Information sources

PubMed, Cochrane reviews and Scopus electronic databases from inception to May 2023 were searched. Our search strategy for PubMed combined both free-text terms and Medical Subject Headings (MeSH) with Boolean operators, as shown in Table 1. The search strategies for the latter two databases consists of free-text terms and Boolean operators only. These are shown in Table 2.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. PubMed search strategy.

https://doi.org/10.1371/journal.pone.0292013.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Scopus and Cochrane reviews.

https://doi.org/10.1371/journal.pone.0292013.t002

Study selection

Two authors (NP and JS) screened the compiled list of search results. These were managed using Endnote 20 (Thomas Reuters, Toronto, Canada) and Excel (Microsoft, Redmond, Washington, USA). After removing duplicates, the articles were independently screened by two authors (NP and JS) in parallel, first by title and abstract. Any disagreements were reviewed by the third and fourth author (BL and JAH). Those remaining were studied in full, prior to a final shortlisting based on pre-specified inclusion/exclusion criteria listed below. To ensure all relevant studies were identified, the reference lists of all included articles and previous systematic reviews were routinely checked. This process is displayed diagrammatically as a PRISMA flowchart in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. PRISMA flowchart.

https://doi.org/10.1371/journal.pone.0292013.g001

Eligibility criteria

Inclusion criteria:

1. Studies utilising photon/electron radiation
2. Studies utilising adipose tissue fractions or single cell groups isolated from adipose tissue as their intervention
3. In vitro/in vivo studies with primary data
4. Studies published in English language

Exclusion criteria:

1. Studies solely focused on fibrosis of visceral organs
2. Clinical trials
3. Review papers and other non-primary data sources
4. Studies focusing on differentiated cells or stem cells derived from non-adipose sources e.g., bone marrow- derived mesenchymal stem cells

Selected articles

The search was performed on 13^th May 2023. The search strategy (Tables 1 and 2) yielded 2909 studies initially, and then 2273 after duplicates removed. After title and abstract screening, 90 articles underwent full-text review for eligibility. Of these articles, 59 were excluded in-line with our inclusion/exclusion criteria. In total, 31 papers underwent analysis in the final systematic review.

Data collection process

All data from the final studies was extracted onto an electronic proforma using Excel (Microsoft, Redmond, Washington, USA) by two authors (NP and JS). This proforma was pre-designed by BL and trialled on three articles. It was deemed to provide a consistent and structured method of data extraction and appraisal. Any disagreements with extracted data were rectified by BL and JAH.

For studies which provided information on multiple interventions, skin conditions and/or research outcomes, only data relating to the current research objective was extracted. Where appropriate, authors were contacted to provide missing information.

Data items

Data items extracted included study design (in vivo/in vitro), intervention preparation, cell types used, disease model, radiation protocol e.g., dosage regimens (Grays, fractionation), date of last follow-up and recorded outcomes. Where mechanisms behind AFT in RIF of the skin were suggested in studies, this and corresponding evidence were also collated.

Risk of bias assessment

All included studies involved a non-randomised design comparing both effector/intervention and comparator/control groups. Without randomisation these observational results may be inherently biased. The risk of bias for in-vivo studies was assessed using the Systematic Review Centre for Laboratory Animal Experimentation (SYRCLE) assessment tool [21]. The Office of Health Assessment and Translation (OHAT) tool designed by the United States national toxicology program was used for in-vitro studies [22].

Two authors (NP and JS) independently appraised each study, assigned individual scores for each tool domain, and then presented this information in risk of bias tables. These are shown in S1 File. Any discrepancies between authors were resolved by consensus with authors BL and JAH.

Data synthesis

As there is no standardised protocol or unit of analysis for assessing the effects of AFT in RIF, methodologies vary between studies. An overarching meta-analysis and quantitative synthesis was not conducted on account of this between-study heterogeneity. Instead, outcome measures were evaluated using simple descriptive statistics and results collated in a narrative synthesis. The results were split into in vivo and in vitro experiments and separated further by effector cell (fibroblast, immune cells, vascular endothelial cells (VEC) etc.). The effect of AFTs on vascular function, immunomodulation and fibrosis were summarised in the review.

Literature search summary

In total, 31 studies were included for final analysis. Of these, nine conducted in vitro experiments utilising a co-culture model whilst 25 studies conducted in vivo experiments. The results are summarised below in Tables 3 and 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. In-vitro studies.

https://doi.org/10.1371/journal.pone.0292013.t003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. In-vivo studies.

https://doi.org/10.1371/journal.pone.0292013.t004

ADSC preparation and characterisation

Across all studies, ADSCs were prepared in a similar fashion. Adipose tissue was isolated from the target tissue and then digested with collagenases. After undergoing cycles of filtration and centrifugation, ADSCs were obtained and cultured through different passage numbers. In 18 studies, ADSCs were further characterised, and this was either achieved through flow cytometry, fluorescence-activated cell sorting (FACS) or differentiation assays to show ADSC cell surface markers or prove their differentiation potential.

In-vitro study design

All nine in vitro experiments used a co-culture set-up where two different cell types were cultured together either directly in the same well or indirectly by using well inserts [23–31]. Four in vitro studies adopted ADSCs as their treatment. Only Xiao et al. utilised mature adipocytes. Shukla et al., Yao et al., Saijo et al., and Sörgel et al. used ADSC-cultured media (ADSC-CM). Yu et al. utilised SVF. A variety of different cell types were used in the co-culture experiments: human microvascular dermal endothelial cell (HDMEC) (n = 1), normal human dermal fibroblasts (NHDF)(n = 2), human foreskin fibroblasts (HFF)(n = 1), bone marrow derived stem cells (BMSC)(n = 1), human immortalized keratinocyte (HaCaT)(n = 3), normal human oral squamous epithelial cells (NOK)(n = 1), human dermal lymphatic endothelial cells (HDLEC)(n = 1), and human skin fibroblast cells (WS1)(n = 2). The radiation protocol generally involved a single exposure of 2–20 Grays (Gy) (median 6Gy), with some studies using varying radiation doses. The methodology of the different studies included proliferation assays (n = 4), enzyme-linked immunosorbent assays (ELISA) (n = 3), Western Blotting (n = 5), reverse transcriptase reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) (n = 2), and migration assays (n = 3). Other experiments included: tube formation assay, Luminex assay, Terminal deoxynucleotidyl transferase dUTP Nick-End Labelling (TUNEL) apoptotic cell assay and Tandem Mass Tag (TMT)- based proteomic quantification.

In-vivo study design

In the twenty-five in vivo studies, different species were utilised: mice (n = 14), rats (n = 4), mini pigs (n = 6), rabbit (n = 1), and human tissue (n = 1) [26,27,30,32–54]. Fifteen studies utilised a single exposure of 10 to 90Gy (median: 50.6Gy) and seven studies a cumulative exposure of 30Gy fractionated as six 5Gy doses over 12 days. Experimental design usually involved irradiation of animals and a transitionary period for RIF to develop (median 25 days), followed by one/multiple AFT procedures involving ADSC, SVF, whole fat or microfat. Control groups were either given phosphate buffered saline (PBS) or ringer’s lactate solution. Two studies looked at the augmentation of lipotransfer using ADSC sub-types. Furthermore, two studies investigated supplementing platelet-rich plasma (PRP) to improve AFT outcomes.The average observational period across studies was 76 days. Outcomes recorded included clinical evaluation, histological assessment (n = 20), mechanical strength testing (n = 5), fat graft volume retention using micro-CT (n = 2), immunohistochemistry (IHC) (n = 15), immunofluorescence (IF) (n = 4), Picrosirius red collagen staining (n = 3) and Massons trichrome collagen staining (n = 8).

Risk of bias assessment

For in-vitro studies (OHAT assessment tool), a low risk of bias was noted in the domains of experimental conditions, exposure characterisation, incomplete data, and reporting on all measured outcomes. Authors provided sufficient information on plate conditions, media and solvents, and they were similar across control and intervention groups. However, they score lower on domains two and five, only due to limited information on whether assessor blinding had taken place. Automated systems where the cells are transferred to assays without assessor handling would mitigate the need for blinding.

Across the in-vivo studies (SYRCLE assessment tool), there was a low risk of selection, attrition, and reporting biases highlighted. Authors ensured similar baseline animal characteristics, measured primary/secondary outcomes in all intervention and control groups, and included all animals in the final analyses. However, we noted that most studies had insufficient details to assess domains three to six e.g., allocation concealment and assessor blinding. As this issue was common, it was not grounds for exclusion, however, this is an area for future studies to improve on.

In-vitro studies

In-vivo studies

Vascularisation

Radiotherapy is known to induce a local inflammatory microenvinroment with increased vascular endothelial dysfunction and permeability. A microscopic hallmark of RIF is microvascular obliteration–a consequence of induced local inflammation and excessive collagen deposition. The disruption and loss of micro-vessels in combination with an increased demand for oxygen, due to an influx of inflammatory and mesenchymal cells to the site of injury, leads to the accumulation of reactive oxygen species (ROS). While ROS are necessary due to their antimicrobial activity in normal wound healing, they cause damage to the tissue in the dysregulated fibrotic environment. Furthermore, ROS strongly activate the TGF-β/SMAD axis and lead to additional downstream collagen accumulation [57]. Hypoxia, caused by post-radiation vasculopathy, is considered a major stimulus of the fibrotic response seen in RIF [58]. Hence, it would seem beneficial for AFTs to enhance tissue vascularisation to counteract this phenomenon.

However, it may not be a question of vascular density but rather the quality and functionality of the present vasculature, indicated for example by their patency [59]. Sultan et al. found increased vascular density four weeks after irradiation injury compared to non-irradiated controls, indicated by increased staining for CD31 and vWF both in sham and fat-grafted mice [60]. In addition, levels of pro-angiogenic molecules such as vascular-endothelial growth factor (VEGF) and stromal cell-derived factor-1 (SDF-1) were higher in the fat-grafted group [60]. After eight weeks, however, less CD31 and vWF were measured in the fat-grafted group, while presenting with a 42% lower scar index compared to the sham-grafted group. This possibly indicates that AFT promotes more proficient vascularisation, leading to closer resemblance to normal skin in the pre-irradiation state. Similarly in other scarring and fibrotic skin conditions, AFT normalized abnormal microvasculature architecture [61,62]. In addition, Lindegren et al. studied genetic pathways in irradiated human breast samples, finding that the hypoxia gene pathway was severely aberrant after radiation and AFT restored this back to normal [39].

ADSCs alone showed ambiguous effects on vascularisation in vivo. One study noted them to increase vascular density three weeks post-grafting, while another study claimed there was no difference between PBS- and ADSC-injected rats after four weeks [43,47]. Huang et al. showed that ADSCs co-localized with vWF and CD31 markers at day 14 post-fat graft [47]. Unrelated to RIF of the skin, Nie et al. demonstrated similar findings using an excisional wound healing model [63]. In this study, ADSCs activated wound healing processes and angiogenesis, as well as being present in newly formed capillaries. It remained unclear whether the ADSCs directly differentiated into VEC, increased VEC proliferation, or supported VEC function by fusing with them. Although the potential of ADSCs to differentiate into VEC in vivo has been demonstrated repeatedly, it remains unclear if this mechanism attenuates RIF of the skin by promoting or regulating angiogenesis [63,64].

Another potential way ADSCs may promote angiogenesis is through the secretion of growth factors at the injection site. Previous research has proven the ability of ADSCs to secrete proangiogenic factors including different VEGFs as well as basic fibroblast growth factor (FGF2) and hepatocyte growth factor (HGF), which have also been found to promote angiogenesis [23,63,65,66]. HGF may play an important role in the angiogenic potential of ADSCs. Cai et al. achieved HGF-knockdown in ADSCs and observed a significantly reduced capacity to promote endothelial cell maturation, migration, and angiogenesis in a mouse hindlimb ischemia model [67]. Yet, no in vivo study has examined the expression of HGF and VEGF in skin harvested from fat-grafted animals for RIF. This warrants further investigation. Haubner et al. also found that FGF2 levels were increased in ADSC and HDMEC co-culture experiments, however, the relative increase was lower in the co-culture as compared to ADSCs alone [23]. Whilst this does suggest a role for FGF2, overexpression might be counterproductive with ADSCs potentially regulating ideal levels. Similarly, FGF2 is believed to have positive effects on wound healing by accelerating proliferation and granulation, while overexpression may lead to uncontrolled formation of granulation tissue [68].

Radiotherapy is known to induce a local inflammatory microenvinroment with increased vascular endothelial dysfunction and permeability. Coupled with the secretion of pro-inflammatory chemotactic cyotkines such as IL-6, leukocytes adhere to and transmigrate across the vascular endothelium into surrounding cells with more ease [69]. In particular, monocyte chemoattractant protein-1 (MCP-1) and IL-8 can trigger firm adhesion of monocytes to the vascular endothelium. Haubner et al. also noted that radiation-induced HDMEC dysfunction involved the upregulation of pro-atherogenic adhesion molecules ICAM-1 and VCAM-1 [70]. This upregulation was attenuated by ADSC co-culture, as well as IL-6 expression. ADSCs could help ameliorate VEC dysfunction in this way.

Lymphangiogenesis

Radiotherapy is also an independent risk factor for developing lymphoedema, and ADSCs may also play a significant role in lymphangiogenesis post-irradiation [71]. Saijo et al. demonstrated that, when co-cultured with ADSCs and ADSC-CM, irradiated HDLECs efficiently proliferated, migrated, and formed tubes [29]. Outside of our study sample, Ahmadzadeh et al. showed a similar effect of ADSCs on non-irradiated HDLECs. They demonstrated more effective proliferation, migration and formation of tube-like structures than after stimulation with either VEGF-C, HGF or FGF alone–suggesting the combination of factors in the ADSC-CM to be most effective [72]. Saijo et al. theorised a paracrine mechanism of action through upregulation of FGF2, which is both a pro-lymphangiogenic and pro-angiogenic factor. FGF2-knockdown in these HDLECs significantly decreased its lymphangiogenic capabilities.

Reactive oxygen species

Outside of our study sample, Kim et al. measured the anti-oxidative properties of ADSC-CM and found it be equally potent as 100 μM ascorbic acid [73]. They treated human fibroblasts with tert­-butyl hydroperoxide (mimicking the oxidising effects of radiotherapy) and incubated them with ADSC-CM, resulting in improved survival and increased SOD and glutathione peroxidase (GPx) activity in fibroblasts. Proteomic analysis of the cultured media further showed different anti-oxidative compounds including superoxide dismutase (SOD), insulin-like growth factor (IGF), hepatocyte growth factor (HGF), platelet-derived growth factor (PDGF), keratinocyte growth factor (KGF), and FGF among others.

Immunomodulation

Cytokines, chemokines, and damage-associated molecular patterns (DAMPs) are produced by epithelial and endothelial cells damaged by radiation. This causes an influx of inflammatory cells, predominately macrophages and neutrophils, which are supposed to eliminate dead cells, debris, and invading pathogens [2,74]. Repetitive trauma or radiation and a disproportionate immune response with prolonged, unresolved inflammation ultimately lead to excessive amounts of ECM deposition [2]. Macrophages are usually found near fibroblasts and repeatedly described as master regulators of fibrosis [75,76]. Depending on their phenotype, they have diverse functions ranging from coordinating wound healing by promoting angiogenesis, tissue remodelling, and resolving inflammation (M2 type), to recruiting immune cells, causing inflammation, and tissue destruction (M1 type) [77]. However, while early fibrosis is characterised by high numbers of inflammatory cells and activated, large fibroblasts with less compact collagen fibres, later stages of fibrosis tend to show compact and organized collagen fibres and often less inflammation with less inflammatory immune cells [78]. Hence, possible immunomodulatory effects of AFT may also depend on what state of RIF is present–early vs late.

The SVF as part of a lipotransfer consists of a variety of cell types including macrophages, endothelial cells, ADSCs, and so on. However, lymphocytic immune cells account for the largest group of cells in the SVF [79,80]. This results in a variety of possible mechanisms by which AFTs can restore tissue homeostasis in RIF. The immunomodulatory effects of AFTs and its inherent cell types have been investigated in several studies. In the context of radiation injury, Lindegren et al. explored the expression of a multitude of gene pathways in irradiated human breast samples, finding that the interferon-gamma (IFN-γ) gene pathway was severely dysregulated post-radiation and subsequently downregulated the most by AFT [39]. IFN-γ is a pro-inflammatory cytokine secreted by T helper 1 (Th1) cells and heavily involved in the development of RIF [81–83].

ADSCs have demonstrated great immunomodulatory capacity. They have been found to suppress lymphocyte proliferation, suppress natural killer (NK) cell cytotoxicity, inhibit differentiation of monocyte-derived immature dendritic cells, and promote macrophage polarization towards an anti-inflammatory M2 phenotype [26,84]. In RIF, some of these effects have been reproduced. In irradiated minipig muscle, Riccobono et al. displayed that ADSC injection led to a greater proportion of M2 to M1 macrophages relative to irradiated PBS-injected control, as well as higher amounts of IL-10 and TGF-β [44]. Both IL-10 and TFG-β have been linked to M2 polarization, however, TGF-β is also one of the main drivers of tissue fibrosis [85]. Other study settings have proposed IL-6 and colony-stimulating factor 1 (CSF1) as possible mediators of ADSC-induced M2 polarization [86,87]. These M2 macrophages then help drive repair of radiated tissues. If large numbers of M1 macrophages were present, they would likely to shift the skin from a state of repair to continued inflammation, myofibroblast survival, and fibrosis [44,88].

M2 macrophages can be split further into sub-types (M2a, M2b, M2c and M2d) with varying roles [85]. Further research is required into these specific subtypes and how ADSCs can modulate their function, since literature to date has only focused on M2 macrophages as a collective whole. Furthermore, expression of tumour necrosis factor-inducible gene 6 protein (TSG-6) and other anti-inflammatory chemokines should be studied in greater detail. In microglia, Hu et al. demonstrated that ADSC-secreted micro-RNAs upregulated TSG-6 expression, which then inhibited immune-mediated inflammation and encouraged tissue repair [89]. Whilst this immunomodulation shows promise in future treatment of inflammatory central nervous disorders such as multiple sclerosis, it is not known whether this applies to RIF.

Extracellular matrix remodelling

Excessive extracellular matrix (ECM) deposition–especially collagen–is the hallmark feature of fibrosis. Remodelling of the ECM after injury is essential for tissue homeostasis and in the case of RIF is strongly dysregulated with fibroblasts producing vast amounts of ECM. AFTs could possibly remodel the ECM directly through ECM synthesis, degradation, and modification or indirectly by changing the behaviour of tissue resident cells which produce ECM e.g., fibroblasts.

AFTs have been shown to reduce, remodel, and realign collagen fibres, as well as reduce α-SMA, dermal thickness and scar size [18]. For RIF of the skin specifically, this review provides further evidence that collagen density and dermal thickness significantly decrease after fat grafting (20, 23, 27) as well as ADSC injections alone (32). ADSCs have shown to prevent fibroblast to myofibroblast transition, cause myofibroblast apoptosis as well as generally decrease the production of collagen from fibroblasts in research of fibrotic skin conditions and scarring [18,19,90]. Further, ADSCs in co-culture with irradiated fibroblasts led to a decrease in α-SMA, collagen, and TGF-β production [23,27].

Secreted HGF, as well as FGF2, have been suggested as possible ways by which ADSCs can modulate fibroblast activity and remodel the ECM. They are believed to be part of an antifibrotic FGF2-JNK-HGF pathway [91]. FGF2 supposedly causes myofibroblast apoptosis and inhibits the TGF-β1/Smad–pathway with subsequent reduction of α-SMA expression and collagen production. It is also believed to increase MMP-1 and decrease TIMP-1 [92–94]. FGF2 administration to hypertrophic scars in animal models reduced epidermal thickness and collagen levels [95]. HGF was present in ADSC monoculture alone and co-culture of ADSCs with irradiated fibroblasts increased HGF levels. When recombinant HGF was added to an irradiated fibroblast monoculture, a dose-dependent reduction in TGF-β levels was also recorded [27]. According to Ejaz et al., ADSC-secreted HGF may also stimulate the influx of endothelial, bone marrow stem cells and other cell types that collectively remodel the extracellular matrix [27]. The authors showed an influx of GFP⁺ Luc⁺ labelled BMSC in AFT-treated mice, but none were identified in saline-grafted mice. HGF has been shown to recruit BMSCs via interactions with their c-met receptor [96]. After they are recruited, ADSCs may affect the expression profiles of these BMSCs, as Ejaz et al. noted significant downregulation in the expression of pro-fibrotic proteins (TGF-β, COL1, CTGF) in ADSC/BMSC co-culture.

Irradiation leads to a dysregulation of MMPs and TIMPs–a further driver of RIF [97,98]. In chronic RIF of the skin, it would seem beneficial if AFTs or ADSCs would increase the amount of active MMPs while decreasing TIMPs. This would enable tissue remodelling with a decrease in ECM. Indeed, a liquid chromatography (LC-MS) study of the ADSC secretome revealed the highest enrichment score and protein counts for ECM proteins, cell adhesion proteins as well as MMPs and TIMPs [95]. However, the functions of TIMPs and MMPs go beyond tissue remodelling and their regulation is complex [99]. They can promote inflammation as well as decrease inflammation. Further, MMPs are tissue specific, regulated by immune and stromal cells, and their functions depend on disease and tissue context [100]. For example, some MMPs–such as MMP8 –are profibrotic while others are believed to be antifibrotic [99]. Further, ECM degradation by MMPs also leads to the release of dormant TGF-β. Excessive MMP activity mat even facilitate the development of chronic wounds, by breaking down newly formed ECM of granulation tissue too quickly, before healing can begin [101]. Hence, it is not as simple as upregulating MMPs and downregulating TIMPs to treat fibrosis. TGF-β decreases the activity of MMPs, specifically MMP-2, MMP-9, and MMP-1, within fibroblasts [102,103]. ADSCs counteracted this effect with increased MMP1 and MMP2 levels, but also increased TIMP1 and TIMP2 levels detected in co-cultures with irradiated fibroblasts [25].

The dynamics of ECM remodelling in fibrotic skin conditions are complex and no simple answer exists on how AFTs or ADSCs may reinstate tissue homeostasis. However, we assume RIF is counteracted by reinstating the physiological balance between ECM synthesis and degradation. This is achieved directly by downregulating profibrotic MMPs and secreting anti-fibrotic MMPs, as well as indirectly by inhibiting excessive ECM production from fibroblasts. This is by preventing their activation and transition into myofibroblasts.

Regenerative ability

The variety of cell types such as endothelial cells or immune in AFTs could possibly replace damaged cells in RIF of the skin. Here, emphasis has been placed mainly on ADSCs. As a form of mesenchymal stem cell (MSCs) they are capable of self-renewal, show long-term viability, and multilineage potential. They can differentiate into various cell types of the mesenchymal lineage including adipocytes, fibroblasts, myocytes, keratinocytes, chondrocytes, osteoblasts but also of other lineages such as neural cells, endothelial cells, and hepatocytes [104,105]. An in vivo study by Haubner et al. showed that when ADSCs were cultured in HDMEC medium, ADSC count decreased progressively during the follow-up period. One suggested theory was the ADSCs were actively differentiating into endothelial cells [23]. Hence, ADSCs could possibly directly replace cell types damaged by radiation. It remains unclear if this mechanism is part of the clinical effect observed after AFTs.

Anti-apoptosis

Radiation-induced cell death of epithelial and endothelial cells is a key driver of RIF of the skin. This leads to the production of cytokines, chemokines, and DAMPs, and subsequent inflammation and influx of immune cells; all of which impair tissue function [106]. AFT may decrease cell death and henceforth prevent further disease progression. When added to irradiated fibroblasts, SVF upregulated DMTF1, MFAP5, and P116, all of which regulate key cellular apoptotic pathways [30]. Similarly, ADSCs led to downregulation of pro-apoptotic proteins (CTSF, Bid, BAX) and fewer cells undergoing apoptosis as a result in co-culture with human immortalized keratinocytes (HaCat) and human oral squamous epithelial cells (Nok) (14). Evin et al. showed increased anti-apoptotic Bcl-2 expression by hair follicles was in their ADSC/PRP and ADSC groups, relative to the control [49]. Rehman et al. also reported direct secretion of anti-apoptotic factors by ADSCs [66]. Huang et al. showed that injected ADSCs co-localized with the cell proliferation marker Ki67 in vivo, providing further evidence that ADSCs may promote cell survival and preserve tissue function and integrity [47].

Augmentation of fat grafts

A prominent limitation of fat grafting in RIF is its retention in hostile, irradiated tissue. In studies of autologous fat grafts in non-irradiated sites, retention rates already vary between 40–60% [107]. On the other hand, irradiated tissue is a fibrotic, hypovascular and inflammatory environment. Grafted fat is therefore susceptible to central necrosis and subsequent inflammation [15]. This can lead to cyst formation, further fibrosis and local infection [108]. Augmentation techniques may help overcome these limitations.

CAL is the enrichment of fat grafts with cells from SVF or with culture-expanded ADSCs. This technique has already shown efficacy in human trials for autologous fat grafts and cosmetic breast augmentation [109,110]. CAL resulted in greater fat volume retention, reduced post-operative atrophy and better cosmetic outcomes. Recently, Luan et al. has also demonstrated its efficacy in RIF treatment, with regards to reduced collagen deposition and increased angiogenesis (CD31) [38].

SVF consists of several cell types, including pericytes, endothelial, haematopoietic, stromal and stem cells. ADSCs only make up less than 5% of SVF. As Yu et al. demonstrated, all these cell types can affect the extracellular matrix through various pathways. Upregulated proteins included microfibrillar-associated proteins (MFAPs) and MMP-1 which complement the regenerative effects of ADSCs [30]. Practical advantages of using SVF over ADSCs alone stems from its simpler isolation process, meaning higher yields can be produced. In addition, no ex-vivo amplification is required which thereby reduces the risk of cell contamination and potential neoplasia [111].

PRP is a concentrate of platelets, obtained after centrifugation of whole blood. It is abundant in pro-regenerative growth factors including PDGF, TGFB, VEGF and epidermal growth factor [49]. Augmentation with PRP is an exciting area of research because its fibrin provides a three-dimensional structure which helps to retain both ADSCs and their secreted growth factors. This allows for appropriate cell migration and enhanced paracrine effects of these growth factors [112]. In particular, the work by Evin et al. on the effects of ADSC/PRP on irradiated hair follicles is promising for the treatment of radiation-induced alopecia in head and neck cancer. The potential benefits of PRF goes even further. This is a second-generation platelet concentrate which is easier to prepare, less expensive and carries a lower risk of immunological rejection compared to PRP [113].

Following from above, several bioactive structural supports in injectable form are currently under investigation, in attempts to better simulate ECM properties. In particular, in vitro research with conventional monolayer cell cultures is significantly limited, as it cannot recreate the mechanical and biochemical communication networks which aid ADSC proliferation and differentiation. Decellularized adipose matrices (DAMs) are acellular, albeit natural 3-D structures that have shown promise even if they do not resemble the donor site’s anatomy [114]. In our included study by Adem et al., demonstrated retention of anti-fibrotic cytokines FGF-2 and MIF, and pro-angiogenic cytokines such as PDGF-AA [52,115]. However, DAMs were inferior to fat grafting alone. This highlights the importance of ADSCs in facilitating de-novo adipogenesis and tissue regeneration. DAMs enriched with further growth factors and ADSCs to levels higher than native grafts may be an avenue to explore. For example, micronized cellular adipose matrices (MCAMs) contain a high proportion of ADSCs, minimal adipocytes and some capillaries [116]. Sowa et al. demonstrated MCAMs had similar anti-fibrotic effects to SVF and superior adipogenic capability to fat grafting alone [53]. In addition, the preparation of MCAMs is less intensive, quicker, and not subject to the same regulations as SVF. Finally, Kim et al. proved the efficacy of cultured, ADSC-seeded spheroids in restoring functionality of irradiated murine salivary glands [51]. Prior to transplant, these cellular microenvironments and its biochemical pathways can be sequentially engineered. For example, the authors treated the spheroids initially with a Wnt/β-catenin pathway activator followed by FGF7 at intervals, to stimulate ADSC differentiation into salivary gland stem cells and paracrine factor secretion respectively.

Additional supplementation of these structures with biocompatible materials and compounds may prove beneficial. Research has shown how ADSCs embedded fibrin glue can enhance wound healing through stimulating angiogenesis, fibroblast migration and ECM organisation [117]. Abbas et al. have highlighted the potent anti-inflammatory and anti-oxidative effects of Vitamin E [50]. Although it has no pro-angiogenic properties itself, Vitamin E likely enhanced ADSC-induced angiogenesis through the above effects on the local microenvironment. In addition, its fat solubility is inherently advantageous in lipotransfer.

ADSC subpopulations

Although often considered as a single entity, ADSCs are a heterogeneous mix of subpopulations with different cellular functions and secretomes. Previous research has identified angiogenic (CD248+), adipogenic (BMPR-1A+), osteogenic (CD105-/‘endoglin’) ADSCs [107]. The CD90+/CD73+/CD105+/CD45-/CD31-/ CD146+(−) sub-types are also known to secrete a large variety of ECM proteins [95].

In our study sample, Borrelli et al. identified promising antifibrotic and anti-inflammatory potential of CD74+ ADSCs in RIF [32]. CD74 activates the anti-inflammatory AMPK signalling pathway through the macrophage inhibitory factor (MIF) receptor [118]. The antifibrotic effects were theorised to be a result of increased fibrillin expression. Fibrillin-1 proteins form the structure of connective tissue microfibrils and are known to regulate TGF-β levels in the ECM [119]. Mutations in these proteins, pathognomonic to Marfan’s syndrome, results in elevated TGF-β and increased susceptibility to lung and kidney fibrosis in particular [120,121]. However, its exact role in RIF of the skin is unclear.

Deleon et al. noted increased expression of CD31 (angiogenesis), perilipin (adipogenesis) and fibrillin by CD146+ ADSCs [42]. The authors theorised that increased fibrillin expression may have enhanced elasticity of irradiated tissue overlying their CD146+ ASC-enriched fat grafts. Perilipin 1-deficient adipocytes have been previously shown to promote inflammation by stimulating monocyte migration and IL-6, iNOS, and IL-1β expression by macrophages [122]. Increased perilipin expression may counteract these inflammatory changes. Continued research into CD146+ ADSCs has shown further proangiogenic potential, through upregulation of VEGF and ANGPTI [123]. The latter stabilises VEGF-induced vascular endothelial cell proliferation and prevents blood vessel leakage [124].

Further proteomic studies are warranted to investigate individual ADSC subtypes (categorised by cell surface markers) and how they can be applied to treatment of RIF of the skin. These purified cells may improve lipotransfer outcomes whilst reducing potential side effects.