Effect of Leukocyte Concentration on the Efficacy of PRP in the Treatment of Knee OA: Letter to the Editor

Dear Editor: We read with interest the network meta-analysis by Riboh and colleagues titled ‘‘Effect of Leukocyte Concentration on the Efficacy of Platelet-Rich Plasma in the Treatment of Knee Osteoarthritis,’’ which compared intra-articular injection of leukocyte-rich platelet-rich plasma (LR-PRP), leukocyte-poor platelet-rich plasma (LP-PRP), and hyaluronic acid (HA) for treatment of knee osteoarthritis (OA). While the authors should be applauded for the considerable effort involved in the conduct of this research, the methodology and conclusions are questionable and warrant further clarification. The primary concern relates to the paucity of available data. The number of studies reporting Western Ontario and McMaster Universities Arthritis Index (WOMAC) outcome scores were 3 for HA, 1 for LR-PRP, 3 for LP-PRP, and 1 for placebo. Similarly, International Knee Documentation Committee (IKDC) scores were provided in 2 HA studies, 3 LR-PRP studies, 2 LP-PRP studies, and 1 placebo study. Lastly, adverse reactions were included for 2 LR-PRP studies, 3 LP-PRP studies, 4 HA studies, and 1 placebo study. Our view is that no valid conclusions can be derived from a network meta-analysis with such limited data. A second concern relates to the method of analyzing WOMAC and IKDC data. For each outcome, the reported statistic was the absolute change from baseline to final follow-up. Yet mean baseline scores in the included studies ranged from 39 to 80 for WOMAC and from 41 to 56 for IKDC. Given these differences in baseline symptom severity among studies, a given absolute WOMAC improvement from a baseline score of 39 has a vastly different interpretation than the same absolute improvement from a baseline of 80. In fact, the likely reason that LP-PRP was deemed most efficacious was because the study with the greatest WOMAC improvement also had an unusually high baseline WOMAC score. A third concern relates to study selection criteria and heterogeneity in patient characteristics. Each of the 3 LR-PRP studies that reported IKDC scores were performed by the same group using the same protocol involving repeat injections with devitalized freeze-thawed PRP, which is not characteristic of clinical practice. Comparing outcomes from studies using freshly isolated LP-PRP comprising living cells with studies that used freezethawed LR-PRP containing dead cells is not appropriate. In addition, the study by Patel et al that compared 1 versus 2 PRP injections in bilateral OA patients and the study by Kon et al that compared lowversus high-molecularweight HA should have been excluded based on the authors’ study selection criteria (exclusion 3). An example of the deleterious synergistic effects of these concerns relative to WOMAC scores is demonstrated in Table 1, which shows extremely limited data and tremendous differences in baseline symptom severity such that reporting absolute improvement as the outcome of interest is flawed and pooling data among studies is questionable. The main findings of this network metaanalysis are dominated by outcomes where no or limited direct evidence is available, which suggests high risk of bias. Inclusion of nonrandomized trials in this systematic review compounds this risk. Finally, since all surface under the cumulative ranking curve (SUCRA) data were derived from the pooled meta-analysis data that suffer from these same data limitations, the SUCRA data should be considered unreliable. In summary, a more appropriate conclusion to this manuscript would have been that conduct of a network metaanalysis is inappropriate and no reasonable conclusions can be drawn regarding the effect of leukocyte concentration on the safety and efficacy of PRP in knee osteoarthritis.