Laparoscopic versus open pancreatoduodenectomy (leopard-2): a multicenter, patient-blinded, randomized controlled trial

INTRODUCTION: Laparoscopic pancreatoduodenectomy (LPD) is gaining popularity because of potential advantages including less operative blood loss, less delayed gastric emptying, and shorter hospital stay but concerns exist regarding increased pancreatic fistula rates and, in low‐volume centers, increased postoperative mortality. Multicenter randomized trials investigating this subject are lacking. In the Netherlands, LPD was introduced according to the IDEAL framework for surgical innovation including a training program (LAELAPS‐2) with a 3.5% 90‐day mortality and 34% pancreatic fistula rate in the first 114 LPDs in 4 centers. According to the IDEAL framework, we hereafter initiated an RCT to assess whether LPD could reduce time to functional recovery. AIMS AND METHODS: A multicenter randomized controlled, patient‐blinded, trial comparing LPD with OPD was performed in 4 centers that each perform ≥20 pancreatoduodenectomies annually (median 37 (range 23‐77)), completed the LPD training program, and had performed at least 20 LPDs during the 18 months training period (range 23‐34). Adult patients with an indication for pancreatoduodenectomy because of a neoplasm without signs of vascular involvement were included. Primary outcome was time (days) to functional recovery. RESULTS: The data safety monitoring board (DSMB) recommended early termination of the trial because of a difference in 90‐day complication‐related mortality (LPD 5 (10%) vs. OPD 1 (2%), p=0.2). The DSMB stated that the possible harm for patients did not outweigh the hypothesized patient benefit of faster recovery. In total, 99 of the projected 136 patients were included (50 LPD and 49 OPD). Time to functional recovery was 9 days (95% CI 6‐12) after LPD vs. 8 days (95% CI 6‐10) after OPD (p=0.9). The conversion rate with LPD was 20% (N=10). Operative blood loss was 300 mL (IQR 200‐519) vs. 400 mL (300‐975) (p=0.13) and operative time 383 (IQR 231‐454) vs. 215 (IQR 180‐299) minutes (p<0.001) for LPD and OPD respectively. Clavien‐Dindo grade ≥3 complications were seen in 25 (50%) vs. 19 patients (39%) (p=0.3), pancreatic fistula grade B/C in 14 (28%) vs. 12 (24%) (p=0.8), hepaticojejunostomy leakage grade B/C in 6 (12%) vs. 5 (10%) (p=0.8) and postpancreatectomy haemorrhage grade B/C in 5 (10%) vs. 7 (14%) (p=0.5) for LPD and OPD respectively. Hospital stay was 11 days (IQR 7‐20) after LPD and 10 (IQR 7‐20) after OPD (p=0.8). The median annual volume of LPD per center during the trial was 11 (range 6‐ 15). CONCLUSION: This early terminated first multicenter RCT showed comparable time to functional recovery and morbidity after LPD versus OPD. The mortality rate after LPD raises concerns, especially in the absence of patient benefits, even in trained centers. Potentially, the learning curve for LPD is longer than previously perceived and possibly a greater annual volume is required.