This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [21]. The MEDLINE database (PubMed) and Google Scholar were queried for relevant articles published until November 13th, 2024. All studies had to be written in English. Only articles presenting original data were included. Only articles discussing experimental swine models for vascularized composite allotransplantation and immunosuppression were eligible.

The search strategy for PubMed/MEDLINE and Google Scholar was developed (Supplementary Table S1). Two reviewers (LK, FK) independently screened all articles by title and abstract. Articles were subsequently analyzed in greater depth through full-text assessment to determine eligibility. Any disagreements regarding the inclusion of individual studies were resolved through consultation with a third author (MK). For included articles, citation searching was carried out on Google Scholar. Data extraction was performed independently by two authors (LK, FK) to ensure accuracy and consistency. During the blinded, dual-review process, we extracted the following variables for each study included: Digital Object Identifier (DOI), first author, study title, year of publication, region of publication, number of animals, mean age, gender, follow-up (mean and range), and the specifics of performed procedures. To evaluate the quality and risk of bias of the included studies, the SYRCLE risk of bias (RoB) tool for animal studies was employed [22]. The detailed risk of bias assessments for all studies are presented in Supplementary Table S2.

To complement the findings of this systematic review, we included a representative case report describing a novel heterotopic hemifacial VCA model in swine. This model was developed in response to gaps identified in the literature, particularly the lack of large-animal models incorporating facial tissue and permitting mucosal assessment. The case report provides detailed procedural insights and demonstrates the feasibility of serial mucosal and skin biopsies in a controlled, minimally invasive manner. Its inclusion offers practical context and supports the translational relevance of emerging strategies for immune monitoring in facial allotransplantation.

We identified various swine breeds that were employed to examine VCA. The primary models included MGH miniature swine [23–26], Yucatan miniature swine [27, 28], and outbred domestic swine [29–31], which were selected for their genetic similarities to human immunologic responses. In particular, MGH miniature swine and Yucatan miniature swine were frequently chosen for their manageable size and robust immunological profiles. Other studies utilized outbred Yorkshire swine and Swiss Landrace pigs [32–34], adding diversity in immune response due to genetic variability, which allowed for a comprehensive analysis of transplantation outcomes across multiple immune phenotypes. Details are reported in Table 1.

Various transplantation models were utilized, including heterotopic hind-limb transplantation, gracilis myocutaneous flaps, vertical rectus abdominis myocutaneous (VRAM) flaps, osteomyocutaneous flaps, partial hindlimb models, forelimb models, and tibial VCA. Of the n = 22 studies included, n = 16 (73%) applied heterotopic VCA models, while n = 5 (23%) used orthotopic models [23, 24, 26–32, 34–44]. Notably, n = 1 study (5%) did not categorize the approach as either heterotopic or orthotopic and n = 10 (45%) studies included osteomyocutaneous VCAs [27, 28, 30–37]. Furthermore, n = 2 (9%) studies performed hemi-facial VCAs, with n = 1 involving transplantation of the maxillo-mandibular complex [29, 44]. Lastly, assessment of mucosal tissue was reported in n = 1 (5%) of studies included.

Orthotopic models were employed by fewer authors. Fries et al. utilized an orthotopic mismatched porcine forelimb VCA model in SH-mismatched Yucatan miniature pigs [35]. Kotsougiani et al. implemented an orthotopic tibial defect VCA model in SLA- and blood type-compatible Yucatan miniature pigs [28]. Tratnig-Frankl et al. used an orthotopic gracilis myocutaneous free flap model in MHC-defined miniature swine to assess the impact of antioxidant therapies on graft survival [39]. Interestingly, Kuo et al. employed an orthotopic hemi-facial chondromyocutaneous flap, including skin, muscle, ear cartilage, and parotid gland in Lan-Yu miniature swine to study rejection dynamics and Park et al. utilized an orthotopic hemi-facial osteochondrocutaneous flap, incorporating skin, mucosa, subcutaneous tissue, ear cartilage and the maxillo-mandibular complex in domestic swine to investigate vascular and skeletal fixation techniques [29, 44]. More information is provided in Table 1.

Multiple immunosuppressive strategies were employed across different VCA models. These approaches included total body irradiation (TBI), thymic irradiation, T-cell depletion, bone marrow transplantation (BMT), and targeted drug therapies such as tacrolimus (TAC), cyclosporine A (CXA), mycophenolate mofetil (MMF), and mTOR inhibitors (e.g., rapamycin). Outcomes varied based on the immunosuppressive regimen and dosages used.

Starting with Barone et al., the authors combined low-dose total body irradiation (100cGy 2 days prior to surgery or 200cGy divided in 2 × 100 cGy doses on preoperative day 2 and 3), T-cell depletion with CD3 immunotoxin (0.05 mg/kg i.v., twice daily from preoperative day 4 to day 0), CXA (target level 400–800 ng/mL), and donor bone marrow cell infusion (7.8 × 10⁸ to 4 × 10⁹ cells/kg of recipient body weight) alongside VCA to achieve mixed chimerism, though this was insufficient for complete tolerance induction [23]. Ibrahim et al. employed short-term TAC monotherapy (target levels of 10–15 ng/mL) in a VCA model with intact vascularized bone marrow, demonstrating long-term graft survival with viable vascularized bone marrow and successful immune monitoring [36]. Kim et al. utilized a 30-day TAC course combined with adipose-derived stem cell (ASC) therapy (1.0 × 10⁶ cells/kg administered intravenously on postoperative day (POD) 7), achieving rejection-free survival for over 200 days while significantly upregulating T-regulatory cells and donor-specific unresponsiveness. Elgendy et al. compared the efficacy of mTOR inhibitors, finding that TAC (0.1–0.125 mg/kg) significantly delayed acute rejection (grade I AR on POD 30 and grade IV AR on POD 74) compared to rapamycin (0.02–0.2 mg/kg), which led to rapid rejection (grade IV AR by POD 17-20) [26]. Conversely, Fries et al. employed low-dose TAC (49 mg) administered via an enzyme-responsive hydrogel platform, which prolonged graft survival, whereas high doses (91 mg) caused poor tolerance and complications such as weight loss and pancreatitis [35]. Kotsougiani et al. used a combination of TAC (target levels of 5–30 ng/mL), MMF (target levels of 1–3.5 ng/mL), and methylprednisolone (tapered to 0.1 mL for maintenance), achieving graft survival and enhancing vascular remodeling without rejection during the 4-month follow-up [28]. Meanwhile, Kuo et al. combined irradiation, BMT, and CXA with mesenchymal stem cells (MSCs) in variying dosages, resulting in significantly prolonged graft survival and reduced acute rejection. Here, increased regulatory T-cell populations (CD4⁺/CD25⁺ and CD4⁺/FoxP3⁺) were found [30, 31, 37]. Leonard et al. applied 100 cGy total body irradiation, T-cell depletion with CD3 immunotoxin (50 μg/kg), and hematopoietic cell transplantation (15 × 10⁹ cells/kg), achieving stable mixed chimerism and long-term graft acceptance without signs of rejection up to POD 504 [24]. Mathes et al. pioneered an in utero bone marrow transplantation approach, achieving multilineage macrochimerism and donor-specific tolerance without prolonged post-transplant immunosuppression. The authors relied on CXA (target levels of 400–800 ng/mL) post-bone marrow infusion (2 × 10⁹ cells/kg) to maintain donor-specific tolerance, demonstrating effective rejection prevention in chimeric animals [32]. Furthermore, Shanmugarajah et al. utilized T-cell depletion with CD3 immunotoxin (50 μg/kg), 100 cGy TBI, and a 45-day CXA regimen (target levels of 400–800 ng/mL) to achieve immune tolerance in MHC class II mismatched chimeras, although MHC class I mismatched animals experienced rejection [38]. Meanwhile, Kuo et al. demonstrated that CXA delayed rejection from POD 7 to 28 in untreated controls to POD 38 to 49 in their hemi-facial VCA model [29]. Additionally, strategies explored by Wu et al. focused on enzyme-responsive and TAC-eluting hydrogels. The authors demonstrated prolonged survival using hydrogel-administered TAC (28 mg/4 cc and 49 mg/4 cc), effectively delaying grade IV AR to POD 20 and 28 [42].

Overall, five studies did not administer immunosuppressive therapies. For instance, Blades et al. observed flap rejection between POD 5 and 9 without immunosuppressive treatment and Park et al. by POD 14 to 18 [43, 44]. Tratnig-Frankl et al. and Wang et al. did not administer immunosuppression to avoid skewing of rejection periods in novel treatment approaches. Tratnig-Frankl et al. investigated H₂S and NaI treatments but observed no significant differences in graft survival or immunological outcomes compared to saline controls [39]. In Wang et al. the experimental group received hyperoxygenated University of Wisconsin solution and showed significantly later onset of grade 1 AR, compared to the control group [45]. Lastly, in Berkane et al. the study protocol did not foresee immunosuppression [33]. Further information can be found in Table 2.

In several models, immunosuppressive therapies and interventions significantly improved graft survival, with some protocols achieving long-term graft acceptance and reduced acute rejection (AR). For instance, Ibrahim et al. reported long-term graft survival exceeding 150 days with short-term TAC therapy and bone marrow infusion, highlighting the effectiveness of combining localized and systemic immunosuppression [36]. Similarly, Kim et al. observed prolonged rejection-free graft survival beyond 200 days using adipose-derived stem cell therapy combined with TAC, correlating the upregulation of regulatory T-cells (Tregs) with sustained graft tolerance [25]. Leonard et al. demonstrated stable chimerism and long-term graft survival up to 504 days following hematopoietic cell transplantation with irradiation and T-cell depletion, suggesting the importance of chimerism in inducing robust immune tolerance [24]. Additionally, Mathes et al. demonstrated that cyclosporine treatment combined with in utero bone marrow transplantation resulted in long-term chimeric stability and VCA acceptance, highlighting the effectiveness of early hematopoietic intervention [32].

Most studies reported significant success in prolonging graft survival with tailored immunosuppressive regimens. For example, Wang et al. observed a delay in acute rejection in flaps treated with systemic immunosuppression, with 75% of experimental flaps showing no rejection by day 15 [45]. Treatments combining stem cells with immunosuppressantsoften resulted in prolonged graft survival with reduced rejection rates, as seen in the work of Kuo et al. [30, 31, 37] In contrast, Elgendy et al. found that TAC significantly delayed AR compared to rapamycin, where rapid AR was observed [26]. Additionally, Fries et al. revealed that while low-dose TAC via enzyme-responsive hydrogels prolonged graft survival, high doses led to poor outcomes, including weight loss, pancreatitis, and early euthanasia [35]. Localized immunosuppressive delivery methods demonstrated promising results. Wu et al. utilized tacrolimus-eluting hydrogels, which effectively prolonged graft survival and delayed grade IV AR [42]. Similarly, Zhang et al. employed a localized tacrolimus-loaded drug delivery system, resulting in repeated intra-graft administration that significantly extended graft survival [34]. Kuo et al. found CXA to significantly delay rejection of hemi-facial flaps. Lastly, some treatments failed to demonstrate significant efficacy: Tratnig-Frankl et al. found no significant difference in graft survival or histological outcomes after antixodative therapy compared to controls [39]. Details are provided in Table 2.

Complications varied with the immunosuppressive approach and transplant model. While many studies reported successful outcomes in terms of prolonged graft survival and delayed rejection, complications often arose from either the intervention protocols themselves or the adverse effects of immunosuppressive regimens. For example, Fries et al. reported weight loss, poor feeding, and pancreatitis in animals subjected to high-dose TAC therapy, with some requiring early euthanasia [35]. Similarly, Elgendy et al. noted rapid rejection with rapamycin treatment compared to TAC [26]. Furthermore, vascular complications such as venous thrombosis were reported. These complications were resolved through re-anastomosis without long-term graft loss [36]. Blades et al. observed flap rejection between POD 5 and 9, despite initial adequate perfusion, and noted minimal erythema as an early rejection marker [43]. Additionally, Tratnig-Frankl et al. reported a technical failure in one saline subgroup, emphasizing the role of surgical precision in preventing graft loss [39]. Systemic complications related to immunosuppression were also observed. Shanmugarajah et al. documented chronic graft-versus-host disease (GvHD) in one animal, requiring euthanasia by POD 190 [38]. Similarly, Kuo et al. noted GvHD in animals treated with cyclophosphamide and irradiation, indicating the risks associated with preconditioning regimens [30, 31, 37]. Additionally, Wachtman et al. reported histological evidence of graft rejection in skin and muscle components, despite long-term survival in other grafted tissues [27]. Surgical mortality due to anesthesia-related complications was also recorded. Waldner et al. described an intraoperative cardiac arrest in one recipient, as well as venous compromise leading to flap loss in another case [40]. Furthermore, complications such as poor perfusion, erythema, and edema were commonly cited as markers of early graft rejection, requiring close monitoring for timely intervention [40, 43]. More information is presented in Table 2.

In summary, these studies underline the potential of swine models to explore VCA and immunosuppressive strategies, revealing that combinations of traditional drugs like TAC and cyclosporine with novel agents or delivery systems can extend graft survival and reduce immune responses.

Based on these findings, we decided to perform a heterotopic hemiface transplantation procedure using an MHC-defined Yucatan Sinclair strain to establish a novel swine model consisting of heterotopic hemiface vascularized composite allotransplantation (VCA) to the groin area. This model specifically enables frequent biopsies due to the accessibility of the flap but more importantly including donor mucosa while allowing the receipient to ingeste, feeding, or persue related activities. In contrast, orthotopic transplantation risks confounding mucosal assessment, as the animal may chew on or manipulate the graft. This setup enables detailed analysis of immunological interactions at the skin and mucosa interfaces, thus providing valuable insights into tissue rejection dynamics and tolerance in a way that conventional graft sites may not accommodate as effectively. To conclude, the Yucatan Sinclair strain is furthermore well-suited for this purpose, given its immunologic compatibility in modeling human responses.

A heterotopic hemiface vascularized composite allotransplantation to the groin area was performed from a male donor pig to a female recipient of MHC-defined Yucatan Sinclair strain. We performed the study following the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health [46]. Experiments were conducted according to a protocol approved by Yale University’s Institutional Animal Care and Use Committee (protocol number 2022-20476). More details are depicted in Figure 2.

The donor pig was positioned supine on heat support, under isoflurane anesthesia (0.8%–2%). Following connection to monitoring equipment and IV fluid administration, the donor received prophylactic antibiotics (cefazolin) and analgesics (meloxicam, buprenorphine) alongside local anesthesia with bupivacaine at key surgical sites. Antiseptic preparation with povidone-iodine was applied to the head and neck.

A hemifacial flap was carefully marked on the donor pig’s face. Skin incisions were made along the brachiocephalic muscle and the neck, sparing the ear and eye, while advancing dissection above the periosteal plane in the nasal and fronto-parietal areas. The dissection proceeded superiorly to the mandible, preserving the external jugular vein. In the facial region, meticulous incisions were made around the auricular, eyelid, and oral areas, incorporating buccal mucosa and securing the salivary glands. Further, the submandibular gland was removed after ligating its vascular branches, and the facial artery and nerve were identified. The facial nerve was transected near the stylomastoid foramen, and the external carotid artery and external jugular vein served as the flap’s vascular pedicle. Following tissue elevations along the masseter muscle and excisions in the neck area, the sternomastoideus muscle was detached, exposing central vessels including the common carotid and its branches. Key arteries, such as the internal carotid, were ligated and transected.

The graft was perfused in situ until the recipient’s vasculature was ready. The donor’s central vessels were ligated, and the graft was flushed with heparin solution, followed by euthanasia with sodium pentobarbital as per established veterinary protocols.

The recipient pig was anesthetized and positioned supine with a 30° rotation to expose the dorsolateral side, allowing simultaneous preparation with the donor. A groin incision exposed the femoral vessels, isolated to allow anastomosis. A subcutaneous pocket was created from the groin to the dorsolateral abdominal wall, where the graft would be placed. The hemiface flap was inset dorsolaterally to facilitate immune monitoring.

Following ligation of the donor’s femoral vessels, the graft was prepared for anastomosis. Venous anastomoses were conducted with a vascular coupling device (2.5 mm size), while arterial anastomoses were sutured with 9-0 nylon. Once vascular patency was confirmed, the graft was secured in place with sutures to the abdominal wall muscles while the skin and mucosa paddle were exteriorized for monitoring. The groin incision was closed in layers and covered with a Tegaderm^® patch to prevent infection. Analgesia was administered via a fentanyl patch, and postoperative antibiotics were given. The recipient pig was monitored until full recovery.

The recipient pig recovered from the operation without any complications, exhibiting normal eating and drinking behavior and full mobility. Frequent monitoring revealed a viable flap, with the recipient site in the groin well-tolerated by the pig. After 24 h, the pig was euthanized according to protocol. These findings demonstrate the feasibility of our model for heterotopic hemiface vascularized composite allotransplantation in evaluating graft viability and immune response in a controlled and accessible site.

This study is limited by the inherent heterogeneity of the swine models and experimental protocols reviewed, which complicates direct comparisons and generalizations across studies. The small sample sizes across the included studies reduce the generalizability of our findings and limit our ability to perform a robust quantitative meta-analysis. The rarity of VCA studies in swine also introduces potential publication bias, as studies with negative or inconclusive outcomes may be underreported.

Additionally, our heterotopic hemiface model, while valuable for serial biopsies, may not fully represent the complexity of vascular integration seen in orthotopic models, potentially limiting its direct applicability to specific clinical scenarios in VCA. Moreover, while the described heterotopic hemifacial VCA model was primarily designed to ensure surgical feasibility and facilitate serial mucosal and skin biopsies, we acknowledge the limitation that the transplanted mucosa is no longer located within its native anatomical environment. As such, it is exposed to non-physiological conditions, including external microbial flora and mechanical influences at the abdominal implantation site. These factors may affect local immune responses and limit the interpretability of biopsy-derived data with respect to natural mucosal immunity. Nevertheless, the model remains valuable for studying epithelial immune activation and early alloimmune events in a controlled and accessible setting, and it offers an important proof-of-concept for future refinements toward orthotopic models.