Keywords: Flap surgery animal model; ICG-NIR fluorescence imaging; MicroCT seroma imaging; Seroma prevention; Tissue adhesives.

Abbreviations: FG: Fibrin Glue; HES: Hematoxylin Eosin and Saffron; ICG: Indocyanine Green; LDM: Latissimus Dorsi Muscle; NBCA: N-Butyl-Cyanoacrylates Adhesive; NIR: Near Infrared; PBS: Phosphate-Buffered Saline; POD: Postoperative Day; QS: Quilting Sutures; TA: Tissue Adhesives; RLU: Relative Light Units; ROI: Region of Interest.

Introduction

Aggressive oncologic surgical procedures or flap elevations for reconstructive surgery are procedures removing considerable amounts of tissue thus damaging lymphatics and vascular channels and leaving a potential large space. These factors may result in formation of postoperative seroma [1]. Seroma is the most frequent surgical complication after procedures with large tissue elevation such as mastectomy [2], abdominoplasty, [3] or latissimus dorsi flap elevation with a frequency of as much as 79% [4,5]. It increases the risk of flap necrosis, wound infection and delayed wound healing and may cause pain or general patient discomfort [3]. They also often require multiple fluid aspirations and possibly additional surgical procedures thus resulting in socioeconomic costs, and possible extended hospital stay. Because of its perioperative morbidity and medico-social consequences, various strategies have been assessed to decrease the incidence of postoperative seroma. Postoperative compression [6], use of closed suction drain [7,8] obliteration of dead space with various flap fixation techniques, use of sclerosants [9,10], talc [11], tranexamic acid [12], fibrin glue and sealants [13-15] associated or not to progressive tension sutures [16] have been attempted with conflicting results and none have been consistent [17] These approaches were generally successful in reducing total seroma output and drainage duration, but not in totally preventing seroma formation [18,19]. To date, no defined protocol nor guidelines exist and surgical prevention technique varies among surgeons. Seroma remains a current clinical issue and may delay adjuvant therapies or impair patients’ quality of life. Efforts are still to be made, to find an innovative Tissue Adhesive (TA) that would prevent seroma production. An ideal TA should act as a sealant compressing vascular and lymphatics disrupted vessels and decreasing fluid accumulation. By bonding tissue layers, it may occlude the dead space that would lead to seroma formation, provide strength to the overlying skin flap and promote the flap immobilization thus preventing shearing forces responsible for the inflammatory effect [20,21]. As much as preventing seroma, it should also promote wound healing with an uneventful recovery. Difficulty in TA development lies in the absence of a defined experimental protocol to objectively and comprehensively evaluate a method of seroma prevention. The current models described in the literature focus essentially on the production or not of seroma with iterative punctures, and the evaluation of morbidity is limited to histological analysis [13], [22-26]. Moreover, the animal model mostly described is based on a mastectomy in rats, as described by Lindsey and Harada [22,23]. This model was not reproducible in our hands, due to both frequent reopening of the wounds by the rats themselves and failure to produce seroma consistently. Our aim was thus to develop a reliable experimental animal model as close as possible to the clinical situation and to precise the followup protocol to assess the efficacy and tolerance of various seroma prevention methods.

Materials and Methods

Ethics

All animal experiments were performed in accordance with the institutional guidelines of the European Community (EU Directive 2010/63/EU) for the use of experimental animals and were approved by an ethic committee (Cometh38 Grenoble, France) and the French Ministry of Higher Education and Research under the reference: APAFIS#34251-2022020915237986. The present study assessed various seroma prevention methods on 50 rats to which right Latissimus Dorsi Muscle (LDM) and axillary nodes harvest was performed. Fifty female Wistar rats between 280 and 350g were used. Comparison included 5 groups (n = 10 per group) including 3 control groups: group I no seroma prevention (Control Group, CG), group II internal Quilting Sutures (QS), group III fibrin glue (FG) (TISSEEL^® of high-thrombin formulation 500 IU thrombin/mL: Baxter Healthcare Corp, Deerfield, IL, USA) - which are the current methods used in practice, and 2 test groups: group IV VENASEAL^® (Medtronic, Inc, Minneapolis, Minn) and group V NEXPOWDER^® (Medtronic, Inc, Minneapolis, Minn).

Surgical Model

All rats were anesthetized with volatile anesthesia. Induction was performed with isoflurane 4% and anesthesia was maintained with isoflurane 2.5%. Each animal was premedicated with a subcutaneous injection of buprenorphine 0.01 mg/kg. The animals were placed in a left lateral decubitus position and the area of interest was shaved, scrubbed with povidone-iodine and locally infiltrated with lidocaine 5 mg/kg. A mediodorsal arched incision was made from the scapular angle to the last thoracic vertebrae. The right Latissimus Dorsi Muscle (LDM) was elevated from its origin along the thoracic and first lumbar vertebrae, up to its humeral insertion. Loose areolar tissue dissection was carried out by peeling the muscle from the subcutaneous tissue, alternating sharp and blunt dissection with fine scissors and gauze pad (Figure 1A). The muscle was individualized on its pedicle (Figure 1B). When identified, the thoracodorsal pedicle was ligated and the muscle removed. The surface of the harvested muscle was measured in cm². Homolateral axillary lymph nodes were removed. Ten subcutaneous scarifications were made on the deep surface of the skin flap to traumatized subcutaneous lymphatic vascularization and increase the risk of seroma (Figure 1C). Meticulous hemostasis was performed by digital pressure or ligation when needed. No electrocoagulation was used. At this point, the procedure varies according to group: no seroma prevention was performed in group I, quilting sutures with 8 stitches of absorbable 4-0 evenly distributed over the entire operated area in group II, application of 1 mL of fibrin glue with pressure of 1 min (group III), application of 8 points of VENASEAL^® distributed as the quilting sutures with a pressure of 5 min (group IV), and application of NEXPOWDER^® dusted all over the operated area with a pressure of 5 min (group V) (Figure 2). The wound was closed with a double-layer closure (subcutaneous and skin suture) with absorbable 4-0 interrupted suture. Postoperative analgesia was scheduled at 6 hours and 24 hours after surgery using buprenorphine (0.01 mg/kg). All procedures were performed by the same surgeon to prevent operator-dependent conditions. Each operative session included the same number of rats from each group to avoid a learning curve for the operator all along the study. After surgery, rats were housed in individual cages for the first 2 months and then grouped in pairs for the rest of the follow-up.

Figure 1: A: Skin flap aspect after peeling the muscle from the subcutaneous tissue. B: Exposure of the LDM and the thoracodorsal pedicle. C: subcutaneous scarifications made all over the flap. Black arrow points a subcutaneous scarification.

Figure 2: Operative area. Blue line: mediodorsal incision. Red line: surface covered by fibrin glue or NEXPOWDER^®. Yellow spots: placement of quilting sutures or VENASEAL^® drops.

Experimental Design

Rats were daily exanimated to assess seroma formation, wound infection or opening, flap necrosis, pain or functional impairment. Postoperative follow-up protocol was performed at Postoperative Day (POD) 7, 30 and 90, and included the following stages. All stages were performed under volatile general anesthesia as previously described.

Camera Documentation: Clinical skin assessment was done with standardized photographs taken 20 cm above the animal placed in a left lateral decubitus position so as to see the skin flap entirely. The analysis of the skin flap aspect was based on a skin damages gradation that we made (Table 1). It takes into account both the local inflammatory effect of the therapy used as well as the scratching lesions, related to both the surgery and the means of preventing seroma.

Fluorescence Imaging: Images acquisition was carried out using the Fluobeam^® 800 system (Fluoptics, Grenoble, France). The camera was placed at a fixed working distance 15 cm above the rat. An intravenous injection of 400 µL of a 500 mM Indocyanine Green (ICG) solution was performed. Near-Infrared (NIR) fluorescence was recorded in real time from ICG injection over a period of 250 seconds with an image every 2 seconds. The dynamic series of images was then exported and analyzed with the Wasabi!^® software (Hamamatsu Photonics, Germany) to quantify the signal intensity. A region of interest (ROI) corresponding to the whole skin flap of the operated site was determined by the examiner. The fluorescence intensity was quantified in relative lights units per 100 milliseconds (RLU/100ms). For each image, the maximum ICG fluorescence value was calculated after subtraction of the background baseline value. All data were compared to the skin fluorescence of five non-operated rats, on the same ROI (future operated skin flap area) to characterize ICG vascular kinetics in normal skin.

Microct-Scan: Seroma volumes were quantified by performing a microCT-scan (vivaCT80, Scanco Medical, Bruttisellen, Switzerland). Images were acquired using a dedicated scan program (energy 45 keV, intensity 114 µA, field of view 79.9 mm diameter, voxel size 100.3 µm, resolution/projection 796/500). If seroma was visually evident or palpable, 1 mL of a contrast agent (iodixanol 320 mg/mL) was injected in the fluid collection with a 29G needle and delicate digital pressure was done to create a clot preventing any fluid extrusion. When no collection was palpable, no injection was performed. The seroma volume was quantified in mm³.

Puncture: Clinical seroma quantification was performed after the microCT-scan, by draining the fluid collection with an 18G needle.

The collected volume was measured in mL. When needed, additive punctures were performed at POD 14 and 21 when the seroma was clinically evident to prevent skin necrosis and potential disabling for the rats. Volumes of these additive punctures were added to the POD30 volume.

Histopathological Analysis: For each group, rats were killed at POD7 (n = 4), POD30 (n = 3) and POD90 (n = 3). Full-thickness biopsies of the skin and chest wall were performed on each side of the animal (operated area, and non-operated side). Biopsies were fixed in 4% buffered formalin for 24h, then rinsed with phosphatebuffered saline (PBS) and put in ethanol 70%. Each fragment was embedded in paraffin and sectioned at 5 µm. standard staining with Hematoxylin Eosin and Saffron (HES) was performed.

Table 1: Skin damages clinical gradation.

Statistical Analysis

Study data were prospectively collected in Microsoft Excel 2016 (Redmond, Washington, USA) and analyzed. All data were analyzed with GraphPad Prism 8 (GraphPad Prism Software Inc., San Diego, CA). Surface of harvested LDM was compared between groups with a one-way ANOVA test to assure group comparability. Seroma volumes were compared using a KruskalWallis test (post hoc Dunn’s analysis) for both microCT-scans and punctures. Data are presented as mean ± SEM. Maximum fluorescence intensities of each group were compared with a oneway ANOVA test (post-hoc test: Tukey test). Data are presented as mean ± SEM. Comparison of decreasing intensity signal over time was performed with a two-test ANOVA. Differences were regarded statistically significant when p £ 0.05. Relevant trends were indicated if p < 0.10. Statistical analyses were performed for POD7 and POD30 exclusively, as n per group was too small at POD90.

Results

General

No rat presented any movement limitation after surgery nor sign of pain requiring supplementary analgesia or animal sacrifice. The mean surface of the harvested LDM was 19.73 ± 0.53 cm² with no significant difference between groups (p=0.1379). No rats developed any allergy to contrast agent nor to any glue. All rats of the NEXPOWDER^® group presented a whole flap necrosis with consequent wound dehiscence and large skin defect imposing a sacrifice between POD9 and 13. For ethical matters, we chose to stop this group at n=5. For all the other groups, no rat presented any wound dehiscence or infection, died or developed a problem that required exclusion from the study. One rat presented partial skin flap necrosis in control group and benefited of necrosis excision and closure.

Seroma Formation

All rats of the control group presented seroma formation at POD7 (Figure 3). All fluid collections were serosanguinous in character. At POD7, total seroma volume measured by microCTscan was 11.7 times lower in QS group (3.15 x10³ ± 0.83 x10³ mm³), 12.5 times lower in FG group (2.97 x10³ ± 1.09 x10³ mm³), 27.6 times lower in VENASEAL^® group (1.34 x10³ ± 0.90 x10³ mm³) and 3.3 times lower in NEXPOWDER^® group (11.26 x10³ ± 3.58 x10³ mm³) than in control group (36.99 x10³ ± 5.27 x10³ mm³). The use of QS, FG and VENASEAL^® showed a significant decrease in seroma volume compared with control group (respectively p=0.0016, p=0.0005 and p<0.0001) and no significant difference was found between these 3 prevention groups. There was no significant difference between NEXPOWDER^® and control group. At POD7, total seroma volume collected by puncture was 12.1 times lower in QS group (3.24 ± 0.92 mL), 13.4 times lower in FG (2.94 ± 1.13 mL), 23.4 times lower in VENASEAL^® group (1.68 ± 0.91 mL) and 3.6 times lower in NEXPOWDER^® group (11.00 ± 2.85 mL) than in control group (39.30 ± 5.10 mL). The use of QS, FG and VENASEAL^® showed a significant decrease in seroma volume compared with control group (respectively p=0.0017, p=0.0004 and p<0.0001) and no significant difference was found between QS, FG and VENASEAL^® groups (Figure 4). There was no significant difference between NEXPOWDER^® and control group. No rat of the tests group presented seroma at POD30. However, as we chose to add punctures of POD14 and POD21 to those of POD30, quantification of puncture at POD30 differed from microCTscan measurements. At POD30, total seroma volume collected with puncture was 16.5 times lower in QS group (1.99 ± 1.52 mL), 11.6 times lower in FG group (2.83 ± 2.64 mL), 131.4 times lower in VENASEAL^® group (0.25 ± 0.17 mL) than in control group (32.85 ± 14.35 mL). There was no significant difference between groups (Figure 4). There was a relevant trend between control group and VENASEAL^® group (p=0.0528). At POD30 total seroma volume on microCT-scan was 13.85 x10³ ± 5.49 x10³ mm³ in control group; 0.00 mm³ in every test group. There was a significant difference between control group and each prevention group (p=0.0006 for each). No difference was found between each test group. No rat presented seroma at POD90.

Figure 3: seroma prior to aspiration at POD7.

Figure 4: A: Seroma volumes (mean ± SEM) at POD7, in control group (n=10), QS group (n=10), FG group (n = 10), VENASEAL^® group (n = 10), and NEXPOWDER^® group (n=5). Statistical analysis: * p<0.05; ** p<0.01, *** p<0.001, **** p<0.0001. B: Seroma volumes at POD30 (mean ± SEM; n = 6 animals/group) in control group, QS group, FG group, and VENASEAL^® group. Volumes of POD14 and POD21 additive punctures were added to the POD30 volume.

Skin Lesion Gradation

At POD7, rats in control group presented in majority no or mild skin lesions. In the QS group, the predominant skin damage was mild to moderate. In the FG and VENASEAL^® groups, skin damages were mostly moderate to severe. 20% of the control group, 10% of the FG group and 20% of the VENASEAL^® group and 100% of the NEXPOWDER^® group presented necrosis of the skin flap. At POD30, skin lesions were mainly absent or mild in control and QS groups. It was moderate in FG group. Grades I to IV were found non-homogeneously in the VENASEAL^® group. No rat showed skin necrosis. At POD90, skin lesions were absent in the entire control group and predominantly in the QS group. Most of the FG and VENASEAL^® groups showed minimal or no skin involvement. One third of the VENASEAL^® group still showed moderate involvement.

ICG-NIR Fluorescence Imaging

At POD7, the maximum fluorescence intensity was 48.9 ± 8.4% (control), 151.1 ± 10.4% (QS), 144.1 ± 6.9% (FG), 147.6 ± 10.9% (VENASEAL^®) and 60.9 ± 17.0% (NEXPOWDER^®) of the one of the non-operated group and these differences were significant for control (p=0.0289), QS (p=0.0289), and VENASEAL^® (p=0.0436) groups (Figure 5A). Maximum intensity was 309.2 ± 21.2% (QS), 294.7 ± 14.1% (FG), and 302.1 ± 22.4% (VENASEAL^®) and 124.6 ± 34.8% (NEXPOWDER^®) of the one of the control group. Maximum fluorescence intensity was highly significantly different between control group and QS (p<0.0001), FG (p<0.0001), and VENASEAL^® (p<0.0001) groups. It was significantly different between QS and VENASEAL^® groups with non-operated group (p=0.0289 and p=0.0436 respectively). There was a relevant trend but no significant difference between FG and non-operated group (p=0.0841). The fluorescence signal kinetic of the control group was clearly slowed down compared to the one of the non-operated group (p=0.0063) illustrating a longer blood circulation and drainage time. (Figure 6A). Similarly, the NEXPOWDER^® group also displayed a slowed down kinetic compared to the non-operated group but not statistically significant (p=0.1851). On the contrary, the fluorescence signal kinetics of QS, FG and VENASEAL^® groups were close to the one of non-operated group and were thus very different from the one of the control group (QS: p=0.0007; FG: p<0.0001; VENASEAL^®: p<0.0001). There also was a significant difference in signal kinetics between FG and VENASEAL^® groups with non-operated group (respectively p=0.0184 and p=0.0011). There was no significant difference between QS group and non-operated group (p=0.3807). At POD30, maximum intensity signal was 79.8 ± 6.2% (control), 144.1 ± 11.3% (QS), 136.2 ± 12.5% (FG) and 149.9 ± 7.9% (VENASEAL^®) of the one of the non-operated group but these differences were not significant (Figure 5B). Nevertheless, maximum intensity signal was 180.6 ± 14.1% (QS), 170.7 ± 15.7% (FG) and 187.9 ± 10.0% (VENASEAL^®) of the one of the control group. QS, FG and VENASEAL^® groups displayed significant differences compared to the control group (p=0.0129, p=0.0467 and p=0.0047 respectively).

At POD30, fluorescence signal kinetics showed a trend toward normalization for control, QS, FG and VENASEAL^® groups compared to the non-operated group (Figure 6B).

Figure 5: Maximum fluorescence signal intensity after intravenous injection of ICG (mean ± SEM) at POD7, in control group (n=10), QS group (n=10), FG group (n=10), VENASEAL^® group (n=10), NEXPOWDER^® group (n=5) and non-operated rats (n=5). Statistical analysis: * p<0.05, ** p<0.01, *** p<0.001, **** p<0.0001. Maximum fluorescence signal intensity (mean and SEM) at POD30, in control group (n=6), QS group (n=6), FG group (n=6), VENASEAL^® group (n=6) and non-operated rats (n=5). *: significant (p<0.05) difference between indicated groups.

Figure 6: Decreasing percentage of fluorescence signal after intravenous injection of ICG at POD7 (A) in control group (n=10), QS group (n=10), FG group (n=10), VENASEAL^® group (n=10), NEXPOWDER^® group (n= 5) and non-operated rats (n=5); and at POD30 (B) in control group (n= 6), QS group (n=6), FG group (n= 6), VENASEAL^® group (n= 6) and non-operated rats (n=5).

Necropsy

No microorganism, edema, or hemorrhage was detected in any of the samples examined. At POD7, histological analysis showed an inflammatory reaction in control group and the skin flap was completely separated from the chest wall, forming a large dead space with a ragged floor. In contrast, skin flap was firmly adherent to the chest wall when a prevention method was used. Major inflammation was present in FG as much as in VENASEAL^® groups, while it appeared to be moderate in QS group. For NEXPOWDER^® group, all the tissue layers were subject to severe inflammation and necrosis (Figure 7). Inflammatory cells were mostly fibroblasts, neutrophils and macrophages in all groups. At POD30, inflammatory cell infiltration was milder in control and QS groups. In FG group inflammatory reaction was moderate, while it was severe in VENASEAL^® group. The seroma cavity was persistent in control group (Figure 8). At POD90, control and QS groups were no subject to inflammation and tissues were healing well with mild focal fibrosis visible. The appearance of the control group showed a return to normal structures organization, with the flap was adherent to the chest wall. Inflammation persisted and the epidermis was more damaged for FG and VENASEAL^® groups (Figure 8).

Figure 7: HES-stained tissue in the rat LDM harvest model at original magnification x50 in NEXPOWDER^® group illustrating both necrotized zones and severe inflammation.

Figure 8: HES-stained tissue in the rat LDM harvest model at original magnification x50. A: control group at POD30 and B: control group at POD90. Black arrow shows the persistent seroma space. C: QS group at POD30 and D: QS group at POD90. E: FG group at POD30 and F: FG group at POD90. G: VENASEAL^® group at POD30 and H: VENASEAL^® group at POD90. I: healthy tissue from symmetric non-operated area.

Discussion

The objective of our study was to develop a complete and transversal approach for assessment of seroma prevention methods and their possible side effects. The first strong point of our study lies in our reliable animal model. We had initially planned the technique of mastectomy in rats by jugulo-xiphoid incision as described in the literature [22,23] but in our case this model had the major disadvantage of not being reproducible. The scar was easily accessible and could be opened by the rat itself in the first days after surgery, despite a two-layer suture. Although new sutures were performed at each re-opening, we found that seroma production was inconsistent or even almost non-existent in the resutured rats. Given the relatively low incidence of seroma with the mastectomy technique, we chose to carry out a different surgical model. To produce seroma formation, we performed a LDM harvest as inspired from scientific literature [13,26]. Our model associating LDM harvest, removing of axillary nodes and subcutaneous scarifications presents the great advantage of providing reliably seroma as 100% of the operated rats presented seroma in large quantity when no prevention was carried out. This surgical technique is reliable, simply reproducible, with a constant anatomy and an easy approach.

It causes no functional impotence nor significant pain to the rat. Finally, it has the major advantage of a dorsal approach that is inaccessible to rats and therefore not subject to wound opening. The continuation of our work consisted in a methodological development for the follow-up of seroma formation and the monitoring of possible side effects of prevention methods. It has the advantage of being transversal and complete. In the first place, standardized photographs permitted a macroscopic and clinical analysis of the skin flap and we defined a dedicated lesion grading by taking into account the sequalae created by seroma as much as the inflammatory impact of the prevention method used. Then, in order to objectify the clinical examination and to go further with quantified analyses, we associated an exploration of microvascular perfusion by ICG-based NIR fluorescence imaging. ICG is a suitable fluorescent tracer for non-invasive evaluation of cutaneous blood flow and may be used as an index of tissue perfusion [2729], as it permits to quantify skin perfusion down to a depth of 3 mm [30]. Another key point brought by our methodology is the addition of a CT-scan to the puncture. CT-scans for animals are mainly used for bone density analysis and are not very sensitive to differentiate fluids from soft tissues. A seroma study found in the literature quantified the seroma with CT-scan but their protocol was not detailed enough to be reproducible [31]. After several acquisition attempts at different voxel size, resolution, energy or intensity, and additional subcutaneous injection of PBS (phosphate-buffered saline) on sacrificed rats, we were unable to obtain a reliable quantification of the liquid volume.

The addition of a standard X-ray contrast agent (iodixanol 320 mg/mL) finally permitted to quantify the volume of the seroma in a simple, reliable and efficient way (Figure 9). The reliability of CT quantification might be a tool to assess the natural history of seroma production without the need for iterative punctures. Finally, histopathological analysis concludes the follow-up protocol, as it provides microscopic confirmation of the results obtained from photographic documentation and ICG-NIR fluorescence imaging, regarding potential TA side effects. Our results at POD7 showed a tendency to inflammation and discomfort for the rats in all test groups whereas control group presented relatively lower skin lesions. Necrosis was macroscopically assessed and found both in control group and all TA groups, with a majority in NEXPOWDER^® group. This suggests a higher necrosis risk associated with TA or resulting as a sequalae in case of great seroma formation.