麻豆传媒

Laparoscopic pancreatoduodenectomy was associated with more complication-related deaths than open pancreatoduodenectomy, but also conferred no benefits in time to functional recovery or improved quality of life, a Dutch phase II/III randomized study found.

The trial, called , was terminated early at 73% of the planned sample size because of a 90-day mortality rate of 15% in the laparoscopic group versus 0% in the open group. The participating centers no longer perform the Whipple procedure laparoscopically, noted Marc G. Besselink, MD, of the University of Amsterdam Medical Center, and colleagues.

"These safety concerns were unexpected and worrisome, especially in the setting of trained multicenter surgeons working in centers performing 20 or more pancreatoduodenectomies annually," the researchers wrote online in .

They added that the impact of experience, learning curve, and annual volume warrants further study.

While laparoscopic surgery is known to generally shorten postoperative recovery, concerns remain that the extensive learning curve needed for this complex procedure could increase the risk of complications.

In the Phase II study, 42 patients were randomized at four centers during 2016 to have either open or laparoscopic surgery. All participating surgeons had performed 50 or more advanced laparoscopic gastrointestinal procedures, including 50 or more pancreatoduodenectomies, either laparoscopic or open.

Two of these patients did not have surgery, and three died within 90 days of laparoscopic pancreatoduodenectomy, compared with none of the 20 patients who had open surgery.

For the Phase III study, conducted during 2017, more patients were randomized, and four patients did not undergo surgery. The mean age of patients was 67 in the laparoscopic arm and 66 in the open arm, while 60% and 49%, respectively, were female.

After randomization of 105 patients from both phases of the trial, of whom 99 underwent surgery, the study was again prematurely terminated because of a difference in 90-day complication-related mortality: five of 50 patients in the laparoscopic group versus one of 49 in the open group, for a risk ratio (RR) of 4.90 (95% CI 0.59-40.44, P=0.20).

Moreover, median time to functional recovery was actually longer after laparoscopic surgery: 10 days (95% CI 5-15) versus 8 days (95% CI 7-9) after the open procedure (log-rank P=0.80).

Postoperative complications, costs, and quality of life were comparable overall, the investigators found. Clavien-Dindo grade III or higher complications affected 25 of 50 patients (50%) after laparoscopic pancreatoduodenectomy versus 19 of 49 patients (39%) after an open procedure, for an RR of 1.29 (95% CI 0.82-2.02, P=0.26). Grade B/C postoperative pancreatic fistulas affected 28% versus 24%, respectively, for an RR of 1.14 (95% CI 0.59-2.22, P=0.69).

Median operative time was 136 minutes longer for laparoscopic surgery at 410 (252-481) versus 274 minutes (212-317). Twelve laparoscopic patients (24%) were admitted to intensive care (unplanned) compared with seven (14%) open-surgery patients.

The findings contrasted with those of two recent single-center randomized trials, and , which found shorter length of hospital stay after laparoscopic than after open pancreatoduodenectomy: 7 versus 13 days in 64 patients and 14 versus 17 days in 66 patients, respectively.

The Dutch trial's negative outcome should be interpreted in light of several factors, the authors stated, including that it was underpowered because of premature termination; in addition, the fact that there was no difference in the rate of postoperative complications such as pancreatic fistulas could explain the observed difference in mortality.

"Renewed uptake of this technique, if considered in the future, should include (very) high-volume centers within a well-controlled setting to provide optimal transparency," Besselink and co-authors wrote. "In the meantime, the results of the LEOPARD-2 trial should cause surgeons, surgical departments, and centers worldwide to consider carefully whether they are equipped to perform laparoscopic pancreatoduodenectomy safely."

They noted that a recent in esophagectomy showed that more than 100 procedures were required to reach the learning curve plateau, suggesting that the learning curve for laparoscopic pancreatoduodenectomy may be much longer than previously presumed.

In an , Oliver Strobel, MD, and Markus W. Büchler, MD, both of Heidelberg University Hospital in Germany, commended the researchers for conducting the first multicenter, patient-blinded, randomized, controlled trial of laparoscopic versus open pancreatoduodenectomy.

However, the commentary noted, in addition to an almost five times greater risk of mortality, the laparoscopic procedure required conversion in two cases to open surgery because of uncontrollable vascular damage, which led to death from bowel ischemia. Moreover, two laparoscopic pancreatoduodenectomies had leakage of the gastroenterostomy.

Strobel and Büchler pointed out that in a review of 41 videos of laparoscopic Whipple procedures, nine of the surgeons received below-average technical summary scores. "Taken together, the LEOPARD-2 trial shows that even for well-trained pancreatic surgeons who have done at least 20 procedures, laparoscopic pancreatoduodenectomy comes with increased risks but no clear benefits," the commentary authors wrote. They suggested that robot-assisted minimally invasive pancreatoduodenectomy might be best to use to facilitate reconstruction, enhance safety, and shorten the learning curve.

"A sensible step forward is to shift a larger part of the learning curve from patient treatment to preclinical simulation training," Strobel and Büchler suggested, adding that for now, open pancreatoduodenectomy remains the standard of care, and minimally invasive pancreatoduodenectomy should be reserved for specialized surgeons working in high-volume settings.

Study limitations, Besselink and co-authors said, included the potential effect of the learning curve and the reduction in the number of laparoscopic procedures performed per center. In addition, although patients were blinded to the procedure, the care providers and assessors were not, so future studies should also include blinding of the evaluators.