% and that with a high GNRI was 84% (p<0.001), whilst the corresponding CSS rates were 67% and 89%, respectively (p<0.001).

The 5-year OS and CSS stratified based on pStage are shown in Figure 2. The low GNRI group had significantly poorer 5-year OS than that with high GNRI among patients with pStage I (p<0.001), pStage II (p=0.015), and pStage III (p=0.003) (Figure 2A). Although there was no significant difference in the 5-year CSS between the two groups in patients with pStage I disease (p=0.579), the rate in the group with low GNRI tended to be worse in patients with pStage II (p=0.063) and was significantly worse in those with pStage III disease (p=0.003) (Figure 2B).

Investigation of prognostic factors for 5-year OS. Univariate and multivariate analysis of prognostic factors for 5-year OS are shown in Table III. Age ≥75 years (p<0.001), tumor size ≥45 mm (p<0.001), advanced pathological T- and N-stage (p<0.001), lymphatic invasion and venous invasion (p<0.001), and low GNRI (p<0.001) were significantly associated with poor prognosis. Multivariate analysis revealed that age ≥75 years (HR=1.20, 95% CI=1.02-1.41, p=0.027), lymph node metastasis (HR=1.22, 95% CI=1.01-1.46, p=0.034), and low GNRI (HR=2.34, 95% CI=1.47-3.64, p<0.001) were independent prognostic factors for poorer survival.

Association between GNRI and prognosis stratified by age. Figure 3 shows OS and CSS curves of the GNRI groups stratified based on age (cutoff=75 years). The low GNRI group had poorer prognosis the high GNRI group for those under 75 years and 75 years or older in OS (Figure 3A, both p<0.001) and in CSS (Figure 3B, p<0.001 and p=0.005, respectively). A low GNRI was associated with poorer prognosis regardless of age.

Association between GNRI and postoperative complications. Table IV shows the association between GNRI and the frequency of postoperative complications. There was a significant inverse association between GNRI and the occurrence of complications overall (p=0.002). Table V shows the associations between GNRI and the frequency of postoperative complications stratified based on age. A low GNRI was significantly associated with increased postoperative complications in patients aged ≥75 years (p=0.042). However, in patients aged <75 years, low GNRI was not significantly associated with postoperative complications (p=0.063).

In patients with GC, malnutrition before surgery significantly contributes to postoperative mortality, morbidity, and length of total hospital stay (24,25). Previous reports mentioned that body weight loss (26), sarcopenia (27), and malnutrition before surgery (28,29) were associated with poorer prognosis after curative surgery for GC. Therefore, it is important to evaluate nutritional status before surgery. Various nutritional indices have been developed to predict the outcome of cancer, including the prognostic nutritional index (30), the Glasgow prognostic score (31), and the Controlling Nutritional Status (32). These indices were constructed using only laboratory data. The GNRI is a simple nutritional index which can be calculated using the serum albumin level and BMI (10). As the GNRI includes a physical measurement, BMI, it may reflect nutritional status better than other indices. Recent reports revealed that the GNRI was a risk factor of morbidity and mortality in elderly patients with GC (18-20). Although the GNRI was first used in elderly patients (10), it may be effective in patients of all ages. To the best of our knowledge, there are no reports that analyzed the utility of the GNRI in patients with GC of all ages. The present study analyzed the predictive ability of the GNRI for prognosis and the occurrence of complications in patients of all ages who underwent curative surgery for GC. Furthermore, we stratified patients based on age and compared the utility of the GNRI between patients aged ≥75 years and those aged <75 years.

Malnutrition is associated with various risk factors in patients with GC. Immunosuppression is induced by malnutrition (8) and is recognized as a risk factor of poor prognosis in GC after curative surgery (9). Malnutrition before surgery has also been reported to be a risk factor of postoperative complications (33,34). We previously reported that postoperative complications were correlated with poor prognosis in patients with GC (35). Postoperative complications induce systemic inflammation that is related to immunosuppression (36). In addition, Lerut et al. reported that systemic inflammation activated micro-residual tumor and led to postoperative recurrence (37). Thus, preoperative malnutrition leads to poor prognosis after curative surgery.

We demonstrated that prognosis after surgery for patients with GC was clearly stratified based on the GNRI. The group with low GNRI had a poorer prognosis than did the high GNRI group (Figure 1). We revealed that the GNRI was an independent prognostic factor in patients with GC by multivariate analysis. Low GNRI was also poor prognostic factor in patients when stratified by age (Figure 3). Patients were divided into four risk groups based on the GNRI: Major, moderate, low, and no risk. In patients aged ≥75 years, a no-risk GNRI tended to be associated with a better prognosis than low-risk GNRI (Figure 4A, p=0.082), whereas there was no difference in prognosis between no-risk GNRI and low-risk GNRI in patients aged <75 years (Figure 4B, p=0.700). Elderly patients have low reserve capacity of nutrition, so that even a slight decrease of nutritional status may affect the outcome after curative surgery. A low GNRI was significantly associated with an increased frequency of overall complications. Furthermore, the GNRI was more effective for predicting the possibility of postoperative complications in patients aged ≥75 years than in younger patients (Table V). Hence, the results suggest that the GNRI is a useful nutritional index for evaluating short- and long-term outcomes in patients with GC of all ages after curative surgery, and may be more useful for elderly patients.

The GNRI was associated with prognosis for each stage of GC (Figure 2). In patients with pStage I GC, the GNRI was not a prognosis factor for 5-year CSS (p=0.579), and was associated with poor 5-year OS (p<0.001). Hence, the GNRI may be a suitable index to predict mortality from other diseases in patients with pStage I GC. Previous studies reported that a lower GNRI was a poor prognostic factor in patients with pneumonia (38), heart failure (39), and chronic kidney disease (40). Various conditions, including immunosuppression, cardiac dysfunction, and protein energy wasting were induced by malnutrition and lead to poor prognosis (38-40). Patients with early-stage cancer may be more strongly affected by these factors than by cancer progression. These findings suggest that the GNRI was a prognostic factor even in patients with early-stage cancer. Therefore, patients with a low GNRI need strict follow-up not only for oncology but also for other systemic diseases, especially those with pStage I GC.

Our study had several limitations that need to be considered. This was a retrospective study performed at a single institution; therefore, a large-scale, multicenter study may be necessary for further analysis. In addition, an interventional study with preoperative nutrition treatment is needed to confirm this result.

In conclusion, we investigated the utility of the GNRI in predicting the outcome after curative surgery for GC and occurrence of postoperative complications. The GNRI was identified as an independent prognostic factor of 5-year OS and CSS. A low GNRI was a risk factor of postoperative complications, especially in elderly patients with GC. Thus, the GNRI may be an effective nutritional index for evaluating short- and long-term outcomes in patients of all ages after curative surgery for GC.

The Authors declare that they have no conflicts of interest to disclose.

This study was designed by H.F., D.M., T.K., and E.O.; H.F. and D.M. performed statistical analyses. The clinical information and materials were collected and kept by H.F., D.M., T.K., J.K., H.K., T.O., H.S., T.A., Y.Y., H.K., R.M., A.S., Y.K., H.I., H.F., and K.O.; H.F., D.M., and T.K. drafted the manuscript. T.K. edited and revised the manuscript. E.O. approved the final version of the manuscript.