6. Surgical Therapy – Pancreatic Cancer

Background

Surgical therapy is the only potentially curative treatment option for pancreatic cancer. Chemotherapy and/or radiotherapy alone cannot achieve a cure for this tumor. A randomized trial demonstrated the inferiority of radiochemotherapy compared with surgery for locally invasive pancreatic cancer classified as resectable [235]. Based on an interim analysis showing a benefit of surgery over radiochemotherapy in median overall survival (22.6 months, 95% CI 12.5 – 32.7 vs. 10.8 months, 95% CI 8.8 – 12.7), the trial was stopped early. Nevertheless, the differences in median overall survival and 3-year survival were statistically significant.

Background

Infections after ERCP and stenting are seen in up to 73% of cases [261]. Infections of the biliary system at the time of pancreatic resection are associated with increased morbidity [262], [263], [264], [265]. In numerous small studies and one larger randomized clinical trial, the rate of severe complications has been shown to be significantly lower with early surgery without biliary drainage than with primary biliary drainage and later surgery (39% severe complications without drainage vs. 74% with drainage [266]) (see IMEBI evidence report at www.leitlinienprogramm-onkologie.de). Therefore, in the absence of cholangitis, biliary drainage is not indicated if surgery is carried out promptly. However, if cholangitis is present, interventional biliary drainage should be carried out quickly to avoid complications of infection.

Background

There is no evidence-based definition of the term „promptly“. Expert consensus is a maximum period of 2 weeks. Optimal interdisciplinary consultation is important to minimize delays until actual surgery once a decision to operate has been made. A planned delay in the timing of surgery with indication for stent placement in cholestasis arises, for example, if surgery is not performed immediately due to the implementation of neoadjuvant therapy.

Background

In PDAC, there is an early potential for metastasis both locally in lymph nodes and peritoneally and hepatically. Metastasis to the peritoneum and/or liver should therefore be ruled out before a laparotomy with the aim of R0 resection. However, occult metastases can remain hidden on imaging [267], [268].

According to the International Association of Pancreatology (IAP) consensus and the ABC criteria for resectability [271], diagnostic laparoscopy is recommended before laparotomy if the tumor size is > 3 cm or the CA 19-9 value is > 500 U/ml or ascites is present (if a hepatic or portal-venous cause is excluded), especially if no clear organ metastasis is visible on imaging (CT/MRI/MRCP/PET-CT). As far as tumor size is concerned, studies have shown that a tumor size of > 3 cm in combinationwith unintentional weight loss of more than 5 kg increases the risk of organ metastasis  to 30% [Ta, R et al. 2019].

In up to 20% of cases, diagnostic laparoscopy can detect previously occult metastasis to the peritoneum and/or liver, which remain hidden on imaging due to their size. This leads to the avoidance of an exploratory laparotomy with a subsequent delay in chemotherapy/multimodal therapy.

Diagnostic laparoscopy with intraoperative ultrasound is possible if the tumor has been unsuccessful to date [269]. As part of the laparoscopic evaluation, a punch for histological tumor confirmation can also be obtained from metastases and the primary [Ta, R et al. 2019]. This will also enable NGS analysis in the future.

Laparoscopic evaluation is a safe and low-complication procedure with a complication rate of 0-0.8% [267].
In addition to obtaining tumor tissue to confirm the diagnosis, e.g. for neoadjuvant treatment strategies, diagnostic laparoscopy can also be used to provide an indication for simultaneous cholecystectomy if the gallbladder is abnormal and an intravenous port can be inserted for systemic CTx under anesthesia.

Background

R0 gives the patient the greatest chance of long-term survival [272], [273], [275]. However, there is no uniform definition of what is meant by a „resection in healthy tissue“ (in sano resection) in pancreatic cancer. Thus, there is currently no uniform literature recommendation regarding a minimal resection margin to be observed, or its prognostic significance. Therefore, the goal is curative resection with the greatest possible safety margin. This should be stated in the histopathological findings.

The prognostic significance of the R-classification of resected pancreatic carcinomas is currently interpreted differently in the literature. A closer look at studies dealing with the question of whether curative resection is prognostically significant reveals that the R classification is not applied correctly or uniformly in individual studies. This circumstance certainly contributes to the fact that in the literature the rate of R1 resections is reported with a wide variation – between 16% and 75% – [276] . In general, the R classification must be applied uniformly. One can only speak of an R0 situation if no carcinoma cells are definitely detectable microscopically at the resection margin. R1 means that carcinoma cells are found at the definitive resection margin. In some publications, this definition is not adhered to – the authors speak of a curative (R0 resection only if the tumor is 0.1 cm from the definitive resection margin. In general, the further the tumor is from the resection margin, the better the prognosis. However, classification systems (R-classification) must be strictly separated from putative prognostic issues (distance from resection margin).

Some authors also evaluate lymph node metastases located at the resection margin as an R1 situation. It is not always specified which resection margin is meant. In the literature, there is also no uniform nomenclature for the resection margins of a partial pancreaticoduodenectomy to date.

Due to the tumor-biologic and thus morphologic peculiarities of pancreatic ductal adenocarcinomas with discontinuous spread, desmoplastic stromal reaction, and mostly extensive perineural or lymphatic vessel invasion, the exact assessment of a resection margin can be difficult [274], [277], [278], [279], [280], [281], [282], [283], [284].

In order to achieve a valid, reproducible, and unambiguous diagnosis, a standardized histopathological workup of the relevant resection margins is recommended. Furthermore, a conceptual extension of the R-classification (the so-called CRM concept, see below) is proposed.

The recommendations in this regard can be found in the sections below.

Background

In the absence of multimodality therapy concepts, synchronous resections of the primary tumor and liver metastases show no survival benefit compared to palliative bypass surgery, albeit with comparable morbidity and mortality [338], [339]. As early as 2008, a systematic review provided evidence that patients with simultaneous resection of the primary tumor and liver metastases could achieve comparable survival to resected patients without distant metastases with appropriate patient selection [329]. This suggestion is supported by other retrospective studies: in a retrospective analysis, Shi et al. showed identical survival of resected patients with hepatic metastatic PDAC and nonmetastatic PDAC: 15.7 vs. 16.9 months, p<0.085. A single-center, retrospective study from South Korea in intraoperatively detected hepatic “surprise metastases” compared 35 resected vs. 35 unresected (bypass or exploration and trial excisions) patients: resected patients had longer surgery with higher blood loss, longer length of stay, and more postoperative complications. However, the survival of resected patients was significantly longer than that of non-resected patients - and this was independent of chemotherapy [332]. In summary, there are indications, mainly from retrospective analyses, that synchronous resection of the primary tumor and metastases detected only intraoperatively could yield better results with regard to overall survival. However, modern chemotherapy concepts are not considered and prospective clinical data from randomized trials are lacking. Therefore, in case of distant metastases that can be detected only intraoperatively, resection of the primary tumor should be omitted despite given resectability.

Background

There are no prospective studies of resection of metastases from pancreatic ductal carcinoma. The existing evidence is based on small retrospectively analyzed case series with unclear selection criteria [328], [353].

The presence of distant metastases significantly worsens the prognosis in pancreatic ductal carcinoma. Only in the last 10 years due to the establishment of multimodal therapeutic concepts - based on inductive chemotherapy with modern regimens such as FOLFIRINOX and gemcitabine/nab-paclitaxel – have case series with better outcomes been reported. For example, Crippa et al performed synchronous resection of the primary tumor and liver metastases in 11 of 127 chemotherapy-pretreated patients with hepatic metastatic PDAC and achieved a median survival of 39 months in this subgroup. Survival in unresected patients with chemotherapy alone was only 12 months, p<0.0001 [330] Mortality and morbidity of such simultaneous resections are not higher compared to standard oncologic pancreatic resections [330], [333]. According to a registry study, distant metastases occur most commonly in the liver (88%) and lungs (10%). The frequency of osseous metastases is 2%; cerebral metastases occur in 0.15% of cases. Thereby, the prognosis for isolated pulmonary and osseous distant metastases is better than for hepatic and cerebral metastases [354], [355], [356], [357], [358], especially if resection is performed. A systematic review of 13 retrospective studies reported survival of up to 121 months for resection of metachronous lung metastases and up to 40 months for resection of liver metastases in adequately selected patients [359]. However, prospective data from randomized trials and specific data on subgroups that may benefit from such an approach are also lacking.

Background

Antibiotic prophylaxis should be carried out regularly during hepatobiliary surgical procedures, as this significantly reduces postoperative infectious complications including wound infections (10-25% vs. 5%) [363], [360], [361]. Perioperative prophylaxis should not be continued postoperatively because there are no advantages, but the risk of complications - including the development of resistance and allergies - increases [362]. In case of cholangitis, antibiotic therapy is administered.

Background

The literature is heterogeneous regarding somatostatin prophylaxis: all studies have shown no differences regarding postoperative lethality [364], [365], [366], [367], [368], [369], [370]. However, a meta-analysis of 10 randomized trials with a total of 1,918 patients shows that somatostatin significantly reduces morbidity and pancreas-specific complications, including the frequency of biochemically detectable fistulas [371]. Further differentiation with respect to resection procedures in subgroup analyses again shows no advantages of somatostatin prophylaxis, so that currently the patient group that benefits from perioperative secretion inhibition with somatostatin cannot yet be identified.

When considering the value of laparoscopic or robotic surgery for pancreatic cancer, it is important to consider that the data situation remains different for interventions at different locations (pancreatic head carcinoma vs. pancreatic body/tail carcinoma located on the left side of the mesenteric-portal axis)  .

Minimally invasive (laparoscopic/robot-assisted) distal pancreatectomy

The clinical outcome quality of minimally invasive distal pancreatectomy was recently assessed in a multicenter randomized controlled clinical trial (LEOPARD) in the Netherlands. This patient-blinded study, conducted from 2015 to 2017 at 14 national centers, compared the time to functional recovery of patients who underwent either minimally invasive (n=51) or open (n=56) distal pancreatectomy for tumor confined to the left side of the pancreas without vascular involvement (LEOPARD; NTR5689). The primary end point, time to functional recovery, was significantly shorter at 4 days for minimally invasive pancreatectomy compared with 6 days for open pancreatectomy. The conversion rate was 8%. The overall complication rate after Clavien-Dindo > III was not significantly different, but fewer gastric emptying disorders and better quality of life without increasing costs were noted for minimally invasive distal pancreatectomy [471]. Surgical quality control took place before patients were included in the study because it involved only surgeons who had performed 50 complex minimally invasive GI procedures, > 20 distal pancreatectomies, and > 5 minimally invasive distal pancreatectomies in a training program [472].

The results of the Dutch Pancreatic Cancer Group data, (17 centers; 2005–2016) were then compared with the American College of Surgeons’ National Quality Improvement Program (ACS-NSQIP) database (88 centers; 2014–2016). In this international cohort study, serious 30-day morbidity, including mortality, was evaluated according to surgical procedure (either minimally invasive or open distal pancreatectomy). The composite endpoint of serious 30-day morbidity and mortality was defined as any of the following events within 30 days postoperatively: death, reoperation, pancreatic fistula, stroke, myocardial infarction, pulmonary embolism, septic shock, etc. Of 2,921 ACS-NSQIP patients, 1,562 (53%) received minimally invasive distal pancreatectomy with 18% conversion rate, and 1,359 (47%) received open distal pancreatectomy. Minimally invasive surgical technique was independently associated with reduced 30-day serious morbidity including mortality (see above for definition) for distal pancreatectomy [473].

In another retrospective analysis of the prospectively maintained database of the ACS-NSQIP, morbidity and mortality after minimally invasive (n=166; 33.1%) versus open (n=335; 66.9%) distal pancreatectomy with comparable preoperative comorbidity and pathologic staging were assessed in 501 patients with pancreatic cancer and distal pancreatectomy. Overall morbidity, transfusion rate, pneumonia rate, surgical wound infections, sepsis, and hospital length of stay were lower with minimally invasive distal pancreatectomy. Mortality, pancreatic fistulae, and gastric emptying disorders were comparable. Thus, short-term postoperative outcomes appear to be more favorable with minimally invasive distal pancreatectomy [474].

In a large population-based retrospective cohort study, a variety of perioperative factors were assessed for differences between minimally invasive and open distal pancreatectomy. A total of 8,575 open surgeries were compared with 382 minimally invasive distal pancreatectomies. There was a lower number of general perioperative complications (39.0% vs. 30.1%, p<0.001), especially postoperative bleeding (20.6% vs. 13.6%, p<0.001) and a shorter hospital stay with minimally invasive distal pancreatectomy [475].

Oncologic outcome after minimally invasive distal pancreatectomy for pancreatic cancer has been investigated in several retrospective studies. A monocenter analysis from a high-volume center (n=422 distal pancreatectomies from 2005-2014) analyzed overall survival and recurrence rates in 79 comparable patients with PDAC after laparoscopic (n=33) versus open (n=46) distal pancreatectomy. With comparable intraoperative and pathologic variables such as operative time, duct size, glandular texture, tumor size, type of pancreatic occlusion, number of lymph nodes removed, tumor stage, and R status at the resection margin, 1-, 3-, and 5-year survival and local recurrence or distant metastasis rates were similar in the two groups [476].

Another monocentric propensity score-matched analysis examined overall oncologic survival in a total of 1,947 patients with pancreatic cancer, each comparing 563 patients after laparoscopic distal pancreatectomy with the same number of patients after open surgical technique. There was no difference in 3-year survival between open and laparoscopic distal pancreatectomy (41.6% vs. 36.0%; hazard ratio 0.93, 95% CI 0.77–1.12; p=0.457). The start of adjuvant chemotherapy was identical at 50 days postoperatively in both groups, as was the number of lymph nodes resected (median 12 lymph nodes in both groups). In addition, there was no difference in 30- and 90-day mortality. The authors conclude that in terms of oncologic outcomes, a minimally invasive approach is an alternative to open surgery [470].

Robot-assisted surgical technique

Huang B. et al. performed a meta-analysis comparing robot-assisted with laparoscopic distal pancreatectomy in adult patients with malignant, borderline malignant, and also benign disease. Primary endpoints were conversions to open surgery, transfusion rate, splenic preservation, operative time, complications (pancreatic fistula) and length of hospital stay. A total of 9 studies with a total population of 1,167 patients were analyzed, including 929 patients with laparoscopic and 238 patients with robotic-assisted distal pancreatectomy. Overall, there was no significant difference between laparoscopic versus robotic-assisted distal pancreatectomy for any named endpoint. Nevertheless, despite the small number of robot-assisted distal pancreatectomies, the authors rated them as safe and effective compared with the laparoscopic surgical technique [477].

In another meta-analysis of all studies published until the end of 2017, the clinical outcome quality after robot-assisted pancreaticoduodenectomy as well as distal pancreatectomy compared to the respective open surgical technique was investigated. A total of 15 nonrandomized controlled trials with 3,690 patients (11 trials of robot-assisted vs. open pancreaticoduodenectomy, 4 trials of robot-assisted vs. open pancreatic left resection) were included in the analysis. There was no significant difference between robotic-assisted and open pancreaticoduodenectomy with regard to lymph node status as well as postoperative complications (pancreatic fistula, postoperative gastric emptying disorder), reoperation rate, hospital length of stay, and mortality. For robotic-assisted pancreaticoduodenectomy, the operative time was significantly longer, whereas blood loss and wound infections, as well as R1 status at the resection margin were significantly lower than those for the open surgical technique. Compared with open distal pancreatectomy, the robotic-assisted procedure resulted in a lower overall complication rate, lower blood loss, shorter hospital stay, but also a lower number of lymph nodes removed. There were no significant differences with regard to splenic preservation, R1 status, mortality, and especially pancreatic fistulae [478], [479].

In the German context, if laparoscopic resection of pancreatic cancer is carried out, it should be reported to the laparoscopic pancreatic surgery registry of the German Society for General and Visceral Surgery (http://www.dgav.de/studoq/weitere-register.html).

Statements on minimally invasive (laparoscopic/robot-assisted) surgical technique

One systematic review of papers published during the period 2000-2016 compared total robotic to open pancreatoduodenectomy for various conditions. A total of 13 non-randomized controlled trials with 692 robotic pancreatoduodenectomies were included in this review. The incidence of complications (biliary fistula, pancreatic fistula, postoperative hemorrhage), reoperations and mortality were comparable, although the results were not considered representative, particularly for complications, due to a lack of data from large series. The number of conversions (average 6.5–7.8%) to open surgery as well as the operative time decreased with increasing number of robotic pancreatoduodenectomy. The number of resected lymph nodes was comparable, but showed a lower percentage of R1 resections with robotic pancreatoduodenectomy compared with open surgery, but this may be due to selection bias. The authors concluded that in high-volume centers, robotic pancreatoduodenectomy is safely feasible. However, no data on long-term oncological survival as well as the cost-effectiveness of this surgical technique are available to date [480].

In a propensity score matched analysis on robotic-assisted vs. open pancreatoduodenectomies, it was shown that the incidence of clinically relevant pancreatic fistulas (POPF grade B/C) was lower in the robotic-assisted group than in open surgery (OR, 0.4; 95% CI 0.2-0.7; p=0.002). In all other secondary endpoints, robotic-assisted procedures and open surgery were comparable without significant differences (overall complication rate (73.7% vs. 66.4%; p=0.21), hospital stay (median: 8 vs. 8.5 days; p=0.31), 30-day readmission (22.4% vs. 21.7%; p>0.99), 90-day mortality (3.3% vs. 1.3%; p=0.38) [481]).

One systematic review with meta-analysis compared perioperative outcomes for laparoscopic versus open elective pancreatoduodenectomy in patients with benign or malignant pancreatic disease from 3 randomized controlled clinical trials with a total of 224 patients. The meta-analysis revealed a significant difference only in operative time in favor of open pancreatoduodenectomy and less blood loss with the laparoscopic surgical technique. All other primary and secondary outcome parameters were not significantly different with an overall low to moderate level of evidence. Based on these results, there is currently no advantage for laparoscopic pancreatoduodenectomy over the open surgical technique. When evaluating perioperative outcome quality, the learning curve in the respective medical technology as well as the increasing implementation of robotic surgery must be taken into account [482], [483], [484], [485].

Another systematic review with meta-analysis (12 studies; a total of 2,186 patients, 705 with minimally invasive and 1,481 with open pancreatoduodenectomy) on minimally invasive (robotic-assisted or laparoscopic) pancreatoduodenectomy compared with the open surgical method also showed no significant difference in overall morbidity. Only lower blood loss, fewer gastric emptying disorders, and shorter hospital stay after minimally invasive pancreatoduodenectomy were reported [486].

Only retrospective studies have been conducted on oncological outcome after minimally invasive pancreatoduodenectomy for pancreatic head carcinoma. One monocentric analysis investigated pathologic and oncologic outcome parameters in addition to perioperative outcome in a total of 1,623 minimally invasive pancreatoduodenectomies (1,458 laparoscopic, 165 robotic-assisted surgeries). It was shown that robot-assisted procedures were more likely to be carried out at high-volume centers and university-based institutions. There were no differences between laparoscopic and robot-assisted procedures with regard to the target parameters studied. Only the conversion rate was lower in the robot-assisted group (17.0% vs. 27.6%, p=0.003). The results in terms of resection status (R0/R1), number of lymph nodes examined, hospital length of stay, 90-day mortality, and median overall survival (laparoscopic 20.7 months vs. roto-assisted 22.7 months; log-rank p=0.445) showed no statistically significant difference [487].

One retrospective non-inferiority propensity scored-matched analysis was performed to investigate the impact of surgical method on R status after pancreatic head resections. Twenty robotic-assisted versus 24 open pancreatoduodenectomies were compared. The robotic-assisted group showed an R1 rate of 55.0% versus 41.7% in the open surgery group (p=0.38). Furthermore, no difference was found in the secondary endpoints (number of lymph nodes examined, blood transfusions, performance of adjuvant chemotherapy, overall survival, disease-free survival) [466].

Therefore, based on the above data, it can be summarized that at the present time laparoscopic/robotic pancreatoduodenectomy should be carried out only in the context of clinical trials or with comprehensive quality assurance (e.g., certification by professional societies) at designated centers. The topic of laparoscopic/robotic pancreatoduodenectomy will be revisited in the Living Guidelines on exocrine pancreatic cancer, taking into account the data available at that time, and appropriate recommendations will be issued.

Background

Due to the growth pattern of pancreatic ductal adenocarcinomas with strong fibrosis and so-called desmoplastic stromal reaction and the anatomy of the surgical site, the histopathological work-up of the circumferential resection margins can be difficult. The following resection margins require special workup (see also the diagram below in Figure 2):

• Medial resection margin: uncinate process (E) and vascular fossa (A./V. mes. sup.) (green).
• Dorsal resection area (former retropancreatic resection area) (blue)
• Ventral marking (if tumor is macroscopically localized ventrally).

The procedure should be coordinated with the surgeons, since the pathologist may have difficulties in locating the relevant structures, especially if a formalin-fixed specimen is sent.

Background

In the studies published on this question, intraoperative frozen section examination led to an increase in the rate of R0-resected pancreatic carcinomas [488], [489].  Secondary resection is suggested mainly if tumor cells are found at the parenchymal resection margin in frozen section. However, no studies could be identified whose results show a prolongation of survival based on the performance of a frozen section and, if necessary, subsequent secondary resection [489].

Background

In addition to the correct application of the internationally valid R-classification, uniform standards must apply to the histopathological workup of Whipple's resected pancreatic cancer. Special attention must be paid to the circumferential resection margin, since an R1 situation is particularly frequent in this case. Within the framework of a standardized histopathological workup, it is advisable to mark these resection margins in the macroscopic section, possibly in cooperation with the surgeon. The circumferential resection margin is composed of an anterior, medial, and posterior resection margin or the corresponding resection area. In addition, the posterior surface of the uncinate process requires special resection. The vascular axis is particularly important; furthermore, tumor-infiltrated lymph nodes in the superior mesenteric artery level must be included in the workup. It has proven useful in daily practice to mark these various resection margins in color, to examine them, and to identify the color markings in the histological section. Under the microscope, it is thus possible to measure exactly how far the tumor cells are from which resection margin. This measurement is added to the findings and thus communicated to the surgeon. To account for the fact that carcinoma cells relatively often approach this circumferential resection margin, a new concept is recommended in analogy to the classification of rectal cancer: the “concept of circumferential resection margin” (Circumferential Resection Margin, CRM). R0-resected pancreatic carcinomas are classified as CRM-positive in analogy to the CRM concept of the rectum if the distance of the tumor cells to the resection margin is less than 1 mm but does not reach this margin (the correct diagnosis is then: R0, CRM-positive = R0 narrow). If the carcinoma cells are more than 1 mm from the definitive resection margin, a “CRM-negative”, the classification is R0 (R0 wide, see Figure 3: Concept of circumferential resection margin (CRM) in pancreatic cancer). The question of which distance to the resection margin is actually of prognostic importance cannot currently be answered conclusively. In one large-scale study, it was shown that patients who underwent R0 resection and whose tumor cells were more than 1.5 mm from the resection margin had a better prognosis [490].

By applying the CRM concept with indication of the minimal distance of the carcinoma cells to the circumferential resection margin, data can be created for the first time that allow a conclusive evaluation of surgical treatment procedures for pancreatic cancer with the aim of better estimating the recurrence risk and prognosis [491], [492], [493], [494].

Background

UICC tumor stage is the most important prognostic parameter after resection of pancreatic cancer with curative intent.

The presence of lymph node metastases is a crucial prognostic parameter [499], [447], [449], [451], [452], [453], [454], [455], [458], [460], [462], [500], [501], [502]. At least twelve lymph nodes should be removed to perform correct pN0 categorization (see Recommendation 6.29) [453].

The lymph node ratio (number of lymph nodes affected to the number of lymph nodes examined) should be reported, as this ratio has prognostic significance (see Recommendation 6.30) [447], [460], [449], [451], [452], [454], [455], [456], [458], [462].

The grading (tumor differentiation grade) of pancreatic ductal adenocarcinoma is defined in the 2019 WHO classification [503] and is based on histological and cytological criteria. In addition to glandular differentiation, mucin production, mitotic count and nuclear atypia enter into the three-tier grading. If consistently applied, histologic differentiation grade correlates with prognosis [456], [460], [500], [504].

Regression grading is not currently established.

Background

For lymphatic vessel invasion, perineural sheath infiltration and blood vessel invasion, prognostic significance is documented in individual studies. However, data on the frequency of these parameters vary widely and require further validation [512], [491], [513], [499], [447], [448], [449], [450], [451], [452], [453], [454], [455], [456], [457], [458], [460], [462], [500], [501], [502], [504], [514], [515], [516], [517], [518], [519].