Background
Intra-articular injection of Platelet-rich plasma (PRP) has shown benefits for pain relief and functional improvement compared with placebo in individuals with osteoarthritis (OA). However, two recently published meta-analyses conclude that intra-articular injection of PRP, compared with the hyaluronic acid (HA) injection, exerted inconsistent effects on joint symptoms or structural changes. The inconsistencies in treatment effects observed across PRP trials may be attributed to heterogeneity in participant characteristics and PRP preparations. Identifying treatment-responsive subpopulations of OA that are most likely to benefit from PRP treatment, and determining the optimal parameters for PRP preparation and administration, are urgently needed.
MBMA and traditional meta-analysis is that the former infuses pharmacologic rationality into the statistical rigor of meta-analysis data integration8. The advantages of MBMA over conventional subgroup analysis include: First, by establishing a covariate model of MBMA, heterogeneity between studies can be adjusted, reducing the bias caused by factors such as patient characteristics and study design. Second, MBMA can handle multiple treatment comparisons simultaneously, while conventional subgroup analysis can only compare two treatments at a time. Third, MBMA can incorporate effect modifiers into the model, which allows for a more precise estimation of treatment effects and identification of subpopulations that are more likely to benefit from a particular treatment.
Therefore, the aim of the current study is to compare the MBMA and conventional subgroup analysis methodologies for identifying treatment efficacy moderators using individual patient data (IPD) from randomized controlled trials (RCTs) of PRP treatment for symptoms in individuals with knee or hip OA. This study will provide valuable guidance for clinical practice and the design of future trials.
Methods
We will conduct the analyses by integrating IPD of relevant published RCTs and MBMA method, which can provide more reliable, precise, and detailed results by reducing bias, increasing power, and improving transparency. A systematic literature search will be performed through several literature databases. We will include interventions involving intra-articular injection of PRP for knee or hip OA treatment, in comparison to any intra-articular injection of placebo, positive control drug treatments or other non-surgical treatments. The primary outcome will be change in pain from baseline at short (3 months), and long-term (24 months). The Visual Analogue Scale (VAS) pain score will be preferentially used for the analysis. If unavailable, the pain subscales from The Western Ontario and McMaster Universities Arthritis Index (WOMAC), Knee Injury and Osteoarthritis Outcomes Score, or other Likert pain scores will be converted to a 0 to 100 scale as per former OA Trial Bank protocols. Secondary outcomes will be WOMAC function, stiffness, incidence of adverse events, structural changes (cartilage volume, cartilage defects, bone marrow lesions, effusion-synovitis, etc) and global patient assessment at each time point.
We will select an appropriate structural model based on the data characteristics to depict the temporal changes of efficacy indicators. Covariate models will be established to explore whether there are potential influences on pharmacodynamic parameters and eliminate heterogeneity between trials. The covariates will include: age, sex, BMI, disease duration, baseline K-L grade, treatment duration, injection site (knee or hip), injection frequency, leukocyte removed from PRP or not, PRP activated or not (addition of calcium gluconate or CaCl2).
Subgroup analyses will be performed under the MBMA method frame for the primary outcome at both short-term (3 months) and long-term (24 months) follow-up. We will conduct subgroup analyses of potential influencing factors that are of clinical interest.
The analytical method will consist of two steps. First, individual parameters and SEs for each study will be obtained after eliminating covariate effect differences using Bayesian post hoc estimation. Second, a meta-analysis with a random-effect model will be used to summarize the typical values of the pharmacodynamic parameters and their SEs for each predefined subgroup. Based on the results of these analyses, the typical response for each subgroup will be obtained using 1000 Monte Carlo simulations (PROSPERO ID: 628441).
Status
Ongoing. We are currently screening studies for inclusion to our review