Introduction

Knee osteoarthritis (OA) is the commonest articular disease and usually presents in aging populations with chronic joint pain, brief morning stiffness, impaired knee function, difficulty in activities of daily living, and loss of mobility. Examination may be characterized by audible or palpable coarse crepitus, bony enlargements, joint line tenderness, and deformities of the knee joints [1]. It can be defined radiographically and/or symptomatically, and such definitions can affect epidemiological estimates of knee OA [2]. Its estimated global prevalence in persons over 40 years of age is at 22·9% in 2020 (correspondingly 654·1 million individuals) [3], imposing considerable socioeconomic costs as the direct and indirect costs for OA management being 1 to 2.5% of the gross national product (GNP) in most of developed countries [4].

The current narrative review was aimed at summarizing the efficacy and adverse effects of the currently available IA agents in knee OA such as corticosteroids and hyaluronic acid, platelet-rich plasma, stem cells, and prolotherapy. Moreover, the accuracy rates of available methods of drug injection into knee joint are briefly discussed: blind or landmark-guided method vs ultrasound guidance.

The literature approach was based on the PubMed database over 2 years from 1 January 2021 to 31 December 2023 to reflect current research findings with a search strategy for randomized controlled trials (RCTs) conducted in human beings and written in the English language, focusing on but not limited to the terms “knee osteoarthritis” or “knee arthrosis” in combination with “intra-articular injections” or “injection-based therapy” or “corticosteroid” or “hyaluronic acid” or “platelet-rich plasma” or “stem cells” or “prolotherapy” ( see Online Supplementary Material for the full search strategy). Additional articles were identified by using the bibliographies of each paper. We identified 1515 papers from the search results and tried to focus on articles deemed to provide a purposeful increase in our knowledge base.

Recommended management of knee osteoarthritis

No drugs are yet available to modify the structural manifestations of the disease course of knee OA [5]. Current OA management focuses on symptomatic improvement only [6] and is largely palliative in approach despite the OA disease course typically being slowly progressive over years/decades [7]. Treatment options include (1) non-pharmacological management such as weight reduction, life-style changes, dieting, and exercises [8], (2) pharmacological options such as paracetamol, nonsteroidal anti-inflammatory drugs (NSAIDs), opioid analgesics, and intra-articular therapies such as steroids and hyaluronic acids, and (3) surgical interventions which are typically reserved only for end-stage OA, as a last resort [9••]. Recently, paracetamol and opioids have been only conditionally or not recommended by several scientific guidelines [6, 9••, 10]. In addition, the existing treatments have shown only modest efficacy at best [2], and long-term use of commonly used analgesics are not recommended due to adverse effects in the gastrointestinal, cardiac, or renal systems as patients with knee OA usually have multiple comorbid diseases [9••].

Intra-articular therapies

The IA administration of drugs possesses the advantage of high efficiency while limiting the systemic exposure and off-target effects as drugs are locally administered into the joints [11]. Most available guidelines for knee OA management would not typically advocate the use of IA medications until the second or third line [6, 12] as it is an invasive procedure and not without uncommon but serious complications such as septic arthritis (10–40 persons per 100,000 injections) [13]. Table 1 summarizes recommendations from NICE [9••], AAOS [14••], ACR [10], OARSI [6], ESCEO [12], and PANLAR [15] guidelines related to available IA therapies for knee OA as part of conservative management. IA therapeutics that possess anti-inflammatory properties might be appropriate, for example, during a flare of knee OA and the presence of a large effusion. The patient should be provided with essential information such as the nature of the IA procedure (blind or imaging guided), the potential benefits and risks of IA therapies, as well as post-injection care such as 24- to 48-h post-injection immobilization [16••, 17•].

Table 1 Recommendations from NICE, AAOS, ESCEO, OARSI, ACR, and PANLAR guidelines related to intra-articular therapies for knee OA
Full size table

On administration of IA therapeutics into the knee joints, either landmark-guided or imaging-guided approach could be used [16••] (Table 2). In the literature, there is strong evidence that ultrasound guidance IA injection provides significantly higher accuracy in administrating the injectates into the knee joints than landmark-guided (blind) injections (96% vs 73%, n=89) [18]; (96% vs 84%, n=99) [19]. Among 3 different approaches, the superolateral portal (100%) and mid-lateral portal (95%) showed significantly higher accuracy than injections in the medial portal (75%) (n=126) [20]. The enhanced injection accuracy achieved with ultrasound needle guidance [21] directly improves patient-reported clinical outcomes, cost-effectiveness [22], and patient satisfaction [23].

Table 2 Published studies investigating intra-articular injectates for the treatment of knee osteoarthritis since 1 Jan 2021
Full size table

Available IA injectates

The commonly used IA injectables in daily practice include corticosteroids, hyaluronic acid (HA), platelet-rich plasma (PRP), stem cell, and prolotherapy.

Corticosteroid

In clinical practice, IA corticosteroid injections are commonly used procedure in knee OA, especially when pain is refractory to physical treatments and oral medications, or to support therapeutic exercise [9••]. The choice of the drug type generally may be triamcinolone, methylprednisolone, or dexamethasone, depending on the clinical experience and preference of the physician. We identified five studies investigating IA corticosteroids in knee OA. In 4 studies, each study used IA normal saline (as a placebo) [24], ketorolac (NSAID) [25], PRP [26], or HA (vs steroid plus HA co-injection) [27] as the comparator group and the remaining study compared the routes of administration (IA vs intramuscular corticosteroid) [28].

Hunter et al. demonstrated significant improvement in WOMAC pain (mean difference = −0.35, p = 0.004) and function (mean difference = −0.26, p = 0.045) in knee OA with radiographic KL grade 2 or 3 when IA injection of 12 mg of liposome formulation (to prolong the local joint residence time) of dexamethasone sodium phosphate (TLC599) was compared to IA normal saline injection at 24-week follow-up (n=76). The percentage of clinically durable responders (defined as > 30% pain reduction as measured by the WOMAC-Pain scale) at 24 weeks was greater in TLC599 group (52% vs 22%; p = 0.0143); however, there was no significant improvement in quality of life on EQ-5D questionnaires. No major or unexpected safety issues were detected [24]. A larger and well-designed pivotal study (ClinicalTrials.gov identifier: NCT04123561) is currently ongoing to confirm this efficacy profile. In a 3-month study comparing the IA administration of 80 mg of triamcinolone with 30 mg of ketorolac under ultrasound guidance in patients with hip (n=52) or knee OA (n=58), no significant between-group differences in HOOS or KOOS and VAS scores with minimal adverse effects were reported [25]. In a 52-week study comparing the single IA injection among three groups (40 mg of triamcinolone hexacetonide (n=33) vs PRP (n=34) vs the saline solution (n=33)), no significant differences were found in the inter-group comparison over time except for superiority of the triamcinolone hexacetonide group over the other two groups at 4 weeks). The PRP group showed the lowest radiographic progression from KL grade 2 to 3 over 52 weeks among the three groups [triamcinolone from 52 to 73%; saline 52 to 91%; PRP from 59 to 62%] [26]. However, the study has several limitations such as small sample size, multiple comparison with no statistical adjustment and questionable sensitivities of radiographic scores in assessing the structural progression of knee OA.

In a 6-month study in knee OA, weekly IA co-injections of corticosteroids plus hyaluronic acid (HA) for 3 weeks (n=28) provided statistically significant improvement in WOMAC-pain (P = .005) and physical function (P = .005), chair-rising time (P = .032), and KOOS-pain (P = .001) from 1 week to 6 months, compared with HA injections alone (n=29) [27]; however, the outcome longer than 6 months is unknown. In a multicenter, open-label, randomized clinical noninferiority trial comparing the IA vs IM administration routes for steroids in symptomatic knee OA, noninferiority could not be declared with between-group mean difference in the KOOS pain score was −3.4 (95% CI, −10.1 to 3.3; effect size = −0.17) favoring the IA route at the primary time point (4 weeks) [28]. Lack of a placebo-controlled group and being an open-level study are study limitations.

Summary

The IA corticosteroid therapy provided short-term symptomatic benefits in knee OA but long-term benefits seem to be less likely even in combination therapies with other IA therapeutics. Extended release IA steroids which can prolong joint residence time are currently under intense research.

Hyaluronic acid

Viscosupplementation with hyaluronic acid (HA) in knee OA has been extensively researched since 1971 [29] and 2438 papers has been published between 2002 and 2021 [30]; however, its role in knee OA management still generates debate in terms of clinical benefits [9••, 30] as the published data were limited by the high heterogeneity of effect sizes perhaps due to different HA formulations, different periods of follow-up, and differences in injection schedules and IA techniques [17•]. In the recent literature, four papers were identified, comparing HA with placebo (n=2) [31, 32] and PRP (n=2) [33, 34].

A 26-week study in Chinese patients with knee OA (n=440) did not establish superiority of single 6 ml Hylan G-F 20 injection over IA placebo using the WOMAC pain score while walking. Percentages of positive responders over 26 weeks, defined as a ≥ 2-point improvement from baseline, was not significantly different (67% in the treatment arm vs. 68% in the placebo arm) [31]. As a note, the placebo effects were marked in the Chinese clinical compared with the European trial (44% vs 29%), which used a similar trial design and methodology (n=253) [35]. When diclofenac etalhyaluronate (DF-HA), which is diclofenac covalently linked to HA, was administered once every 4 weeks for 20 weeks (a total of 6 injections) in Japanese patients with knee OA, a statistically significant improvement in the WOMAC pain subscale, measured on a 100-mm VAS, in the active treatment group was detected at 12 weeks compared to the placebo group, with a difference of −6.1 mm (95% confidence interval −9.4, −2.8; P < 0.001); however, there was no significant difference at week 24 (n=440). Anaphylactic reactions were observed in 2 subjects receiving DF-HA [32]. The limitations are the need to frequent IA injection, which may lead to an increased risk of joint infections [17•] and the lack of evaluation of chondrotoxicity caused by NSAIDs which have been suggested to be deleterious to joint cartilage [36].

A single IA injection of HA (3 ml, 20 mg/ml) followed by 3 ml PRP (the combined-injection group) showed statistically significant pain reduction on a VAS score (mean difference: 7.9; p = 0.020), compared with a single injection of 3 ml PRP (the one-injection group) at 6-month follow-up (n=78). No serious adverse events occurred following injections. As a note, at 3-month follow-up, the subgroup of patients with baseline VAS pain > 56.4 mm in the one-injection group revealed a significantly greater reduction in VAS pain score than the combined injection group [33]. In another 24-month study, PRP combined with HA (48 knees) is significantly effective compared with PRP alone (40 knees) or HA alone (34 knees) at improving pain and function and acts through inhibiting synovial inflammation detected on ultrasound and reducing inflammatory cytokine content[34]. In a recent systematic review and meta-analysis including 7 studies (n=941), PRP combined with HA provided better clinical improvement (standardized mean difference≥0.30), such as pain and function beyond 6-month follow-up than PRP alone with no significant difference in the incidence of side effects [37], presumably due to their synergistic therapeutic effects on the proliferation of chondrocytes and cartilage repair demonstrated in a rabbit model [38].

Summary

While symptomatic benefits of IA HA therapies may be statistically significant, most clinical outcomes do not reach clinical significance. They are frequently being used in combination with other IA agents, expecting the chances of longer-term benefits compared with single-therapy regime. More research in the cost-effectiveness of such combination therapies is required.

Platelet-rich plasma

Platelet-rich plasma (PRP) definitions vary but one group suggested it be defined by its absolute platelet concentration > 1 ×106/μL or a fivefold increase in platelet concentration from baseline [39]. PRP may have the capacity to reverse pro-inflammatory processes and promote tissue repair via the release of many biologically active factors, such as growth factors and cytokines for restoration of the articular homeostasis [40]. Nine recent papers have been identified for IA PRP in knee OA.

In some studies, PRP has been found to be significantly effective in reducing the symptoms [41,42,43,44] and MRI-detected cartilage loss [45], compared with HA [41,42,43] or ozone [42] or normal saline placebo [44, 45] up to 36 [43] or 60 [45] months. However, in the other studies, it was reported that PRP is not superior to the placebo [46,47,48] or HA [49] in improving pain and function [46,47,48,49] and preventing the structural progression of the disease [46] up to 12 months [46, 48]. Regarding the frequency of single-dose PRP injection, PRP injections could be repeated at 6-month intervals as the effect of a single-dose PRP decreased significantly after 6 months[44]. In a recent meta-analysis of IA PRP versus comparative cohorts retrieved until 1 December 2021 which included 24 PRP clinical trials with HA(n=11), corticosteroid (n=6), normal saline (n=5), exercise therapy (n=3), and clinical relevance of outcome improvements (VAS and WOMAC scores) cannot be determined despite statistical significant findings and examination of structural changes showed no difference between the comparative groups. In addition, a paucity of high-quality studies, substantial heterogeneity of included studies, mostly small sample sizes, relatively short-term follow-up (mostly 6-month follow-ups) were noted [50••]. Recently, the American Academy of Orthopaedic Surgeons downgraded their strength of recommendation of PRP to “limited” due to inconsistent evidence [14••].

Summary

Current evidence is of low quality and is based on clinical trials with high risk of bias, great heterogeneity among clinical trials, and serious methodological flaws. Future high-quality studies using larger sample sizes, longer study durations and good methodologies are required.

Stem cell therapy

Due to multilineage differentiation potential, stem cell injections have been proposed as an innovative regenerative therapy for knee OA. However, a recent systematic review reported a paucity of RCT in this area, high risk of bias in the available studies, and a lack of long-term results [51]. We identify two papers related to stem cell therapies compared with PRP [52] or HA [53] in knee OA.

In a 24-month RCT comparing bone marrow aspiration concentrate (BMAC) and PRP in knee OA (n=84), WOMAC scores at 24 months improved by 14.5 points (41%) from baseline in the BMAC group and 12.4 points (38%) in the PRP group with no significant inter-group differences, disputing the beneficial effects of BMAC, given the added morbidity and expense of a bone marrow aspirate in the general OA population (an average cost of US $714 for PRP and US $3000 for BMAC). The study was limited by a high loss to follow-up (24% and 32% at 24 months for the BMAC and PRP groups) and no placebo group[52]. In another study conducted in 56 patients with bilateral knee OA, single IA BMAC injection was administered into one knee and single IA HA injection into the contralateral knee, each patient thereby acting as their own control and eliminating the need for sham bone marrow aspirations to maintain blindness. Compared to HA, BMAC showed a greater improvement in VAS pain at 12 months (2.2 vs 1.7, p = 0.04) and 24 months (2.2 vs 1.4, p <0.01) with no serious adverse events [53].

Summary

Conflicting results are noted between the comparative groups (BMAC vs PRP or HA) in two recent clinical trials, highlighting inconclusive evidence and the need for future high-quality studies.

Prolotherapy

In prolotherapy, small amounts of an irritant solution such as hyperosmolar dextrose (d-glucose) with concentrations ranging from 12.5 to 25% are injected into painful joints to restore joint stability by promoting the tensile strength of joint stabilizing structures, such as ligaments, tendons, and joint capsules [54]. We identified two recent studies.

In a small study conducted in knee OA (n=47), IA dextrose prolotherapy demonstrated a significant reduction in NRS pain scores over 12 weeks, compared with HA injection (−3 vs −2 for prolotherapy vs HA groups on NRS; p=0.04) [55]. In another 6-month study (n=104), HA plus dextrose co-injections under ultrasound guidance provided more significant improvements in stair climbing time and physical function at 6 months, compared with HA plus normal saline co-injection[56].

Summary

Despite statistically significant results in the recent studies, there is uncertain clinical relevance/significance due to small sample size, short-term follow-ups, a lack of control group, and absence of structural evaluations.

Opinion and future directions

Placebo effects

IA saline is a commonly used placebo in control groups in RCTs of IA therapies, and it can produce remarkable pain relief that may reach the values of minimal clinically important difference (MCID) [57]. In a recent meta-analysis, the placebo effects of IA saline at 6 month follow-up generated a significant improvement on 0–100 VAS pain score [−13.4 (−21.7/−5.1)], in WOMAC function sub-score [−10.1 (−12.2,-8.0)] and the pooled responder rate was 56% by using the OMERACT-OARSI criteria [58]. Therefore, in the clinical trials where pain and function are used as the primary endpoints, the placebo effects of IA saline should be accounted for in planning the trial design [59]. As the IA saline injection may be more than a “mere” placebo due to dilution effects in the joint environment [60], the underlying mechanisms of placebo effects and their predictors as well as the comparative effects of sham vs saline injections should be examined in robust study designs[5].

Trial quality and reasons why many injectates are not recommended in guidelines

In evidence-based medicine, the quality of clinical trials is one of the core factors for translating the research results into clinical practice. Increased risk of bias at the individual study level, inconsistent results among included clinical trials for a particular intervention, potential for small study effects and imprecision of the effect estimates often leads to the downgrading of the evidence in formulating the treatment recommendations[6]. In HA and PRP clinical trials, the use of different formulations, trade names, preparation methods, and frequency of injections leads to conflicting results among the studies, thereby reducing the strength of recommendations for these treatment options. Another core factor is cost-effectiveness or in terms of out-of-pocket costs. As an example, a single PRP injection has an average out-of-pocket cost of $714 and BMAC costs on average $3000. Given the degree and duration of efficacy between the two treatment options are not different, patients could elect to receive 4 PRP injections over the course of treatment for the same cost as a single BMC injection [52]. Similarly, IA HA cost per injection ranges from $300 to $600 USD, while a single steroid injection ranges from $10 to $50 USD [25].

Drug delivery systems for sustained release

As IA therapies directly targets the recognized pathogenetic tissues within the joint, lower doses than oral formulations are required due to higher bioavailability. Despite this advantage, there are several issues for IA administration which include pain and swelling during/after injection, uncommon but serious complication of septic arthritis and a short joint residence time due to rapid clearance of the IA therapeutics by the body. Therefore, a variety of drug delivery systems (DDS) have been developed with the aim of increasing the joint residence time of IA therapeutics. An ideal DDS should have such properties as controlled and/or sustained release of IA therapeutics for long-term effectiveness without a need of frequent injections [61] and adequate disease-modifying capacity, biocompatibility, and biodegradability [62]. New smart drug delivery strategies, which utilized hydrogels methods, nanoparticles and microparticles may enhance the opportunity for detecting the ideal long-lasting IA therapeutics [63].

Concomitant use of local anesthetics and chondrotoxicity

As IA injection may have pain and some discomfort to the patients, local anesthetics such as lidocaine are often added to the injectates in many studies described above. In recent systematic reviews, dose-dependent and time-dependent chondrotoxic effects of local anesthetics, presumably through mitochondrial dysfunction have been reported in vitro experimental and in vivo animal studies [64, 65]. However, it is uncertain whether these data from chondrocyte cultures or animal models might be transferrable to human tissues [64]. In the most recent in vivo study, single intra-articular knee injection of 10 ml of 2% lidocaine did not influence the chondrocyte viability and showed fast post-injection reduction of synovial lidocaine concentration (more than 40 times) [66]. Future studies which determine the chondrotoxicity of local anesthetics should be conducted.

Conclusion

Despite quick improvement in pain and function, IA corticosteroid has questionable efficacy on head-to-head comparisons with other comparators such as IA PRP, ketorolac, or normal saline in the long term. Combination therapy of IA corticosteroid plus HA may be more effective than a single IA corticosteroid therapy but needs further confirmation study in larger sample size. Although IA HA failed to show symptomatic improvement compared with normal saline, significant improvement in pain and function at least over 6 months was detected in providing combined therapies of HA with PRP or diclofenac, compared with single HA therapy in small studies, which similarly require further research. Conflicting results were reported over IA PRP using a variety of comparators such as HA, ozone and normal saline with some studies having follow-up durations of 36 or 60 months. Similarly, divergent results are reported for stem cells therapies. Therefore, standardization of PRP and stem cell products are in urgent need. Prolotherapy may be effective in single or combination regimes in small studies. None of the IA therapies included in the review demonstrated serious adverse effects nor septic arthritis. Joint injection accuracy and clinical outcomes are higher with imaging-guidance. Future studies should address the issues of product standardization, placebo effects, possible adverse effects of IA NSAID or local anesthetics on the cartilage, drug delivery systems, and cost-effectiveness of different IA therapeutics.