Evaluation of adipose-derived stromal vascular fraction or bone marrow-derived mesenchymal stem cells for treatment of osteoarthritis

Evaluation of Adipose-Derived Stromal Vascular Fraction orBone Marrow-Derived Mesenchymal Stem Cells for Treatment ofOsteoarthritis David D. Frisbie, John D. Kisiday, Chris E. Kawcak, Natasha M. Werpy, C. Wayne McIlwraith Equine Orthopaedic Research Center, Department of Clinical Sciences, Colorado State University, 300 West Drake Road, Fort Collins, Colorado80523 Received 19 December 2008; accepted 7 May 2009Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/jor.20933 ABSTRACT: The purpose of this study was the assessment of clinical, biochemical, and histologic effects of intraarticular administeredadipose-derived stromal vascular fraction or bone marrow-derived mesenchymal stem cells for treatment of osteoarthritis. Osteoarthritiswas induced arthroscopically in the middle carpal joint of all horses, the contralateral joint being sham-operated. All horses receivedtreatment on Day 14. Eight horses received placebo treatment and eight horses received adipose-derived stromal vascular fraction in theirosteoarthritis-affected joint. The final eight horses were treated the in osteoarthritis-affected joint with bone marrow-derived mesenchymalstem cells. Evaluations included clinical, radiographic, synovial fluid analysis, gross, histologic, histochemical, and biochemical evaluations.
No adverse treatment-related events were observed. The model induced a significant change in all but two parameters, no significanttreatment effects were demonstrated, with the exception of improvement in synovial fluid effusion PGE2 levels with bone marrow-derivedmesenchymal stem cells when compared to placebo. A greater improvement was seen with bone marrow-derived mesenchymal stem cellswhen compared to adipose-derived stromal vascular fraction and placebo treatment. Overall, the findings of this study were not significantenough to recommend the use of stem cells for the treatment of osteoarthritis represented in this model. ß 2009 Orthopaedic ResearchSociety. Published by Wiley Periodicals, Inc. J Orthop Res bone marrow-derived mesenchymal stem cells; stromal vascular fraction; osteoarthritis; in vivo model; equine Joint disease and specifically osteoarthritis (OA) is one study10 has been published limiting the overall trans- of the most prevalent and debilitating diseases clinically lational information. The goal of this study was to assess affecting both humans1 and horses.2,3 Furthermore, BMDMSC and ADSVF ability to decrease the progres- similarities in joint disease between the two species sion of OA without joint instability, as well as compare have allowed translational research to be conducted in the horse.4 Specifically, models of cartilage healing andOA have been developed in the horse allowing controlled studies to be performed on therapeutic interventions Experimental Design and Induction of Osteoarthritis that have clinical relevance to both human and equine Twenty-four skeletally mature 2–5-year-old horses, free of patients.4,5 To date,12 other studies assessing clinically musculoskeletal abnormalities [pain, range of motion, and relevant therapeutic interventions have been published joint effusion in the carpal joints (front knee)], were utilizedin the study. Horses were randomly assigned to one of three using a randomized blinded placebo controlled model treatments groups: ADSVF (N ¼ 8), BMDMSC (N ¼ 8), or of OA in the horse. Currently, no one therapeutic placebo (PCB) (N ¼ 8). All evaluators were unaware of treat- intervention for OA in any species has proven effective at long-term symptom-modifying or disease-modifying As previously described,5,11 on Day 0, following anesthesia effects.6 Recently, anecdotal reports have described full and routine preparation for surgery, each horse underwent return to athletic function in 70% of equine OA patients bilateral arthroscopic surgery of the middle carpal joints to treated using intraarticular (IA) administration of ensure that there were no preexisting abnormalities. During adipose-derived stromal vascular fraction (ADSVF) this procedure, an 8-mm osteochondral fragment was created (R. Harman et al., 2007, personal communication, in one randomly selected middle carpal joint. The fragment was http://www.vet-stem.com). Also, an uncontrolled multi- allowed to remain adhered to the joint capsule proximally. Amotorized arthroburr was used to debride the exposed sub- center equine clinical trial using IA treatment of bone chondral bone between the fragment and parent bone creating a marrow-derived mesenchymal stem cells (BMDMSC) 15-mm defect. The debris was not actively flushed from the for inoperable meniscal lesions has shown early prom- joint, thereby participating in the induction of osteoarthritis.
ise.7 These results, coupled with in vivo studies that This joint was designated as the OA-affected joint; the sham- have shown significant disease-modifying effects using operated contralateral joint was used as the control joint. The BMDMSC for the treatment of joint instability8 and arthroscopic portals were closed routinely. Horses assigned to collagen-induced arthritis,9 have fueled new enthusi- the BMDMSC group also had a 40–50 mL aspirate of bone asm for mesenchymal stem cells (MSC) as a novel marrow suspended in 3,000 units of sodium heparin which was treatment for OA. To date, one controlled clinical aseptically harvested from the sternum. Postoperative carefollowed routine clinical standards.
Correspondence to: David D. Frisbie (T: 970-297-4555; F: 970-297- Routine harvest of adipose tissue occurred for the ADSVF horses (Vet-StemTM, Poway, CA). Briefly, using systemic ß 2009 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.
sedation, routine surgical preparation, and local anesthetic, a 6-cm skin incision parallel to the spine at the level of the tail Tumor Necrosis Factor-a (TNF-a) concentrations in serum head was made and 10–20 g of adipose tissue was harvested.
and synovial fluid samples were determined using a commer- A routine closure of the incision was performed. The adipose cially available indirect ELISA kit (Endogen, Rockford, IL) that was shipped to the manufacturer, where the tissue was washed has been previously validated in equine samples.15,16 The assay with phosphate buffered saline (PBS), minced, then digested in was used according to manufacturer’s instructions, and low glucose Dulbecco’s modified Eagle’s medium (DMEM) absorbance was measured at A450. The upper and lower containing 10% fetal bovine serum (FBS) and 0.1% collagenase detection limits were 5,702.196 and 2.1 pg/mL, respectively.
for 3–4 h with agitation. Nucleated adipose cells were pelleted,washed, and resuspended in PBS for injection; finally, the cells were shipped to CSU. The total nucleated cell count normalized Following euthanasia on Day 70 with an overdose of pento- to the lowest quantity obtained, which was 16.3 million barbital, both middle carpal joints were specifically examined for degree and location of articular cartilage fibrillation orerosion, as well as synovial membrane hemorrhage.
BMDMSC Culture TechniqueMarrow aspirates were washed in PBS and then mixed with0.8% ammonium chloride. The cell pellet was rinsed with PBS, resuspended in low glucose DMEM containing 10% FBS, and Synovial membrane was harvested and placed in neutral- seeded in flasks at a concentration of 0.66 Â 106 nucleated cells/ buffered 10% formalin, embedded in paraffin, 5-mm sections cm2. Confluent BMDMSC colonies developed over 10–12 days, created and stained with hematoxylin and eosin (H&E).
at which point the cells were reseeded and expanded in growth Sections were assessed for cellular infiltration, synovial medium containing 1 ng/mL FGF-2. BMDMSC cultures were intimal hyperplasia, subintimal edema, subintimal fibrosis, passaged at a split ratio of 1:3 twice prior to treatment.12 Articular cartilage pieces (5 mm2) were obtained from each joint; samples were fixed neutral-buffered 10% formalin Horses were housed in stalls (3.65 Â 3.65 m each). Beginning embedded in paraffin, 5-mm sections created and stained with on Day 15, horses were exercised on a high-speed treadmill both H&E or Safranin O, fast green (SOFG). H&E sections 5 days each week throughout the study. Horses were trotted were evaluated for articular cartilage fibrillation, chondrocyte (16–19 km/h) for 2 min, galloped (approximately 32 km/h) necrosis, chondrone formation, and focal cell loss.11 SOFG for 2 min, and trotted again (16–19 km/h) for 2 min daily to sections were evaluated for intensity of staining in each simulate the strenuous exercise of race training.
All horses were treated on Day 14 postsurgery. PCB horses Articular cartilage proteoglycan content was estimated by use received 2 mL 0.9% NaCl in their OA-affected joint. The of a 1,9-dimethyl-methylene blue technique14 on samples ADSVF horses received 16.3 million total nucleated cells obtained from each joint that were stored at À808C. For suspended in 2 mL of 0.9% buffered NaCl in their OA-affected analysis of cartilage matrix metabolism, articular cartilage joint. OA affected joints of BMDMSC horses were treated samples were aseptically collected, and radiolabeled SO4 with a mean of 10.5 million (SEM ¼ 1.1 million cells, range of (35SO4) incorporation was measured by use of previously 5.6–15 million cells) culture expanded BMDMSC suspended in Data were evaluated using an ANOVA framework with PROC Clinical examinations of both forelimbs were performed GLIMMIX of SAS17 with the horse as a random variable. The every 2 weeks throughout the study period. Pain was graded ANOVA tables were used to determine significant main effect on a standardized 0–5 scale.13 All other clinical, histologic, and interactions between main effect variables. When indi- and histochemical outcomes were graded on a 0–4 scale vidual comparisons were made, a least square means was (0 represented normal, 4 represented severe change). Joint utilized and a p-value less than or equal to 0.05 was considered effusion was measured as an indication of inflammation, and significant. Data was tested for normality using residual plots joint range of motion was measured through carpal flexion.
and, when required, natural log transformations performed to Bilateral radiographic carpal evaluation occurred on ensure normality. When transformation was performed, data Day À7, Day 14, and Day 70. Images were evaluated for boney proliferation at the joint capsule attachment, subchondral bonelysis, and osteophytes.
Synovial fluid was collected weekly from both middle carpal joints. Samples were assessed for total protein concentration using a refractometer, cytologic evaluation, total white blood ( p < 0.0001) in the OA-affected [2.33 Æ 0.06 (mean - cell (WBC) count, or stored at À808C for biochemical protein Æ SE)] limb when compared to the sham-operated limb A modified 1,9-dimethyl-methylene blue dye-binding assay (0.00 Æ 0.06) on Day 14 (prior to treatment). Change in was used to determine glycosaminoglycan (GAG) concentra- pain values were calculated using Day 14 (the last tion.14 Synovial fluid concentration of prostaglandin E2 (PGE2) pretreatment evaluation) as the post-osteoarthritis but was also assessed (PGE2 Kit, Assay Designs, Ann Arbor, MI).
pretreatment baseline (a positive change score indicates MESENCHYMAL STEM CELLS FOR OSTEOARTHRITIS by a significantly ( p < 0.0001) higher cumulative radio-graphic score in the OA-affected (1.19 Æ 0.12) comparedto the sham-operated (0.06 Æ 0.12) joints. No significanttreatment effects were detectable.
Synovial FluidRoutine synovial fluid analysis indicated, as expected,the total protein concentration increased significantly( p < 0.0001) with induction of OA throughout the studyperiod when sham-operated (2.08 Æ 0.09) were com-pared to OA-affected joints (2.70 Æ 0.09). Synovial fluid Raw lameness scores (mean Æ SEM) plotted by time for WBC counts were significantly increased ( p < 0.0001) each treatment group as well as for the osteoarthritis-affected and by the induction of OA, with OA-affected joints having a Sham limbs. No significant differences were noted in any compar-ison except for an increase in osteoarthritis joints postsurgery.
higher WBC count compared to sham-operated joints(245 Æ 32 vs. 169 Æ 32 cell/dL, respectively). Therewere no significant treatment effects seen in synovial improvement). There was no significant improvement in total protein or WBC counts. Based on the cytology pain score with respect to treatment (Fig. 1).
of the synovial fluid WBCs, there were significantly( p ¼ 0.0070) less lymphocytes in the ADSVF OA- affected joints compared to all other joints (Table 1).
All horses showed a significant increase ( p < 0.0001) in Synovial GAG concentrations were significantly flexion score (representing a decreased range of motion) ( p < 0.0001) increased with induction of OA; OA-affected in the OA-affected (2.50 Æ 0.08) limb when compared joints (4.45 Æ 0.02 Ln mg GAG/mL) had an increase GAG to the sham-operated limb (0.08 Æ 0.08) for Day 14.
concentration when compared to sham-operated joints Change in flexion was calculated using Day 14 as the pretreatment (4.33 Æ 0.02 Ln mg GAG/mL). No signifi- post-osteoarthritis but pretreatment baseline, and sig- cant treatment effects were seen in synovial fluid GAG nificant improvements ( p ¼ 0.0013) based on treatment group and joint were seen. Improvement in the Synovial fluid PGE2 concentrations were signifi- OA-affected limbs that received PCB and BMDMSC cantly ( p < 0.0001) increased with induction of OA treatments were seen when compared to ADSVF (4.24 Æ 0.09 Ln pg/mL), compared to the sham-operated joints (3.23 Æ 0.09 Ln pg/mL). A significant ( p ¼ 0.0423)decrease in synovial PGE2 concentration was seen inBMDMSC compared to PCB treatment horses starting on Day 35 (Fig. 3a). This affect was independent of joint, All horses showed a significant increase ( p < 0.0001) in although synovial fluid from OA-affected joints demon- effusion score in the OA-affected (2.25 Æ 0.08) joints strated a more pronounced difference (Fig. 3b). Synovial compared to the sham-operated joints (0.13 Æ 0.08) fluid TNF concentrations were significantly ( p ¼ 0.0005) for Day 14. Change in joint effusion was calculated increased with induction of OA (2.18 Æ 0.3 Ln pg/mL), using Day 14 as the post-osteoarthritis but pretreat- compared to sham-operated joints (1.73 Æ 0.29 Ln pg/ ment baseline; no significant differences were observed mL). The only significant ( p ¼ 0.0194) treatment differ- ence was higher TNF concentrations in OA-affected(2.69 Æ 0.52 Ln pg/mL) compared to sham-operated (1.72 Æ 0.51 Ln pg/mL) joints of ADSVF-treated horses.
A significant increase in radiographic joint pathologywas induced following induction of OA as demonstrated Cytology of the Synovial Fluid White Blood Cells, Specifically the Percentage of White Blood Cells thatWere Lymphocytes* OA, osteoarthritis; BMDMSC, bone marrow-derived mesenchymalstem cells; ADSVF, adipose-derived stromal vascular fraction.
*Significantly (p ¼ 0.0070) less lymphocytes were noted in the The grade improvement in flexion score (mean Æ SEM) adipose-derived stromal vascular fraction osteoarthritis-affected plotted by treatment group. Different letters indicate a significant joint compared to all other joints (different letters indicate a concentration plotted by Day for eachtreatment group (average of boththe osteoarthritis-affected and Shamjoints). An asterisk represents asignificant difference between thecomparison. (b) Natural log of PGE2concentration plotted by Day for eachtreatment group (for both the osteo-arthritis-affected and Sham joints).
At necropsy, hemorrhage within the synovial mem- Evaluation of articular cartilage for SOFG staining did brane was significantly ( p ¼ 0.0002) increased in OA- not demonstrated a significant difference when OA- affected (1.79 Æ 0.13) compared with sham-operated affected (8.35 Æ 0.42) were compared to sham-operated (1.13 Æ 0.13) joints. Similarly, articular cartilage total (8.28 Æ 0.42) joints or treatment comparisons.
erosion scores were significantly ( p < 0.0001) increasedin OA-affected joints (2.42 Æ 0.14) compared with sham- operated joints (1.38 Æ 0.14). No significant treatment No significant difference was noted with the cartilage effects were seen for any of the gross pathologic GAG content or GAG synthesis with respect to induc- Histologic ExaminationsSynovial Membrane H&E There was a significant ( p ¼ 0.0061) increase in the The current study used a model of OA that effectively cumulative pathology score for the synovial membrane induced significant clinical, gross, histologic, and bio- in OA-affected (5.17 Æ 0.45) when compared to sham- chemical changes indicative of OA. The authors believe operated (3.25 Æ 0.45) joints. No significant treatment this is the first controlled study to assess clinical musculoskeletal pain following the treatment of OAwith MSCs. While all of the clinical parameters were significantly increased following the induction of OA, Cartilage stained with H&E showed a significant improvements were not demonstrated with ADSVF or ( p < 0.0131) increase in the modified Mankin score BMDMSC treatment. This is in contrast to anecdotal (cumulative score of all four outcome parameters) when reports presented by Vet StemTM following the clinical OA-affected (3.43 Æ 0.53) were compared with sham- use of ADSVF (R Harman et al., 2007, personal operated (1.63 Æ 0.53) joints based on location. No communication, http://www.vet-stem.com). The authors significant treatment effects were observed based on assume the uncontrolled nature of case selection, the total modified Mankin score or individual outcome variability in clinical disease progression, as well as lack of treatment uniformity in the Vet-StemTM cases MESENCHYMAL STEM CELLS FOR OSTEOARTHRITIS are most likely responsible for the disparity. In fact, previous failure of medical treatment, as well as a poor more improvement in range of motion (measured by prognosis following diagnostic arthroscopy.7 joint flexion) was gained with placebo treatment than A secondary goal of the current study was to compare with either ADSVF or BMDMSC, with the least treatment effects of ADSVF and BMDMSC. Because only significant response being seen in the ADSVF-treated significant improvement in synovial PGE2 concentra- horses. It is difficult to explain these findings, given that tions could be demonstrated with BMDMSC, this other outcome parameters that typically accompany comparison is somewhat limited; finding a greater change in joint flexion (significant change in synovial magnitude of an effect with bone marrow- versus membrane pathology and radiographic pathology at adipose-derived cells has also been seen with multiple joint capsule margin) were not seen in this study.
other studies involving musculoskeletal tissues and is, Induction of OA significantly increased synovial fluid PGE2 concentrations. This finding would be In summary, no adverse effects were noted with expected with joint disease as a marker of inflamma- tion5,18 and can be driven by a host of proinflammatory ADSVF was associated with increased synovial fluid cytokines. While a decrease in synovial fluid PGE2 TNF-a concentration, which is worrisome. The only was seen with ADSVF treatment, this difference was significant beneficial effect was noted in synovial PGE2 not statistically significant. Conversely, treatment with level reduction following BMDMSC treatment. Overall, BMDMSC not only significantly decreased synovial fluid this modest improvement based on the number of other PGE2 levels in the OA-affected limb, but also showed a outcome parameters dampens the enthusiasm of the systemic effect through significantly decreasing the authors’ use of unmodified MSC in the treatment of OA.
PGE2 concentration in the sham-operated limb as well.
Further modification using gene therapy or selection of The reduction of PGE2 represents a decrease in overall subpopulations of MSC needs to be explored for the joint inflammation and has historically been positively correlated with a decrease in pain,18–21 although nosignificant reduction in pain was observed as a result of BMDMSC or ADSVF treatment. Previously in this Vet-StemTM Poway, CA provided partial funding for this model, when a significant reduction of PGE2 has been study. None of the authors’ professional or financial affiliations noted, a decrease in clinical pain has also been observed have biased this presentation. The authors thank all of thestaff and volunteers at the Equine Orthopaedic Research as was the case following IA administration of cortico- Center at Colorado State University for their help and steroids.22,23 Treatment using interleukin-1 receptor antagonist11,24 demonstrated both symptom- and dis-ease-modifying effects without reduction of synovial PGE2 levels. Thus, PGE2 reduction is not a prerequisite 1. Lawrence R, Helmick C, Arnett F, et al. 1998. Estimates of the for symptom-modifying effects. While a decrease of TNF- prevalence of arthritis and selected musculoskeletal disorders alpha has been shown to occur following peritoneal in the United States. Arthritis Rheum 41:778–779.
2. Rossdale PD, Hopes R, Digby NJW, et al. 1985. Epidemio- treatment with allogenic BMDMSC in mice with colla- logical study of wastage among racehorses 1982 and 1983. Vet gen-induced arthritis,9 leading to a decrease in inflam- mation measured through paw size and discoloration, 3. USDA. 2000. Lameness and laminitis in U.S. horses. Fort TNF-a was not decreased in the current study. In fact, Collins, CO: USDA:APHIA:VS, CEAH, National Animal treatment with ADSVF increased synovial fluid TNF-a.
Murphy et al. observed a decrease in progression of 4. McIlwraith CW, Frisbie DD, Kawcak CE, et al. 2009.
OA following treatment with autologous BMDMSC Recommendation of criteria for the evaluation of macroscopicand histological changes occurring in equine osteoarthritis.
and hyaluronic acid using a medial meniscectomy and Osteoarthritis Cartilage (in press).
anterior cruciate deficient model in the goat.8 This 5. Frisbie DD, Al-Sobayil F, Billinghurst RC, et al. 2008.
study also demonstrated a neomeniscal tissue formation Changes in synovial fluid and serum biomarkers with exercise associated with animals showing a decreased OA. Barry and early osteoarthritis in horses. Osteoarthritis Cartilage postulated that the increase in stability afforded from the neomeniscal tissue was most likely responsible for the 6. Frisbie DD. 2005. Future directions in treatment of joint disease in horses. Vet Clin North Am Equine Pract 21:713– decreased progression of OA,25 a fact that is supported by the lack of decreased OA progression in the current OA 7. Frisbie DD, Hague BA, Kisiday JD. 2007. Stem cells as a model which does not have an instability component. It treatment for osteoarthritis. Presented at American College should be noted that the current study did not use the of Veterinary Surgeons Veterinary Symposium, Chicago, IL addition of hyaluronic acid with the MSC administration [available in digital format from the corresponding author].
and, thus, may have decreased the resident time of the 8. Murphy JM, Fink DJ, Hunziker EB, et al. 2003. Stem cell cells in the joint space. It is of interest to the authors that therapy in a caprine model of osteoarthritis. Arthritis Rheum48:3464–3474.
a short-term follow-up (6–18 months) on equine patients 9. Augello A, Tasso R, Negrini SM, et al. 2007. Cell therapy using suffering from meniscal disease, and treated with allogeneic bone marrow mesenchymal stem cells prevents BMDMSC plus hyaluronic acid, did show a better than tissue damage in collagen-induced arthritis. Arthritis Rheum expected return to full athletic work (67%) following 10. Black LL, Gaynor J, Gahring D, et al. 2007. Effect of adipose- 20. May SA, Hooke RE, Lees P. 1991. Adverse conditions in vitro derived mesenchymal stem and regenerative cells on lame- stimulate chondrocytes to produce prostaglandin E2 and ness in dogs with chronic osteoarthritis of the coxofemoral stomelysin. Equine Vet J 23:380–382.
joints: a randomized, double-blinded, multicenter, controlled 21. May SA, Hooke RE, Lees P. 1992. Inhibition of interleukin-1 activity by equine synovial fluid. Equine Vet J 24:99–102.
11. Frisbie DD, Ghivizzani SC, Robbins PD, et al. 2002. Treat- 22. Frisbie DD, Kawcak CE, Trotter GW, et al. 1998. The effects ment of experimental equine osteoarthritis by in vivo delivery of 6-alpha methylprednisolone acetate on an in vivo equine of the equine interleukin-1 receptor antagonist gene. Gene osteochondral fragment exercise model. Am J Vet Res 12: 12. Kisiday JD, Kopesky PW, Evans CH, et al. 2008. Evaluation of 23. Frisbie DD, Kawcak CE, Trotter GW, et al. 1997. The effects of adult equine bone marrow- and adipose-derived progenitor triamcinolone acetate on an in vivo equine osteochondral cell chondrogenesis in hydrogel cultures. J Orthop Res 26: fragment exercise model. Equine Vet J 29:349–359.
24. Frisbie DD, Kawcak CE, Werpy NM, et al. 2007. Clinical, 13. Anonymous. 1991. Definition and classification of lameness.
biochemical and histologic effects of intra-articular admin- In: Guide for veterinary service and judging of equestrian istration of autologous conditioned serum in horses with experimentally induced osteoarthritis. Am J Vet Res 68:290–296.
14. Farndale RW, Buttle DJ, Barrett AJ. 1986. Improved 25. Barry FP. 2003. Mesenchymal stem cell therapy in joint quantitation and discrimination of sulphated glycosaminogly- disease. Novartis Found Symp 249:86–96; discussion –102, cans by use of dimethylmethylene blue. Biochem Biophys Acta 26. Im GI, Shin YW, Lee KB. 2005. Do adipose tissue-derived 15. Vick MM, Adams AA, Murphy BA, et al. 2007. Relationships mesenchymal stem cells have the same osteogenic and among inflammatory cytokines, obesity, and insulin sensitiv- chondrogenic potential as bone marrow-derived cells? Osteo- ity in the horse J Anim Sci. 85:1144–1155.
16. McFarlane D, Holbrook TC. 2008. Cytokine dysregulation 27. Hennig T, Lorenz H, Thiel A, et al. 2007. Reduced chondro- in aged horses and horses with pituitary pars intermedia genic potential of adipose tissue derived stromal cells dysfunction. J Vet Intern Med 22:436–442.
correlates with an altered TGFbeta receptor and BMP profile 17. Anonymous. 2006. The GLIMMIX Procedure. Cary, NC: SAS and is overcome by BMP-6. J Cell Physiol 211:682–691.
28. Liu TM, Martina M, Hutmacher DW, et al. 2007. Identifica- 18. May SA, Hooke RE. 1987. The identity of the E-series tion of common pathways mediating differentiation of bone prostaglandins produced by chondrocytes and synovial cells marrow- and adipose tissue-derived human mesenchymal in response to a variety of stimuli. Proc Br Equine Vet Assoc stem cells into three mesenchymal lineages. Stem Cells 25: 19. May SA, Hooke RE, Lees P. 1989. Identity of the E-series 29. Noel D, Caton D, Roche S, et al. 2008. Cell specific differences prostaglandin produced by equine chondrocytes and synovial between human adipose-derived and mesenchymal-stromal cells in response to a variety of stimuli. Res Vet Sci 46: cells despite similar differentiation potentials. Exp Cell Res

Source: http://www.2f-stemcells.de/downloads/frisbie-JO-%202009%20-stem-cell-OA.pdf