The effect of bone marrow mononuclear cells on vascularization and bone regeneration in steroid-induced osteonecrosis of the femoral head

Joint Bone Spine 76 (2009) 685–690

Original article

The effect of bone marrow mononuclear cells on vascularization and bone regeneration in steroid-induced osteonecrosis of the femoral head Yuan Sun , Yong Feng , Changqing Zhang ∗ Department of Orthopaedics, Shanghai Sixth People’s Hospital, 600 Yishan Road, Shanghai Jiao Tong University, School of Medicine, Shanghai 200233, China Accepted 2 April 2009 Available online 2 July 2009

Abstract Objective: We examined whether implantation of autologous bone marrow mononuclear cells (BM-MNC) can augment neovascularization and bone regeneration in steroid-induced osteonecrosis of the femoral head. Methods: Sixty-five 28-week-old male New Zealand white rabbits were divided into group I (left untreated, N = 20), group II (core decompression, N = 20) and group III (core decompression + autologous bone marrow cells implantation, N = 25) after receiving an established inductive protocol for inducing steroid-associated ON. Four weeks later, these rabbits were euthanized, bilateral femora were dissected for micro-CT-based microangiography to assess vascularization, and then the osteonecrotic changes and repair processes were examined histopathologically. Results: Quantitative analysis showed that new vessel formation in group III was significantly greater compared with other groups at 4 weeks after treatment. Penetrating capillary vessels number vessels number in group III (44.5 ± 5.11) was significantly larger than that of group II (11.4 ± 2.46) and group I (3.10 ± 0.33) (p < 0.01). The histologic and histomorphometric analysis revealed that the new bone volume was significantly higher in the group III than in the group I and II, 4 weeks after treatment. Conclusion: In this animal model, a combination of bone marrow mononuclear cells and core decompression enhance the neovascularization and the osteoinductive ability, resulting in bone regeneration. These findings confirm the preliminary clinical results obtained in humans that the implantation of bone marrow mononuclear cells is an effective and feasible method for treating early osteonecrosis. © 2009 Société franc¸aise de rhumatologie. Published by Elsevier Masson SAS. All rights reserved. Keywords: Bone marrow mononuclear cells (BM-MNC); Osteonecrosis; Vascularization; Micro-CT; Bone regeneration

1. Introduction Osteonecrosis of the femoral head (ONFH) has been reported to occur in patients who have received corticosteroids as treatment for underlying diseases, such as systemic lupus erythematosus, nephrotic syndrome, and renal transplantation [1]. One of the common treatment options for patients with advanced ON is total hip replacement. Several studies, however, have observed poor prosthetic durability in younger patients with ON [2]. The early intervention prior to the collapse would therefore be a better strategy for such patients. Treatment options for early stage ONFH include electrical stimulation, core decompression, rotational osteotomy, nonvascularized and vas-

∗

Corresponding author. E-mail addresses: [email protected] (Y. Sun), [email protected] (C. Zhang).

cularized bone grafting [3,4]. Core decompression used to be the most widespread procedure to treat the ONFH; its efficacy still remains controversial [5]. The vascularized fibular graft has a satisfactory success rate as high as 90% in early stage osteonecrosis and is superior to core decompression, however, there is still a great concern with its complications [6]. This has prompted the investigation into a novel method for treatment of ONFH. Hernigou and Beaujean firstly tried to apply bone marrow grafting to treat osteonecrosis and achieved good results [7]. Recently, Gangji et al. suggested that the beneficial effects of one procedure result from implantation of autologous bone marrow mononuclear cells (BM-MNC). In their study, implantation of autologous BM-MNC appears to be a safe and effective treatment for early stages of ONFH [8]. Although the findings of this study are promising, their interpretation is limited because of the small number of patients and the short duration of follow-up. At the same time, the explanation that injected marrow stromal cells

1297-319X/$ – see front matter © 2009 Société franc¸aise de rhumatologie. Published by Elsevier Masson SAS. All rights reserved. doi:10.1016/j.jbspin.2009.04.002

686

Y. Sun et al. / Joint Bone Spine 76 (2009) 685–690

increased angiogenesis and subsequent improvement in osteogenesis for the therapeutic effect was postulating. So, further study is needed to confirm the results. The purpose of this study was to determine whether BMMNC implantation truly plays a positive role in angiogenesis and bone regeneration of femoral head in a steroid-induced rabbit model. 2. Methods All experimental procedures adhered to the recommendations of the US Department of Health for the care and use of laboratory animals, and were approved by the Ethics Committee of Shanghai Jiao Tong University. 2.1. Animal ON model and treatment protocol Sixty-five 28-week-old male New Zealand white rabbits with body weight of 3.5–4.5 kg were housed at the Experimental Animal Center of the investigators’ hospital and received a standard laboratory diet and water ad libitum. An ON model of the femoral head was established according to the previous inductive protocols [9]. In brief, one injections of 10 ␮g/kg body weight of lipopolysaccharide (LPS) (Sigma) were given intravenously, and then three injections of 20 mg/kg body weight of methylprednisolone (MPS) (Pfizer, USA) were given intramuscularly, at a time interval of 24 h. It was reported that ON gradually developed 6 weeks after injection of MPS, which was similar to stage II of ON clinically (Ficat and Arlet classification system). No rabbits died of the inductive protocol throughout the experiment period. Six weeks after the last injection of MPS, the rabbits were randomly divided into three groups. Group I (N = 20) served as controls and did not receive any therapy. Group II (N = 20) only underwent bilateral core decompression of the femoral head. Group III (N = 25) underwent bilateral core decompression of the femoral head and autologous BM-MNC transplantation. Then 4 weeks later, the femoral head of all rabbits were assessed by microangiography and histological examination. All of the treated animals lost weight and several rabbits had to be killed or died precociously. Thus, only 16 of the 20 animals (32 femoral heads) in group II and 20 of the 25 animals (40 femoral heads) in group III were available for study. 2.2. Surgical procedure For the core decompression, the animals were anesthetized with intravenous pentobarbital and operated according to the modified previous technique [10]. Normal saline was infused intravenously through an ear vein during the procedure. A standard lateral approach to expose the lateral aspect of the femur just distal to the greater trochanter was made under aseptic conditions. A drill with an outer diameter of 1 mm was inserted at the flare of the greater trochanter and into the femoral neck and head. The location was confirmed radiographically. Both sides of femoral head were performed. The wound was then closed layer by layer.

2.3. Isolation and implantation of rabbit BM-MNC For the rabbits in group III, the bone marrow harvesting was performed during the same operative session as the core decompression. Bone marrow cells (5 ml) were aspirated from the ilium and placed in heparinized phosphate buffered saline (PBS). Rabbit BM-MNC were isolated by Percoll (Gibco, USA, 1.077 g/L) density-gradient centrifugation. A total of 5 × 106 BM-MNC per animal were resuspended in 1 ml PBS. Half of the composite solution (0.5 ml) was slowly injected into the right femoral head through the tunnel made by drill; another 0.5 ml was injected into the left side. The hole was sealed by an absorbable collagen sponge plug. 2.4. Assessment of femoral head vascularization Femoral head blood vascularization was measured using a new technique- the micro-CT technique. Eight rabbits of each group were assessed at 4 weeks after treatment. 2.4.1. Perfusion and decalcification Under general anesthesia with 3% sodium pentobarbital (1 ml/kg), the abdomen cavity of the animals was opened (Fig. 1A), and a scurf needle with 25 mm syringe was inserted in the abdominal aorta distal to the heart with ligation of that proximal to the heart. The vasculature was flushed with 50 U/ml heparinized normal saline at 37 ◦ C and at a flow speed of 20 mm/min via a syringe. As soon as the outflow from an incision of the abdominal vein was limpidness, 10% neutral buffered formalin (37 ◦ C) was pumped into the vasculature to fix the nourished skeletal specimen. The formalin was then flushed from the vasculature using the heparinized normal saline, and the vasculature was injected with Microfil, a lead chromate-containing confected radiopaque silicone rubber compound based on the manufacturer’s protocol (Microfil MV-122, Flow Tech; Carver, MA, USA) (Fig. 1B). Animals were then euthanized with an overdose of sodium pentobarbital and stored at 4 ◦ C for 1 h to ensure polymerization of the contrast agent before microangiography. Bilateral femoral samples were then harvested and fixed in paraformaldehyde (4%) for decalcified with ethylenediaminetetraacetic acid (EDTA) (10%, pH 7.4). Success of decalcification was confirmed by anteroposterior view radiographs taken using a cabinet X-ray system (Specimen Radiography System, Faxitron 43855C, Fraxitron X-ray Corporation, Wheeling, IL, USA) under an exposure condition of 40 kV/30 s. Then, both proximal 1/3 of bilateral femoral samples of each group were obtained for evaluations. 2.4.2. Microangiography With the help of one experienced micro-CT application specialist, proximal part of the bilateral femoral samples of each rabbit were placed into a polymethylmethacrylat (PMMA) sample tube, respectively. The femoral shaft was fixed in the tube with its long axis perpendicular to the bottom of the tube for micro-CT scanning using micro-CT (GE, USA). The scan was preformed at a resolution of 36 ␮m per voxel with 1024 × 1024 pixel image matrix. For segmentation of blood vessels from

Y. Sun et al. / Joint Bone Spine 76 (2009) 685–690

687

in the specified microfil range represented the vessel volume (VV). After microangiography, the decalcified samples were embedded in paraffin, cut into 5-␮m-thick sections along the coronal plane for the proximal parts. Sections were stained with hematoxylin–eosin (HE) for evaluation of osteonecrosis and repair process. 2.5. Tissue preparation and assessment of bone regeneration around the necrotic zone Other rabbits of each group were euthanized at 4 weeks after treatment, and bilateral proximal one-thirds were harvested. After fixing in 10% neutral formalin, the extirpated bone samples were decalcified in 10% EDTA, pH 7.4. Decalcified bones were embedded in paraffin and sectioned at 5 ␮m, and stained with HE. Histological analysis was performed to observe the osteonecrotic changes and repair processes in the femoral heads at 4 weeks after treatment. New bone was measured with image-analysis software (NIH image) and new bone density was defined as the ratio of new bone area to total implant area × 100. Osteoid tissue was excluded from new bone calculations, since the border between the new bone and osteoid tissue is unclear, and inclusion of osteoid tissue leads to overestimation. A pathologist, blinded to the treatment conditions, evaluated all sections. 2.6. Statistical analysis Data are expressed as means ± standard deviations; Statistical analysis of the data was performed using ANOVA with a Bonferroni post-hoc analysis for multiple analysis and Pearson correlation coefficients were calculated using GraphPad Instat Software (GraphPad Software, Inc., San Diego, CA, USA). The results were taken to be significant at a probability level of p < 0.05. 3. Results 3.1. Assessment of femoral head neovascularization Fig. 1. Surgical technique for Microfil infusion and visualization of vessel networks. A. Exposure of the abdominal aorta (arrow) for cannulation and injection of heparin. B. Direct visualization of microfil contrast agent within the intestine or other arteries.

background, noise was removed using a low pass Gaussian filter (sigma: 1.2, support: 2) and blood vessels were then defined at a threshold of 85. In order to reconstruct the three-dimensional (3-D) architecture of vasculature in the proximal femur, the blood vessels filled with Microfil were included with semiautomatically drawn contour at each two-dimensional (2-D) section by built-in “Contouring Program” for automatic reconstruction of 3-D image of vasculature in the decalcified sample. In addition axial slices through the samples were sequentially visualized and the number of vessels (main trunk only) penetrating the bone vessels was collated. All voxels counted

Blood vessel microarchitecture of each group was reconstructed in three dimensions for presentation. The samples of group I showed that these vasculatures were not visible in and around the necrotic lesion of femoral head, the samples of group II showed lightly increasing capillary vessels, the samples of group III showed intensive vascular architecture (Fig. 2A–C). Quantitatively, Fig. 3A shows the average number of vessels penetrating the proximal femur at 4 weeks after treatment. It was: 3.10 ± 0.33 in group I; 11.4 ± 2.46 in the group II; 44.5 ± 5.11 in group III. Penetrating capillary vessels number between group III and group II showed apparent difference. Capillary vessels number in group III was significantly larger than that of group II and group I (p < 0.0001). Capillary vessels number in group II was also significantly large compared with group I (p < 0.01). Fig. 3B showed the similar results.

688

Y. Sun et al. / Joint Bone Spine 76 (2009) 685–690

Fig. 2. Representative images of micro-CT reconstructed 3-D microangiography of proximal femur from group I (A), II (B) and III (C) rabbit. As compared with other groups, the density of vessel in femoral head of group III was obviously increased.

3.2. Formation of new bone surrounding the necrotic zone Subchondral bone of the femoral heads in the group I showed sparser trabecular bone with empty lacunae or pyknotic nuclei of osteocytes, and the medulla was full of fatty cells. In the rabbits of group II, empty lacunae and pyknotic nuclei of osteocytes within the bone trabeculae were also observed, but some of the marrow fat cells have decreased in size compared to that of group I and capillary formation, hemorrhage in the medulla and lots of osteoblasts were observed (Fig. 4A–B), but new bone formation was not prominent. The histologic observations suggested that BM-MNC had potential to regenerate bone and repair the ON. In the group III, the reparative response resulted in progressive development of a reactive margin, or interface, between the dead zone and

Fig. 4. Histological findings of the femoral head 4 weeks after treatment. A. Representative photomicrograph from the group I showed massive lamellar trabeculae had become empty (arrows), surrounded by more marrow fat cells with increased size dominantly occupying marrow space. B. In group II, Limited repair-appositional bone formation with lining cells (osteoblasts) (indicated with black arrow) around the necrotic bone and hemorrhage (indicated with white arrow) in the medullary cavity was observed. C and D. Photomicrographs showing development of new bone in the group III, 4 weeks after treatment of BM-MNC. During the repair process of osteonecrosis, primitive mesenchymal cells invade dead trabecular bone, differentiate into osteoblasts, and lay down new living bone (black arrows) to surround dead trabecular bone. Amounts of new vessels which contain intravascular perfusion substance (Microfil) also can be observed around the immature bone (white arrows). Table 1 Comparisons of new bone volume among three groups after 4 weeks. Group

New bone density (%)

I II III

4.19 ± 1.4 12.3 ± 5.6 29.7 ± 8.3*,**

* **

Compared to group I, p < 0.001. Compared to group II, p < 0.01.

adjacent viable tissues (Fig. 4C–D). During the repair process of ON, primitive mesenchymal cells and capillaries proliferated and invaded dead trabecular bone, differentiated into osteoblasts, and laid down new living bone surrounding dead trabecular bone, which was later remodeled. Histomorphometric analysis revealed that the new bone volume was significantly higher in the group III than in group I, II at 4 weeks (Table 1). 4. Discussion

Fig. 3. Micro-CT quantification of new vessel formation in the femoral head. Mean number of blood vessels (A) and vessel volume (B) of group III were significantly higher than that of other groups(*p < 0.05).

The best approach for precollapse ONFH remains unanswered. To date, vascularized fibular grafts have demonstrated the highest rates of success in treating early-stage ONFH, but it is a complex procedure with a higher rate of morbidity [6,11]. There has been a growing interest in the implantation of BMMNC for ONFH in recent years, and its clinical effectiveness has been demonstrated by several studies [7,8,12,13]. However, the mechanisms of the implanted BM-MNC for treating osteonecro-

Y. Sun et al. / Joint Bone Spine 76 (2009) 685–690

sis remain unclear. Gangji et al. once speculated that increased osteogenic capacity and angiogenesis contribute to this effectiveness. But due to the small number of patients and the ethical or technical problems, they did not confirm their speculation in their study. Further investigation is needed to pave the way for this therapy with respect to the osteogenic capacity and vascularization of the implanted BM-MNC in the avascular necrotic area. Our study showed that implantation of BM-MNC could increase vascularization and promote bone regeneration in steroid-induced ONFH in rabbit model. The current study applied a novel technique, using micro-CT imaging, to visualize and quantify new blood vessel formation and vascularization in rabbit femoral head. Traditionally, approaches to measure vessel formation and structure have relied on 2-D qualitative strategies. In recent years, contrast-enhanced micro-CT has been utilized to assess tissue microvasculature networks in many fields [14–19]. This technique is quantitative and effective for assessing vascularization. But the application in bone vascularization is scarce. Here, we quantified vessel formation within femoral head by applying this technique. This study demonstrated a significant increase in blood VV, penetrating vessel number in the group III compared to the group I, II. This result was similar to Hisatome et al. report [20]. As we known, in the adult, vascularization encompasses two separate mechanisms using cells from different sources [21]. Either differentiated vascular cells derived from pre-existing vessels are utilized, a process known as angiogenesis, or undifferentiated bone marrow-derived cells are recruited and incorporated into a growing vessel wall, a process termed vasculogenesis. Bone marrow contains sub-populations of various types of progenitor cell, and endothelial cells are thought to be mobilized from the bone marrow in adults. Asahara et al. established murine models of bone marrow implants that received bone marrow from transgenic mice expressing ␤-galactosidase under the transcriptional control of an endothelial cell-specific promoter. They demonstrated that lacZ-positive endothelial progenitor cells derived from the bone marrow were incorporated into sites of neovascularization where there was complete differentiation into endothelial cells [22]. Reyes et al. reported that human BM-MNC differentiated not only into mesenchymal cells but also into endothelial cells [23]. Hisatome et al. also found that the autologous BM-MNC expressed CD31, an endothelial lineage cell marker, and induced efficient neovascularization after implantation in bone marrow [20]. So, implanted BM-MNC directly participate in the vasculogenesis of femoral head. In addition, In the present study, by semi-quantitative immunohistochemistry analysis, we found vascular endothelial growth factor (VEGF) expression in group III was more prominent than in other groups (data not shown). Many researchers also confirmed that BM-MNC excrete many growth factors stimulates angiogenesis and the growth of smooth muscle cells and fibroblasts as well as endothelial cells [24–28]. Yan et al. once performed a similar study, they thought that mesenchymal stem cells (MSCs) transplantation enhanced bone formation, but did not increase vascularization [29]. We think the following reason may be attributed to this difference. First and foremost,

689

BM-MNC are different from MSCs. Apart from MSCs, BMMNC contains sub-populations of various types of progenitor cell including endothelial progenitor cells which are thought to play an important role in angiogenesis [30]. In their study, single MSCs did not take part in the vessel formation. Secondly, the animal model is different between the two experiments. We applied high dose steroid to induce nontraumatic ONFH in rabbit. Traumatic ONFH model was used in their study. The pathophysiology of the two models is completely different. Bone formation is an important physiological process in regulating bone remodeling. During repair of dead compact bone, resorption of necrotic bone precedes the formation of new bone. Assessment of bone regeneration of the necrotic femoral head was conducted by histological and histomorphometric analysis at Week 4 post-treatment. BM-MNC treatment enhanced osteogenic activity in the femoral heads of the ON rabbits. On histology slides, there was characterization of the repair process of ON, primitive mesenchymal cells had invaded necrotic trabecular bone, differentiated into osteoblasts, and laid down new living bone covering dead trabeculae, which created the characteristic appearance of “creeping substitution”. Histomorphometric analysis revealed that the new bone volume was significantly greater in the BM-MNC group than the other control group. BM-MNC ultimately led to an increase of bone volume, which was similar to the Gangji et al.’s clinic result [8]. In their study, the ratio of the volume of the necrotic lesion decreased by a mean of 35% in the bone marrow-graft group, whereas it increased by a mean of 23% in the control group. Recent research suggested that a decrease in the mesenchymal stem-cell pool of the proximal aspect of the femur might not provide enough osteoblasts to meet the needs of bone-remodeling in the early stage of the disease [31]. An insufficiency of osteogenic cells could explain the inadequate repair mechanism that, it is postulated, leads to femoral head collapse. Our results were consistent with this postulation. When BM-MNC were added, a lot of osteoblasts were observed and bone regeneration was prominent in group III. This phenomenon may be related to the availability of stem cells endowed with osteogenic properties, arising from an increase in the supply of such cells to the femoral head through BM-MNC implantation. On the other hand, accelerated vascularization of femoral head may also contribute to bone regeneration. In summary, we have confirmed the implantation of BMMNC could increase vascularization and bone regeneration in the early stages of ONFH in rabbit model. The implantation of BM-MNC is an effective and feasible method for treating early osteonecrosis. Conflicts of interests None of the authors has any conflicts of interest to declare. Acknowledgements This work was supported by the Key Scientific Foundation of Health Ministry PR China.

690

Y. Sun et al. / Joint Bone Spine 76 (2009) 685–690

References [1] Mont MA, Jones LC, David S, et al. Nontraumatic osteonecrosis of the femoral head: Ten years later. J Bone Joint Surg Am 2006;88:1117–32. [2] Chiu KH, Shen WY, Ko CK, et al. Osteonecrosis of the femoral head treated with cementless total hip arthroplasty: a comparison with other diagnoses. J Arthroplasty 1997;12:683–8. [3] Petrigliano FA, Lieberman JR. Osteonecrosis of the hip. Clin Orthop Relat Res 2007;465:53–62. [4] Mont MA, Jones LC, Hungerford DS. Nontraumatic osteonecrosis of the femoral head: ten years later. J Bone Joint Surg Am 2006;88A:1117–32. [5] Lieberman JR. Core decompression for osteonecrosis of the hip. Clin Orthop Relat Res 2004;418:29–33. [6] Scully SP, Aaron RK, Urbaniak JR. Survival analysis of hips treated with core decompression or vascularized fibular grafting because of avascular necrosis. J Bone Joint Surg Am 1998;80A:1270–5. [7] Hernigou P, Beaujean F. Treatment of osteonecrosis with autologous bone marrow grafting. Clin Orthop Relat Res 2002;405:14–23. [8] Gangji V, Hauzeur JP, Matos C, et al. Treatment of osteonecrosis of the femoral head with implantation of autologous bone marrow cells: A pilot study. J Bone Joint Surg Am 2004;86A:1153–60. [9] Qin Ling, Zhang G, Sheng H, et al. Multiple imaging modalities in evaluation of experimental osteonecrosis induced by a combination of lipopolysaccharide and methylprednisolone. Bone 2006;39: 863–71. [10] Wang GJ, Dughman SS, Reger SI, et al. The effect of core decompression on femoral head blood flow in steroid-induced avascular necrosis of the femoral head. J Bone Joint Surg Am 1985;67:121–4. [11] Urbaniak JR, Coogan PG, Gunneson EB, et al. Treatment of osteonecrosis of the femoral head with free vascularized fibular grafting, A. longterm follow-up study of 103 hips. J Bone Joint Surg Am 1995;77A: 681–94. [12] Gangji V, Toungouz M, Hauzeur JP. Stem cell therapy for osteonecrosis of the femoral head. Expert Opin Biol Ther 2005;5:437–42. [13] Hernigou P, Poignard A, Manicom O, et al. The use of percutaneous autologous bone marrow transplantation in nonunion and avascular necrosis of bone. J Bone Joint Surg Br 2005;87:896–902. [14] Toyota E, Fujimoto K, Ogasawara Y, et al. Dynamic changes in threedimensional architecture and vascular volume of transmural coronary microvasculature between diastolic- and systolic-arrested rat hearts. Circulation 2002;105:621–6. [15] Heinzer S, Krucker T, Stampanoni M, et al. Hierarchical microimaging for multiscale analysis of large vascular networks. Neuroimage 2006;32:626–36. [16] Malyar NM, Gossl M, Beighley PE, et al. Relationship between arterial diameter and perfused tissue volume in myocardial microcir-

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29] [30] [31]

culation: a micro-CT-based analysis. Am J Physiol Heart Circ Physiol 2004;286:H2386–92. Kindlmann GL, Weinstein DM, Jones GM, et al. Practical vessel imaging by computed tomography in live transgenic mouse models for human tumors. Mol Imaging 2005;4:417–24. Moore DC, Leblanc CW, Muller R, et al. Physiologic weight-bearing increases new vessel formation during distraction osteogenesis: a microtomographic imaging study. J Orthop Res 2003;21:489–96. Bolland BJ, Kanczler JM, Dunlop DG, et al. Development of in vivo microCT evaluation of neovascularisation in tissue engineered bone constructs. Bone 2008;43:195–202. Hisatome T, Yasunagaa Y, Yanadaa S, et al. Neovascularization and bone regeneration by implantation of autologous bone marrow mononuclear cells. Biomaterials 2005;26:4550–6. Ball SG, Shuttleworth CA, Kielty CM. Mesenchymal stem cells and neovascularization: role of platelet-derived growth factor receptors. J Cell Mol Med 2007;11:1012–30. Asahara T, Masuda H, Takahashi T, et al. Bone marrow origin of endothelial progenitor cells responsible for postnatal vasculogenesis in physiological and pathological neovascularization. Circ Res 1999;85:221–8. Reyes M, Lund T, Lenvik T, et al. Purification and ex vivo expansion of postnatal human marrow mesodermal progenitor cells. Blood 2001;98:2615–25. Kamihata H, Matsubara H, Nishiue T, et al. Implantation of bone marrow mononuclear cells into ischemic myocardium enhances collateral perfusion and regional function via side supply of angioblasts, angiogenic ligands, and cytokines. Circulation 2001;104:1046–52. Kinnaird T, Stabile E, Burnett MS, et al. Local delivery of marrow-derived stromal cells augments collateral perfusion through paracrine mechanisms. Circulation 2004;109:1543–9. Beckermann BM, Kallifatidis G, Groth A, et al. VEGF expression by mesenchymal stem cells contributes to angiogenesis in pancreatic carcinoma. Br J Cancer 2008;99:622–31. Al-Khaldi A, Eliopoulos N, Martineau D, et al. Postnatal bone marrow stromal cells elicit a potent VEGF-dependent neoangiogenic response in vivo. Gene Ther 2003;10:621–9. Shibata T, Naruse K, Kamiya H, et al. Transplantation of bone marrowderived mesenchymal stem cells improves diabetic polyneuropathy in rats. Diabetes 2008;57:3099–107. Yan Z, Hang D, Guo C, et al. Fate of mesenchymal stem cells transplanted to osteonecrosis of femoral head. J Orthop Res 2009;27:442–6. Zampetaki A, Kirton JP, Xu Q. Vascular repair by endothelial progenitor cells. Cardiovasc Res 2008;78:413–21. Hernigou P, Beaujean F, Lambotte JC. Decrease in the mesenchymal stem cell pool in the proximal femur in corticosteroid-induced osteonecrosis. J Bone Joint Surg Br 1999;81:349–55.