INTRODUCTION

Energy-based devices allow safe and effective sealing and closure of blood vessels (¹-³). They are widely used in laparoscopic surgery where conventional ligation techniques can extend procedural times (⁴,⁵) and even reduce intraoperative bleeding (⁴,⁶).

Temperature-controlled bipolar electrocoagulation (BE) and the harmonic scalpel (HS) are among the main energy-based hemostatic devices currently available. BE achieves tissue sealing and coagulation through thermal energy that is generated based on the impedance of clamped tissues and estimated by the instrument itself. It can effectively seal blood vessels with diameters < 7 mm (⁴).

The harmonic scalpel (HS) achieves hemostasis by utilizing ultrasound energy produced by a piezoelectric material at the tip of the forceps that generates enough thermal energy for tissue coagulation and transection. This device allows the hemostasis of vessels up to 5 mm in diameter (⁷).

The effectiveness of both devices sealing blood vessels and other tissues is widely recognized in the field of abdominal, thoracic, and cervical surgery based on former studies (³,⁸,⁹). In the field of coronary revascularization surgery, these devices have also been used for the endoscopic dissection of the internal saphenous vein or radial artery resulting in good graft patency and a lower rate of wound infection (¹⁰,¹¹).

However, in lower limb revascularization surgery, the endoscopic dissection of the saphenous vein assisted by sealing systems with BE has shown inconsistent results. In some of the series published, lower graft patency has been observed possibly associated with the excessive manipulation of the vein with dissection tools (¹²-¹⁴). However, more recent studies have demonstrated that with greater experience in this technique, similar results to those obtained in open safenectomy can be achieved (¹⁵,¹⁶).

Nevertheless, none of these studies analyze the sealing strength of venous collaterals, and the mid-term efficacy and safety profile of different energy-based hemostatic systems for venous collaterals—subject to high pressure after their arterialization—is unknown to this day.

In a previous study conducted by our group, the efficacy and safety profile of both hemostatic devices were evaluated in venous collaterals of the saphenous vein (SV) in an in vitro model. Venous fragments were interposed in a circuit with a roller pump to simulate pulsatile flow, and pressure inside the system increased gradually until rupture occurred. Bursting pressure was recorded for both devices, and supraphysiological pressures were obtained (788.9 mmHg ± 455 mmHg for BE and 602.5 mmHg ± 363.1 mmHg for HS) without statistically significant differences between them. However, differences were seen in the location of the leakage area. Grafts sealed with BE were more resistant (¹⁷).

However, whether resistance of the sealing area after the healing period and the potential risk of aneurysmal degeneration in the graft can lead to mid-term complications is unknown.

To be able to evaluate the characteristics of these sealing devices in vivo, an effective animal model with the highest possible survival rate is necessary. This model should allow us to assess the mid-term efficacy profile of sealing saphenous vein collaterals in an arterial system. With this goal in mind, in this study, we have designed 2 different animal models and analyzed their feasibility in terms of ischemic time, survival, and postoperative complications.

METHODS

PROCEDURE

Surgical technique

All animals were anesthetized by veterinary personnel from IISLAFE. Preoperative medication included intravenous (IV) enrofloxacin 10 mg/kg, IV metoclopramide 0.5 mg/kg, IV ranitidine 2 mg/kg to prevent gastric ulcers, and IV meloxicam 0.25 mg/kg as an analgesic drug. Anesthesia was induced using intramuscular (IM) ketamine (50 mg/kg) and IM xylazine (5 mg/kg). Prior to the administration of drugs, the animal was immobilized by wrapping it with a cloth and injecting the premedication into the femoral biceps. Afterwards, the ear marginal vein was cannulated, and general anesthesia was maintained using 2.5 % to 3 % sevoflurane and IM boluses of fentanyl-fluanisone 0.2-0.3 ml/kg every 30 to 45 min.

The surgical technique consisted of a midline laparotomy approach, exposure of the retroperitoneum, and dissection of the infrarenal aorta and lumbar branches. After systemic heparinization with a bolus of sodium heparin (0.5 mg/kg of body weight), the infrarenal aorta was clamped immediately above the common iliac arteries. A fragment of the human SV was interposed according to the animal aorta models suggested.

In the first model described an aorto-aortic bypass was performed with the interposition of a venous fragment of approximately 3 cm with the collaterals under study (Fig. 1). The grafts were end-to-end sutured with both proximal and distal anastomoses being performed using continuous 6/0 polypropylene sutures with × 2.5 magnification.

Figure 1. Intraoperative image of the interposition graft model. 

In the second model described, an infrarenal abdominal aortoplasty was performed using a venous fragment with the collaterals to be sealed (Fig. 2). Continuous suturing of the graft was performed using the same material described above.

Figure 2. Intraoperative image of the aortoplasty model with SV patch. 

After completing the vascular sutures, the clamps were removed following prior flushing with 1 % heparinized saline solution. Afterwards, the venous collateral was sealed using one of the devices (BE or HS). Ligasure® (Maryland Sealer/Divider, Covidien, CO, United States) and the Valleylab® LS10 generator were used for BE sealing while the Sonicision® device (Cordless Ultrasonic Dissector, Covidien, CO, United States) was used for the HS. The integrity of the sealing was immediately checked after unclamping.

Hemostasis of the surgical site was verified, and the abdominal wall closed in layers using continuous sutures of 3/0 polyglycolic acid for the deeper tissues and continuous intradermal sutures with 4/0 polyglycolic acid for the skin.

RESULTS

The first aorta model (end-to-end interposition of an SV graft into the infrarenal aorta) was initially used as the technique of choice in the first 5 rabbits. However, 1 animal died while anesthesia was being induceed. The individual results for each subject are shown on table IA.

Table IA. Results and individual characteristics of experimental animals with the interposition graft model. 

The mean procedural time for the aorto-aortic bypass model was 75 min ± 8.23 min, with a mean ischemia time of 42 min ± 10.61 min.

The survival rate in this group at 7 days was 50 % (2 subjects). The rate of paraplegia in all rabbits operated using this model was 100 %. In the 2 surviving rabbits, early euthanasia had to be performed 3 weeks after surgery due to a loss of animal welfare due to weight decrease > 25 % following paraplegia. No intra-abdominal hemorrhage was reported in the autopsies of the 2 spontaneously deceased rabbits or in the 2 rabbits that were euthanized. In all the cases presented, both the sealing of the collateral and patency of the aorto-iliac sector were confirmed.

In this animal model, the resistance test of the sealing area could not be performed or bursting pressure was obtained as no animal survived the necessary 4-week postoperative period to complete the healing process.

After considering the possible causes of mortality and the higher rate of paraplegia than expected, experimentation was interrupted to reassess the protocol and refine the surgical procedure. Among the different alternative models analyzed to minimize peri-aortic dissection and clamp time, the model of infrarenal aortoplasty with SV patch was eventually the chosen one. With this model, the remaining 11 rabbits were operated on. The individual results for each subject are shown on table IB.

Table IB. Results and individual characteristics of experimental animals with the aortoplasty model. 

In this model, the mean procedural time was 66.33 min ± 13.53 min, which was shorter ­compared to the interposition graft model (p = 0.076). The mean ischemia time was 37.11 min ± 8.1 min, shorter compared to the aorto-aortic bypass model (42 min ± 10.61 min), but without statistically significant differences (p = 0.414).

The 7-day survival rate was 5 out of 9 animals (55.56 %), as there were 2 losses due to preoperative mortality during the induction of anesthesia. The rate of paraplegia was 25 %, which is significantly lower compared to the interposition graft model (100 %) (p = 0.03). The 4-week survival rate was 4 out of 9 animals (44.44 %). In 1 of the animals, early euthanasia had to be performed 7 days after the surgery due to a loss of animal welfare, and significant weight loss due to paraplegia. In all autopsies, intra-abdominal bleeding was ruled out with presence of fibrotic reaction around the collateral sealing area.

In the 4 survivors, a resistance test of the sealing area was run with mean rupture pressure for the BE device (n = 2) of 491.29 mmHg ± 132.59 mmHg and mean bursting pressure for the HS sealing system (n = 2) of 1608.88 mmHg ± 132.59 mmHg without statistically significant differences being reported.

Regarding the location of the leakage, for BE, in 1 case it occurred at the sealing area and in the other case at the aortic lateral wall while for the HS, 1 burst occurred at the area sealed by the device, and the other at the connections of the pressure recording system at 1515 mmHg.

No infectious complications or other adverse events to immunosuppression were reported in any of groups studied.

DISCUSSION

The use of energy-based sealing devices is effective to achieve hemostasis in arteries or veins in procedures like nephrectomy (⁸), hysterectomy (¹⁸) or thoracotomy (¹⁹). However, different bursting pressures have been reported between venous and arterial territories with venous pressures being lower (²⁰). Therefore, studies are needed to assess the efficacy profile of sealing in the mid-term in venous collaterals undergoing high-pressure systems, as in the case of lower limb revascularization surgery.

In a former study conducted by our group, non-inferiority of these sealing systems compared to conventional ligation was demonstrated in venous collaterals undergoing arterial pressures and with supraphysiological bursting pressures in an in vitro hemodynamic circuit (¹⁷).

However, to demonstrate their durability and, therefore, their mid-term efficacy profile, it's necessary to demonstrate their performance in an in vivo model that allows us to assess resistance to rupture in collaterals under physiological conditions once the usual inflammatory and healing periods have been met beyond the 3-week mark (²¹).

The rabbit aorta model is the best one thanks to its anatomical characteristics in its abdominal section with mean length and diameters (3 cm × 3.7 mm), which is consistent with segments of the ISV. Additionally, this model is cheaper and easier to use and maintain compared to the alternative canine carotid artery model.

In the study conducted by Ribas et al. (²²), a polydimethylsiloxane graft was implanted in New Zealand rabbits as a prosthetic bypass model in small-sized arteries. In that study, the survival rate was 46.9 %, and the paraplegia rate, 17.2 % in survivors.

In our study, the end-to-end aortic interposition graft model had a 50 % mortality rate at 7 days and a 100 % mortality rate at 4 weeks, as well as a 100 % paraplegia rate. This high rate of neurological complications due to spinal cord ischemia resulted in a postoperative period with more comorbidities due to difficulties moving the animal, lack of spontaneous urination that requiring daily emptying maneuvers, and progressive weight loss > 25 %. All these factors led to a loss of animal welfare and, therefore, the fulfillment of the endpoint criteria for euthanasia purposes.

A similar rate of complications has been described in other experimental studies with rabbits as a model of spinal cord ischemia. In the study conducted by Saba et al. (²³), 37 rabbits underwent endoluminal aortic clamping for 30 min to create a model of spinal cord ischemia. In the group without neuroprotection, the rate of paraplegia was 100 % while in the intra-arterial diltiazem group while the rate of paraparesis was 12.5 % (Tarlov score < 3) at 48 hours. In the study conducted by Oyar et al. (²⁴), the experiment was performed with 27 rabbits with rates of paraplegia of 100 % in the group without neuroprotection and 75 % at 48 hours in the adrenomedullin group. The severity of paraparesis was greater in animals with ischemia time > 30 min, which explains the higher rate of neurological complications associated with time (²³,²⁴).

Unlike the study conducted by Ribas et al. (²²) where the arterial model consisted of an aortic bypass and a new prosthetic material with 2 anastomoses in an end-to-end configuration to allow tissue reperfusion after the first anastomosis, our first model involved end-to-end anastomosis with straight interposition of the venous graft resulting in a longer ischemia time due to clamping of the aorta during both vascular sutures. This may partially justify the higher rates of mortality and paraplegia described.

Given the high morbidity and mortality rates reported with this model, and following the principle of refinement in laboratory animals, a second animal model was proposed involving an infrarenal aortoplasty with a human saphenous vein graft. This new model required 1 vascular anastomosis only and, therefore, a shorter spinal cord ischemia time. This, added to the fewer number of lumbar arteries that required temporary occlusion during the ischemic period, resulted in lower rates of paraplegia (25 %) in the surviving subjects.

The anatomy of the rabbit's arteries is similar to the human arteries in terms of the bifurcation of the abdominal aorta into common iliac arteries and the presence of lumbar branches (²⁵). However, in rabbits, the lumbar arteries have limited collateral circulation, thus contributing to more neurological complications during manipulation and temporary occlusion of the infrarenal aorta and its branches (²⁶).

The rabbit has 7 pairs of lumbar arteries all of which originate from the aorta posterior wall. The seventh pair is often the distal branch of the median sacral artery that, in 50 % of the cases, is caudal to L6. Aortic bifurcations are found in 53.3 % of cases and are cranial to L7 in rabbits (²⁷). This means that there is limited space for dissection at distal infrarenal aortic level if all lumbar branches are to be spared to prevent spinal cord ischemia.

In the second model, which involves a shorter clamped length of the aorta and only 1 anastomosis, a lower rate of paraplegia was seen. Additionally, the aortoplasty was performed as caudally as possible to spare the median sacral artery from which the last pair of lumbar arteries arise. In cases where the arterial ostium was located more caudally, reflux was controlled by placing 2 opposing 3/0 silk ligatures without circumferentially enclosing each artery as shown on figure 3. This reduced the reflux of the median sacral artery and allowed venous graft anastomosis in the aorta under optimal conditions.

Figure 3. Intraoperative image of middle sacral artery with reflux control using a pair of 3/0 silk ligatures without completely encircling the artery to prevent injuries. 

Both in the arterial interposition graft and in the aortoplasty model, effective hemostasis was demonstrated with no evidence of abdominal bleeding in all autopsies. However, only in the latter arterial model survivors were obtained after 4 weeks to run the resistance test of the sealing area. Bursting pressure with both devices was always supraphysiological with mean pressures of 491.29 mmHg for BE and 1608.88 mmHg for HS. However, no statistically significant differences were found.

In all autopsies, patency of the aortoiliac sector treated was confirmed except 1 case where early aortic thrombosis was found during the procedure and early euthanasia was decided based on the predetermined endpoint criteria. No bleeding or intra-abdominal hemorrhage were identified, and the sealing area remained intact in both the interposition graft and aortoplasty models.

As study limitations we should mention that the sample size was small for the bursting pressure test. Despite an initial sample size of 16 animals, the test could only be run on 4 rabbits. One of the main limitations was the low survival rate in the end-to-end bypass animal model. Nevertheless, the bursting pressure obtained was always supraphysiological (> 300 mmHg). However, no significant differences were found between the 2 devices due to the insufficient number of subjects used to run the resistance test.

In future studies, the animal model should be refined to obtain more survivors after the 4-week healing process. Although rabbits have a higher risk of spinal cord ischemia, they are considered the most suitable animal model due to their similar aortic diameter compared to the saphenous vein and because they meet the reduction criterion of avoiding using larger species, which would involve more technical difficulties from the veterinary postoperative care standpoint, as well as a series of bio-ethical implications.

This preliminary study allowed us to fine tune the protocol to assess the efficacy profile of sealing venous collaterals using energy-based devices with acceptable mortality and paraplegia rates in an in vivo rabbit aorta model.

However, despite partially addressing the challenge of early spinal cord ischemia, an insufficient number of subjects was obtained to allow proper comparisons between the 2 devices regarding bursting pressure due to overpressure. Therefore, in the coming future, the study will continue with the development of this second aorta model until a sufficient number of survivors is achieved to verify the mid-term efficacy of collateral sealing using energy-based systems.