Intra-articular Injection of Mesenchymal Stem Cells for the Treatment of Osteoarthritis of the Knee: A Proof-of-Concept Clinical Trial
Chris Hyunchul Jo1, Young Gil Lee1, Won Hyoung Shin1, Hyang Kim1, Jee Won Chai2, Eui Cheol Jeong3, Ji Eun Kim4, Hackjoon Shim5, Ji Sun Shin1, Il Seob Shin6, Jeong Chan Ra6, Sohee Oh7, and Kang Sup Yoon1*
1Department of Orthopedic Surgery, 2Department of Radiology, 3Department of Plastic and Reconstructive Surgery,
4Department of Pathology, Seoul National University College of Medicine, SMG-SNU Boramae Medical Center, 20
Boramae-ro 5-gil, Dongjak-gu, Seoul 156-707, Korea; 5Cardiovascular Research Institute, Yonsei University College
of Medicine, 250 Seongsanno, Seodaemung-gu, Seoul, 120-752, Korea; 6Stem Cell Research Center, K-STEM CELL,
1596-7 Nakseongdae-dong, Gwanak-gu, Seoul, 151-835, Korea; 7Department of Biostatistics, SMG-SNU Boramae
Medical Center, Seoul, Korea.
Key words. Osteoarthritis • adipose-tissue derived mesenchymal stem cells • intra-articular injection • cartilage
regeneration

 

ABSTRACT
Mesenchymal stem cells are known to have a potential for articular cartilage regeneration. However, most studies focused on focal cartilage defect through surgical implantation. For the treatment of generalized cartilage loss in
osteoarthritis, an alternative delivery strategy would be more appropriate. The purpose of this study was to assess the safety and efficacy of intra-articular injection of autologous adipose tissue derived MSCs (AD-MSCs) for knee osteoarthritis. We enrolled 18 patients with osteoarthritis of the knee and injected AD MSCs into the knee. The phase I study consists of 3 dose-escalation cohorts; the low-dose (1.0×107 cells), mid-dose (5.0×107) and high-dose (1.0×108) group with 3 patients each. The phase II included 9 patients receiving the high-dose. The primary outcomes were the safety and the Western Ontario and McMaster  Universities Osteoarthritis index (WOMAC) at 6 months. Secondary outcomes included clinical, radiological, arthroscopic, and histological evaluations. There was no treatment-related adverse event. The WOMAC score improved at 6 months after injection in the high-dose group. The size of cartilage defect decreased while the volume of cartilage increased in the medial femoral and tibial condyles of the high-dose group. Arthroscopy showed that the size of cartilage defect decreased in the medial femoral and medial tibial condyles of the
high-dose group. Histology demonstrated thick, hyaline-like cartilage regeneration. These results showed that intra-articular injection of 1.0×108 AD MSCs into the osteoarthritic knee improved function and pain of the knee joint without causing adverse events, and reduced cartilage defects by regeneration of hyaline-like articular cartilage.

Author contributions: C.H.J.: Conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; Y.G.L., W.H.S.: Collection and/or assembly of data, analysis of radiographic data, administrative support; H.K.: Collection and assembly of histological samples, administrative support; J.W.C.: Collection, assembly, and analysis of MRI data; E.C.J.: Harvest adipose tissue, administrative support; J.E.K.: Data analysis of histological samples; H.S.: Data analysis and interpretation of the cartilage volume measures; J.S.S.: Collection and/or assembly of clinical and arthroscopic data, administrative support; I.S.S. and J.C.R.: Conception and design, financial support, provision of study material or patients; S.O.: Statistical data analysis; K.S.Y.: Conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript. *Corresponding author: Kang Sup Yoon, MD, Department of Orthopedic Surgery, Seoul National University College of Medicine, SMG-SNU Boramae Medical Center, 20 Boramae-ro 5-gil, Dongjak-gu, 156-707 Seoul, Korea, Tel: 82-2-870-2315, Fax: 82-2-870-3866, E-mail:
kangsupyoon@gmail.com; Acknowledgement: This work was funded by the Bio & Medical Technology Development Program (No.2011-0019773) and the Basic Science Research Program (No.2011-0022184) of the National Research Foundation (NRF) funded by the Korean government (MEST) and by K-STEM CELL; Received July 22, 2013; accepted for publication November 16, 2013; 1066-5099/2014/$30.00/0 as doi: 10.1002/stem.1634
This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as Doi: 10.1002/stem.1634

INTRODUCTION

Osteoarthritis of the knee is the most common form of arthritis that cause pain, stiffness, and decreased function, and one of leading causes of disability among noninstitutionalized adults [1, 2]. More than 50 modalities of pharmacological, non-pharmacological, and surgical treatment are reported in the literature [3]. However, the current most common treatments for osteoarthritis except for joint replacement have at best modest albeit clinically relevant effects and can endanger substantial adverse events or costs, or both [4]. Furthermore, these treatments are generally intended to decrease pain, maintain or improve joint function, and minimize disability, not to regenerate articular cartilage, whereas
osteoarthritis is characterized by the degeneration of the extracellular matrix resulting in loss of articular cartilage [5, 6].

For regeneration of articular cartilage, various efforts including cell therapy and tissue engineering have been tried. Chondrocytes are one of the most extensively investigated cells showing positive clinical outcomes [7-10]. Nevertheless, chondrocyte implantation has inherent disadvantages such as a two-stage surgical procedure that may cause further cartilage damage and degeneration [8, 10, 11], and chondrocyte dedifferentiation during culture that might result in fibrocartilage rather than hyaline cartilage [8, 12]. Moreover, its use has been limited to focal cartilage defect caused by injury while generalized cartilage loss seen in osteoarthritis has been its exclusion criterion [8, 10], suggesting the need to find a different approach for cartilage regeneration in osteoarthritis.

Mesenchymal stem cells have also been focused as an emerging regime for cartilage regeneration. Unlike chondrocytes implantation, the use of MSCs for regeneration of human articular cartilage is still investigational [13-15]. Recently, some authors reported results of direct intra-articular injection of MSCs into the knee for the treatment of focal defect or more generalized cartilage loss in osteoarthritis [16-21]. Direct intra-articular injection of MSCs would offer great advantages if it could be translated into clinical practice as it would avoid surgeries and associated side-effects, such as hypertrophy and ossification of periosteal coverage, immune reaction and disease transmission caused by xenograft coverage. More
importantly, simplicity and ease of the injection could provide better treatment opportunities, especially for the elderly with comorbidity. Despite this potential, no clinical trials have been performed but a few case reports. Therefore, we conducted a proof-ofconcept phase I/II clinical trial to assess the safety and the efficacy of intra-articular injection of autologous adipose tissue derived mesenchymal stem cells (AD MSCs) in patients with knee osteoarthritis. We report the
clinical, radiological, arthroscopic and histological results.

MATERIALS AND METHODS

Study design and patients This study is a phase I/II clinical trial with no active control conducted between March 2009 and September 2011 at SMG-SNU Boramae Medical Center, Seoul, Korea. The protocol was approved by the institutional review board of our institute. All participants provided written informed consent.

The phase I study consisted of three doseescalation cohorts; the low-, mid-, and highdose group with three patients each. Patients in each dose group received 1.0×107, 5.0×107, and 1.0×108 cells in 3ml of saline, respectively. After three patients in each cohort were followed up for 28 days after injection, a safety review was done before moving to the next dose or phase (Supplemental method 1). The phase II included nine patients receiving the high-dose. Therefore, 18 patients were granted by the Korean Food and Drug Administration (KFDA) and were consecutively enrolled in the trial.

Eligible patients were between 18 and 75 years of age with idiopathic osteoarthritis of the knee of grade 2 or more according to Kellgren–Lawrence criteria and had an average pain intensity of grade 4 or more on a 10 point visual analog scale for at least 4 months. Details of inclusion and exclusion criteria are listed in the Supplemental method 2.

Patients underwent physical examination, laboratory tests including routine blood and urine tests, serologic tests, tumor screening, and the pregnant test if indicated, and MRI of the knee at screening after providing informed consent. All pain medications except the rescue analgesic, acetaminophen, were discontinued (Supplemental method 3). Eligible patients returned to the hospital within 1 week for liposuction. Arthroscopy and cell injection was performed 3 weeks after liposuction. Patients were followed up at 1, 2, 3, and 6 months after injection. At each visit, the safety and efficacy assessments were performed. Furthermore, MRI of the knee was obtained at 3 and 6 months after injection. Second-look
arthroscopy was performed at 6 months after injection. A 2mm-punch biopsy specimen was obtained from the center of the cartilage defect of the medial femoral condyle at the first arthroscopy, and from the adjacent area to the first biopsy site at the second-look arthroscopy in patients who gave consent in the high-dose group. Independent safety and data monitors oversaw the overall trial process.

Mesenchymal stem cell preparation
AD MSCs (Jointstem; K-STEM CELL, Seoul, Korea) were prepared from the abdominal subcutaneous fats by liposuction under good manufacturing practice conditions, as previously described (Supplemental method 4) [22]. Cells were tested before shipping for cell number, viability, purity (CD31, CD34, CD45), identity (CD73, CD90), sterility, endotoxin, and mycoplasma (Supplemental table 1).

Arthroscopy and stem cell injection
All procedures were performed in the supine position under spinal anesthesia. A single orthopedic surgeon performed all procedures. Standard arthroscopic examination of the knee was performed; articular cartilage lesions were
measured with a calibrated arthroscopic probe, and graded according to the international cartilage repair society (ICRS) cartilage injury classification [23]. After diagnostic exploration, AD MSCs in 3ml of saline were injected into
the knee joint through the medial portal via 22G spinal needle. No debridement, synovectomy or meniscectomy was performed during arthroscopy, and no drainage was used. Postoperative rehabilitation is described in the Supplemental method 5.

Outcome measures
Primary outcomes were the safety and the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) at 6 months after injection [24]. The safety was assessed with vital signs, physical examination, laboratory tests, adverse events, and serious adverse events. Adverse events were categorized using National Cancer Institute – Common Terminology Criteria for Adverse Events version 4.0 scale (NCI – CTCAE v4.0). The WOMAC is a validated, self-administered
outcome measure designed to evaluate knee and hip osteoarthritis; higher scores mean increased pain, stiffness, and decreased function [24].

Secondary outcomes included four categories: clinical, radiological, arthroscopic, and histological. Clinical outcomes included a visual analog scale for knee pain on a scale from 0 to 10, and Knee Society Clinical Rating System (KSS) score [25]. Radiological outcomes were measured with Kellgren-Lawrence grade [26], joint space width of the medial compartment [27], mechanical axis with weight bearing line [28], and anatomical axis using x-ray. The size, depth of cartilage
defect, and signal intensity of regenerated cartilage was also measured using MRI by a blinded musculoskeletal radiologist as previously described (Supplemental method 6) [29, 30]. In addition, changes of the cartilage volume of the knee joint were measured using a semi-automated segmentation method by a blinded researcher (Supplemental method 7) [31]. Arthroscopy was performed to evaluate any change in cartilage defect at the time of cell injection, and at 6 months after injection. The size and ICRS grade of cartilage defect was measured. If cartilage was regenerated at second-look arthroscopy, ICRS grade of the defect was changed only when regenerated cartilage covered more than 50% of the
original defect. For histological assessment, biopsy specimens were subject to safranin O staining and immunohistochemistry for type I and II collagen as previously described with slight modification (Supplemental method 8) [32]. Thickness of regenerated cartilage measured, and specimens evaluated with ICRS II by a histopathologist [33].

Statistical analysis
The sample size (18 patients) was decided in consultation with the Korean FDA. Outcome measures were analyzed based on the intention-to-treat population. Missing data were replaced with multiple imputations (10 sets) under a missing-at-random assumption [34]. Ten imputed datasets were generated, analyzed separately for each outcome measure, and then combined into a single set of estimates according to the Rubin rules [35]. For sensitivity test, single imputation using the last-observation-carried forward method and a complete-case analysis were additionally performed [36]. Because all of the three methods did not yield meaningful changes in each measurement, we presented only the imputation analyses. Changes from baseline in all the measures that were scale variables were determined with a paired t-test. Kellgren-Lawrence grade, depth of the cartilage defect measured by MRI, and ICRS grade determined with arthroscopy were determined with a Wilcoxon signed rank test. The analysis was performed using SAS version 9.2 (SAS Institute, Inc., Cary, NC).

Results

Demographics of patients
Twenty-five patients were assessed for eligibility and 18 patients were consecutively allocated to treatment groups and received AD MSCs (Fig. 1). Generally, all the patients enrolled in the study showed similar baseline characteristics of age, height, weight, body mass index, and radiographic grade of osteoarthritis. One patient in the mid-dose group withdrew consent after cell injection. Another patient completed follow-up except for the second-look arthroscopy. The other
sixteen patients completed 6 months of followup. Analysis was performed according to the level of cell doses (low-, mid-, and high-dose), not to the phase of the trial, and according to the intention-to-treat principle in clinical, radiological, and arthroscopic assessments. Histological assessments were performed in specimens from eight patients in the high-dose group who gave consent for biopsies at both arthroscopies.

Patients in each group had similar baseline characteristics (Table 1). Generally, females aged 60 years with an average body-mass index around 26 who suffered for more than 5 years despite conservative treatments were included in the study. All patients had osteoarthritis of the knee of Kellgren-Lawrence grade 3 or 4. Baseline cartilage defect of the medial femoral condyle measured with MRI was 407.0, 535.0, and 497.9 mm2 in the low-, mid-, and high-dose group, respectively.

Safety
Adverse events occurred in two (67%), two (67%), and five (42%) patients in the low-, mid-, and high-dose group, respectively (Table 2). None of them was grade 3 or 4 by NCICTCAE scale, nor treatment-related. The most common adverse event was nasopharyngitis, which developed in one patient in each group (Supplemental table 2). There was one serious adverse event, urinary stone, which occurred in a patient in the low-dose group with a previous history. He was treated with extracorporeal shock wave lithotripsy and medicine. He was fully recovered and completed follow-up. Two patients reported arthralgia. One patient in the mid-dose group reported bilateral knee pain; the ipsilateral pain and tenderness was due to pes bursitis which has been known commonly accompanied in the osteoarthritis knee. And the contralateral pain was due to osteoarthritis of the contralateral knee. The other patient in the high-dose group also reported pain and
tenderness in the pes anserinus of the ipsilateral knee. Both patients were managed with knee stretching and quadriceps setting exercise and intermittent acetaminophen. Both of them completed follow-up. No patients were discontinued from the study because of adverse events. There were no clinically important trends in the results of physical examination, vital signs, laboratory test during the study.

Clinical outcomes
AD MSCs injection was associated with improvement of the WOMAC score at 6 months after injection as compared with baseline in the high-dose groups (Fig. 2A and Supplemental table 3). The mean reduction from the baseline over 6 months was 39% in the high-dose group, from 54.2 ± 5.2 to 32.8 ± 6.3 (P = 0.003). Patients in the low- and middose group did not improve over 6 months.

Visual analog scale for knee pain significantly decreased from 79.6 ± 2.2 to 44.2 ± 6.3 in the high-dose group only (45% decrease; P <0.001) (Fig. 2B).

The knee score of KSS significantly increased in the low-dose group from 41.3 ± 6.8 to 79.0 ±12.5 (91% increase; P = 0.025) and in the highdose group from 47.2 ± 2.6 to 71.0 ± 4.4 (50% increase; P < 0.001) (Fig. 2C). Meanwhile, the
function score of KSS significantly increased in the low-dose group only from 60.0 ± 5.8 to 83.3 ± 8.8 (39% increase; P = 0.020) (Fig. 2D).

Radiological outcomes
Kellgren-Lawrence grade, joint space width, mechanical axis, and anatomical axis did not change significantly over 6 months in all dose groups (Supplemental table 4).

Serial MRI examinations found gradual regeneration of articular cartilage in the medial femoral and tibial condyles over 6 months (Fig. 3A). At 3 months, thin cartilage was noticed in the both condyles. It thickened and became mature with isointensity at 6 months.

The size of cartilage defect measured with MRI significantly decreased both in the medial femoral and tibial condyles as well as in the lateral femoral and tibial condyles at 6 months in the high-dose group. (Fig. 3A and Supplemental table 5); from 497.9 ± 29.7 mm2 to 297.9 ± 51.2 mm2 in the medial femoral condyle (40% decrease; P = 0.004), from 333.2 ± 51.2 mm2 to 170.6 ± 48.2 mm2 in the medial tibial condyle (49% decrease; P < 0.001), from 103.6 ± 27.1 mm2 to 51.1 ± 24.9 mm2 in the lateral femoral condyle (51% decrease; P = 0.011), and from 19.4 ± 7.3 mm2 to 10.4 ± 4.2 mm2 in lateral tibial condyle (46% decrease; P = 0.041), but not in the patella, from 93.3 ± 33.3 mm2 to 79.1 ± 27.6 mm2 (15% decrease; P = 0.340). There were no significant changes in the other dose groups.

The depth of the cartilage defect did not show significant changes over 6 months in all dose groups (Supplemental table 6).

The signal intensity of regenerated cartilage in each compartment had a slight tendency to become isointense over 6 months in the highdose group, but without a statistical significance (Supplemental table 7).

The cartilage volume also increased gradually over time till 6 months both in the medial femoral and tibial condyles in the high-dose group (Fig. 3B and Supplemental table 8); from 3313.7 ± 304.1 mm3 to 3780.6 ± 284.4 mm3 in the medial femoral condyle (14% increase; P = 0.044) and from 1157.5 ± 145.8 mm3 to 1407.7 ± 150.5 mm3 in the medial tibial condyle (22% increase; P = 0.047). Meanwhile, patients in the low-dose group temporarily also showed increased cartilage
volume from 3315.0 ± 104.3 mm3 to 3959.7 ± 55.9 mm3 at 3 months (27% increase; P = 0.026) in the medial femoral condyle. The cartilage volume of the lateral femoral and tibial condyles and the patella did not change in all dose-groups over 6 months).

Second-look arthroscopy
As a gold standard for articular cartilage assessment, arthroscopy before and 6 months after AD MSCs injection demonstrated findings consistent with clinical and radiological outcomes. Macroscopically,  regenerated cartilage formed in the most severely degenerated area with ICRS grade 3 in the medial femoral and tibial condyles, whereas it was hardly seen in the less severely degenerated area in the lateral compartment and the patella (Fig. 4A-C). Regenerated cartilage looked glossy white with a smooth surface. With a probe, it felt firm like healthy articular cartilage in the medial femoral condyle, whereas it was less firm in the medial tibial condyle. No loose body, hypertrophy or abnormal calcification was identified.

The size of cartilage defect measured with a calibrated probe demonstrated a significant reduction of the cartilage defect from 1225.7 ± 282.8 mm2 to 837.8 ± 278.9 mm2 in the medial femoral condyle (32% decrease; P = 0.003) and from 352.3 ± 77.6 mm2 to 126.3 ± 43.8 mm2 in the medial tibial condyle (64% decrease; P = 0.008) in the high-dose group (Fig. 4D). The size of cartilage defect in the lateral femoral and tibial condyle, and the patella did not change in all dose-groups over 6 months (Supplemental table 9).

The ICRS grade of the cartilage defect significantly improved in the medial femoral and tibial condyle in the high-dose group at second-look arthroscopy (Fig. 4E). No significant change was found in the lateral femoral and tibial condyles, and the patella did not change in all dose-groups (Supplemental table 10).
Histological outcomes
Generally, biopsy specimens from the medial femoral condyles had no articular cartilage before injection (ICRS 3C) (Fig. 5A). At 6 months after injection, articular cartilage with a thick, glossy white matrix and smooth surface
was regenerated and was well integrated with the subchondral bone. In the lower half of the middle zone and the deep zone, safranin O and type II collagen positive hyaline-like cartilage was clearly demonstrated, whereas type I
collagen positive fibrocartilage was identified in the superficial and the upper half of the middle zone. Collagen fibrils in the superficial and middle zone run parallel and oblique to articular surface, respectively, whereas those in
the deep zone run vertically. Chondrocytes are flattened in the superficial zone, and round in the middle and deep zones similar to those in the deep zone of hyaline cartilage. Small chondrocytes are also found in the in the calcified cartilage zone. However, typical columnar chondrocytes or tide mark is not definite, suggesting that maturation is still in process [37]. In some patients with ICRS 3B before injection, hyaline-like articular cartilage similar to Fig. 5A was also regenerated (Fig. 5B). Meanwhile, relatively thin fibrocartilage with minimal safranin O and type II collagen positive matrix was formed in the worst case (Fig. 5C). Additional histological data are available in the Supplemental figure.
The ICRS II scores changed significantly after AD MSCs injection in four parameters: surface architecture, and surface, mid, and overall assessments (Supplemental table 11). The mean thickness of articular cartilage increased from 0.4 ± 0.3 mm before injection, which increased to 1.6 ± 0.8 mm after injection (300%; P = 0.004). The mean thickness of regenerated cartilage in four patients who had no cartilage (ICRS grade 3C) before injection was also 1.6 ± 0.5 mm.

Discussion

This proof-of-concept trial reached its predetermined primary outcomes, that is, intraarticular injection of AD MSCs into osteoarthritic knee was not associated with apparent adverse events, but improved function of the knee measured with WOMAC over 6 months of follow-up. Patients in the high-dose group demonstrated significantly improved WOMAC score with a clinically meaningful pain reduction which is approximately 30% from the baseline [38]. Evaluation with MRI
and second-look arthroscopy identified regenerated articular cartilage consistently in the high-dose group. Histological evaluation revealed that regenerated cartilage had a thick, glossy white matrix with a smooth surface, and was well integrated with the subchondral bone. In the upper half of the middle and the deep zones, safranin O and type II collagen positive hyaline-like cartilage was clearly demonstrated, whereas type I collagen positive fibrocartilage was identified in the superficial and the upper half of the middle zones. Patients in the middose group showed improvement in some clinical outcomes, but those in the low-dose group did not show improvement in most outcome measures. These results would be due to regeneration of articular cartilage as well as via paracrine effects, and that the effects were closely related to the number of injected ADMSCs.

We agree that osteoarthritis is a mesenchymal disease, that is to say, a condition in which the activity, phenotype or mobilization of MSC population is altered, leading to an absence of repair and increased degeneration [39]. In osteoarthritis, MSCs are depleted and have reduced proliferative capacity and reduced ability to differentiate [40]. Therefore, provision of an adequate number of healthy and functional MSCs would be helpful for enhancing repair or inhibit the progression of cartilage loss [18]. Potential mechanisms of MSCs for the treatment of osteoarthritis are believed through two ways. One is direct differentiation into chondrocytes, and the other is paracrine effects of secreted bioactive materials [39, 41]. Early studies have focused the differentiation potentials of MSCs which were examined with small surgically created chondral defects in animal models [15, 42]. Recent studies also showed that MSCs contributed to the repair of damaged articular cartilage through homing, engraftment and production of cartilage matrix [16, 18, 43] in osteoarthritis models. Differentiation of delivered MSCs into chondrocytes appeared to be induced by the local environment of the
homing site [43, 44]. Meanwhile, a surging paradigm suggests that direct differentiation might not be the only mechanism, but paracrine effects through secretion of bioactive materials should involve [39, 45]. MSCs are known to stimulate chondrocytes to proliferate and synthesize extracellular matrix [46-48], to induce anti-inflammatory cytokine production [44, 49-51], and to possess immunomodulatory properties [52, 53]. These studies together suggest that MSCs modulate inflammation and provide environment for tissue regeneration either by direct secretion of bioactive materials, or by controlling cytokine and growth factor production from endogenous cells [41, 49, 54- 57]. The results of this study provide robust evidences for both mechanisms. Regeneration of hyaline-like articular cartilage after injection is clearly demonstrated in this study by MRI, arthroscopy and histology. Evidences of previous studies showing that injected cells participated in regeneration of articular cartilage suggest that injected MSCs rather than endogenous cells recruited by paracrine mechanism were supposed to regenerate articular cartilage [16, 43, 58] while we did not track. Furthermore, as even few MSCs could trigger paracrine effects [44], better clinical and structural results in the high-dose group should more support that regeneration occurred mainly via direct differentiation. However, improved clinical outcomes not only in the high-dose group but also in the low- and middose group suggest that paracrine effects should also work. Nevertheless, we still do not have enough knowledge about details; when and how much each mechanism contributes, which mechanism is more important to patients with different conditions, optimal cellular dose and condition for each mechanism, and etc. Additional researches need to be done for elucidation of these questions. We used AD MSCs in this trial with already proven safety [22]. In comparison with bone marrow MSCs, AD MSCs have several advantages including feasibility of harvesting in a large amount by a simple, repeatable, and minimally invasive method, the highest frequency of MSCs [59], easy and rapid expansion in culture, and higher passage cells still retaining stem cell phenotypes and pluripotency [60]. However, the main benefits of AD MSCs are that they have less effect of age or morbidity of patients on quality in contrast to bone marrow MSCs [40, 61-64]. Despite some concerns about inferior chondrogenic potential of AD MSCs [65, 66], several experimental studies showed that AD MSCs reduced hypertrophy and
dedifferentiation of chondrocytes [67], inhibit synovial thickening, and protect against joint destruction [68], and decreased the development and progression of osteoarthritis [69, 70]. The results of this study are consistent with previous experimental studies, and suggest that AD MSCs are an appealing source for the treatment of osteoarthritis.

Most previous studies that investigated potentials of MSCs for regeneration of articular cartilage have used acute chondral defects models through surgical implantation [13, 15, 17, 71, 72]. Those defects are usually small with defined dimensions, surrounded by relatively normal cartilage, and thus would simulate cartilage injury caused by trauma. However, cartilage lesions associated with osteoarthritis are chronic, large, complex in shape and thickness and surrounded by degenerative cartilage. Therefore, alternate strategies other than direct implantation would be more appropriate [16-18, 73]. MSCs are known to home and are preferentially attracted to diseased tissue rather than to intact tissue [58, 74-76]. Using this homing ability, some authors demonstrated that intra-articularly injected MSCs attached to cartilage defect, proliferated, and participated in regeneration of articular cartilage [16, 17, 43], decreased synovial fluid concentration of prostaglandin E2 [50], and retard the progression of osteoarthritis [18, 77]. A few case reports in human also described encouraging early clinical outcomes of intra-articular injection of bone marrow MSCs [19-21]. In line with previous experimental studies and clinical case reports, the present study demonstrated a great promise of intra-articular injection of AD MSCs with details of clinical, radiological, arthroscopic and histological results. Current medical treatment for osteoarthritis are commonly associated with gastrointestinal, hepatic, renal, or cardiac side-effects [78], and surgery is inevitably invasive no matter how minimal it is. This makes intra-articular injection a valuable option, especially in the elderly. Considering very low incidence of infection, 0.002% [79], and feasibility of the procedure, intra-articular injection of MSCs would be a valuable therapy for osteoarthritis if evidences accumulate. One of important findings in this study is that most of regenerated cartilage was found in the medial femoral and tibial condyles, both of which were the most severely degenerated site in the knee. The results are consistent with studies reporting that injected cells adhere diseased rather than intact articular cartilage [18, 80, 81]. Also these results would confirm the homing ability of AD MSCs that actually work in human osteoarthritis. Meanwhile, little change was found in the other compartments such as the lateral femoral and tibial condyles, and patella in which less degenerated cartilage existed. Considering that earlier injection of MSCs during the progression of osteoarthritis would be more beneficial [16], investigations for enhancing homing and engraftment of MSCs not only to as most degenerated location but also to less degenerated site should be necessary.

Patients in the high-dose group showed significantly improved outcomes in most clinical, radiological and arthroscopic measures whereas those in the low- and middose group did not. These results suggested that a sufficiently adequate number of MSCs should be delivered to the lesion for the best results. The importance as well as concerns of the cell dose has been raised by several authors [58, 65, 82]. Some reported that injection of 1.0×107 MSCs generated free bodies of scar tissue in the rat knee [58], whereas others reported insufficient numbers of applied cells showed inferior results [65]. Therefore, the optimal cell dose needed to be clarified for achieving efficacy balanced with safety. The present study showed that at least the total number of 1.0×108 MSCs per injection would be a prerequisite for consistently good results. Nevertheless, they might not be the best results; regenerated cartilage did not completely cover the original defect of the medial femoral and tibial condyles even in the high-dose group, and there were little changes in the other compartments. Therefore, further studies would be necessary for optimal results, and repeated injections at intervals could be a
good option.

There are some limitations of the study. First, there is no control in the study. A larger scale study with an appropriate control would be necessary for clinical application. Second, while regeneration of articular cartilage was clearly identified with MRI, arthroscopic and histological measures, the 6-month of followup would be short especially for the assessment of clinical outcomes as certain clinical outcomes such as VAS pain in the mid-dose group increased at the final follow-up. Further study with longer follow-up would be necessary. Third, the results in the high-dose group might not be the best. As increasing the number of injected cells more may be practically difficult and would raise concerns such as fibrous foreign body formation, another approaches including repetition of the injection and enhancement of homing ability of MSCs would be more promising. Fourth, the period of non-weight bearing after injection would not be optimized. As a proof-of-concept study, we focused more on regeneration of articular cartilage than on early return to daily activity. Thus, we recommended non-weight bearing with only toe-touch for 8 weeks that may be similar with the period used in other treatments for cartilage regeneration [83, 84]. Whereas this prolonged period of non-weight bearing might allow some native repair, it decreased and delayed recovery of the knee function after injection as evidenced by initial
decline of the function score of KSS (Fig. 2D). Therefore, an optimal rehabilitation protocol for intra-articular injection of MSCs needs to be further investigated. Fifth, clinical researches need to use a validated questionnaire that is specific for the condition being studied. While WOMAC is a widely used, validated self-administered instrument specifically designed to evaluate knee and hip osteoarthritis [24], it might not be specific for evaluating patients after intra-articular
injection of AD MSCs which has never been studied before. Finally, the quality of regenerated cartilage would be not optimal as demonstrated in the histological results. Further investigations for enhancing chondrogenic differentiation would be necessary for better results.

Conclusions

In summary, intra-articular injection of 1.0×108 AD MSCs into the osteoarthritic knee improved function and pain of the knee joint without causing adverse events. Radiological, arthroscopic, and histological measures consistently demonstrated decreased of articular cartilage defects by regeneration of hyaline-like articular cartilage. These results are promising to encourage large randomized clinical trials, and we are cautiously optimistic about this new step for the treatment of osteoarthritis of the knee.

Acknowledgment 

We express our sincere appreciation to the patients who participated in this trial. We also thank Jae Seob Jung, Myung Hwa Lee, Ji Eun Kang, Ji Tae Jung, Dae Woong Jo, Sung Sook Yoo, Min Hee Kim and Youngtaek Hong for
valuable assistance.

Disclosure of potential conflicts of interest
KSY reported receiving research grant and travel support from K-STEM CELL. ECJ reported receiving research grant from KSTEM CELL. ISS and JCR are employees of K-STEM CELL and reported owning stock and/or stock options. CHJ reported receiving grants from the National Research Foundation funded by the Korean Ministry of Education and Science Technology. K-STEM CELL is a company manufacturing stem cells and researching their application. No other authors reported any financial disclosures.

References

.1. Lawrence RC, Felson DT, Helmick CG et al. Estimates of the prevalence of arthritis and other rheumatic conditions in the United States. Part II. Arthritis and rheumatism. 2008;58:26-35.

2. Dillon CF, Rasch EK, Gu Q et al. Prevalence of knee osteoarthritis in the United States: arthritis data from the Third National Health and Nutrition Examination Survey 1991-94. The Journal of rheumatology. 2006;33:2271-2279.

3. Zhang W, Moskowitz RW, Nuki G et al. OARSI recommendations for the management of hip and knee osteoarthritis, Part II: OARSI evidence-based, expert consensus guidelines. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2008;16:137-162.

4. Lohmander LS, Roos EM. Clinical update: treating osteoarthritis. Lancet. 2007;370:2082-2084.

5. Jo H, Park JS, Kim EM et al. The in vitro effects of dehydroepiandrosterone on human osteoarthritic chondrocytes. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2003;11:585-594.

6. Wilson JF. To stop osteoarthritis, fixing cartilage may not be enough. Annals of internal medicine. 2007;147:437-439.

7. Grande DA, Pitman MI, Peterson L et al. The repair of experimentally produced defects in rabbit articular cartilage by autologous chondrocyte transplantation. J Orthop Res. 1989;7:208-218.

8. Knutsen G, Drogset JO, Engebretsen L et al. A randomized trial comparing autologous chondrocyte implantation with microfracture. Findings at five years. J Bone Joint Surg Am. 2007;89:2105-2112.

9. Vanlauwe J, Saris DB, Victor J et al. Five-year outcome of characterized chondrocyte implantation versus microfracture for symptomatic cartilage  defects of the knee: early treatment matters. Am J Sports Med. 2011;39:2566-2574.

10. Brittberg M, Lindahl A, Nilsson A et al. Treatment of deep cartilage defects in the knee with autologous chondrocyte transplantation. The New England journal of medicine. 1994;331:889-895.

11. Lee CR, Grodzinsky AJ, Hsu HP et al. Effects of harvest and selected cartilage repair procedures on the physical and biochemical properties of articular cartilage in the canine knee. J Orthop Res. 2000;18:790-799.

12. von der Mark K, Gauss V, von der Mark H et al. Relationship between cell shape and type of collagen synthesised as chondrocytes lose their cartilage phenotype in culture. Nature. 1977;267:531-532.

13. Wakitani S, Imoto K, Yamamoto T et al. Human autologous culture expanded bone marrow mesenchymal cell transplantation for repair of cartilage defects in osteoarthritic knees. Osteoarthritis and cartilage / OARS, Osteoarthritis
Research Society. 2002;10:199-206.

14. Nejadnik H, Hui JH, Feng Choong EP et al. Autologous bone marrow-derived mesenchymal stem cells versus autologous chondrocyte implantation: an observational cohort study. Am J Sports Med. 2010;38:1110-1116.

15. Kuroda R, Ishida K, Matsumoto T et al. Treatment of a full-thickness articular cartilage defect in the femoral condyle of an athlete with autologous bone marrow stromal cells. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2007;15:226-231.

16. Mokbel AN, El Tookhy OS, Shamaa AA et al. Homing and reparative effect of intra-articular injection of autologus mesenchymal stem cells in osteoarthritic animal model. BMC musculoskeletal disorders. 2011;12:259.

17. Lee KB, Hui JH, Song IC et al. Injectable mesenchymal stem cell therapy for large cartilage defects–a porcine model. Stem cells (Dayton, Ohio). 2007;25:2964-2971.

18. Murphy JM, Fink DJ, Hunziker EB et al. Stem cell therapy in a caprine model of osteoarthritis. Arthritis and rheumatism. 2003;48:3464-3474.

19. Centeno CJ, Busse D, Kisiday J et al. Increased knee cartilage volume in degenerative joint disease using percutaneously implanted, autologous mesenchymal stem cells. Pain physician. 2008;11:343-353.

20. Davatchi F, Abdollahi BS, Mohyeddin M et al. Mesenchymal stem cell therapy for knee osteoarthritis. Preliminary report of four patients.  International journal of rheumatic diseases. 2011;14:211-215.

21. Emadedin M, Aghdami N, Taghiyar L et al. Intraarticular injection of autologous mesenchymal stem cells in six patients with knee osteoarthritis. Archives of Iranian medicine. 2012;15:422-428.

22. Ra JC, Shin IS, Kim SH et al. Safety of intravenous infusion of human adipose tissue-derived mesenchymal stem cells in animals and humans. Stem cells and development. 2011;20:1297-1308.

23. Brittberg M, Peterson L. Introduction of an articular cartilage classification. ICRS Newsletter. 1998;1:5-8.

24. Bellamy N, Buchanan WW, Goldsmith CH et al. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. The Journal of rheumatology. 1988;15:1833-1840.

25. Insall JN, Dorr LD, Scott RD et al. Rationale of the Knee Society clinical rating system. Clin Orthop Relat Res. 1989:13-14.

26. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Annals of the rheumatic diseases. 1957;16:494-502.

27. Buckland-Wright JC, Wolfe F, Ward RJ et al. Substantial superiority of semiflexed (MTP) views in knee osteoarthritis: a comparative radiographic study, without fluoroscopy, of standing extended, semiflexed (MTP), and schuss views. The Journal of rheumatology. 1999;26:2664-2674.

28. Johnson F, Leitl S, Waugh W. The distribution of load across the knee. A comparison of static and dynamic measurements. The Journal of bone and joint surgery British volume. 1980;62:346-349.

29. Marlovits S, Striessnig G, Resinger CT et al. Definition of pertinent parameters for the evaluation of articular cartilage repair tissue with highresolution magnetic resonance imaging. European journal of radiology. 2004;52:310-319.

30. Yulish BS, Montanez J, Goodfellow DB et al. Chondromalacia patellae: assessment with MR imaging. Radiology. 1987;164:763-766.

31. Bae KT, Shim H, Tao C et al. Intra- and interobserver reproducibility of volume measurement of knee cartilage segmented from the OAI MR image set using a novel semi-automated segmentation method. Osteoarthritis and cartilage / OARS, Osteoarthritis Research Society. 2009;17:1589-1597.

32. Jo CH, Ahn HJ, Kim HJ et al. Surface characterization and chondrogenic differentiation of mesenchymal stromal cells derived from synovium. Cytotherapy. 2007;9:316-327.

33. Mainil-Varlet P, Van Damme B, Nesic D et al. A new histology scoring system for the assessment of the quality of human cartilage repair: ICRS II. Am J Sports Med. 2010;38:880-890.

34. Gadbury GL, Coffey CS, Allison DB. Modern statistical methods for handling missing repeated measurements in obesity trial data: beyond LOCF. Obesity reviews : an official journal of the International Association for the Study of Obesity. 2003;4:175-184.

35. Rubin DB. Multiple imputation for nonresponse in surveys: J Wiley; 1987.

36. White IR, Thompson SG. Adjusting for partially missing baseline measurements in randomized trials. Statistics in Medicine. 2005;24:993-1007.

37. Temenoff JS, Mikos AG. Review: tissue engineering for regeneration of articular cartilage. Biomaterials. 2000;21:431-440.

38. Rowbotham MC. What is a “clinically meaningful” reduction in pain? Pain. 2001;94:131-132.

39. Barry F, Murphy M. Mesenchymal stem cells in joint disease and repair. Nature reviews Rheumatology. 2013.

40. Murphy JM, Dixon K, Beck S et al. Reduced chondrogenic and adipogenic activity of mesenchymal stem cells from patients with advanced osteoarthritis. Arthritis and rheumatism. 2002;46:704-713.

41. Caplan AI, Dennis JE. Mesenchymal stem cells as trophic mediators. Journal of cellular biochemistry. 2006;98:1076-1084.

42. Wakitani S, Goto T, Pineda SJ et al. Mesenchymal cell-based repair of large, full-thickness defects of articular cartilage. J Bone Joint Surg Am. 1994;76:579-592.

43. Sato M, Uchida K, Nakajima H et al. Direct transplantation of mesenchymal stem cells into the knee joints of Hartley strain guinea pigs with spontaneous osteoarthritis. Arthritis research & therapy. 2012;14:R31.

44. Yang S-H, Wu C-C, Shih TT-F et al. In vitro study on interaction between human nucleus pulposus cells and mesenchymal stem cells through paracrine stimulation. Spine. 2008;33:1951-1957.

45. Horie M, Choi H, Lee RH et al. Intra-articular injection of human mesenchymal stem cells (MSCs) promote rat meniscal regeneration by being activated to express Indian hedgehog that enhances expression of type II collagen. Osteoarthritis and cartilage /OARS, Osteoarthritis Research Society. 2012;20:1197-1207.

46. Acharya C, Adesida A, Zajac P et al. Enhanced chondrocyte proliferation and mesenchymal stromal cells chondrogenesis in coculture pellets mediate improved cartilage formation. Journal of cellular physiology. 2012;227:88-97.

47. Wu L, Prins H-J, Helder MN et al. Trophic effects of mesenchymal stem cells in chondrocyte co-cultures are independent of culture conditions and cell  sources. Tissue engineering Part A. 2012;18:1542-1551.

48. Qing C, Wei-ding C, Wei-min F. Co-culture of chondrocytes and bone marrow mesenchymal stem cells in vitro enhances the expression of cartilaginous extracellular matrix components. Brazilian Journal of Medical and Biological Research. 2011;44:303-310.

49. Vézina R, Lavoie-Lamoureux A, Lavoie J-P et al. Inflammatory Stimuli Differentially Modulate The Transcription Of Paracrine Signaling Molecules Of Equine Bone Marrow Multipotent Mesenchymal Stromal Cells. Osteoarthritis and Cartilage. 2013.

50. Frisbie DD, Kisiday JD, Kawcak CE et al. Evaluation of adipose-derived stromal vascular fraction or bone marrow-derived mesenchymal stem cells for treatment of osteoarthritis. J Orthop Res. 2009;27:1675-1680.

51. Ortiz LA, DuTreil M, Fattman C et al. Interleukin 1 receptor antagonist mediates the antiinflammatory and antifibrotic effect of mesenchymal stem cells during lung injury. Proceedings of the National Academy of Sciences. 2007;104:11002-11007.

52. Le Blanc K, Ringden O. Immunomodulation by mesenchymal stem cells and clinical experience. Journal of internal medicine. 2007;262:509-525.

53. Chen X, Armstrong MA, Li G. Mesenchymal stem cells in immunoregulation. Immunology and cell biology. 2006;84:413-421.

54. Diekman BO, Wu CL, Louer CR et al. Intra-articular Delivery of Purified Mesenchymal Stem Cells From C57BL/6 or MRL/MpJ Superhealer Mice Prevents Posttraumatic Arthritis. Cell Transplant. 2013;22:1395-1408.

55. Phinney DG, Prockop DJ. Concise review: mesenchymal stem/multipotent stromal cells: the state of transdifferentiation and modes of tissue repair—current views. Stem cells (Dayton, Ohio). 2007;25:2896-2902.

56. Rehman J, Traktuev D, Li J et al. Secretion of angiogenic and antiapoptotic factors by human adipose stromal cells. Circulation. 2004;109:1292- 1298.

57. Leijs MJ, van Buul GM, Lubberts E et al. Effect of Arthritic Synovial Fluids on the Expression of Immunomodulatory Factors by Mesenchymal Stem Cells: An Explorative in vitro Study. Frontiers in immunology. 2012;3:231.

58. Agung M, Ochi M, Yanada S et al. Mobilization of bone marrow-derived mesenchymal stem cells into the injured tissues after intraarticular injection and their contribution to tissue regeneration. Knee Surg Sports Traumatol Arthrosc. 2006;14:1307-1314.

59. Kern S, Eichler H, Stoeve J et al. Comparative analysis of mesenchymal stem cells from bone marrow, umbilical cord blood, or adipose tissue. Stem cells (Dayton, Ohio). 2006;24:1294-1301.

60. Zhu Y, Liu T, Song K et al. Adipose-derived stem cell: a better stem cell than BMSC. Cell biochemistry and function. 2008;26:664-675.

61. Izadpanah R, Trygg C, Patel B et al. Biologic properties of mesenchymal stem cells derived from bone marrow and adipose tissue. Journal of cellular biochemistry. 2006;99:1285-1297.

62. Barry FP. Biology and clinical applications of mesenchymal stem cells. Birth defects research Part C, Embryo today : reviews. 2003;69:250-256.

63. Chen HT, Lee MJ, Chen CH et al. Proliferation and differentiation potential of human adiposederived mesenchymal stem cells isolated from elderly patients with osteoporotic fractures. Journal of cellular and molecular medicine. 2012;16:582-592.

64. Mirsaidi A, Kleinhans KN, Rimann M et al. Telomere length, telomerase activity and osteogenic differentiation are maintained in adiposederived stromal cells from senile osteoporotic SAMP6 mice. Journal of tissue engineering and regenerative medicine. 2012;6:378-390.

65. Koga H, Muneta T, Nagase T et al. Comparison of mesenchymal tissues-derived stem cells for in vivo chondrogenesis: suitable conditions for cell therapy of cartilage defects in rabbit. Cell and tissue research. 2008;333:207-215.

66. Winter A, Breit S, Parsch D et al. Cartilagelike gene expression in differentiated human stem cell spheroids: A comparison of bone marrow–derived and adipose tissue–derived stromal cells. Arthritis & Rheumatism. 2003;48:418-429.

67. Maumus M, Manferdini C, Toupet K et al. Adipose mesenchymal stem cells protect chondrocytes from degeneration associated with osteoarthritis. Stem Cell Research. 2013.

68. ter Huurne M, Schelbergen R, Blattes R et al. Antiinflammatory and chondroprotective effects of intraarticular injection of adipose-derived stem cells in experimental osteoarthritis. Arthritis & Rheumatism. 2012;64:3604-3613.

69. Toghraie F, Chenari N, Gholipour M et al. Treatment of osteoarthritis with infrapatellar fat pad derived mesenchymal stem cells in Rabbit. The Knee. 2011;18:71-75.

70. Desando G, Cavallo C, Sartoni F et al. Intra-articular delivery of adipose derived stromal cells attenuates osteoarthritis progression in an experimental rabbit model. Arthritis research & therapy. 2013;15:R22.

71. Uematsu K, Hattori K, Ishimoto Y et al. Cartilage regeneration using mesenchymal stem cells and a three-dimensional poly-lactic-glycolic acid (PLGA)  scaffold. Biomaterials. 2005;26:4273-4279.

72. Fan H, Hu Y, Zhang C et al. Cartilage regeneration using mesenchymal stem cells and a PLGA– gelatin/chondroitin/hyaluronate hybrid scaffold. Biomaterials. 2006;27:4573-4580.

73. Lee CH, Cook JL, Mendelson A et al. Regeneration of the articular surface of the rabbit synovial joint by cell homing: a proof of concept study. Lancet. 2010;376:440-448.

74. Sordi V. Mesenchymal stem cell homing capacity. Transplantation. 2009;87:S42-45.

75. Karp JM, Leng Teo GS. Mesenchymal stem cell homing: the devil is in the details. Cell stem cell. 2009;4:206-216.

76. van Buul GM, Kotek G, Wielopolski PA et al. Clinically translatable cell tracking and quantification by MRI in cartilage repair using superparamagnetic iron oxides. PloS one. 2011;6:e17001.

77. Al Faqeh H, Nor Hamdan BM, Chen HC et al. The potential of intra-articular injection of chondrogenicinduced bone marrow stem cells to retard the progression of osteoarthritis in a sheep model. Experimental gerontology. 2012;47:458-464.

78. Singh JA. Stem cells and other innovative intraarticular therapies for osteoarthritis: what does the future hold? BMC medicine. 2012;10:44.

79. Bellamy N, Campbell J, Robinson V et al. Intraarticular corticosteroid for treatment of osteoarthritis of the knee. The Cochrane database of systematic reviews. 2006:CD005328.

80. Barbash IM, Chouraqui P, Baron J et al. Systemic delivery of bone marrow–derived mesenchymal stem cells to the infarcted myocardium feasibility, cell migration, and body distribution. Circulation. 2003;108:863-868.

81. Jo CH, Kim EM, Ahn HJ et al. Degree of degeneration and chondroitinase ABC treatment of human articular cartilage affect adhesion of chondrocytes. Tissue Eng. 2006;12:167-176.

82. Qi Y, Feng G, Yan W. Mesenchymal stem cell-based treatment for cartilage defects in osteoarthritis. Molecular biology reports. 2012;39:5683-5689.

83. Steadman JR, Rodkey WG, Rodrigo JJ. Microfracture: surgical technique and rehabilitation to treat chondral defects. Clinical orthopaedics and related research. 2001;391:S362-S369.

84. Hoffmann A, Gross G. Tendon and ligament engineering: from cell biology to in vivo application. Regenerative medicine. 2006;1:563-574.