Introduction

Pancreatic surgery represents one of the most technically demanding abdominal procedures. During the last five decades, several advancements in surgical techniques, the introduction of new surgical platforms and a greater understanding of the biology of the various pancreatic lesions have led to a rapid expansion of the indications and modalities of pancreatic surgery [1,2,3,4]. The utilization of minimally invasive distal pancreatectomy (MIDP) is increasing, and several retrospective studies, metanalysis and one published randomized controlled trial reported on its safety and feasibility in mixed cohorts of benign and malignant pancreatic disease [5,6,7,8]. Nevertheless, debate remains regarding MIDP oncologic safety in the setting of pancreatic adenocarcinoma (PDAC) [9, 10]. The DIPLOMA study, a pan-European propensity score matched study, demonstrated comparable survival after MIDP and ODP for PDAC [5]. Nonetheless, van Hilst and al. observed a lower rate of lymph node retrieval and Gerota's fascia resection in the MIDP compared to the ODP. Importantly, in this study, the cohort of MIDP (356 cases) was chiefly represented by laparoscopic DP (LDP), while the robotic-assisted (RDP) cohort consisted of only a minority of cases (4%) [5].

In a recent meta-analysis, Zhou et al. conducted a comprehensive evaluation of RDP compared to ODP including seven studies conducted at high-volume robotic surgery centers and demonstrated an advantage for the former in terms of lower estimated blood loss and lower postoperative mortality rates without any major differences for operating time, number of lymph nodes harvested, positive margin rate, spleen preservation rate and severe morbidity rate between the two groups. [11] However, the studies selected in the meta-analysis by Zhou et al. reported on a mixed cohort of patients with benign and malignant pancreatic disease alike. [11] Therefore, a conclusion on the benefits and drawbacks of RDP compared to ODP in the setting of PDAC cannot be reached. Herein we present our institutional experience with RDP, LDP, and ODP in the setting of distal pancreatic adenocarcinoma and evaluate differences in perioperative and postoperative outcomes.

Methods

Retrospective review of a prospectively collected single institutional database of patients undergoing consecutive ODP, LDP, and RDP for left-sided PDAC between January 2008 and December 2019 at the University of Pittsburgh Medical Center (UPMC). Lesions classified preoperatively as T4 based on American Joint Committee on Cancer (AJCC) 8th edition [12] were excluded from the analysis to allow for a more homogeneous group of distal pancreatic adenocarcinoma and minimize the contamination with patients that required multivisceral resection. The study was approved by the University of Pittsburgh's institutional review board. Procedures were performed by 15 surgeons, most with specialized training in pancreatic surgery, yet over 85% of cases were performed by four surgeons. UPMC is a high-volume center for pancreatic surgery—as evidenced by 2159 pancreatic resections during the study period (including pancreatoduodenectomy, distal pancreatectomy, Appleby procedure, total pancreatectomy, central pancreatectomy, and enucleation, with pancreatoduodenectomy accounting for 64.1%). Moreover, UPMC is recognized as a high-volume center for pancreatic adenocarcinoma resection—evidenced by 950 resections performed during the study period, among which 75.8% were pancreatoduodenectomies.

RDPs were performed by four surgeons with specialized training in robotic pancreatic surgery. Selection of operative technique is non-randomized and at the discretion of the surgeon. Analysis was performed on an intention-to-treat basis (all conversions were included in the laparoscopic or robotic groups).

Preoperative, operative, postoperative, and oncologic factors are collected in the database by a dedicated database manager. Neoadjuvant therapy (NAT) refers to any cancer-specific treatment (i.e., systemic chemotherapy or chemo-radiotherapy) administered prior to surgical resection. It is our practice to offer NAT to patients with resectable as well as borderline resectable disease; the decision to administer NAT is based on a number of factors in addition to the clinical stage, including patient functional status, availability of NAT trials, and patient preference. NAT was recorded as either neoadjuvant chemotherapy (NAC; where patients received only systemic therapy preoperatively) or neoadjuvant chemo-radiotherapy (NCRT; where patients received systemic chemotherapy and local radiotherapy prior to resection). All patients with PDAC are considered for adjuvant therapy—however, some patients fail to receive adjuvant therapy due to post-operative morbidity, poor functional status, and unwillingness to continue treatment. All systemic therapy administered after surgery, prior to recurrence is considered adjuvant, while systemic therapy administered after development of recurrence is recorded as salvage therapy.

Negative margin resection was defined as tumor distance from the closest margin > 1 mm. Operative time was calculated from first incision to closure of last incision and taken from the intraoperative nursing record. Pancreatic fistula was graded according to the International Study Group on Pancreatic Fistula (ISGPF) classification. Postoperative complications were assessed by the Clavien-Dindo score with the worst complication used to grade overall severity. Readmission was defined as any inpatient hospitalization. Perioperative mortality was defined as death within 30 days or during the index hospitalization after surgery. Data were tested for normalcy and reported with mean and standard deviation (SD). Non-parametric testing was used for all variables without a normal distribution and reported with a median and interquartile range (IQR). Fischer exact or chi-squared testing was used for all categorical variables, and proportions were reported. Survival analysis, including DFS and OS, was performed using Kaplan–Meier survival curves compared by log-rank sum test. Backward Cox-proportional hazard regression analysis was performed to determine if the operative approach was an independent predictor of recurrence and overall survival. Only statistically significant variables were included in the final model. Statistical analysis was performed using SPSS version 26, and all statistical tests were two-sided, and p < 0.05 was considered statistically significant.

Results

Over 12 years, a total of 146 consecutive distal pancreatectomies for PDAC were performed, of which 28.1% were ODP, 60.3% were RDP, and 11.6% were LDP. There were no statistical differences in patient demographics, including gender, comorbidities, and prior abdominal surgeries (Table 1).

Table 1 Demographics and treatment variables
Full size table

A significant number of patients underwent neoadjuvant chemotherapy (43.9% ODP, 47.7% RDP and 23.5% LDP, p = 0.184). On the other hand, a small cohort of patients underwent neoadjuvant radiation (14.6% ODP, 4.5% RDP and 0% LDP, p = 0.053; Table 1).

Patients in the ODP cohort were more likely to have vascular resection (24.4%) compared to RDP (6.8%) or LDP (0%, p = 0.004; Table 1). The estimated blood loss (EBL) was lower in the RDP compared to LDP and ODP, although this did not reach statistical significance (p = 0.119). Remarkably, the RDP group had the lowest median operative time among the three groups (RDP = 213 min, LDP = 238 min, ODP = 348 min; p < 0.001). Finally, robotic surgery was associated with a lower conversion rate (2.3%) compared to laparoscopic surgery (23.5%, p < 0.001; Table 1). Pathologically, there was no difference in the percentage of AJCC stage, T stage, N stage, LN positivity, lymphovascular invasion, perineural invasion, and positive margin between the three groups (Table 2). The majority of patients received adjuvant therapy (70.7% for ODP, 71.6% for RDP, and 82.4% for LDP, p = 0.628; Table 2).

Table 2 Pathological and Oncological outcomes
Full size table

Postoperatively, there was no difference in the frequency of major complications (19.5% for ODP, 15.9% for RDP and 29.4% for LDP, p = 0.414), CR-POPF (9.8% for ODP, 9.1% for RDP and 17.6% for LDP, p = 0.563) or DGE (9.8% for ODP, 3.4% for RDP and 0% for LDP, p = 0.179). A higher rate of postoperative bleeding and reoperation was observed in the LDP cohort compared to the two others, but there was no difference in 90-day mortality among the three groups (Table 3).

Table 3 Post-operative outcomes
Full size table

The median follow-up for the entire cohort was 32.7 months. The median overall survival was 28.4 months for ODP, 34.6 months for RDP, and 32.5 months for LDP (Log Rank p = 0.914; Fig. 1 and Table 4). The median progression-free survival was 16.9 months for ODP, 16.6 months for RDP, and 18.8 months for LDP (Log Rank p = 0.681; Fig. 2 and Table 4). On multivariate Cox proportional hazard analysis, the operative approach was not associated with overall survival (Table 5). Factors associated with worse survival were the presence of comorbidities, lymphovascular invasion, and stage III disease. The receipt of neoadjuvant therapy was associated with improved survival in this unmatched retrospective cohort.

Fig. 1
figure 1

Median overall survival for all patients undergoing open (ODP), robotic (RDP) and laparoscopic (LDP) distal pancreatectomy for adenocarcinoma. Median OS from the time of diagnosis for the entire cohort of resected distal PDAC: mOS = 32.9 months (IQR 27.1–38.6); Open: mOS = 28.4 (IQR 17.7–39.1); Robotic: mOS = 34.6 (IQR 22.7–45.4); Laparoscopic: mOS 32.5 (IQR 15.5–49.5) Log-rank p = 0.914

Full size image
Table 4 Overall survival of patients with pancreatic ductal adenocarcinoma stratified by operative treatment
Full size table
Fig. 2
figure 2

Progression free survival for all patients undergoing open (ODP), robotic (RDP) and laparoscopic (LDP) distal pancreatectomy for adenocarcinoma. Median PFS from the time of diagnosis for the entire cohort of resected distal PDAC: mPFS = 16.9 months (IQR 14.7–19.1); Open: mPFS = 16.9 months (IQR 14.2–19.7); Robotic: mPFS = 16.6 (IQR 12.5–20.8); Laparoscopic: 18.8 (4.2–33.5); Log-rank = 0.681

Full size image
Table 5 Cox-proportional hazard regression analysis for overall survival
Full size table

Discussion

The overall safety of left-sided pancreatic surgery has increased tremendously compared to the early reports of the twentieth century [13, 14]. Yet, the introduction of minimally invasive pancreatic surgery during the 1990s has generated much debate, and many remain skeptical of its merit, especially in the setting of pancreatic ductal adenocarcinoma [5, 9]. The current study describes the perioperative and oncologic outcomes for open, laparoscopic, and robotic distal pancreatectomy for PDAC performed at UPMC between 2008 and 2019.

Our cohort is heavily biased toward MIS techniques, and over 60% of cases were performed using the robotic platform. It must be noted that LDP has been largely abandoned at our institution following the introduction of RDP as institutional data demonstrated superiority of outcomes for RDP compared to LDP in a mixed cohort of benign and malignant diseases. [3, 15, 16] Moreover, a comprehensive comparative analysis of cost-effectiveness and perioperative outcomes between MIS platforms and classic ODP demonstrated superiority of the RDP compared to both ODP and LDP in a mixed cohort of benign and malignant pancreatic disease, these data have been previously published. [17]

In the current intention-to-treat analysis, we found no evidence of inferiority of survival after RDP compared to LDP and ODP performed specifically for PDAC. All cancer-related outcomes remained independent of surgical technique, with margin-negative resection and lymph node harvest similar to previously reported series. RDP was associated with a lower conversion rate than LDP, although the LDP represented only a small portion of our cohort. [18] The decreased conversion rate is at least partly explained by the increased instrument dexterity offered by the robotic platform and the enhanced three-dimensional vision, which are valuable additions to the lack of versatility of the classic laparoscopic instrumentation. As a result, tissue dissection is facilitated, especially in the setting of malignant disease when classic surgical planes are distorted by direct tumor invasion or desmoplastic reaction. Yet, the most common reasons for conversion in both groups included elevated BMI, intraoperative bleeding, tumor proximity to major vasculature, and surgeon experience.

Nonetheless, the majority of vascular resections were still preferentially performed in the ODP group compared to the RDP, and none was performed in the LDP. When considering a MIDP approach for a PDAC with mesenteric venous involvement, it is paramount to ascertain the extension of venous involvement and anticipate the necessity for segmental venous resection rather than a more conservative partial venous resection. It has been our preference to proceed with ODP when the need for a segmental venous resection is anticipated based on preoperative imaging.

Although our group and others have previously described the safety and feasibility of robotic-assisted distal pancreatectomy with celiac axis resection, this procedure is technically demanding, especially in the setting of concomitant portal vein (PV) or superior mesenteric vein involvement (SMV) [19,20,21]. In the setting of PDAC, when both the celiac axis and the PV or the SMV are involved, an open approach might be preferable to facilitate vascular dissection and reconstruction.

Neoadjuvant chemotherapy (NAT) was utilized in nearly half of the entire cohort without significant differences between the ODP and RDP. No consensus on the role of NAT for left-sided PDAC currently exists in the literature [22]. At our institution, the use of NAT in distal PDAC has evolved during the study period, and most recently, NAT is offered to all patients with acceptable performance status (ECOG 0–1) based on multidisciplinary consensus and eligibility for clinical trials.

Postoperative major complications rates were similar between the groups, although the LDP cohorts had an overall higher rate of complications, which did not reach statistical significance. The major complication and reoperation rates for the LDP—recorded in this series—are higher than what was reported by our institution before the introduction of RDP, these results are at least partially biased by the small sample size in the current series. [16]

An important observation is the high rate of adjuvant therapy administration in all three cohorts ODP, RDP, and LDP, which occurred in over 70% of cases. Candidacy for adjuvant therapy is commonly predicated on overall patient performance status and satisfactory recovery from the surgical procedure [23]. The overall rate of adjuvant therapy in this study is encouraging and echoes the rate reported in the literature suggesting non-inferiority of minimally invasive platforms compared to ODP [22].

Ultimately, overall survival was similar between the three groups and comparable with published reports from high-volume pancreatic cancer centers [5]. Although this is an unmatched cohort analyzing three different surgical techniques, we attempted to control for major confounding factors in our Cox model. The outcomes obtained in the RDP group suggest a valuable role for this surgical modality, even in the setting of PDAC.

This analysis has several important limitations. First, as a retrospective study, it is subject to the inherent bias of this study design. Second, a significant portion of the ODP cohort underwent vascular resection in a disproportional ratio compared to RDP and LDP. Most of the vascular resections involved the PV/SMV vessels and were done for centrally located pancreatic lesions. Therefore, at least a portion of the ODP group underwent more extensive surgery than the RDP and LDP. Third, the limited number of LDP included in this study hinders the conclusions derived from the comparison of LDP with RDP and ODP in both the univariate and Cox model. Fourth, the NAT regimens and selection criteria evolved during the study period, and no definitive conclusion can be made on the role of NAT in distal PDAC. Fifth, the exclusion of T4 lesions (AJCC8th)—as an attempt to limit contamination from very large tumors requiring multivisceral resection—limits the generalizability of this study.

In conclusion, this limited institutional series suggests that RDP is safe and feasible in well-selected left-sided PDAC when performed by experienced pancreatic surgeons in a high-volume pancreatic center. Moreover, RDP was not inferior to ODP and LDP in oncology-specific outcomes and overall survival. Further studies evaluating the merits of MIS pancreatic surgery for PDAC should strive to separate the robotic platform from the purely laparoscopic-assisted technique as a growing body of evidence suggests that more favorable outcomes are associated with RDP compared to LDP.