- Email: [email protected]
A multicenter randomized controlled trial comparing pancreatic leaks after TissueLink versus SEAMGUARD after distal pancreatectomy (PLATS) NCT01051856
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
A multicenter randomized controlled trial comparing pancreatic leaks after TissueLink versus SEAMGUARD after distal pancreatectomy (PLATS) NCT01051856 Christopher R. Shubert, MD,a Christina R. Ferrone, MD,b Carlos Fernandez-del Castillo, MD,b Michael L. Kendrick, MD,a Michael B. Farnell, MD,a Rory L. Smoot, MD,a Mark J. Truty, MD,a and Florencia G. Que, MDa,* a
Mayo Clinic, Department of Surgery, Division of Subspecialty General Surgery, Section of Hepatobiliary and Pancreas Surgery, Rochester, Minnesota b Department of Surgery, Massachusetts General Hospital and Harvard Medical School, General and Gastrointestinal Surgery, Boston, Massachusetts
article info
abstract
Article history:
Background: Pancreatic leak is common after distal pancreatectomy. This trial sought to
Received 9 March 2016
compare TissueLink closure of the pancreatic stump to that of SEAMGUARD.
Received in revised form
Methods: A multicenter, prospective, trial of patients undergoing distal pancreatectomy
6 May 2016
randomized to either TissueLink or SEAMGUARD.
Accepted 10 June 2016
Results: Enrollment was closed early due to poor accrual. Overall, 67 patients were enrolled,
Available online 17 June 2016
35 TissueLink and 32 SEAMGUARD. The two groups differed in American Society of Anesthesiologist class and diagnosis at baseline and were relatively balanced otherwise.
Keywords:
Overall, 37 of 67 patients (55%) experienced a leak of any grade, 15 (46.9%) in the
Distal pancreatectomy
SEAMGUARD arm and 22 (62.9%) in the TissueLink arm (P ¼ 0.19). The clinically significant
TissueLink
leak rate was 17.9%; 22.9% for TissueLink and 12.5% for SEAMGUARD (P ¼ 0.35). There were
SEAMGUARD
no statistically significant differences in major or any pancreatic fistulaerelated morbidity
Postoperative pancreatic fistula
between the two groups. Conclusions: This is the first multicentered randomized trial evaluating leak rate after distal pancreatectomy between two common transection methods. Although a difference in leak rates was observed, it was not statistically significant and therefore does not provide evidence of the superiority of one technique over the other. Choice should remain based on surgeon comfort, experience, and pancreas characteristics. ª 2016 Published by Elsevier Inc.
Article written in accordance with CONSORT 2010 Guidelines for Trial Reporting. Poster Presentation: DDW 2015, Clinical: Pancreas III. * Corresponding author. Mayo Clinic, 200 First St SW, Rochester, MN 55905. Tel.: 507-284-1529; fax: þ1 507 284-5196. E-mail address: [email protected] (F.G. Que). 0022-4804/$ e see front matter ª 2016 Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jss.2016.06.034
shubert et al trial comparing plats
33
Introduction
Trial design
Distal pancreatectomy is performed for broad variety of indications including both benign and malignant conditions. Specifically, distal pancreatectomy refers to resection of the portion of pancreas to the left of the superior mesenteric vein/ portal vein trunk, excluding the duodenum and distal bile duct. Leakage from the pancreatic duct at the resection margin remains one of the most common complications after distal pancreatectomy. This complication prolongs inpatient and outpatient care and results in significant detriments to the patient’s operative experience and increases in the financial burden of pancreatic surgery. Several series have reported pancreatic leak rates ranging from 5% to 40%.1-3 Laparoscopic distal pancreatectomy is reported to achieve similar perioperative outcomes when compared with the traditional open method.4 Newer studies report decreased pancreatic leaks with absorbable mesh reinforcement of the stapled pancreatic transection line (SEAMGUARD).5,6 Thaker et al.6 studied the use of SEAM GUARD reinforcement of the stapled pancreatic transection in 80 patients undergoing distal pancreatectomy. Laparoscopic and open procedures were included into the study. They observed a significant reduction in postoperative pancreatic leak rate from 22% to 3.5% with the use of the SEAMGUARD reinforcement. Jimenez et al.5 did not observe any pancreatic leak in 13 consecutive patients undergoing distal pancreatectomy with the use of SEAMGUARD reinforcement, whereas the leak rate in the historical control group without SEAMGUARD reinforcement was 39%. However, Ferrone et al.,7 demonstrated a 29% fistula rate that did not differ significantly regardless of method of pancreas transection, including stapling with SEAMGUARD.7 Blansfield et al.,8 report a 10% fistula rate with TissueLink compared with a 36% incidence using other methods. This is consistent with our previously published data showing a decreased pancreatic leak after distal pancreatectomy in a porcine model using TissueLink.9 Given the variation in reported outcomes, the best method of transection of the pancreatic stump to reduce postoperative complications is unknown. Thus, a randomized prospective human trial comparing TissueLink versus SEAMGUARD reinforced stapler is needed to determine if one method has superior outcomes for transecting the pancreas during distal pancreatectomy. The objective of this trial is to compare the effectiveness of TissueLink closure of pancreatic stump after distal pancreatectomy to that of SEAMGUARD reinforced stapler closure in preventing pancreatic leaks at the transection margin.
This was a multicenter prospective randomized controlled trial evaluating outcomes after elective distal pancreatectomy for human patients. This trail is comprised of two parallel treatment arms, with patients randomized to either TissueLink or SEAMGUARD closure of the remnant pancreatic stump at the time of distal pancreatectomy. The Mayo Clinic Internal Review Board reviewed and approved the conduct of this study. This study was registered with clinicaltrials.gov as NCT01051856.
Participants All patients were reviewed for study eligibility if they were aged 18 y or older and recommended for elective distal pancreatectomy for any indication. Patients were then excluded if they were currently on immunosuppressive therapy, had undergone chemotherapy within 2 wk before operation, had received bevacizumab (Avastin) treatment within 6 wk before operation, radiotherapy before operation, inability to follow the instructions given by the investigator, lack of compliance, any persons unable or unwilling to give informed consent to participate in this study, pregnant women, prisoners, or institutionalized individuals. Individuals were also excluded if the thickness of pancreas on preoperative computed tomography scan images was greater than 2.0 cm at the site of expected transection in anterioreposterior diameter. Pancreatic thickness was reviewed by the operative surgeon to predetermine the appropriateness of the patient for the study.
Interventions Standardized surgical abdominal approach The abdomen was approached by midline or transverse laparotomy or by laparoscopy. Complete exploration of the abdomen was then performed to determine feasibility of resection. The decision of whether to perform a splenectomy or other concurrent resections of involved organs in addition to the planned/performed distal pancreatectomy was made by the surgeon, taking into account the patient’s underlying disease and relative health status. No additional treatment or covering of the pancreatic remnant is permitted.
Saline-coupled radiofrequency ablation (TissueLink) In the TissueLink treatment arm, after pancreatic transection, with any method chosen by the operating surgeon, the pancreatic remnant was treated with TissueLink alone. Completeness of treatment was assessed visually and by direct or indirect palpation as the treated stump becomes pale, stiff, and rubbery. No suture was allowed to be used in the pancreas.
Methods Reinforced stapler (SEAMGUARD) Hypothesis The rate of pancreatic duct leak at the resection margin (pancreatic fistula [PF]) is lower with TissueLink than that with SEAMGUARD reinforced stapler closure.
In the SEAMGUARD treatment arm, pancreatic transection of the pancreatic body was executed using an endoscopic linear stapling device, ENDO GIA Universal Straight, or Roticulator, 45 or 60 mm (Covidien, Mansfield, MA). The choice staple depth (2.0, 2.5, 3.5, or 4.8 mm) was at the discretion of the
34
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
operating surgeon. Bioabsorbable mesh sleeves (SEAMGUARD, W.L Gore, Flagstaff, AZ) specifically manufactured for the chosen staple depth and cartridge length were placed over the stapler before firing. The operating surgeon must pause for 20 s after closure of the stapler before firing the SEAMGUARD reinforced stapler.
defined as both grades B and C. Table 1 defines the clinical severity of the leak into Grades A-C as in the original ISGPF article. The online Pancreas Club calculator, which is based on the ISGPF definitions, was used for the purposes of this study.12
Sample size determination Other treatment Intraperitoneal drainage All patients received intraperitoneal drainage at the end of the surgical procedure. The decision about type, size, number, and location of the drain was made by the surgeon. Drain amylase was checked 72 h (postoperative day 3) after the surgical procedure. If after postoperative day 3 the clinical situation warrants, additional drain amylase could be checked as well. Volume of drain output was recorded daily until drain removal.
Power for the study was estimated assuming a two-sided, chisquare test comparing the two study arms with the power set at 80%, and significance level 0.05. Specifically, with an enrollment of 200 patients in the two arms (i.e., a total enrollment of 400 patients), there was 80% power to detect an odds ratio (OR) of 2.25 or larger, assuming a pancreatic remnant leak rate (of any grade) of 10% in the saline-coupled radiofrequency ablation group and a leak rate of 20% in the mesh-reinforced stapler group. Accounting for a 10% loss to follow-up, target enrollment was 223 patients in each arm.
Octreotide Octreotide is a synthetic somatostatin analogue inhibiting pancreatic exocrine secretion, which is said to influence the development of PF. Because there is no clear evidence in favor of octreotide treatment in preventing fistula, the use was not allowed unless a pancreatic leak was diagnosed based on International Study Group of Pancreatic Fistula (ISGPF) criteria.10,11 Once a leak was diagnosed, the decision of whether to use octreotide is made by the surgeon.
Primary end point The primary end point is the development of a postoperative pancreatic duct leak at the resection margin (PF) of any grade within 90 d from the operation. For this study, we used the definition, diagnosis, and grading of postoperative PF as outlined in the landmark article by ISGPF.11
Definition PF was defined as amylase-rich (>3 upper limit of normal serum amylase for the treating institution) fluid either in the operatively placed drain or on reinsertion of an image-guided drain for postoperative fluid collection.
Randomization Enrollment was performed preoperatively at the time of scheduling for surgery. Patients were randomized after meeting criteria and having given their written informed consent to participate in the study. Patients were randomized to one of the two treatment arms, either mesh-reinforced stapler or saline-coupled radiofrequency ablation, at the time of enrollment. Surgeons were made aware of the patients’ randomization status at the time of operation. Patients were not blinded and were informed postoperatively to which arm they were randomized. Randomization was stratified by enrolling site. The 1:1 randomization was performed using a dynamic allocation approach based on the PocockeSimon method to ensure a balance in the treatment groups across the enrolling sites. The randomization assignment for each patient was obtained on entering the site at which the patient was enrolled into a web-based application created and supported by the Division of Biomedical Statistics and Informatics at Mayo Clinic in Rochester, Minnesota.
Statistical methods Diagnosis of leak All patients had a drain amylase, on or after postoperative day 3, measured via an operatively placed drain (or a subsequently placed, percutaneous drain) of any measurable volume. A leak was defined by drain amylase content greater than three times the upper limit of normal serum value for the institution. Radiologic documentation of a communication between the drain and the pancreatic duct was not required for diagnosis. Patients returned to their respective participating sites at 30 d for routine follow-up, patients with a postoperative PF were treated as clinically indicated or if they were unable to return, they will be contacted by telephone and queried using a standardized patient questionnaire form regarding signs or symptoms of pancreatic leaks.
Grading Clinical significance of the pancreatic leak will be defined as outlined by the ISGPF 2005.11 Clinically significant leaks were
Continuous variables were summarized with mean, standard deviation, median, lower quartile, upper quartile, and range (minimum, maximum). The continuous baseline characteristics were compared between the two groups using a two-sample t-test for continuous variables or Wilcoxon ranksum test (whichever was appropriate). Categorical variables were estimated with a binomial point estimate and 95% confidence interval (CI). The categorical variables were compared between groups using a chi-square test or Fisher’s exact test in the case that there were too few expected events for a reliable chi-square result. The tests were two sided with a 0.05 level of significance. All analyses were performed on an intent-to-treat basis. The primary outcome is presence of any grade of pancreatic leak within 90 d of surgery. Leak rates were compared between randomized groups using a Fisher’s exact test. Multivariable logistic regression models were used to evaluate the
35
shubert et al trial comparing plats
Table 1 e PF grade. PF (leak) grade
No leak
Grade
Grade
Grade
A
B
C
No
Yes
Yes or No*
Yes or No*
Radiographic documentation of perianastomotic extravasation or drain sinogram connection
No
Yes*
Yes or No*
Yes or No*
Persistent drainage > 3 wk
No
No
Yes
Yes
No
No
Yes
Yes
No
No
Yes
Yes
Amylase > 3 upper limit At any time on or after POD3
“Persistent” as defined by operating surgeon Any specific treatment for PF Including: (1) Partial or total parenteral nutrition (2) Antibiotics (3) Somatostatin analogues Any sign of infection Including: (1) Purulent or sinister drainage (2) Wound infection Readmission for PF*
No
No
Yes
Yes
Percutaneous drainage around pancreatic anastomosis
No
No
Yes
Yes
(downgraded from original definition where this minimally invasive procedure was defined as grade C) Pseudoaneurysm due to PF
No
No
No
Yes
Reoperation for PF
No
No
No
Yes
Sepsis & organ dysfunction* (no definition given in original description; use the generally accepted definition in the following) Documented infection plus evidence of acute organ dysfunction plus two or more of the following: (1) Temperature > 38 C or <35 C, (2) Pulse > 90 beats/min, (3) Respiratory rate > 20 or PaCO2 < 32 mm Hg, (4) Leucocyte count > 12,000 cells/mm3or <4000 cells/mm3
No
No
No
Yes
Death “related to PF”
No
No
No
Yes
Including (1) Repair of site of leakage with wide peripancreatic drainage (2) Conversion to alternative pancreatic-enteric anastomosis (3) Completion pancreatectomy (4) Control of hemorrhage due to pseudoaneurysm
Examples here would include death related to PF as death associated with (1) Intra-abdominal abscess (2) Intra-abdominal bleeding (3) Percutaneous drainage (4) Reoperation for PF *
Had to be interpreted as listed in this table because not defined specifically in the original ISGPF 2005 definition (Bassi, 2005#11).
association of surgical group with a leak event while controlling for the stratification variable and variables found to be imbalanced between the groups at baseline. Note that because there are a small number of events, at most two variables could be used to avoid overfitting and ensuring model convergence. As such, multivariable models were created with the treatment assignment and one additional variable to determine the impact on the treatment arm assignment. In the secondary analysis comparing the 90-d leak grade (no leak, A, B, C) between treatment groups, unavailable analysis was done using a chi-square test. Multivariable
logistic regression was performed to assess the association between surgical group and 90-d leak grade (none or A, B, C) to determine the impact of the stratification variable and to determine the impact of imbalances in the baseline variables, American Society of Anesthesiologist (ASA) class (class 3 versus 1, 2), enrollment site, and diagnosis group (high risk versus low risk). Additional secondary analysis was performed to compare the rate of major PF-related morbidity (MPFRM) between the two treatment groups. MPFRM was defined as having any of the following events: readmission for PF, percutaneous
36
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
drainage, pseudoaneurysm, intensive care unit and/or progressive care unit stay, grade C PF, grade 3A/3B complication, postoperative hemorrhage, or reoperation for PF. Any morbidity was defined as MPFRM or any of the following events: any drainage longer than 3 wk, drain amylase greater than three times the upper limit of normal, specific treatment for PF, signs of infection, PF documented by X-ray, any grade of complication, antibiotic treatment, and grade B PF.
and were consented and randomized. All 68 received the allocated treatment. One patient withdrew from the study after surgery. Overall, 67 patients were assigned to a treatment arm and completed follow-up (Figure): 35 on the TissueLink arm and 32 on the SEAMGUARD arm. Of the two institutions, institution A enrolled 22 patients and institution B enrolled 45 patients.
Demographics and preoperative characteristics
Results The trial enrolled patients from January 2010 to March 2014, inclusive. The trial was closed before targeted enrollment was reached due to slow rate of patient accrual. There were 316 patients assessed for trial eligibility of which 248 patients were excluded: 85 patients due to predetermined exclusionary criteria, 32 due to patient preference, and 131 due to various other reasons. Sixty-eight patients met all inclusion criteria
Overall, 58.2% of patients were female; 65.7% for TissueLink and 50% SEAMGUARD (P ¼ 0.19). Median age for the overall cohort was 57 y; 54 for TissueLink and 63 for SEAMGUARD (P ¼ 0.26). Median body mass index was 27.58; 28.6 for TissueLink and 27.38 for SEAMGUARD (P ¼ 0.37; Table 2). Between treatment arms, there was no statistically significant difference with respect to the incidence of preoperative coronary artery disease, hypertension, chronic obstructive pulmonary disease, or diabetes (all P > 0.05). The
Figure e Study designdrandomization and follow-up. (Color version of figure is available online.)
37
shubert et al trial comparing plats
Table 2 e Demographics and preoperative characteristics. All (n ¼ 67)
SEAMGUARD (n ¼ 32)
TissueLink (n ¼ 35)
Gender, no. (%)
0.19
Male
28 (41.8)
16 (50.0)
12 (34.3)
Female
39 (58.2)
16 (50.0)
23 (65.7)
Age Mean (SD) Median (IQR)
0.26 57.99 (12.79) 57 (49-69)
59.84 (12.73)
56.29 (12.79)
63 (53.5-69)
54 (46-70)
20-78
20-78
34-78
Mean (SD)
28.19 (5.17)
27.60 (5.08)
28.74 (5.27)
Median (IQR)
27.58 (24.3-31.65)
27.38 (23.65-30.1)
28.6 (24.6-34.2)
18.14-40.4
19.5-40.4
18.14-38.6
Range BMI
Range
P value
0.37
<0.01
Diagnosis, no. (%) Adenocarcinoma
11 (16.4)
7 (21.9)
Neuroendocrine
19 (28.4)
12 (37.5)
4 (11.4) 7 (20.0)
Metastasis
6 (9.0)
3 (9.4)
3 (8.6)
IPMN
5 (7.5)
2 (6.3)
3 (8.6)
Mucinous cystic neoplasia
7 (10.4)
0 (0.0)
7 (20.0)
Cystic lesion other than IPMN/MCN
5 (7.5)
1 (3.1)
4 (11.4)
Chronic pancreatitis
4 (6.0)
4 (12.5)
0 (0.0)
10 (14.9)
3 (9.4)
7 (20.0)
5 (7.5)
3 (9.4)
2 (5.7)
34 (50.7)
17 (53.1)
17 (48.6)
0.71
5 (7.5)
1 (3.1)
4 (11.4)
0.36
14 (20.9)
7 (21.9)
7 (20.0)
Other CAD, no. (%) Hypertension, no. (%) COPD, no. (%) Diabetes, no. (%) ASA class, no. (%)
2 (3.0)
0 (0.0)
2 (5.7)
2
45 (67.2)
18 (56.3)
27 (77.1)
3
20 (29.9)
14 (43.8)
6 (17.1)
Serum albumin
Mean (SD) Median (IQR) Range
0.85 0.02
1
Missing*
0.66
0.64 34 4.42 (0.41)
23 4.37 (0.36)
21 4.45 (0.45)
4.5 (4.2-4.7)
4.2 (4.2-4.5)
4.5 (4.3-4.7)
3.2-5.1
3.9-5.1
3.2-5
BMI ¼ body mass index; CAD ¼ coronary artery disease; COPD ¼ chronic obstructive pulmonary disease; IPMN ¼ intraductal papillary mucinous neoplasm; MCN ¼ mucinous cystic neoplasm; SD ¼ standard deviation. * No. (%) used reported when not 100%.
ASA status was found to differ between the treatment arms with the SEAMGUARD cohort having a greater proportion of ASA class III patients (P ¼ 0.02). Median preoperative serum albumin was 4.5 (interquartile range [IQR] 4.2-4.7) and was not significantly different between treatment arms (P ¼ 0.64). Indications for operation were varied and were found to statistically differ between the treatment arms (P < 0.01; Table 2).
Operative characteristics Overall, 68.7% of operations were performed laparoscopically; 77.1% TissueLink and 59.4% SEAMGUARD (P ¼ 0.12). The operations were performed by a number of different surgeons with no one surgeon having performing significantly more
operations than the others (P ¼ 0.98). Conversion to an open operation was performed in 15.9% of patients: 8.0% TissueLink and 26.3% SEAMGUARD (P ¼ 0.21). The median operative duration was 199 min (IQR 141-262), and median estimated blood loss was 200 cc (IQR 100-400) for both arms combined; these variables did not differ between the treatment arms (Table 3). Most patients (73%) had concurrent organ resections at the time of distal pancreatectomy, 63% TissueLink and 84% SEAMGUARD (P ¼ 0.06). The most common solitary concurrent organ resection was splenectomy (n ¼ 38, 56.7 %). In addition to splenectomy, 10 (15%) patients had additional organs resected at the time of distal pancreatectomy. When compared between treatment arms, there was no difference in concurrent resections (Table 3).
38
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
Table 3 e Operative characteristics. All (n ¼ 67)
SEAMGUARD (n ¼ 32)
TissueLink (n ¼ 35)
Open versus laparoscopic, no. (%)
0.12
Open
21 (31.3)
13 (40.6)
8 (22.9)
Lap
46 (68.7)
19 (59.4)
27 (77.1)
10 (14.9)
5 (15.6)
5 (14.3)
Multiple organs resected, no. (%) Number of organs resected Mean (SD)
1.29 (0.68) 1 (1,1)
Range
1 to 5
1.22 (0.42) 1 (1,1) 1 to 2
1.38 (0.92) 1 (1,1) 1 to 5
Additional organs resected, no. (%)
Spleen
0.06 1 (1.5)
0 (0.0)
1 (2.9)
38 (56.7)
22 (68.8)
16 (45.7)
Spleen, portion of diaphragm, segment II liver, and gastric wedge resection
1 (1.5)
0 (0.0)
1 (2.9)
Spleen and Gallbladder
6 (9.0)
5 (15.6)
1 (2.9)
Spleen and partial gastrectomy
2 (3.0)
0 (0.0)
2 (5.7)
Spleen and adrenal gland
1 (1.5)
0 (0.0)
1 (2.9)
18 (26.9)
5 (15.6)
13 (37.1)
1
4 (6.0)
3 (9.4)
1 (2.9)
3
1 (1.5)
1 (3.1)
0 (0.0)
8
17 (25.4)
9 (28.1)
8 (22.9)
9
13 (19.4)
4 (12.5)
9 (25.7)
10
5 (7.5)
2 (6.3)
3 (8.6)
12
5 (7.5)
3 (9.4)
2 (5.7)
13
8 (11.9)
2 (6.3)
5 (14.3)
14
14 (20.9)
7 (21.9)
7 (20.0)
None Surgeon, no. (col %)
0.98
Duration of operation (min) Mean (SD) Median (IQR) Range
0.54 215.82 (84.53) 199 (141-265) 97-528
222.53 (91.22) 204 (155.5-265.5) 97-528
209.69 (78.75) 194 (138-259) 116-434
Est blood loss (mL) Mean (SD) Median (IQR) Range
1.00 0.43
Median (IQR)
Nephrectomy
P value
0.50 331.42 (351.71) 200 (100-400) 0-1700
362.87 (373.43) 200 (100-600) 0-1700
303.57 (334.27) 200 (100-350) 0-1600
SD ¼ standard deviation.
Pancreatic leak Overall, 37 of 67 (55%) patients experienced a leak of any grade, (95% CI: 37.46%-73.68%). Fifteen (46.9%; 95% CI: 27.08%-64.65%) in the SEAMGUARD arm and 22 (62.9%; 95% CI: 34.89%-90.08%) in the TissueLink arm. The difference in experienced leak rate of any grade between the groups was 16.0% (95% CI: 7.81%-25.43% P ¼ 0.19). Twelve patients experienced a clinically significant leak yielding a rate of 17.9% (95% CI: 8.72%-27.08%). Eight patients (22.9%; 95% CI: 8.99%-36.8%) in the TissueLink group experienced a clinically significant leak compared with four patients (12.5%; 95% CI: 1.04%-23.96%) in the SEAMGUARD group. The difference in clinically significant leak
rate between the groups was 10.4% (95% CI: 1.22%-19.58% P ¼ 0.35). The univariable OR for a clinically significant leak (TissueLeak versus SEAMGUARD) was 2.07 (95% CI: 0.567.70). The OR for treatment group in a multivariable model adjusting for enrolling site was 2.11 (95% CI: 0.57-7.89), the OR for treatment group in a multivariable model adjusting for ASA class was 2.39 (95% CI: 0.59-9.63), and the OR for treatment group in a multivariable model adjusting for indication for surgery (i.e., diagnosis) was 1.71 (95% CI: 0.44-6.57). All these estimates of treatment effect were similar, indicating little impact of baseline imbalances. The observed difference in leak rates between the arms was not statistically significant, which may be due to low power.
39
shubert et al trial comparing plats
Table 4 e Postoperative outcomes. Outcomes no. (col%)
No. (col %) All (n ¼ 67)
SEAMGUARD (n ¼ 32)
TissueLink (n ¼ 35)
P value
PFdany grade
37 (55.2)
15 (46.9)
22 (62.9)
0.19
Clinically significant PF
12 (17.9)
4 (12.5)
8 (22.9)
0.35
*
Drainage longer than 3 wk Missing Yes
3
2
1
14 (21.9)
5 (16.7)
9 (26.5)
0.34
Drained amylase 3 normal upper limit
38 (56.7)
15 (46.9)
23 (65.7)
0.12
Specific treatment for PF
0.35
12 (17.9)
4 (12.5)
8 (22.9)
Readmissions for PF
9 (13.4)
4 (12.5)
5 (14.3)
1.00
Percutaneous drainage
9 (13.4)
4 (12.5)
5 (14.3)
1.00
PF documented by X-ray
9 (13.4)
4 (12.5)
5 (14.3)
1.00
Pseudoaneurysm due to PF
1 (1.5)
0 (0.0)
1 (2.9)
1.00
26 (38.8)
12 (37.5)
14 (40.0)
0.83
5 (7.5)
1 (3.1)
4 (11.4)
0.36
Postoperative CT scan Postoperative sonogram Postoperative antibiotic treatment
15 (22.4)
7 (21.9)
8 (22.9)
0.92
4 (6.0)
3 (9.4)
1 (2.9)
0.34
Any morbidity
48 (71.6)
21 (65.6)
27 (77.1)
0.30
Major PF-related morbidity
19 (28.4)
10 (31.3)
9 (25.7)
ICU or PCU stay
Postoperative LOS Mean (SD) Range
0.62 0.31
5.45 (2.61)
5.94 (3.09)
5.00 (2.03)
3-15
3-13
3-15
CT ¼ computed tomography; ICU/PCU ¼ intensive care unit/progressive care unit; LOS ¼ length of stay; SD ¼ standard deviation. * No. (%) used reported when not 100%.
Other postoperative outcomes There were no postoperative mortalities in either treatment arms. Overall, 38 patients (56.7%) were found to have a drain amylase three times the upper limit of normal. Of the 12 patients (17.9%) requiring specific treatment for PF, nine (13.4%) required percutaneous drainage. Twenty-six patients (38.8%) required postoperative computed tomography scan, and 15 (22.4%) required postoperative antibiotics. Overall, four patients (6.0%) required care in a monitored setting (intensive or progressive care unit). Only one patient experienced pseudoaneurysm (Table 4). Average length of stay was 5.5 d (range 3-15), and readmission occurred in nine patients (13.4%); however, neither of these variables statistically differed between treatment arms (Table 4). The rate of MPFRM was found to be 28.4% (n ¼ 19): 25.7% (n ¼ 9) for TissueLink and 31.3% (n ¼ 10) for SEAMGUARD (P ¼ 0.62). The rate of any morbidity was 71.6% (n ¼ 48): 77.1% (n ¼ 27) for TissueLink and 65.6% (n ¼ 21) for SEAMGUARD (P ¼ 0.30; Table 4). Multivariable analysis (adjusting for enrolling site, ASA, and diagnosis) comparing TissueLink with SEAMGUARD did not reveal a significant association of treatment arm with any grade PF (OR ¼ 0.69, 95% CI, 0.23-2.04; P ¼ 0.50); ASA class (OR ¼ 1.42, 95% CI, 0.43-4.67; P ¼ 0.57); enrollment site (OR ¼ 2.80, 95% CI, 0.87-8.97; P ¼ 0.08); or diagnosis group (OR ¼ 0.67, 95% CI, 0.20-2.22; P ¼ 0.52). Clinically significant
fistula was also not found to be independently associated with treatment arm (OR ¼ 1.93, 95% CI, 0.47-7.99; P ¼ 0.37).
Conditional power analysis Conditional power analysis was performed given the findings and early closure of the study. Under the conditions of the study and the current observed difference in clinically significant leak rates between the two treatment groups, it would require 422 total patients to reach statistical significance assuming a two-sided, two-sample t-test comparing the two study arms with 80% power and 0.05 alpha level. This estimation was found to be similar to the power estimation determined before enrollment.
Discussion This is the first multicenter, randomized trial to evaluate the incidence of leak rate after distal pancreatectomy between two separate methods of pancreatic transection and remnant stump management. Although this study closed early due to poor patient enrollment, this remains a significant scientific effort among two large academic medical centers and, at the time of the study, the first ever contribution of level 1 evidence to this source of morbidity after distal pancreatectomy. Although there was an absolute difference in clinically significant leak rates between arms (22.9% TissueLink versus 12.5% SEAMGUARD), statistical significance was not reached.
40
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
The lack of statistical significance is likely due to a lack of power (insufficient sample size). This can be seen by the 95% CI of the difference in leak rates between groups, which contains clinically meaningful values and the results of the conditional power analysis, which suggest that a statistically significant difference would be obtainable had a sufficient number of patients been enrolled. This is also different than the results of our previous porcine model.9 The reason for this difference is not certain, although plausible to be related to the differences in porcine and human pancreas and the fact that this study was performed on diseased pancreatic parenchyma, which is likely to affect the incidence of fibrosis and ductal dilatation.
Limitations This study has the obvious limitation of insufficient sample size. However, given the accrual rate of patients over the time of the study, the observed difference in leak rates and results of the conditional power analysis, it would have been impractical, and neither financially nor technically feasible, to continue the study. In addition, had the exclusion criteria for pancreatic thickness been set at a level greater than the chosen 2 cm, it is plausible that accrual could have been greater. Although, this cutoff was chosen to create a homogeneous population with equally efficacious use of either treatment, as pancreatic parenchyma greater than 2-cm thick can create a contraindication for stapler transection. In addition, a standard staple height could have been mandated, although many surgeons do in fact vary staple height based on pancreasspecific factors, and we wanted to allow for this as it is current practice.
Conclusions Despite the study limitations and a nonsignificant difference in clinically significant leak rates between the treatment arms, this represents the first ever multicenter randomized controlled trial investigating the outcomes of distal pancreatectomy using two contemporary treatment approaches. We have here demonstrated that if the trail had continued and reached target enrollment, a significant difference in leak rate between treatment arms may have been possible given the observed differences; it is unlikely that this sample size is feasible given the accrual rates observed and power required. Therefore, reaching statistical significance in a prospective randomized controlled fashion using these two treatment approaches is limited. Given lack of these definitive results, it is likely that both treatment options will continue to represent the current standard of care and should be chosen based on surgeon comfort, experience, and pancreas-related factors.
Acknowledgment Authors’ contributions: C.R.F., C.F.-d.C., M.L.K., M.B.F., R.L.S., M.J.T., and F.G.Q. contributed to hypothesis generation, trial design, patient enrollment, and trial participation. C.R.S., C.R.F., C.F.-d.C., and F.G.Q. contributed to data collection and statistical analysis. All the authors contributed to preparation of the article.
Disclosure The authors have nothing to disclose.
references
1. Fahy BN, Frey CF, Ho HS, Beckett L, Bold RJ. Morbidity, mortality, and technical factors of distal pancreatectomy. Am J Surg. 2002;183:237e241. 2. Ferna´ndez-del Castillo C, Rattner DW, Warshaw AL. Standards for pancreatic resection in the 1990s. Arch Surg. 1995;130:295e299. discussion 299-300. 3. Lillemoe KD, Kaushal S, Cameron JL, Sohn TA, Pitt HA, Yeo CJ. Distal pancreatectomy: indications and outcomes in 235 patients. Ann Surg. 1999;229:693e698. discussion 698-700. 4. Kooby DA, Gillespie T, Bentrem D, et al. Left-sided pancreatectomy: a multicenter comparison of laparoscopic and open approaches. Ann Surg. 2008;248:438e446. 5. Jimenez RE, Mavanur A, MacAulay WP. Staple line reinforcement reduces postoperative pancreatic stump leak after distal pancreatectomy. J Gastrointest Surg. 2007;11:345e349. 6. Thaker RI, Matthews BD, Linehan DC, Strasberg SM, Eagon JC, Hawkins WG. Absorbable mesh reinforcement of a stapled pancreatic transection line reduces the leak rate with distal pancreatectomy. J Gastrointest Surg. 2007;11:59e65. 7. Ferrone CR, Warshaw AL, Rattner DW, et al. Pancreatic fistula rates after 462 distal pancreatectomies: staplers do not decrease fistula rates. J Gastrointest Surg. 2008;12:1691e1697. 8. Blansfield JA, Rapp MM, Chokshi RJ, et al. Novel method of stump closure for distal pancreatectomy with a 75% reduction in pancreatic fistula rate. J Gastrointest Surg. 2012;16:524e528. 9. Truty MJ, Sawyer MD, Que FG. Decreasing pancreatic leak after distal pancreatectomy: saline-coupled radiofrequency ablation in a porcine model. J Gastrointest Surg. 2007;11:998e1007. 10. Connor S, Alexakis N, Garden OJ, Leandros E, Bramis J, Wigmore SJ. Meta-analysis of the value of somatostatin and its analogues in reducing complications associated with pancreatic surgery. Br J Surg. 2005;92:1059e1067. 11. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery. 2005;138:8e13. 12. Pancreas Club Calculator ISGPS leak definition. Available at: http://pancreasclub.com/calculators/isgps-calculator/
Available online at www.sciencedirect.com
ScienceDirect journal homepage: www.JournalofSurgicalResearch.com
A multicenter randomized controlled trial comparing pancreatic leaks after TissueLink versus SEAMGUARD after distal pancreatectomy (PLATS) NCT01051856 Christopher R. Shubert, MD,a Christina R. Ferrone, MD,b Carlos Fernandez-del Castillo, MD,b Michael L. Kendrick, MD,a Michael B. Farnell, MD,a Rory L. Smoot, MD,a Mark J. Truty, MD,a and Florencia G. Que, MDa,* a
Mayo Clinic, Department of Surgery, Division of Subspecialty General Surgery, Section of Hepatobiliary and Pancreas Surgery, Rochester, Minnesota b Department of Surgery, Massachusetts General Hospital and Harvard Medical School, General and Gastrointestinal Surgery, Boston, Massachusetts
article info
abstract
Article history:
Background: Pancreatic leak is common after distal pancreatectomy. This trial sought to
Received 9 March 2016
compare TissueLink closure of the pancreatic stump to that of SEAMGUARD.
Received in revised form
Methods: A multicenter, prospective, trial of patients undergoing distal pancreatectomy
6 May 2016
randomized to either TissueLink or SEAMGUARD.
Accepted 10 June 2016
Results: Enrollment was closed early due to poor accrual. Overall, 67 patients were enrolled,
Available online 17 June 2016
35 TissueLink and 32 SEAMGUARD. The two groups differed in American Society of Anesthesiologist class and diagnosis at baseline and were relatively balanced otherwise.
Keywords:
Overall, 37 of 67 patients (55%) experienced a leak of any grade, 15 (46.9%) in the
Distal pancreatectomy
SEAMGUARD arm and 22 (62.9%) in the TissueLink arm (P ¼ 0.19). The clinically significant
TissueLink
leak rate was 17.9%; 22.9% for TissueLink and 12.5% for SEAMGUARD (P ¼ 0.35). There were
SEAMGUARD
no statistically significant differences in major or any pancreatic fistulaerelated morbidity
Postoperative pancreatic fistula
between the two groups. Conclusions: This is the first multicentered randomized trial evaluating leak rate after distal pancreatectomy between two common transection methods. Although a difference in leak rates was observed, it was not statistically significant and therefore does not provide evidence of the superiority of one technique over the other. Choice should remain based on surgeon comfort, experience, and pancreas characteristics. ª 2016 Published by Elsevier Inc.
Article written in accordance with CONSORT 2010 Guidelines for Trial Reporting. Poster Presentation: DDW 2015, Clinical: Pancreas III. * Corresponding author. Mayo Clinic, 200 First St SW, Rochester, MN 55905. Tel.: 507-284-1529; fax: þ1 507 284-5196. E-mail address: [email protected] (F.G. Que). 0022-4804/$ e see front matter ª 2016 Published by Elsevier Inc. http://dx.doi.org/10.1016/j.jss.2016.06.034
shubert et al trial comparing plats
33
Introduction
Trial design
Distal pancreatectomy is performed for broad variety of indications including both benign and malignant conditions. Specifically, distal pancreatectomy refers to resection of the portion of pancreas to the left of the superior mesenteric vein/ portal vein trunk, excluding the duodenum and distal bile duct. Leakage from the pancreatic duct at the resection margin remains one of the most common complications after distal pancreatectomy. This complication prolongs inpatient and outpatient care and results in significant detriments to the patient’s operative experience and increases in the financial burden of pancreatic surgery. Several series have reported pancreatic leak rates ranging from 5% to 40%.1-3 Laparoscopic distal pancreatectomy is reported to achieve similar perioperative outcomes when compared with the traditional open method.4 Newer studies report decreased pancreatic leaks with absorbable mesh reinforcement of the stapled pancreatic transection line (SEAMGUARD).5,6 Thaker et al.6 studied the use of SEAM GUARD reinforcement of the stapled pancreatic transection in 80 patients undergoing distal pancreatectomy. Laparoscopic and open procedures were included into the study. They observed a significant reduction in postoperative pancreatic leak rate from 22% to 3.5% with the use of the SEAMGUARD reinforcement. Jimenez et al.5 did not observe any pancreatic leak in 13 consecutive patients undergoing distal pancreatectomy with the use of SEAMGUARD reinforcement, whereas the leak rate in the historical control group without SEAMGUARD reinforcement was 39%. However, Ferrone et al.,7 demonstrated a 29% fistula rate that did not differ significantly regardless of method of pancreas transection, including stapling with SEAMGUARD.7 Blansfield et al.,8 report a 10% fistula rate with TissueLink compared with a 36% incidence using other methods. This is consistent with our previously published data showing a decreased pancreatic leak after distal pancreatectomy in a porcine model using TissueLink.9 Given the variation in reported outcomes, the best method of transection of the pancreatic stump to reduce postoperative complications is unknown. Thus, a randomized prospective human trial comparing TissueLink versus SEAMGUARD reinforced stapler is needed to determine if one method has superior outcomes for transecting the pancreas during distal pancreatectomy. The objective of this trial is to compare the effectiveness of TissueLink closure of pancreatic stump after distal pancreatectomy to that of SEAMGUARD reinforced stapler closure in preventing pancreatic leaks at the transection margin.
This was a multicenter prospective randomized controlled trial evaluating outcomes after elective distal pancreatectomy for human patients. This trail is comprised of two parallel treatment arms, with patients randomized to either TissueLink or SEAMGUARD closure of the remnant pancreatic stump at the time of distal pancreatectomy. The Mayo Clinic Internal Review Board reviewed and approved the conduct of this study. This study was registered with clinicaltrials.gov as NCT01051856.
Participants All patients were reviewed for study eligibility if they were aged 18 y or older and recommended for elective distal pancreatectomy for any indication. Patients were then excluded if they were currently on immunosuppressive therapy, had undergone chemotherapy within 2 wk before operation, had received bevacizumab (Avastin) treatment within 6 wk before operation, radiotherapy before operation, inability to follow the instructions given by the investigator, lack of compliance, any persons unable or unwilling to give informed consent to participate in this study, pregnant women, prisoners, or institutionalized individuals. Individuals were also excluded if the thickness of pancreas on preoperative computed tomography scan images was greater than 2.0 cm at the site of expected transection in anterioreposterior diameter. Pancreatic thickness was reviewed by the operative surgeon to predetermine the appropriateness of the patient for the study.
Interventions Standardized surgical abdominal approach The abdomen was approached by midline or transverse laparotomy or by laparoscopy. Complete exploration of the abdomen was then performed to determine feasibility of resection. The decision of whether to perform a splenectomy or other concurrent resections of involved organs in addition to the planned/performed distal pancreatectomy was made by the surgeon, taking into account the patient’s underlying disease and relative health status. No additional treatment or covering of the pancreatic remnant is permitted.
Saline-coupled radiofrequency ablation (TissueLink) In the TissueLink treatment arm, after pancreatic transection, with any method chosen by the operating surgeon, the pancreatic remnant was treated with TissueLink alone. Completeness of treatment was assessed visually and by direct or indirect palpation as the treated stump becomes pale, stiff, and rubbery. No suture was allowed to be used in the pancreas.
Methods Reinforced stapler (SEAMGUARD) Hypothesis The rate of pancreatic duct leak at the resection margin (pancreatic fistula [PF]) is lower with TissueLink than that with SEAMGUARD reinforced stapler closure.
In the SEAMGUARD treatment arm, pancreatic transection of the pancreatic body was executed using an endoscopic linear stapling device, ENDO GIA Universal Straight, or Roticulator, 45 or 60 mm (Covidien, Mansfield, MA). The choice staple depth (2.0, 2.5, 3.5, or 4.8 mm) was at the discretion of the
34
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
operating surgeon. Bioabsorbable mesh sleeves (SEAMGUARD, W.L Gore, Flagstaff, AZ) specifically manufactured for the chosen staple depth and cartridge length were placed over the stapler before firing. The operating surgeon must pause for 20 s after closure of the stapler before firing the SEAMGUARD reinforced stapler.
defined as both grades B and C. Table 1 defines the clinical severity of the leak into Grades A-C as in the original ISGPF article. The online Pancreas Club calculator, which is based on the ISGPF definitions, was used for the purposes of this study.12
Sample size determination Other treatment Intraperitoneal drainage All patients received intraperitoneal drainage at the end of the surgical procedure. The decision about type, size, number, and location of the drain was made by the surgeon. Drain amylase was checked 72 h (postoperative day 3) after the surgical procedure. If after postoperative day 3 the clinical situation warrants, additional drain amylase could be checked as well. Volume of drain output was recorded daily until drain removal.
Power for the study was estimated assuming a two-sided, chisquare test comparing the two study arms with the power set at 80%, and significance level 0.05. Specifically, with an enrollment of 200 patients in the two arms (i.e., a total enrollment of 400 patients), there was 80% power to detect an odds ratio (OR) of 2.25 or larger, assuming a pancreatic remnant leak rate (of any grade) of 10% in the saline-coupled radiofrequency ablation group and a leak rate of 20% in the mesh-reinforced stapler group. Accounting for a 10% loss to follow-up, target enrollment was 223 patients in each arm.
Octreotide Octreotide is a synthetic somatostatin analogue inhibiting pancreatic exocrine secretion, which is said to influence the development of PF. Because there is no clear evidence in favor of octreotide treatment in preventing fistula, the use was not allowed unless a pancreatic leak was diagnosed based on International Study Group of Pancreatic Fistula (ISGPF) criteria.10,11 Once a leak was diagnosed, the decision of whether to use octreotide is made by the surgeon.
Primary end point The primary end point is the development of a postoperative pancreatic duct leak at the resection margin (PF) of any grade within 90 d from the operation. For this study, we used the definition, diagnosis, and grading of postoperative PF as outlined in the landmark article by ISGPF.11
Definition PF was defined as amylase-rich (>3 upper limit of normal serum amylase for the treating institution) fluid either in the operatively placed drain or on reinsertion of an image-guided drain for postoperative fluid collection.
Randomization Enrollment was performed preoperatively at the time of scheduling for surgery. Patients were randomized after meeting criteria and having given their written informed consent to participate in the study. Patients were randomized to one of the two treatment arms, either mesh-reinforced stapler or saline-coupled radiofrequency ablation, at the time of enrollment. Surgeons were made aware of the patients’ randomization status at the time of operation. Patients were not blinded and were informed postoperatively to which arm they were randomized. Randomization was stratified by enrolling site. The 1:1 randomization was performed using a dynamic allocation approach based on the PocockeSimon method to ensure a balance in the treatment groups across the enrolling sites. The randomization assignment for each patient was obtained on entering the site at which the patient was enrolled into a web-based application created and supported by the Division of Biomedical Statistics and Informatics at Mayo Clinic in Rochester, Minnesota.
Statistical methods Diagnosis of leak All patients had a drain amylase, on or after postoperative day 3, measured via an operatively placed drain (or a subsequently placed, percutaneous drain) of any measurable volume. A leak was defined by drain amylase content greater than three times the upper limit of normal serum value for the institution. Radiologic documentation of a communication between the drain and the pancreatic duct was not required for diagnosis. Patients returned to their respective participating sites at 30 d for routine follow-up, patients with a postoperative PF were treated as clinically indicated or if they were unable to return, they will be contacted by telephone and queried using a standardized patient questionnaire form regarding signs or symptoms of pancreatic leaks.
Grading Clinical significance of the pancreatic leak will be defined as outlined by the ISGPF 2005.11 Clinically significant leaks were
Continuous variables were summarized with mean, standard deviation, median, lower quartile, upper quartile, and range (minimum, maximum). The continuous baseline characteristics were compared between the two groups using a two-sample t-test for continuous variables or Wilcoxon ranksum test (whichever was appropriate). Categorical variables were estimated with a binomial point estimate and 95% confidence interval (CI). The categorical variables were compared between groups using a chi-square test or Fisher’s exact test in the case that there were too few expected events for a reliable chi-square result. The tests were two sided with a 0.05 level of significance. All analyses were performed on an intent-to-treat basis. The primary outcome is presence of any grade of pancreatic leak within 90 d of surgery. Leak rates were compared between randomized groups using a Fisher’s exact test. Multivariable logistic regression models were used to evaluate the
35
shubert et al trial comparing plats
Table 1 e PF grade. PF (leak) grade
No leak
Grade
Grade
Grade
A
B
C
No
Yes
Yes or No*
Yes or No*
Radiographic documentation of perianastomotic extravasation or drain sinogram connection
No
Yes*
Yes or No*
Yes or No*
Persistent drainage > 3 wk
No
No
Yes
Yes
No
No
Yes
Yes
No
No
Yes
Yes
Amylase > 3 upper limit At any time on or after POD3
“Persistent” as defined by operating surgeon Any specific treatment for PF Including: (1) Partial or total parenteral nutrition (2) Antibiotics (3) Somatostatin analogues Any sign of infection Including: (1) Purulent or sinister drainage (2) Wound infection Readmission for PF*
No
No
Yes
Yes
Percutaneous drainage around pancreatic anastomosis
No
No
Yes
Yes
(downgraded from original definition where this minimally invasive procedure was defined as grade C) Pseudoaneurysm due to PF
No
No
No
Yes
Reoperation for PF
No
No
No
Yes
Sepsis & organ dysfunction* (no definition given in original description; use the generally accepted definition in the following) Documented infection plus evidence of acute organ dysfunction plus two or more of the following: (1) Temperature > 38 C or <35 C, (2) Pulse > 90 beats/min, (3) Respiratory rate > 20 or PaCO2 < 32 mm Hg, (4) Leucocyte count > 12,000 cells/mm3or <4000 cells/mm3
No
No
No
Yes
Death “related to PF”
No
No
No
Yes
Including (1) Repair of site of leakage with wide peripancreatic drainage (2) Conversion to alternative pancreatic-enteric anastomosis (3) Completion pancreatectomy (4) Control of hemorrhage due to pseudoaneurysm
Examples here would include death related to PF as death associated with (1) Intra-abdominal abscess (2) Intra-abdominal bleeding (3) Percutaneous drainage (4) Reoperation for PF *
Had to be interpreted as listed in this table because not defined specifically in the original ISGPF 2005 definition (Bassi, 2005#11).
association of surgical group with a leak event while controlling for the stratification variable and variables found to be imbalanced between the groups at baseline. Note that because there are a small number of events, at most two variables could be used to avoid overfitting and ensuring model convergence. As such, multivariable models were created with the treatment assignment and one additional variable to determine the impact on the treatment arm assignment. In the secondary analysis comparing the 90-d leak grade (no leak, A, B, C) between treatment groups, unavailable analysis was done using a chi-square test. Multivariable
logistic regression was performed to assess the association between surgical group and 90-d leak grade (none or A, B, C) to determine the impact of the stratification variable and to determine the impact of imbalances in the baseline variables, American Society of Anesthesiologist (ASA) class (class 3 versus 1, 2), enrollment site, and diagnosis group (high risk versus low risk). Additional secondary analysis was performed to compare the rate of major PF-related morbidity (MPFRM) between the two treatment groups. MPFRM was defined as having any of the following events: readmission for PF, percutaneous
36
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
drainage, pseudoaneurysm, intensive care unit and/or progressive care unit stay, grade C PF, grade 3A/3B complication, postoperative hemorrhage, or reoperation for PF. Any morbidity was defined as MPFRM or any of the following events: any drainage longer than 3 wk, drain amylase greater than three times the upper limit of normal, specific treatment for PF, signs of infection, PF documented by X-ray, any grade of complication, antibiotic treatment, and grade B PF.
and were consented and randomized. All 68 received the allocated treatment. One patient withdrew from the study after surgery. Overall, 67 patients were assigned to a treatment arm and completed follow-up (Figure): 35 on the TissueLink arm and 32 on the SEAMGUARD arm. Of the two institutions, institution A enrolled 22 patients and institution B enrolled 45 patients.
Demographics and preoperative characteristics
Results The trial enrolled patients from January 2010 to March 2014, inclusive. The trial was closed before targeted enrollment was reached due to slow rate of patient accrual. There were 316 patients assessed for trial eligibility of which 248 patients were excluded: 85 patients due to predetermined exclusionary criteria, 32 due to patient preference, and 131 due to various other reasons. Sixty-eight patients met all inclusion criteria
Overall, 58.2% of patients were female; 65.7% for TissueLink and 50% SEAMGUARD (P ¼ 0.19). Median age for the overall cohort was 57 y; 54 for TissueLink and 63 for SEAMGUARD (P ¼ 0.26). Median body mass index was 27.58; 28.6 for TissueLink and 27.38 for SEAMGUARD (P ¼ 0.37; Table 2). Between treatment arms, there was no statistically significant difference with respect to the incidence of preoperative coronary artery disease, hypertension, chronic obstructive pulmonary disease, or diabetes (all P > 0.05). The
Figure e Study designdrandomization and follow-up. (Color version of figure is available online.)
37
shubert et al trial comparing plats
Table 2 e Demographics and preoperative characteristics. All (n ¼ 67)
SEAMGUARD (n ¼ 32)
TissueLink (n ¼ 35)
Gender, no. (%)
0.19
Male
28 (41.8)
16 (50.0)
12 (34.3)
Female
39 (58.2)
16 (50.0)
23 (65.7)
Age Mean (SD) Median (IQR)
0.26 57.99 (12.79) 57 (49-69)
59.84 (12.73)
56.29 (12.79)
63 (53.5-69)
54 (46-70)
20-78
20-78
34-78
Mean (SD)
28.19 (5.17)
27.60 (5.08)
28.74 (5.27)
Median (IQR)
27.58 (24.3-31.65)
27.38 (23.65-30.1)
28.6 (24.6-34.2)
18.14-40.4
19.5-40.4
18.14-38.6
Range BMI
Range
P value
0.37
<0.01
Diagnosis, no. (%) Adenocarcinoma
11 (16.4)
7 (21.9)
Neuroendocrine
19 (28.4)
12 (37.5)
4 (11.4) 7 (20.0)
Metastasis
6 (9.0)
3 (9.4)
3 (8.6)
IPMN
5 (7.5)
2 (6.3)
3 (8.6)
Mucinous cystic neoplasia
7 (10.4)
0 (0.0)
7 (20.0)
Cystic lesion other than IPMN/MCN
5 (7.5)
1 (3.1)
4 (11.4)
Chronic pancreatitis
4 (6.0)
4 (12.5)
0 (0.0)
10 (14.9)
3 (9.4)
7 (20.0)
5 (7.5)
3 (9.4)
2 (5.7)
34 (50.7)
17 (53.1)
17 (48.6)
0.71
5 (7.5)
1 (3.1)
4 (11.4)
0.36
14 (20.9)
7 (21.9)
7 (20.0)
Other CAD, no. (%) Hypertension, no. (%) COPD, no. (%) Diabetes, no. (%) ASA class, no. (%)
2 (3.0)
0 (0.0)
2 (5.7)
2
45 (67.2)
18 (56.3)
27 (77.1)
3
20 (29.9)
14 (43.8)
6 (17.1)
Serum albumin
Mean (SD) Median (IQR) Range
0.85 0.02
1
Missing*
0.66
0.64 34 4.42 (0.41)
23 4.37 (0.36)
21 4.45 (0.45)
4.5 (4.2-4.7)
4.2 (4.2-4.5)
4.5 (4.3-4.7)
3.2-5.1
3.9-5.1
3.2-5
BMI ¼ body mass index; CAD ¼ coronary artery disease; COPD ¼ chronic obstructive pulmonary disease; IPMN ¼ intraductal papillary mucinous neoplasm; MCN ¼ mucinous cystic neoplasm; SD ¼ standard deviation. * No. (%) used reported when not 100%.
ASA status was found to differ between the treatment arms with the SEAMGUARD cohort having a greater proportion of ASA class III patients (P ¼ 0.02). Median preoperative serum albumin was 4.5 (interquartile range [IQR] 4.2-4.7) and was not significantly different between treatment arms (P ¼ 0.64). Indications for operation were varied and were found to statistically differ between the treatment arms (P < 0.01; Table 2).
Operative characteristics Overall, 68.7% of operations were performed laparoscopically; 77.1% TissueLink and 59.4% SEAMGUARD (P ¼ 0.12). The operations were performed by a number of different surgeons with no one surgeon having performing significantly more
operations than the others (P ¼ 0.98). Conversion to an open operation was performed in 15.9% of patients: 8.0% TissueLink and 26.3% SEAMGUARD (P ¼ 0.21). The median operative duration was 199 min (IQR 141-262), and median estimated blood loss was 200 cc (IQR 100-400) for both arms combined; these variables did not differ between the treatment arms (Table 3). Most patients (73%) had concurrent organ resections at the time of distal pancreatectomy, 63% TissueLink and 84% SEAMGUARD (P ¼ 0.06). The most common solitary concurrent organ resection was splenectomy (n ¼ 38, 56.7 %). In addition to splenectomy, 10 (15%) patients had additional organs resected at the time of distal pancreatectomy. When compared between treatment arms, there was no difference in concurrent resections (Table 3).
38
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
Table 3 e Operative characteristics. All (n ¼ 67)
SEAMGUARD (n ¼ 32)
TissueLink (n ¼ 35)
Open versus laparoscopic, no. (%)
0.12
Open
21 (31.3)
13 (40.6)
8 (22.9)
Lap
46 (68.7)
19 (59.4)
27 (77.1)
10 (14.9)
5 (15.6)
5 (14.3)
Multiple organs resected, no. (%) Number of organs resected Mean (SD)
1.29 (0.68) 1 (1,1)
Range
1 to 5
1.22 (0.42) 1 (1,1) 1 to 2
1.38 (0.92) 1 (1,1) 1 to 5
Additional organs resected, no. (%)
Spleen
0.06 1 (1.5)
0 (0.0)
1 (2.9)
38 (56.7)
22 (68.8)
16 (45.7)
Spleen, portion of diaphragm, segment II liver, and gastric wedge resection
1 (1.5)
0 (0.0)
1 (2.9)
Spleen and Gallbladder
6 (9.0)
5 (15.6)
1 (2.9)
Spleen and partial gastrectomy
2 (3.0)
0 (0.0)
2 (5.7)
Spleen and adrenal gland
1 (1.5)
0 (0.0)
1 (2.9)
18 (26.9)
5 (15.6)
13 (37.1)
1
4 (6.0)
3 (9.4)
1 (2.9)
3
1 (1.5)
1 (3.1)
0 (0.0)
8
17 (25.4)
9 (28.1)
8 (22.9)
9
13 (19.4)
4 (12.5)
9 (25.7)
10
5 (7.5)
2 (6.3)
3 (8.6)
12
5 (7.5)
3 (9.4)
2 (5.7)
13
8 (11.9)
2 (6.3)
5 (14.3)
14
14 (20.9)
7 (21.9)
7 (20.0)
None Surgeon, no. (col %)
0.98
Duration of operation (min) Mean (SD) Median (IQR) Range
0.54 215.82 (84.53) 199 (141-265) 97-528
222.53 (91.22) 204 (155.5-265.5) 97-528
209.69 (78.75) 194 (138-259) 116-434
Est blood loss (mL) Mean (SD) Median (IQR) Range
1.00 0.43
Median (IQR)
Nephrectomy
P value
0.50 331.42 (351.71) 200 (100-400) 0-1700
362.87 (373.43) 200 (100-600) 0-1700
303.57 (334.27) 200 (100-350) 0-1600
SD ¼ standard deviation.
Pancreatic leak Overall, 37 of 67 (55%) patients experienced a leak of any grade, (95% CI: 37.46%-73.68%). Fifteen (46.9%; 95% CI: 27.08%-64.65%) in the SEAMGUARD arm and 22 (62.9%; 95% CI: 34.89%-90.08%) in the TissueLink arm. The difference in experienced leak rate of any grade between the groups was 16.0% (95% CI: 7.81%-25.43% P ¼ 0.19). Twelve patients experienced a clinically significant leak yielding a rate of 17.9% (95% CI: 8.72%-27.08%). Eight patients (22.9%; 95% CI: 8.99%-36.8%) in the TissueLink group experienced a clinically significant leak compared with four patients (12.5%; 95% CI: 1.04%-23.96%) in the SEAMGUARD group. The difference in clinically significant leak
rate between the groups was 10.4% (95% CI: 1.22%-19.58% P ¼ 0.35). The univariable OR for a clinically significant leak (TissueLeak versus SEAMGUARD) was 2.07 (95% CI: 0.567.70). The OR for treatment group in a multivariable model adjusting for enrolling site was 2.11 (95% CI: 0.57-7.89), the OR for treatment group in a multivariable model adjusting for ASA class was 2.39 (95% CI: 0.59-9.63), and the OR for treatment group in a multivariable model adjusting for indication for surgery (i.e., diagnosis) was 1.71 (95% CI: 0.44-6.57). All these estimates of treatment effect were similar, indicating little impact of baseline imbalances. The observed difference in leak rates between the arms was not statistically significant, which may be due to low power.
39
shubert et al trial comparing plats
Table 4 e Postoperative outcomes. Outcomes no. (col%)
No. (col %) All (n ¼ 67)
SEAMGUARD (n ¼ 32)
TissueLink (n ¼ 35)
P value
PFdany grade
37 (55.2)
15 (46.9)
22 (62.9)
0.19
Clinically significant PF
12 (17.9)
4 (12.5)
8 (22.9)
0.35
*
Drainage longer than 3 wk Missing Yes
3
2
1
14 (21.9)
5 (16.7)
9 (26.5)
0.34
Drained amylase 3 normal upper limit
38 (56.7)
15 (46.9)
23 (65.7)
0.12
Specific treatment for PF
0.35
12 (17.9)
4 (12.5)
8 (22.9)
Readmissions for PF
9 (13.4)
4 (12.5)
5 (14.3)
1.00
Percutaneous drainage
9 (13.4)
4 (12.5)
5 (14.3)
1.00
PF documented by X-ray
9 (13.4)
4 (12.5)
5 (14.3)
1.00
Pseudoaneurysm due to PF
1 (1.5)
0 (0.0)
1 (2.9)
1.00
26 (38.8)
12 (37.5)
14 (40.0)
0.83
5 (7.5)
1 (3.1)
4 (11.4)
0.36
Postoperative CT scan Postoperative sonogram Postoperative antibiotic treatment
15 (22.4)
7 (21.9)
8 (22.9)
0.92
4 (6.0)
3 (9.4)
1 (2.9)
0.34
Any morbidity
48 (71.6)
21 (65.6)
27 (77.1)
0.30
Major PF-related morbidity
19 (28.4)
10 (31.3)
9 (25.7)
ICU or PCU stay
Postoperative LOS Mean (SD) Range
0.62 0.31
5.45 (2.61)
5.94 (3.09)
5.00 (2.03)
3-15
3-13
3-15
CT ¼ computed tomography; ICU/PCU ¼ intensive care unit/progressive care unit; LOS ¼ length of stay; SD ¼ standard deviation. * No. (%) used reported when not 100%.
Other postoperative outcomes There were no postoperative mortalities in either treatment arms. Overall, 38 patients (56.7%) were found to have a drain amylase three times the upper limit of normal. Of the 12 patients (17.9%) requiring specific treatment for PF, nine (13.4%) required percutaneous drainage. Twenty-six patients (38.8%) required postoperative computed tomography scan, and 15 (22.4%) required postoperative antibiotics. Overall, four patients (6.0%) required care in a monitored setting (intensive or progressive care unit). Only one patient experienced pseudoaneurysm (Table 4). Average length of stay was 5.5 d (range 3-15), and readmission occurred in nine patients (13.4%); however, neither of these variables statistically differed between treatment arms (Table 4). The rate of MPFRM was found to be 28.4% (n ¼ 19): 25.7% (n ¼ 9) for TissueLink and 31.3% (n ¼ 10) for SEAMGUARD (P ¼ 0.62). The rate of any morbidity was 71.6% (n ¼ 48): 77.1% (n ¼ 27) for TissueLink and 65.6% (n ¼ 21) for SEAMGUARD (P ¼ 0.30; Table 4). Multivariable analysis (adjusting for enrolling site, ASA, and diagnosis) comparing TissueLink with SEAMGUARD did not reveal a significant association of treatment arm with any grade PF (OR ¼ 0.69, 95% CI, 0.23-2.04; P ¼ 0.50); ASA class (OR ¼ 1.42, 95% CI, 0.43-4.67; P ¼ 0.57); enrollment site (OR ¼ 2.80, 95% CI, 0.87-8.97; P ¼ 0.08); or diagnosis group (OR ¼ 0.67, 95% CI, 0.20-2.22; P ¼ 0.52). Clinically significant
fistula was also not found to be independently associated with treatment arm (OR ¼ 1.93, 95% CI, 0.47-7.99; P ¼ 0.37).
Conditional power analysis Conditional power analysis was performed given the findings and early closure of the study. Under the conditions of the study and the current observed difference in clinically significant leak rates between the two treatment groups, it would require 422 total patients to reach statistical significance assuming a two-sided, two-sample t-test comparing the two study arms with 80% power and 0.05 alpha level. This estimation was found to be similar to the power estimation determined before enrollment.
Discussion This is the first multicenter, randomized trial to evaluate the incidence of leak rate after distal pancreatectomy between two separate methods of pancreatic transection and remnant stump management. Although this study closed early due to poor patient enrollment, this remains a significant scientific effort among two large academic medical centers and, at the time of the study, the first ever contribution of level 1 evidence to this source of morbidity after distal pancreatectomy. Although there was an absolute difference in clinically significant leak rates between arms (22.9% TissueLink versus 12.5% SEAMGUARD), statistical significance was not reached.
40
j o u r n a l o f s u r g i c a l r e s e a r c h n o v e m b e r 2 0 1 6 ( 2 0 6 ) 3 2 e4 0
The lack of statistical significance is likely due to a lack of power (insufficient sample size). This can be seen by the 95% CI of the difference in leak rates between groups, which contains clinically meaningful values and the results of the conditional power analysis, which suggest that a statistically significant difference would be obtainable had a sufficient number of patients been enrolled. This is also different than the results of our previous porcine model.9 The reason for this difference is not certain, although plausible to be related to the differences in porcine and human pancreas and the fact that this study was performed on diseased pancreatic parenchyma, which is likely to affect the incidence of fibrosis and ductal dilatation.
Limitations This study has the obvious limitation of insufficient sample size. However, given the accrual rate of patients over the time of the study, the observed difference in leak rates and results of the conditional power analysis, it would have been impractical, and neither financially nor technically feasible, to continue the study. In addition, had the exclusion criteria for pancreatic thickness been set at a level greater than the chosen 2 cm, it is plausible that accrual could have been greater. Although, this cutoff was chosen to create a homogeneous population with equally efficacious use of either treatment, as pancreatic parenchyma greater than 2-cm thick can create a contraindication for stapler transection. In addition, a standard staple height could have been mandated, although many surgeons do in fact vary staple height based on pancreasspecific factors, and we wanted to allow for this as it is current practice.
Conclusions Despite the study limitations and a nonsignificant difference in clinically significant leak rates between the treatment arms, this represents the first ever multicenter randomized controlled trial investigating the outcomes of distal pancreatectomy using two contemporary treatment approaches. We have here demonstrated that if the trail had continued and reached target enrollment, a significant difference in leak rate between treatment arms may have been possible given the observed differences; it is unlikely that this sample size is feasible given the accrual rates observed and power required. Therefore, reaching statistical significance in a prospective randomized controlled fashion using these two treatment approaches is limited. Given lack of these definitive results, it is likely that both treatment options will continue to represent the current standard of care and should be chosen based on surgeon comfort, experience, and pancreas-related factors.
Acknowledgment Authors’ contributions: C.R.F., C.F.-d.C., M.L.K., M.B.F., R.L.S., M.J.T., and F.G.Q. contributed to hypothesis generation, trial design, patient enrollment, and trial participation. C.R.S., C.R.F., C.F.-d.C., and F.G.Q. contributed to data collection and statistical analysis. All the authors contributed to preparation of the article.
Disclosure The authors have nothing to disclose.
references
1. Fahy BN, Frey CF, Ho HS, Beckett L, Bold RJ. Morbidity, mortality, and technical factors of distal pancreatectomy. Am J Surg. 2002;183:237e241. 2. Ferna´ndez-del Castillo C, Rattner DW, Warshaw AL. Standards for pancreatic resection in the 1990s. Arch Surg. 1995;130:295e299. discussion 299-300. 3. Lillemoe KD, Kaushal S, Cameron JL, Sohn TA, Pitt HA, Yeo CJ. Distal pancreatectomy: indications and outcomes in 235 patients. Ann Surg. 1999;229:693e698. discussion 698-700. 4. Kooby DA, Gillespie T, Bentrem D, et al. Left-sided pancreatectomy: a multicenter comparison of laparoscopic and open approaches. Ann Surg. 2008;248:438e446. 5. Jimenez RE, Mavanur A, MacAulay WP. Staple line reinforcement reduces postoperative pancreatic stump leak after distal pancreatectomy. J Gastrointest Surg. 2007;11:345e349. 6. Thaker RI, Matthews BD, Linehan DC, Strasberg SM, Eagon JC, Hawkins WG. Absorbable mesh reinforcement of a stapled pancreatic transection line reduces the leak rate with distal pancreatectomy. J Gastrointest Surg. 2007;11:59e65. 7. Ferrone CR, Warshaw AL, Rattner DW, et al. Pancreatic fistula rates after 462 distal pancreatectomies: staplers do not decrease fistula rates. J Gastrointest Surg. 2008;12:1691e1697. 8. Blansfield JA, Rapp MM, Chokshi RJ, et al. Novel method of stump closure for distal pancreatectomy with a 75% reduction in pancreatic fistula rate. J Gastrointest Surg. 2012;16:524e528. 9. Truty MJ, Sawyer MD, Que FG. Decreasing pancreatic leak after distal pancreatectomy: saline-coupled radiofrequency ablation in a porcine model. J Gastrointest Surg. 2007;11:998e1007. 10. Connor S, Alexakis N, Garden OJ, Leandros E, Bramis J, Wigmore SJ. Meta-analysis of the value of somatostatin and its analogues in reducing complications associated with pancreatic surgery. Br J Surg. 2005;92:1059e1067. 11. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery. 2005;138:8e13. 12. Pancreas Club Calculator ISGPS leak definition. Available at: http://pancreasclub.com/calculators/isgps-calculator/