Osteoarthritis is a significant and costly cause of pain for both humans and horses. The horse has been identified as a suitable model for human osteoarthritis. Regenerative therapy with allogeneic mesenchymal stem cells (MSCs) is a promising treatment, but the safety of this procedure continues to be debated. The aim of this study is to evaluate the safety of intra-articular injections of allogeneic MSCs on healthy joints by comparing two different dosages and two different tissue sources, namely, bone marrow and umbilical cord blood, with a placebo treatment on the same individuals. We also assessed the influence of autologous versus allogeneic cells for bone marrow-derived MSC treatment. Twelve clinically sound horses were subjected to injections in their 4 fetlock joints. Each of the three fetlocks was administered a different MSC type, and the remaining fetlock was injected with phosphate-buffered saline as a control. Six horses received 10 million cells per joint, and the 6 other horses received 20 million cells per joint. Clinical and ultrasound monitoring revealed that allogeneic bone marrow-derived MSCs induced significantly more synovial effusion compared to umbilical cord blood-derived MSCs but no significant difference was noted within the synovial fluid parameters. The administration of 10 million cells in horses triggered significantly more inflammatory signs than the administration of 20 million cells. Mesenchymal stem cell injections induced mild to moderate local inflammatory signs compared to the placebo, with individual variability in the sensitivity to the same line of MSCs. Understanding the behavior of stem cells when injected alone is a step towards the safer use of new strategies in stem cell therapy, where the use of either MSC secretome or MSCs combined with biomaterials could enhance their viability and metabolic activity.
Osteoarthritis (OA) is a joint disease characterized by cartilage breakdown, subchondral bone failure, periarticular bone remodeling, and synovitis [
Stem cell-based regenerative medicine is a promising strategy given the lack of spontaneous regenerative capacity in the articular cartilage. The expected effects are explained by the chondrogenic differentiation potential of stem cells and their immunomodulatory and paracrine signaling effects [
The objectives of this study are twofold. Firstly, we sought to compare the safety of intra-articular injections in the healthy joints of 2 different allogeneic MSC tissue sources using cells of controlled quality and viability and consisting of BM-MSCs and UCB-MSCs with phosphate-buffered saline (PBS) as the vehicle control. The influence of autologous versus allogeneic cells was also assessed for BM-MSCs. Secondly, we compared the safety of these intra-articular injections with 10 million cells versus 20 million cells.
Twelve clinically sound French Standardbreds owned by the Center of Imaging and Research on Equine Locomotor Pathology (CIRALE) were included in the study. There were 7 geldings and 5 mares, all aged from 2 to 4 years old. None of the horses had a history of pregnancy, had received a blood transfusion before recruitment, or had kin relationships. Each horse was evaluated clinically with static and dynamic evaluation and underwent radiographic and ultrasonographic examination of its four fetlock joints to rule out the presence of preexisting signs of articular disease. The study protocol was approved by the ComEth Anses/ENVA/UPEC Ethical Committee (permit number: 10/06/14-8). All horses received a single injection in all 4 fetlock joints, i.e., in both metacarpophalangeal and both metatarsophalangeal joints. A uniform distribution of the different stem cell types was performed in which each horse had one front/hind fetlock joint injected with autologous BM-MSCs and its contralateral fetlock joint injected with allogeneic BM-MSCs and one front/hind fetlock joint injected with UCB-MSCs and its contralateral fetlock joint injected with the same volume of the MSC transport medium (Gibco Phosphate-Buffered Saline, Fisher Scientific SAS, Illkirch, France) as a control. Each horse was randomly assigned to one of the six stem cell treatment distributions (see Supplementary
Sternal bone marrow was taken from each of the 12 horses after sedation (detomidine 0.01 mg/kg, IV; butorphanol 0.02 mg/kg, IV) and local skin anesthesia, as previously described
MSC isolation and culture were performed in the same manner and using the same culture medium as previously described in
Cell expansion was performed in low-glucose Dulbecco’s modified Eagle medium containing 20% fetal calf serum (FCS, Invitrogen Life Technologies). The culture medium was changed three times per week and cells were passaged at 80% confluency until passage 4 (P4). At this stage, cells kept per cell lines were counted, centrifuged, and then suspended in a cryopreserved medium (6 to 10 million cells/mL) composed of 90% FCS and 10% dimethyl-sulfoxide (Sigma-Aldrich). Freezing was performed using CoolCell freezing containers and cells were stored in liquid nitrogen until needed for the study.
Immunophenotyping and trilineage differentiation were performed as previously described [
The capacity of equine MSCs to differentiate into osteogenic, chondrogenic, and adipogenic lineages was determined at P4 using culture and fixation methods described previously [
Allogeneic sources of cells were randomly selected from available cell lines with 2 lines of BM-MSCs per group of horses (group 1: dosage 20 million; group 2: dosage 10 million) and 2 lines of UCB-MSCs per group. Approximately one week before the intra-articular injections (7 days for BM-MSCs and 10 days for UCB-MSCs), cells were thawed rapidly by friction, seeded at 5000 cells/cm2, and cultured with the medium described in the cell culture section. Two to three hours before injection, cells were detached using trypsin/EDTA then suspended in 50 mL of PBS to completely remove the culture medium and especially the residual fetal bovine serum (FBS). Cells were then counted before a second centrifugation. Six batches of each MSC source containing 10 million cells/mL of PBS were prepared for injection for group 1, and 6 batches containing 5 million cells/mL of PBS were prepared for injection for group 2. For allogeneic injections, cell lines were randomly assigned to recipient horses (see paring in Supplementary
All horses were confined to
Each fetlock was clinically evaluated on day 0 (before injection) then on days 1, 3, 7, 14, and 28. Clinical evaluation was performed blindly by the same operator (SJ). Firstly, the fetlock joint circumference was measured at the middle section of the proximal sesamoid bones and midmetacarpal/metatarsal area using a shaved skin landmark. Both measurements were taken in triplicate and averaged. Secondly, sensitivity to digital flexion tests and joint effusion were evaluated using a five-point scale from normal to severe (0: normal, 1: mild, 2: moderate, 3: substantial, and 4: severe). Presence of subcutaneous oedema was also evaluated and scored using the same five-point scale (see Table
Fetlock joint effusion grading system.
Score | Physical criteria | |
---|---|---|
0 | Normal | Concave aspect of the proximopalmar recess of the metacarpo(tarso)phalangeal joint. No lateral swelling when a medial pressure is applied on the recess. |
1 | Mild | Flat aspect of the proximopalmar recess of the metacarpo(tarso)phalangeal joint. Mild lateral swelling when a medial pressure is applied on the recess. |
2 | Moderate | Convex aspect of the proximopalmar recess of the metacarpo(tarso)phalangeal joint. Lateral swelling easy to obtain when a medial pressure is applied on the recess. |
3 | Substantial | Convex aspect of the proximopalmar recess of the metacarpo(tarso)phalangeal joint exceeding the suspensory ligament branches (third interosseous muscle). Soft consistency of the recess on palpation. |
4 | Severe | Convex aspect of the proximopalmar recess of the metacarpo(tarso)phalangeal joint exceeding the suspensory ligament branches (third interosseous muscle) with a hard consistency of the recess on palpation indicating synovial pressure. Synovial distension of the dorsal recess of the joint. |
Ultrasonographic examination was performed on day 0 (before injection) then days 1, 3, 7, 14, and 28. The dorsal and collateral aspects of each fetlock were examined in transverse and longitudinal scans, using a 7.5 MHz linear transducer (Hitachi Medical Systems, Saint-Priest, France) (see Supplementary
Synovial fluid effusion grading system measured by ultrasound.
Score | Ultrasound criteria | |
---|---|---|
0 | Normal | Mild amount of fluid in the proximopalmar recess of the metacarpo(tarso)phalangeal joint. Concave aspect of the skin. No motion of the fluid when pressing on the recess. |
1 | Mild | Mild amount of fluid in the proximopalmar recess of the metacarpo(tarso)phalangeal joint. Concave to flat aspect of the skin. Motion of the fluid when pressing on the recess. |
2 | Moderate | Moderate amount of fluid in the proximopalmar recess of the metacarpo(tarso)phalangeal joint. Convex aspect of the skin. Motion of the fluid when pressing on the recess. |
3 | Substantial | Substantial amount of fluid in the proximopalmar recess of the metacarpo(tarso)phalangeal joint and mild amount of fluid on the dorsal recess of the joint. Convex aspect of the skin. |
4 | Severe | Marked amount of fluid in the proximopalmar recess of the metacarpo(tarso)phalangeal joint and substantial amount of fluid on the dorsal recess of the joint with synovial pressure. Convex aspect of the skin. |
Samples of synovial fluid (1 mL) were taken from each joint on day 0 (before injection) then days 7, 14, and 28 using the technique previously described for MSC injections (Supplementary
Statistical analysis was performed after the data has been visually assessed for normality. The mean and standard deviation of normally distributed outcomes and the median and quartiles of the others were calculated at each time point for each treatment and dosage groups (see Supplementary “Tolerated” when the assigned grades (sensitivity to digital flexion tests, joint effusion, degree of lameness, subcutaneous oedema, and synovial fluid effusion on ultrasound) were ≤1 out of 4 (or 5 for lameness). These signs were considered as acceptable signs of pain and inflammation “Not tolerated” when assigned grades were >1 out of 4 (or 5 for lameness). These were considered as excessive signs of pain and inflammation (see Supplementary
When the outcome was binary, logistic regression models were performed using generalized estimating equations (GEE) to adjust for correlated repeated measurements within each horse [
MSCs were successfully isolated and expanded in culture. Characterization by flow cytometry revealed the detection of CD29, CD44, and CD90 expression for all cell populations of each strain and the absence of CD45 and MHC class II expression. CD73 expression was donor-dependent, with expression by only part of the cell population. Similarly, CD105 was not detected for UCB-MSCs and was weakly expressed by BM-MSCs at P4 (data shown in other published reports [
All the BM-MSC strains used in the present study had high proliferative capacity and possessed multipotency capacity to differentiate into osteoblasts, adipocytes, and chondrocytes, as previously described [
Intra-articular injections were successfully performed in all fetlock joints. The cell lines from which each horse received allogeneic cells are available in Supplementary
Mean viability of MSCs after culture and preparation for injection at 25°C.
Body temperature, heart rate, respiratory rate, and appetite remained normal for all the horses throughout the study. None of the 12 horses showed sensitivity to digital flexion tests throughout the study. Measurements of fetlock circumferences did not reveal any significant changes except for UCB-MSC-treated fetlocks that showed a significantly higher mean circumference than the control fetlocks (Table
Differences in means (95% CI;
Quantitative outcomes: difference in means (95% CI); |
Binary outcomes: OR (95% CI); | ||||||
---|---|---|---|---|---|---|---|
Fetlock circumference | Total protein | Log10 (PGE2) | Log10 (CTX-II) | Total nucleated cell counts | Joint effusion | Synovial fluid effusion | |
Autologous BM-MSCs vs. placebo | 0.2 (-0.7 to 1.1); 0.66 | -0.07 (-0.3 to 0.1); 0.50 | NC | NC | |||
Allogenic BM-MSCs vs. placebo | 0.1 (-0.8 to 1); 0.77 | -0.15 (-0.4 to 0.08); 0.20 | -0.1 (-0.2 to 0.01); 0.06 | NC | NC | ||
Allogenic UCB-MSCs vs. placebo | -0.08 (-0.2 to 0.04); 0.20 | NC | NC | ||||
Autologous vs. allogenic BM-MSCs | 0.07 (-0.1 to 0.2); 0.38 | 0.2 (-0.002 to 0.4); 0.052 | 0.08 (-0.05 to 0.2); 0.24 | -0.03 (-0.1 to 0.1); 0.67 | 0.8 (0.3 to 2.3); 0.74 | 0.6 (0.3 to 1.1); 0.12 | |
Autologous BM vs. allogenic UCB-MSCs | 0.001 (-1 to 1); 0.10 | -0.01 (-0.3 to 0.3); 0.94 | 0.001 (-0.2 to 0.2); 0.99 | 0.02 (-0.1 to 0.1); 0.65 | 0.6 (0.1 to 2.5); 0.44 | 0.9 (0.3 to 1.9); 0.75 | 1.1 (0.6 to 2); 0.68 |
Allogenic BM vs. allogenic UCB-MSCs | -0.07 (-1 to 0.9); 0.88 | -0.2 (-0.5 to 0.03); 0.08 | -0.07 (-0.3 to 0.1); 0.50 | 0.04 (-0.04 to 0.1); 0.30 | 0.7 (0.2 to 2.6); 0.55 | ||
10 vs. 20 million cells | 0.5 (-0.9 to 2); 0.48 | 0.2 (-0.1 to 0.5); 0.26 | 0.1 (-0.1 to 0.3); 0.2 |
OR: odds ratio; CI: confidence interval; NC: not comparable with the model used;
Results of clinical and ultrasound parameters of fetlock joint inflammation after intra-articular injection of MSCs or control PBS. (a) Fetlock joint circumference (in cm) expressed as the
Results of clinical and ultrasound parameters of fetlock joint inflammation after intra-articular injection of 2 different dosages of BM-MSCs or UCB-MSCs. (a) Fetlock joint circumference (in cm) expressed as the
Photographs of the lateral aspect of the fetlock joint of 4 horses taken on day 0 (D0) before MSC injections (a, c, e, g), on day 1 (D1) after injection (b, d, f, h), and on day 7 (D7) after injection (i) showing the different grades of joint effusion observed. See the proximopalmar recess of the metacarpophalangeal joint (light arrow) and the distention of the dorsal recess of the joint (thick arrow). (a, b) Left front fetlock of horse 1 injected with allogeneic BM-MSCs showing grade 0 fetlock joint effusion on D0 and D1. (c, d) Left front fetlock of horse 3 injected with autologous BM-MSCs showing grade 0 fetlock joint effusion on D0 (c) and grade 1 on D1 (d). (e, f) Right front fetlock of horse 6 injected with allogeneic UCB-MSCs showing grade 0 fetlock joint effusion on D0 (e) and grade 2 on D1 (f). (g, h, i) Left front fetlock of horse 7 injected with allogeneic BM-MSCs showing grade 0 fetlock joint effusion on D0 (g), grade 4 on D1 (h), and grade 3 on D7.
Transverse ultrasound images of the lateral aspect (a-h) and dorsal aspect (i, j) of the fetlock joint of 4 horses taken on day 0 (D0) before MSC injections (a, c, e, g), on day 1 (D1) after injection (b, d, h, j), and on day 3 (D3) after injection (f, i) showing the different grades of synovial effusion measured on ultrasound. Note the aspect of the skin (arrow) from concave to convex and the amount of fluid (
Seventeen fetlocks from 8 horses injected with MSCs showed a >1/4 joint effusion grade on clinical examination (Supplementary
According to the clinical follow-up of joint effusion, synovial fluid effusion measured by ultrasound revealed that 18 fetlocks injected with MSCs from 8 horses showed >1/4 effusion and were classified in the “not tolerated” category. These “not tolerated” fetlocks were also significantly more frequent in allogeneic BM-MSC-treated fetlocks compared to UCB-MSC-treated fetlocks, but they were also more frequent in allogeneic BM-MSCs compared to autologous MSCs (Table
Synovial fluid was successfully obtained at each time point for 32/48 fetlocks of the study and for at least 3 out of 4 time points for the others. When considering the frequency of fetlocks above the threshold of 200 nucleated cells and the mean total protein concentration, all MSC-injected fetlocks showed significantly higher levels than the PBS control-injected fetlocks (Table
Results of synovial fluid analysis from fetlock joints after intra-articular injection of MSCs or PBS control. (a) Total protein concentration (g/100 mL) expressed as the
Results of synovial fluid analysis from fetlock joints after intra-articular injection of 2 different doses of BM-MSCs or UCB-MSCs. (a) Total protein concentration (g/100 mL) expressed as the
This study has two main findings. The first is the presence of significant differences between the inflammatory responses induced by allogeneic UCB-MSCs and allogeneic BM-MSCs when injected into the healthy joints of the same individuals, but only when considering synovial effusion (measured both clinically and by ultrasound), not regarding other local clinical and ultrasound signs as well as synovial fluid parameters. Allogeneic sources were also compared to autologous BM-MSCs and significant differences were noted but only between BM sources for synovial effusion measured by ultrasound and total protein concentration with higher levels after allogeneic BM-MSC injections. This confirms one previously reported finding [
PBS injections were used to monitor the response of the joint to arthrocentesis and to the transport medium. As previously reported [
As regards the clinical parameters, the most reactive parameter after injections was synovial effusion. Of the 36 fetlocks treated with MSCs, 17 (47%) had been assigned a
The results of the present study suggest that intra-articular injection of a same line of MSCs on healthy joints induces a variable degree of local inflammatory response according to the individual. Even if only one of the 12 horses (8.3%) in this study (horse 7) exhibited severe signs of pain and inflammation, this kind of reaction could be observed on any client-owned animal participating in a clinical trial. Growing evidence indicates that future clinical applications of stem cell therapies will not aim at injecting MSCs alone or simply diluted in PBS but rather suggests that MSCs would gain therapeutic interest if combined with extracellular matrix substitutes like hydrogels in order to enhance their viability and metabolic activity [
The cause of the inflammatory responses observed is still not perfectly understood. Sepsis was excluded based on synovial fluid analysis and clinical evolution (see Supplementary
Even if no dramatic increase in total nucleated cell counts was measured and every line of injected MSCs was tested for the absence of MHC class II, the signs observed could be explained by an immune reaction. Indeed, it has been demonstrated that equine allogeneic MSCs are capable of eliciting antibody responses
Fetal bovine serum should also be considered as a potential cause of the different reactions observed in the horses, despite the effort that had been made to remove all the culture media. FBS was used in our culture and a residual contamination with xenoproteins could have occurred, as highlighted by a recent study [
This study has two main limitations. First, the design implied a small number of horses in which several MSC sources were injected within the same horse. Therefore, the results of the statistical analysis should be taken with caution given that and because no correction for multiple testing has been performed. Another consequence of the study design is that an objective evaluation of the degree of lameness of each limb was not possible, meaning that this parameter could not be considered in the data analysis. Finally, this study design could potentially induce a systemic effect and affect the responses seen within the individual. The immunomodulatory capabilities of the MSCs injected were not measured in this study. Although the immunomodulatory properties and immunogenicity of MSCs have been widely reported and are already used in clinical settings [
The second limitation of this study is the absence of MHC haplotype analysis and immunophenotyping of donor and recipient horses to know if pairs were MHC matches or mismatches. If available, such analysis would be beneficial to improve the understanding of the mechanisms leading to the occurrence of horse-dependent adverse reactions. To confirm or disprove the occurrence of an immune reaction, it would also have been interesting to perform repeated injections.
This study has shown that injections into the healthy joints of the same individuals of allogeneic BM-MSCs induce significantly more synovial effusion of the joint (measured both clinically and by ultrasound) compared to allogeneic UCB-MSCs without significant differences within the synovial fluid parameters. A significant transient increase in total protein concentration and nucleated cell counts with individual variations in strength was observed following MSC injections whatever the tissue source. Despite this elevation of synovial fluid parameters, CTX-II and PGE2 concentrations show that MSCs seemed to have a protective effect on cartilage degradation rather than being deleterious. In the present study, this effect was more pronounced after injection of UCB-MSCs. This study also revealed that injection of 10 million MSCs per joint induces significantly more clinical and ultrasound signs of synovial effusion and increase in total nucleated cell count than the injection of 20 million MSCs. However, the results of this study should be taken with caution given the small number of horses, the concurrent administration of multiple MSC sources on the same individuals, and the lack of information regarding donor and recipient matching, lameness grades, and synovial fluid analysis during the first week of study. As a step toward a safer use of stem cell therapies in the future, either combined to biomaterials as vehicles or used for their secretome, this study nevertheless highlights the need for further investigation to better understand the causes of reactive arthritis induced by intra-articular injection of MSCs on the horse in order to find new preventive methods. The results of this study will also be useful in human medicine, as the horse is recognized as a reliable model for human osteoarthritis.
The database used to support the findings of this study is included within the supplementary information materials.
The authors declare that there is no conflict of interest regarding the publication of this paper.
Lélia Bertoni, Thomas Branly, Philippe Galéra, and Magali Demoor contributed equally to this work.
We thank the Ecuries Lebourgeois and the veterinary clinics CAP’VET (Saint-Julien sur Sarthe, France) for collecting equine umbilical cord blood. This research was funded by ERDF (European Regional Development Funds) grants to SJ, FA, PG, and MD (HIPPOCART 1 no. 2897/33535, 917RB148; HIPPOCART 917CB174), by a Normandy County Council program to SJ, FA, PG, and MD (HIPPOCART no. 2013-AGRI-236/13P07492, 917CB166), by Fonds Eperon to SJ, FA, PG, and MD (EQUISTEM, N80-2014, 917CB194), by the French National Research Agency (ANR) and by Normandy County Council through the ANR TecSan PROMOCART program to PG (917RB020 and 917RB072, respectively), by the French Ministry of Research and Technology to PG, and by CENTAURE European project cofunded by Normandy County Council, European Union, in the framework of the ERDF-ESF operational program 2014-2020. MDes was a recipient of a PhD fellowship from the Normandy County Council. TB was supported by a PhD scholarship from the French Ministry of Research and Technology.
Dataset S1: viability tests of MSC results.
Dataset S2: scores and values of clinical, ultrasound, and synovial fluid parameters.
Dataset S3: mean (standard deviation) or median (first quartile, third quartile) scores and values of clinical, ultrasound, and synovial fluid parameters within the 48 fetlocks from day 0 to day 28.
Figure S1: synovial fluid sampling followed by intra-articular injection of MSCs or placebo in a metatarsophalangeal joint of one horse using a lateral approach on the flexed limb.
Figure S2: ultrasound technique and reference images.
Figure S3: characterization of equine bone marrow MSCs and umbilical cord blood MSCs.
Table S1: distribution of MSC treatments. Table S2: quality control of MSCs. Table S3: sensitivity to digital flexion tests grading system. Table S4: subcutaneous oedema grading system.