Colonic Anastomotic Leak Model in Swine Open Access

Anastomotic leaks (ALs) are linked to increased mortality, higher reoperation rate, and prolonged hospital stay after gastrointestinal (GI) surgery. Colorectal anastomosis typically has higher leak rates ranging from 0.5 to 30% [1‒8] and mortality rates of 27–67% [4, 6, 7]. Experimental models involving several animal species have been developed to investigate colonic ALs (CALs), but pigs are favored due to the anatomic and physiologic similarities of the GI tract of humans and pigs [9, 10]. However, animal models of CAL have been, for the most part, unreliable with respect to time of suture dehiscence, peritonitis, and leakage [11‒14].

Previous experimental models based on unabridged anastomoses have proven ineffective, considering the low rate of spontaneous leaks reported at 25% in pigs that underwent lower anterior rectal resections [11‒14]. Another swine model introduced 5–21 mm tubes directly through the suture line into the colon to create an anastomotic defect [13]. Results showed that all animals subjected to a 21-mm defect developed an AL [13]. However, this model was limited by the need for intraoperative induction of leakage and the lack of reproducibility of results. Accordingly, two studies on colonic anastomosis showed different incidence of leaks (25% vs. 33.3%) despite the same defect size (18 mm) [12, 13]. In the present study, we describe a clinically relevant model of postoperative colonic AL in swine that could contribute to research in this field.

This protocol followed the guidelines of the Animal Welfare Act of 1966 and was approved by the Keenan Research Centre in the Li Ka Shing Knowledge Institute of St. Michael’s Hospital Animal Care Committee; protocol number 950. Seventeen Yorkshire pigs of both sexes with an average age of 3.8 months and an average weight of 29.9 kg (standard deviation [SD]: 3.5 kg) were divided into two groups: experimental (n = 11) and control (n = 6) (shown in online suppl. Fig. 1; for all online suppl. material, see https://doi.org/10.1159/000534580). Group size was determined based on the number of animals required to show statistical significance, taking into account sample sizes necessary for significance in previous swine models [14]. Allocation to experimental and control groups were completed using randomization. An exteriorized suture was pulled by vivarium technicians to induce CAL in the experimental group (shown in Fig. 1e) and left undisturbed in the control group. The surgeon remained blinded to group allocation. The total observation period of the study was 4 postoperative days. The endpoints were clinical, laboratory, behavioral, and macroscopic indicators of CAL, confirmed through relaparotomy.

Animals arrived at the vivarium facility 7 days prior to the initial laparotomy for acclimatization. Pigs were individually housed in double-sized stainless steel cages equipped with video surveillance. Their health status was assessed based on activity levels, appetite, signs of pain/discomfort, and emesis. Daily vital assessments were performed to ensure a normal heart rate, respiratory rate, and external temperature. The pigs were kept on a standard diet, controlled by animal technicians, and were fasted for 12 h prior to surgery to prevent aspiration; water was provided ad libitum.

Animals received an intramuscular administration of ketamine at 20 mg/kg (6–7 mL) (Ketalean, Bimeda-MTC Animal Health Inc., Distributed by Vetoquinol N.-A. Inc., Lavaltrie, QC, Canada), xylazine at 2 mg/kg (3–3.5 mL) (Rompun, Elanco Canada Limited, Mississauga, ON, Canada), and atropine sulfate at 1 mg/25 kg (1–2 mL) (Atropine, Teligent Canada, Mississauga, ON, Canada). Subsequently, the animals were intubated and mechanically ventilated with a tidal volume of 10 mL/kg (Ohmeda 7000, Division of Canadian Oxygen Limited, Rexdale, ON, Canada). General anesthesia was maintained with 2–3% isoflurane for the duration of the procedure. An intravenous catheter was inserted in an ear vein using a 20-g intravenous catheter and 0.9% saline was administered. Anesthesia level and cardiopulmonary function were monitored using jaw tone, pulse oximetry, and ECG.

The anterior abdominal skin was prepared with a povidone-iodine solution (Teva Canada Ltd., ON, Canada) and draped in sterile fashion. A 10-cm midline incision was performed and midportion of the descending colon was exteriorized (shown in Fig. 1b; online suppl. Fig. 2A). The total circumference of the colon was measured and recorded, and an AARON 1250 electrocautery machine (Bovie Medical, FL, USA) was utilized to perform an enterotomy encompassing 80% of the circumference of the colon (shown in Fig. 1c). A handsewn, single-layer, continuous suture anastomosis was performed using non-absorbable 2-0 polypropylene sutures (Ethicon, Somerville, NJ, USA). A needle was used to pierce two holes in a 4 cm × 4 cm sterile plastic film to pass both ends of the polypropylene suture. The plastic film was interposed between the parietal peritoneum and the anastomosis to prevent adhesion of the anastomotic site to the lateral abdominal wall (shown in Fig. 1d; online suppl. Fig. 2C). The ends of the suture were externalized through the right abdominal wall, passed through a small latex tube (Bard Catheters, Covington, GA, USA), and tied (shown in Fig. 1e; online suppl. Fig. 2B) to anchor and maintained the suture taut (shown in Fig. 1f).

A retro-rectus sheath nerve block was performed with bupivacaine hydrochloride (5 mg/mL) (Sterimax Inc., Oakville, ON, Canada) prior to closing the laparotomy. The linea alba was closed with continuous number 1 polyglactin 910 single-layer suture (Vicryl Johnson & Johnson Intl., Somerville NJ, USA). The skin was sutured separately.

A transdermal fentanyl patch (75 μg/h) (Sandoz Canada, Boucherville, QC, Canada) was applied to the animal’s skin for postoperative pain management. The fentanyl patch was secured using a bandage (3M Vetrap TM St-Paul, MN, USA) to minimize tampering. The animals were placed in their cages upon completion of the procedures.

Anastomotic dehiscence was performed 3–4 h after the surgical procedure by cutting and pulling the external portion of the sutures used for the anastomosis from the latex tube. In the control group, the plastic tube and the suture were left untampered to maintain the integrity of the anastomosis. All pigs were continuously monitored through the live feed surveillance cameras for behavioral changes related to CAL (online suppl. Table 2). Blood samples were obtained for white blood cell count (CBC), and potassium except when collection was not feasible due to poor blood circulation or overactivity.

Upon presentation of CAL symptoms or at the end of the 4-day monitoring period, the animals underwent exploratory relaparotomy. The midline incision was opened to assess for macroscopic signs of an AL including, spillage of enteric contents, fecal odor, purulent secretion, and local inflammation/peritonitis (online suppl. Table 1). The anastomosis was exposed for direct visualization of the suture line. The animals were euthanized with pentobarbital sodium (0.2 mL/kg) intravenously.

Continuous variables are reported using means and SD and categorical variables as frequencies. The Shapiro-Wilk test was conducted to determine normality for continuous variables. Non-parametric data were evaluated using the Mann-Whitney Wilcox test, whereas normally distributed data were evaluated using a t test. If a priori assumption existed for a variable in the literature, the appropriate one-sided test was conducted. Categorical variables were analyzed using a 2-sample test for equality of proportions allowing comparisons between the control and experimental cohorts. p value <0.05 was considered statistically significant.

All six pigs in the control group and three in the experimental group were females. Baseline characteristics of the animals are outlined in Table 1. Baseline and operative characteristics were similar among the control and experimental groups.

At least one behavioral indicator of CAL was present in eight experimental group animals, while no indicators were present in control animals. Three or more signs of CAL were present in 5/11 experimental animals. Therefore, relaparotomy was performed significantly earlier in experimental group compared to control, respectively, 1.82 days (SD: 1.08) versus 3.33 days (SD: 1.03), (p = 0.0159). Behavioral data, clinical parameters, laboratory data from blood samples obtained prior to the exploratory relaparotomy, and intraoperative findings from the relaparotomy are reported in Table 2 and online suppl. Table 3.

During the exploratory relaparotomy, CAL was confirmed in eight experimental pigs with gross contamination and severe inflammation/peritonitis. Seven animals in the experimental group presented fecal contamination at the midline incision (shown in Fig. 2). Complete dehiscence of the anastomosis was seen in all experimental animals except one. This was attributed to inadequate suture pull-out caused by a broken stitch. Furthermore, 3/11 of the experimental pigs showed no symptoms of CAL despite adequate suture pull-out. Assessment during relaparotomy showed that the leaks were walled off by surrounding viscera despite the interposition of the plastic film.

There were no leaks detected in the control group animals during the exploratory relaparotomy and the suture remained intact in all animals. One control animal presented with local inflammation at the midline and serosal fluid in the surrounding peritoneum. That animal had partial dehiscence of the midline laparotomy suture.

Colorectal anastomotic leakage is a severe postoperative complication that can lead to adverse outcomes such as peritonitis, sepsis, multiple organ failure, and emergency reoperations [1‒8]. Intraoperatively, ischemia and tension at the site of anastomosis are leading factor associated with the development of CAL [1, 15, 16]. The lack of clinically relevant animal models has constrained research on new technologies and interventions to reduce CAL and improve early diagnosis. Most models use healthy animals with significant physiologic resilience that prevents the occurrence of leakage at the anastomosis. Therefore, abridged anastomoses and the induction of local ischemia have been used to increase the likelihood of leak.

The present experimental model resulted in successful leaks in nearly three-quarters of experimental pigs (8/11) as confirmed during relaparotomy. This leak rate is much higher than the 23–50% reported in previous studies [11‒13, 17]. We attribute the higher leak rate to the pull-out of the entire suture and the use of the plastic film interposed between the colon and the parietal peritoneum. The latter prevented tamponade of the leak by surrounding tissue and viscera.

Our model resulted in behavioral changes, clinical symptoms, and laboratory findings of CAL in all experimental animals. The model also produced consistent results in the control group animals as they did not develop CAL or associated clinical signs. Interestingly, previous publications pertaining to CAL lacked a control group and did not outline their initial and final assessments [12, 13]. The similarity in intraoperative management of the two groups further corroborates the effectiveness of our model.

The information provided by live feed surveillance cameras facilitated continuous monitoring of the behavioral parameters of the animals and enabled detection of early clinical signs indicative of CAL, which is lacking in previous studies [12, 13]. Other studies reported early symptom onset at the 6th postoperative day [11‒13, 17]. In contrast, data collected from the present model showed that behavioral changes were noted within 1.82 (SD: 1.08) days of the surgical procedure in the experimental group. The earlier symptoms could be attributed to actively induced suture dehiscence of a previously intact anastomosis as opposed to awaiting spontaneous leakage. Therefore, the overall monitoring period in this study was relatively shorter than the minimum 7-day period outlined by previous literature [12, 13].

Concurrent with the results from previous studies, hypokalemia was shown in 44% of the experimental group, whereas potassium levels were unchanged or increased in the control group [11‒13, 17]. Additional clinical relevance was achieved by avoiding ischemia at the level of the anastomosis. Accordingly, surgeons refrain from performing anastomoses of ischemic segments of the GI tract. Surgeons also devote serious effort to obtaining seamless sutures during an anastomosis. Therefore, research models involving abridged sutures contradict current surgical practices.

We speculate that the technique used for AL in this model could allow researchers to control the time of leak onset and contribute to future investigation in this field. Although this model demonstrates clinical relevance, it has limitations inherent to animal studies. The need to interpose the plastic film as a foreign body between the colon and the parietal peritoneum could interfere with the inflammatory process during anastomotic healing. Moreover, the lack of histopathological analysis of the specimens limited investigation pertaining to the inflammatory response at the anastomosis. Longer observation before the relaparotomy and different time points for suture pull-out could provide critical information regarding the effectiveness of the model. Furthermore, image diagnoses used in clinical practice such as contrast computed tomography scans were not performed in our study.

In conclusion, this swine model allowed for the simulation of a clinically relevant AL with a success rate of approximately three-quarters, assessed through relaparotomy. The model allows the researcher to provoke anastomotic dehiscence at specific times during the postoperative period. Those important assets could contribute to future research on anastomotic leakage.

We sincerely thank Dr. Joao Rezende-Neto (St. Michael’s Hospital) for his surgical expertise and clinical oversight of the study, in addition to Danielle Bince and the rest of the vivarium staff at the Research Vivarium, St. Michael’s Hospital for their significant contribution to the conduct of this research.

The protocol for this study followed the guidelines of the Animal Welfare Act of 1966 and was approved by the Keenan Research Centre in the Li Ka Shing Knowledge Institute of St. Michael’s Hospital Animal Care Committee; protocol number 950. The ARRIVE II Author Checklist is appended as online suppl. material (Supp. 3).

Nour Helwa, Manaswi Sharma, Youssef Helwa, and Abdallah El-Falou are current employees of FluidAI Medical. This study received funding from FluidAI Medical. The funder was involved with study design, data collection and analysis, decision to publish, and preparation of the manuscript. All authors declare no other competing interests.

This research was funded in full by FluidAI Medical.

Nour Helwa, Youssef Helwa, and Abdallah El-Falou were responsible for study design, and Nour Helwa prepared the study protocol. Nour Helwa, Manaswi Sharma, Youssef Helwa, and Abdallah El-Falou carried out all study activities including sample collection, analysis, and data entry. Nour Helwa and Manasvi Sai Vanama conducted data management and analysis, with guidance from Abdallah El-Falou. Manasvi Sai Vanama and Nour Helwa drafted the main text, while Manaswi Sharma and Nour Helwa prepared the figures, and Manaswi Sharma carried out manuscript preparation. All authors critically reviewed and approved the manuscript.

The raw data supporting the conclusions of this article are available with this article’s online supplementary material (Supp. 2). Further inquiries can be directed to the corresponding author.