Abstract
Aim: To investigate the influence of clinical characteristics including nutritional markers on postoperative survival in patients undergoing total gastrectomy (TG) for gastric cancer (GC). Patients and Methods: One hundred fifty-four patients were enrolled. Uni- and multivariate analyses using the Cox proportional hazard model were performed to explore the most valuable clinical characteristic that was associated with postoperative survival. Results: Multivariate analysis using twelve clinical characteristics selected from univariate analyses revealed that age (≤72/>72), carcinoembryonic antigen (≤20/>20) (ng/ml), white blood cell count (≤9.5/>9.5) (×103/mm3), prognostic nutritional index (PNI) (≤45/>45) and lymph node metastasis (negative/positive) were associated with postoperative survival. Kaplan-Meier analysis and log-rank test showed that patients with higher PNI (>45) had a higher postoperative survival rate than those with lower PNI (≤45) (p<0.001). Conclusion: PNI is associated with postoperative survival of patients undergoing TG for GC and is able to divide such patients into two independent groups before surgery.
Abbreviations: AUROC curve; Area under the receiver operating characteristic curve; BMI; Body-mass index; CA19-9; Carbohydrate antigen 19-9; CEA; Carcinoembryonic antigen; CRP; C-reactive protein; GC; Gastric cancer; GIC; Gastrointestinal cancer; IL-6; Interleukin-6; NLR; Neutrophil to lymphocyte ratio; NRI; Nutritional risk index; PNI; Prognostic nutritional index; ROC curve; Receiver operating characteristic curve; SIR; Systemic inflammatory response; TG; Total gastrectomy.
Although obesity has been considered a risk factor for not only poor outcomes after a variety of surgical procedures but also carcinogenesis of several types of cancer (1), a number of studies of critically- and chronically-ill patients have suggested that overweight and obese patients may paradoxically have better outcomes than “normal”-weight patients (2). This so-called “obesity paradox” is well-known for several diseases including gastrointestinal cancer (GIC). In fact, recent collaborative analysis of data from almost 900,000 adults in 57 prospective studies has disclosed that a moderate body-mass index (BMI) (22.5-25 kg/m2) is associated with lowest mortality. However, in that study, most of the subjects analyzed were from Europe, Israel, USA or Australia, and only 8% were Japanese (3).
On the other hand, it is well-known that gastric cancer (GC) patients undergoing total gastrectomy (TG) cannot easily increase their BMI to the preoperative level because they lost the capacity of reservoir for food and drink. In particular, recent studies revealed that stomach is a main organ which produces the appetite-related peptide hormone “ghrelin” (4). Because ghrelin stimulates hypothalamus and increases appetite, it is difficult for patients who underwent TG to increase their appetite fully in terms of the lack of ghrelin production. Therefore, it is well acceptable that preoperative BMI would affect postoperative BMI or the nutritional status of such patients.
In addition, recent studies have revealed that among several nutritional markers, the prognostic nutritional index (PNI) is not only a simple index of nutritional condition, calculated using the serum albumin level and total lymphocyte count, but also a useful prognostic indicator in patients with gastrointestinal malignancies (5). PNI was established in 1988 by Onodera et al. (5) and its clinical utility is broadly investigated in several types of cancer patients (6-8). Although PNI was originally proposed to assess the perioperative immunonutritional status and surgical risk in patients undergoing gastrointestinal surgery, it has been shown to be a prognostic marker for various malignancies. Actually, recent studies revealed that PNI is a simple and useful marker for predicting the outcomes of GC patients (9, 10).
Therefore, the aim of the present study was to investigate the influence of clinical characteristics including PNI on postoperative survival of patients undergoing TG for GC in a Japanese University Hospital.
Patients and Methods
Patients. We retrospectively reviewed a database of 154 patients (men: women=118: 36) who had undergone TG for GC with Roux-en-Y reconstruction, performed by the same trained surgical team at the Department of Gastroenterological Surgery, Dokkyo Medical University Hospital, between October 2000 and February 2008. All patients underwent preoperative routine laboratory tests including PNI on the day of admission. The PNI was calculated as 10 × serum albumin level (g/dl) + 0.005 × total lymphocyte count (/mm3) (5). Although the day of admission and the day of surgery were not same day, none of the patients received chemotherapy and irradiation before surgery.
The cut-off values of clinicolaboratory characteristics were determined using receiver operating characteristic (ROC) curve analyses. The recommended cut-off values of such characteristics were based on the most prominent point on the ROC curve for “sensitivity” and “1-specificity”, respectively. Then, the ideal cut-off values were defined using the Youden index (maximum (sensitivity+specificity-1)) (11). ROC curve revealed that most prominent point disclosed two parameters such as sensitivity (0.596) and specificity (0.635). These two parameters indicated maximum Youden index. Because these two parameters also indicated a recommended cut-off value as 45.7, we defined a cut-off value of PNI as 45. The area under the ROC (AUROC) curve was also calculated.
Univariate analyses were performed to examine the relationship between overall survival and various clinical characteristics such as gender (men/women), age (≤72/>72) (yr), operation time (≤350/>350) (min), bleeding volume (<355/>355) (ml), the serum levels of carcinoembryonic antigen (CEA) (≤20/>20) (ng/ml), carbohydrate antigen 19-9 (CA19-9) (≤15/>15) (U/ml), C-reactive protein (CRP) (≤0.3/>0.3) (mg/dl) and albumin (≤3.5/>3.5) (g/dl), white blood cell (WBC) count (≤9.5/>9.5) (×103/mm3), neutrophil ratio (≤56/>56) (%), monocyte ratio (≤5.8/>5.8) (%), lymphocyte ratio (≤25/>25) (%), platelet count (≤24/>24) (×104/mm3), neutrophil to lymphocyte ratio (NLR) (12) (≤2.6/>2.6), body-mass index (BMI) (≤24/>24) (kg/m2), PNI (≤45/>45), type of tumor (3, 4, 5/0, 1, 2), number of tumors (≥2/1), pathology (others/pap, tub1, tub2; see below for definitions), lymphatic invasion (presence/absence), venous invasion (presence/absence) and lymph node metastasis (presence/absence). All these cut-off values were defined using ROC curve analyses, except for pathology-related characteristics such as tumor type (3, 4, 5/0, 1, 2), number of tumors (≥2/1), pathology (others/pap, tub1, tub2), lymphatic invasion (presence/absence), venous invasion (presence/absence) and lymph node metastasis (presence/absence).
In order to detect the clinical characteristics most closely associated with postoperative survival, multivariate analysis was performed using variables selected from the results of univariate analysis on the basis of a cut-off probability value of p<0.05.
Definition of macroscopic tumor types and pathological findings. Macroscopic tumor types are classified according to the Japanese classification of gastric carcinoma-2nd English edition (13) as: Type 0, superficial tumor; Type 1, polypoid tumor; Type 2, ulcerated tumor with a clear margin; Type 3, ulcerated tumor with infiltration; Type 4, diffusely infiltrating tumor; and Type 5, unclassified tumor. The pathological types of tumors are also defined as: pap, papillary adenocarcinoma; tub1, well differentiated adenocarcinoma; tub2, moderately differentiated adenocarcinoma; por, poorly differentiated adenocarcinoma; muc, mucinous adenocarcinoma; sig, signet-ring cell carcinoma.
Invasion of vessels, i.e. lymphatic invasion (ly) and venous invasion (v), is diagnosed as: ly0 (v0), no invasion; ly1 (v1), minimal invasion; ly2 (v2), moderate invasion; ly3 (v3) and severe invasion.
The depth of tumor invasion (T) is recorded as the T-category: TX, depth of tumor unknown; T1, tumor invasion of mucosa and/or muscularis mucosa or submucosa; T2, tumor invasion of muscularis propria or subserosa; and T3, tumor penetration of serosa; T4, tumor invasion of adjacent structures.
Stage grouping is defined as from I to IV according to the Japanese classification of gastric carcinoma-2nd English edition (13).
Lymph node dissection (D). The lymph node dissection (D) is recorded as the D-category: D0, no dissection or incomplete dissection of the Group 1 nodes; D1, dissection of all the Group 1 nodes; D2, dissection of all the Group 1 and Group 2 nodes; and D3, dissection of all the Group 1, Group 2 and Group 3 nodes.
Grouping of lymph nodes is defined as from 1 to 3 according to the Japanese classification of gastric carcinoma-2nd English edition (13).
Statistical analysis. Data are presented as mean±standard deviation (SD). Differences between groups were analyzed using the chi-squared test and Mann-Whitney U-test. Hazard ratio with 95% confidence intervals (C.I.) was calculated using univariate or multivariate analysis. Multivariate analysis was performed using clinical characteristics with p<0.05 selected in the univariate analysis to assess the most variable characteristic of postoperative survival using Cox proportional hazard test.
Deaths before 31 March 2009 were included in this analysis. Statistical analyses were performed using the SPSS statistical software package, version 16.0 (SPSS Inc., Chicago, IL, USA) at a significance level of p<0.05.
Results
Table I shows the background clinicopathological characteristics of the patients undergoing TG for GC. The patients were divided into two groups using the cut-off value of PNI (>45/≤45). There were no significant differences between the two groups, except for gender (men/women) (p=0.037), age (≤72/>72) (p=0.034), venous invasion (absence/presence) (p=0.007), lymph node metastasis (absence/presence) (p=0.012), tumor depth (T1/T2/T3/T4) (p=0.047), lymph node dissection (D0/D1/D2/D3) (p=0.014) and stage (I/II/III/IV) (p=0.006).
Table II shows the background clinicolaboratory characteristics of patients undergoing TG for GC. The patients were also divided into two groups using the cut-off value of PNI (>45/≤45). There were no significant differences between the two groups, except for age (years) (p=0.006), operation time (min) (p=0.005), serum level of albumin (g/dl) (p<0.001), neutrophil ratio (%) (p<0.001), lymphocyte ratio (%) (p<0.001), NLR (p<0.001), BMI (kg/m2) (p<0.001) and survival period (days) (p=0.006).
Relationships between background characteristics and PNI in patients undergoing total gastrectomy for gastric cancer.
Univariate analyses were performed to evaluate the relationship between the clinical characteristics and overall survival. All such characteristics were divided into two groups using cut-off values.
Under such conditions, the results of univariate analyses demonstrated that age (≤72/>72) (years) (hazard ratio, 2.175; 95% C.I., 1.381-3.427; p<0.001), operation time (≤350/>350) (min) (hazard ratio, 0.557; 95% C.I., 0.354-0.941; p=0.028), the serum levels of CEA (≤20/>20) (ng/ml) (hazard ratio, 2.968; 95% C.I., 1.536-5.736; p=0.001) and albumin (≤3.5/>3.5) (g/dl) (hazard ratio, 0.395; 95% C.I., 0.249-0.627; p<0.001), WBC count (≤9.5/>9.5) (×103/mm3) (hazard ratio, 2.800; 95% C.I., 1.387-5.651; p=0.004), lymphocyte ratio (≤25/>25) (%) (hazard ratio, 0.437; 95% C.I., 0.277-0.689; p<0.001), NLR (≤2.6/>2.6) (hazard ratio, 2.252; 95% C.I., 1.423-3.562; p<0.001), PNI (≤45/>45) (hazard ratio, 0.396; 95% C.I., 0.246-0.639; p<0.001), type of tumor (3, 4, 5/0, 1, 2) (hazard ratio, 0.440; 95% C.I., 0.275-0.704; p<0.001), lymphatic invasion (presence/absence) (hazard ratio, 0.181; 95% C.I., 0.066-0.497; p<0.001), venous invasion (presence/absence) (hazard ratio, 0.255; 95% C.I., 0.134-0.487; p<0.001) and lymph node metastasis (presence/absence) (hazard ratio, 0.169; 95% C.I., 0.088-0.323; p<0.001) were associated with overall survival, respectively (Table III).
Relationships between clinical characteristics and PNI in patients undergoing total gastrectomy for gastric cancer.
Then, multivariate analysis was performed using these twelve selected clinical characteristics found to have a significance level of p<0.05 in the univariate analysis to clarify those most closely associated with overall survival of patients undergoing TG for GC. The results demonstrated that PNI (≤45/>45) (hazard ratio=0.301; 95% C.I.=0.129-0.705; p=0.006) had an association with overall survival, along with age (≤72/>72) (years) (hazard ratio=1.829; 95% C.I.=1.054-3.172; p=0.032), CEA (≤20/>20) (ng/ml) (hazard ratio=2.782; 95% C.I.=1.313-5.894; p=0.008), WBC count (≤9.5/>9.5) (×103/mm3) (hazard ratio, 2.848; 95% C.I., 1.164-6.970; p=0.022) and lymph node metastasis (presence/absence) (hazard ratio, 0.318; 95% C.I., 0.134-0.754; p=0.009) (Table IV). All of these five clinical characteristics were evaluable before surgery, except for pathologically-diagnosed lymph node metastasis.
Among the four preoperatively evaluable clinical characteristics, ROC curve analysis revealed that PNI (0.594) had the largest AUROC curve in comparison with the other characteristics (age 0.589, CEA 0.569, and WBC count 0.534), with a sensitivity of 59.6% and a specificity of 63.5%, respectively.
The median and maximum follow-up periods for survivors were 1,246 and 3,590 days, respectively. The mean postoperative survival period was 1,314±931 days (mean±SD).
There was a significant difference in postoperative survival period between the two groups (PNI>45/ PNI≤45) (Table II). Kaplan-Meier analysis and log rank test demonstrated that there was a significant difference in overall survival between these two groups (PNI>45/ PNI≤45). Patients with higher PNI (>45) had a higher postoperative survival rate than those with lower PNI (≤45) (p<0.001) (Figure 1). The PNI was thus able to divide such patients into two independent groups before surgery.
Discussion
The PNI was first reported by Onodera et al. in 1984. In their report, the cut-off values of PNI indicated several types of operative risk: resection and anastomosis of the gastrointestinal tract was safely feasible for PNI >45; the same procedure appeared to be dangerous for PNI ≤40 -≤45; and such surgery appeared to be contraindicated for PNI <40 (5). On the basis of these cut-off values, our present cut-off values seem reasonable as PNI (45.7) was able to preoperatively divide GC patients undergoing TG into two subgroups with a better and worse outcome. Recently, this index has been re-estimated in patients with GC. Watanabe et al. clearly showed that PNI is an independent prognostic factor in elderly patients undergoing surgery for GC. Interestingly, they determined the PNI cut-off value as 44.7 (14). This strongly supports the result we obtained by ROC curve analysis, which indicated that the ideal cut-off value of PNI was 45.7, the two cut-off values being almost the same.
Nozoe et al. also decided the cut-off value of PNI using the mean PNI value of 49.7 (range 29.8-65.8). Although, their value appeared to be higher than ours (44.5±7.4 (mean±SD); median=45.5, range=15.8-77.8), they reached a similar conclusion, i.e. “The PNI value can be a useful tool to predict the prognosis of patients with gastric carcinoma” (15).
On the other hand, although BMI is one of the important nutritional markers in patients with several types of cancers, two recent studies (16, 17) have revealed that there was no significant relationship between preoperative BMI and postoperative survival. Nozoe et al. reported the impact of BMI in patients with GC. According to them, although no significant correlation was found between obesity and tumor-related factors, BMI might represent the physical condition brought about by the extent of tumor progression, rather than a factor influencing the tumor-related features of GC. They reported that there were no significant differences in overall survival among three groups of GC patients (BMI<18.5, 18.5≤BMI≤25, BMI>25) (16).
Relationship between PNI (patients with PNI (>45) and PNI (≤45) from top to bottom) and overall survival in patients undergoing TG for GC.
Oh et al. also reported the effect of being overweight on postoperative morbidity and long-term outcome of patients with proximal GC undergoing TG. They also stated that although being overweight was associated with a longer operation time and a higher risk of complications in GC patients undergoing curative TG, it had no effect on recurrence or long-term survival. They concluded that there was no significant difference in overall survival between the two groups of patients with proximal GC (BMI≤25, BMI>25) (17).
On the basis of the above evidence, PNI might be superior to BMI for prognostication of patients with GC, as demonstrated by our finding that, although PNI and BMI had a close relationship to each other, univariate analyses revealed no relationship between BMI and overall survival in GC patients undergoing TG.
Unfortunately, although there have been no such reports from Western countries, the above four studies of patients with GC appeared to reach the same conclusion. In particular, the results of multivariate analyses revealed interesting results as follows.
Watanabe et al. demonstrated that six clinical characteristics – histology, tumor depth, lymph node metastasis, neo- and adjuvant chemotherapy, and PNI – were associated with overall survival in elderly patients with GC (14).
Univariate analyses of clinical characteristics in relation to overall survival.
Multivariate analysis of clinical characteristics in relation to overall survival.
Oh et al. also reported that four clinical characteristics – age (>55/≤55), tumor size (5.5>/≤5.5 cm), serosal invasion (presence/absence) and lymph node metastasis (presence/absence) – were associated with overall survival in patients with proximal GC undergoing TG (17).
Comparison between the former and latter studies showed that tumor depth and lymph node metastasis might be two valuable and important clinical characteristics associated with overall survival of GC patients undergoing surgery, because our study demonstrated that lymph node metastasis was one of the important clinical characteristics associated with overall survival as well as PNI. Therefore, it is quite acceptable to conclude that PNI is associated with postoperative survival in patients undergoing TG for GC as well as pathologically-diagnosed lymph node metastasis, since the above two studies support our results.
Several nutritional indices have been reported, such as the nutritional risk index (NRI) (NRI=1.519 × serum albumin level (g/l) + 0.417 × (current weight/usual weight) ×100) (18), nutritional assessment index (NAI) (NAI=2.64 × arm circumference (cm) + serum pre-albumin level (mg/dl) + 3.7 × serum retinol-binding protein level (mg/dl) x purified protein derivative (PPD) skin test (mm2) - 53.8) (19) and prognostic nutritional index for surgery (PNIS) (PNIS=−0.147 × weight change (%) + 0.046 × ideal body weight (%) + 0.010 × (actual triceps skinfold thickness as a % of the standard value) + 0.051 × hepaplastin test (%)) (20), including PNI (4).
Among these NRI, the PNI is not only the easiest index to estimate but also a repeatable index before surgery because PNI consists of only two routine laboratory characteristics: the serum level of albumin and the total count of lymphocytes in peripheral blood. In fact, the results of univariate analyses revealed that these two laboratory characteristics had a significant relationship to overall survival as well as WBC count, lymphocyte ratio and NLR. Especially, these laboratory characteristics are considered to be related to the systemic inflammatory response (SIR) (21), since SIR might reflect hypercytokinemia resulting from immunoreactions based on tumor versus host interactions (22). Among several SIR-related cytokines, interleukin-6 (IL-6) is known to play a crucial role in protein synthesis in the liver (23). IL-6 stimulates hepatocytes and exerts an adverse effect, not only down-regulating the synthesis of albumin but also up-regulating that of CRP (24). Similarly, IL-6 is known to stimulate cell proliferation in the bone marrow, promoting not only neutrophil proliferation but also reduction of the lymphocyte ratio (25). Therefore, it is acceptable that PNI might be able to indicate not only nutritional status but also SIR.
Although it is difficult to explain all of the parameters that showed significant inter-group differences, it seems reasonable that patients with lower PNI (≤45) would have had not only more marked tumor-related characteristics such as presence of venous invasion, presence of lymph node metastasis, tumor depth, lymph node dissection and tumor stage, but also SIR-related characteristics such as a lower serum albumin level, higher neutrophil ratio, lower lymphocyte ratio, higher NLR and lower BMI than patients with higher PNI (>45). These situations reflect tumor progression because cachexia status is induced by SIR-related hypercytokinemia through tumor versus host interactions (26, 27).
Thus, our hypothesis that preoperative PNI, as well as a better preoperative nutritional status, is associated with the survival of GC patients undergoing TG appears to have been validated. Evaluation of preoperative PNI would be useful for selecting patients with a poor prognosis because PNI is able to divide such patients into two independent groups before surgery.
Footnotes
-
Conflicts of Interest
The Authors have no conflicts of interest to declare. Furthermore, no funding/grant support for this study was received.
- Received April 11, 2014.
- Revision received May 31, 2014.
- Accepted June 3, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
