Study settings
All hospitals belonging to the GIRCG are eligible to participate in the ADiGe Trial, irrespective of their hospital volume. A total of 9 Centers applied to participate, and, at the time of this publication, the study has already been approved by the local IRB of each Center (Verona University Hospital, the leading Center, Orbassano Hospital of Turin, San Raffaele Hospital of Milan, Morgagni Hospital of Forlì, Federico II University Hospital of Naples, Niguarda Hospital of Milan, S. Orsola-Malpighi Hospital of Bologna, Parma University Hospital, Modena University Hospital).
Eligibility criteria
All consecutive patients that undergo total (TG) or subtotal gastrectomy (STG) with curative intent, for histologically proven gastric cancer or esophago-gastric junction cancer (Siewert type II or III), are eligible in participating Centers from the beginning of the study until they reach reaching the accrual number for each arm.
Inclusion criteria:
- Esophageal involvement <= 2 cm
- Patients planned for upfront surgery or treated with a neoadjuvant/perioperative chemotherapy or chemo-radiotherapy
- Open, hybrid, laparoscopic or robotic approach
- All types of anastomoses (circular stapled, linear stapled, hand sewn)
Exclusion criteria:
- Refuse to sign informed consent
- Age <18
- severe heart disease (Heart failure New York Heart Association - NYHA Class IV)
- severe liver disease (Child-Pugh >= B7) [10]
- Pregnancy
- Metastatic disease
- Gastrectomy performed in an emergency condition
- Palliative surgery
- Operation different from total or subtotal oncological gastrectomy (e.g. pylorus preserving, proximal gastrectomy)
- <D1 lymph nodal dissection according to the Japanese Gastric Cancer Association guidelines 5th edition [11]
- Reconstruction different from Roux-en-Y or Billroth II
- Multiple organ resection (except for cholecystectomy)
- Gastric cancer with duodenal involvement
- Intraoperative Hyperthermic Intraperitoneal Chemotherapy
We included in the trial patients with limited esophageal involvement defining a threshold of 2 cm in order to guarantee a proper oncological treatment. According with the available evidences, a tumor that invades the esophagus for more than 2 cm has an increased incidence of mediastinal nodal involvement, thus requiring a thoracic approach [12,13].
Compared to 3 RCTs already published on this topic [14,15,16], we also included patients that undergo neoadjuvant/perioperative therapies, as these treatments are currently the standard of care for locally advanced gastric tumours. A 2018 meta-analysis on 9 studies evidenced that neoadjuvant therapy is not associated with increased morbidity and mortality as compared to upfront surgery [17]. These results have been confirmed also in a recent propensity score matching analysis [18].
No restriction on a previous abdominal surgery has been planned since data available in literature suggest no correlation with postoperative morbidity [19,20]. Moreover, considering that minimally invasive and open surgery outcomes are nowadays comparable, we decided to include both type of approaches.
We included only Roux-en-Y and Billroth II reconstruction because they have similar vulnerabilities: the anastomosis and the duodenal stump. On the other hand, Billroth I reconstruction has different morbidity due to the lack of a duodenal stump and a direct suture between the gastric remnant and the duodenum.
Cholecystectomy was not considered as an exclusion criterion because some evidence suggests its utility in a subset of patients undergoing gastrectomy [21] and it seems not to add morbidity or mortality to the main operation [22].
Who will take informed consent
Patients eligible for participation are informed about the trial by one of the surgeons during the preoperative visit at the outpatient clinic. Patients can agree to participate until the day before the operation by signing the informed consent.
Interventions
Intervention description
Operation and Intraoperative drop out
Randomized patients included in the control group (Group A) receive one abdominal drain (any type of drainage is allowed) placed below the liver, passing by the duodenal stump with the apex posterior to the esophago-jejunal or gastro-jejunal anastomosis. No drainage is placed in patients assigned to the experimental group (Group B). Towards the end of the operation the leading surgeon can classify a patient as an intraoperative drop out before knowing the randomisation arm, if one or more of the following criteria are met:
- Intraoperative finding of non-radically resectable disease (R2 resection - gross residual disease with gross residual tumor that was not resected) [23]
- Need for unplanned multiple organ resection (except for cholecystectomy)
- High risk anastomosis defined as: intraoperative test (e.g. methylene blue or pneumatic test) positive for leak or intraoperative evidence of a positive resection margin.
- Other intraoperative complication that has to be specified in the case report form (unintended intraoperative damage to major vessels and/or organs requiring reconstruction or resection, intraoperative bleeding requiring urgent transfusion, unexpected medical conditions interrupting or changing the planned procedure [24]).
Postoperative procedure
All patients follow the usual postoperative pathway of each participating Center except for abdominal drain management. Prophylactic drain in group A patients is daily evaluated checking for suspicious debt. In patients with a normal debt, a methylene blue test (the patient is asked to drink 200 ml of water/tea with 5 ml of methylene blue) is performed on postoperative day (POD) 4 and the test is considered negative if no blue staining of drain output is apparent within 60 minutes (the 200 ml coloured solution stays in the esophageal/gastric lumen and this is considered as an indirect sign of the absence of an anastomotic leak). After the blue test is confirmed as negative, the drain can be removed. Each Center is allowed to remove drain on whatever POD after a negative test. Drain management in patients with a suspicious debt or positive blue test is left to Center preference (no further blue tests are planned). Postoperative complications are recorded until POD 30 or in hospital if hospital stay is longer than 30 days.
Reoperation under general anaesthesia and/or percutaneous abdominal/thoracic drain placement during the first 30 PODs for any cause related to the previous gastrectomy are considered as an event in both groups. The cause-effect relationship decision is left to the lead site surgeon and in case of doubt the investigator can ask for support from the steering committee. This specification was considered mainly necessary in case of unrelated pathology acquired after discharge but within POD 30 (e.g. operation for road accident). After discharge, patients are evaluated in the outpatient clinic on POD 30 or in the subsequent week. Follow up calls are conducted by authorised medical or nursing staff at POD 60±7 and 90 (max +7 days) in order to assess patient’s clinical status.
Outcomes
The primary composite endpoint is the cumulative incidence of reoperation and/or percutaneous drainage placement by the 30th postoperative day.
Secondary endpoints are:
- Incidence, severity and time to diagnosis of anastomotic and duodenal leak
- Length of hospital stay
- Overall 30 days morbidity or in hospital if longer than 30 days
- Overall 90 days mortality
- 30 days readmission rate
Length of hospital stay is considered from the day of operation until discharge at home or at other facility, or death. Postoperative complications, including anastomotic and duodenal leak, are classified according to the International Consensus on a Complications List After Gastrectomy for Cancer [ 24] and stratified by severity using the Clavien-Dindo classification [25].
Participant timeline
The timeline of study events is displayed in Fig.1 and Fig.2. After a patient’s inclusion, a medical authorized staff member randomizes the participant using an online secure module on the International Gastrectomy Complications Database website (www.gastrodata.org), the day before the operation. Upon filling out the randomization form, an immediate reply is obtained, containing the study group. “Group A” includes patients with prophylactic drain placed at the end of the operation, “Group B” includes patients without any abdominal drain at the end of the operation. The ratio 1:1 is obtained using a computer-generated randomization scheme, equally stratified (1:1 ratio) for type of operation (STG or TG). Enrolment type is competitive and was capped by type of surgery.
Sample size
According to a systematic review published in May 2020 [9], the cumulative incidence of reoperation in two RCTs was 7/115=6.1% (95% CI 2.5-12.1%) in the drain group and 3/115=2.6% (95% CI 0.5-7.4%) in the no drain group. Cumulative incidence of additional drain was 1/86=1.2% (95% CI 0.03-6.3%) in the drainage group and 2/84=2.4% (95% CI 0.3-8.3%) in the no drain group. Assuming a 70% of overlap between the two procedures in the drain group, the estimated proportion of the composite outcome is 6.44%, while assuming a 35% overlap in the no drain group, the estimated proportion of the composite outcome is 4.16%.
Hence, we assumed a reference group proportion of 6.44%, and a treated group proportion of 10% under the null hypothesis of inferiority and 4.16% under the alternative hypothesis of non-inferiority. A sample size of 182 in each group, corresponding to 364 patients overall, achieves 80% power to detect a non-inferiority margin difference between the group proportions of 3.56% (= 10 - 6.44%), with a one-sided significance level of 5%. A power curve for treated group proportion ranging from 3% to 7% under the alternative hypothesis of non-inferiority is presented in Figure 3.
Considering a 10% drop-out rate, 404 patients (202 in each group) are needed. In order to have a balanced percentage between total and subtotal gastrectomy, recruitment will end at 202 patients for each type of gastrectomy.
Assignment of interventions
Sequence generation and concealment mechanism
Randomization plan has been generated using www.randomization.com program and no one of the investigators has access to the list. Participants are randomized in a 1:1 ratio, using a secure web-based randomization system to receive either intervention or routine care, stratified by type of gastrectomy.
Blinding
The leading surgeon and the patient are blinded to the arm assigned until gastrointestinal reconstruction is completed (and anastomosis integrity test is done, if applicable). Before discovering the arm assigned, the surgeon will decide whether the patient meet or not the dropout criteria. There is no blinding for the patient, care providers, or coordinating researcher after the operation.
Data Collection and management
All study-related information will be stored securely at each participating Center. Patient’s information will be stored in locked file cabinets in areas with limited access. Digital files are kept in password-protected applications.
A section on the www.Gastrodata.org secure web-based platform was developed by a specialized software firm (www.Fluxedo.com) to facilitate multicenter data collection. All data, including Center, surgeon, and patient data, are strictly anonymous and managed through secure codes. Cloud servers are compliant with the General Data Protection Regulation. An SSL certificate is installed to encrypt data exchanged between applicative and users. Dataset and applicative are accessible only with personal credentials owned by authorized investigators. Each participating Center have access only to its own dataset.
Data are collected until POD 91. Before recruitment process starts, investigators from each participating Center will practice with the Leading Center data manager in order to increase proficiency in data entry and regular audits will be held with all participating Centers.
Anonymization was guaranteed by removing personally direct identifiers from the dataset (e.g. using age instead of date of birth) and recoding indirect identifiers when considered at risk for a possible identification.
Statistical methods
Analysis population
The primary analysis will be performed on a modified Intention-To-Treat (mITT) population, including all randomized patients who have undergone gastrectomy. Patients will be analyzed according to the treatment assigned at pre-operative randomization.
A secondary analysis will be performed on the As-Treated Population, including the same patients enclosed in the mITT population, classified according to whether they actually underwent the prophylactic drain placement or not.
Another secondary analysis will be carried out on the Per-Protocol (PP) population, including only patients undergoing surgery according to the arm assigned at randomization, and followed-up for at least 30 days. Also patients dying before 30 days will be considered for the PP analysis.
Statistical analysis
Categorical variables will be presented as absolute and percent frequencies, continuous variables as mean and standard deviation when approximately symmetrically distributed and as median and interquartile range otherwise.
The proportion of patients undergoing re-intervention or percutaneous drain placement (primary endpoint) will be computed for each group with the corresponding 95% confidence interval. The difference between the primary end-point proportion in the treated group and in the control group will be computed with the corresponding confidence interval. If the upper limit of the 90% confidence interval of this proportion does not exceed the non-inferiority margin difference of 4.5%, the null hypothesis of inferiority will be rejected.
Further statistical analysis on the primary endpoint will follow a hierarchical approach: if the non-inferiority null hypothesis will be rejected, significance of differences in the primary endpoint between control and experimental arms will be further investigated with a Fisher’s exact test. The relation between the primary endpoint and treatment (drain placement or avoidance) will be further investigated using a multivariable Firth’s logistic model, adjusting for Center, sex, age, tumor stage, type of gastrectomy (total/subtotal).
To evaluate significance of differences between control and experimental arms, Fisher’s exact test or chi-square test will be used for nominal variables (reoperation, placement of additional percutaneous drain, anastomotic or duodenal stump fistula, postoperative mortality, hospital readmission), Wilcoxon-Mann-Whitney test for ordinal variables (fistula or complication severity), and t test (or the corresponding non-parametric test) for continuous variables (length of hospital stay, time from surgery to fistula detection). Statistical significance for secondary analyses will be set at p<0.05. Statistical analyses will be performed using Stata®/IC 16.0 for windows (StataCorp LLC, College Station, Texas, USA).
Oversight and monitoring
The Sponsor in collaboration with the Steering Committee decided not to establish a formal Data Monitoring Committee for this trial considering the following 5 points:
- This trial aims to evaluate the uselessness of prophylactic drain placement after gastrectomy, a clinical practice that is still widely used but that has been already been discouraged by international guidelines
- evidence in literature does not suggest a highly favorable or unfavorable result. The design of a non-inferiority trial was deemed necessary due to the small estimated difference between the two groups on the outcome considered for this study
- the inclusion criteria and the study design (leaving the decision on the POD of drain removal to each participating centre according to their usual clinical practice) consistently reduce the (already limited) risks
- allowing any type of drain in the study protocol (leaving the decision to each Center according to their usual clinical practice) does not define this study as a device study
- Patient population is above 18 years old
As all the participating Centers are part of GIRCG quality of surgery and postoperative care were considerate adequate for this Trial according to our previous studies [15,16 26,27].
The wide inclusion criteria and the balancing between type of resection aim to result in a trial as close as possible to the everyday Western clinical practice.
Considering that both arms include treatment commonly applied in daily clinical practice (drain placement and drain avoidance) under the inclusion criteria considered in this Trial, the IRBs (of each participating Center), in accordance with the legal service of all the participating Center, agreed to ensure ADiGe Trial using the hospital insurance of each Center. Therefore, in case of claims for damages the responsible would be the insurance coverage of assistance activities of the Center that recruited and treated the patient.
Safety and adverse event Monitoring
Any untoward medical occurrence in a patient without regard to the possibility of a causal relationship has been defined as an adverse event (AE). Investigators from each participating Center will record any AE from the day of operation until the end of the observation period. All AEs will be documented in the patient’s medical record and AE form including any workup or treatment needed and grading (according to the Common Terminology Criteria for Adverse Events), according to everyday clinical practice.
AE, expected to be related with the study treatment (bleeding from drain site, skin infection around the drain site and pain) will be also recorded in the study CRF. Any unexpected AE, considered by the investigator as causally related with the treatment will be recorded in the study CRF and reported to the steering committee for safety reporting purposes.
AE that meets the following criteria, will be recorded as serious adverse event (SAE): life threatening condition (immediate risk of death), severe or permanent disability and prolonged hospitalisation. All SAE will be recorded in the study CRF. Moreover, the investigators will report to the local IRB and the steering committee, as soon as possible, but in no event later than 2 working days, any SAE that is deemed by the local investigator to be probably or definitely related to the study treatment. The steering committee will then review the report and send it to the promoting Center’s IRB within 2 working days. SAEs, occurring after a patient is discontinued from the study, will not be reported unless the investigators will determine that they are relate to the study treatment. If the number of SAE is higher than reported in the recently published Gastrodata study on complications after gastrectomy [28], patient enrollment will be terminated immediately, and the steering committee will reassess the safety of the trial.
Ethical Committee approval
The design of ADiGe Trial was approved by the scientific committee of GIRCG. The IRB of Verona has approved the present protocol with the following code: 2245CESC. Prior to implementation, any protocol amendment will be approved by GIRCG scientific committee and IRBs of each participating Center. Study information in ClinicalTrials.gov will be updated accordingly.
Dissemination policy
Results of this trial will be published in a peer-reviewed journal. The abstract will be submitted for presentations at different congresses. Within 6 months of completion of the trial, the ClinicalTrials.gov site will be updated to include summary results.
Authorship is granted to authors who make important contributions to the creation of the final publication in accordance with recommendations from the International Committee of Medical Journal Editors and GIRCG group internal policy. Authors can contribute via written or physical help in this clinical trial.
The datasets used during the current study will be available from the corresponding author on reasonable request within 1 year after completion of the Trial.
A timeline of the overall study design is displayed in Figure 4.