Clinical predictive factors of long‐term survival after curative resection of pancreatic cancer: a retrospective study

Abstract Pancreatic ductal adenocarcinoma (PDAC) continues to have the poorest prognosis of all gastrointestinal malignancies, even after the tumor has been completely resected. However, only a proportion of patients achieve 5‐year survival after resection. The factors predictive of achieving 5‐year survival remain unclear. The aim of this study was to investigate the pre‐ and postoperative clinicopathological characteristics of PDAC patients with a >5‐year survival after curative resection. We retrospectively reviewed patients who underwent pancreatectomy for PDAC between January 1995 and December 2011. Logistic regression analysis was performed to determine the predictive factors for 5‐year survival. One hundred and fifty‐one patients were enrolled, including 38 patients with 5‐year survival (actual 5‐year survival rate, 25.2%). The independent preoperative factors predictive of achieving 5‐year survival included serum albumin levels (odds ratio [OR]: 5.06, 95.0% confidence interval [CI]: 1.49–17.19; P = 0.009) and neoadjuvant chemoradiotherapy (OR: 3.02, 95.0% CI: 1.00–9.08; P = 0.049). Venous infiltration (OR: 2.99, 95.0% CI: 1.09–8.25; P = 0.034), liver recurrence (OR: 0.17, 95.0% CI: 0.04–0.69; P = 0.013), and perioperative portal vein infusion chemotherapy (OR: 3.06, 95.0% CI: 1.09–8.25; P = 0.028) were found to be independent postoperative predictive factors for achieving 5‐year survival. Serum albumin levels could be a biomarker for predicting the prognosis of PDAC patients after curative resection. Liver recurrence and perioperative portal vein infusion chemotherapy were independent postoperative factors, suggesting that perioperative portal vein infusion chemotherapy could be promising for improving the survival rate of PDAC patients after curative resection.


Introduction
The 5-year survival rate of pancreatic ductal adenocarcinoma (PDAC) patients remains poor relative to other gastrointestinal malignancies [1,2], even after complete curative resection. Recently, the 5-year survival rates have improved due to various developments in pre-and postoperative chemo-and chemoradiotherapy. Neoadjuvant chemoradiotherapy (NACRT) is considered an acceptable treatment for locally advanced PDAC [3,4], and postoperative adjuvant chemotherapy has been widely administered [5,6]. However, the 5-year survival rate of PDAC patients remains low despite the development of these multimodal therapies.
Several retrospective studies [7][8][9] have reported that a proportion of patients who underwent curative resection for PDAC have survived for >5 years. Almost all the factors predictive of 5-year survival in these studies [7][8][9] are associated with pathological data (e.g., resection margin and lymph node metastasis status). However, clinical factors that

ORIGINAL RESEARCH
Clinical predictive factors of long-term survival after curative resection of pancreatic cancer: a retrospective study predict long-term survival in PDAC patients, especially preand postoperative chemo-or chemoradiotherapy and preoperative laboratory data, have yet to be fully elucidated.
In this study, we aimed to investigate pre-and postoperative clinicopathological characteristics of PDAC patients who survived for >5 years after curative resection, focusing on pre-and postoperative chemo-or chemoradiotherapy and preoperative laboratory data.

Patients
Patients who underwent intended curative pancreatectomy for PDAC at our institution between January 1995 and December 2011 were retrospectively reviewed. All patients were histologically confirmed as having PDAC. We excluded patients who underwent R2 resections. All study participants had provided informed written consent. The Human Experimentation Committee of our institution approved this study. Research was conducted in accordance with the Declaration of Helsinki.
Patients were stratified into two groups according to survival: the 5-year survival group (patients who survived for >5 years) and the <5-year survival group (patients who died or were lost to follow-up within 5 years of curative pancreatectomy).

Surgical resection and pathology
Surgical resection included pancreaticoduodenectomy, distal pancreatectomy, and total pancreatectomy. D2 lymph node dissections were performed in all patients.
Pathological staging of all specimens was determined according to the UICC TNM Classification of Malignant Tumors (seventh edition). R0 resections were defined as cases without gross or microscopic evidence of residue disease. R1 resections had microscopically positive margins and R2 resections still contained some gross tumor matter. Pathological features associated with prognosis included histologically assessed tumor size; distal bile duct, duodenal, serosal, retropancreatic tissue, portal venous or arterial system, and extrapancreatic nerve plexus invasion; other organ invasion; lymph node metastasis; and lymphatic, venous, or intrapancreatic neural infiltration [16].

Postoperative assessment
Since the year 1986, perioperative portal vein infusion (PVI) chemotherapy has been performed as a standard treatment to prevent liver recurrence and improve survival in pancreatic cancer patients undergoing potentially curative resection at our institution. PVI was adopted in only those patients who gave informed consent for this treatment. PVI chemotherapy consisted of 5-FU and heparin (250.0 mg/day and 2000 IU/[body•day], days 1-7 each week for 4 weeks), mitomycin C (4.0 mg/[body•day], days 6, 13, 20, and 27), and cisplatin (10.0 mg/[body•day], days 7, 14, 21, and 28) [17,18]. S-1, gemcitabine, 5-FU, or mitomycin C was administered as adjuvant chemotherapy at the physician's discretion. Postoperative complications were evaluated according to the Clavien-Dindo classification. In this study, pancreatic fistulas, intraabdominal bleeding, delayed gastric emptying, and fluid collection were also investigated.

Follow-up
Patients were followed-up 1, 3, 6, and 12 months after surgery. Patients were also subject to semiannual reviews. Clinical examinations, laboratory investigations, and abdominal computed tomography scans (to detect tumor recurrence) were performed. Disease-free survival (DFS) was calculated from the date of surgery to the date of recurrence or last follow-up and overall survival (OS) Y. Nakanov et al. Predictors of Achieving 5-year Survival in PDAC was calculated from the date of surgery to the date of death or last follow-up. Patients were followed up until death or December 2016.

Statistical analyses
The clinicopathological characteristics between the 5-year survival and <5-year survival groups were evaluated and factors predictive of achieving 5-year survival were investigated using logistic regression analysis. The optimal cutoff points for predicting 5-year survival were estimated from a receiver operating characteristic (ROC) curve analysis. Survival curves were plotted using the Kaplan-Meier method and compared by the log-rank test. Variables that were significant in the univariate analysis were included in the multivariate analysis to identify prognostic factors of survival. All statistical analyses were conducted using Statistical Package for the Social Sciences for Macintosh, software version 23.0 (IBM Corp., Armonk, NY). A P < 0.05 was considered statistically significant.

Patient characteristics and surgical procedures
In total, 153 patients underwent curative resection for PDAC between January 1995 and December 2011. Two patients who underwent R2 resections were excluded. Therefore, 151 patients were enrolled in the final study. The patient characteristics are summarized in Table 1.
Of The preoperative data are summarized in Table 2. Several factors were identified in the univariate analysis as being significantly different between the 5-year survival and <5year survival groups, including NACRT, serum albumin levels, and the WBC and lymphocyte counts.

Preoperative albumin levels and survival
ROC curve analysis demonstrated that a preoperative serum albumin level of 3.9 g/dL (area under the curve, 0.670; P = 0.002) was the optimal cut-off for 5-year survival. In the survival analysis according to the Kaplan-Meier method, patients with preoperative serum albumin levels of ≥3.9 g/dL had a significantly longer DFS compared to patients with preoperative serum albumin levels of <3.9 g/dL ( Fig. 1A; P = 0.029). The OS of these patients was also longer, although this was not significant (Fig. 1B).

Other predictive factors and survival
Patients receiving NACRT did not have a significantly longer DFS (Fig. 2A). However, they did have a significantly longer OS ( Fig. 2B; P = 0.045). Adjuvant PVI chemotherapy was not a significant predictor of DFS (Fig. 2C), but was predictive of OS ( Fig. 2D; P = 0.05).

Discussion
This study investigated the pre-and postoperative predictive factors related to survival in patients with 5-year survival compared to those with <5-year survival. Our findings demonstrated that NACRT and preoperative serum albumin levels were independent preoperative predictive factors for achieving 5-year survival. Additionally, venous infiltration, perioperative PVI chemotherapy, and liver recurrence were confirmed as independent postoperative predictive factors for achieving 5-year survival.
By analyzing patients who survived for >5 years, factors for achieving 5-year survival could be identified more accurately. Previous retrospective studies [7][8][9] have identified preoperative serum CA 19-9 levels, multiple operations, lymph node metastasis, and R0 resections as prognostic factors in PDAC patients with an actual 5-year survival after curative resection. These studies [7][8][9] also  NACRT has been used to reduce tumor volume and improve margin-negative resection rates in patients with locally advanced PDAC [4,19]. A clinical trial [20] has made progress. However, standardized protocols have yet to be established. At our institution, we began administering NACRT in 2003. We have previously reported [19] on two cases of PDAC with pathological complete responses to NACRT and have included these cases in our 5-year survival group. Katz et al. [9] emphasized the importance of staging and patient selection, a standardized operative approach, and the routine use of multimodal therapy for achieving 5-year survival rates.
Previous studies [21,22] have suggested that in patients with non-small-cell lung cancer or epithelial ovarian cancer, preoperative serum albumin levels are associated with a predictive survival rate. Low serum albumin levels are associated with a risk of postoperative complications and poor survival outcomes because of malnutrition [23]. The prognostic factors related to serum albumin levels (e.g., CRPto-albumin ratio, PNI, and mGPS) [12][13][14] have previously been identified in PDAC patients. However, few studies have reported that preoperative serum albumin levels alone can predict survival. Using prospective data from two randomized controlled trials [24,25], Stocken et al. [26] have identified serum albumin as one of the prognostic factors for survival in PDAC. We believe that preoperative serum albumin levels may represent a useful biomarker for predicting the prognosis of PDAC patients after resection.
We hypothesize that preoperative albumin levels may be influenced by several factors, including systemic inflammation, disease state, nutritional status, liver cirrhosis, and the quantity of skeletal muscle [27][28][29][30][31]. Several studies [32][33][34] have demonstrated that preoperative inflammation and postoperative complications can lead to poor survival outcomes in patients with various types of tumors. In our study, preoperative CRP levels, WBC counts, and fibrinogen levels, measured acute or chronic inflammatory markers, did not correlate with preoperative albumin levels (data not shown). Therefore, we speculate that preoperative serum albumin levels may reflect nutritional depletion and loss of skeletal muscle.
It remains unclear whether improving preoperative serum albumin levels can lead to a favorable prognosis. The criteria of the European Society for Clinical Nutrition and Metabolism [35] demonstrate that low serum albumin levels is an established marker of malnutrition, and improving perioperative nutrition, especially immunonutrition, can result in reduced postoperative complications and hospital stays. In pancreatectomy, preoperative providing oral immunonutrition helps in reducing risk of postoperative complications such as infections, and also the length

A B
Y. Nakanov et al. Predictors of Achieving 5-year Survival in PDAC of hospital stays [36]. From a molecular viewpoint, the overexpression of caspases and the reduced levels of antiapoptotic proteins in PDAC patients may lead to lymphocyte apoptosis and dysfunction and immune system suppression, especially after pancreatectomy [37]. Therefore, improving perioperative immunonutrition may reduce the risk of immune disorders in patients with PDAC.
We consider it important to control for disease state and cancer recurrence. Among the 5-year survivors, recurrence (especially liver recurrence) was a poor prognostic factor. To prevent liver recurrence and improve survival in PDAC patients who undergo potentially curative resections, we have administered perioperative PVI chemotherapy with 5-FU at our institution since 1986 [17]. We have already reported that perioperative PVI chemotherapy is beneficial in preventing liver recurrence [18], and in this study, perioperative PVI chemotherapy was identified as an independent postoperative predictive factor for achieving 5-year survival.
In our univariate analysis, the pattern of initial recurrence (except lung recurrence) was a significant risk factor for achieving 5-year survival. Several studies [38][39][40] have revealed that patients with isolated lung recurrence as the initial recurrence followed a better course than patients with other recurrences [38] for a variety of reasons. First, lung recurrence was more likely to occur late with a longer time to development than other sites of recurrence [39]. Second, lung recurrence may be associated with unique and different phenotypes to those with other patterns of lung metastases in PDAC patients [38]. Finally, the outcomes of patients who underwent reoperation for isolated lung recurrence were superior [40]. In this study, patients with lung recurrence as the initial recurrence survived longer than patients with other patterns of recurrence (median OS: 45 vs. 18 months; P = 0.008).
There are several limitations of this study. First, this was a retrospective study conducted at a single institution. Second, it was unclear whether improving preoperative serum albumin levels resulted in long-term survival benefits after resection. Although the relationship between improving serum albumin levels and variations in oncological behavior is not well understood, several studies [26,28,29,41] have demonstrated an association between hypoalbuminemia and poor postoperative short-/long-term survival in patients with various types of malignancies. We believe that preoperative serum albumin levels represent a potential biomarker for predicting prognosis in PDAC patients after curative resection. However, further studies will need to be conducted in order to evaluate the relationship between serum albumin levels and improved survival after pancreatectomy.
In conclusion, preoperative serum albumin levels may represent a useful biomarker for predicting prognosis in PDAC patients after curative resection. This factor was more straightforward in clinical situations than other prognostic factors that are related to serum albumin levels (e.g., the CRP to albumin ratio, PNI, and mGPS). NACRT and perioperative PVI chemotherapy were also identified as significant predictive factors. Patients with liver recurrence were more likely to have a poor prognosis. Therefore, perioperative chemotherapy (e.g., PVI) is well suited to improve survival after pancreatectomy.