Knee osteoarthritis (OA) is one of the most general irreversible articulus diseases globally, and presents with features of incremental cartilage defect and articular space narrowing. Approximately 11.8% to 12.7% of the global population are affected by knee OA, according to World Health Organization [1]. The condition is similar in China, where the number of OA patients nearly increased 2.35-fold over the past three decades, and approximately 61.2 million individuals suffered from symptomatic OA in 2017, with a percentage of mild, moderate, and severe OA of 47%, 35.9%, and 17.1% respectively [2]. Despite the high prevalence of knee OA, effective and permanent interventions to halt or reverse the degenerative progression have not yet been developed [3]. Intra-articular chronic inflammation accompanied with joint pain and dysfunction is the main pathological features of knee OA, which necessitate long-term management. Widely applied pharmacotherapies aimed at anti-inflammation and pain reduction are limited to acetaminophen and non-steroidal anti-inflammatory drugs (NSAIDs) [4]. While conventional medications have only a marginal effect on pain, with no significant impact on joint function. Adverse events (AEs) that may occur in digestive and cardiovascular systems also restrict the feasibility of long-term administration of NSAIDS [5–7]. Hence, the exploration of alternative options with good safety and efficacy profiles for knee OA has been delved into traditional herbal medicine [8–10]. Notably, curcumin, extracted from the rhizome of Curcuma longa L. [Zingiberaceae] (CL), is a botanical extract with promising clinical values [11].

Analogues comprising curcumin, bisdemethoxycurcumin, demethoxycurcumin and cyclocurcumin are collectively referred to as curcuminoids (CURs) [12], which constitute the principal ingredients of CL—an herbal plant used in east Asia to alleviate pain and inflammation. CURs are natural polyphenols which have been shown to exert anti-inflammatory and anti-oxidant effects in vivo and vitro studies by downregulating inflammation-related nuclear factor kappa-B (NF-κB) signaling pathway, scavenging free radicals, and inhibiting the activity of enzymes, such as cyclooxygenase-2 (COX-2), 5-lipoxygenase (5-LOX), and nitric oxide synthase (NOS), which exacerbate the oxidative stress in OA condition [13–15]. Normal NSAIDS are of critical safety concerns due to simultaneously inhibition of COX-1 and COX-2 enzymes in arthritis, while CURs can reduce the synthesis of COX-2 tendentiously [16], which may result in better safety profiles. Furthermore, CURs exhibit chondroprotective properties by stimulating extracellular matrix synthesis, down-regulating the synthesis of matrix metalloproteinases (MMPs) [17]. And CURs were shown to postpone joint contracture progress via inhibiting the proliferation of myofibroblasts from the joint capsule [18]. Considering that the pathophysiology of knee OA is characterized by inflammation and degeneration with prominent symptoms of pain and dysfunction, alleviating local inflammation and oxidative stress, stimulating cartilage regeneration and delaying joint contracture may be conducive to the condition, and CURs have emerged as an attractive treatment option for knee OA.

Several animal studies have assessed the efficacy of CURs administered via nano-scale drug carriers for knee OA, demonstrating that CURs have potent anti-inflammatory and anti-arthritic activity, both with and without biological materials [19–23]. Although pre-clinical studies have revealed promising results, the clinical efficacy, safety, dosage, and treatment duration of CURs for knee OA remain equivocal. Thus far, the therapeutic effects of CURs for knee osteoarthritis (OA) uncovered by multiple reviews remained uncertain due to broadly involving trials with different agents-combined or CURs-free interventions [8, 24–26], and evidence to reveal the clinical significance of CURs alone for knee OA is insufficient. Consequently, we aimed to summarize the evidence to date on the clinical effectiveness of CURs alone in alleviating pain and dysfunction for knee OA by a systematic review and meta-analysis. We postulated that CURs have superior efficacy in pain relief and functional promotion compared to control measures.

The research was performed according to our pre-registered protocol (CRD42021266888, PROSPERO) with some amendments in the selection and assessment of outcomes.We adopted the concept of the minimum clinically important difference (MCID) [27] to assess the clinical significance of CURs for treating knee OA. The study was conducted by the guidance of the Cochran Handbook for Systematic Review of Interventions [28], and reported according to the Preferred Reporting Items for Systematic Review and Meta-Analysis checking list (Supplementary Table 1) [29].

The principal finding of our study was that CURs were associated with better effectiveness than placebo and not inferior to NSAIDs in terms of pain reduction and functional promotion for knee OA. The pooled analyses found that CURs were more effective than placebo in the improvement of VAS for pain, WOMAC total score, WOMAC pain score, WOMAC function score and WOMAC stiffness score, while there was no significant difference found between CURs and NSAIDs. We used the MCID as a threshold in this meta-analysis to assess the clinical significance of the difference between CURs and the control groups, instead of rely solely on the statistical significance. The MCID can be calculated by anchor-based and distribution-based methods, we applied the anchor-based method to set the threshold at 20% based on previous research [27, 33–36]. The significance test of clinical benefits found that only VAS for pain and WOMAC total score achieved clinical significance by exceeding their MCID, while WOMAC pain score, WOMAC function score and WOMAC stiffness score did not. We also found that CURs did not induce an increase of AEs compared with placebo and NSAIDs. The total incidences of AEs in CURs and control groups were 25.06% and 35.57%. Diarrhea and/or constipation and stomach pain (5.8% and 8.17%) were the most frequent mild AEs in CURs and control groups respectively (Table 4).

Pain and dysfunction were the leading causes for medical care use and clinical decision making for knee OA [73]. Novel disease-modifying treatments targeting the pathological process of OA are in development to solve the treatment dilemma of symptom-relieving drugs (pain-killers or NSAIDs) [74]. Among which, CURs have attracted much attention of medical researchers and clinicians [8–10]. CURs have been shown to possess therapeutic effects on knee OA as a result of their anti-inflammatory and anti-oxidant properties [15]. The regulation of inflammation- and catabolism-related pathways is the main mechanism underlying the anti-inflammatory and chondroprotective properties of CURs [75]. CURs exhibit anti-apoptotic and antioxidant effect on chondrocytes and induce mesenchymal stem cells chondrogenic proliferation. Thus far, many pre-clinical and clinical studies [76] have identified CURs as being effective for treating knee OA. Despite the highly pleiotropy in knee OA, the application of CURs is controversial due to poor oral bioavailability. Numerous studies have focused on methods to optimize the pharmacokinetics of CURs [77]. Among the included studies, ten used bioavailable CURs, such as nanocurcumin [61, 70, 71], liposome CURs complexes [58, 67] and so on, while the other five used pure extracts from CL. Theoretically, bio-optimized CURs should be superior to pure extracts given their higher absorptivity and lower metabolism [77]. According to our study, we found that the result of the bio-optimized extracts subgroup increased to exceed the MCID of WOMAC pain score, but the result of the pure extracts subgroup was neither statistically nor clinically significant. Besides, the effect sizes of both VAS for pain and WOMAC function score in the bio-optimized extracts subgroup exceeded their MCID, while those in the pure extracts group did not. These findings indicated that bio-optimized CURs may have better clinical applicability for knee OA than pure CURs. However, a recent meta-analysis of Wang et al. [24] found no significant difference between the enhanced and normal CL extracts in pain and physical function related outcomes. Two trials [44, 59] applying normal CURs as adjuvants to NSAIDs were included in their placebo-controlled group for quantitative synthesis, which may cause the divergence in the priority of the enhanced CURs given that the added effects of NSAIDS were neglected. Each study in this meta-analysis applied different metrics and tactics to remodel the bioavailability of CURs, direct comparisons between different CURs products are essential to verify our findings and seek a cost-effective agent. Predictably, bio-optimization techniques with more than one approach to conquer the hindrances (e. g., poor water solubility, rapid metabolism, and instability) to oral bioavailability would achieve significant improvement in the effectiveness of CURs.

The appropriate dosage of CURs for treating knee OA remains uncertain. Previous meta-analyses by Daily et al. [10] and Onakpoya et al. [9] demonstrated the typical dose of 1,000 mg/day as effective in the improvement of inflammation-related symptoms. According to our study, there was no statistically significant difference in the main outcomes between CURs and placebo in all subgroups of different doses (daily dose < or ≥ 1,000 mg and total dose < or ≥ 50 g). Theoretically, the optimal dose of a drug is closely associated with its safety and bioavailability. Various studies focused on diverse diseases have proved that CURs are effective without major safety concerns even at high doses such as 6 g/day [78], which indicated that the main limitation of optimal dosage is the bioavailability of CURs. Thus theoretically, the requirement for CURs of lower dosage and better compliance without affecting curative effects for knee OA may be met by the optimization of bioavailability. Actually, we found that the effect sizes of VAS for pain, WOMAC pain score and WOMAC function score in low-dose (daily dose < 1,000 mg and total dose < 50 g) subgroup exceeded their MCID, while those in high-dose (daily dose ≥ 1,000 mg and total dose ≥ 50 g) subgroup did not achieve clinical significance. The observed difference in clinical values between low- and high-dose subgroup may be caused by the limited number of studies in each subgroup or the administration of bio-optimized CURs for all studies in low-dose group.

A recently published review by Zeng et al. [26] suggested that CURs could not exhibit significant therapeutic effects until the duration of administration lasted for more than 12 weeks. Given the simultaneous inclusion of active-controlled [44, 56, 57, 59, 79] and placebo-controlled trials in their subgroup analyses, the effect sizes at different time points may be weakened by effects of NSAIDs-controlled arms. Thus, we removed the data of active-controlled trials in subgroup analyses, and found that CURs showed favorable improvement in VAS for pain, WOMAC pain score, and WOMAC function score compared to placebo at each time points (follow-up duration < or ≥ 12 weeks). Besides, the MCID was exceeded by the effect sizes of all main outcomes in the time < 12 weeks subgroup, but the effect sizes of the WOMAC pain score and WOMAC function score in the time ≥ 12 weeks subgroup did not achieve clinical significance. The reason for the difference in clinical values between short- and long-term subgroup may be that all three studies [62, 67, 71] in the short-term subgroup applied bio-optimized CURs, and two [65, 69] of the three [65, 66, 69] studies in long-term subgroup used normal CURs. Overall, in terms of alleviating pain and other symptoms, bio-optimized CURs may be sufficiently potent to lower dosage and shorten medication cycle. Besides, larger effect sizes with clinical significance of both pain reduction and functional promotion were observed in trials performed in Asia compared with those in other countries, which was in accordance with the result of a recently published review by Wang et al. [24].

The quality of our findings was evaluated using the GRADE system [37]. All pain and function related outcomes were downgraded to have a moderate to very low quality of evidence duo to inconsistency, risk of bias, and publication bias, while AEs were defined as high-quality evidence (Supplementary Table 3). The US Food and Drug Administration (FDA) defined CURs as nutraceuticals under “Generally Recognized as Safe” (GRAS) [80], and good safety and tolerability properties have been revealed by multitudinous studies at cellular level, in animals and even in human subjects [78], but it is still worth noting that nutraceuticals like CURs generally lack a systemic safety assessment before being used for medicinal purposes [81],therefore the potential dose- and time-dependent side effects of CURs on human body should be weighted carefully when facing the current benefits and potential values for broader clinical use of CURs.

CURs alone can be expected to achieve considerable analgesic and functional promotion effects for patients with symptomatic knee OA in short-term, without inducing an increase of adverse events. However, considering the low quality and substantial heterogeneity of present studies, a cautious and conservative recommendation for broader clinical use of CURs should still be made. Further high-quality studies are necessary to investigate the impact of different dosages, optimization techniques and administration approaches on long-term safety and efficacy of CURs, so as to strengthen clinical decision making for patients with symptomatic knee OA.