A Randomized Trial of Two Surgical Techniques for Pancreaticojejunostomy in Patients Undergoing Pancreaticoduodenectomy

Pancreaticoduodenectomy (PD) is a commonly performed operative procedure which is used in
selected patients with benign and malignant diseases of the pancreas and periampullary
region. The procedure involves regional resection of the pancreatic head, neck, and uncinate
process en-bloc with the duodenum, distal bile duct, and lymph nodes. The standard 'Whipple'
operation also adds a distal gastrectomy to the above procedures, while a pylorus-preserving
pancreaticoduodenectomy (PPPD) spares the distal stomach. The indications for PD include
neoplastic processes confined to the periampullary region, such as pancreatic cancer, distal
common bile duct cancer, duodenal cancer, ampullary cancer, neuroendocrine tumors, cystic
tumors, etc. A small number of benign conditions, such as chronic pancreatitis and benign
neoplasms are also treated with PD. Upon completion of the pancreatic resection, 3
anastomoses are used to re-establish GI continuity—a pancreatic-enteric anastomosis, a
biliary-enteric anastomosis, and a gastro or duodeno-enteric anastomosis. The
pancreatic-enteric anastomosis has traditionally been the most troubling of these anastomoses
because of a failure to heal and resultant fistulas and leaks.

The operative mortality rate for PD is usually less than 5% in major surgical centers with
significant experience with the procedure. The leading causes of mortality include
hemorrhage, cardiac events, and sepsis (often related to a pancreatic-enteric fistula). In
contrast to this low mortality rate, the morbidity rate is still quite high with one review
showing a rate of 40%. One of the most common causes of morbidity is a leak or pancreatic
fistula from the pancreatic-enteric anastomosis. A recent review estimated the incidence of
this complication to be 10% to 28.5%. A pancreatic fistula is currently defined by the
International Study Group for Pancreatic Fistulas (ISGPF) as drain amylase levels that are ≥3
times normal amylase levels from the third postoperative day onward, if drain output is ≥
10ml, and if the color of the drained fluid is altered (non-serous). Several large single
institution series from the Mannheim, Lahey, and Mayo Clinics have shown leak rates of
11-15%. The Mannheim Clinic series demonstrated that 20% of the pancreatic fistulas were
directly responsible for postoperative deaths.

There have been several randomized prospective trials by investigators at the Johns Hopkins
Hospital which have tested various interventions attempting to improve the leakage rates. In
one trial, they determined that leak rates were similar (11-12%) whether the
pancreatic-enteric anastomosis was a pancreaticojejunostomy (PJ) or a pancreaticogastrostomy
(PG). In another trial, these investigators evaluated the use of prophylactic octreotide as
an agent to reduce pancreatic fistula rates—in this study there was no decrease in fistula
rates with the use of octreotide. Finally, these authors most recently performed a
randomized, prospective trial of stenting the pancreatic-enteric anastomosis. In this trial,
the fistula rates were not changed by the placement of a perioperative stent across the
anastomosis.

There are two widely used methods for the PJ reconstruction after PD—invagination or
'dunking' the pancreatic remnant or end-to-side duct-to-mucosa PJ. In the invagination
technique, the cut end of the pancreas in sewn into an opening in the side of the jejunum
using two layers of suture—an outer layer of permanent suture on the pancreas capsule and
bowel serosa and muscle, and an inner layer of running dissolvable suture on the duct and
pancreatic parenchyma and full thickness of the bowel wall. In the duct-to-mucosa technique,
there is again an outer layer of interrupted permanent suture. However, the inner layer is an
interrupted anastomosis between the pancreatic duct and the bowel mucosa. In one single
institution study utilizing the duct-to-mucosa technique and an internal stent, Strasberg et
al demonstrated a pancreatic fistula rate of 1.6% in 123 patients. In another review by Tani
et al, the fistula rate was 11% for the stented duct-to-mucosa technique and 6.5% for the 2
layer end-to-side externally stented technique.

There has been only one small randomized, prospective trial evaluating a duct-to-mucosa
versus an end-to-side PJ reported in the literature. In this trial, the authors randomized
144 patients undergoing PD to either a 2-layer duct-to-mucosa anastomosis or a single layer
end-to-side anastomosis which was not invaginated. Pancreatic fistulas were seen in 14% of
patients—13% in the duct-to-mucosa group and 15% in the end-to-side group and there was no
difference in complications between groups. It is not entirely clear from this study how
these anastomoses were performed, but it does not appear that their construction was
compatible to the methods that are most commonly used today.

Therefore, we propose to perform a randomized, prospective, controlled study comparing these
two techniques. This study will be offered to all patients at Thomas Jefferson Hospital
undergoing PD. Patients will be recruited on the basis of the preoperative anticipation of
pancreaticoduodenal resection and preoperative consent will be obtained. Stratification and
randomization will be performed intraoperatively, following pancreaticoduodenal resection.
Because many studies have demonstrated that leak rates are directly related to pancreatic
texture, we will stratify into two groups: soft (normal) texture (predicted fistula rate of
15-30%) and hard (fibrotic) texture (predicted fistula rate of 0-15%). Patients will be
randomized to one of two groups: 1) pancreatic duct to jejunal mucosa, two-layer anastomosis
or 2) end-to-side, two-layer, invagination technique.

The intraoperative management of the patients will not be influenced by this study and will
be under the direction of the attending surgeon. The perioperative care of the patient,
including the use of prophylactic antibiotics, gastric acid secretory inhibition agents,
nasogastric tubes, the timing of removal of operatively placed closed-suction drains, and the
restoration of oral intake will remain under the direction of the attending surgeon. If a
postoperative pancreatic fistula does occur, the attending surgeon will manage the fistula
appropriately. We have a Critical Pathway in place which we will use to standardize patient
care and insure uniform postoperative management.

Immediately following the PD, the attending surgeon will complete a short questionnaire
documenting the type of resection performed, the type of anastomosis, the character of the
remnant pancreas, the size of the pancreatic duct, how the pancreas was transected (stapler,
electrocautery, etc), and other details of the operative procedure. Other routine data that
will be collected includes further details of the operative procedure (from the operative
report), pathology of the resected specimen (from the pathology report), and occurrence of
postoperative morbidity and mortality.

Inclusion Criteria:

- Patients must have undergone pancreaticoduodenectomy.

- Must have pancreatic remnant remaining in place (body and tail).

- Must be candidate for reconstruction by pancreaticojejunostomy by one of the two
techniques described below.

- Must have an identifiable pancreatic duct which can be used for a duct-to-mucosa
anastomosis.

Exclusion Criteria:

- Patients undergoing total pancreatectomy.

- Patients undergoing PD who have had previous left-sided pancreatic resection.

- Failure to sign informed consent.

- Failure to identify the pancreatic duct.

- Pregnant patients.