Introduction

Osteoarthritis (OA) is a highly prevalent, progressive, debilitating joint disease characterized by gradual loss of articular cartilage, damage to the subchondral bone and the surrounding soft tissue [1]. The avascular nature of articular cartilage limits its capacity for self-repair, resulting in progressive cartilage loss ultimately contributing to widespread degeneration of the affected joint [2]. Consequently, OA has become a primary focus for orthopaedic surgeons to provide a biologic milieu that will facilitate some form of endogenous cartilage healing [3].

Various biologic adjuncts have been described that may affect repair including growth factors and stem cells. Much of the recent literature has focused on bone marrow-derived mesenchymal stem cells (MSC) for chondrogenesis [4]. However, the clinical use of bone marrow MSCs has presented problems, including donor site morbidity and pain and low cell number upon harvest [5]. This has led to the investigation of alternative sources for MSCs [6]. Several potential donor sites have been identified for harvesting MSCs, including periosteum, muscle, synovial membrane and adipose tissue [7]. All MSCs share similar characteristics in that they have the capacity to differentiate into chondrocytes, osteoblasts, myoblasts, adipocytes, and fibroblasts depending on their differentiation potential and the conditions in which they are stimulated.

Adipose tissue has become an attractive alternative source of MSCs because of its relatively easy accessibility and abundance [8]. Although adipose may not have the same degree of differentiation as bone marrow MSCs, the relative abundance of MSCs may be more advantageous, but there is evidence suggesting that adipose derived mesenchymal stem cells (ADSCs) have reduced chondogrenic potential relating to its bone morphogenic protein characteristics and the increased age of the patient [5, 8,9,10]. However, recent evidence suggests that ADSCs obtained from lipoaspirates can be induced to express gene and matrix markers associated with chondrocyte pathways under specific conditions [11]. Significant chondrogenic effects from the application of ADSCs have been shown in in-vitro studies, proposing that ADSCs possess the CD73, CD90, CD105 and CD106 markers, which are surface markers required for cell differentiation into cartilage [12]. In addition, recent animal model studies confirmed the chondrogenesis effect of ADSCs in vivo [13]. There is also a paracrine effect of ADSCs on OA chondrocytes as they promote inhibitory macrophages and T regulatory cells, which may decrease inflammatory markers and improve clinical outcomes alongside potential cartilage regeneration [14]. This has prompted several studies to investigate the role of ADSCs in human clinical trials.

However, due to the relative novelty of using ADSCs, there has been no systematic review that demonstrates its efficacy in osteoarthritic joints to date. The purpose of this systematic review is to evaluate the effects of ADSCs in the treatment of OA in the clinical setting. The hypothesis is that ADSCs would produce effective outcome on OA repair in clinical settings supported by clinical evidence.

Materials and methods

Search strategy

The following search terms were used in MEDLINE, EMBASE and the Cochrane Library Database, databases on January 7th 2017: “(cartilage OR cartilage injury OR cartilage damage OR cartilage repair OR cartilage defect OR osteochondral injury OR osteoarthritis) AND (adipose stem OR adipose stem cell OR adipose stem cells OR adipose derived OR adipose derived stem cell OR adipose derived stem cells OR adipose derived mesenchymal stem OR adipose derived mesenchymal stem cell OR adipose derived mesenchymal stem cells OR adipose derived msc OR svf OR stromal vascular fraction of adipose tissue)”. No time limit was given to publication date.

Eligibility criteria

The inclusion criteria were: (1) clinical studies demonstrating the effects of ADSCs on OA, (2) published in a peer reviewed journal, (3) written in English, and (4) full-text of studies available. Exclusion criteria were the following: (1) review studies, (2) case report, (3) in vitro studies, and (4) animal studies.

Study selection

Two independent reviewers performed the literature search and reviewed the search results. The title and abstract were reviewed for all search results, and potentially eligible studies received full text review. In addition, the reference lists of all publications, including systematic reviews found in the search results, were manually screened for additional articles, which met the inclusion criteria that were potentially not identified through our electronic search. If a consensus could not be reached, a senior author was consulted who had the final decision.

Data extraction/analysis

The level of evidence (LOE) was evaluated based on previously published criteria [6]. The methodological index for non-randomized studies (MINORS) score [15] was used to evaluate the methodological quality of the clinical evidence (MQOE), with a score of 0–8 for case series and 0–12 for cohort studies.

The data of each clinical study was then extracted using a standardized data sheet with a list of 30 standardized variables (Table 1) [16]. The following variables were analyzed: donor site locations, volume of adipose tissue, donor site complications, adjunctive biologic therapies, concomitant treatment and clinical outcomes. In addition, the preparation and delivery methods of ADSCs was also recorded and analyzed.

Table 1 Data reported
Full size table

Results

Search and literature selection

The literature search revealed 2475 total studies and 1728 studies after duplicates were removed. Sixteen clinical studies with 635 joints were included in this review (Fig. 1).

Fig. 1
figure 1

PRISMA study selection flow diagram

Full size image

Reported clinical variability in included studies

The clinical variability of the included 16 studies is shown in Table 1. A mean of 61% of the characteristic data was reported. General demographics including age and sex were reported in 94% of studies. While the study design and patient satisfaction were all well reported variables with 88 and 77%, respectively, patient history, clinical variables and imaging data were the least reported criteria with 16, 36, and 38% of the data being reported, respectively.

Donor site locations, volume of adipose tissue, and complications

All of the studies reported the location of the harvesting of adipose tissue. The most common sites included the buttock (nine studies [2, 17,18,19,20,21,22,23, 30]), abdomen (four studies [25,26,27,28]), and infrapatellar fat pad (two studies [24, 29]). One study did not report the harvesting site [31]. The weighted average volume of harvested adipose tissue in the buttock (reported in 6/9 studies [17,18,19,20, 22, 30]) was 140 ml, in the abdomen (reported in 4/4 studies [25,26,27,28]) was 103 ml, and in the infrapatellar fat pad (two studies [24, 29]) was 9.26 g. All of the studies using the buttock or abdomen harvested with liposuction and no study reported complications in the liposuction procedure. The two studies utilizing the infrapateller fat pad harvested via open excision [29, 30].

Preparation method of SVF

All of the studies prepared ADSCs in the form of the stromal vascular fraction (SVF). SVF is a component of the lipoaspirate obtained by liposuction, comprised of ADSCs as well as variety of other cells including pericytes, vascular adventitia cells, fibroblasts, preadipocytes, monocytes, macrophages, and red blood cells [31]. Eleven studies used the Zuk et al. method of ADSC preparation [32]. Eight studies [2, 19,20,21,22,23, 29, 30] reported liposuction the day before surgery, four studies [24,25,26, 28] reported liposuction the day of, and four studies [18, 21, 27, 31] did not report the date of liposuction. After the liposuction, the procedure of the adipose tissue was transported to the laboratory, where adipocytes and connective tissue were separated from the stromal vascular fraction (SVF) by centrifugation, according to the method by Zuk et al. [32]. The majority of the studies analyzed the SVF cells to confirm chondrogenic potential, including 12 studies via flow cytometry, three studies via cell count, 13 studies via culture in a medium. Only two studies did not report analysis of SVF cells [25, 28]. The mean concentration number of ADSCs in the SVF was 6.3 × 106 (range; 1.2 × 106 − 4.2 × 107) in the ten studies that reported this [2, 17,18,19,20, 22, 23, 28,29,30]. The mean percentage of ADSCs in the SVF was 9.2% (range 8.5–9.7%) in the 5 studies [2, 18, 20, 23, 30] that reported this. One study [31] used three different doses with dose escalation: low dose (2 × 106 cells), medium dose (10 × 106), and high dose (50 × 106).

Treatment

Of the 16 studies that treated patients using the SVF, 5 were case series (Table 2) involving OA knees using arthroscopic debridement, with 4 case series using injections in multiple OA joints [17, 18, 22, 24, 25, 26, 28, 29, 31]. All ten studies involved patients with OA joints and reported significant improvement in clinical outcome measures. In six studies, the SVF was used with biologics (PRP or fibrin) [17, 18, 24, 25, 28, 29]. The mean lesion size was reported by three of the ten studies, and ranged between 5.4 and 6.2 cm [2, 17, 22]. 

Table 2 Study characteristics—case series
Full size table

 

Of the 16 studies that treated patients using the SVF, 7 were comparative studies [2, 19, 20, 21, 23, 27, 30] (Table 3), All 7 comparative studies reported significant improvement of functional outcomes including three studies that compared SVF to no SVF. Six studies involved OA knees [2, 20, 21, 23, 27, 30], and one study involved OA ankles [19]. All studies in OA knees used SVF with a biologic (PRP or fibrin), while the study in the ankle compared SVF alone to no biologics. The mean lesion size was reported by three of the seven studies, and ranged between 4.6 and 6.4 cm [2, 20, 30]. The clinical outcomes of the comparative studies are reported in Table 4.

Table 3 Study characteristics—comparative studies
Full size table
Table 4 Study results—comparative studies
Full size table
Table 5 Study results—radiographic evidence
Full size table

Radiological outcomes

Eight of the included studies reported the radiological outcomes using MRI or X-ray [18, 24, 25, 26, 27, 28, 30, 31]. The reported outcome measures were mixed, but the evidence was promising. MRI findings showed improved cartilage thickness in the majority of the included studies, with only one study finding no difference between the pre and post-treatment scores. Additionally, no study found any evidence of tumor formation. The radiological outcomes reported are shown in Table 5.

Patient satisfaction and complication rate

Five of the included studies reported the satisfaction rate, all of which were above 90% [2, 17, 23, 24, 26]. Seven of the included studies reported the complication rate. Two studies [28, 31] reported a 10 and 37% complication rate of swelling and four studies reported no complications [21, 24, 25, 27]. The other included studies did not mention any complications.

Discussion

The most important finding of this systematic review was that ADSCs in the form of SVF produce favorable clinical outcomes with minimal complications in the treatment of OA. Additionally, several studies showed that patients receiving SVF had better radiographic outcomes compared to the baseline [18, 24, 25, 27, 29]. However, caution should be taken to interpret the outcomes of the present study. This systematic review demonstrates the variability in both the data collected and the use of biological adjuvants between studies, thereby limiting proper cross-study comparisons and potentially confounding the effects of ADSCs. Therefore, while SVF may produce favorable clinical outcomes for patients with OA, the confounding factors limit our understanding of the effectiveness of ADSCs.

There have been many studies evaluating the effects of ADSCs in cartilage repair; however, the terminology describing the use of ADSCs has been inconsistent and confusing with terms including adipose-derived adult stem (ADAS) cells, adipose-derived mesenchymal stem cells (AD-MSCs), adipose MSCs (AMSCs), and adipose stromal/stem cells (ASCs). The results from this systematic review demonstrated inconsistencies on what components of ADSC were used in the various studies reviewed when evaluating the effects of ADSCs. The results demonstrate that analysis has focused on the SVF rather than ADSCs. The SVF contains different proportions of ADSCs, pericytes, vascular adventitia cells, fibroblasts, preadipocytes, monocytes, macrophages, and red blood cells [33]. Therefore, analysis of the SVF limits the ability to evaluate the effectiveness of only the stem cell component of the fraction. There has been limited data evaluating other components of SVF such as pericytes and the vascular fraction and therefore these human trials cannot determine the role each factor has in cartilage regeneration.

The results of this systematic review also demonstrate the use of an adjunctive procedure in the majority of the studies evaluated (12/16 studies). These studies utilized either PRP (7/16) or fibrin (4/16), with one study comparing both biologic adjuncts. These studies demonstrated favorable outcomes when SVF and a biologic were used. Few studies have compared SVF to the adjunctive biologic itself, confounding meaningful analysis. There is evidence that PRP contains growth factors that increases chondrocyte viability and differentiation, as well as the synthetic capacity of MSCs, which may prove beneficial in cartilage repair [34, 35]. There have been several level I and II clinical studies demonstrating significant effects and positive clinical outcomes of PRP applications in the treatment of OA and osteochondral defects [36, 37]. There is also evidence of fibrin as a stem cell carrier [34], however, its effect on cartilage repair is largely unknown. Both PRP and fibrin can act as a scaffold and may enhance ADSC adherence to cartilage lesions and promote their proliferation [38]. The application of either PRP or fibrin confounds the effects of ADSCs in OA; therefore, the advantages of ADSCs beyond the use of these adjuncts cannot be determined from this study.

The results from the current study demonstrate large variability in the data reported thereby limiting the ability to analyze the results through meta-analytical methods. Two different sites including the knee and ankle were evaluated in the literature, which result in variable responses in cartilage repair. In addition, many studies failed to report lesion size, which is an important prognostic factor in the repair process of cartilage [39]. Another reported variable was in the donor sites location, which varied between the buttock and infra-patellar fat pad. Although liposuction has been shown to produce a higher percentage of viable cells in lipoaspirates when compared to surgical resection of adipose tissue [35, 37], the number of nucleated cells can range from 500,000 to 2,000,000 cells per gram of adipose tissue [40]. Recent evidence has shown that different individuals have variable density in different locations of adipose tissue [41] and that there may be differences in the multi-potency of ADSCs depending on the harvest location [42, 43]. In addition to the extraction methods and the differences in adipose tissue, the process and preparation of SVF are an important assessment area in this systematic review.

There was significant variability in the preparation methods reported in the included studies, and in the reported methods of assessing chondrogenic potential. The reported percentage of ADSCs in the SVF was only 9.2% in the studies assessing this, showing that the majority of the injected material was in fact the pericytes, vascular adventitia cells, fibroblasts, preadipocytes, monocytes, macrophages, and red blood cells. Additionally, those studies found a wide-ranging difference in the percentage of included ADSCS in SVF, although no analysis of factors relating to differences where performed. Previously, studies have found that older patients have decreased growth factors and chondrogenic potential in autologous blood products compared to younger patients [44]. While the Zuk method of SVF preparation [32] originally reported using collagenase to separate the ADSC from the surrounding area, the authors in the included studies did not report using any collagenase indicating that the method may have been altered. It is worth noting that collagenase digestion cannot be used in the United States. The concentration and incubation time of collagenase can affect the yield of ADSCs as higher dosages or exposures have been shown to be toxic to ADSCs [38]. Excess amounts of collagenase can decrease the ADSC viability while insufficient amounts may result in inefficient and inadequate amounts of ADSC yield [12, 38]. Although all of the included studies reported improved clinical outcomes, the lack of standardization in the clinical outcome measures, the operative treatment, and concomitant biologic use limits adequate comparison thus confounding any meaningful clinical analysis.

Five of the studies (42%) reported on patient’s reported outcomes with all above 90% satisfaction. However, of the five studies reporting on patient satisfaction only 2 reported radiological findings, with mixed results. Koh et al. found high patient satisfaction alongside improved radiological outcome measures, while Fodor et al. showed high patient satisfaction with no change in radiological measures at 3 months [23, 26]. These findings suggest that the patient satisfaction may be at least partially due to the paracrine effect of the SVF rather than true cartilage repair alone. Six (50%) of the included studies reported an improvement in VAS scores. The complication rate of the liposuction procedure was not reported in any of the included studies. Previous evidence has shown this to be a safe procedure, with complication rates of only 0.1% in a national survey of 112,756 reported patient procedures [45]. The outcomes from this study, therefore, suggest that injection of ADSCs in the form of SVF may be a safe procedure with a low complication rate.

This systematic review has several inherent limitations and potential biases. The search criterion was limited to MEDLINE, EMBASE and The Cochrane Library Database articles published exclusively in English. Eleven of the surgical studies came from one center: The Center for Stem Cell & Arthritis Research, Department of Orthopaedic Surgery, Yonsei Sarang Hospital, Seoul, Korea. This may suggest that the results could be partially attributed to the surgeons’ skill sets, as there is a lack of variance in the operating surgeons for the procedures compared to normal systematic reviews. There were few studies that included a true control, as the majority compared the results to patients treated with PRP alone, which limits the comparison. There is also evidence that ADSCs have a paracrine effect on OA chondrocytes by producing cytokines, anti-inflammatory mediators, immunoregulatory molecules, and there is a potential placebo effect from the injections, both which may explain at least partially explain the improvement in clinical symptoms [14]. The overall quality of evidence of published studies was variable in this review. The LOE in the included studies was low with 14 (82%) of the studies being LOE III or IV, with no reported randomized clinical trials.

Conclusions

This systematic review demonstrates that the majority of studies in the current literature utilize ADSCs in the form of SVF when evaluating the outcomes in the treatment of OA. While SVF may produce favorable clinical outcomes with minimal risk of side effects on osteoarthritis, the variability in the data and the use of biological adjuvants have confounded the effectiveness of ADSCs.