Intra-articular injection of expanded autologous bone marrow mesenchymal cells in moderate and severe knee osteoarthritis is safe: a phase I/II study

This is a phase I prospective open-label safety study. The study was prospectively registered in clinicaltrial.​gov (reference NCT02118519).

After an IRB approval, a signed informed consent was obtained in accordance with the latest version of Helsinki Declaration. From February 2014 to August 2014, 13 adult patients, 7 females and 6 males with moderate to moderately severe knee, were enrolled in a prospective study using autologous expanded bone marrow mesenchymal stem cells (BM-MSCs) delivered by percutaneous intra-articular injection using lateral tibio-femoral approach by an experienced orthopedic surgeon.

KOA staging was done in accordance with the Kellgren and Lawrence classification [11] using standard knee x-ray imaging with the standing anteroposterior projection and horizontal lateral projection. Image interpretation and staging were independently done by two radiologists. No patient with significant varus or valgus malalignment or significant effusion of either knee or both knees was included.

Inclusion and exclusion criteria are shown in Table 1.

Bone marrow (BM) aspirate was done in an outpatient setting using local anesthetic of 2% lidocaine. A total of 35–40 ml of bone marrow was obtained in multiple small aspirate of 3–5 ml each from the iliac crest. The samples were collected in sterile citrated tubes of 3.8%. Prior to collection, the patient had to have normal prothrombin time (PT), partial thromboplastin time (PTT), and platelet count.

BM aspirates were diluted in a 1:1 ratio with phosphate-buffered saline (PBS) pH 7.4 (Gibco, USA, Cat # 10010-015). Mononuclear cells (MNCs) were separated by density gradient centrifugation using Ficoll-Paque (Histopaque 1077, Sigma, Cat. 10771). MNCs were counted and seeded at a density of 0.16 × 10⁶ cells/cm² in T1 75 cm² tissue culture flask (NUNC, USA) in complete media. The complete media consist of α-minimum essential medium (α-MEM) (Gibco, Cat. 22561-021) supplemented with 100 IU penicillin and 100 IU streptomycin (Gibco), 10% FBS qualified (Gibco, Cat. 12763017), and 2 Mm L-glutamine (Gibco, Cat. 25030081). Cells were allowed to attach for 24 h before changing media. Subsequently, the culture medium was changed twice a week. When cultures reached 70–80% confluence, subculturing was performed using trypsin-EDTA 0.25% (Gibco, USA, Cat. 25200056). After the primary passage, cells were seeded at a density of 4 × 10³ cells/cm². Cells were cultured until the number reaches the clinical grade with an average number of 30.5 × 10⁶ cells per dose injected per patient. At harvest, cells of all patients were in passages lower or equal to 4. Prior to injection, MSCs were tested for endotoxin, mycoplasma, and microbial contamination. For injection, MSCs were washed and suspended in 5 ml 0.9% normal saline.

Patients were followed for adverse events by direct questioning on days 1, 7, 14, 28, 60, and then every 6 months until month 24. The patient was assessed by physical examination just before the second injection and 2 months after the second injection. Blood count and clinical chemistry were done 3 and 24 months after the first injection. For outcome, normalized KOOS [13] was used at baseline before injections and at months 1, 2, 4, 6, 12, and 24 after the first injection.

MRI scans were done at baseline, 6 and 12 months using 3 T Siemens scanner. Standard knee MRI imaging protocol was obtained in axial, coronal, and sagittal planes, in addition to using a specific cartilage sequence which is T1-weighted FS spoiled 3D gradient echo in axial and sagittal planes. Detailed measurements were obtained from each compartment from three points: anterior, central, and posterior. The mean thickness was calculated. Identical sequences and measurement sites were done on the follow-up scans.

For statistical analysis, IBM SPSS software version 20 was used. The data was described as mean and standard deviation. Univariate analysis between baseline and pre-specified time points was performed. Confidence interval was set at 95%.

All isolated patients of BM-MSCs were positive for MSC signature markers determined by ISCT; CD90, CD105, CD73, and CD44 and were negative for CD34, CD45, CD11b, CD19, and HLA-DR (Fig. 1).

The data shown are the representative cell phenotype analyzed at passage 4. Gray peaks correspond to the isotype control, and the pink peaks to the antibody of interest.

Upon induction of differentiation, BM-MSCs were differentiated into adipocytes and osteocytes (Fig. 2).

Total of 13 adult patients were enrolled, 7 females and 6 males, with stage II (5 patients) and stage III (8 patients) with mean age of 50 years (range 34–63 years were included in the study. Details of patients are shown in Table 2.

There were total of three adverse events; all were local events. Two instances in which patients had pain in the injected joint within 2 h which needed cold compresses and resting the joint for several hours. The pain disappeared within 24 h. There was one instance in which the patient developed moderate pain and mild knee swelling 6 h after receiving the injection needing cold compresses and mild oral analgesia. The pain disappeared within 48 h. No clinical or biochemical adverse events were noticed after 2 years of follow-up.

Table 3 shows the positive changes related to all five major areas measured by normalized KOOS. All were significantly better at 6, 12, and 24 months post first injection.

At 6 months, there was no significant change in the cartilage thickness by MRI. At 12 months, a significant improvement in the thickness of knee cartilage in the femoral and tibia plates was noticed as shown in Table 4. Only one female patient deteriorated by MRI despite of KOOS improvement.