Surgical Outcome After Distal Pancreatectomy With and Without Portomesenteric Venous Resection in Patients with Pancreatic Adenocarcinoma: A Transatlantic Evaluation of Patients in North America, Germany, Sweden, and The Netherlands (GAPASURG)

This retrospective, observational, multicenter study was performed following the Strengthening the Reporting of Observational Studies (STROBE) guidelines.²⁶ Informed consent was not necessary owing to the use of anonymous datasets.

A total of 9801 DPs were performed during the study period, of which 6877 patients (70.2%) were excluded because of a diagnosis other than pancreatic adenocarcinoma (n = 6177, 63.0%), had missing data regarding the diagnosis (n = 260, 2.7%), had missing data on whether a vascular resection was performed (n = 45, 0.5%), or underwent DP with concomitant arterial resection[s] (n = 168, 1.7%) or multivisceral resection[s] (n = 227, 2.3%).

The final study cohort comprised 2924 patients, of whom 8.2% (n = 241) underwent a concomitant PVR. In North America, Germany, Sweden, and The Netherlands, the proportion of patients who underwent a concomitant PVR among the included patients was 9.2% (n = 199/2165), 5.9% (n = 27/461), 1.9% (n = 3/157), and 8.5% (n = 12/141), respectively (p = 0.002).

To date, this largest observational international study in 2924 patients with pancreatic adenocarcinoma demonstrated that PVR during DP is associated with higher rates of in-hospital/30-day major morbidity (24% vs. 18%) and PPH grade B/C (10% vs. 3%), whereas it did not increase the in-hospital mortality/30-day major morbidity (1.7% vs. 1.2%). PVR was an independent predictor for developing major morbidity, but did not impact the mortality. In contrast, older age, chronic obstructive pulmonary disease, and conversion from minimally invasive to open surgery were associated with mortality. Nevertheless, no mortality occurred among patients who underwent DP-PVR after conversion from minimally invasive to open surgery.

Previously, two single-center, observational studies investigated the surgical outcome after DP-PVR versus standard DP.^9,11 Loos et al. compared 79 patients after DP-PVR with 1383 patients who underwent standard DP for any indication, without concomitant arterial and/or multivisceral resection(s).¹¹ Of note, the DP-PVR group was defined as either a concomitant PVR or left renal vein resection. Their univariable logistic regression analyses supported the findings from the current study that DP-PVR is associated with major morbidity but not with short-term mortality.¹¹ However, no distinction was made between standard DP and DP-PVR in the multivariable analyses. Although Loos et al. did not perform direct statistical comparisons from DP-PVR versus standard DP, differences in surgical outcome between both procedures in that study seemed in line with our study: major morbidity (41% vs. 22%), PPH grade C (5% vs. 2%), and 90-day mortality (1% vs. 1%).¹¹ A second single-center, observational study compared DP-PVR (n = 41) with standard DP (n = 82) without concomitant arterial resection(s), again performed for any indication.⁹ Both 90-day mortality (0% vs. 3%) and major morbidity (17% vs. 13%) did not differ significantly. However, the rate of PPH was significantly higher after DP-PVR (17% vs. 4%), as was also found in our study.⁹

Among patients who developed POPF grade B/C in this study, the rate of PPH grade B/C was higher after DP-PVR compared with standard DP (30% vs. 8%), but no information was available on the timing and cause of PPH. Unfortunately, only two of four registries/audits collected data on both PPH and intraoperative drain management, leaving too few patients to sufficiently investigate patterns in drain management between DP-PVR and standard DP. No difference in the rate of PPH grade B/C was observed in the international PANDORINA trial wherein 282 patients who underwent a DP were randomized between drain versus no drain placement.³⁹ Since the detection of POPF grade B/C was lower in the no drain group, drain placement during DP-PVR might detect POPF in an earlier stage, hypothetically reducing the risk for PPH grade B/C from the reconstructed vein.⁴⁰ Another potential cause of PPH is the use (and type) of thromboprophylaxis,^41,42 but no data on the use of anti-coagulance were available—an important understudied topic with variances in local protocols.⁴²

The findings from both single-center series^9,11 suggest that, in particular, the increased risk for PPH after DP-PVR contributes to the higher major morbidity. The lack of significance in the current study from the differences in ICU admission and relaparotomy rates suggests that the complications related to PVR can mostly be managed with minimally invasive interventions (i.e., Clavien–Dindo grade IIIa³⁴). In contrast, the results from the Heidelberg series suggested that the rates of relaparotomy and ICU admission are higher after PD-PVR compared with standard DP, although this observation could also be related to the higher rates of POPF after DP-PVR compared with standard DP (33% vs. 20%).¹¹

The technical challenge of a DP-PVR is the limited mobilization of the portomesenteric venous axis, in contrast to a PD or TP due to removal of the pancreatic head and the Kocher and Cattell–Braasch maneuvers.⁴³ As a consequence, a DP-PVR for a pancreatic body tumor is mostly only performed in the presence of limited left-sided venous involvement. Segmental venous resections are performed in 22–56% of the DP-PVRs.^9,14,18,19 Hypothetically, the limited capability to mobilize the portomesenteric veins increases the risk for narrowing of the venous reconstruction and thus the risk for early portomesenteric venous thrombosis.¹² Other known risk factors for thrombosis are the use of synthetic grafts for the venous reconstruction and performing a segmental venous resection.^14,44,45 Three studies reported on the incidence of thrombosis after DP-PVR.^12,17,46 A single-center experience from Roch et al. revealed that 4 of 14 patients (29%) developed portomesenteric venous thrombosis after DP-PVR for any indication.¹⁷ Another single-center study reported a 7% incidence of portomesenteric venous thrombosis among 15 patients who underwent DP-PVR for a locally advanced pancreatic neuroendocrine tumor,⁴⁶ while a third single-center study described high rates of occlusion (22%) and > 50% narrowing (28%) among 32 patients who underwent a DP-PVR for pancreatic adenocarcinoma.¹² Based on these findings, Maekawa et al. proposed to perform a PD-PVR instead of DP-PVR if the length of venous involvement is ≥ 30 mm.¹² Unfortunately, information regarding the incidence of thrombosis was not available in the four registries used for the present study.

The results from the current study suggest that DP-PVR is associated with a significant but limited increased risk for major morbidity without an increase in short-term mortality in comparison with a standard DP. Therefore, DP-PVR seems to be a reasonably safe and feasible procedure in patients with pancreatic adenocarcinoma located in the pancreatic body, with limited portomesenteric venous involvement when performed by experienced surgeons. Importantly, the seemingly limited additional risks of DP-PVR over standard DP could be related to more high-volume surgeons/centers performing these procedures. However, this remains speculative because volume data are not available within the NSQIP. Nevertheless, this limited risk elevation seems to outbalance the risks and long-term sequalae from alternative procedures (i.e., extended PD-PVR or TP-PVR). Established benchmarks for major morbidity (≤ 28%) and in-hospital mortality (≤ 4%) after PD-PVR performed in a high-volume center are higher than the outcomes after DP-PVR in our study.⁶ Another option is a TP-PVR, which is associated with similar surgical outcomes as the benchmarks for PD-PVR when performed in high-volume centers.^47,48 However, TP should be avoided when possible considering the associated endocrine and exocrine insufficiencies.²² Furthermore, TP-PVR is associated with gastric venous congestion, particularly when combined with left gastric vein ligation.⁴⁹ Of note, it might also be needed to sacrifice the left gastric vein as part of DP-PVR, after which the gastric venous drainage depends on the right gastric vein, right gastroepiploic vein, and distal esophageal veins, which might be associated with gastric venous congestion. Eventually, the left gastric vein could be reconstructed.⁵⁰ Since the extent of tumor involvement with the portomesenteric venous axis was not available in the currently used registries, it is possible that an extended PD or TP is a more safe and feasible procedure in the case of more extensive venous involvement, depending on the surgeon’s preference.

Over the last decade, the use of minimally invasive DP for benign and (pre-)malignant pancreatic diseases has increased.⁵¹ It is likely this trend will continue based on the results from the DIPLOMA trial, which demonstrated the non-inferiority of minimally invasive DP over open DP in patients with resectable pancreatic adenocarcinoma.⁵² The present observational study showed that conversion from minimally invasive to open surgery is independently associated with both in-hospital/30-day major morbidity and mortality. As vascular involvement is known to be the main indication to convert, caution is required for adenocarcinomas located in the pancreatic body with proximity to the portomesenteric venous axis.⁵³ Unfortunately, the reasons for conversion to open surgery (e.g., unexpected vascular involvement, intraoperative bleeding) are not registered in the four audits/registries. It would seem prudent to use recent imaging when embarking on a minimally invasive DP with potential vascular involvement. The DIPLOMA trial required computed tomography at a maximum of 4 weeks before surgery.⁴⁸

The present study showed that preoperative chemo(radio)therapy was associated with lower odds of developing major morbidity. This might be related to a lower rate of POPF as a consequence of radiotherapy, however data regarding treatment with preoperative radiotherapy were not available in all four registries.⁵⁴

The results from this study should be seen in the light of several limitations. First, no data were available regarding either the extent of venous involvement or the type of PVR, which was why no proper comparison could be made with the GAPASURG data to compare DP-PVR with either TP-PVR or PD-PVR. Both PD and TP procedures can be performed for a wide spectrum of indications and with various concomitant procedures, associated with different levels of complexity and outcomes.^47,48,55 Second, only ISPGS type 2–4 resections were registered as PVR in the NSQIP registry, which might have influenced the surgical outcome in this study. Third, no data were available regarding anti-coagulance therapy, short- and long-term patency of the venous reconstruction, timing and origin of PPH, and detailed data on perioperative oncological treatments and survival. Fourth, the rule of thumb for the logistic regression analysis on predictors for mortality was more or less violated, testing one more dependent variable than allowed, considering the number of events. Fifth, even though all patients who underwent an arterial resection were excluded, a minority of patients was classified as having (y)pT4, most likely due to a misclassification by the pathologist. Sixth, the small numbers of patients who underwent DP-PVR in Germany, The Netherlands, and Sweden did not allow comparative analyses between DP-PVR and standard DP for each of the four registries separately; therefore, the results could have been influenced by (potential) differences in clinical practices among countries (e.g., regarding patient selection, length of hospital stay) and the structure from the registries. Important differences between the registries were (1) voluntary and multicenter registries (North America and Germany) versus mandatory and national registries (Sweden and The Netherlands); (2) auditing strategies; and (3) length of follow-up. These variances might explain why the registry was associated with major morbidity. The current study tried to correct for these potential influences by testing the association of the four registries in the logistic regression analyses. Seventh, no data on adjuvant therapy and survival are collected in the four audits/registries, whereby it was not possible to assess the impact of complications on the receipt of adjuvant chemotherapy and overall survival.

Nevertheless, the major strength of this study is that this relates to the first multicenter study that compared the surgical outcome after DP-PVR versus standard DP, providing valuable evidence regarding the surgical risks of DP-PVR in a large international cohort. These new insights can guide surgeons in preoperative and intraoperative decision making. Future studies should focus on the comparison of DP-PVR with either extended PD-PVR or TP-PVR, hereby taking into account the extent of venous involvement and the type of PVR, and investigating the surgical outcome, including the short- and long-term venous patency as well as the long-term oncological outcome.

Pancreatic adenocarcinoma located in the pancreatic body with portomesenteric venous involvement can be safely managed with DP-PVR, considering the elevated but limited risk for additional major morbidity (24% vs. 18%) compared with standard DP without additional risk for mortality (1.7% vs. 1.2%).