Methods
Study design
Patients undergoing elective DP will be randomly allocated in a 1:1 ratio to no prophylactic abdominal drainage or prophylactic abdominal drainage after surgery, stratified based on the risk of POPF. This protocol was developed according to the SPIRIT guidelines [
20]. Total inclusion time of the study is planned to be 24 months from start of recruitment and the total study time 36 months. The study structure includes setup of sites (4–6 months), enrollment (24–30 months), and data analysis and reporting results (4 months).
In case of readmission related to surgery within 90 days after initial discharge, follow-up for secondary outcomes will be extended to the entire duration of readmission.
Study population
Adult patients with an indication for elective DP, with or without splenectomy, minimally invasive or open, for any indication.
Inclusion criteria
In order to be eligible to participate to the study, a patient must meet all of the following criteria:
-
° At least 18 years old;
-
° Elective indication for DP, with or without splenectomy, minimally invasive or open, with and without sparing of splenic vessels, for all indications;
-
° Fit to undergo surgery;
-
° Oral and written informed consent.
Exclusion criteria
Patients who meet any of the following criteria will be excluded from participation in this study:
-
° Pregnancy;
-
° DP as a secondary procedure during gastric or colonic resection;
-
° Colonic resection required for cancer extension (gastric resection allowed);
-
° Additional hepatic resection;
-
° Participation to another study with interference with study outcome;
-
° ASA 4 / WHO 3;
-
° Arterial resection other than splenic artery.
Indications for DP
Numerous indications for elective DP exist, and the most common indications are as follows: ductal adenocarcinoma, mucinous cystic neoplasm, intraductal papillary mucinous neoplasm, neuroendocrine tumor, solid-pseudopapillary neoplasms, chronic pancreatitis. All indications are included in this trial.
Sample size calculation
The PANDORINA trial is as a non-inferiority randomized trial, hypothesizing that the outcome (i.e., postoperative complications Clavien-Dindo score ≥ 3 and, secondarily, clinically relevant POPF, grades B and C) in patients undergoing DP without prophylactic abdominal drainage are non-inferior to those of patients with routine prophylactic abdominal drainage after surgery. The sample size is calculated according to the formula (available at
https://www.sealedenvelope.com/power/binary-noninferior/):
$$n=f\left(\alpha, \beta \right)\times \left[{\pi}_{\mathrm{s}}\times \left(100-{\pi}_{\mathrm{s}}\right)+{\pi}_{\mathrm{e}}\times \left(100-{\pi}_{\mathrm{e}}\right)\right]/{\left({\pi}_{\mathrm{s}}-{\pi}_{\mathrm{e}}-d\right)}^2$$
where
πs and
πe are the true percent “success” in the “drain” and “no drain” groups respectively.
$$F\left(\alpha, \beta \right)={\left[{\varPhi}^{-1}\left(\alpha \right)+{\varPhi}^{-1}\left(\beta \right)\right]}^2$$
And φ−1 is the cumulative distribution function of a standardized normal deviate.
Patients recruited in the PANDORINA trial will be randomized within the referred Institute with a 1:1 allocation ratio. The primary endpoint of the present trial is the rate of Clavien-Dindo ≥3 major complications, following the assumption: 2.5% one-sided significance level (
α), 80% power (1−
β), and a non-inferiority level of 8%. The expected success percentage in the intervention arm (no prophylactic abdominal drainage) of 77% and a success percentage in the control group of 70% (prophylactic abdominal drainage, based on the multicenter LEOPARD trial with a majority of minimally invasive procedures) [
21]. The needed number of patients was calculated with a sealed envelope: a total of 272 patients are required. Including a 3% of possible drop-out after randomization, the total required sample size is 282 patients (141 patients per arm).
The most relevant secondary endpoint is the rate of clinically relevant POPF (grades B and C). The sample size was calculated with the following assumption: 2.5% one-sided significance level (
α), 80% power (1−
β), and a non-inferiority level of 8%. The expected success percentage in the intervention arm (no prophylactic abdominal drainage) is 81%, and a success percentage in the control group (prophylactic abdominal drainage) is 75%. This difference of 6% less grade B/C POPF without abdominal drainage is based on the Van Buren trial, and the baseline risk of 25% grade B/C POPF is based on the recently published distal fistula risk score (D-FRS), a combined Dutch/Verona multicenter study [
22]. The needed number of patients was calculated with sealed envelope: a total of 274 patients. Including a 3% of possible drop-out after randomization, the total required sample size is 282 patients (141 patients per arm).
Based on the second (slightly lager) sample size calculation, we will include 282 patients so that conclusions can be drawn for both endpoints.
Stratification
The patients included in the present trial will be stratified in preoperative estimated high and low risk for grade B/C POPF. High-risk patients is defined as a pancreatic duct diameter > 3 mm and/or pancreatic height (at the neck) > 19 mm based on the D-FRS [
22]. Other patients are considered as low risk. Patients will also be stratified based on annual hospital volume of DP. Stratification will be as follows: high volume > 40 DPs annual and low volume will be ≤ 39 DPs annual.
Quality
Participating centers should perform at least 10 DPs per year for any diagnosis. Surgeons should have performed > 50 pancreatic resections (any type, any diagnosis) in the past 5 years and > 20 DPs for any diagnosis ever. All sixteen centers participating in the Dutch Pancreatic Cancer Group are invited.
Methods
Definitions
All definitions are displayed in the Additional file
1. The left pancreas is defined as the pancreatic portion (body and tail) located on the left side of the porto-mesenteric vein. Complications are classified according to the Clavien-Dindo score ≥ 3 [
26]. Grade B/C POPF [
19], delayed gastric emptying (DGE) [
27], and postpancreatectomy hemorrhage (PPH) [
18] are classified using the ISGPS definitions. Surgical site infection (SSI) is classified according to the Center for Disease Control and Prevention definition [
28].
Study parameters/endpoints
Main parameters/endpoint
Primary endpoint is the rate of Clavien-Dindo score ≥ 3 complications [
26].
Secondary parameters/endpoint
Secondary outcomes are grade B/C POPF, reoperation, catheter drainage, abdominal collections, wound infection, DGE, PPH, blood transfusion, length of stay (LOS), in-hospital mortality, 90-day mortality, readmission within 90 days, and start of adjuvant chemotherapy. Outcomes will be based on the Dutch Pancreatic Cancer Audit registry.
-
Intraoperative parameters:
-
° Date of operation;
-
° Splenectomy;
-
° Conversion;
-
° Reason for conversion;
-
° Vessel resection (excluding splenic vessels); if so, type of venous resection/reconstruction;
-
° Operative time (from first incision to closure of the abdomen), minutes;
-
° Intraoperative blood loss, mL (suction canister and weight of gauzes);
-
° Intraoperative blood transfusion, mL fresh frozen plasma or packed cells.
-
Postoperative parameters:
-
° POPF (grade B/C)*;
-
° Major complications (CD ≥ 3);
-
° DGE (grade B/C)*;
-
° PPH (grade B/C)*;
-
° SSI*;
-
° Postoperative intervention (radiology, endoscopy, surgery);
-
° Reason for postoperative intervention.
-
Hospitalization parameters:
-
° LOS, days;
-
° Readmission;
-
° Intensive care admission;
-
° Reason for intensive care admission;
-
° Duration of intensive care admission, days;
-
° CRP on 3rd postoperative day;
-
° Amount of days with drain in place.
-
See Additional file
1 for detailed definitions of study outcomes.
Other study parameters
Other study parameters are baseline characteristics, including:
-
Patient study identification number;
-
Date of birth;
-
Sex;
-
Performance status (Karnofsky score)*;
-
ASA physical status*;
-
Body mass index (BMI), kg/m2);
-
Abdominal surgery in history;
-
Diabetes mellitus;
-
Diagnosis based on preoperative imaging.
*See Additional file
1 for detailed definitions of study outcomes.
Randomization and treatment allocation
Eligible patients for the study will be identified at the outpatient clinic and in this stage informed consent will be obtained; the randomization will take place directly after start of the operation, when the decision is made to proceed with the resection. The surgeon performing the operation will call the study coordinator to perform the randomization in Castor EDC. Patients will be analyzed according to the allocated treatment, as per intention-to-treat principles. If during the operation is chosen not to act conform the randomization, specification of this choice is required. A per-protocol analysis will also be performed for the primary endpoint.
A data safety monitoring committee will assess safety endpoints per 50 randomized patients. All patients will be randomized centrally using an online computer-controlled permuted-block randomization module in a 1:1 ratio. The block sizes itself will be subject to random variation with block sizes varying from 4 to 8 patients. The entire randomization will be concealed to all involved investigators except the trial coordinators. Patients will be coded by a numeric randomization code and the principal investigator will be the only one with access to it and investigators and patients are not blinded. Patients will be stratified according to high or low risk and minimally invasive and open procedure. The source data will be stored digitally and will be kept by the project leader for 15 years after the inclusion of the last patient.
Study procedures
Patient screening procedures
Screening of patients consists of standard procedures, including high-quality computed tomography (CT) of the pancreas. Screening procedures will be according to the local treating team’s preference.
Data collection
Baseline characteristics will be recorded before randomization. The required clinical data, i.e., primary and secondary outcomes, will be collected after randomization, i.e., from hospitalization up to 6 months postoperatively using standardized case report forms, and in Dutch centers according to the mandatory Dutch Pancreatic Cancer Audit which is coded. All complications will be scored using the Clavien-Dindo score of surgical complications (see Additional file
1).
Registration of patients not included in the PANDORINA trial
All patients undergoing elective DP but not included in the trial will be logged using a screening log. These patients will be registered anonymously using a standardized prospectively collected database. Due to this, it will be possible to assess the presence of patient selection by indication.
Withdrawal of individual subjects
Subjects can withdraw from the study at any time for any reason if they wish to do so without any consequences. The investigator can decide to withdraw a subject from the study for urgent medical reasons.
Replacement of individual subjects after withdrawal
Patients withdrawn after surgery will not be replaced. If a patient does not receive a resection due to intraoperative metastasis or other reasons for not performing a DP, he will not be randomized since the randomization process takes place only after the intention to resect. Patients withdrawn because of treatment-related reasons will not be replaced.
Follow-up of subjects withdrawn from treatment
All subjects will be analyzed according to intention-to-treat principles and follow-up will be 6 months after DP.
Statistical analysis
Primary and secondary endpoints will be cross checked with data from primary sources and a blinded adjudication committee will check them against the definitions, which are established before the start of this study. Frequencies will be presented for dichotomous data. The primary endpoint Clavien-Dindo score ≥ 3 complications will be tested for non-inferiority using the chi-square test. The distribution of variables will be determined using several plots (boxplot, Q-Q plot, and histogram) and the Kolmogorov-Smirnov, Shapiro-Wilk, and Levene’s tests. For comparison of normally distributed continuous variables, the independent samples t-test will be used and values will be expressed as means (standard deviations). Continuous non-normally distributed variables will be compared using the Mann-Whitney U test and values will be expressed as medians (interquartile ranges). Categorical variables will be compared by chi-square or Fisher’s exact test as appropriate, and values will be expressed as proportions. A two-tailed p-value <0.05 will be considered statistically significant. Where possible, risk ratios with 95% confidence intervals will be reported. For the primary study outcome, a two-sided 95% confidence interval will be reported. Time to event endpoints, such as survival, will be calculated using Kaplan-Meier estimations. A Cox regression analysis will be performed to investigate predictors of postoperative survival. All parameters with a p-value <0.1 in a univariable analysis are included in the multivariable Cox regression analysis. Additionally, multivariable analyses are performed to determine predictors for primary and secondary study outcomes, for example R0 resection, the occurrence of postoperative pancreatic fistula. Intraoperative details and primary endpoint of this study are expected to be complete. All data entries should be sent to the project leader immediately after the final examination. In case of missing data, explanation should be given to the project leader. For subjects who are lost to follow-up, a sensitivity analysis will be performed to determine best-case/worst-case scenarios. A detailed statistical analysis plan will be drafted prior to database lock. Despite all prior preventive measures taken, a complex multinational trial may still evoke unforeseen situations after database lock that threaten data integrity and can only be resolved by unlocking the database prior to the final analysis. For purpose of transparency and reproducibility, the statistical analysis plan will therefore also describe the procedure to be followed when such situations arise.
Ethical considerations
Regulation statement
The PANDORINA trial will be conducted according to the principles of the Declaration of Helsinki (64th version, October 2013) and in accordance with the local laws and regulations, such as in the Netherlands the Medical Research Involving Human Subjects Act. The local principal investigator is responsible for making sure that local laws and regulations are followed.
Recruitment and consent
All subjects will be recruited at the outpatient clinic by one of the principal investigators. The principal investigator may be replaced by an assigned substitute, who is fully informed and aware of the study requirements and procedures, e.g., the study coordinator, local treating physician, or a study nurse. The approached patient will be given at least 24 hours’ time to consider informed consent and to make a decision. If indicated, surgery will be performed within 4 weeks after determination of the diagnosis, so there is sufficient time for explanation of the disease severity. Furthermore, due to this the patient will have sufficient time to explain her/his prospects and to consider the (dis)advantages of participating in the study. Participant retention and complete follow-up was promoted by periodic meetings and site visits with outcome data and corresponding time windows.
Objections by minors or incapacitated subjects
Minors and incapacitated patients will not be included in this study.
Benefits and risks assessment group
Prophylactic abdominal drainage can be useful for early detection of pancreatic juice in case of a POPF but drains might themselves also contribute to these complications. Patients will not undergo additional investigations and interventions due to participation in the PANDORINA trial and therefore risks to subjects involved in this trial are similar to every other patient undergoing DP in routine clinical practice, especially since the Van Buren multicenter RCT already demonstrated the safety of omitting prophylactic abdominal drainage. Potential benefits for subjects in the investigational treatment arm could be fewer major complications, less abdominal pain, and less discomfort in case of no prophylactic drainage.
Public disclosure and publication policy
No arrangements have been made concerning public disclosure and publication of the research data and outcomes. The trial was registered within Netherlands Trial Register (
https://www.trialregister.nl/trial/9116) with trial number NL9116 on 11-12-2020. The results of this trial will be submitted to a high-impact peer-reviewed medical journal regardless of the study outcome. Authorship will be based on the most recent international ICMJE guidelines. Next to these ICMJE guidelines, a minimum of 5 randomized patients is required for 1 co-authorship per participating center, a minimum of 20 randomized patients for 2 co-authorships and a minimum 40 randomized patients for 3 authorships. Per site, it is internally determined which local investigator will be author as long as this person fulfils the international ICMJE guidelines for authorship.
The study PhD coordinators will be the first authors. There is one trial coordinator responsible for the coordination of the trial for all Dutch centers (EAVB) and one trial coordinator for all Italian centers (AB). This includes recruitment, randomization, and follow-up of the study subjects as well as communication with the study sites. The penultimate and last authorships are for the three principal investigators (CB, CvE, MGB). All other authors will be listed in alphabetical order. Clinicians who are involved in this study and do not fulfil the previously mentioned authorship criteria will be listed as “collaborator” in the final manuscript and the medical journal will be asked to present the names of these collaborators in PubMed. For purposes protocol amendment, abstract presentation, and publication, any secondary publication will be delegated to the appropriate principal authors.
Discussion
PANDORINA is the first binational, multicenter, randomized controlled non-inferiority trial with the primary objective to evaluate the hypothesis that omitting prophylactic abdominal drainage after DP does not worsen the risk of postoperative severe complications [
18,
19]. The focus of this study is to assess if operative placed drains lead to less complicated POPF which need a reintervention or reoperation. However, POPF B/C without intervention cannot be measured in the no drain arm of our study since this is because of prolonged operative drainage. Therefore, the primary endpoint is chosen to be severe morbidity, while this is caused by a POPF grade B/C, on which the clinical focus will lie. Most of the published studies on drain placement after pancreatectomy focus on both pancreatoduodenectomy and DP, but these two entities present are associated with different complications and therefore deserve separate evaluation [
29,
30]. The PANDORINA trial is innovative since it takes the preoperative risk on POPF into account based on the D-FRS and it warrants homogenous stump closing by using the same graded compression technique and same stapling devive [
22,
24].
Several studies provide better insight into the value of abdominal drainage after PD. In a cohort of 69 patients after DP (39 with and 30 without a drain), Paulus et al. [
31] did not observe a decrease in the incidence of intra-abdominal abscess, POPF [
32], or pseudocyst formation in case of intra-abdominal drainage. In a randomized trial by Conlon et al [
1] in 179 patients undergoing pancreatic resection (among them 40 distal pancreatectomies), no benefit was seen for prophylactic abdominal drainage
. However, the statistical analysis was conducted without distinction between pancreatoduodenectomy and DP. Behrman et al. [
6] used propensity scored matching to compare 706 patients undergoing DP with or without prophylactic abdominal drainage. They observed a higher incidence of POPF in those patients who received drains; however, their analysis included the clinically non-relevant POPF Grade A [
33].
A randomized trial by Bassi et al. reported the outcome of early vs late removal of the intra-abdominal drain after pancreatectomy in 114 patients of whom 39 underwent DP [
33]. They found an increase incidence of POPF, intra-abdominal infected collections, pulmonary complications, length of stay (LOS), and readmission in the group of late drain removal. They concluded that late drain removal is a risk factor for POPF. This finding was confirmed in a recent study by Seykora et al. in which in 5581 DPs (POPF grade B/C rate 17%) early drain removal (in 716 patients) was associated with improved outcomes [
34].
Recently, Van Buren et al. reported the first multicenter randomized controlled trial in 344 patients undergoing DP with (
n=174) and without (
n=170) prophylactic intra-abdominal drainage [
2]. Their hypothesis was that DP without routine intra-abdominal drainage does not affect the frequency of grade 2 or higher-grade complication (according to the Common Terminology Criteria for Adverse Event [
23]). The rate of grade 2 complications was comparable between the groups (44% vs. 42%,
p=0.80). The rate of grade B/C POPF did not differ and was 6% lower in the group without abdominal drainage (18% vs 12%,
p=0.11) either. This led to the conclusion that clinical outcomes after DP with or without drain are comparable. The Van Buren trial does not describe a standardized technique of pancreatic stump closure and does not stratify patients to their risk of POPF after DP.
Asbun et al. described the now commonly used graded compression technique for DP [
17,
24]. By compressing the pancreatic tissue gradually based on the experienced resistance, the pancreatic parenchyma is compressed rather than crushed before transecting, which results in low rates of POPF [
17]. A reduction in POPF using this technique was confirmed in two other studies, one study showed that out of 42 patients undergoing DP, 17 were treated with the graded compression technique and did not develop POPF (0%) compared to 28% of the patients in the comparison group [
35,
36]. The intra-abdominal drainage period and the median LOS were significantly shorter in the graded compression group compared to the no-graded compression group.
In order to assess the rationale of prophylactic abdominal drainage after DP, we designed a binational multicenter randomized control trial (RCT). All patients undergoing DP with open and minimally invasive techniques (laparoscopic and robotic), with and without preservation of the spleen, are eligible for the study.
Conclusion
PANDORINA is a binational multicenter, randomized controlled non-inferiority trial with the primary objective to evaluate that omitting prophylactic abdominal drainage after DP does not worsen the risk of postoperative Clavien-Dindo score ≥ 3 complications [
18,
19]. Additional to the current literature, the PANDORINA trial makes prespecified fistula risk groups and guarantees homogenous stump closure.
Trial status
Confirmation of funding of the trial by Ethicon UK (Johnson & Johnson Medical Limited, Edinburgh, UK) was received on September 25, 2017. Ethical approval in the Amsterdam UMC was received on September 25, 2020. The PANDORINA trial was registered in the Netherlands Trial Register on December 11, 2020 (NL9116). The first patient was randomized on November 02, 2020. At the time of submitting this protocol for publication (May 4, 2022), all centers were actively recruiting patients for the trial and 152 out of 282 (54%) have been randomized, which means that inclusion is on schedule.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.