Platelet-rich plasma versus hyaluronic acid in the treatment of knee osteoarthritis: a meta-analysis

We systematically searched electronic databases including PubMed, Embase, Web of Science and the Cochrane Library on January 23, 2020 to identify relevant studies issued in English languages. The search strategy was made for the use of Medical Subject Headings (MeSH) terms and correspondence keywords. The search strategies can be seen in Supplement S1. We also searched relevant reviews and meta-analyses to identify other eligible studies.

Two investigators (Jia Zhu Tang and Ming Jun Nie) independently conducted the initial searches and screened the titles and abstracts for selecting eligible studies. In addition, the reference lists from all the original articles and identified reviews were also manually scanned for additional relevant studies. If either reviewer observed a title or abstract meeting the eligibility criteria, full text of the study was retrieved.

Two reviewers independently extracted data from each study using a standardized data extraction form. Disagreements were resolved by discussion, and those unresolved through discussion were reviewed by a third reviewer. The following variables were included: first author, publication year, country, number of participants, age, sex, body mass index, radiographic classification, and follow-up. Moreover, we collected injection doses, times, and intervals of PRP and HA injections. Primary outcomes included WOMAC pain and WOMAC function scores at 1, 3, 6, and 12 months. Secondary outcomes were WOMAC total scores, WOMAC stiffness scores, IKDC scores, VAS scores, EQ-VAS scores, KOOSs, Lequesne Index scores, and adverse events. We intended to contact authors for detail information when the reported data were inadequate.

Two reviewers (Jian Zhong Zhao and Guang Cheng Zhang) independently evaluated the risk of bias of each RCT by the Cochrane Risk of Bias tool. Each article was assessed based on the following seven items: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective reporting (reporting bias), and other bias. Every item was scored as high, low, or unclear. Any discrepancies shall be settled by consensus between the two reviewers. If necessary, a third reviewer shall be consulted (Bo Wang).

As for outcomes measurements (WOMAC pain, function, and total scores; IKDC scores; and Lequesne scores) in each study, the mean difference between the baseline and post intervention was calculated to compare the efficacy of intervention with the control group. The present study was conducted by using Stata 12.0 (Stata Corp., College Station, TX, USA) for meta-analysis. The standard mean difference (SMD) and 95% confidence interval (CI) were calculated for continuous outcomes (WOMAC pain, function, stiffness, and total scores; IKDC scores; VAS scores; EQ-VAS scores; KOOSs; and Lequesne scores), while risk ratios (RRs) with 95% CIs were adopted for dichotomous outcome (adverse events). Heterogeneity among studies was accessed by using the I² statistic. The I² values of 25%, 50%, and 75% were respectively considered to be the cut-off points for low, moderate, and high heterogeneity. Random effects method was used for all meta-analyses. P value < 0.05 was considered to be of statistical significance.

Figure 1 shows the details of the literature search. A total of 190 records were identified as potentially relevant studies. By removing duplicates, scanning titles, and reading abstracts, 83 full-text articles were assessed for eligibility. Ultimately, 19 RCTs [17, 19, 20, 22‐38] were included for data extraction and meta-analysis.

The study characteristics are available in Table 1. These studies were published from 2011 to 2019 with a total of 1281 patients. The sample size of the studies ranged from 10 to 104. There were 654 patients in the PRP injection group and 627 patients in the HA injection group. The follow-up period ranged from 3 to 18 months. The demographic features between the two groups in each study were similar. The administrated timing and dosage of PRP and HA injections are shown in Table 2, which were varied among these studies.

Figures 2 and 3 reveal the risk of bias summary and graph of all included RCTs. Among the twenty RCTs, random sequence generation were recorded adequately in eighteen studies [17, 19, 20, 22, 23, 25‐28, 31‐39] and unclear in two studies [24, 30]. Allocation concealment were recorded adequately in eleven studies [17, 19, 20, 23, 27, 28, 32, 34, 36‐38], unclear in seven studies [22, 25, 26, 30, 35, 39], and high in two studies [24, 33]. The performance bias was recorded adequately in thirteen studies [17, 19, 20, 23, 27, 28, 30‐32, 34, 38], unclear in five study [26, 35‐37, 39], and high in two studies [24, 33]. Only four studies [24, 27, 33, 35] reveal unclear risk of bias for other bias.

Table 3 shows the results of subgroup analyses for the WOMAC total scores at 12 months. The subgroup analyses based on the number of PRP injections, PRP spinning approach (single or double), PRP classification (LP-PRP or LR-PRP), fresh or frozen PRP, whether an activator was used, and risk of bias (low or unclear/high) were conducted for exploring the WOMAC total scores at 12 months.

The findings of WOMAC pain, stiffness, function, and VAS at 12 months were consistent in all subgroup analyses except for the number of PRP injections, PRP single spinning approach, LP PRP, without using an activator, molecular weight, and structure subgroups. In the subgroups of ≥ 2 injections, double spinning approach, LP-PRP, and activator use, we found that PRP was associated with significantly better WOMAC than HA at 12 months (Table 3).

In the subgroup analyses dividing into low and high molecular weight groups, results showed a PRP significant reduction in WOMAC at 12 months than HA with low molecular weight (SMD − 1.31, 95% CI − 1.97 to − 0.64], P = 0.000, I² = 92.2%, Table 3), but not in HA with high molecular weight (SMD − 1.90, 95% CI = − 4.53 to 0.73, P = 0.157, I² = 97.9%, Table 3). Moreover, there was a PRP significant reduction in WOMAC at 12 months than HA with no cross-link or unclear (SMD − 0.57, 95% CI − 1.78 to 0.63, P = 0.000, I² = 95.0%, Table 3), but not in HA with cross-link (SMD − 2.06, 95% CI = − 3.20 to − 0.92, P = 0.352, I² = 96.2%, Table 3).

The pooled results showed that intra-articular PRP injection appeared to be more efficacious than HA injection for the treatment of KOA in terms of short-term functional recovery. Moreover, PRP injection was superior to HA injection in terms of long-term pain relief and function improvement. In addition, PRP injection did not increase the risk of adverse events when compared with HA injection. The level of evidence, which was undermined by heterogeneity and/or study design limitations, was moderate or low, indicating that the degree of benefit must be studied although the benefit is conclusive. PRP is an autologous concentrate of human platelets isolated through centrifugation of the patient’s blood, containing numerous components containing variety of growth factors, cytokines, and many other bioactive proteins [40].

Based on preclinical research, it is known that PRP ameliorates the degeneration of cartilage by stimulation of mesenchymal stem cell migration, proliferation, and differentiation into articular chondrocytes. PRP affects the progression of KOA via inhibition of inflammatory cytokines and altering the level of enzymatic expression and thus promotes cartilage repair [41]. Moreover, several clinical trials and systematic reviews have demonstrated that PRP have the ability to relive osteoarthritic symptoms up to 12 months postinjection, including pain, stiffness, and function failure [42].

HA is the most important component of articular fluid and responsible for the viscoelastic and lubricant capabilities in joints [43]. It is involved in chondroprotection, proteoglycan, and glycosaminoglycan synthesis as well as anti-inflammation. In addition, intra-articular HA injection can significantly reduce the apoptosis rates of chondrocytes [44]. Clinical researches have shown HA injection in patients with KOA has the potential to reduce knee pain, improve function, and quality of life [39].

Interestingly, numerous studies have focused on the clinical efficacy between PRP and HA in the KOA treatment. Duymus et al. [26] compared the efficacy of intra-articular injections of PRP with HA for KOA treatment. They found that PRP injection was more successful than HA injection in the treatment of mild–moderate knee OA. PRP application could provide at least 12 months of pain-free daily living activities. Similarly, Lin et al. [17] investigated the discrepancy between PRP and HA in therapy of KOA and suggested that intra-articular injections of leukocyte-poor PRP (LP-PRP) improved function recovery for at least 1 year in patients with mild-to-moderate osteoarthritis of the knee. Furthermore, Ahmad et al. explored the clinical outcomes of PRP injection with changes in the ultrasonography structural appearance [22]. They observed that intra-articular injections of PRP were associated with improved synovial hypertrophy and vascularity scores and less effusion. However, PRP injection failed to perform better efficacy than HA injection in several clinical studies. Filardo et al. [19] found that the patients with PRP injection could not obtain a better clinical outcome than those treated with HA injection. With a long-term follow-up of 5 years, Di Martino et al. [25] concluded that PRP injection did not provide an overall superior clinical improvement compared with HA injection in terms of functional improvement at any follow-up point. Although LP-PRP injection showed more effective in terms of clinical improvement with respect to HA injection, there was no influence on the X-ray and MRI performance of cartilage progression at 52 weeks follow-up.

Therefore, it still remains a contradiction whether PRP injection is superior to HA injection in the treatment of KOA. Previous systematic review and meta-analysis also evaluated the efficacy of PRP injection compared with HA injection in the treatment of KOA. Laudy et al. [45] enrolled 10 trials and found that PRP injection performed better clinical outcomes than HA injection on pain reduction at 6 months postinjection. Recent meta-analysis by Han et al. [44] pooled 14 RCTs and suggested that PRP injection might be more effective with respect to HA injection in terms of long-term pain relief and functional improvement. The biggest flaw of this meta-analysis was that they included bilateral knee OA, and thus, a large clinical heterogeneity existed. However, Zhang et al. [13] analyzed 13 studies (10 RCTs and 3 non-randomized studies) and concluded that PRP injection was not obviously superior to HA in KOA. Pooled RCTs and non-RCTs for meta-analysis violates the PRISMA guideline for meta-analysis, and selection bias is ineluctable. The present study is, to our knowledge, the most comprehensive, up-to-date, and with the largest sample size (n = 19; totally, 1281 patients) meta-analysis undertaken to estimate the efficacy and safety of PRP versus HA in OA.

Limitations of current meta-analyses should be noted. Due to the limited evidence available, previous meta-analysis data extracted from retrospective studies [46] and even case series [47], which might bring significant bias for the overall analysis. Most comparisons included only 1 or 2 studies due to the small number of clinical trials pooled for meta-analysis. More RCTs are responsible for the evaluation of the efficacy of PRP injection on pain relief and function improvement compared to HA injection. Our meta-analysis included 20 RCTs to investigate the efficacy of PRP injection on pain relief and function recovery compared with HA injection in patients with KOA. In short-term period postinjection (no more than 3 months), PRP injection resulted in better WOMAC function score at 1 month and WOMAC stiffness function score, stiffness score, and IKDC at 3 months compared with HA injection. PRP injection and HA injection had similar effects with respect to the WOMAC pain scores, WOMAC total scores, and VAS scores at 1 month and 3 months. And also, the patients with PRP injection showed similar effects in IKDC and EQ-VAS scores at 2 months, and KOOS at 3 months. In long-term period postinjection (no less than 6 months), we found that better clinical results were achieved in the PRP injection group compared with the HA injection group in terms of WOMAC pain, function, stiffness, and total scores and VAS scores at 6 months and 12 months. Moreover, the patients with PRP treatment showed better performance with respect to IKDC at 6 months and EQ-VAS at 12 months. Nevertheless, there was no significant difference between groups in terms of bias for the overall analysis. Most comparisons included only 1 or 2 studies due to the small number of clinical trials pooled for meta-analysis. More RCTs are responsible for the evaluation of the efficacy of PRP injection on pain relief and function improvement compared with HA injection.

Another issue that affects the effects of PRP is the leukocyte concentration in PRP composition, which may contain more proinflammatory cytokines and be detrimental to cartilage repair. Subgroup analyses were used to identify the potential heterogeneity of the results. A study conducted by Riboh et al. [48] compared LP-PRP and LR-PRP in the treatment of KOA and found that LP-PRP injections resulted in significantly improved WOMAC scores compared with HA or placebo. These results were consistent with our subgroup findings.

Vilchez-Cavazos et al. [45] conducted a meta-analysis and revealed that single injection was as effective as multiple PRP injections in pain improvement; however, multiple injections seemed more effective in joint functionality than a single injection at 6 months. Similarly, most of these results were also consistent in different PRP spinning approaches, whether with an activator use and PRP species (fresh or frozen), which suggested that these factors might have little influence on the efficacy of PRP.

Some limitations in the current study should be interpreted. Firstly, the main limitation is that most overall analyses are accompanied with high heterogeneity. The high heterogeneity among pooled results has weakened the persuasion of the conclusion. Although we tried to compensate for methodological deficiencies by performing stratified analyses, some results remained inconclusive since several reports lacked the documentation of the key factors. Secondly, although 19 RCTs were included in this study, some indexes such as WOMAC function and stiffness score at 1 month, IKDC at 3 months, EQ-VAS at 2 months, and KOOS at 3 months were analyzed by the data extracted from only two studies. Moreover, similar to the previous meta-analysis studies, we evaluated the efficacy between PRP and HA within 1 year on the account of limited follow-up. Some RCTs explored the long-term follow-up (52 weeks by Buendia-Lopez et al. [23] and 5 years by Di Martino et al. [14]). However, we were unable to pool the long-term results from the limited data. Thirdly, almost all included RCTs used subjective questionnaires to deduce the treatment effects. Objective findings such as magnetic resonance and ultrasound seem to be needed in the efficacy evaluation. Finally, the administration of PRP injection was varied in the included RCTs. The present study failed to recommend the optimal administration dosage and interval because of the insufficient data. Therefore, more well-designed RCTs with long-term follow-up are still necessary.