C-Reactive Protein/Albumin Ratio and Prognostic Nutritional Index Are Strong Prognostic Indicators of Survival in Resected Pancreatic Ductal Adenocarcinoma

Abstract Purpose: We evaluated the clinical importance, such as the occurrence of postoperative pancreatic fistula (POPF) or prognosis, of preoperative serum markers of chronic inflammation, nutrition, and immunity, as well as that of serum tumor markers after curative resection of pancreatic ductal adenocarcinomas (PDACs). Methods: Between 2006 and 2015, 43 patients with PDACs underwent curative resection at Tottori Prefectural Central Hospital. We analyzed which preoperative indicators (i.e., C-reactive protein/albumin ratio [CAR], neutrophil/lymphocyte ratio [NLR], prognostic nutritional index [PNI], carcinoembryonic antigen [CEA], and carbohydrate antigen 19-9 [CA 19-9]) were the most relevant risk factors for occurrence of POPF and poor patient survival. Results: POPF was detected in 8/43 (18.6%) patients. One patient died of pancreatic fistula at 2 months postoperatively. Among nine candidate factors (operative procedure, operation time, tumor stage, preoperative serum amylase, preoperative CAR, NLR, PNI, CEA, and CA 19-9), we did not identify any significant risk factor for the occurrence of POPF. The 5-year overall survival (OS) rate of the 43 patients was 22.4%, and the overall median survival time was 21 months. The multivariate OS analysis demonstrated that high CAR and low PNI were strong preoperative markers of poor prognosis independently of tumor stage. Conclusions: Preoperative CAR and PNI are useful prognostic markers for patients with operable PDACs.


Introduction
The prognosis of patients with pancreatic ductal adenocarcinomas (PDACs) is extremely poor. Even though curative resection is performed, many patients die of cancer recurrence within a few months. 1-3 Yamamoto et al. 4 reported that the median survival of 195 patients who underwent pancreatic resection was 27.1 months, and the 5-year actuarial survival rate was 34.5%. In addition, Sato et al. 5 reported that in patients with resected PDACs in Japan, the estimated 5-year disease-specific survival rate was 23.1%, and the disease-free survival (DFS) rate was 16.8%. Moreover, it has been reported that after resection of the pancreas, postoperative complications occur very frequently. Anastomotic leakages between the pancreatic duct and the jejunum have been observed in 14-17% of patients after pancreatoduodenectomy (PD). [6][7][8] In addition, pancreatic fistula is frequently detected (16-26%) after distal pancreatectomy (DP). [9][10][11] Such anastomotic leakages or pancreatic fistula are often fatal.
Recently, several studies found a strong correlation between poor survival and preoperative chronic inflammation, poor nutrition, and low immunity among patients with various cancers. [12][13][14] Patient preoperative C-reactive protein/albumin ratio (CAR) has been used in combination with other parameters not only to diagnose chronic inflammation but also to assess the nutritional status of patients with cancer. Neutrophil/lymphocyte ratio (NLR) and prognostic nutritional index (PNI) are markers of chronic systemic inflammation and the patients' immune status and nutritional condition. 15,16 Also, serum carcinoembryonic antigen (CEA) and carbohydrate antigen  levels are known to be good progression indicators among patients with PDACs, including followup after resection of tumors. 1,17 In the present study, we evaluated the prognostic and clinical importance of serum markers of chronic inflammation, nutrition, and immunity combined with tumor markers of patients with PDACs.

Patients
Between 2006 and 2015, we enrolled 43 patients who were diagnosed with operable PDAC and underwent curative resection (no residual tumors macroscopically) at Tottori Prefectural Central Hospital. The average age of the 30 male and 13 female patients was 71.6 (range: 37-88) years. Clinical and pathological staging of PDACs was performed using the American Joint Committee on Cancer 6th edition tumor-node-metastasis (TNM) staging system for pancreatic cancer. 18 All patients were followed up at Tottori Prefectural Central Hospital until the end of 2016. The mean follow-up period was 25.2 (range: 1-80) months. No patients received chemotherapy and radiation therapy before the operation. Informed consent for medical treatment and use of clinical data from medical records were obtained from all patients.

Surgical procedure
All patients underwent open laparotomy. According to the tumor location, we performed DP with splenectomy in 14 patients, subtotal stomach-preserving pancreaticoduodenectomy (SSPPD) in 24 patients, SSPPD combined with portal vein partial resection in 4 patients, and total pancreatectomy with splenectomy in 1 patient. After DP, we closed the proximal pancreas stump by suturing with 4-0 prolene in 12 patients and by linear stapler in 2 patients. Reconstruction after SSPPD or SSPPD with partial resection of the portal vein was performed by Child's method (e.g., Braun's anastomosis after pancreaticojeju-nostomy, choledochojejunostomy, and gastrojejunostomy). 19 An internal short stent was placed across a pancreaticojejunostomy, but no stent was placed following choledochojejunostomy.

Postoperative complications
Postoperative complications were analyzed by patient clinical record data. The severity of postoperative complications was graded according to the Clavien-Dindo classification. 20 In addition, a Clavien-Dindo score of III or higher was considered a major complication.

Study design
This study was a retrospective study. We recorded clinical parameters, including operation time, postoperative hospital stay, and occurrence of postoperative complications. Blood samples were taken from each patient routinely at the time of admission to our hospital. C-reactive protein, serum albumin, serum amylase, total lymphocyte count, and tumor markers, (including CEA and CA 19-9), were recorded. In addition, CAR, NLR, and PNI were calculated. The PNI was calculated using the following formula: 10· serum albumin concentration (g/dL) + 0.005 · lymphocyte counts (no./mm 3 ) in peripheral blood. 15 The reference ranges of CAR, NLR, and PNI were determined by investigating the preoperative CAR, NLR, and PNI in 147 patients who underwent radical inguinal hernia operations from 2012 to 2015 at our hospital. None of these patients had either chronic or acute inflammation. The average age of the 147 control patients was 69 years (range: 27-93 years). Ninety-two were male, and 55 were female. The mean CAR, NLR, and PNI among of the 147 control patients were 0.037 (range: 0.002-0.464), 2.5 (range: 0.9-19.4), and 49.9 (range: 35.2-61.7), respectively. To determine the cutoff values corresponding to the CAR, NLR, and PNI, we plotted receiver operating characteristic (ROC) curves for the 147 control patients and 43 patients with PDACs. The Youden index (sensitivity + specificity À 1) was used, and the highest Youden index values were considered the cutoff levels. 21 According to a previous report, the normal range of serum amylase, CEA, and CA 19-9 levels were 37-125 U/L, 0-5 ng/mL, and 0-37 U/mL, respectively. 22 This study was approved by the Ethical Review Board of Tottori Prefectural Central Hospital (approval number: 2016-49).

Statistical analysis
Differences between two normally distributed parameters were compared using the v 2 and Fisher's exact probability tests. The Mann-Whitney U test was used to compare differences between two parameters with nonnormal distribution. Correlations between the parameters and occurrence of postoperative pancreatic fistula (POPF) were estimated using logistic-regression analysis. Long-term OS was calculated using the Kaplan-Meier method and the prognostic difference between the two groups was compared using the logrank test. All data were analyzed by StatView software (Abacus Concepts, Inc., Berkeley, CA). p < 0.05 was regarded as indicating statistical significance.

Results
The clinical characteristics, operative procedure, operation time, postoperative complications, and postoperative hospital stay of all 43 patients with PDACs are presented in Table 1. Postoperative complications were detected in 34.9% of cases. POPF secondary to pancreatic juice leakage at the anastomotic site or at the pancreas stump was frequently detected (8/43, 18.6%). The mean length of postoperative hospital stay of the 15 patients with postoperative complications (61.4 days) was significantly longer than that of the 28 patients without postoperative complications (22.2 days, p < 0.001). One patient died from postoperative peritonitis secondary to POPF 2 months postoperatively. Thus, the operative mortality and morbidity in our series were 2.3% and 34.9%, respectively.
The mean and median (range) of preoperative serum amylase, CAR, NLR, PNI, CEA, and CA 19-9 of all 43 patients with PDACs were as follows: serum amylase We analyzed factors that were thought to influence the occurrence of POPF. The nine candidate factors were operative procedure, operation time, tumor stage, preoperative serum amylase, preoperative CAR, NLR, PNI, CEA, and CA 19-9. We divided the patients into two subgroups (long and short) according to the median values of operation time, but we could not find any significant risk factor for the occurrence of POPF (data not shown).
The 5-year overall survival (OS) rate of the 43 patients was 22.4%, and the overall median survival time (MST) was 21 months. In addition, the 5-year DFS rate of the 43 patients was 9.1%, and the diseasefree MST was 9 months. The correlation between the 10 candidate factors (operative procedure, operation time, tumor stage, occurrence of postoperative complications, preoperative serum amylase, preoperative CAR, NLR, PNI, CEA, and CA 19-9) and the overall MST and disease-free MST of 43 patients are shown in Table 2. We found that low preoperative PNI was important risk factor of poor survival in patients with PDACs and, we found that there was even no statistical significance, high preoperative CAR was also risk factor of poor survival. Figure 1 shows the OS curves of 28 patients with high PNI (solid line) and 15 patients with low PNI (dotted line). The 5-year OS of patients with low PNI (4%) was significantly poorer than that of patients with high PNI (35%, p = 0.001). The multivariate analysis of OS showed that high CAR and low PNI were strong prognostic preoperative markers independently of tumor stage (Table 3). In addition, in multivariate analysis of DFS, we found that high CAR was a significant poor survival and recurrence preoperative marker (CAR: p = 0.018, odds ratio = 2.948, 95% confidence interval = 1.203-7.222), but low PNI was not (PNI: p = 0.124, odds ratio = 2.247, 95% confidence interval = 0.802-6.289).

Discussion
The incidences of severe POPF (Clavien-Dindo score of III or higher) were reported in 14-23% of patients after PD 23 and in 10-30% of patients after DP. 24 Also, some cases of POPF can be fatal. Soft pancreas or pancreatic thickness has been thought to be risk factors for occurrence of POPF. 25,26 In our series, we found POPF in eight cases (18.6%, PD: 5/28, 17.9% and DP: 3/15, 20%), however, we found no significant correlation between POPF and poor survival.
The preoperative estimation of risk factors for poor prognosis of patients with PDACs has been studied. Pathological tumor stage was reported as a strong prognostic factor for patients with PDACs, but preoperative serum tumor markers, such as CA 19-9 and pancreatic cancer-associated antigen (DUPAN-2), were not. 4 However, clinicians may change the treatment strategy for patients with PDACs based on strong preoperative prognostic makers. For patients who have poor prognosis based on preoperative prognostic markers, we recommend chemoradiotherapy before radical surgery, or perhaps even avoiding the surgical procedure altogether because of the high incidence of postoperative complications. In this study, we retrospectively analyzed the prognostic markers of patients with PDACs  at a single center. In doing so, we found that high CAR and low PNI were strong prognostic preoperative markers independently of tumor stage. Also, we found that postoperative complications did not influence the survival of the patients. Two markers of systemic inflammatory response, CRP and albumin, have been used in combination to diagnose not only chronic inflammation but also the nutritional status of cancer patients. In addition, pretreatment CAR was shown to be a significant prognostic indicator in various carcinomas. [27][28][29] PNI is based on albumin and absolute lymphocyte count, which are measured routinely in clinical practice. This index is designed to assess nutritional and immunological status, which may be useful in predicting prognosis. Moreover, a significant correlation between low PNI and poor survival has been reported in various cancers. 16,30,31 In PDACs, Geng et al. 31 reported that low PNI correlated significantly with shorter OS in patients with advanced pancreatic cancer. In this study, multivariate analysis identified PNI as an independent prognostic factor for OS. Thus, we recommended the preoperative assessment of CAR and PNI before planning the treatment strategies for patients with PDACs. Also, it may be possible that improvement of nutrition before surgery may improve patient's postoperative prognosis in PDACs.

Conclusion
High preoperative CAR and low PNI levels strongly correlated with poor survival in patients with PDACs. These markers should be assessed when selecting the treatment strategies for patients with PDACs.