Introduction

Pancreatic cancer ranked 13th in the list of most commonly diagnosed cancers and was the fifth most common cause of cancer death in China in 2006.1 The 5-year survival rate for all pancreatic cancer patients is below 5%.2 Among the various subtypes of pancreatic cancer, pancreatic ductal adenocarcinoma (PDAC) is by far the most common and most important tumor type, accounting for more than 85% of pancreatic tumors.3 Surgical removal of the tumor with negative resection margins remains to be the most potentially curative therapy.3,4 Late presentation and fear of perioperative mortality5 have left only a minority of patients amenable to curative resection6 in the last decade. Efforts have been made towards early detection with biomarkers and advanced imaging techniques such that perioperative outcome has been markedly improved within the past two decades due to advances in both surgical technique and perioperative care.7,8 Today, pancreas resection can be performed in more than 20% of pancreas cancer patients with localized disease resulting in a very low mortality rate (below 5%)7–9 and producing a 5-year survival rate that exceeds 20% with adjuvant therapy.10,11 With the development of diagnostic capacity and environmental influence, the incidence of PDAC in China has increased in the past several decades.1,12 Though highly specialized centers for pancreatic cancers were built up with skilled surgical oncologists and experienced multimodality physicians, the improvement in treatment modalities and patients’ outcomes have not yet been delineated. The purpose of this study is to investigate the decade-based improvement in diagnosis modalities and surgical treatment, and to compare the different outcomes in the same cancer center between two decades.

Materials and Methods

Retrospectively collected patient data from Tianjin Medical University Cancer Institute and Hospital was used to identify all patients with PDAC that underwent surgical treatment (n = 565). Clinical variables include age at diagnosis, gender, serum CA-199, preoperative bilirium, site, American Joint Committee on Cancer (AJCC) stage,13 perioperative blood transfusion, and hospital stay. Pathologic characteristics included tumor size, French Federation of Cancer Centres (FNCLCC) grade,14 Union Internationale Contre Cancer (UICC) margin,15 lymph nodes invasiveness (LNI), vascular involvement (VI), and perineural involvement (PNI).

Treatment modalities included the surgical procedures, radiation therapy, and neoadjuvant/adjuvant chemotherapy. All patients underwent surgical treatment, the surgical type included radical resection (n = 179) and palliative surgery (n = 386). Radical resection is an intraoperative evaluation which refers to an extensive en bloc resection of the tumor and surrounding tissues, as well as lymph drainage and vascular structures. Patients with operable diseases (n = 132) were assigned to radical resection, patients with inoperable lesions were assigned either to neoadjuvant chemotherapy (n = 156) or palliative surgery (n = 277). Neoadjuvant chemotherapy was based on 5-Fu in the last decade and gemcitabine in this decade, including CF, CDDP, and paclitaxel, single dose or combined for 2 to 4 cycles. After neoadjuvant chemotherapy, the lesions were reassessed for resectability by surgical oncologists (JH H and XS H) with comparable pre- and post-neoadjuvant chemotherapy, computed tomography, or magnetic resonance imaging. After the reassessment, patients with operable lesions were assigned to radical resection (n = 47) with postoperative chemotherapy of same regimen; patients with inoperable lesions (n = 109) were treated with palliative procedures. All the pre- and post-adjuvant chemotherapy images were reviewed to identify borderline resectable lesions; their actual treatments depending on patients’ preference were recorded.

SPSS 13.0 statistical software (SPSS, Chicago, IL, USA) was used for statistical analysis, the following variables were considered for their prognostic value: age at diagnosis, sex, tumor site, tumor size, AJCC stage, surgical type, blood transfusion, hospital stay, UICC margin, FNCLCC grade, chemotherapy, PNI, and vascular involvement. Univariate and multivariate analyses were carried out for overall survival (OS). Survival curves were computed by the Kaplan–Meier16 method and compared by the log-rank test. Multivariate analyses based on the stepwise Cox17 proportional hazards model were used to identify the most significant factors related to outcome. A stepwise forward selection procedure was used and a significance level of 5% was chosen as the criterion for entering factors in the multivariate model.

Results

Clinical, Pathologic, and Treatment Variable

All 565 patients underwent surgical treatment for radical or palliative purpose, clinical, pathologic, and treatment variables and their distribution for decades are listed in Table 1. The median follow-up for survivors was 6 months (range, 0–156 months). There were 359 (63.5%) males and 206 (36.5%) females with a median age of 61.0 years (range, 24–88 years). The median tumor size was 4.0 cm (range, 0.8–18 cm). At the time of diagnosis, the tumor stages according to AJCC staging system were as follows: I, 77 patients; II, 172 patients; III, 136 patients, and IV, 176 patients. Local invasiveness of tumor were confirmed by intraoperative exploration or postoperative pathological examination, 166 patients had vascular involvement, either in the artery or vein or both. Ninety-three patients had lymph node involvement and 34 patients had PNI.

Table 1 Relationship between decades and clinical, pathologic, treatment parameters
Full size table

Carbohydrate cantigen 19-9 (CA19-9) was investigated with a cutoff level of 30 U/L in 463 patients. Two hundred sixty-four patients had a positive CA19-9 with a sensitivity of 57.0%. In those with both CA19-9 and tumor size records (n = 289), CA19-9 was positive in 170 patients, and most of them (163/170, 95.9%) had a primary tumor of over 2 cm. Of 552 patients that had a preoperative bilirium test, patients with preoperative bilirium of less than 200 U/L, between 200 and 300 U/L, and over 300 U/L were 372 (372/552, 67.4%), 85 (85/552, 15.4%), and 95 (95/552, 17.2%), respectively.

Palliative resection and exploratory laparotomy were assigned to be R2 resections. Of the 179 patients submitted to radical resection, resection was grossly complete (R0/R1) in 171 patients (94.4%), seven patients (10%) had a tumor rupture intraoperatively (R2), and one patient had missing data. Forty-eight of the 179 patients had undergone a vascular resection and reconstruction; 28 patients had extended lymphadenectomy.

Neoadjuvant chemotherapy was performed in 156 patients with locally advanced or disseminated diseases, after neoadjuvant treatment, 30.1% (47/156) of them underwent radical resection. One hundred eight-four (184/565, 32.6%) patients were treated with adjuvant chemotherapy. Adjuvant chemotherapy was administered after radical resection, palliative resection, and exploratory laparotomy in 82, 73, and 29 patients, respectively. Chemotherapy agents included 5-Fu, CF, CDDP, gemcitabine, and HCPT, single or combined for at least 2 cycles. Fourteen patients underwent postoperative radiotherapy.

Of the 565 patients, 164 underwent perioperative blood transfusion of a mean amount of 1.4 U. Blood transfusion was required in 45.3% (81/179) of patients who underwent radical resection, while this number decreased to 31.5% in patients with other procedures. All the 565 patients had a mean hospital stay of 29.6 days, and expectantly, patients with radical resections had a significantly longer hospital stay than those with non-radical resection (P = 0.002, mean 31.6 days vs 28.7 days).

A Decade-Based Analysis of Early Diagnosis and its Impact on Outcomes

Of all 565 patients, 2-year and 5-year overall survival was 13.5% and 3.6%, respectively. AJCC stage was associated with OS, 2-year and 5-year OS was 38.0% and 8.8% for stage I, 13.2% and 2.6% for stage II, 8.1% and 1.6% for stage III, and 6.7% and 0% for stage IV, respectively (P < 0.001). More patients (208/460, 45.2%) operated on for PDAC in this decade had a stage I/II disease at diagnosis than those (44/105, 40.3%) in the last decade. Furthermore, fewer patients (119/460, 26.1%) operated on for PDAC in this decade were at their terminal stage compared to those (57/105, 54.3%) in the last decade (P < 0.001) (Table 1). Patients with stage I/II PDAC in this decade had a significantly longer median survival of 14 months compared to those of 4 months in last decade (P < 0.001). Accordingly, patients with stage III/IV PDAC in this decade had a significantly longer median survival of 6 months compared to those of 3 months in last decade (P < 0.001) (Fig. 1a).

Fig. 1
figure 1

The curves of overall survival in patients according to clinical and pathological parameters. a tumor site; b AJCC stage; c FNCLCC grade; d UICC margin; e lymph node involvement; f vascular involvement

Full size image

A Decade-Based Analysis of Clinical and Pathological Variables Predictive of Outcome

Patients of the two decades had similar tumor sites (P = 0.203), sizes (P = 0.085), grades (P = 0.701), stages (P = 0.570), PNIs (P = 0.619), VIs (P = 0.649), and LNIs (P = 0.328). Tumor site was an important prognostic factor (Fig. 1b). The 2-year and 5-year OS was 31.4% and 12.6% in patients with a tumor in pancreas body and tail, which was significantly different from 11.2% and 2.0% in those with a tumor in the pancreas head (P < 0.001) (Table 2). Of all 565 patients, FNCLCC grade was a strong prognostic factor for OS (Fig. 1c), 2-year and 5-year OS was 24.3% and 4.0% for grade 1, 19.4% and 0% for grade 2, and 10.8% and 3.6% for grade 3, respectively (P = 0.003) (Table 2). PDAC patients with normal CA19-9 had a poorer prognosis. Two-year and 5-year OS was 18.5% and 4.4% in patients with positive CA19-9 (n = 264) compared to 9.7% and 2.2% in those with normal CA19-9 (n = 172), respectively (P = 0.001). But in this decade, CA19-9 did not seem to be significantly associated with prognosis. Median survival in 260 patients with positive CA19-9 was 10 months, which was similar (P = 0.694) to 11 months in 102 patients with normal CA19-9, indicating that a more sensitive cutoff value should be investigated for prognostic purpose.

Table 2 Univariate and multivariate analysis of variable factors for OS
Full size table

More patients underwent operation for PDAC in this decade (150/453, 33.1%) had a R0/R1 resection than those in the last decade (21/104, 20.2%, P = 0.010). R0/R1 resection was significantly associated with improved OS compared to R2 resection, 2-year and 5-year OS in patients with R0/R1 margin was 29.6% and 6.1%, while in patients with R2 resection, they were 6.7% and 2.6%, respectively (P < 0.001) (Fig. 1d).

We further identified that LNI was also associated with poorer prognosis in PDAC patients (Fig. 1e), as patients without LNI had a median survival of 11 months, which was significantly longer than LNI patients of 5 months (P = 0.005). PNI was not associated with OS (P = 0.803) as indicated in Table 2, but VI was significantly associated with poor prognosis in PDAC patients (Fig. 1f). In 400 patients who had detailed information of vascular status, median OS in patients with VI (n = 166) was 9 months, compared to 12 months in those without (n = 234, P = 0.013).

Blood transfusion was recognized as a prognostic factor for PDAC,18 though not universally accepted.19,20 We next investigated blood transfusion in the 565 patients. For all surgical procedures, blood transfusion was required in 43.8% (46/105) of patients in the last decade with a mean amount of 2.4 U, but it fell to 25.7% (118/420) and 1.2 U in this decade (Table 1). Though blood transfusion was not prognostic for survival in all patients (Table 2), it was of prognostic value in the subset of patients who underwent radical resection (P = 0.032). Decade-based analysis indicated that requirement of blood transfusion for radical resection was significantly higher (P = 0.001) in the last decade (77.3%, 17/22) than that in this decade (40.7%, 64/157).

Hospital stay reflects the capacity of multidisciplinary care in specialized high-volume centers. The mean hospital stay in the last decade was significantly longer than that in this decade (mean 32.5 days vs 29.0 days, P = 0.002). Patients with radical resection had a significantly longer hospital stay than those with other procedures, indicating reasonably that more aggressive surgeries necessitate longer postoperative recovery.

A Decade-Based Analysis of Treatment Variables Predictive of Outcome

Of the 565 patients, more patients (157/460, 34.1%) operated for PDAC in this decade underwent a radical resection than those in last decade (22/105, 21.0%, P = 0.014). Surgery type was significantly associated with OS (Fig. 2a); 2-year and 5-year OS in patients with radical resection was 29.8% and 5.5%, while in patients with non-radical resection, they were 6.4% and 2.7%, respectively (P < 0.001). Neoadjuvant chemotherapy increased the rate of radical resection in patients with inoperable diseases (47/156 vs 8/277, P < 0.001), and this increase was even more dominant in this decade, 36.2% (46/127) of the patients who underwent neoadjuvant chemotherapy for their inoperable diseases were able to have radical resection, indicating that the emergency of first-line gemcitabine for PDAC might have contributed to this improvement.

Fig. 2
figure 2

The curves of overall survival in patients according to treatment variables. a surgery type; b adjuvant chemotherapy; c vascular reconstruction; d extended lymphadenectomy

Full size image

Similarly, of all 565 patients, more patients in this decade (171/460, 37.2%) underwent adjuvant chemotherapy than those in the last decade (13/105, 12.4%, P < 0.001). Adjuvant chemotherapy was significantly associated with improved OS in PDAC patients; 2-year and 5-year OS in patients with chemotherapy was 24.8% and 4.7%, while in those without, they were 7.7% and 3.1%, respectively (P < 0.001, Fig. 2b). Vascular resection and reconstruction performed for the purpose of radical resection mostly occurred in this decade (1/31 vs 47/135, P < 0.001), with a single artery reconstruction in five, single venous reconstruction in 17, and both in 26. Of the 176 patients with radical resection who had a record of VI status, vascular reconstruction had no significant impact on survival; 2-year and 5-year OS was 34.9% and 7.8% in patient with vascular reconstruction, which was similar to 27.0% and 8.4% in those without (P = 0.686). But in all patients with VI, vascular reconstruction was significantly associated with improved OS, as patients with vascular reconstruction had a 2.7-fold decreased risk of death compared to those without (95% CI: 1.786–4.114, P < 0.001, Fig. 2c). Twenty-eight patients underwent extended lymphadenectomy because of LNI, this was performed more in this decade (24/71, 33.8%) than in last decade (4/22, 18.2%). Of all 93 patients who had an LNI, extended lymphadenectomy was significantly associated with improved OS (Fig. 2d) and patients with extended lymphadenectomy had a 2.9-fold (95% CI: 1.690–5.066, P < 0.001) of decreased risk of death compared to those without.

Discussion

Pancreatic adenocarcinoma is one of the leading causes of death due to cancer, with the lowest survival rate.21 Difficulty in diagnosing the disease at an early stage is the major obstacle for improving the outcome in PDAC patients.22 Current methods for early diagnosis are often ineffective and/or inaccurate, especially in identifying smaller, potentially curable lesions.23 In our study, we detected that in patients with tumor size no more than 3 cm, the sensitivity of CA19-9 was only 57.0%, and its positive accuracy was also limited (63/93, 67.7%). Though it was a little higher than Steinberg’s reports of 55% 20 years ago,24 these unsatisfactory results indicated the limited predictive value of CA19-9 in detecting PDAC at early stage cancer, and necessitated other biomarkers and genetic markers to be used as a panel in combination with other modalities for early detection with high sensitivity and specificity in PDAC patients.25

Postoperative margin status in PDAC patients has been reported by various authors to be a prognostic factor associated with survival.26–29 In our study, patients with R2 resection had a 1.60-fold of increased risk of death compared to those with R0/R1 resection (P < 0.001). We concluded here that the grossly total resection of tumor should be the primary goal in attempting a resection. R0/R1 resection was more common in this decade, indicating that more patients were resectable at diagnosis and that surgical techniques have improved in this high-volume center. Furthermore, patients with vascular involvement in our study benefited from vascular reconstruction, which was performed for the purpose of radical resection without extra morbidity. Some outcomes were satisfactorily extended, as vascular reconstruction had improved radical resection rate in patients with vascular involvement and subsequently improved survival. This is in accordance with previous reports that vascular reconstruction might provide a negative surgical margin and better survival in patients with locally advanced pancreatic cancer.30,31 Similarly, more vascular reconstruction was performed in this decade, indicating that the procedures are complicated challenges to surgical oncologists, who should be surgically skilled in high-volume center with experience in perioperative care.

With awareness of surgical techniques and their impact on survival, surgical oncologists have improved survival in PDAC patients significantly in recent years,32,33 but inoperable diseases remain to be the main cause of mortality. In our study, 47 patients were downstaged from inoperable PDAC to radical resections, and most of the downstage occurred to gemcitabine-based neoadjuvant chemotherapy in this decade. With increasing data supporting neoadjuvant chemotherapy in borderline resectable PDAC patients,34,35 we now routinely recommend gemcitabine-based regimen for patients with inoperable PDACs.

Unfortunately, even an R0 resection will not guarantee long-term survival, and many patients will eventually have local and/or system failure and ultimately die of disease progression.36,37 Therefore, the use of adjuvant therapy is a logical strategy for systemic disease and survival. Gemcitabine has proven to be essential in the management of advanced pancreatic cancer for clinical benefit and OS improvement in 1997.38 Updated data from the CONKO-001 trial definitively ascertain the value of adjuvant chemotherapy and provide level 1 evidence supporting the use of adjuvant chemotherapy for patients with resected PDA.11 In our study, median survival in patients with adjuvant chemotherapy was significantly longer than that in patients without (15 months vs 6 months, P < 0.001). More patients underwent chemotherapy in this decade, reflecting that adjuvant chemotherapy has been recognized as an important component in treatment of PDA in this high-volume center.

The late presentation and poor response to chemotherapy has been the main obstacle in improving the outcome of PDAC patients. Sensitive and specific tumor markers are needed for patients to be diagnosed at an early, operable disease stage. To this point, CA glycoproteins are less satisfactory because of limited sensitivity and false-positives. Newly investigated genetic markers are still undergoing evaluation. Efficient evaluation of new markers of pancreatic neoplasia will benefit from recently completed pancreatic cancer genome39 and require greater enrollment of patients with suspected pancreatic disease into clinical trials. These developments in early diagnosis of PDAC in this decade will undoubtedly increase the proportion of early disease and result in improved outcome in the coming decades. Also, with multiple genetic mutations being identified as the precursor to the development of pancreatic cancers and recognized as the target for therapeutic interventions,40 with multiple immunotherapy and targeted therapy being investigated with encouraging results in different clinical trials,41–44 and with high-volume centers being built worldwide, we believe that the results of treatment and outcome in PDAC patients in the next decade will be even more encouraging.