Injection of Adipose-Derived Stromal Cells in the Knee of Patients with Severe Osteoarthritis has a Systemic Effect and Promotes an Anti-Inflammatory Phenotype of Circulating Immune Cells.

Rationale: Recent studies confirmed that osteoarthritis (OA) is associated with systemic inflammation. Adipose-derived stromal cells (ASCs) could become the most promising cell-based therapy in OA, based not only on their differentiation capacities and trophic and paracrine effects on the existing cartilage, but also on their immunomodulatory properties. Here, we wanted to determine the biological effect of autologous ASC intra-articular (IA) injection. Method: To this aim, we monitored the profile of immune cells in fresh peripheral blood after IA injection of autologous ASCs in the knee of 18 patients with severe OA (ADIPOA phase I study). Specifically, we used 8-color flow cytometry antibody panels to characterize the frequencies of innate and adaptive immune cell subsets (monocytes, dendritic cells, regulatory T cells and B cells) in blood samples at baseline (before injection) and one week, one month and three months after ASC injection.
Results: We found that the percentage of CD4+CD25highCD127lowFOXP3+ regulatory T cells was significantly increased at 1 month after ASC injection, and this effect persisted for at least 3 months. Moreover, CD24highCD38high transitional B cells also were increased, whereas the percentage of classical CD14+ monocytes was decreased, at 3 months after ASC injection. These results suggest a global switch toward regulatory immune cells following IA injection of ASCs, underscoring the safety of ASC-based therapy. We did not find any correlation between the scores for the Visual Analogic Scale for pain, the Western Ontario and McMaster Universities Osteoarthritis Index (pain subscale and total score) at baseline and the immune cell profile changes, but this could be due to the small number of analyzed patients.
Conclusion: ASCs may drive an immediate local response by releasing paracrine factors and cytokines, and our results suggest that ASCs could also initiate a cascade resulting in a long-lasting systemic immune modulation.

Introduction

Osteoarthritis (OA), the most prevalent form of arthritis, affects up to 15% of the adult population. This joint disease is characterized by degradation of the articular cartilage, synovitis and osteoclast activation in the subchondral bone 1. Chronic, low-grade inflammation contributes to the symptoms and to disease progression. Indeed, recent studies suggest that OA is not only a local disease of the joint(s), and that is also associated with systemic disorders, such as inflammation, metabolic dysregulation and obesity 2,3. For instance, the knee cartilage volume is negatively correlated with the concentration of circulating inflammatory cytokines, such as interleukin 6 (IL-6), tumor necrosis factor alpha (TNF-α) and C-reactive protein (CRP). Additionally, some inflammatory mediators (chemokines, and cytokines, such as IL-1β and alarmins) affect chondrocyte viability and enhance chondrocyte hypertrophy. A moderate increase of CRP and of some inflammatory mediators (PGE2, IL-1, IL-6) could play a key role in the development of complications in patients with OA, such as cardiovascular damage 4. Indeed, patients with OA have a higher overall mortality rate compared with the general population, mainly due to increased cardiovascular events 5. For all these reasons, OA should be considered not only as a local articular disease, but as a systemic disease.

Mesenchymal stromal or stem cells (MSCs) are multipotent cells that can be isolated from a variety of adult and neonatal tissues, such as bone marrow (BM-MSCs), fat tissue (adipose stem cells, ASCs), placenta and umbilical cord. MSC cell-based therapy in OA is justified by their differentiation capacities and also their paracrine and immunosuppressive properties. MSCs can regulate immune responses and are used to alleviate immune disorders 6. In vitro and in vivo preclinical experiments suggest that MSCs can regulate the activity of many immune cell types, such as T cells, B cells, dendritic cells (DC), macrophages, neutrophils, and natural killer cells 7-11. In clinical settings, MSC-based therapies have been successfully used to reverse graft-versus-host disease (GvHD) in patients receiving allogeneic bone marrow transplantation 12-15. More recently, it has been reported that MSCs can suppress inflammation and reduce tissue damage through the induction of regulatory T (Treg) cells in patients with autoimmune diseases, such as systemic lupus erythematosus 16,17 and Crohn disease 18-20.

At the beginning, it was thought that MSC beneficial effects were mainly explained by their engraftment and tissue regeneration; however, it is now widely accepted that the main MSC therapeutic effects are mediated primarily through the short-term secretion of trophic factors that reduce inflammation and modulate immune cells. Despite a large body of in vitro experimental studies on MSC effects on immune cells, little is known about the biological mechanisms underlying MSC-mediated inhibition of the immune response in vivo.

The goal of this study was to determine the impact of MSC-based therapy on the circulating immune cell profile. To this aim, we used multiparametric flow cytometry to prospectively quantify the different immune cells in fresh whole blood samples collected before and at different time points after therapy. Blood samples were from the 18 patients with severe OA who took part in the ADIPOA Phase I clinical trial. These patients received one intra-articular (IA) injection of autologous ASCs and were then followed for 6 months. This ASC-based therapy showed a good safety profile with preliminary evidence of efficacy on OA-related symptoms 21.

Materials and Methods

Patients and healthy controls

The ADIPOA1 study design was previously described 22. Briefly, it was a phase I, prospective, bi-center (Montpellier University Hospital Center, France, and Orthopedic Department of Würzburg University, Germany), single-arm, open-label study conducted between March 2012 and April 2014. Eighteen consecutive patients with symptomatic, severe knee OA were enrolled. The study was performed without a control arm and involved three different groups (6 patients/each) who received one single IA injection of autologous ASCs at low (2 x 106 cells), medium (10 x 106 cells), or high dose (50 x 106 cells). Autologous ASCs were prepared by a single Good manufacturing practice-certified facility (Etablissement Français du Sang Midi-Pyrénées, France) as described elsewhere 23. Peripheral blood samples were collected during the scheduled visits at baseline (D0), and then one week (D7), one month (M1) and three months (M3) after ASC injection. Pain, stiffness and physical functions before/after treatment were monitored with different scales during the trial, among which the Western Ontario and McMaster University Arthritis Index (WOMAC) 24 and a Visual analog scale (VAS) for pain. The WOMAC total score and pain sub-score and the VAS score were used in the present study for the correlation analyses.

Patients were classified in responders (n=4) and non-responders (n=12) to the ASC-based treatment using the responder criteria index of the OsteoArthritis Research Society International)/ Outcome Measures in Rheumatology (OARSI-OMERACT) 25. Fresh peripheral blood was obtained from healthy age and sex matched volunteers with no history or current pain due to OA (n=6).

Reagents

The FITC-conjugated anti-CD64 (10.1), phycoerythrin (PE)-conjugated anti-CD33 (WM53), PerCP-Cy5.5-conjugated anti-HLA-DR (L243), PE-CyTM7-conjugated anti-CD56 (B159), PE-CyTM7-conjugated anti-CD14 (M5E2), allophycocyanine (APC)-conjugated anti-CD32 (FL18.26), APC-H7-conjugated anti-CD4 (RPA-T4), APC-H7-conjugated anti-CD16 (3G8), V450-conjugated anti-CD3 (UCHT1), V500-conjugated anti-CD45 (HI30), PerCP-CyTM5.5- conjugated anti-CD123 (7G3), PE-CyTM7-conjugated anti-CD45RO (UCHL1), APC-H7- conjugated anti-HLA-DR (L243), V450-conjugated anti-CD11c (B-ly6), V500-conjugated anti-CD4 (RPA-T4), FITC-conjugated anti-CD24 (ML5), PE-conjugated anti-IgD (IA6-2), PerCP-Cy5.5-conjugated anti-CD138 (MI15), PE-CyTM7-conjugated anti-CD38 (HIT2), APC-conjugated anti-CD27 (M-T271), APC-H7-conjugated anti-CD20 (L27), V450-conjugated anti-CD5 (L17F12), AmCyan-conjugated anti-CD19 (SJ25C1), FITC-conjugated anti-CD62L (DREG-56), PE-conjugated anti-CD127 (HIL-7R-M21), PerCP-CyTM5.5-conjugated anti-CD25 (M-A251), V450-conjugated anti-CD4 (RPA-T4), APC-conjugated anti-CD152 (BNI3) antibodies and FACS Lysing solution were purchased from BD Biosciences. The PE-conjugated anti-CD314 (BAT221), PE-conjugated anti-CD1c (AD5-8E7) and APC-conjugated anti-CD303 (AC144) antibodies were purchased from Miltenyi Biotech. The FITC-conjugated anti-human lineage cocktail 1 (lin1) (CD3 clone SK7, CD14 clone Mop9, CD16 clone 3G8, CD19 clone SJ25C1, CD20 clone L27, CD56 clone NCAM16.2) was from Beckmann Coulter and the PE-Cyanine7-conjugated anti-FOXP3 (PCH101) antibody was from eBioscience.

Flow Cytometric Analysis

Samples were incubated with antibodies within 20 hours and more precisely within 4 hours for Montpellier's patients and 20 hours for Würzburg's patients, after blood collection. The characterization of monocytes, DCs, T and B cells were performed on whole blood cells and freshly prepared peripheral blood mononuclear cells (PBMC) were used for the identification of FOXP3+ Treg cells. All the frequencies analyzed in our study were found to be stable over the 20 hours since no significant variation was observed between patients from the 2 clinical centers. Cells in fresh EDTA-treated whole blood samples were labeled using three 8-color flow cytometry panels of conjugated antibodies to quantify the different subsets of monocytes (CD64, CD33, CD314, HLA-DR, CD56, CD14, CD32, CD4, CD16, CD3, CD45), DCs (lin1, CD1c, CD123, CD45RO, CD303, HLA-DR, CD11c, CD4) and B cells (CD24, IgD, CD138, CD38, CD27, CD20, CD5, CD19). All fluorochrome-conjugated antibodies were used according to the manufacturer's recommendations. Following incubation with antibodies, red blood cells were lysed and cells were washed twice (PBS containing 2% fetal calf serum and 0.1% sodium azide) prior to acquisition on a BD FACSCanto II flow cytometer (BD Biosciences, San Jose, CA). In parallel, PBMCs were isolated through a Ficoll Hypaque gradient. Then, 0.5 million of PBMCs were incubated with antibodies against cell surface markers (CD62L, CD127, CD25, HLA-DR, CD4, CD45) for the characterization of Treg cells. Intracellular CD152 and FOXP3 staining was performed after cell fixation and permeabilization using the FoxP3 Staining Buffer Set (eBioscience). Application settings were used to standardize the flow cytometer during the clinical trial. All data acquired with one cytometry panel were analyzed by the same analyst using the BD FACSDiva software. Representative dot plots are shown using the FlowJo software (TreeStar).

Statistical analysis

A general description of the population under study was performed using frequencies for qualitative variables and means with standard deviation and minimal and maximal values for quantitative variables. Analyses of the quantitative changes in the immune cell populations according to the responder criteria index of the OARSI-OMERACT was analyzed using the Mann Whitney test or one-way ANOVA, as applicable. Unilinear correlations were used to examine the link between clinical symptoms (WOMAC and VAS pain scores) and immune status at baseline, and also to correlate changes in clinical symptoms and immune subpopulations during the follow-up. Pearson correlation coefficients were obtained assuming a Gaussian distribution of the data. Responders and non-responders were compared using the Student's t test. All FACS analysis data are presented as the mean ± standard error of the mean. The effect of ASC injection on the various immune cell subsets was analyzed using Wilcoxon matched-pairs test and the significance level was set at 5% for all tests. Analyses were performed with Prism version 6.0c (GraphPad Software Inc., La Jolla, CA, USA).

Results

Changes in Peripheral Innate Immune Cells after ASC Injection in the Knee

As the handling of blood samples has a large impact on cytometry data and based on the experience of other consortia, such as EuroFlow 26, Human Immunology Project 27, and Milieu Interieur 28, cytometry experiment was performed using fresh whole blood samples to avoid variability induced by freeze/thaw cycles, especially in monocyte and DC populations.

For DC characterization, cells were plotted according to their size and granularity followed by doublet exclusion. Lineage-negative cells were first selected, and then CD4+HLA-DR+ double positive cells were gated to detect the two major DC subsets: the myeloid subset (mDC: HLA-DR++CD11c++CD1c+CD123low) and the plasmacytoid subset (pDC: HLA-DR++CD11clowCD123high) (Figures ). The phenotype of both populations was confirmed by labeling with anti-CD303 and -CD45RO and -HLA-DR antibodies (Figures ). The percentage of the mDC (52.2 ± 2.1% at day 0 and 49.8 ± 2.2% at 3 months) and pDC subsets (35.2 ± 2.12% at day 0 and 36.3 ± 2.7% at 3 months) was not affected by ASC injection (Figures ). These results emphasized the reproducibility of our experimental procedure for immune cell quantification and suggest that no alteration of major DC subsets could be monitored following autologous ASCs injection.

For monocyte analysis, debris and doublets were excluded followed by selection of CD45+ leucocytes. Contaminating neutrophils, granulocytes and T cells were excluded from CD45+ monocytic cells. At baseline (before injection), classical (CD14++CD16-), intermediate (CD14+CD16+) and patrolling (also known as non-classical) monocytes (CD14dimCD16+) represented, on average, 82.1 ± 1.2%, 4.8 ± 0.7% and 7.5 ± 0.8%, respectively, of all monocytes (Figure ). After ASC injection, the percentage of classical monocytes was significantly reduced (82.1 ± 1.2% at day 0 versus 79.3 ± 1.6% at 3 months, P=0.0335), but not that of intermediate (4.8 ± 0.7% at day 0 versus 5.9 ± 0.9% at 3 months, P=0.0654) and patrolling monocytes (7.5 ± 0.9% at day 0 versus 8.6 ± 0.9% at 3 months, P=0.1167) (Figures ). This result suggests that ASC injection slightly modulates the distribution of monocyte subsets with a higher impact on classical monocyte.

The classification of monocytes in three groups has shifted attention to the mobilization of non-classical monocytes during inflammatory diseases and on their inflammatory characteristics, particularly the secretion of important inflammatory cytokines (i.e., TNF-α), and supports a more significant role for intermediate monocytes in inflammation 29. In agreement, in our 18 patients, the percentage of classical monocytes was inversely correlated with CRP level (r = -0.39; 95% CI -0.59; -0.14; p=0.0029). Conversely, the percentages of intermediate monocytes (r = 0.46; 95% CI 0.22; 0.64; p=0.0003), and to a lower extent, of patrolling monocytes (r = 0.50; 95% CI -0.01; 0.80; p=0.049) were positively correlated with CRP levels.

Changes in Peripheral Adaptive Immune Cells after ASC Injection in the Knee

B cell characterization was performed using whole blood cell samples. Cells were plotted according to their size and granularity followed by doublets exclusion. Within the CD19+CD20+ B cell population, CD27 and IgD were used to identify switched (CD27+IgD-) and un-switched memory B cells (CD27+IgD+) (Figure ). ASC injection did not significantly change the percentage of CD27+ (23.6 ± 3.1% at day 0 versus 26.9 ± 3.4% at 3 months, Figure ), CD27+ IgD- switched (16.1±2.6% at day 0 versus 18.1±3.1% at 3 months, not significant) and CD27+IgD+ un-switched memory B cells. Similarly, within the subset of CD27- cells, no variation in the percentage of naïve B cells (CD27-IgD+) was observed (90.5 ± 1.5 % at day 0 versus 90.5 ± 2.0% at 3 months, not significant). Conversely, transitional B cells (CD24highCD38high, Figure ), slightly increased after ASC injection (2.5 ± 0.4% at day 0 versus 3.0 ± 0.3% at 3 months, not significant) (Figure ).

Treg characterization was performed using freshly prepared PBMCs. First, lymphocytes were gated according to their size and granularity followed by doublet exclusion. Then, CD45+ lymphocytes were selected based on CD4 expression, and Treg cells were defined as the FOXP3+ subset among all CD25highCD127low cells within the CD4+ population (Figures ). The percentage of CD25highCD127lowFOXP3+ Treg cells was significantly increased at 1 month after ASC injection (4.2 ± 0.3% at day 0 to 5.1 ± 0.3% at 1 month, P=0.0151), and this effect was still observed at 3 months post-treatment (5.4 ± 0.4% at 3 months, P=0.0054) (Figure ). Analysis of the fraction of CD25highCD127lowFOXP3+ Treg cells in healthy controls (4.2±0.5% for controls, n=6) did not show any significant difference with OA patients. Therefore, Treg cells are significantly increased in OA patients after one single ASC injection in the knee, and this effect persists at least for 3 months.

Correlations between Clinical Responses and Frequencies of Immune Cells Subsets

During the ADIPOA1 trial, the different ASC concentrations all led to symptom decrease; however, only patients in the lowest dose group reported a significant improvement in pain and physical functions compared with baseline, up to 6 month after ASC injection 22. To examine whether pain and physical function, were correlated with a specific immune cell signature at baseline, multiple regression analyses were performed using the baseline scores of the VAS pain, WOMAC pain subscale, WOMAC total score, and the baseline values for the Treg, monocyte, and B cell populations (Table 1). No significant correlation between clinical and cellular data was obtained. Similarly, no correlation was found between changes in the pain (VAS) and functional (WOMAC) scores and changes in the different cell subsets between baseline and month 3 post-injection (Table 2). Moreover, analysis of the quantitative changes in the immune cell populations according to the responder criteria index of OARSI-OMERACT revealed no association between the clinical response at month 3 post-ASC injection and Treg changes. Finally, analysis of the relative changes in the immune subsets in function of the number of injected ASCs (low, medium and high dose) also did not highlight any difference (data not shown).

Table 1

Correlation between clinical features and immune cell subsets at baseline

	Correlation (r)	95% CI	p
WOMAC Total
vs. FOXP3+	0.32	-0.23; 0.71	0.24
vs. Tregs	-0.05	-0.54; 0.47	0.85
vs. Classical monocytes	0.06	-0.46; 0.55	0.82
vs. Intermediate monocytes	-0.39	-0.74; 0.13	0.12
vs. Transitional B cells	0.19	-0.35; 0.64	0.19
WOMAC Pain
vs. FOXP3+	0.25	-0.30; 0.67	0.35
vs. Tregs	-0.29	-0.70; 0.26	0.28
vs. Classical monocytes	0.03	-0.49; 0.53	0.92
vs. Intermediate monocytes	-0.29	-0.68; 0.24	0.26
vs. Transitional B cells	0.43	-0.08; 0.76	0.09
VAS Pain
vs. foxp3+	0.16	-0.37; 0.62	0.54
vs. Tregs	-0.09	-0.57; 0.43	0.73
vs. Classical monocytes	0.01	-0.50; 0.51	0.97
vs. Intermediate monocytes	-0.39	-0.74; 0.13	0.13
vs. Transitional B cells	0.19	-0.35; 0.64	0.48

VAS: visual analog scale; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index

Table 2

Correlation between absolute changes in clinical features and immune cell subsets between baseline (D0) and 3 months (M3) post-injection

	Correlation (r)	95% CI	p
∆ WOMAC Total (M3-D0)
vs. ∆ FOXP3+	0.33	-0.21; 0.72	0.21
vs. ∆ Tregs	0.48	-0.04; 0.80	0.06
vs. ∆ Classical monocytes	0.07	-0.46; 0.55	0.81
vs. ∆ Intermediate monocytes	-0.01	-0.51; 0.50	0.98
vs. ∆ Transitional B cells	-0.25	-0.67; 0.23	0.35
∆ WOMAC Pain (M3-D0)
vs. ∆ FOXP3+	0.21	-0.33; 0.65	0.44
vs. ∆ Tregs	0.36	-0.18; 0.73	0.17
vs. ∆ Classical monocytes	0.27	-0.27; 0.69	0.30
vs. ∆ Intermediate monocytes	-0.16	-0.62; 0.38	0.55
vs. ∆ Transitional B cells	-0.16	-0.62; 0.38	0.55
∆ VAS Pain (M3-D0)
vs. ∆ FOXP3+	0.05	-0.47; 0.55	0.85
vs. ∆ Tregs	0.21	-0.34; 0.65	0.44
vs. ∆ Classical monocytes	-0.02	-0.52; 0.49	0.95
vs. ∆ Intermediate monocytes	-0.33	-0.70; 0.18	0.18
vs. ∆ Transitional B cells	-0.07	-0.54; 0.44	0.79

∆: absolute change between baseline and month 3; VAS: visual analog scale; WOMAC: Western Ontario and McMaster Universities Osteoarthritis Index.

Discussion

Our study on the immune monitoring of 18 patients with severe OA after local treatment by autologous ASC injection in the knee found slight but significant changes in circulating immune cells, suggesting a systemic effect that contributed to the good tolerance of this MSC-based therapy. Specifically, the significant increase in the fraction of Treg and the significant reduction of the percentage of classical CD14+ monocytes at 3 months post-IA injection suggest a long-lasting immune modulation, and a global switch toward a more regulatory immune cell profile. These significant changes are also associated with a slight increase in the transitional B cells population which show regulatory capacity 30. No such modification of the frequencies of these 3 populations of interest were monitored within 1 month in untreated OA patients, suggesting that the observed significant increase in Treg and significant decrease in classical monocytes are not due to the normal evolution of the disease (personal unpublished data).

It is thought that ASCs act as coordinators of the immune system by providing a tolerogenic environment for both the innate and the adaptive immune response 34. Emerging evidence has demonstrated the impact of ASCs on key cell types involved in the continuum between innate and adaptive immunity, and in modulating inflammation in acute organ injury, such as acute GvHD, organ transplantation, diabetes, multiple sclerosis and Crohn's disease. However, only few studies have monitored immune cell changes by flow cytometry after MSC-based therapy, particularly local treatment. BM-MSCs strongly inhibit the in vivo activation of T lymphocytes that are particularly involved in organ rejection. For example, BM-MSCs decrease the Th1 response in patients with GvHD 35 or autoimmune diseases, such as systemic lupus erythematosus (SLE). In patients with severe and treatment-refractory SLE, umbilical cord MSC transplantation resulted in a significantly increase of CD4+FOXP3+ Treg cells at month 3 and 6 after infusion and in a strong decrease in the secretion of autoantibodies, especially against dsDNA, suggesting a possible impact on B cell function 16. Dander at al. confirmed these results in patients with GvHD who received multiple infusions of BM-MSCs 37. They found that the T cell fraction was not significantly affected by BM-MSC infusion at different time points over 6 months. Conversely, Treg cell percentage was increased and concomitantly the Th1/Treg and Th17/Treg ratios were decreased, particularly in patients with a better response. These ratios were significantly higher in non-responders before and after BM-MSC infusion. Similar MSC-mediated effects on Th17 cell function and differentiation were also reported following injection of third-party MSCs after allogeneic hematopoietic stem cell transplantation 36. A recent study compared the percentage of CD4+CD25+FOXP3+ Tregs after MSC injection in patients with GvHD and found a significant increase from week 8 post-injection and at least up to month 36 38. Conversely, they did not observe any change in CD19+ B cells. Similarly, monitoring of circulating T lymphocytes after local MSC injection in patients with fistula in inflammatory bowel disease showed that the percentage of CD4+CD25brightFOXP3+ cells was significantly increased after the second injection (1 month) and persisted for 12 months 18. Altogether these data and our results indicate that the circulating Treg cell fraction is increased upon systemic or local MSC injection with a higher Treg increase following systemic injection.

Moreover, the absence of systemic immune cell depletion, which is observed with anti-inflammatory drugs and Janus kinase inhibitors (a new emerging therapy) 31,32, consolidates the safety of autologous ASCs. A recent systematic literature analysis of the reported adverse events upon IA injection of culture-expanded stem cells 33 indicated that there is no compelling argument against its use in human patients.

Does the observed slight but significant increase in Treg frequency is physiologically relevant? To discuss this point, it is interesting to analyze the variation of Treg frequencies in autoimmune settings. Indeed, break of tolerance in autoimmune diseases have been often associated with a decrease either in number or function of Treg cells. In murine model of autoimmune diseases, increased Treg cell number lead to protection of mice. For instance, injection of anti-CD3 antibodies in NOD mice protect mice from diabetes with an increase in Treg cells from 10 to 14% in the pancreatic and mesentheric LN representing a 1.4-fold increase 39. In human, in newly diagnosed immune-mediated diabetes as well as long-standing type 1 diabetic patients, the mean percentage of the CD4+CD25+Treg cells was significantly reduced to 2.6 ± 0.23% and 3.7 ± 0.69%, respectively, compared with 6.9 ± 0.4% and 6.3 ± 0.48% in the normal control and type 2 diabetic groups, respectively 40. These results suggest that less than a 2-fold decrease in Treg cell frequency may have important consequences on the immune homeostasis. In rheumatoid arthritis (RA), results concerning the decreased number of peripheral Treg cells are very controversial because of the heterogeneity of the definition of Treg cells (reviewed in 41). Recent studies showed that the frequencies of Treg cells were significantly lower in both active RA patients (2.89 ± 0.17%) and inactive RA patients (3.21 ± 0.21%) than those in healthy controls (5.83 ± 0.39%) 42. Another study found that the proportion of Treg cells was significantly lower in RA (3.8 ± 1.0%) and Behcet's Disease (3.3 ± 0.5%) compared to heathy controls (5.0 ± 0.5%) with a specific significant decrease in activated Treg 43. Furthermore, in RA, clinical remission following biotherapy treatment using Infliximab has been associated with an increase in Treg number from 2% to 3% at 6 weeks representing a 1.5-fold increase 44. All these results suggest that slight variation in Treg number, even less than 2-fold, may have important consequences on immune homeostasis suggesting that the 1.3-fold increase in Treg cell number, observed in our study, may be physiologically relevant.

Concerning the results obtained for monocytes and B cells, it should be noted that both cell types have shown high intra-assay variability (up to 20% in some cases) among longitudinal studies and among centers 45. Several studies have previously reported an impact of MSCs on B cell behavior in vitro, and recently, Peng et al found an increase in the percentage of IL-10-secreting CD5+ B cells in patients with GvHD upon MSC infusion 46. The clinical improvement after MSC infusion was accompanied by a 1.5-fold increase in CD5+ B cells underscoring that thigh variation of such B cell population may have physiological impact. Concerning the variation in the monocyte subsets, it may be informative to follow the frequencies of such populations in several clinical situations at base line and after treatment. Similar slight but significant frequencies alterations of the various monocyte subsets have been demonstrated 47,48. Decreased proportions of circulating monocytes to leucocytes was recently observed in OA compared with controls mainly due to a decrease in classical monocytes and a higher number of circulating lymphocytes 49. Moreover, changes in monocyte frequencies, phenotype and function occur in the advanced-age that correlate with chronic diseases 50. However, we still do not know if these slight but measurable changes predispose individuals to age-related chronic inflammatory disease.

Finally, we did not find in our study any correlation between baseline clinical features and immune cell profile changes post-treatment. This might be due to the low number of patients that did not allow to do a correlation between the clinical response and the different ASC doses. Our results suggest that the increased immune tolerance after ASC injection is mainly an indication of a good ASC safety profile. Thus, we did not observe any association between increasing Treg cells frequency and a better clinical response although it might be expected by the anti-inflammatory action mediated by ASCs. The present results on the immune monitoring of the 18 patients included in the ADIPOA phase I study need to be consolidated in a large clinical trial to evaluate the immunological impact of MSC-based therapy with long term monitoring, including a placebo-treated group. In the ongoing ADIPOA phase IIb clinical trial (randomized, placebo-controlled; n=150 patients) with two ASC doses, ASC efficacy in OA is evaluated using standardized and established outcome scores. Immune-monitoring of this phase IIb clinical trial, as well as in other immune disorders will allow deciphering MSC long-term biological and immunological impact in various clinical settings.

In conclusion, the present study reveals a global switch of the immune response toward immune homeostasis, following local ASC injection. It is acknowledged that ASCs drive the immediate response by releasing paracrine factors and cytokines, and might also initiate a cascade resulting in a long-term effect on the immune homeostasis. More data on the in vivo immune cell profile changes upon ASC injection could help to improve our understanding of stromal cell mechanisms of action and to better design future clinical studies.