The subjects analyzed in this article are patients included into the “Safe D3 right hemicolectomy for cancer through multi-detector computed tomography (MDCT) angiography” trial (Regional Ethical Committee approval REK Sør-Øst No. 2010/3354) and registered at clinicaltrials.gov (NCT01351714). Patients were required to sign an informed consent form prior to inclusion. The trial includes mandatory preoperative 3D reconstruction of the vascular anatomy, a standardized definition of the D3 volume, a standardized surgical approach to the central (level of dissection III) lymph nodes, and a separate histopathological analysis of the D3 volume. Data collection is prospective. All these data points have been previously published [2,3,4, 9,10,11] but will be addressed in short below. While the regional ethical committee approval allows any mode of access (open , laparoscopic , and robotic assisted), the study started out through open access while laparoscopic and robotic-assisted surgery were introduced at a later point.
The question of chylous ascites was raised after a number of patients in the mentioned trial had undergone surgery, some of which developed CA postoperatively. The awareness of the relevance of chylous ascites after D3 surgery prompted a change in the trial protocol, and the change regarding the introduction of a routine fat-reduced diet as part of the study was submitted to the Regional Ethical Committee.
Consecutive patients operated with right colectomy and D3 extended mesenterectomy at Akershus University Hospital for right-sided colon cancer from September 2014 to January 2017 were prospectively evaluated. Medical records were collected from the electronic medical record systems DIPS (Copyright 1995–2016 DIPS ASA version 7.395) and Panorama (anesthesiology electronic record system included in DIPS). Two surgeons operated all patients.
Preoperative 3D reconstruction of the vascular anatomy
Every patient included in the study had a personalized reconstruction of the vascular anatomy in the central mesentery derived from the preoperative staging CT dataset (Fig. 1). Segmentation is manually performed by BVS using the FDA-approved Osirix MD v. 10.0.2 64-bit image processing application (Pixmeo, Bernex, Switzerland), Mimics medical image processing software ver. 184.108.40.2066, and 3-matic medical software ver. 220.127.116.11, both Windows 7 ultimate edition × 64 2018 (both from Materialize NV, Leuven, Belgium).
Definition of the D3 volume
The third level of dissection (D3) is defined as the volume of tissue limited by four lines connecting anatomical landmarks (vessel origins/confluences). The cranial border lies 5 mm proximal and parallel to the line connecting the confluence of the gastrocolic trunk of Henle and the origin of the middle colic artery (MCA). The medial border lies on the left side of the superior mesenteric artery (SMA). The caudal border is 10 mm distal to the line connecting the ileocolic artery origin (ICA) and ileocolic vein confluence. The lateral border is placed 10 mm parallel to the right-hand side of the superior mesenteric vein (SMV) [2, 3]. This definition was consistently used for postoperative division of the surgical specimen into the respective D3/D2 volumes.
Surgical approach to the central (level of dissection III) lymph nodes
The surgical technique has been previously described and encompasses removal of all fatty tissue surrounding the superior mesenteric vessels en bloc with the surgical specimen [4, 10]. Figure 1 depicts the preoperative 3D image (right) and the dissected D3 area at surgery in the same patient. Notice the vascular sheath of the superior mesenteric artery.
Consecutive patients included in the “Safe D3 right hemicolectomy for cancer through multi-detector computed tomography (MDCT) angiography” trial were divided into two distinct chronological groups (diagnostic and treatment groups). Group 1 (diagnostic group) consists of patients where surgical drains were placed at the time of surgery and biochemical testing was performed on the drain fluid (the volume registered and fluid tested for triglyceride and cholesterol levels) in order to diagnose CA. There were no dietary restrictions, and the diet was not documented.
Group 2 (treatment group) consists of patients after the incidence of chylous ascites was established and the intervention introduced; a routine FRD to all patients operated and lasting for at least three postoperative days. If chylous ascites developed the FRD was continued until the CA was resolved. Surgical drains were routinely placed and the volume of drain discharge was registered and drain fluid was tested for triglyceride and cholesterol levels.
The presence of milky, non-infectious discharge through the abdominal drain with triglyceride levels higher than 1.3 mmol/L and lower cholesterol levels than blood, was defined as chylous ascites. Blood triglyceride and cholesterol levels were taken only on the first postoperative day. The level of triglyceride and cholesterol in the drain discharge was analyzed on the morning of the first-, second- and third postoperative days and prolonged when needed. If the analyses did not raise suspicion of chylous ascites, the abdominal drain was removed on day 3. Volume of drain discharge was registered at 24-h intervals by nurses and noted in the electronic patient journal.
The fat reduced diet
In cooperation with a clinical nutritionist, a FRD was compiled and thereafter introduced as a prophylactic measure. An information pamphlet was made for staff and patients to secure equal treatment. This FRD is not fat free, but the amount of fat is reduced to 10–18 g daily depending on the patients’ nutritional requirements. Recommended duration of the intervention was 3 days, and after drain removal, fat was reintroduced into the diet. In the case of chylous ascites, the FRD was not discontinued until resolution.
Regression modeling was used to analyze differences between the group with FRD and the group with no dietary restriction. Incidence of chylous ascites was analyzed with a logistic regression model. A longitudinal analysis of the amount of drain fluid was conducted using a mixed effect model with patient-wise random intercepts to account for repeated measurements. The amount of drain fluid after the drain was removed was set to 0, in both the statistical analysis and the plots. Natural cubic splines were used to account for the non-linear trends in drain fluid over time. Time to removal of the drain was analyzed using a Cox regression model. The analyses were done in R (version 3.4.0). t test was used for comparison of data between the two groups.