Introduction

Over the past few decades, resectability of pancreatic cancer has extended due to improved surgical morbidity and mortality, advances in imaging methods, and greater efficacy of new chemotherapy regimens in the treatment of advanced pancreatic cancer¹. In the eighties, when our experience began, we only resected tumors with a clear flat plane between the tumor and the mesenteric vessels. Later, in the nineties we went a step further and indicated resection in patients with isolated mesentericoportal axis involvement and absence of contact between the tumor and the superior mesenteric artery. Most of these patients required vascular resections^2,3. In the two thousands, treatment with chemotherapy or chemoradiotherapy before surgery (neoadjuvant therapy) allowed us to extend the indication of tumor resection by including patients in whom the tumor is in contact with the superior mesenteric artery (borderline resectable pancreatic cancer).

The aim of this study is to evaluate morbidity and mortality of pancreatectomy in a series of patients with borderline resectable pancreatic cancer previously treated with neoadjuvant therapy.

Material and methods

Between 2012 and 2018, 24 patients with borderline resectable pancreatic cancer previously treated with neoadjuvant therapy with no evidence of disease progression after restaging were evaluated. Twenty-two patients underwent surgical exploration while 2 refused surgery. Disease progression was confirmed in 7 patients, 3 with liver metastases (disseminated disease) and 4 with a tumor firmly attached to the superior mesenteric artery (locally advanced disease). The remaining 15 patients constitute our study population; 11 of them underwent pancreaticoduodenectomy and 4 underwent left pancreatectomy.

The following variables were evaluated: epidemiological data, preoperative comorbidities according to the American Society of Anesthesiologists (ASA) physical status classification, type of borderline resectability (arterial or venous involvement) according to the National Comprehensive Cancer Network (NCCN)⁴, type of neoadjuvant therapy, radiological response after neoadujuvant therapy based on RECIST 1.1 criteria and surgical data (operative time, type of vascular resection, and transfusions). The evaluation of the surgical specimen and pathologic response after neoadjuvant therapy was performed following the standards of the College of American Pathologists based on cell viability. This protocol classifies the response to neoadjuvant therapy as: grade 0: complete response, no viable cancer cells; grade 1: marked response, minimal residual cancer (simple/small groups of cancer cells); grade 2: moderate response, residual cancer outgrown by fibrosis; grade 3: poor/no response: extensive residual cancer.

Postoperative complications were considered following the Clavien-Dindo classification. Specific mortality was classified according to the International Study Group for Pancreatic Surgery (ISGPS)⁵. In-hospital mortality and mean survival were also recorded.

These 15 patients who underwent pancreatectomy after neoadjuvant therapy were compared with a cohort of 15 patients with resectable pancreatic cancer who underwent pancreatectomy and did not require previous neoadjuvant therapy over the same period (2012-2018). Patients were matched in a 1:1 ratio by age, sex, comorbidities according to the ASA physical status classification, cancer location (head/ tail/body) and procedure (pancreaticoduodenectomy, distal pancreatrectomy). The selection process was randomized, and the investigator was blinded to the other clinical and radiological data and outcomes.

Statistical analysis: The Wilcoxon signed-rank test was used to establish statistically significant differences between continuous variables. Discrete variables were analyzed using the chi-square test with Yates correction. A p value < 0.05 was considered statistically significant. The Kaplan-Meier curves were used to calculate mean survival. Survival between both groups was compared using log-rank test. Both survival and follow-up were considered from surgery to death or to the last visit. Mortality due to any cause was included in survival analysis. Those differences with a p value < 0.05 were considered statistically significant.

Results

The epidemiological data of the 15 patients are detailed in Table 1. The head of pancreas was the most common location in 11 patients, followed by the body in 4. According to the NCCN classification of borderline cancer, 7 patients had tumor contact < 180° with the superior mesenteric artery while 8 only had contact with the mesentericoportal axis (> 180° in 5 and < 180° in 3). The chemotherapy regime FOLFIRINOX (folinic acid, fluorouracil, irinotecan clorhidrate and oxiplatin) was used in 6 patients, followed by a combination of gemcitabine and nab-paclitaxel in 5. The remaining 4 patients received gemcitabine alone (3 patients) and gemcitabine combined with capacitabine in 1 patient. Eight patients underwent neodjuvant chemotherapy plus radiation therapy.

The evaluation of the response to neoadjuvant therapy using RECIST 1.1 criteria showed stable disease in 10 patients and partial response in 4. In the remaining patient, the response could not be evaluated due to poor quality of the imaging tests before neoadjuvant therapy. Mean CA 19-9 levels were 333 ± 248 U/mL before neoadjuvant therapy and significantly decreased to 147 ± 106 U/mL (p < 0.01) before surgery. Mean operative time was 393 minutes (range 260 to 440).

There were no significant differences in opera tive time between patients with and without radiation therapy (361 vs. 364 minutes, p = 0.87). Three patients required transfusions. Inframesocolic superior mesen teric artery first approach was used in 5 patients and supramesocolic superior mesenteric artery first ap proach was used in 10. The supramesocolic approach was anterior in 5 patients, posterior in 3 and lateral in 2. Ten patients (66%) required some type of vascular re section. Of the patients undergoing pancreaticoduode nectomy, 7 (64%) resections of the mesentericoportal axis were performed; 5 patients underwent segmental resection and 2 underwent lateral resection. Segmental resections were reconstructed with end-to-end anasto mosis in 4 patients and with autologous internal jugu lar vein graft in the remaining patient. Among patients with left pancreatectomy, 3 (75%) vascular resections were performed, 2 patients underwent lateral resec tion of the mesentericoportal axis, and the celiac trunk was resected in 1 patient.

Nine patients (60%) developed at least one complication of the Clavien-Dindo classification: grade I, 1 patient; grade II, 3 patients; grade IIIb, 3 patients; grade IVa, 2 patients. The specific complications are detailed in Table 2. Four patients required reoperations, 2 for abdominal bleeding, 1 for sepsis and abdominal infection and 1 for colonic necrosis. On pathological examination, resection margins were R0 in 11 patients and R1 in 4. Lymph nodes were negative in 8 patients. The following pathologic response based on cell viability was recorded: none of the patients presented complete response, 1 patient had marked response (grade 1) and 8 patients had moderate response (grade 2), while the response was poor (grade 3) in 6. No deaths were reported. Seven patients underwent adjuvant therapy after pancreatectomy and mean survival was 23.4 months.

In Table 1 the preoperative data of the 15 patients who received neoadjuvant therapy before pancreatectomy is compared with 15 patients who did not require neoadjuvant therapy before surgery. Vascular resections were significantly more common and operative time was significantly longer in patients with neoadjuvant therapy. There were no significant differences in tumor size, and the R0 resection rate and negative lymph nodes were non significantly greater in patients with neoadjuvant therapy.

The specific morbidity in patients with and without neoadjuvant therapy is compared in Table 2. The number of complications was similar and there were no deaths in both groups; mean survival was also similar (Table 3).

Discussion

Neoadjuvant therapy in patients with borderline pancreatic cancer allows selecting a group of patients with advanced stages for surgical resection. In our series, morbidity and survival in patients undergoing pancreatectomy after neoadjuvant therapy were similar to those of patients undergoing pancreatectomy with less advanced pancreatic cancer who did not require prior neoadjuvant therapy.

From a practical point of view, pancreatic cancer can be classified into localized and metastatic. In recent years, improvements in imaging techniques have allowed more accurate visualization of the anatomy between the tumor and the mesenteric vessels. Partial contact (abutment) is considered when the contact between the tumor and the vessel circumference is < 180°, and total contact (encasement) is considered when it is > 180° (Fig. 1). According to this information, localized pancreatic cancer is classified in resectable, borderline resectable and locally advanced⁵. A tumor is resectable when there is no contact with the mesenteric vessels (Fig. 2 A). Locally advanced cancer presents tumor encasement of the superior mesenteric artery, celiac trunk, hepatic artery or superior mesenteric vein with absence of distal end for venous reconstruction (Fig. 2 D). Borderline resectable tumors fall between resectable and locally advanced cases. These tumors have partial or total contact with the superior mesenteric vein or portal vein with distal end for venous reconstruction or partial contact with the superior mesenteric artery, liver artery or celiac trunk (Fig. 2 B and C).

Borderline resectable tumors include a spectrum of presentations ranging from isolated partial contact with the superior mesenteric vein to partial contact with the superior mesenteric artery. There are many classifications for borderline resectable tumors: MD Anderson Cancer Center (MDACC)⁶, Americas Hepato-Pancreato-Biliary Association/Society of Surgical Oncology/Society for Surgery of the Alimentary Tract (AHPBA/SSO/SSAT)⁷, National Comprehensive Cancer Network NCNN4, and International Study Group of Pancreatic Surgery (ISGPS)⁸ are the most commonly used (Table 4). All these classifications agree that the partial contact of the tumor with the superior mesenteric artery is the upper limit of invasion to include patients as borderline. Yet, they differ in the degree of contact with the superior mesenteric vein. While the MD Anderson classifies borderline tumors as those with full contact with the superior mesenteric vein or portal vein, which are therefore more advanced tumors, the NCCN and AHPB classifications consider borderline tumors as those with partial contact with the mesentericoportal axis, which are less advanced tumors. This lack of agreement on the definition of borderline tumors hinders the comparison between series and the correct evaluation of the efficacy of the different types of neoadjuvant treatment.

Surgical resection of borderline resectable tumors is technically possible; however, in view of the high risk of R1 resection margins, the current recommendation is neoadjuvant therapy before surgery⁹. The goals of neoadjuvant therapy in the treatment of borderline tumors are local control, treatment of occult micrometastases and patient selection^10,11. Local control increases R0 resection rate and treatment of occult micrometastases reduces systemic recurrence rate. The period of neoadjuvant therapy, which varies between 3 and 6 months, allows the selection of patients who have not developed systemic disease and are therefore the most suitable for major resection surgery. There is no consensus on which protocol is the best for neoadjuvant therapy. The choice depends on the experience of the institutions and the availability of radiation therapy. As in most published series, the most common chemotherapy regimens used in our series are FOLFIRINOX and the combination of gemcitabine/nab-placitaxel^12,13.

The response to neoadjuvant therapy is evaluated with three parameters: imaging and laboratory tests and physical status^14,15. Imaging methods have low sensitivity and specificity for assessing resectability, as they cannot differentiate between residual neoplastic tissue or scar tissue and fibrosis secondary to neoadjuvant therapy¹⁶. Initial CT scan provides the necessary information to assess tumor resectability. In general, tumor downsizing or downstaging do not occur after neoadjuvant therapy¹⁷. Tumor staging after neoadjuvant therapy did not change in 67% of the patients in our series. Only 4 patients presented partial response (Fig. 3). CA 19-9 is the most useful marker. A reduction in CA 19-9 levels after neoadjuvant therapy was associated with better resectability and higher survival. Nevertheless, the absence of reduction in CA 19-9 levels does not exclude patients from surgical exploration¹⁸. Preservation of good physical status and weight and absence of pain during neoadjuvant therapy is a reliable indicator of resectability. In summary, the absence of disease progression on imaging tests, a decrease in CA 19-9 levels and a good physical status after neoadjuvant treatment are the best indicators for selecting patients for surgical resection.

In patients selected for surgical exploration, once disseminated disease has been ruled out, resectability varies according to the surgeon’s attitude towards advanced pancreatic cancer. Pancreatectomy after neoadjuvant therapy requires a surgical team with strong determination, as the R0 resection rate is closely related with the surgical technique. These resections are directed to the mesenteric vessels and one of the first maneuvers is to rule out invasion of the superior mesenteric artery, known as superior mesenteric artery first approach. The adequate dissection and separation of the soft tissue on the right side of the superior mesenteric artery is one of the most important surgical steps to achieve an R0 resection. In the East, this technique is called Tora no Ana Strategy, which comes from a Han Dynasty proverb that says that you cannot get the tiger’s cubs without entering the tiger’s den¹⁹. Speaking of pancreatic surgery, this means an R0 resection is not possible without exploring the superior mesenteric artery.

Interestingly, despite the complexity of pancreatectomy after neoadjuvant therapy, surgical morbidity and mortality are similar to those of pancreatectomies without neoadjuvant therapy. In our series, as in several publications, after matching for age, sex and ASA classification, morbidity and mortality of pancreatectomy after neoadjuvant therapy was not significantly different in patients undergoing pancreatectomy without neoadjuvant therapy. This could be explained by the fact that most of the series come from centers with high volume of pancreatic surgeries²⁰. These results may not be extrapolated to centers with low volume of pancreatic surgeries.

The retrospective nature and the small sample size are the main limitations of this study. The diversity of the health care system in our country hinders the implementation of treatment protocols and the enrollment of a higher number of patients. In addition, only patients with good response to neoadjuvant therapy were included which were therefore the most suitable for resection and with the best chances of survival. Finally, the low surgical morbidity and mortality may have been influenced by our previous experience in pancreatectomy and vascular resections.

In conclusion, neoadjuvant therapy of borderline resectable pancreatic cancer allows selecting a group of patients with advanced cancer in whom pancreatectomy results in a survival rate similar to that of patients undergoing pancreatectomy without the need for neoadjuvant therapy. High commitment and previous experience in pancreatectomy and vascular resections are essential to obtain R0 resections. This complex procedure should be performed in centers with a high volume of pancreatic surgeries to minimize operative morbidity and mortality.