Fifty-three RCTs investigating the effects of functional ingredients in OA met the inclusion criteria for this systematic review. The functional ingredients tested in these RCTs were lipids (ASUs, n-3 PUFAs, lipid extracts from New Zealand green-lipped mussel, and cetyl myristoleate), vitamins and minerals (vitamins C, E, B3, and B12, boron, a cocktail of vitamins and selenium, and a cocktail of minerals), plant extracts (bromelain, R. canina, H. procumbens, U. tomentosa and U. guianensis, Salix sp., ginger, turmerics, tipi tea, soy proteins, and B. serrata), a cocktail of plant extracts (SKI306X, Gitadyl, Duhua Jushing Wan, and Articulin-F), and a few other types of ingredients (methylsulfonlymethane, hyperimmune milk, and collagen hydrolysate). Eighteen of these functional ingredients had their efficacy supported by at least one RCT (Table ).
To summarise the strength of scientific evidence behind a functional ingredient, we used a mathematically based best-evidence synthesis. This best-evidence synthesis allowed us to categorise the functional ingredients as having a limited, moderate, or good record of efficacy. According to this best-evidence synthesis (Table ), good evidence exists for ASUs. Moderate evidence exists for methylsulfonylmethane and SKI306X, a cocktail of plant extracts. Limited evidence exists for the Chinese cocktail of plant extracts Duhuo Jisheng Wan, for cetyl myristoleate, for lipids from green-lipped mussels, and for plant extracts from H. procumbens. Limited evidence also exists for vitamins B3 and C and bromelain, but the small effects obtained, the high doses used, or the experimental design employed questions the clinical relevance and/or safety of these functional ingredients. The other interventions lacked scientific evidence either because of their rather poor design or because of contradicting available evidence. Among these interventions that lacked evidence of efficacy, vit E is unique: it is the only nutritional intervention whose lack of symptom-modifying and structure-modifying effects in knee OA is reported in high-quality RCTs. Despite the fact that our best-evidence synthesis considers each functional ingredient as a single entity, the evidence of efficacy and the safety record of plant extracts should be considered to be product-specific given that the composition of an extract from a same plant can vary widely between manufacturers.
All 18 functional ingredients evaluated in Table were tested under a nutraceutical/dietary supplement form in the RCTs, except for hyperimmune milk incorporated in a functional drink. Depending on the regulatory laws of each country, these functional ingredients are sold as drugs, nutraceuticals (dietary supplements), or functional foods in association with health claims of variable strength. Although most ingredients are sold mostly as nutraceuticals today, some such as SKI306X and ASUs require a prescription and are sold as drugs, at least in some countries (Korea for SKI306X and several European countries for ASU). Similarly, the vitamins and some of the lipids reviewed here are sold mostly as nutraceuticals but can also be incorporated in functional foods (up to a country-specific defined maximal dose) because they have GRAS (generally recognised as safe) status. Regarding collagen hydrolysate specifically, its GRAS status and its advertised therapeutic dose (10 g) make it more practical to be used in a functional food than in a nutraceutical. Ideally, the efficacy of a functional food should be directly evaluated in an RCT (by providing to the enrolled patients the final commercial product) because the incorporation of a functional ingredient into a complex food matrix could potentially modify its efficacy, either by increasing or on the contrary by decreasing its bioavailability.