Study Design and Population
This prospective clinical trial, conducted at a single center, involved three study arms. Outpatients with a history of knee pain complaints underwent a thorough evaluation, including clinical history, physical examination, laboratory tests, and X-rays. Participation in the study was contingent upon patients' understanding and agreement with the treatment method, confirmed through the signing of informed consent. Inclusion criteria for the study were established after a meticulous review of test results, requiring (1) adult patients with symptomatic knee osteoarthritis (OA), Kellgren-Lawrence (KL) grade 2 to 4, and (3) symptoms persisting for at least 12 months. Exclusion criteria were also defined, excluding patients with (1) knee instability, (2) severe misalignment, (3) inflammatory arthritis such as rheumatoid arthritis and ankylosing spondylitis, and (4) underlying diseases such as hematologic disorders, septicemia, coagulopathy, neoplasm, active infection, and immune deficiency.
A total of 195 diagnosed OA patients meeting the inclusion criteria were sequentially selected and treated. However, 20 patients were lost to follow-up during the study process and were not included in the final calculations (see Flow diagram in Figure 1.)
In the initial round, patients received treatment with BMAC. In the second round, when PRP sets and HA injections became available, patients were randomly allocated to two groups: HA and PRP. Randomization at this stage utilized R.3.3.0 version software. All groups were monitored using the same methodological criteria. The study adhered to the Declaration of Helsinki for medical research involving human subjects and received approval from the Research Ethics Committee (EC KCV 319-321/16). Registration on ClinicalTrials.gov (NCT03825133) was completed, and the results presented in this paper are an excerpt from this registered study.
BMAC Processing Procedure
For the procedure, the patient was positioned in a supine posture after the tibial tubercle was prepared and draped. Local anesthesia (a combination of 2 mL Lidocaine and 10 mL Marcaine) was administered from the skin to the periosteum following the infiltration. A small incision was made using an 11-blade, and a trocar was inserted into the cancellous part of the bone. Two 50 mL syringes were filled with 10 mL of the anticoagulant acid citrate dextrose formula A (ACD-a). By maneuvering the trocar in a bottom-up direction to create more harvesting sites, approximately 80 mL of autologous bone marrow was aspirated (resulting in two syringes of bone marrow; totaling 40 mL + 10 mL of anticoagulant). The bone marrow underwent filtration through a 150-micron filter to eliminate coagulum and potential bone fragments. The filtered bone marrow was processed using the Arthrex Angel separating system, yielding a product of about 5–6 mL of Bone Marrow Aspirate Concentrate (BMAC). Subsequently, the obtained BMAC was injected into the targeted knee.
PRP Procedure
To prepare Platelet-Rich Plasma (PRP), 60 mL of peripheral blood was drawn into syringes containing ACD-A in a 7:1 ratio. The whole blood underwent processing using a fully automated Angel® system (Arthrex®), where the G-force and spinning time were automatically adjusted to separate the desired volume of blood into three fractions: PRP, Red Blood Cells (RBC), and Platelet Poor Plasma (PPP). The goal was to obtain a PRP substrate with a platelet count 6–8 times higher than the baseline, a setting automatically configured on the device. The average platelet count in the PRP group was 320.37 × 106, and the leukocyte (Le) count was 7.07 × 106. Following the centrifugation and concentration procedure, the average platelet count increased to 2179.31 × 106, and the average leukocyte count was 16.16 × 106. The average concentration rate for platelets was 7.23 times the baseline, and for leukocytes, it was 2.22 times the baseline. Therefore, our formulation can be characterized as Leukocyte-Rich Platelet-Rich Plasma (LR-PRP) in accordance with Ehrenfast classification (21).
HA Injection
For this group, HA with the Cartinorm® brand name was injected. Cartinorm® is manufactured by Goodwill Pharma, Hungary. It is a viscous solution consisting of a high molecular weight fraction of purified 1% natural sodium hyaluronate (4000 kDa) in buffered physiological sodium chloride, having a pH of 6.8–7.6. The sodium hyaluronate was supplied as a sterile, non-pyogenic solution in 2 mL pre-filled syringes containing 20 mg of sodium hyaluronate, 16.6 mg of sodium chloride, 0.52 mg of potassium dihydrogen phosphate, 2.8 mg of dibasic sodium phosphate dihydrate, and up to 2 cc water for injection. Injections were administered to a total of 3 treatments, 1 every week for 3 weeks.
DataCollection
Patients were asked to self-report SF-36 scale before intervention and they were scheduled to be followed-up after 1, 3, 6, 9, and 12 months. Patients were asked to report any potential adverse events starting from the procedure to the end of the study. Adverse events were described as conditions connected to the procedure itself or to the nature of the disease.
Data Analysis
As mentioned previously, the data for this study comprised repeated measurements of the self-reported SF-36 scale. Due to such nature of the data, linear mixed models (LMMs) were used in order to assess which of the three injection types was the most effective in improving the quality of life. LMMs build upon the foundation of linear regression by allowing for correlated observations and the analysis of data collected at multiple time points. Crucially, LMMs have the capability to include both fixed effects, which are consistent across all patients, and random effects, which account for variations within each patient’s repeated measurements.
For the study, we firstly performed exploratory analysis, as to further understand the distribution of the 9 variables measure by the SF-36 scale. This included box plots for physical functioning, role limitations due to physical health, role limitations due to emotional problems, energy/fatigue, emotional wellbeing, social functioning, pain, general health and health change. The values of each of the nine variables were split by injection type, and the measurements were shown for all time points. Secondly, we then fitted nine linear mixed models, where each of the nine variables from the SF-36 scale were considered to be the outcome variables. Additionally, each of the models included time and injection type as covariates with fixed effects, whereas the intercept in all models was considered to be a random effect to allow for patient-specific variation.