Introduction

Knee osteoarthritis (KOA) is a chronic joint condition characterized by cartilage degeneration and an increase in bone growth in the knee joint [1, 2]. The knee joint’s primary symptoms include discomfort, swelling, and mobility problems. As the population ages, more people are developing KOA, which has a major impact on middle-aged and older people’s health and quality of life [3,4,5]. As a multifactorial disease that develops over a long period of time [6], KOA has always been a huge burden on individuals and society as a whole due to its high disability rate [7].

Currently, intra-articular injection (IAI) remains the primary element of non-surgical therapy for KOA [8]. The evidence that is now available demonstrates that this therapy can significantly reduce short-term pain for patients with KOA and improve joint function while also having a minimal risk of patient injury [9, 10]. Interestingly, botulinum toxin and ozone have also been proven to be used for injection into joints to treat KOA [11, 12]. HA, a naturally occurring glycosaminoglycan, serves as a crucial component of synovial fluid in joints, functioning as a lubricant and a shock absorber with elastic properties during joint movement [13]. In addition, A has the following functions: proteoglycan and glycosaminoglycan synthesis, anti-inflammatory, mechanical, subchondral, and analgesic actions [14]. HA is a widely used conservative treatment for OA because of both its indirect and direct analgesic effects on joints. Many clinical studies have shown that HA supplementation has a good effect on KOA, but HA may increase the risk of adverse events, such as transient pain at the injection site [15]. Knee Joint injection of CSC has a lasting effect of weeks to months [16]. The anti-inflammatory and immunosuppressive effects of corticosteroids are obvious [17], and CSC can raise the knee joint’s relative viscosity and HA concentration [18]. Regarding the intra-articular CSC’s effective duration, there is disagreement. IAI of PRP has gained widespread attention in recent years as a novel and successful alternative therapy for patients with KOA [19]. The mechanism of local injection of PRP is that it can relieve joint pain and reduce synovial hyperplasia and effusion in the joint cavity [20]. PRP is considered to have a variety of important physiological functions, such as anti-inflammation, analgesia, and promoting chondrocyte proliferation and cartilage repair. Besides, PRP can also regulate the progression of KOA by regulating WNT and IL-1 signaling [21]. In recent years, scholars have combined them to investigate the possibility of dual therapy [22]. Wang et al. discovered that individuals taking hyaluronan and corticosteroids together had pain alleviation and improved knee function faster than either medication alone. At 6 months, however, there was no discernible difference [23]. Huang et al. discovered that whereas corticosteroids and hyaluronan were equivalent in terms of pain alleviation after three months, PRP injections were superior in terms of long-term pain relief [24]. John et al.‘s study found that PRP has better efficacy than HA [13], but another study found no difference between the two [25]. Overall, there are still many controversies in this field, and there is an urgent need for an article to integrate all the evidence and provide a credible recommendation.

In this study, a Bayesian network meta-analysis of randomized controlled trials (RCTs) was conducted to evaluate the effectiveness and safety of CSC, HA, PRP, and their combination in treating KOA.

Materials and methods

Ethical approval

This meta-analysis did not need ethical approval since no new clinical raw data were collected or used; rather, the analysis was conducted based only on previously published research that had already been granted ethical approval.

Literature search

In accordance with the PRISMA checklist [26], a comprehensive search was carried out in the Web of Science databases, Embase, the Cochrane Library, PubMed, and the Wanfang database to collect English publications until December 2022. The search criteria consisted of keywords such as “corticosteroids OR steroids OR hyaluronic acid OR platelet rich plasma OR PRP OR placebo (PLA)” and the condition of interest, “knee OR osteoarthritis OR KOA”. To find more pertinent literature, a manual search and literature tracking techniques were also performed. Supplemental File 1 provides details of the search strategy.

Inclusion and exclusion criteria of literature

The following were the study’s inclusion criteria: (1) RCTs involving patients with KOA; (2) original research; (3) studies that reported at least two of the following treatments: HA, CSC, PRP, combination therapy, and/or placebo; and (4) includes VAS OR WOMAC outcome scores or the proportion of patients who had adverse effects. The following were the exclusion criteria: (1) literature review; (2) non-randomized studies; (3) failure to get original data; and (4) low-quality or duplicate publications. Two authors conducted an independent search of all references and any disagreements were resolved by a vote of all authors.

Data extraction

Two authors (XQ and LY) conducted data extraction independently, discussed their findings, and reached an agreement in case of any disagreements. Each qualifying study’s first author, publication year, country, methods of treatment, length of time, sample size, outcome measures, and follow-up time points were all recorded.

Methodological quality assessment

Two authors (XL and LY) independently evaluated the quality of the included literature, and a third researcher was invited to help resolve any differences. Review Manager Software5.4 (The Nordic Cochrane Collaboration, Copenhagen)’s risk of bias summary was used to examine the following biases: sequence generation, allocation concealment, blinding, incomplete outcome data, no selective outcome reporting, and other sources of bias. Each criterion was judged to have a low, unclear, or high risk of bias.

Statistical analysis

Data synthesis

Stata 17.0 was used for data processing and analysis, and to draw related graphs [27]. For dichotomous variable data, we estimated the odds ratio (OR) with 95% confidence intervals (CIs), and for continuous variable data, we estimated the standardized mean differences (SMD) with 95% CIs. The initial model update iteration number was set to 10,000, and the continuous update iteration number was set to 10,000. To mitigate the impact of the starting value, the first 10,000 annealing times were utilized, and sampling began after 10,001 times. We calculated the relative ranks of the intervention groups using a consistency model and then displayed the percentages of the surface under the cumulative ranking curve (SUCRA). We conducted a network meta-analysis for each outcome only when the intervention groups could be connected to create a network; however, comparisons of support surfaces allocated to the same group were not excluded from the overall systematic review.

Assessing the certainty of evidence

A detailed review of the completeness of the literature search was used to estimate the possibility of publication bias. This involved creating funnel plots for each paired meta-analysis that contained more than 10 studies, as well as a network-adjusted funnel plot. Furthermore, the depth of the literature search and the amount of unpublished data acquired were considered.

Results

Literature search

Out of 1097 RCTS pertaining to KOA identified through the database search, 1062 were eliminated for diverse reasons, including 385 duplicates, while 712 articles were screened by title and abstract, thereby resulting in the exclusion of 599 irrelevant studies. Afterward, a thorough examination of 113 articles led to the elimination of 8 articles that lacked an index of existing data, 46 articles that did not present the outcome of interest, and 24 articles that were not connected to the outcome. This ultimately brought the meta-analysis down to 35 studies. Figure 1 illustrates the particulars of the literature search.

Fig. 1
figure 1

Flowchart of the study procedure

Full size image

Patient demographics and methodological quality assessment

Included were 35 RCTs with 3104 patients from 16 countries in total. The mean age of the enrolled patients was 59.1 years, and 61.3% of them were female. The course of treatment ranged from 3 to 24 months. Follow-up time reached 3 months in 35 studies, 6 months in 31 studies, 9 months in 14 studies, 12 months in 14 studies. Table 1 presents a comprehensive list of included studies along with their characteristics. The majority of studies utilized blinding techniques. Furthermore, the hazards of attrition, reporting, and unidentified bias are minimal. Methodological evaluations had a minimal risk of bias and were of high quality. Figure 2 depicts the methodological quality evaluation.

Table 1 Characteristics of 36 studies included in the meta-analysis
Full size table
Fig. 2
figure 2

Summary of the risk of bias

Full size image

WOMAC scores

At the 3-month follow-up, 1319 patients were included in the study, with 15 reported WOMAC scores. The PRP groups performed the best in terms of the outcomes (SMD=-8.79; 95% CI-15.69~-1.89), followed by PRP + HA (SUCRA value, 61.2; mean rank, 2.6); HA (SUCRA value, 48.9; mean rank, 3); PLA (SUCRA value, 38.2; mean rank, 3.5); and CSC (SUCRA value, 17.3; mean rank, 4.3).

Twenty reported WOMAC scores at 6 months of follow-up, including a total of 2310 patients, the best outcomes were shown in the PRP groups (SMD=-11.92; 95% CI: -19.16~-4.69), which were followed by PRP + HA (SUCRA value, 64.2; mean rank, 2.4), HA (SUCRA value, 50.2; mean rank, 3.0), PLA (SUCRA value, 39.9; mean rank, 3.4), and CSC (SUCRA value, 6.7; mean rank, 4.7). Ten reported WOMAC scores at 12 months of follow-up, including a total of 1148 patients, the PRP groups performed the best (SMD=-7.04;95% CI: -9.38~-4.70), followed by PRP + HA (SUCRA value, 69.0; mean rank 2.2), HA (SUCRA value, 42.8; mean rank, 3.3), PLA (SUCRA value, 42.0; mean rank, 3.3), CSC (SUCRA value, 0.0; mean rank, 5.0). Table 2; Fig. 3 provide summaries of the network meta-analysis findings. No discrepancy between the direct and indirect effects of any intervention was observed as per the nodal analysis of the intervention measures (P > 0.05). Figure 3 compares the results based on the WOMAC scores at the 3, 6, and 12-month follow-ups.

Table 2 Network meta-analysis treatment ranking results for each of WOMAC scores, VAS scores and adverse effects
Full size table
Fig. 3
figure 3

Overall network comparisons using WOMAC scores

Full size image

VAS scores

In total, 1099 patients reported 15 VAS scores after 3 months of follow-up. PRP + HA had the best outcomes, with a SUCRA value of 70.5 and a mean rank of 2.2, followed by PRP (SUCRA value, 57.6; mean rank, 2.7), CSC (SUCRA value, 48.2; mean rank, 3.1), HA (SUCRA value, 47.4; mean rank, 3.1), and PLA (SUCRA value, 26.3; mean rank, 3.9). Eighteen reported VAS scores at 6 months of follow-up, including a total of 1732 patients, the PRP + HA groups showed the best outcomes (SUCRA value, 81.8; mean rank 1.7), followed by CSC (SUCRA value 56.7; mean rank, 2.7), PRP (SUCRA value, 50.7; mean rank, 3.0), HA (SUCRA value, 48.0; mean rank, 3.1), PLA (SUCRA value, 12.7; mean rank, 4.5). At the 12-month follow-up, a total of 656 patients reported 8 VAS scores, with the PRP + HA groups displaying the most favorable outcomes (SUCRA value, 85.5; mean rank, 1.4), followed by PRP + HA (SUCRA value, 63.7; mean rank, 2.1), HA (SUCRA value, 27.5; mean rank, 3.2), and CSC (SUCRA value, 23.3; mean rank, 3.3). The results of the network meta-analysis are summarized in Table 2; Fig. 4. No discrepancy between the direct and indirect effects of any intervention was observed as per the nodal analysis of the intervention measures (P > 0.05). Figure 4 compares the results based on the VAS scores at the 3, 6, and 12-month follow-ups.

Fig. 4
figure 4

Overall network comparisons using VAS scores

Full size image

Safety

Among the 2576 patients with reported adverse effects, the PRP groups demonstrated the most favorable outcomes with a SUCRA value (81.8) and a mean rank (1.9), followed by PRP + HA (SUCRA value, 81.2; mean rank, 1.9), CSC (SUCRA value, 64.2; mean rank, 2.8), HA (SUCRA value, 27.5; mean rank, 3.2), PRP (SUCRA value, 22.6; mean rank, 4.9), and CSC + HA (SUCRA value, 7.0; mean rank 5.7). The results of the network meta-analysis are summarized in Table 2; Fig. 5. No discrepancy between the direct and indirect effects of any intervention was observed as per the nodal analysis of the intervention measures (P > 0.05). Figure 5 showed a comparison of results based on adverse effects.

Fig. 5
figure 5

Overall network comparisons using adverse effects

Full size image

Discussion

After 3, 6, and 12 months of follow-up, the Bayesian network meta-analysis revealed that PRP and PRP + HA IAIs were superior to CSC, HA, and placebo in alleviating pain and improving joint function. However, no discernible changes between CSC, HA, and placebo were found. Regarding safety, the incidence of adverse events associated with the other interventions was not significantly higher than that of the placebo.

According to this study, PRP proved to be superior to PRP + HA, CSC, HA, and PLA in enhancing joint function. Additionally, PRP + HA was found to be better than PRP, CSC, HA, and PLA in reducing pain. The incidence of adverse events did not significantly increase with other interventions, as compared to placebo. According to a prior network meta-analysis, the PRP group was more effective than CSC, HA, and placebo [28]. In the research by Zhao and his colleagues, the PRP + HA scheme was shown to be more effective than PRP alone in alleviating knee pain and raising the WOMAC overall score [29]. Compared with lower-molecular-weight hyaluronic acid, the highest-molecular-weight hyaluronic acid may be more efficacious in treating knee OA [30]. However, viscosupplementation is associated with an increased risk for serious adverse events [31]. Another study showed that intraarticular CS is more effective on pain relief than intraarticular HA in short term (up to 1 month), while HA is more effective in long term (up to 6 months) [32]. Autologous blood can be subjected to centrifugation to extract PRP, which can increase the platelet concentration by nearly ten times [33]. Upon activation, it exhibits the ability to discharge macrophages and growth factors, consequently promoting the elimination of necrotic tissue, reducing the inflammatory reaction, and facilitating the repair and regeneration of articular cartilage [34, 35]. HA, an essential element of synovial fluid and articular cartilage [10], is a polysaccharide with a high molecular weight. Injecting HA into the knee joint cavity can physically lubricate the joint surface, reduce erosion, biologically nourish the articular cartilage, and stimulate the production of endogenous HA, thereby delaying the onset of additional joint disease [36, 37]. Besides, HA has also been proven effective in obese individuals [38]. According to Marmotti et al. [39], the incorporation of HA into PRP has been found to significantly enhance the growth of chondrocytes and enhance cartilage regeneration capabilities. PRP and HA have been shown in studies to synergistically increase the functioning of signaling molecules such as inflammatory molecules, catabolic enzymes [40], cytokines, and growth factors, thus contributing to the successful treatment of KOA [41].

Some studies are consistent with the results of this study [42, 43], which found that CSC and HA showed similar results compared to placebo. However, there are also other studies that have reached different conclusions [30, 44, 45], finding that CSC and HA are more effective than placebo. The study demonstrated that the analgesic efficacy of the two therapies varied with time. Particularly, the VAS score of the intra-articular CSC group was considerably lower than that of the intra-articular HA group after 1 month, suggesting that CSC had a higher short-term analgesic impact than HA. However, in the long run, HA exhibited a greater analgesic effect than CSC [32]. No significant difference in pain relief was found between HA and placebo(saline)by Colen et al. [46]. According to a meta-analysis, intra-articular corticosteroid injection is an effective treatment for pain relief with no increase in treatment-related adverse reactions when compared to placebo [47]. Najm et al. discovered that CSC decreased pain and increased function early after administration (≤ 6 weeks) compared to placebo. However, there were no clinical improvements when compared to HA [10]. Based on our analysis, the only treatments that clinically showed improvement in both cases were PRP and PRP + HA. The effectiveness of CCS and HA is uncertain. Although treating KOA with PRP and HA combination may be more expensive and difficult, it may still be a preferable option to the expenses and risks of surgery. Nevertheless, there is still a shortage of cost-effectiveness studies that examine the combination of PRP and HA for KOA treatment, as well as studies that investigate PRP or HA alone, indicating a need for further research.

There are several limitations to this study: First of all, the main limiting factor is the lack of available data between the included studies. Secondly, some authors conducted a single injection, whereas others performed repeated injections. Thirdly, the duration of treatment and follow-up was diverse. Fourthly, we only included studies written in English, which may result in the loss of some research data. Lastly, the use of different formulations in different studies of HA may lead to bias.

Conclusions

The study’s SUCRA value backs the application of PRP and PRP + HA for appropriate patients with KOA. PRP is likely the most effective pain-relieving treatment with the lowest incidence of adverse effects, followed by PRP + HA. The differences in treatment effects were minor and might not have any significant impact on clinical outcomes.