The present systematic review focuses on the following two intra-articularly applicable orthobiologic therapeutic concepts for equine OA.

A distinction was made between systematic analysis and meta-analysis. To obtain a general overview, all experimental studies with a follow-up time of more than 6 months were examined in the systematic review. In the meta-analysis, only randomized and controlled trials (RCTs) with a follow-up period of more than 6 months were examined according to the following inclusion and exclusion criteria. To present the results as clearly as possible, the PICO method was used. (a) Population: horses with naturally occurred OA; (b) Interventions: intra-articular therapy by MSCs alone or by MSCs in combination with autologous blood products, or autologous blood products alone; (c) Comparison: degree of lameness before and after intra-articular treatment (comparison of success rate, horses working on competition, horses working at trainings level, lame free horses); (d) Outcome: degree of lameness and adverse effects; (e) Study designs: for the systematic review all experimental studies were included, for the meta-analysis randomized controlled trials were included.

The following studies were excluded: (a) treated animals other than horses and diseases other than OA; (b) use of treatment method other than intra-articular; (c) no clear lameness diagnostics used; (d) not published in English or German; German language was included as this is the authors mother language and articles could be assessed in detail (e) no complete replication of quantitative data of the treated animals (for example individual degree of lameness).

The following research platforms were used (listed according to weighting): PubMed, Google Scholar and CAB direct. Literature searches were carried out using the following keywords: “horse/equine,” “joint/osteoarthritis,” “intra-articular,” “regenerative therapy,” “return/performance.” The search terms could be summarized with the Boolean operators “AND” or “OR” (84). The research was conducted between January 2021 and March 2022. A comprehensive literature search on orthobiologic based joint therapies in horses was undertaken, including all studies published in English and German. This initial investigation summarized 271 findings, of which all studies were examined according to inclusion and exclusion criteria. Subsequently, this initial investigation delivered 86 results. In addition, the reference list of all 86 papers were manually checked for research-relevant studies. To ensure that no meta-analyses relevant to this topic were available, a hit query was performed on PubMed using the two keywords “horse” and “meta-analysis.” The response resulted in 79 meta-analyses. This compares to 18 matches with three real meta-analyses in 2017 (85). These results prove clearly that meta-analysis is becoming more and more relevant in evidence-based medicine. None of these 79 meta-analyses deals in a similar or identical way with the issue investigated in this research. The subsequent table lists the most important studies, sorted by intra-articular administered products, in horses with naturally occurring OA compared to induced OA. In addition, the study duration is indicated >6 months (Table 1).

To meet the aim of the topic, only in vivo studies were analyzed. For the systematic review, all experimental studies were included regardless their level of evidence or design, with and without a control group. The control group was defined as another horse, another leg (contralateral limb) or another treatment method. The following data were examined and listed according to the following aspects: author, year of publication, type of study (RCTs/ No-RCTs), sample size, treatment protocol, treated joint/joints, placebo-controlled, adverse reactions, follow-up time, lameness evaluation (horses working at trainings level/lame free horses/horses working on competition level/success rate).

For the intra-articular treatment regimen with orthobiologic therapeutics, there were no specifications regarding diagnostic methods, treatment frequency, dosage, and preparation of the appropriate therapeutics (intra-articular therapeutics with MSC and/or autologous blood products are allowed). In addition, studies with any joint with naturally occurring OA were included in the systematic review; there were no specifications on a specific localization. Finally, all studies were evaluated based on the lameness examination and classified into either a positive or a negative outcome. The positive outcomes were divided into two groups: horses working at training level and horses returning to competition. Horses with a negative outcome did not respond to treatment or had a relapse during the observation period. For the meta-analysis, the two positive outcome groups (horses working at training level/lame free and horses returning to competition) were combined due to a lack of study numbers.

Each study in the systematic review was examined for the following 7 bias characteristics: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), other source of bias. Regarding each aspect, the studies were classified as high risk, low risk or unclear risk according to the PRISMA guidelines (93). For a better visualization, a traffic light table with “high risk” in red, “low risk” in green, and “unclear risk” in yellow was created. The classification into the category “unclear risk” occurs when relevant details for the classification into bias are not sufficiently substantiated in the respective study (94).

Using the PRISMA guidelines, 13 studies met the inclusion criteria for the systematic review (93). To compare dichotomous outcomes via meta-analysis, an odds ratio (OR) with 95% confidence interval (CI) was calculated using the R program (95). For data evaluation a random-effects model was used describing the overall outcome. Each study with its estimated effect size and corresponding confidence interval is graphically represented in the forest plot. Furthermore, the forest plot illustrates the extent to which the result from the individual study varies (96, 97). This variability is referred to as heterogeneity and is assessed by I² in the following meta-analysis. Heterogeneity was determined to be significant at I² > 50% or p < 0.1. A result was considered significant with p < 0.05.

In the meta-analysis, the results of lameness evaluation at different time periods of the studies were presented in the individual sections. In the short-term follow-up periods, one additional placebo-controlled and randomized trial was examined for better comparability (57). These will be discussed separately. All long-term studies are listed in the last row of the forest plot.

With all instruments that measure the risk of bias in clinical trials, it must be considered that they do not present an exact measurement method. Instead, it is an estimation in which the result always contains a subjective component. The purpose is to compare similar and homogenous treatment groups affected only by random variabilities (75).

All studies in the systemic review were assessed against the listed seven criteria and classified as high, low, or unclear risk (Table 2). The traffic light system (Figure 1) was used to illustrate the overall risk achieved by each study. Six studies avoided selection bias by randomly assigning participants (31, 75, 77, 86, 87, 89). Secrecy of the randomization scheme and blinding of veterinarians and patient owners was met by only two studies, both demonstrate a low risk of bias (31, 86). Blinding of treatment was achieved by generating two groups of examining and dispensing veterinarians at both study sites and owner's absence at administering the agent (86). For comparison, in the other study, the syringe was blinded so that owners and veterinarians did not know which treatment regime was selected. Blinding was maintained throughout the entire duration of the study (31).

All studies in the systematic review reported study discontinuations and missing outcome data. Therefore, almost all studies were considered to have a low risk of incomplete results and selective reporting. In addition, most studies used an owner questionnaire for long-term follow-up. In summary, eleven studies are at a high risk of bias (18, 32, 40, 75, 77, 87–92), due to the lack of blinding. The risk of bias graph shows the authors' assessment of each item in percentage (Figure 2). Overall, <25% of the studies included in the systematic review were found to be at a low risk of bias.

The flowchart (Figure 3) shows the detailed systematic analysis after initial electronic and manual research, with a total of 271 studies. This resulted in 28 studies being assessed for the qualitative synthesis after an initial review. These studies were further assigned to the defined orthobiologic therapies when treatment of equine OA was the scientific focus (Table 3). Following this, the biologic cell source of the selected 28 studies was assessed and listed (Table 4). This states, that based on their sources of orthobiologic therapeutics, 4 studies were included using blood-derived MSCs either non-induced (native) or chondrogenic-induced (40, 57, 77, 86), 5 studies focused on bone marrow-derived MSCs (18, 29, 99–101) from which 2 studies also included MSCs derived from adipose tissue for comparison (100, 101). Five studies included adipose tissue-derived MSCs only (27, 55, 100–102). The effect of umbilical cord-derived MSCs were studied in 2 publications (31, 61). Eight studies describe the use of cell-based autologous blood products as orthobiologic therapeutics (19, 32, 75, 89–91, 103, 104) and 6 studies a cell-free final therapeutic product (30, 81, 87, 88, 92, 105). The following results were obtained: 8 studies used MSCs as a therapeutic agent for naturally occurring OA (18, 31, 40, 55, 61, 77, 86, 102); 6 studies examined the effect of MSCs after inducing OA (27, 29, 57, 99–101); 14 studies treated with autologous blood products, with 1 study inducing OA (30) while the remaining studies examined naturally occurring OA (19, 32, 75, 81, 87–92, 103–105). Of the 14 MSC-related studies, 12 were placebo controlled (27, 29, 31, 55, 57, 61, 77, 86, 99–102) and 7 studies had an outcome with patient follow-up at least 6 months after treatment initiation (18, 27, 29, 31, 77, 86, 99). Comparatively, of the 14 groups treated with autologous blood products, 7 were placebo controlled (19, 30, 75, 87, 89, 92, 103) and 8 studies had a long-term follow-up (32, 75, 87–92). After screening the studies with the specified inclusion and exclusion criteria, 13 studies were examined for systematic analysis, and listed in Table 1 (18, 31, 32, 40, 75, 77, 86–92).

Table 5 lists all 15 studies from the quantitative synthesis that were not included in the systematic review due to the lack of information on the individual degree of lameness. Therefore, an average value was given for the whole group. Other reasons for exclusion were short observation periods, induced OA, or an overall too short observation time.

The age, sex, breed, and disposition of the horses selected for the investigations varied among the studies. Most of the trials in the systematic review examined the effect of orthobiologic therapeutics for the coffin or fetlock joint (Table 1). Five of the 13 studies treated naturally occurring OA with MSCs (18, 31, 40, 77, 86), and 1 study subdivided the treatment groups into f4 subgroups (PRP; native MSCs, native MSCs with PRP; chondrogenic-induced MSCs with PRP) (77). This study used allogenic peripheral blood as MSC source and labeled isolated, non-induced MSCs as “native.” Due to lack of placebo-controlled studies, the PRP-subgroup was compared with the MSC-subgroups of different sources in combination with PRP in the following meta-analysis. The remaining studies treated horses with non-induced (“native”) MSCs, chondrogenic-induced MSCs, umbilical cord-derived MSCs and bone marrow-derived MSCs. None of the 5 reviewed studies using adipose-derived MSCs as orthobiologic agent met the inclusion criteria for meta-analysis (27, 55, 100–102). In general, all MSC-studies demonstrated a heterogenous group regarding manufacturing and processing methods of the particular cell source. Considering possible side effects, 3 studies observed a mild to moderate inflammatory response after intra-articular treatment with MSCs (18, 31, 40). At the final examination, all patients felt well. Therefore, no general side effects were concluded.

Eight of the 13 studies examined treatment outcomes with autologous blood products. Of these, 3 studies used autologous conditioned serum (ACS) products (87, 88, 92). The remaining studies used cellular autologous blood products with a high platelet-rich content (32, 75, 89–91). Mild side effects such as self-limiting local swelling and lameness were noted in 1 ACS study (87).

Concerning the post-treatment, every study designed a particular rehabilitation program. All horses received a 1- (86, 89) to 8-week (31) hand-walking program at the end of treatment, followed by individual retraining. Most studies graded the severity of lameness according to the AAEP (American Association of Equine Practitioners) scoring system (18, 32, 75, 77, 86–90).

Although several placebo-controlled studies were included, most of them lack long-term follow-up or control was not maintained throughout the entire duration of the study. For example, in one RCT, horse owners in the placebo group were offered treatment with autologous protein solution (APS) 14 days after the placebo treatment. The randomized controlled study was well-structured, but the observation time of the control group was too short to be included in our meta-analysis. The APS group improved significantly after the treatment compared with baseline or control group scores (75). In total, 5 studies were placebo controlled over the entire observation period, 4 of which were randomized (77, 86, 87, 89). These studies have also been included in the meta-analysis.

In summary, an average of 65% improvement in lameness grade was achieved after the treatment with intra-articular applied orthobiologic therapeutics, regardless of which therapeutic agent was used (Figure 4). Eleven studies showed a general positive effect after treatment, with horses working at trainings level or horses returning to competition (Table 1). Two outliers could be detected, that showed initial improvement in the first 7–8 months after treatment but then returned to their initial degree of lameness (32, 89). In both studies, the majority of horses responded positively at the beginning and maintained their high level of performance over a period of at least 6 months. Furthermore, horses showed no adverse reactions. This outcome suggests that platelet lysate (PL) and autologous platelet concentrate (PC) can be an efficient short-term therapy for horses suffering from OA (32, 89) (Figure 5). Looking at the average proportions without outliers, 80% of the horses involved in the studies showed lameness reduction after treatment with orthobiologic therapies (Figure 5).

As previously stated, in 1 study treatment groups were divided into 4 subgroups, and interestingly, the group treated with chondrogenic induced MSCs had the most successful result, with 80% lameness-free horses and horses working at trainings level (77). Another pilot study compared non-induced (native) MSCs with chondrogenic induced MSCs, both in combination with PRP. This resulted in a higher average score for the beneficial effects using chondrogenic induced MSCs. However, the result was statistically non-significant (40).

Promising results with MSCs and ACS were justified over a 24-months follow-up period (18, 88). After intra-articular administration of MSCs postoperatively after arthroscopy of the stifle, 42% of horses returned to their previous level of work, and 33% returned to work after a mean follow-up period of 24 months (18). In a retrospective study from Warner et al. 31% of the horses returned to their previous level of work and 15% performed at exercise level after a period of at least 2 years following the ACS treatment of the coffin joint (88). Both studies were not blinded and without a control group, which significantly limits their validity. However, both studies provide indication of a long-term effect of MSC and ACS treatment in OA.

Four RCTs (77, 86, 87, 89) out of the 13 trials were included in the long-term meta-analysis with a follow-up time >6 months. The control groups were treated with saline (86, 89), other orthobiologic therapeutic agents (PRP) (77) or corticosteroids and hyaluronic acid (87). All studies included horses of different breed, sex, age, and level of performance. Moreover, the diagnosed and treated OA occurred in different joints, ranging from low to high motion joints (Table 1). Due to the scarcity of studies, no restrictions were made.

Figure 6 demonstrates a forest plot with outcomes at different time points. The focus was the set inclusion criteria of a follow-up period >6 months. Three studies (77, 86, 87) showed a positive impact of orthobiologic therapeutics compared to their control groups.

One study reported a regression to its initial lameness level after an observation period of 1 year (89). In this study, no side effects were noted in the first 6 months after treatment and 9 out of 10 horses treated with PL returned to their normal activity. Lameness recurred from the 7th month, and all horses relapsed to their initial degree of lameness at the end of the study period. This study illustrates the correlation between duration of follow-up and recurrence of lameness. Within 6 months, horses returned to their previous level of performance. However, all horses relapsed to their initial degree of lameness, therefore only a temporary positive effect could be observed.

The forest plot illustrates the common effect model and the random effects model and whether heterogeneity could be stated as significant. Data demonstrating I² > 50% were assigned to the random effects model. As demonstrated in the last row of the plot, all long-term follow-up studies showed moderate heterogeneity with I² = 55% and p = 0.11 (Figure 6). A random effects model was used due to the assumption of moderate differences among study design and implementation in the clinical studies.

An odds ratio (OR) of 1 indicates no difference between the treatment and control group, whereas an OR > 1 indicates that lameness is more likely to be reduced in the experimental group. All studies with an OR values higher than 1 favor the experimental group (Figure 6: OR 17.02; 95% CI: 8.5474 to 33.8849 p < 0.0001). None of the studies crossed the line into ineffectiveness, suggesting that the treatment effect was estimated to be similar across studies.

The diamond square represents the average of all individual studies. If the limit of ineffectiveness is not exceeded, a significant difference in lameness reduction between the experimental and control groups is stated. It can be summarized, that the included orthobiologic therapeutics are safe and showed significant improvement in lameness reduction compared to their control groups. Three studies (77, 86, 87) showed a long and constant improvement over 6 months.

One short-term RCT was included in the forest plot for comparative reasons (57). We assessed at what time point the trial showed significance for treatment with an orthobiologic therapy and how effective the short-term trial was. Treatment success with chondrogenic induced MSCs in an induced OA model was demonstrated to be a time dependent factor, with decreasing lameness levels from 2 weeks after treatment throughout the observation period of 11 weeks (57).

In summary, the use of intra-articular administered orthobiologic therapeutic agents show an incidence of lameness reduction by 73% compared to the control in the long-term follow-up, whereas in the control group lameness was reduced by 17% (77, 86, 87, 89). According to the included studies, horses with naturally occurring OA demonstrated a significantly reduced degree of lameness after intra-articular treatment with orthobiologic therapeutics compared with the control in the long-term follow-up.

Publication bias occurs when the probability of a study being published depends significantly on its outcome. This means that it is more likely, that a study will be published if the results are consistent with the hypothesis or if the study results are significant (106).

The occurrence of publication bias can be tested by creating a funnel plot. Ideally, the individual data points form a symmetrical, inverted funnel. On the x-axis, the treatment effect is plotted against the study size on the y-axis. The largest studies are located at the top of the graph and plotted near the average. The smaller studies are distributed on both sides of the average and lie close to the x-axis.

The funnel plot showed almost the desired symmetrical shape, with the studies close to the midline. It is important to note, that studies that conducted lameness examinations at different time points are considered as individual studies. For instance, a study by Broeckx et al. (57) is plotted 4 times in the funnel plot, at each study time point. The hypothesis that studies with a smaller number of participants are more likely to be in the bottom range is correct. Overall, the publication bias can be classified as low (Figure 7).