Stem cell therapy for the treatment of early stage avascular necrosis of the femoral head: a systematic review

Background Avascular necrosis (AVN) of the femoral head (FH) is believed to be caused by a multitude of etiologic factors and is associated with significant morbidity in younger populations. Eventually, the disease progresses and results in FH collapse. Thus, a focus on early disease management aimed at joint preservation by preventing or delaying progression is key. The use of stem cells (SC) for the treatment of AVN of the FH has been proposed. We undertook a systematic review of the medical literature examining the use of SC for the treatment of early stage (precollapse) AVN of the FH, in both pre-clinical and clinical studies. Methods Data collected included: Pre-clinical studies – model of AVN, variety and dosage of SC, histologic and imaging analyses. Clinical studies – study design, classification and etiology of AVN, SC dosage and treatment protocol, incidence of disease progression, patient reported outcomes, volume of necrotic lesion and hip survivorship. Results In pre-clinical studies, the use of SC uniformly demonstrated improvements in osteogenesis and angiogenesis, yet source of implanted SC was variable. In clinical studies, groups treated with SC showed significant improvements in patient reported outcomes; however hip survivorship was not affected. Discrepancies regarding dose of SC, AVN etiology and disease severity were present. Conclusions Routine use of this treatment method will first require further research into dose and quality optimization as well as confirmed improvements in hip survivorship.

SC have been shown to promote bone formation [4,6,11,12] and neovascularization [6,11] in vitro. Additionally, patients treated with SC in conjunction with CD demonstrated significant improvements in Harris Hip Scores (HHS) [15] as well as decreased hip pain and symptoms compared to those treated with CD alone [2]. Yan et al. documented that stem cells implanted into the necrotic FH not only survive, but thrive and proliferate [15]. Although the pathogenesis of AVN is unclear [17] many hypothesize that SC work to improve early stage AVN potentially as a function of their critical role in the regulation and improvement of osteogenesis and angiogenesis [11,12]. Furthermore, it is thought that mesenchymal SC implanted into the necrotic FH may differentiate into osteoblasts or vascular endothelial cells, thereby promoting bone repair and regeneration [12]. Despite encouraging results in preclinical (basic science) and clinical studies, improvements in hip survivorship or time to THA has not been uniformly reported and remains controversial [2,8,14].
The purpose of our study was to perform a systematic review of the current medical literature on the treatment of early stage AVN of the FH using SC implanted via CD. We examined both preclinical studies and clinical studies. We reported bone healing outcomes (histologic and imaging outcomes) from preclinical papers and all examined outcomes from available clinical papers.

Eligibility criteria
Manuscripts were deemed eligible for our review if they evaluated treatment of early stage AVN of the FH with SC implanted via CD. We defined early stage AVN as precollapse of the FH. Both clinical and preclinical manuscripts were selected. For clinical trials, we included studies on patients age > 18. All types of clinical studies were eligible for inclusion to this review. Studies of all languages were eligible for inclusion to this review. For studies reporting on the same group of patients at multiple follow up periods, the most recent publication was used in this review. For preclinical studies, manuscripts were eligible if they examined bone healing either histologically, or by imaging techniques. If studies examined other treatments such as vascularized fibular grafting or bone morphogenic proteins, they were excluded unless the data on SC and CD were presented separately from the other treatments, to allow us to examine the effect of SC specifically.

Study identification
A systematic, computerized search for potential manuscripts was performed by three independent reviewers (HE, SM, RL). Pubmed (−July 2012), Ovid Medline (−July 2012) and EMBASE (−July 2012) databases were used to identify studies. Key words used for the search were: AVN or avascular necrosis or osteonecrosis AND stem cells; AVN or avascular necrosis or osteonecrosis AND autologous bone marrow. Abstracts were retrieved for all manuscripts considered relevant by title. Abstracts were independently reviewed and any disagreements were resolved by discussion. Full length articles of relevant abstracts were reviewed for inclusion. Bibliographies of the full length articles were also searched for other potential studies and full length articles were retrieved.
Outcomes data were extracted by two reviewers (HE, RL) using prearranged summary tables. Data extraction for preclinical studies included study design, animal model, type of SC used, sample size, and outcomes measured. For the clinical studies, study type, sample size, potential biases, AVN classification, AVN etiology, SC dose and cell type, and outcomes measured were recorded.

Study selection
We identified 215 abstracts using our electronic search. Thirty-four met the initial screening inclusion criteria and the full-length articles were retrieved and reviewed. Following the full-length reviews, 16 studies (11 preclinical, 5 clinical) met our inclusion criteria and were retained for this review. Eighteen studies were excluded: 3 were review articles [18][19][20], one was a surgical technique article [21], 3 articles were early/pilot results which were subsequently included in future articles [7,16,22], 6 articles examined SC treatment in conjunction with therapies other than CD (i.e. vascularized fibula) [23][24][25][26][27][28], two articles used SC transplanted intravenously instead of by CD [13,29], 1 article examined prevention of AVN not treatment [5], and two articles included patients with post collapse AVN of the FH [1,15] (Figure 1).
Five studies performing a histologic assessment of neovascularization following SC treatment found improvement (p < 0.05) in the SC treatment groups compared to CD alone [6,11,[30][31][32]. One study found higher levels of neovascularization compared to no treatment [17]. One study did not perform statistical analysis but reported higher levels of neovascularization [34] (Table 1).

Imaging (6 studies)
On CT imaging, Wen et al. found that the SC group had imaging that closely resembled the normal hip in comparison to CD alone [30]. On MRI imaging they found decreased FH necrosis volume compared to no treatment, and on CT perfusion imaging they found blood volume to be similar to normal hips (no statistics reported) [30]. Abudusaimi et al., Xie et al. and Aimaiti et al. found significantly higher bone volume, bone mineral density and trabecular volume in the SC group compared to CD alone (p < 0.05) on micro CT [10,32,34].
Sun et al. performed micro CT angiography and found improved angiogenesis (p < 0.05) [6]. Hang et al. found that the CD only group had decreased uptake of radioactivity on SPECT compared to SC treatment group, indicative of worse perfusion [11]. Radiographs also revealed that the CD only group had worse results, with irregular articular surfaces and heterogenous density of the FH in comparison to the SC group (no statistics reported) [11] ( Table 1).

Sample size
Sample size (n) was recorded as number of hips. Combined, the 5 studies comprised 763 hips. Across studies, n varied from a minimum of 24 [2] to a maximum of 534 [35] ( Table 2). None of the studies reported a power calculation.

Stem cell treatment protocol
Four studies utilized concentrated BMMNC harvested from the iliac crests of each patient as their SC treatment [2,9,14,35]. Zhao et al. used cultured autologous BMMNC harvested from the proximal femur of each patient as their SC treatment [8]. After bone marrow was harvested, in vitro expansion of the mesenchymal SC was performed for 2 weeks prior to reimplantation. SC treatment was  performed in conjunction with CD for all clinical studies ( Table 2).

Stem cell dose
SC dose was measured using a variety of methods -total mononucleated cell count (TMNCC), CD34+ cell count, number of fibroblast colony forming units (F-CFU) and total mesenchymal SC count (TMSCC). TMNCC is a measure of all the nucleated cells including mature nucleated cells and mesenchymal SC. CD34+ cells are a marker of hematopoeitic precursor cells and F-CFU are a measure of mesenchymal SC (each F-CFU is thought to arise from clonal expansion of mesenchymal SC). TMSCC is an approximate count of the number of SC and was used by Zhao et al. prior to reimplantation of cultured mesenchymal SC [8]. TMNCC was measured in 4 studies and ranged from 5 × 10 8 to 19 × 10 8 cells [2,9,14]. CD34+ cell counts were measured in two studies (1.9 × 10 7 to 5 × 10 7 cells) [2,14]. F-CFU counts were measured in two studies (mean F-CFU count of 1.76 × 10 3 to 2.4 × 10 3 ) [2,35]. TMSCC was reported in one study (2.0 × 10 6 cells) [8] ( Table 3).

Patient reported outcomes (5 studies)
Four of the studies reported HHS [8,9,14,35]. Zhao et al. found a statistically significant increase in HHS 60 months following SC therapy compared to CD only in patients with ARCO stage 1C/2B/2C (p < 0.05, raw score not reported), and a trend for ARCO stage 2A (p = 0.06, raw score not reported) [8]. Sen et al. found a statistically significant difference in HHS for the SC treated patients compared to CD only at 12 months follow up (83.65 ± 8.04 vs. 76.68 ± 13.86, p < 0.05) [14]. At 24 months, overall HHS was not significantly improved between the two groups (82.42 ± 9.63 vs. 77.39 ± 16.98, p = 0.09), but the pain and deformity domains of the HHS were still in favor of the SC therapy group (p < 0.05, raw score not reported) [14]. Wang [35]. They reported HHS of 70 preoperatively and 88 post   SC treatment at an average of 13 years follow up (no p value reported) [35].

Progression of avascular necrosis (3 studies)
Two of 3 studies found a significant decrease in the proportion of patients progressing to ARCO stage 3 or 4 following SC treatment compared to CD only treatment, 60 months post procedure (p < 0.05) [ Table 4).

Volume of necrotic lesion (3 studies)
Zhao et al. reported a significant decrease in lesion volume in SC treated hips compared to CD only treated hips with a pretreatment ARCO stage of 2B and 2C (ARCO 2B = 6.5% vs 13.3% of FH, ARCO 2C = 13.8% vs 29.3% of FH, p < 0.05) at 60 months post treatment [8]. Gangji et al. reported a significant decrease in lesion volume after SC treatment compared to CD only at 24 months follow up (SC group volume decreased 42%, CD group volume decreased 1%, p < 0.05), and a trend towards decreased lesion volume at 60 months (SC group volume decreased 42%, CD group volume decreased 22%, p = 0.06) [2]. Hernigou et al. reported on lesion volume in 371 of 534 hips in their case series [35]. They found a decrease in lesion size from 26 cm 3 to 12 cm 3 at an average of 12 years follow up (no statistical analysis reported) [35] ( Table 4).

Discussion
This review systematically examined the current literature on SC therapy for the treatment of early stage (precollapse) AVN of the FH including clinical and preclinical studies. Preclinical studies yielded encouraging results for treatment of AVN of the FH with SC. Although the source of SC varied among studies, SC treatment of AVN uniformly demonstrated improvements in osteogenesis and vascularization. All 11 studies showed positive effects with respect to bone formation in groups treated with SC. Furthermore, reported X-ray, SPECT, CT and MR outcomes from all studies favoured the SC treatment group. Bone marrow was the most common source of SC but other sources such as adipose and dental pulp were identified. SC isolated from dental pulp represents a relatively new treatment option with noteworthy potential for use in orthopaedics [4]. Adipose derived SC are another potential alternative to SC from bone marrow. Advantages of adipose derived SC include abundance, ease of isolation, rapid expansion, and multipotency [10].
Despite positive results, relevance of animal models described in preclinical studies should be considered. Corticosteroid and liquid nitrogen induced AVN of the FH are widely recognized means for induction of AVN in numerous animal models; both lead to ischemic conditions within the FM and eventual osseous infarction producing changes phenotypically similar and clinically relevant to human disease [36]. Some studies, however, have addressed the significance of larger animal models, particularly with respect to translational medicine, as they may better replicate conditions in human AVN. An ovine model of AVN of the FH may better reflect articulation in early-stage human AVN as compared to other disease models [4]. For liquid nitrogen induced disease, the bone defect produced, which is non-negligible in animals with small diameter FHs, has been proposed as a limitation due to its absence from human pathology. This has led some researchers to reject use of this method on rabbits [37].
Results of clinical studies were also encouraging. In our review, the clinical studies used CD as a means for implanting SC directly into the necrotic region of the FH, in the form of a cell suspension. CD works by decreasing intra-osseous pressure and improving circulation and vascularization [9]. Used alone, however, CD exhibits inconsistent outcomes including poor lesion reconstruction, ultimately leading to FH collapse [9,14]. The progression of AVN of the FH occurs in consequence of a limited capacity for articular tissue self-repair [3,11,14], including decreased osteogenesis [11,14] and vascularization [3]. This may occur as a result of inadequate numbers of progenitor cells in the proximal femur of patients with AVN of the FH [38]. It is thought that SC implanted into the necrotic region of the FH work to repopulate the low numbers of progenitor cells [20]. Pluripotent, mesenchymal SC differentiate into various cell types, namely osteoblasts, thereby improving repair mechanisms and potentially reversing damage to bone [11,12,14]. In addition to directly increasing bone formation by differentiating into osteoblasts, it is hypothesized that mesenchymal SC have an indirect effect by the expression of cytokines which influence osteogenesis and neovascularization [39,40]. In general, clinical studies reported improvements in patient reported outcomes for those treated with SC; notably, the HHS. Similarly, studies that examined progression to more advanced disease, and lesion volume reported improvements for the SC treatment group. Participants treated with SC did not experience consistent improvements in hip survivorship across studies. None of the studies using a comparative group found worse outcomes for SC treatment.
Considerable variations and inconsistent reporting among clinical studies were observed regarding the dose of SC, etiology of AVN, lesion size, and severity/classification of disease making comparisons between studies challenging. However, there are currently limited numbers of clinical studies addressing SC therapy for treatment of AVN of the FH, and even fewer addressing early-stage disease and administration of SC by CD. Accordingly, we were unable to perform meta-analysis on study results. Quantitative assessment will be a prerequisite to making definitive conclusions on vital therapyrelated factors such as SC dose and quality.
Standardization of SC dose has proven difficult due to a lack of uniformly accepted, reliable cell markers which can be used to identify mesenchymal SC [39]. However, the dose of SC used has been reported to impact disease outcome [7]. Both SC dose and quality are also known to affect their clinical potential. Density of SC transplanted to the necrotic FH was shown to affect the rate of osteogenesis, and thereby bone repair [12]. Quality of transplanted cells affects their proliferative capacity [41]. Prior to routine use of combined SC/CD therapy, defined standards of SC dose and quality, such as CD34+ or CFU counts [42], will likely have to be set in order to accurately evaluate the effect of each therapy. However, as a result of presently observed inconsistencies, and a paucity of studies in this area, further research, examining both SC dose and quality will be prerequisite to routine clinical use of this therapy.
Though not specifically addressed by studies assessed in this review, the impact on treatment outcome of whether cells were derived from a concentrate or a culture may also represent an area for future research. Pre-clinical and clinical studies included examples of both concentrated and cultured cells. Concentrated cells contain all cells and cell types present in the tissue from which they have been derived, not only SC. Concentrated BMMNC from bone marrow aspirate contain hematopoietic progenitor cells, lymphocytes, leucocytes, in addition to nonhematopoietic cells including MSC, EPC, embryonic-like SC expressing pluripotent markers, and other multipotent or committed cells [43]. Cultured cells, conversely, represent an isolated pool of SC. The comparative regenerative capacity of concentrated vs. cultured cells remains unclear. Despite positive results observed for both treatments within this review, other studies explicitly assessing differences between the two have displayed mixed findings. Use of pure, cultured MSC led to greater improvements in ischemic limb diseases, compared to concentrated BMMNC, in both human and rat models [44,45]. Alternatively, BMMNC use displayed beneficial outcomes in treatment of spinal cord injury when compared to MSC [46]. Cost and feasibility must also be considered when selecting an appropriate treatment. Indeed, cultured cells require greater preparation times and are associated with increased cost [44,47]. Ultimately, the outcomes of concentrated vs. cultured cells should be assessed for the specific treatment of AVN of the FH in order to develop future robust clinical guidelines for cellular intervention in this disease.
Etiologic risk factors of AVN are also known to significantly affect treatment outcomes [2]. It has been demonstrated that the capacity for SC to differentiate into the necessary osteogenic cells for bone repair and remodeling is limited in patients with alcohol and steroid induced AVN of the FH due to differences in the ischemic environment [12,48]. The size of the osteonecrotic lesion is also known to affect overall patient outcome no matter the method of treatment used [8,49]. Future studies should aim to use the same AVN classification system as well as account for AVN etiology and lesion size as potential confounding variables.
Several other reviews [18,38,[50][51][52][53] have been published discussing the use of SC for the treatment of AVN of the femoral head. However to our knowledge, ours is the first systematic review that includes data from several recently published clinical trials [2,8,14]. Our review included data from over 700 hips, more than previously published reviews. Additionally, our review included both clinical and pre-clinical studies, furthering the breadth of our review. A limitation of any systematic review is in the quality of the papers available for review. Clinical studies included in our review did not provide sample size and power calculations. Preclinical studies did not always use a classification system to identify stage of AVN of the FH. There were a limited number of comparative trials, and only two RCTs. We included all types of clinical studies, potentially introducing confounding and selection bias. We felt that inclusion of these studies would provide a more comprehensive review of the literature surrounding this topic. Furthermore, meta-analysis was not performed due to the limited number of comparative trials and variable methodology employed in the studies.

Conclusions
AVN of the FH primarily affects younger, working age individuals and thus leads to increased morbidity and functional disability in this population [13,30]. Treatments aimed at halting or delaying progression of disease would provide a welcome alternative to those faced with progression to joint collapse and hip replacement surgery. Combining CD, the most widely used treatment for AVN of the FH to date [30], with SC, could result in a novel long lasting hip preserving treatment option.
Pre-clinical studies have demonstrated the potential of SC for reversing the debilitating damage to the femoral head associated with AVN in vitro. Current clinical studies have suggested beneficial effects on patient reported outcomes, but definitive conclusions regarding hip survival and disease progression cannot be made. Further, more refined clinical studies are needed to establish the effectiveness of SC treatment in AVN of the FH. Quantitative studies including meta-analyses aimed at addressing SC dose and quality standards will be necessary to make fundamental treatment-related deductions. The effects of concentration and culture based preparatory methods for SC therapy should be compared in the context of treating AVN of the FH in order to establish suitable protocols. Demonstrated improvement in hip survivorship is prerequisite to the future of this treatment.

Competing interests
There are no competing interests for any author of this study.
Authors' contributions RL contributed to the design of the study, performed collection and assembly of data, analysis and interpretation of data and drafting and critical revision of the article. RG conceived and designed the study, drafted and critically revised the article, and gave final approval of the article. HE performed data collection and assembly and drafting of the article. SM performed data collection and assembly and drafting of the article. NM participated in study design and coordination, revised and gave final approval of the article. All authors read and approved the final manuscript.
Role of the funding source No external funding was provided for the purposes of this study.