Osteoarthritis (OA) is a chronic disease involving progressive degeneration of the articular cartilage and subchondral bone, accompanied by synovitis [1]. Due to its avascular nature and the limited self-renewal capacity of chondrocytes, adult articular cartilage presents limited repair capability [2]. Current treatment options for articular cartilage injury and osteoarthritis are aimed to relieve inflammation and pain, but have no effect on the natural progression of the disease [3]. To date, in severe cases of knee OA, knee replacement is the only therapeutic option [4].

During the last two decades focal cartilage defects have been treated using cell therapy and tissue engineering approaches. In this context, autologous chondrocyte implantation (ACI) or matrix-induced autologous chondrocyte (MACI) implantation techniques have been applied with promising results, although the large non-contained cartilage defects found in OA and its own pathogenesis cannot be treated using ACI or MACI [57]. The use of intraarticular injections of mesenchymal stromal cells (MSCs) may represent some advantages over chondrocytes in patients with OA. First, because of their ability for self-renewal, the number of cells that can be obtained is increased without cartilage donor site morbidity and with reduced cost [68]. Second, MSCs are responsible for the normal turnover and maintenance of adult mesenchymal tissues, including cartilage, and has been suggested that the number of MSCs present in the subchondral bone decreases with age and OA grade, suggesting that such MSCs deficit could prime the degenerative process [912]. It has also been proposed that during tissue injury MSCs migrate to participate in the reparative process, giving MSCs a potential therapeutic value when added exogenously [13, 14]. Additionally, cultured MSCs induce in vitro chondrocyte proliferation and extracellular matrix protein synthesis, including aggrecan and type II collagen, which support their critical role in cartilage tissue repair [15, 16].

There is an increasing number of reports on the treatment of OA using MSC, but these are methodologically heterogeneous in dose, cell source, coadjuvants and cell processing methods, which makes it difficult to compare the different studies [17]. In many cases, treatments consist of the administration of bone marrow concentrates as a source of MSCs. However, it is well known that only 0.001 % of the mononuclear cells found in the bone marrow could be considered as MSCs as defined by the ICRS in 2006. Therefore, their number in a bone marrow concentrate is very limited compared to that obtained upon culturing MSCs [1820]. Only a few studies using MSCs produced by good manufacture practices (GMP) such as advanced cell-therapy products have been reported [2124]. In addition, there is a need to explore the effect of different cell doses in a randomized way to gain insight into the ideal conditions for knee OA patients to take advantage of MSC therapy. For these reasons, the purpose of this study was to randomly assess the safety, feasibility and efficacy of the intra-articular injection of two different doses of GMP-produced autologous bone marrow MSCs (BM-MSCs) with hyaluronic acid (HA) in patients with knee OA.


Participants and study design

This is a phase I/II randomized clinical trial with active control conducted between August 2012 and October 2014, involving the Clínica Universidad de Navarra (Pamplona, Spain) and IBSAL-Hospital Universitario de Salamanca (Salamanca, Spain). All the procedures were approved by the Institutional Review Board of Navarra and the Spanish Agency of Medicines and Medical Devices (Nº EudraCT: 2009-017624-72, Clinical identifier: NCT02123368). All participants provided written informed consent.

Criteria for eligibility of patients

Inclusion criteria were as follows: males and females aged 50–80, diagnosis of knee OA according to American College of Rheumatology criteria, visual analogue scale (VAS) joint pain ≥2.5, Kellgren–Lawrence radiological classification scale ≥2, body mass index between 20 and 35 kg/m [2], and availability to be followed during the study period; exclusion criteria were: previous diagnosis of polyarticular disease, severe mechanical extra-articular deformation (>15° varus/15° valgus), systemic autoimmune rheumatic disease, arthroscopy or intraarticular infiltration in the last 6 months, chronic treatment with immunosuppressive or anticoagulant drugs, corticosteroids treatment in the 3 last months, nonsteroidal anti-inflammatory drugs therapy in the last 15 days, bilateral knee OA requiring treatment in both knees, poorly controlled diabetes mellitus, blood dyscrasias, and allergy to HA or bird proteins.

Treatment groups

Participants were assigned to comparison groups by an unblinded computer-generated list, based on unrestricted randomization, which was maintained centrally by staff with no clinical involvement in the trial so no center knew the treatment allocation of any patient until the patient had been recruited into the trial.

Three groups were created:

Control group, constituted by patients who received a single intra-articular injection of 60 mg HA (Hyalone®) in a final volume of 4 ml.

Low-dose BM-MSCs group, constituted by patients who received a single intra-articular injection of 10 × 106 autologous cultured BM-MSC in 1.5 ml Ringer’s lactate solution, followed by an intraarticular injection of 4 ml HA.

High-dose BM-MSCs group, constituted by patients who received a single intra-articular injection of 100 × 106 autologous cultured BM-MSCs in 3 ml Ringer’s lactate solution, followed by an intraarticular injection of 4 ml HA.

Sample size calculation

We estimated that a sample size of ten patients per group was required to detect an effect size of 0.6 with a power of 80 %, assuming a balanced allocation to treatment groups, and a 5 % type I error probability.

Cell culture

BM-MSCs were generated under good manufacturing practice conditions (GMP) with standard operating procedures. Briefly, bone marrow (100 ml) was harvested from the pelvic bone (iliac crest) under sterile conditions. The mononuclear cell fraction was isolated by Ficoll density gradient centrifugation (Ficoll-Paque, GE Healthcare Bio-Sciences AB, Uppsala, Sweden). Cells, ranging between 20 × 106 and 60 × 106, were subsequently seeded in 175 cm2 flasks with growth medium, which consisted of αMEM without ribonucleosides (Gibco, Life Technologies, Carlsbad, CA, USA) supplemented with 5 % platelet lisate, 2 units/ml heparin, penicillin–streptomycin at 1 % (Gibco) and 1 ng/ml human fibroblast growth factor (bFGF) (Sigma-Aldrich, St. Louis, MO, USA). The flasks were maintained in culture at 37 °C in 5 % CO2 atmosphere. The growth medium was changed every 3–4 days. About 10–15 days later, colonies were formed and the cells were split with TrypLE Select™ (Life Technologies) and seeded at 3000–5000 cells/cm2. Once 70–80 % confluence was reached, cells were split again and cultured until they were available at the amounts required to be administered to patients. Finally, cells were harvested with TrypLE Select™, washed three times with PBS and resuspended in Ringer’s lactate buffer (Grifols, Barcelona, Spain) containing 1 % human albumin (Grifols), to be administered within 24 h of harvesting of the cells. Cells were characterized according to ISCT criteria. Cells were then analyzed by flow cytometry (FACSCalibur, BD Biosciences, San José, CA, USA) with the appropriate antibodies (BD Biosciences) to confirm expression of surface markers CD90, CD73 and CD44, as well as absence of CD34 and CD45.

Cell injection

Cell injection was performed without radiographic guidance through a lateral patellar approach by three different orthopaedic surgeons from both involved centers (Additional file 1: Figure S1), 3–4 weeks after the iliac crest biopsy had been performed. In 90 % of the patients, cells were administered within the first hour after being harvested. For this purpose, a 19 G needle was used in two consecutive intraarticular injections. In the first one, 10 × 106 (low dose) or 100 × 106 (high dose) BM-MSCs were administered in 1.5 and 3 ml doses respectively. Subsequently, 4 ml HA (Hyalone®) were injected using the same via.

Outcomes of interest

The occurrence of complications and/or adverse effects during the study was registered. In addition, the response to the intra-articular infusion of HA with or without BM-MSCs was assessed using the following procedures:

A goniometer-based evaluation of the articular range of motion at baseline i.e. before treatment administration, and 3, 6 and 12 months after treatment.

Two scale-based methods Visual Analog Scale (VAS) [25] and the Likert version of the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) [26, 27], evaluated at baseline and 3, 6 and 12 months after treatment, to clinically assess pain and function. VAS ranges from 0 (maximum relief, i.e., no pain) to 10 (no relief, i.e., maximal pain). WOMAC comprises three subscores: pain, which includes 5 items; stiffness, with 2 items; and physical function, with 17 items. According to previous literature, patients were considered WOMAC responders when they reported an improvement of 20 % in at least two items together with an improvement of ten points in the overall scale [28].

Rosenberg X-ray projections at baseline and 6 and 12 months afterwards to provide a radiographic assessment of the joint space width. A custom methacrylate patient positioner was used to achieve a comparative view (Additional file 2: Figure S2).

A magnetic resonance imaging (MRI) study at baseline and 6 and 12 months after treatment. Two experienced radiologists evaluated MRI images in a blinded manner by assessing the number and location of the lesions, cartilage thickness, signal intensity, and subchondral bone alteration and volume, following the Whole-Organ Magnetic Resonance Imaging Score (WORMS) protocol, in which higher score values indicate more damage [29]. 3T Magnetom TRIO equipment (Siemens, Erlangen, Germany) was used following a protocol which included an axial T1 weighted image (WI) with slice thickness of 5 mm, coronal T1 WI (4 mm), sagittal T1 WI (4 mm), sagittal T2 FS WI (4 mm) and sagittal gradient echo 3D (DESS) (2 mm).


The analyses were performed according to treatment assignment, and all available data from all patients were included in the analyses, following the intention-to-treat principle. Descriptive data summaries are presented as median [interquartile range (IQR)] or percentages. Within each group, the comparison of each clinical and radiographic endpoint between the value obtained at 6 or 12 months and the baseline value, i.e. the one obtained immediately before the administration of the treatment, was performed using the Mann–Whitney U test. Changes in the same end points over time were determined calculating the differences between the measurements collected at the 6 or 12-month follow-up visit and the baseline visit. Subsequently, comparisons between treatment groups were carried out using the Kruskal–Wallis test and the Mann–Whitney U test. All tests were two-tailed. A p value of 0.05 was considered to indicate statistical significance, without adjustment for multiple testing. All analyses were performed using Stata 14 (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP) and IBM SPSS Statistics 20 (IBM Corp. Released 2011. IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp).


Demographics of patients

Thirty-two patients were assessed for eligibility, and were consecutively randomized to treatment groups (Fig. 1). Two patients who had been randomly assigned to the control group withdrew consent and were excluded from the trial. All the groups showed similar baseline characteristics of age and body mass index. Patients in the three groups showed an uneven distribution according to the Kellgren–Lawrence scale but without statistical significance (p = 0.585, Table 1).

Fig. 1
figure 1

Study flow diagram. Patients were screened in the two participating centers by using the inclusion and exclusion criteria

Table 1 Baseline characteristics of patients


No serious adverse events or complications derived from the procedures or treatments were noted. There were no clinically important trends in the results of physical examination, vital signs and laboratory tests during the study. Articular pain requiring anti-inflammatory treatment during the first 24 h after infiltration was observed in 1, 3 and 6 patients in the control, low-dose BM-MSC and high-dose BM-MSC groups respectively. All patients recovered completely without sequelae and no treatment group-dependent differences were detected in the dose of required anti-inflammatory drug or in the time that passed until recovery.

Clinical assessment of pain and function

VAS and WOMAC clinical scores were used in order to obtain the best picture of how patients perceived their own evolution. Evaluations were performed before the administration of treatment and 3, 6 and 12 months afterwards, and the results are summarized in Fig. 2, Additional file 3: Table S1 (VAS) and Table 2 (WOMAC). The patients that were solely given HA did not show changes during follow up in their pain status according to VAS (Fig. 2; Additional file 3: Table S1). Furthermore, although they initially perceived some improvement according to the WOMAC pain and physical function subscores, this perception was not significantly sustained in the long term (Table 2). Inatraarticular delivery of BM-MSCs, specially when used at high dose, enabled patients to perceive an improvement in their perception of pain in their daily activity. On one hand, the VAS score value was significantly reduced upon treatment with low and high BM-MSC doses at all follow-up times (Fig. 2; Additional file 3: Table S1). Furthermore, treatment with 100 × 106 cells was associated with a significant improvement in all WOMAC subscores at 12 months (Table 2). It is important to note that, when the overall WOMAC value at 12 months was subtracted from the baseline value in each patient, the median decrease in the score, i.e. the relief of the symptoms, was notably larger if patients had been treated with BM-MSCs [−6.5 (−19, 4), −14 (−27, 4), and −14 (−15, −8), median (IQR), for control, low-dose and high-dose BM-MSCs groups respectively]. Thus, only the patients who had been treated with BM-MSCs met criteria to be considered WOMAC responders in the long term.

Fig. 2
figure 2

VAS scores along the study. The median values of VAS in the three groups before administration of treatments and 3, 6 and 12 months afterwards are presented. *p < 0.05; **p < 0.01 with respect to the baseline value of the same group

Table 2 WOMAC score before administration of treatments and 3, 6 and 12 months afterwards

Effect of treatments on the range of knee motion

The knee flexion and extension ranges of motion were significantly improved in the patients who were treated with BM-MSCs and the effect was seen earlier in patients receiving the higher doses of BM-MSC. No improvement was seen in patients receiving HA alone (Fig. 3; Additional file 4: Table S2).

Fig. 3
figure 3

Knee range of motion along the study. The median values expressed in degrees of the goniometric measurements of the knee flexion (top) and extension (bottom) ranges of motion before administration of treatments and 3, 6 and 12 months afterwards are presented. *p < 0.05; **p < 0.01 with respect to the baseline value of the high-dose group. #p < 0.05 with respect to the baseline valued of the low-dose group

Radiological and MRI findings

The analysis of the knee joint space by X-rays during follow up showed a borderline reduction in the control group (p = 0.05 at 12 months), which was not observed in patients treated with high dose BM-MSC (Table 3; Additional file 5: Table S3). The assessment in the low dose group was not possible because the baseline value was 0. These results suggest that BM-MSC may halt the progressive loss of cartilage observed in patients with OA despite the use of HA.

Table 3 X-ray measurement of the evolution of the knee articular interline at 6 and 12 months after the administration of treatments

Consistent with the X-Ray results, the analysis of the MRI following the WORMS protocol showed a reduction in the score value during follow up (Table 4). Patients treated with high dose BM-MSCs showed a median improvement of 4 points at 12 months, with an improvement of 22 points in 25 % of patients, while there were no signs of improvement either in the control or in the low BM-MSC group.

Table 4 WORMS score before administration of treatments and 6 and 12 months afterwards


The interest in the clinical use of MSCs for the treatment of knee OA has recently grown. However, the optimal dose and source of cells, as well as the use of coadjuvants, are not yet established. In the present clinical trial we used two single doses of BM-MSCs, 10 and 100 × 106 cells, coadministered with HA, and compared their effects with the single administration of HA in patients with knee OA.

We found that the use of BM-MSCs resulted in a significant relief of pain symptoms in the long term. According to the VAS scores, when BM-MSCs had been administered, an improvement was seen from the earliest evaluation and was maintained until the last one, at 12 months, at which time point the highest effect was observed. Interestingly, this pain reduction was independent of the dose of BM-MSCs administered. On the other hand, no significant changes in VAS were detected in the control group, and the value at 12 months was similar to the one registered before the administration of the treatments. Accordingly, the analysis of the information provided by WOMAC score confirmed that BM-MSCs induced relief of pain symptoms. It is interesting to note that, although treatment with HA alone was able to reduce the WOMAC score during the first 6 months, this improvement was not sustained in the long term, but when patients received BM-MSCs, a significant reduction in WOMAC score was detected at 12 months. In addition, unlike what was observed with VAS, only the high dose of BM-MSCs showed an efficient reduction in the WOMAC score. Furthermore, it is notable that only patients treated with high-dose BM-MSCs met the criteria to be considered WOMAC responders [28].

The effect of MSCs on pain improvement in knee OA is controversial and the literature provides differing accounts. One metaanalysis and a comprehensive review have been recently published on this topic. Xia et al. [17] performed a metaanalysis by managing the results of seven clinical trials, concluding that cell treatments were not able to reduce pain scores. Unfortunately, the heterogeneity in the methodology used in the different studies, with different cell production methods and dosage, precludes these authors from drawing solid conclusions. On the other hand, Rodríguez-Merchán [30] reviewed 25 articles that reported the use of intra-articular injection of MSCs in knee OA, finding that MSCs induce pain relief and functional improvement in three randomized clinical trials which, however, were not comparable to ours methodologically. One of them used bone marrow concentrate, another used peripheral blood progenitor cells, while the third one used cultured autologous BM-MSCs together with a high tibial osteotomy, which is a surgical treatment with a well-known impact on pain relief [3133]. The number of clinical randomized trials comparing different treatment and dosage is limited. In an interesting study, Orozco et al. [24, 34] reported an improvement in pain and function with the use of a single intra-articular injection of 40 × 106 cultured autologous MSCs in twelve patients. In a more comparable randomized clinical trial, using allogenic MSCs, Vega et al. reported good clinical outcomes in pain control and function when comparing the use of a single intra-articular injection of 40 × 106 cultured allogenic MSCs against a single intraarticular injection of HA [23].

Osteoarthritis is not considered a classical inflammatory arthropathy due to the absence of neutrophils in the synovial fluid and the lack of systemic manifestations of inflammation [35]. However, it is frequently associated with inflammation signs and symptoms such as joint pain, swelling and stiffness, leading to significant functional impairment and disability [36]. The improvement in pain scores together with the mild effect on function and MRI scores suggests that the positive effect of BM-MSCs that we have observed may rely on their paracrine function. In support of this notion, MSC antiinflammatory properties have been correlated with pain reduction elsewhere [3740]. In addition, the reduction in pain scores may explain the positive changes in flexion and extension. Although such changes are small, it must be noted that a limitation of only a few degrees in flexo-extension may severely compromise the daily functional activity. These improvements together with the findings in the image analyses, suggest that MSC-based therapies may be indicated in asymptomatic patients with mild OA grade, in whom the injected MSCs could be more effective through their paracrine function when a healthier cartilage is still present.

The maintenance of the knee joint space width has been related to an appropriate cartilage thickness [41]. Unlike what happened in the patients that were treated with HA only, who experienced a reduction of this space over the time of the study, the space width was preserved when BM-MSC were also administered, even though the results obtained in the patients that had received the low dose must be taken cautiously since the baseline value in 25 % of them was already zero, which precludes suitable follow-up. Nevertheless, a difference could be observed between the high dose and control groups, which did exhibit comparable baseline values. This finding is consistent with MRI observations and is in agreement with previous reports that also investigated the role of cultured MSCs or MSCs embedded in scaffolds in knee OA [23, 24, 4244].

The required dose of MSCs to treat knee OA efficiently is a topic of active research. Recently Jo et al. [12] performed a pilot study comparing three doses of cultured adipose tissue-derived MSCs (1 × 106, n = 3; 50 × 106, n = 3; and 100 × 106, n = 3). They found a significant reduction in the VAS score only in the high dose group at 6 months, in spite of the small number of patients included. Since results were better with the highest dose, they focused on this in a second phase of the study 100 × 106 (n = 9), with promising results. Our findings also suggest that it is preferable to administer 100 × 106 rather than 10 × 106 cells. However, we have to bear in mind that, despite randomization, the OA degree at recruitment was more severe in the patients who received only 10 × 106 cells, which may obscure our interpretation of this result.

It is accepted that OA patients have a MSC deficit that leads to a degenerative process, and the number, in vitro proliferation and differentiation potential of BM-MSCs present in the subchondral bone decreases with age and OA grade [10, 11, 45]. However, we were able to obtain a sufficient amount of BM-MSCs in osteoarthritis patients, regardless of their age or grade of disease [4649]. We have not identified any problems during the process of production of autologous BM-MSCs, achieving the number of autologous BM-MSCs proposed, even though the mean age of patients was around 60 years.

The present study is not exempt from limitations. First, ethical issues prevented us from performing a double-blinded trial. In order to minimize this inconvenience, subjective clinical scores were contrasted with objective measures to minimize bias. In addition, two independent radiologists carried out the MRI analyses in a blinded manner. Second, the relatively short duration of the study prevented us from analyzing the efficiency of the treatments beyond 1 year after the administration of the treatments. Finally, as anticipated, the severe initial condition of a portion of patients who were going to be administered the low dose of cells may have stopped these exerting more beneficial effects.


Our study shows that the single intraarticular injection of in vitro expanded autologous BM-MSCs together with HA is a safe and feasible procedure that results in a clinical and functional improvement of knee OA, especially when 100 × 106 cells are administered. These results pave the way for a future phase III clinical trial.