Single and repeated intra-articular injections in the tarsocrural joint with allogeneic and autologous equine bone marrow-derived mesenchymal stem cells are safe, but did not reduce acute inflammation in an experimental interleukin-1β model of synovitis

Summary

Background

Allogeneic and autologous bone marrow-derived mesenchymal stem cells (BMDMSCs) have been administered in equine joints for their anti-inflammatory effects. However, allogeneic BMDMSC offer multiple clinical and practical advantages. Therefore, it is important to determine the relative effectiveness of allogeneic versus autologous BMDMSCs.

Objectives

The objective of the study was to compare the inflamed joint response to autologous versus allogeneic BMDMSCs injections, and to determine if either treatment generated an anti-inflammatory effect.

Study design

Randomised-controlled study.

Method

Bone marrow was harvested from eight horses. Autologous BMDMSCs and pooled-allogeneic BMDMSCs were culture expanded, cryopreserved and thawed immediately prior to administration. Ten million autologous BMDMSCs were administered with 75 ng rIL-1β into one tarsocrural joint and the contralateral tarsocrural joint received allogeneic BMDMSC plus 75 ng rIL-1β. Repeat injections were performed with the same treatment administered into the same joint. Four additional horses received 75 ng rIL-1β alone in a single tarsocrural joint. Clinical parameters (lameness, joint circumference and joint effusion) and synovial fluid parameters including nucleated cell count (NCC), differential cell count, total protein (TP), prostaglandin E₂ (PGE₂) and C-reactive protein (CRP) were measured at baseline, 6, 12, 24, 72, 168, and 336 hours post-injection.

Results

No difference was detected between autologous and allogeneic treatment groups with respect to subjective lameness, joint effusion, joint circumference, NCC, TP, differential cell count, CRP or PGE₂. Neither autologous nor allogeneic treatments resulted in an improvement in clinical or cytological parameters over that elicited by rIL-1β alone.

Main limitations

A single dose of rIL-1β was evaluated and resulted in a severe synovitis which may have been too severe to observe a BMDMSC-mediated effect.

Conclusions

This study revealed allogeneic and autologous BMDMSCs resulted in an equivalent clinical and cytological response. Allogeneic and autologous BMDMSCs were equally ineffective in reducing the inflammatory response from acute rIL-1β-induced joint inflammation in horses.

Introduction

Bone marrow-derived mesenchymal stem cells (BMDMSCs) have shown promise in the treatment of inflammatory musculoskeletal conditions including osteoarthritis, desmitis and tendonitis. [1–6] In vitro studies have documented the anti-inflammatory effects of both allogeneic and autologous BMDMSCs. [7,8]

Interleukin-1β (IL-1β) is an important inflammatory mediator in naturally-occurring equine, osteoarthritis (OA) which within the joint environment results in the production of matrix metalloproteases and prostaglandin-E₂. [9,10] Because of this, treatments directed against IL-1β such as IL-1β receptor antagonist protein have resulted in improved clinical outcomes and disease modifying effects. [11,12] Although BMDMSCs have been increasingly used as an anti-inflammatory joint therapy, intra-articular administration has been found to result in an inflammatory response in normal joints. [13–15] Further, no studies have investigated the effect of BMDMSCs in an inflammatory model of disease. These controlled, experimental models are important for assessing BMDMSCs as an anti-inflammatory therapy.

Researchers have long argued over the use of allogeneic BMDMSCs for intra-articular injection. Allogeneic BMDMSCs have been used to treat horses with joint disease without inducing obvious negative effects. [1,16] Experimental studies comparing allogeneic and autologous BMDMSCs have yielded conflicting results. [14,17–19] A study by Joswig et al. suggested that allogeneic BMDMSCs may be more inflammatory. [14] However, the study showed only a single statistically significant difference between allogeneic and autologous BMDMSC, with a higher nucleated cell count (NCC) for a single day following the second injection of BMDMSCs. [14] This study and other studies have compared allogeneic and autologous BMDMSCs by administering the treatment in different cohorts of animals. With variable individual reactions of horses to intra-articular treatments, this can be difficult to interpret. Therefore, the present study sought to compare allogeneic and autologous BMDMSCs within the same cohort of animals.

The objectives of this study were three-fold. First, the authors sought to compare the reaction of the inflammatory joint to allogeneic and autologous BMDMSCs. Second, we sought to determine whether a repeat injection changed the intra-articular response to BMDMSC injection within inflamed joints. Third, we sought to determine whether autologous or allogeneic BMDMSCs elicit an equivalent anti-inflammatory effect when injected into joints with recombinant equine interleukin-1β (rIL-1β) induced inflammation. We hypothesised that there would be no difference in the reaction of the inflamed joint to autologous versus allogeneic BMDMSCs after a single injection or two repeated injections, and that both autologous and allogeneic BMDMSCs would result in an anti-inflammatory effect.

Materials and Methods

Animals

Eight horses (16 joints) were used in the study. Horses were of mixed breed and ranged in age from 2–5 years old. All horses were determined to be sound prior to entering the study with no response to joint flexions and no effusion present in the tarsocrural joints. Treatment limbs were randomised, and all investigators and staff were unaware of treatment assignment with the exception of the first author.

Treatment groups

BMDMSC isolation, culture, expansion and cryopreservation details are included in Supplementary Item 1. Commercially available equine recombinant IL-1β^a was used. The autologous treatment (AUTO) limb was given 75 ng of rIL-1β with 10 million autologous BMDMSCs in 1 ml of freeze media. The allogeneic treatment (ALLO) limb was given 75 ng of rIL-1β with 10 million pooled-allogeneic BMDMSCs in 1 ml freeze media. Eight horses received one treatment in one tarsocrural joint and the other treatment in the contralateral tarsocrural joint a week later. All treatments were administered into the dorsal pouch of the tarsocrural joint. Treatments were then repeated at 2 weeks in the same limb they were given previously (again, with a one-week interval between treatments). Whether the horse received ALLO or AUTO for the first treatment was determined randomly using a random number generator (www.random.org). Four additional horses were treated once with 75 ng of rIL-1β in the dorsal pouch of the tarsocrural joint with no other treatment (previously published data).[20] These horses underwent the same clinical and cytological analysis as the AUTO and ALLO treatment groups and were used to evaluate the effect of BMDMSCs on the synovitis created by the rIL-1β model. Only the first injection of AUTO and ALLO were compared to rIL-1β alone. The study design is described in Supplementary Item 2. Methods of treatment administration are described in Supplementary Item 1.

Evaluation of clinical response to treatment

Clinical evaluations were performed at 0, 6, 12, 24, 72, 168 (1 week), and 336 (2 weeks) post-injection hours (PIH) including a physical examination and lameness. Clinical assessment and arthrocentesis were performed prior to treatment administration. At each time point, joints were evaluated for joint circumference, joint effusion, and heat. A board-certified equine surgeon (A.C.) performed all subjective lameness evaluations. Subjective lameness was graded using the AAEP lameness scale; half points were awarded at the discretion of the evaluator. [21] Wireless motion analysis system (Lameness Locator®)^b was used for objective lameness evaluation.

The location of joint circumference measurement was chosen by palpation (as the middle of the joint pouch) prior to beginning the study and marked by clipping the hair. Joint circumference (cm) was measured three times, consecutively, at the same location on the limb at each time point. For each time point, the three values were averaged. Joint effusion was given a subjective clinical grade with grade 0 indicating no effusion, grade 1 indicating slight effusion, grade 2 indicating mild effusion, grade 3 indicating moderate effusion, and grade 4 indicating severe effusion. All measurements were conducted by a single observer.

Synovial fluid analysis

At each time point (0, 6, 12, 24, 72, 168, 336 PIH), arthrocentesis was performed aseptically following clinical assessment. Horses were sedated using detomidine hydrochloride (0.01 mg/kg i.v.) and butorphanol tartrate (0.01 mg/kg i.v.) and synovial fluid was harvested from the dorsal pouch of the tarsocrural joint prior to treatment. Synovial fluid was immediately placed in plain glass tubes and processed within one hour of collection. A portion of the aspirate was used for direct smear and cytospin analysis for determination of differential neutrophil, monocyte, lymphocyte and eosinophil counts. Hyaluronidase digestion was performed prior to using an automated cell counter to determine total nucleated cell count (NCC). Total protein (TP) content was determined using a refractometer. The remainder of the synovial fluid was centrifuged for 10 min at 1000 x g and the supernatants were stored at −80°C in Eppendorf tubes. Multiple aliquots were frozen to prevent freeze-thaw cycles until c-reactive protein (CRP) and Prostaglandin-E2 (PGE₂) analysis could be performed.

Enzyme-linked immunosorbent assays

Synovial Prostaglandin E2 (PGE₂) was evaluated as previously described [22], using a commercially available equine specific PGE₂ Enzyme-linked immunosorbent assay (ELISA) kit^c. Synovial C-reactive protein (CRP) was evaluated using a commercially available ELISA kit^d. Synovial PGE₂ and CRP were evaluated at baseline, 24 hours, 72 hours and 168 hours following injection of AUTO, ALLO, and rIL-1β alone.

Data analysis

An a priori power analysis was performed. The power calculation was based on prior joint studies with described differences in clinical parameters, as well as synovial cytokine levels, total protein and nucleated cell counts. [6,20,22–24] The power calculation suggested that 8 horses would achieve a power of 0.8 and an alpha error rate of 0.05. Clinical and synovial fluid data were compared using a two-way ANOVA for repeated measures with time defined as the within subjects factor, and the treatment (AUTO vs. ALLO) defined as a between-subjects effect. In order to compare the 8 horses given BMDMSCs versus the 4 horses only administered rIL-1β, a two-way ANOVA (without repeated measures) was performed. Means were compared between treatments at each time point using Sidak’s multiple comparison test. Significance was set at P≤0.05. Normality was assessed by evaluating diagnostic plots of the residuals for each variable. Log transformation was performed for nucleated and differential cell count data. Statistical analysis was conducted using GraphPad Prism (version 7.03)^e.

Results

Clinical responses

Physical examination

Administration of rIL-1β did not result in a significant increase in temperature, pulse or respiration from baseline in any of the groups. When comparing horses that received rIL-1β alone with the ALLO and AUTO treatment groups, two statistically significant differences were noted. Twelve hours following the first injection, the heart rate was higher in the group receiving rIL-1β alone compared to AUTO (P = 0.02) and ALLO (P = 0.002). In addition, 24 hours following the first injection, the respiratory rate was higher in the group receiving rIL-1β alone compared to AUTO (P = 0.02) and ALLO (P = 0.02). No other significant differences were detected in physical examination parameters between horses receiving rIL-1β alone and those receiving rIL-1β with BMDMSCs. No difference was detected in the temperature, respiratory rate or heart rate between AUTO and ALLO treatment groups at any time point following the first or second injection (Fig 1).

Fig 1: Physical examination parameters.

Twelve hours following the first injection, the heart rate was higher in the group receiving rIL-1β alone compared to AUTO (P = 0.02) and ALLO (P = 0.002). At 24 hours following the first injection, the respiratory rate was higher in the group receiving rIL-1β alone compared to AUTO (P = 0.02) and ALLO (P = 0.02). Significance was set at P≤0.05. Individual data points are plotted. Different letters indicate a significant difference between bars. Bar graphs indicate the mean and standard deviation.

Lameness

Administration of rIL-1β resulted in an increase in subjective lameness by 12 hours post-injection compared to baseline (P = 0.05). Both treatment groups (ALLO and AUTO) showed an increase in subjective lameness score by 6 hours post-injection (P<0.001) and lameness continued 24 hours post-injection for the AUTO treatment group (P<0.001) and 72 hours post-injection for the ALLO treatment group (P = 0.02). When the subjective lameness scores of the AUTO and ALLO group were compared to horses receiving rIL-1β alone, the AUTO (P = 0.008) and ALLO (P = 0.04) treatment groups had a greater increase in subjective lameness score at 6 hours post-injection (Table 1). After the first injection, no difference was found in change in subjective lameness between AUTO and ALLO treatments at any time point (Table 1). Following the second injection, both treatment groups showed an increase in subjective lameness score for 24 hours post-injection (P<0.001). No difference was found in change in subjective lameness between AUTO and ALLO treatments at any time point (Table 1).

Table 1: Change in subjective lameness score.

Increase in subjective lameness score did not differ between AUTO and ALLO at any time point for the first or second injection. When the first injection of AUTO and ALLO were compared to a single injection of rIL-1β alone, there was a significant increase in lameness at 6 hours for both the AUTO (P = 0.0082) and ALLO group (P = 0.0387). The second injection of AUTO and ALLO was not compared to rIL-1β alone as only a single injection of rIL-1β was performed. Significance was set at P<0.05.

Time	1st injection: AUTO mean (s.d.)	1st injection: ALLO mean (s.d.)	2nd injection: AUTO mean (s.d.)	2nd injection: ALLO mean (s.d.)	Single injection: 75ng rIL-1β mean (s.d.)	P-value
6	2.6c (0.7)	2.0b (1.9)	3.7 (1.1)	4.3 (0.3)	1.5a (1.0)	P(ab) = 0.04 P(ac) = 0.008
12	3.8 (0.4)	3.6 (0.7)	3.9 (0.9)	4.1 (0.2)	3.1 (1.1)
24	3.6 (0.5)	3.3 (0.7)	2.7 (1.1)	3.2 (0.5)	1.9 (1.2)
72	0.8 (1.0)	1.0 (1.1)	0 (0.9)	0.4 (0.5)	0.8 (0.5)
168	0.4 (1.1)	0.5 (1.1)	0 (0.8)	0.6 (1.0)	0 (0.5)

For Lameness Locator® data, the DiffMax and DiffMin for each time point were compared to baseline and reported as the change in DiffMax and DiffMin (mm). Administration of rIL-1β alone did not lead to statistically significant increases in DiffMin or Diffmax compared to baseline. There was no significant difference in the Lameness Locator measurements (DiffMin, DiffMax) between the animals treated with only rIL-1β and those treated with rIL-1β and BMDMSCs (Fig 2). For the first injection, the change in DiffMax and DiffMin were greater in the AUTO group versus the ALLO group at 6 hours post-injection (DiffMax: P = 0.01, DiffMin: P = 0.04) (Fig 2). After the second injection, the change in DiffMax and DiffMin were greater for the AUTO group versus the ALLO group at 12 hours (DiffMax: P = 0.0001, DiffMin: P = 0.002) (Fig 2).

Fig 2: Lameness Locator®.

Six hours following the first injection, the change in DiffMax (P = 0.0147) and DiffMin (P = 0.0428) was greater in the AUTO versus the ALLO treatment group. Twelve hours following the second injection, the change in DiffMax (P<0.001) and DiffMin (P = 0.002) was greater for the AUTO treatment group versus the ALLO treatment. No significant difference was found between the horses receiving rIL-1β alone and the AUTO group 6 hours following the first injection (DiffMin, P = 0.0622; DiffMax, P = 0.07). Significance was set at P<0.05. Individual data points are plotted. Different letters indicate a significant difference between bars. Bar graphs indicate the mean and standard deviation.

Joint circumference and effusion score

Joint circumference and a subjective joint effusion score were used to evaluate joint distention post-injection. Administration of rIL-1β resulted in a statistically significant increase in joint circumference by 72 hours post-injection (P = 0.05). Joint circumference was significantly increased 6 hours post-injection for both treatment groups (AUTO P<0.0001, ALLO: P = 0.0007). Change in joint circumference continued to be increased from baseline for both groups through 336 hours following the first injection (P<0.001). Joint circumferences were not improved in the AUTO or ALLO groups compared to control joints administered rIL-1β alone (Fig 3). Following the first injection, no difference was detected in the change in joint circumference between the AUTO and ALLO treatment groups at any time point (Fig 3).

Fig 3: Change in joint circumference.

There was no significant difference in the change in joint circumference (cm) between AUTO and ALLO at any time point following the first or second injection. There was no significant difference in change in joint circumference between horses which received rIL-1β alone and those in the ALLO or AUTO group. Significance was set at P<0.05. Individual data points are plotted. Bar graphs indicate the mean and standard deviation.

Following the second injection, joint circumference was significantly increased for animals in the AUTO treatment group by 12 hours (P<0.001) and the ALLO treatment group by 6 hours (P<0.001). There was no significant difference in the change in joint circumference between AUTO and ALLO treatment groups at any time point (Fig 3).

Subjective joint effusion score was assessed at every time point by the same observer. Administration of rIL-1β resulted in a statistically significant increase in joint effusion by 6 hours post-injection (P = 0.03). Both AUTO and ALLO treatment groups showed a significant change in subjective joint effusion score from baseline by 6 hours following the first injection (P<0.0001) and second injection (P<0.001). There was no improvement in joint effusion score in treatment groups administered BMDMSCs versus rIL-1β alone (Table 2). After the first injection and second injection, no difference was detected between AUTO and ALLO treatment groups at any time point (Table 2).

Table 2: Change in subjective effusion score.

There was no significant difference in the change in subjective joint effusion score between AUTO and ALLO at any time point following the first or second injection. No difference was fosund in the change in subjective effusion score between horses receiving rIL-1β alone and those receiving BMDMSCs with rIL-1β. Significance was set at P<0.05.

Time	1st injection: AUTO mean (cm)	1st injection: ALLO mean (cm)	2nd injection: AUTO mean (cm)	2nd injection: ALLO mean (cm)	Single injection: 75ng rIL-1β mean (s.d.)
6	2.1 (0.6)	1.8 (0.9)	1.9 (0.5)	2.1 (0.6)	1.5 (0.6)
12	3.4 (0.4)	2.7 (0.5)	3.0 (0)	2.6 (0.5)	3.0 (0.8)
24	3.1 (0.9)	2.6 (0.5)	2.1 (0.4)	1.9 (0.4)	3.3 (0.5)
72	2.0 (0.5)	1.4 (0.9)	1.0 (0)	1.1 (0.4)	2.0 (0.0)
168	1.6 (0.5)	1.1 (0.4)	0.4 (0.5)	0.7 (0.8)	1.3 (0.5)
336	1.1 (0.4)	0.6 (0.5)	0.1 (0.4)	0.4 (0.5)	1.3 (0.5)

Synovial fluid analysis

Arthrocentesis was performed at each time point and synovial fluid was assessed for NCC, TP and differential cell counts. In horses administered rIL-1β alone, NCC was increased by 6 hours (P = 0.04). Following the first injection, both AUTO and ALLO treatment groups had an increase in NCC compared to baseline 6 hours post-injection (P<0.001) which persisted for 72 hours for both treatment groups (P<0.001). Administration of BMDMSCs (AUTO or ALLO) did not result in a significant decrease in inflammation as determined by the NCC. In fact, the NCC was significantly higher at 24 hours (P = 0.009) and 72 hours (P = 0.02) in the AUTO treatment group compared to rIL-1β alone. Further, the NCC was significantly higher at 24 hours (P = 0.03) in the ALLO group compared to rIL-1β alone (Fig 4). Following the second injection, there was a significant increase in NCC for both groups by 6 hours (AUTO, ALLO: P<0.001) compared to baseline and persisting for 72 hours post-injection (AUTO, ALLO: P<0.001).There was no significant difference in the NCC between AUTO and ALLO treatment groups at any time point after the first or second injection (Fig 4).

Fig 4: Synovial nucleated cell count and total protein.

There was no significant difference in the nucleated cell count or total protein between the AUTO and ALLO treatment groups at any time point following the first injection or second injection. The NCC was significantly higher at 24 hours (P = 0.0087) and 72 hours (P = 0.0198) in the AUTO treatment group versus rIL-1β alone. The ALLO group had a significantly higher NCC at 24 hours (P = 0.0272) compared to rIL-1β alone. There was no significant difference in the TP at any time point between AUTO, ALLO, and rIL-1β alone. Significance was set at P≤0.05. Individual data points are plotted. Different letters indicate a significant difference between bars. Bar graphs indicate the mean and standard deviation.

Differential cell counts were determined for each time point post-injection. In horses administered rIL-1β alone, monocytes were increased by 12 hours post-injection (P = 0.004).

The AUTO (P = 0.05) and ALLO (P = 0.04) treatment groups had a significantly higher monocyte count at 24 hours compared to horses which received rIL-1β alone (Fig 5). There was no difference in monocyte count between AUTO and ALLO treatment groups at any timepoint following the first or second injection (Fig 5). Monocyte counts increased significantly for both AUTO and ALLO treatment groups by 6 hours post-injection (AUTO, ALLO: P<0.001) and continued to be increased for 72 hours post-injection (AUTO: P<0.001, ALLO: P = 0.01). Following the second injection, there was a significant increase in monocytes for 24 hours following injection (P = 0.001) in the ALLO group, and 72 hours (P<0.001) post-injection in the AUTO group.

Fig 5: Synovial differential cell count.

There was no significant difference in the differential cell counts between AUTO and ALLO treatment groups at any time point following the first or second injection. The monocyte count was significantly higher in the AUTO and ALLO group versus horses receiving rIL-1β alone at 24 hours following the first injection. Significance was set at P≤0.05. Individual data points are plotted. Bar graphs indicate the mean and standard deviation.

In horses administered rIL-1β alone, neutrophils were increased by 6 hours post-injection (P = 0.03). The neutrophil count was significantly increased for both treatment groups by 6 hours following the first injection (P<0.001) and continuing through 72 hours post-injection (P<0.0001). Following the first injection, no significant difference was found in the neutrophil counts between horses receiving AUTO and ALLO versus rIL-1β alone (Fig 5). After the second injection, the neutrophils significantly increased by 6 hours (P<0.001) for both groups (ALLO and AUTO) and continued to be increased until 72 hours in both groups (P<0.001) and through one week in the ALLO group (P = 0.02). No significant difference was found in the neutrophil counts between horses receiving AUTO and ALLO at any time point following the first or second injection (Fig 5). There was no difference in lymphocyte or eosinophil counts between AUTO and ALLO at any time point after the first or second injection, and neither lymphocytes nor eosinophils increased significantly from baseline (data not shown).

In horses administered rIL-1β alone, total protein was increased by 6 hours post-injection (P = 0.02). There was no significant difference in the synovial fluid TP in horses receiving BMDMSCs versus r-IL1β alone (Fig 4). Following the first injection, both treatment groups had a significant increase in synovial fluid TP at 6 hours post-injection (AUTO, ALLO: P<0.001) which persisted for 2 weeks following injection (AUTO: P<0.001, ALLO: P<0.001). There was no difference in the synovial fluid TP between AUTO and ALLO groups for any time point following the first or second injection (Fig 4). As the synovial fluid TP had not returned to baseline by 2 weeks following the first injection, following the second injection, a significant change from baseline was noted in the AUTO group only at 6 hours (P = 0.04), 12 hours (P<0.001) and 24 hours (P<0.001) post-injection and in the ALLO group only at 12 hours (P<0.001) and 24 hours post-injection (P = 0.003).

In horses that received rIL-1β alone, there was no significant increase in CRP, but a significant increase was noted in synovial PGE₂ by 72 hours post-injection (P = 0.03). In the ALLO and AUTO groups, synovial PGE₂ levels were increased by 24 hours following the first injection (AUTO: P = 0.01, ALLO: P = 0.03) and second injection (AUTO: P = 0.001, ALLO: P<0.001) (Fig 6). Following the first injection, synovial CRP increased mildly but significantly at 24 hours (compared to baseline) in the AUTO group only (P = 0.02). Following the second injection, synovial CRP increased mildly but significantly (compared to baseline) in the ALLO group only (P = 0.02) (Fig 6). There was no significant difference in the change in synovial CRP or PGE₂ when horses received rIL-1β alone versus those which received BMDMSCs and rIL-1β. No significant difference was identified in the change in synovial CRP and PGE₂ between AUTO and ALLO treatment groups at any time point following the first or second injection.

Fig 6: Change in synovial PGE₂ and synovial CRP.

There was no significant difference in the change in synovial PGE₂ (pg/ml) or synovial CRP (ng/ml) between ALLO and AUTO at any time point following the first or second injection. There was no difference in the change in synovial PGE₂ or CRP between horses receiving rIL-1β alone and those receiving ALLO or AUTO following the first injection. Significance was set at P≤0.05. Individual data points are plotted. Bar graphs indicate the mean and standard deviation.

Discussion

The use of allogeneic BMDMSCs have multiple practical clinical advantages for the treatment of joint injuries and inflammation in horses, including being an off-the-shelf therapy and potentially less expensive, while also increasing the overall availability of stem cell therapy to veterinarians. They also have potential medical advantages; allogeneic cells may be screened and characterised for their healing abilities prior to administration. Age and disease state negatively affect stem cell health and efficacy; [25,26] allogeneic cells would surpass these barriers and provide potentially better cells for healing.

Allogeneic stem cells have been evaluated alone and compared to autologous stem cells in vitro and in vivo in horses. [7,14,27–29] These studies have yielded conflicting results. A single in vivo study comparing intra-articular injection of allogeneic and autologous BMDMSCs identified an increase in the synovial total nucleated cell count when allogeneic cells were repeatedly administered; however, this was a transient response for a single day and there were no differences in clinical parameters. [14] Likewise, a single study of intravenous administration revealed an increase in CD8+ T cells following injection of allogeneic BMDMSCs but no clinical effects. [27] In fact, large clinical trials of intravenous and intra-articular administration of allogeneic cells have demonstrated no adverse effects. [1,2,28,29] One potential reason for conflicting results in small experimental studies may be the variability in individual horses’ responses to any intra-articular medication including the administration of mesenchymal stem cells (MSCs). [4,14,30] Therefore, the authors of the current study sought to compare the intra-articular response after intra-articular injection of autologous and allogeneic BMDMSCs within the same cohort of animals.

Unlike most pharmaceutical drugs, in the case of cellular therapies, living cells can respond and react to their environment. In vitro studies have shown BMDMSCs appear to be primed by inflammation creating a more anti-inflammatory phenotype. [31,32] However, this may be stimuli dependent as a different in vitro study found BMDMSCs increase expression of inflammatory mediators when exposed to inflammatory stimuli. [33] Therefore, it is imperative that stem cells be evaluated experimentally in both the normal and inflammatory joint environment. Studies in normal joints cannot be effectively extrapolated to predict the results of BMDMSC treatment in the clinical, inflamed joint. In the present study these cells were evaluated in a well-established model of synovitis. [20,22,34]

No significant differences between the two treatment groups (ALLO versus AUTO) were identified in clinical parameters including subjective lameness, joint effusion or joint circumference at any time point following the first or second injection. The only clinical difference identified between the ALLO and AUTO groups was an increase in objective lameness parameters in the AUTO group compared to the ALLO group at one time point (6 hours) following the first and one timepoint (12 hours) following the second injection. This difference should be interpreted with caution, as the change in subjective lameness grades was not different and no other significant clinical or cytological differences were identified. No differences were found in the synovial fluid NCC, TP, differential cell counts, synovial PGE₂ or synovial C-reactive protein between AUTO and ALLO treatment groups at any time point following the first or second injection. The study found no appreciable difference in the inflammatory joint clinical or cytological reaction to AUTO versus ALLO stem cells at any time point following the first or second injection.

In our study, AUTO and ALLO BMDMSCs resulted in no clinical or cytological improvement in comparison to joints treated with rIL-1β alone. In fact, at a single time point (6 hours) post-injection an increase in subjective lameness was appreciated in both the AUTO and ALLO treated groups compared to horses treated with rIL-1β alone. Although not substantiated by objective lameness data this result is consistent with previous studies of intra-articular administration which find BMDMSCs may result in an inflammatory effect. [13–15] A single previous study in horses evaluated the use of umbilical derived mesenchymal stem cells in a lipopolysaccharide (LPS) induced inflammatory joint model. This study showed a decrease in nucleated cell counts when MSCs were injected together with LPS in the tarsocrural joint. [17] Although this study revealed promising data which supported allogeneic stem cell use, an in vitro study with equine BMDMSCs revealed that MSCs may respond differently to different inflammatory stimuli. [33] IL-1β is a cytokine produced by joint tissues (unlike LPS), therefore, the rIL-1β model may be more clinically relevant. [22,35] In addition, MSCs derived from various tissues may respond differently to inflammation. [36] BMDMSCs are commonly used in equine practice. Therefore, understanding the response of BMDMSCs to joint inflammation is particularly important. The authors are unaware of an additional experimental study comparing autologous and allogeneic MSC in an inflamed joint model.

In the present study, the ALLO and AUTO treatment groups appeared to transiently increase inflammation when compared to rIL-1β administration alone. This was evident as a transient increase in NCC, monocyte count, and subjective lameness in horses treated with BMDMSCs versus those administered rIL-1β alone. Traditionally, rIL-1β causes a neutrophilic inflammation. [20,22] Therefore, it is particularly interesting that BMDMSCs have resulted in an increase in monocytes over rIL-1β alone. A sustained monocytic inflammation has been found in previous studies of intra-articular administration of BMDMSCs.[13] This finding does not discount the significant clinical success that has been reported by others, [1,29] as BMDMSCs mechanism of action is still largely unknown; anti-inflammatory properties may not be what is responsible for clinical healing. In addition, the clinical effect of a transient monocytic inflammation is unknown. The study does indicate that AUTO and ALLO BMDMSCs are not effective in reducing the severe inflammation induced by rIL-1β at the dose (75 ng) administered in this study.

The potential reasons for the lack of response to BMDMSC treatment in this model are several. For example, it is possible the inflammation induced was so severe as to override the milder anti-inflammatory effects of BMDMSC. In our model, a single dose (75 ng) of rIL-1β was utilised. The dose administered was lower than previous doses described in the literature. [22,35] The lower dose was chosen as previous researchers found the tarsocrural joint to be more sensitive to rIL-1β administration. [34] In addition, it is difficult to compare doses of rIL-1β between studies. Potency of rIL-1β depends on dilution and storage, and potency is variable between lots due to manufacturing processes and testing. This dose resulted in an acute and severe synovitis. It is possible the inflammation was too great for the anti-inflammatory properties of BMDMSCs and did not effectively mimic the level of inflammation present in a typical osteoarthritic joint. In addition, the acute synovitis caused by rIL-1β may not be the most appropriate for determining long-term efficacy of BMDMSCs. Clinically and experimentally, BMDMSCs appear to have an anti-inflammatory effect, [1,7] therefore, a lesser degree of synovitis may have resulted in evidence of this anti-inflammatory effect. In addition to controversy regarding dose of rIL-1β, the optimal time for mesenchymal stem cell administration remains unclear. Therefore, BMDMSCs may have been ineffective in reducing inflammation because of inappropriate timing of administration. BMDMSCs may have been more effective if administered prior to the onset of inflammation or following the initial inflammatory phase. Unfortunately, rIL-1β induced inflammation has a rapid onset and short-term of action. Because of this, the authors did not delay administration of BMDMSCs and administered BMDMSCs concurrently with rIL-1β. Administration prior to onset of inflammation could have been performed but is not clinically applicable.

Opposite treatments (ALLO vs. AUTO) were administered into contralateral tarsocrural joints at one-week intervals; repeat treatments (into the same tarsocrural joint) were administered at 2 weeks (Supplementary Item 2). Baseline lameness returned to a mean of less than 1 out of 5 (AAEP lameness scale) prior to treatment of the contralateral limb. However, we recognise that residual lameness in the contralateral limb may complicate interpretation of the lameness data. Due to this concern, all lameness is reported as a change from baseline lameness (with baseline lameness as the lameness at the time of treatment administration). However, a small degree of residual lameness at the time of contralateral limb treatment is recognised as a limitation of study. The two-week interval for repeat treatment administration was chosen to mimic a clinically relevant inflammation with an initial inflammatory response and gradual resolution followed by a second acute inflammation and gradual resolution. A complete resolution of inflammation was not achieved nor expected prior to the second treatment administration.

The optimal dose of BMDMSCs for inflammation has not been determined. Clinically, the authors routinely administer 10 – 20 × 10^6 cells per joint. [4,6] The ability of the BMDMSCs to decrease inflammation may also be related to dose. Studies investigating the use of BMDMSCs for human knee osteoarthritis have used much larger doses of BMDMSCs. [37] It is difficult to speculate whether a larger dose of BMDMSCs would result in an anti-inflammatory effect or promote a greater inflammation. A dose escalation and titration of BMDMSCs may be warranted to further determine the effect of BMDMSCs on rIL-1β induced synovitis.

Culture conditions must be discussed when BMDMSCs are used in any study; both culture expansion methods and cryopreservation can have a significant effect on immunophenotype of mesenchymal stem cells. [38,39] Although our stem cells were cryopreserved in equine serum, fetal bovine serum was used in our culture period. Recent studies have reported inflammatory reactions because of intra-articular administration following culture expansion in fetal bovine serum. [14] However, because all cells were treated the same and AUTO and ALLO were given in the same animal, individual inflammatory reaction to fetal bovine serum would have been equal between treatments.

Some recent in vitro studies have identified allogeneic MHCII positive cells as immunogenic. [19,40] However, no in vivo studies have been able to correlate MHCII expression with a negative outcome or intra-articular inflammatory response. In addition, the majority of studies report a lack of MHCII expression by equine BMDMSCs [40–42], and BMDMSCs in our laboratory have a routinely low to absent level of expression of MHCII. [7] Our study did not evaluate the expression of MHCII by pooled allogeneic MSCs. Although the importance of MHCII expression by BMDMSCs administered in vivo is still unknown, not determining MHCII expression by the BMDMSCs is a limitation of our study.

This study used the same cohort of animals to evaluate the response to AUTO and ALLO BMDMSCs. Stem cell tracking studies in equine tendonitis models have identified stem cell migration, [43] and a murine study identified migration of intravenously administered MSCs to inflamed joints. [44] However, recent studies following intra-articular administration of stem cells have shown prolonged retention in the joint. [45–47] In fact, a recent study in rats was able to identify retention of stem cells in inflammatory (surgical) joints for 10 weeks following intra-articular administration. [48] Likewise, a study in sheep identified MSCs 12 weeks after intra-articular administration. [47] Although a limitation of the present study, the authors felt the possibility of stem cell migration was offset by the importance of controlling for individual variation in response to BMDMSC administration.

In conclusion, the current study did not identify significant clinical or cytological differences between horses treated with AUTO or ALLO BMDMSCs in an rIL-1β model of synovitis. Neither AUTO nor ALLO BMDMSCs reduced inflammation induced by 75 ng of rIL-1β. In fact, a transient increase in NCC, monocyte count, and subjective lameness resulted from BMDMSC treatment. The current study would suggest that BMDMSCs are unsuccessful at mitigating acute, severe, synovitis. However, there was no difference in the inflamed joint response to intra-articular AUTO and ALLO BMDMSCs after the first or second injection. Although the rIL-1β model of synovitis is valuable, joint pathology is multifaceted and future in vivo studies modeling natural joint pathology are necessary to determine the clinical utility of BMDMSCs.

Supplementary Material

supp info1

Supplementary Item 1: BMDMSC isolation, culture, expansion and cryopreservation and method of treatment administration.

supp info2

Supplementary Item 2: Study design flow chart.