Abstract
Objective To systematically review the evidence for the efficacy of mesenchymal stem cell (MSC) injections in improving osteoarthritis (OA)-related structural outcomes.
Methods Ovid Medline and EMBASE were searched from their inceptions to April 2020 using MeSH terms and key words. Independent reviewers extracted data and assessed methodological quality. Qualitative evidence synthesis was performed due to the heterogeneity of interventions and outcome measures.
Results Thirteen randomized controlled trials (phase I or II) were identified: 10 in OA populations and 3 in populations at risk of OA, with low (n = 9), moderate (n = 3), or high (n = 1) risk of bias. Seven studies used allogeneic MSCs (4 bone marrow, 1 umbilical cord, 1 placenta, 1 adipose tissue), 6 studies used autologous MSCs (3 adipose tissue, 2 bone marrow, 1 peripheral blood). Among the 11 studies examining cartilage outcomes, 10 found a benefit of MSCs on cartilage volume, morphology, quality, regeneration, and repair, assessed by magnetic resonance imaging, arthroscopy, or histology. The evidence for subchondral bone was consistent in all 3 studies in populations at risk of OA, showing beneficial effects. Sixteen unpublished, eligible trials were identified by searching trial registries, including 8 with actual or estimated completion dates before 2016.
Conclusion Our systematic review of early-phase clinical trials demonstrated consistent evidence of a beneficial effect of intraarticular MSC injections on articular cartilage and subchondral bone. Due to the heterogeneity of MSCs, modest sample sizes, methodological limitations, and potential for publication bias, further work is needed before this therapy is recommended in the management of OA.
Osteoarthritis (OA) causes disability, impaired quality of life, and significant financial burden1,2. Current treatment modalities, including analgesics, nonsteroidal antiinflammatory drugs, opiates, intraarticular injections of steroids and hyaluronans, and physical therapies3,4,5, only alleviate symptoms with short-term, small-to-moderate effects6. No drugs have shown enough of an effect on slowing structural progression of OA to be approved as disease-modifying OA drugs7.
Adult mesenchymal stem cells (MSCs) are multipotent, undifferentiated cells that can be isolated from bone marrow, adipose tissue, muscle, or synovium, and readily culture expanded without undergoing differentiation8. MSCs have been investigated as a promising treatment for OA due to their ability to differentiate into cartilage, bone, adipose, tendon, and other cells of the mesenchymal lineage, and their antiinflammatory and immunomodulatory activities8,9,10,11. While the use of MSCs has gained momentum in recent decades, their potential as a treatment for OA remains unclear, as studies have shown that few stem cells survive after injection12,13,14, and there is a lack of data on the long-term safety and efficacy from larger clinical trials15,16,17.
Several systematic reviews that focus on patient-reported outcomes have shown the safety and effectiveness of intraarticular MSC injections in improving pain and function in OA17,18,19,20,21,22,23,24. While previous studies on stem cell therapy are based on moderate numbers of participants, the effect of MSCs on patient-reported outcomes is critical information for clinical decision-making and future research. A number of clinical trials have examined the effect of MSCs on OA-related structural outcomes25–37. A recent systematic review that included 6 clinical trials of knee OA demonstrated beneficial effects of MSCs on improving radiological, histological, and arthroscopic outcomes, but all studies had high risk of bias and large clinical heterogeneity17. Normal joints, established OA, and end-stage OA are on the same continuum, and preclinical diseases, such as focal chondral defect, partial meniscectomy, and anterior cruciate ligament injury, identify those at risk of OA in whom therapies such as MSC may be beneficial. Therefore, we systematically reviewed the evidence for the efficacy of stem cell injections in improving structural outcomes of the knee, hip, and spine in individuals with OA or at risk of OA, specifically focusing on OA-related structural outcomes assessed objectively in studies with a control group.
MATERIALS AND METHODS
Our systematic review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Review and Meta-Analyses) guidelines38.
Search strategy. Ovid Medline and EMBASE databases were searched from their inceptions to April 2020 using MeSH terms and keywords to identify studies examining the effect of stem cell injections on joint structures (Table 1). Searches were limited to human and English-language studies. The references of identified manuscripts were searched for additional studies.
Study selection. Two authors (initial search: RV and LC; updated search: JF and YW) independently reviewed records to assess the eligibility of studies by title, abstract, and full text, using a 3-stage determination method according to the inclusion and exclusion criteria (Table 1). Any disagreement between the 2 authors was resolved by discussion.
Data extraction and synthesis. Two authors (JG and JF) extracted data on target population; sex, age, and number of study participants; type, source, and immunophenotypic characterization of stem cells; route of administration; number of injections; outcome measures; duration of follow-up; source of funding; and effect of stem cell injections on structural outcomes. Qualitative synthesis was performed due to the heterogeneity in interventions and outcome measures.
Risk of bias assessment. Two authors (SMH and YW) independently assessed the risk of bias using the Cochrane Collaboration tool for assessing risk of bias in randomized trials39. This tool covers 6 domains of bias: selection bias, performance bias, detection bias, attrition bias, reporting bias, and other bias. Studies were assessed as “high,” “low,” or “unclear” risk of bias for each item, with an overall risk of bias being scored as “low,” “moderate,” or “high”17 (Supplementary Table 1, available from the authors on request). The agreement between the 2 authors was 86%. Differing assessments were discussed to get a consensus.
Search of trial registers and registries for unpublished studies. One author (YW) searched trial registers and registries for clinical trials with “completed” or “unknown” status that were eligible for our current systematic review but not published: US National Institutes of Health Trial Register (www.ClinicalTrials.gov), World Health Organization International Clinical Trials Registry Platform (apps.who.int), European Union Clinical Trial Register (www.clinicaltrialsregister.eu), Australian New Zealand Clinical Trials Registry (www.anzctr.org.au), and International Standard Randomized Controlled Trial Number registry (www.isrctn.com).
RESULTS
Study selection. Figure 1 shows the breakdown of the study selection. After removal of duplicates, 1250 articles were screened. Full text was reviewed for 32 studies, with 14 eligible studies identified (13 on knee, 1 on spine). No additional articles were found after searching the references of published research or review articles. A study on degenerative disc disease40 was further excluded since a single study precludes a comparison with other studies and lacks the robustness to draw any reliable conclusion.
Description of included studies. Table 2 provides an overview of the 13 studies published between 2013 and 2019; all were phase I or II randomized controlled trials25–37. Three studies originated from Australia31,32,35, 2 from Spain28,30, and a single study each from Malaysia25, Singapore26, USA27, India29, Chile33, Iran34, South Korea36, and China37. The mean ages of participants ranged from 26 to 66 years and percentage of men ranged 10–71%. Ten studies included patients with knee OA, defined using the Kellgren-Lawrence grading scale28,29,30,32–37 or criteria not clearly specified26. Other studies examined patients with International Cartilage Repair Society (ICRS) grade 3–4 cartilage lesions25, partial meniscectomy27, or unilateral anterior cruciate ligament injury31. The follow-up was at 6 months34,36, 12 months26,28,29,30,32,33,35,37, 18 months25, or 24 months27,31. Six studies were funded by companies27,29,31,32,35,36, 4 studies by governments25,28,30,34, 1 study by a company and government37, and 2 studies did not report the funders26,33.
Interventions. Stem cells were sourced through allogeneic or autologous method. Seven studies used allogeneic MSCs, derived from bone marrow27,28,29,31, umbilical cord33, placenta34, or adipose tissue32. Six studies used autologous MSCs, derived from adipose tissue35,36,37, bone marrow26,30, or peripheral blood25. Twelve trials performed immunophenotypic characterization of MSCs25–31,33,34,35,36,37, reporting positive CD73, CD90, or CD10526–30,33,34,35,37, and negative CD14, CD19, CD34, CD35, or HLA-D related (HLA-DR)26–30,33,34,35,37. One study reported positive CD105 and CD3425. Two studies did not report the details31,36. All stem cell treatment was administrated through intraarticular injection of varying doses. Eleven studies involved a single injection26–36, with 2 studies also involving 2 injections at baseline and 6 months33,35. One study applied 8 injections25. One study involved 2 injections at 0 and 3 weeks37. Seven studies used a single dose group25,26,28,31,34,36,37, 5 studies had 2 dose groups27,30,32,33,35, and 1 study had 4 dose groups29. MSCs were suspended in different media, including hyaluronic acid (HA) only25,26,31,37; Plasma-Lyte A only29; normal saline only34,35,36; HA, human serum albumin, and Plasma-Lyte A27; Ringer lactate containing human albumin28,30; or saline with AB plasma33. One study did not report the suspension medium32. The control group received intraarticular injection of HA25,26,27,28,30,31,33,37, normal saline34,36, Plasma-Lyte A29, or cell culture media and cryopreservative32. One study used standard care as the control35 (Table 3).
Assessment of structural outcomes. Structural outcomes were the primary outcome in 4 studies25,27,30,34 and the secondary outcome in 9 studies26,28,29,31,32,33,35–37 (Supplementary Table 2, available from the authors on request). Knee structure was assessed in 8 studies by magnetic resonance imaging (MRI) only26,27,28,32,33,34,35,37, 4 studies by both MRI and radiograph29,30,31,36, and 1 study by MRI and second-look arthroscopy with chondral core biopsy25. Articular cartilage outcomes were cartilage volume/thickness31,32,34,37, cartilage defects32,35,36, cartilage quality28,33, cartilage repair25,26, meniscal volume27, and meniscal pathology35, assessed using MRI, and cartilage repair, assessed using validated arthroscopy grading systems25. Subchondral bone outcomes were tibial bone area31,32, bone marrow lesions25,32,35, and subchondral bone sclerosis and osteophyte formation27,34,35, assessed by MRI. Composite MRI scores of multiple features were assessed using the Whole-Organ Magnetic Resonance Imaging Score (WORMS)27,29,30,33, MRI Osteoarthritis Knee Score35, or a scoring system developed for morphological evaluation25. Radiograph outcome was either joint space width30,31,36 or not specified29.
Risk of bias assessment. The overall risk of bias was low in 9 trials25,27,29,30,32–35,37, moderate in 3 trials26,31,36, and high in 1 trial28 (Table 4). The study population and research question were clearly defined and participants and personnel were blinded in all the studies. Some studies did not have adequate allocation concealment26,28,33,36 or complete outcome data25,27,28,31. Some studies had unclear risk of bias for random sequence generation28,31,36, blinding of outcome assessment29, or selective reporting as they were not registered in trial registries26.
Effect of MSCs on articular cartilage outcomes. Eight studies examined cartilage volume, quality, regeneration, and repair in OA populations26,28,32,33,34,35,36,37 (Table 3). Wong, et al showed significantly better Magnetic Resonance Observation of Cartilage Repair Tissue score and more prevalent cartilage coverage (complete and > 50%), as well as complete integration of regenerated cartilage in the intervention group compared with the control group after 1 year26. Vega, et al found a significant decrease in poor cartilage index in the intervention group but not the control group, with improvement against baseline score not significantly different between the 2 groups at 12 months28. Kuah, et al’s study showed no significant decrease in lateral tibial cartilage volume in the Progenza 3.9M group but a significant cartilage loss in the control group after 12 months32. Khalifeh Soltani, et al showed increased cartilage thickness in the intervention group but no significant change in the control group over a 24-week period; no significant change in meniscus lesions was seen in either group34. Freitag, et al found significantly reduced progression of cartilage loss in those treated with 2 MSC injections (11%), compared with those treated with 1 MSC injection (30%) or controls (67%) at 12 months35. Lee, et al demonstrated a significant increase in cartilage defect size in the control group but not in the MSC group at 6 months36. Lu, et al found a significant increase in knee cartilage volume at 12 months in the MSC group, whereas the control group had a significant reduction in cartilage volume37. In contrast, Matas, et al found no significant difference in articular cartilage or meniscal integrity scores between the intervention and control groups over 6 or 12 months33.
Three studies examined articular cartilage in populations at risk of OA25,27,31 (Table 3). In Saw, et al’s study, a second look arthroscopy with chondral biopsy and histologic evaluation at 18 months after the initial surgery showed a significantly higher ICRS II score in the intervention group compared with the control group25. The intervention group scored 14% higher on flush morphologic features, 23% higher on good repaired cartilage fill, and 20% higher on no-gap integration than the control group at 18 months25. In the Vangsness, et al study, while no patients in the control group met the 15% threshold for increased meniscal volume, significant increase in meniscal volume was observed in 24% of patients treated with 50 million MSCs and 6% of patients treated with 150 million MSCs at 12 months27. At the 2-year follow-up, 18% of patients treated with 50 million MSCs had significant increase in meniscal volume that was not observed in the 150 million MSC group or control group, with no significant differences between either MSC or control groups27. Wang, et al found no significant difference in tibial cartilage volume loss at 6, 12, and 24 months between the intervention group treated with mesenchymal precursor cells (MPC) and the control group31. There was a trend in which the MPC group had a reduced rate of medial tibial cartilage volume loss over the first 6 months31.
Effect of MSCs on subchondral bone outcomes. Three studies examined subchondral bone in OA populations32,34,35 (Table 3). Freitag, et al found a nonsignificant trend of less extension of osteophyte formation over 12 months in patients receiving 2 MSC injections (11%), compared with those receiving 1 MSC injection (50%) or the control group (56%), with no significant difference in bone marrow lesions between groups35. Kuah, et al found no significant difference in the change in tibial bone area or bone marrow lesions among Progenza 3.9M, Progenza 6.7M, or control groups over 12 months32. Khalifeh Soltani, et al found no significant change in spur or erosion in either group over 24 weeks34.
Three studies examined subchondral bone in populations at risk of OA25,27,31 (Table 3). Wang, et al found a significantly reduced rate of tibial bone expansion in the MPC group compared with the control group over 6 months, with the trend maintained over 12 and 24 months31. Saw, et al found that moderate to severe edema was 2% in the intervention group vs 10% in the control group at 18 months25. In Vangsness, et al’s study, subchondral bone sclerosis and osteophyte formation were found in 6% of the MSC group and 21% of the control group at 1-year follow-up27.
Effect of MSCs on composite MRI scores of the knee. Four studies examined composite MRI scores in populations with OA29,30,33 and at risk of OA25 (Table 3). Saw, et al found morphological MRI grading was significantly higher in the intervention group than the control group at 18 months25. Lamo-Espinosa, et al found a median improvement of 4 points in WORMS score in the 100M MSC group at 12 months, with 25% of patients having an improvement of 22 points, and no improvement in either the 10M MSC or control group30. Studies by Gupta, et al and Matas, et al showed no significant differences in WORMS score between intervention and control group at 6 or 12 months29,33.
Effect of MSCs on radiograph outcomes. Three studies assessed joint space width in populations with OA30,36 and at risk of OA31 (Table 3). Wang, et al showed a greater increase in joint space width at 12, 18, and 24 months in the MPC + HA group than in the HA alone group31. Lamo-Espinosa, et al found no significant change in joint space width in the MSC groups at 12 months, but a borderline reduction in the control group30. Lee, et al showed no significant change in joint space width in either group over 6 months36. Gupta, et al’s study found no clinically meaningful changes in radiograph parameters (details not reported) at 3 and 6 months in either group29.
Unpublished studies. Searches of trial registers and registries yielded a further 16 possible eligible trials for which no additional full-text reports could be obtained (Supplementary Table 3, available from the authors on request). Eight trials had an actual or estimated completion date prior to 2016, and 1 trial started in 2013 but lacked a recorded completion date. Seven trials had the actual or estimated completion dates between May 2017 and June 2019.
DISCUSSION
We systematically reviewed the evidence for the efficacy of MSC injections in improving OA-related structural outcomes. The evidence syntheses were derived from 13 phase I or II randomized controlled trials comprising 513 participants: 9 of high quality25,27,29,30,32–35,37, 3 of moderate quality26,31,36, and 1 of low quality28. There was consistent evidence that MSC treatment improved cartilage outcomes assessed by MRI, arthroscopy, or histology, and that it has beneficial effects on subchondral bone in populations at risk of OA. However, there was significant heterogeneity in injected MSCs, modest sample sizes, methodological limitations, and potential for publication bias.
We found consistent evidence for a beneficial effect of MSC therapy on articular cartilage. Among the 11 studies examining cartilage using MRI or arthroscopy, 10 studies showed a beneficial effect of MSC injections25,26,27,28,31,32,34–37, evidenced by improved cartilage volume/thickness27,31,32,34,37, morphology35,36, quality28, and regeneration and repair25,26, assessed by MRI, arthroscopy, or histology. Results tended to be similar, regardless of the type (allogeneic or autologous) and origin (bone marrow, adipose tissue, peripheral blood, or placenta) of MSCs, and the differences in study population (stage of OA).
Six studies examined subchondral bone from MRI25,27,31,32,34,35. There was consistent evidence for a beneficial effect of MSC therapy on subchondral bone in populations at risk of knee OA, with all 3 studies showing an effect on bone expansion31, edema25, sclerosis, and osteophyte formation27. The evidence in OA populations was conflicting, with 1 study showing a beneficial effect on osteophyte formation35. Although the other 2 OA studies found no effect of MSC injections on tibial bone area, bone marrow lesions32, spur, or erosion34, the follow-up of the latter study was only 24 weeks, which may not be enough time to demonstrate an effect on subchondral bone. Bone manifestations are varied and may not be influenced by the same factors.
Four studies examining the effect of MSCs on composite MRI scores of the knee reported inconsistent results, with 2 studies reporting beneficial effect25,30 and 2 studies reporting no effect29,33. Although the overall effect of MSCs on knee structures can be assessed using the composite scores of the whole knee, this method cannot differentiate the effect of MSCs on different joint structures.
Three studies reported inconsistent results for the effect of MSCs on joint space width. While 1 study showed an effect of MSCs on increasing joint space width over 24 months31, 2 studies found no effect over 6 or 12 months30,36. Another study reported no clinically meaningful change in radiograph parameters after 6 months29. A follow-up period of up to 12 months might not be enough time to observe meaningful change in radiographic outcomes.
Our systematic review has limitations. Due to the heterogeneity in study populations; sources and contents of MSCs; dose, frequency, and schedule of MSC injections; media in which MSCs were suspended before administration; treatment modalities in the control group; and structural outcome measures, performing a metaanalysis was not possible, so a qualitative evidence synthesis was performed instead. The media in which stem cells were suspended was used as the control intervention in 6 studies25,26,27,29,31,36. Although these heterogeneities may limit the ability of our study to draw reliable conclusions, we found consistent evidence that MSC treatment improved cartilage outcomes. However, there was a lack of high-level evidence to support this due to the methodological issues in some studies. Future studies will need to reduce the bias commonly identified in previous studies. It is important to consider that all the studies included in our systematic review were phase I or II trials with modest sample sizes. Given that efficacy is generally not the main aim of phase I or II trials, and that all systematic reviews examining stem cells, including ours, have been based on early-stage clinical trials, we conducted a review of clinical trials databases to examine the potential of publication bias (e.g., only those studies with positive findings being published). We identified a further 8 possible eligible trials with an actual or estimated completion date before 2016 and 1 trial beginning in 2013 that have not been published. The reason these studies have not been published is unknown. However, this needs to be considered, as it may have inflated the effect of stem cell therapy. A further 7 studies were supported by industry funders27,29,31,32,35,36,37, which might introduce reporting bias. MSC and cell concentrate nomenclature tends to be used interchangeably in the literature, despite the fact that they are different products. It has been suggested that commonly used cell concentrates should be distinguished from laboratory-purified stem cells41,42. In our study, we only included studies of laboratory-purified/expanded stem cells.
The ability of MSCs to produce trophic factors for neuronal development and stimulate local tissue repair are key hallmarks for their increasing popularity as an intervention in degenerative diseases43,44,45. Inflammation plays an important role in cartilage damage and structural progression in OA46,47,48. MSCs may have beneficial effects on articular cartilage and subchondral bone through their antiinflammatory and immunomodulatory properties, since intraarticular injections of MSCs may affect the local environment of the joint8,9,10,11, as supported by data from animal studies49. However, the MSC metabolism and related therapeutic effects are complex, and the composition of injected MSCs is unclear and likely to be highly variable, with few stem cells surviving after injection12,13,14. The optimal tissue source, type, dose, and duration of MSC treatment is unknown, as demonstrated by the variation in intervention in this review, and a dose-response relationship has not been established.
Our systematic review, based on 13 phase I or II clinical trials, found consistent evidence for a beneficial effect of intraarticular injections of MSCs on articular cartilage and subchondral bone, irrespective of the source or contents of the MSCs. Due to the heterogeneity in the source and composition of injected MSCs, the early stage of the trials, modest sample sizes, methodological limitations, and potential for publication bias, more work is needed before the therapy is recommended in the management of OA.
Footnotes
SMH is the recipient of National Health and Medical Research Council (NHMRC) Early Career Fellowship (APP1142198). LC is the recipient of an Australian Postgraduate Award and Arthritis Foundation Scholarship. AEW and YW are the recipients of NHMRC Translating Research into Practice Fellowship (APP1150102 and APP1168185, respectively).
The authors have no conflicts of interest.
- Accepted for publication September 24, 2020.
- Copyright © 2021 by the Journal of Rheumatology