Avocado-soybean unsaponifiables (ASU)

Avocado-soybean unsaponifiables are a 300 mg standardized extract, sold as a drug in some countries and a supplement in others, and derived from two foods that nobody suggests you eat for the purpose. What makes it worth an entry is not the symptom data — plenty of things have symptom data — but that somebody funded a three-year radiographic trial, which almost nothing in this section can say.

The structural trials

ERADIAS randomized 399 people with symptomatic hip osteoarthritis and a joint space width between 1 and 4 mm to 300 mg a day of ASU or placebo, and followed them with annual radiographs for three years. Three hundred and forty-five remained in the analysis set.

The primary endpoint — mean change in joint space width at three years — came out at −0.638 mm on ASU and −0.672 mm on placebo. p = 0.72. Nothing.

A pre-specified secondary analysis counted "progressors", defined as anyone losing at least 0.5 mm: 40% on ASU against 50% on placebo, p = 0.040. No clinical outcome differed between the groups over three years. Safety was excellent. The authors' own conclusion described a potential structure-modifying effect "to be confirmed", whose clinical relevance "requires further assessment".

A trial twelve years earlier had run the same shape. In 163 patients with hip osteoarthritis over two years, overall joint space change did not differ (p = 0.9). A post-hoc split at the median baseline joint space found that the more severely narrowed half lost 0.43 mm on ASU against 0.86 mm on placebo (p < 0.01), with nothing in the less affected half — a finding the authors were careful to label as post-hoc in a pilot that had failed to demonstrate a structural effect.

ERADIAS was, in effect, the confirmatory trial for that subgroup. It did not confirm it on its own primary endpoint.

The symptom data

Two meta-analyses, and both find something.

The 2008 pooling covered four trials and 664 patients with hip or knee osteoarthritis on 300 mg a day for an average of six months. Pain favoured ASU with an effect size of 0.39 (95% CI 0.01–0.76, p = 0.04) and the Lequesne index by 0.45 (0.21–0.70, p = 0.0003), with a responder odds ratio of 2.19 and a number needed to treat of six. Two things about it: the heterogeneity on the pain estimate was I² = 83.5%, meaning the trials substantially disagreed with each other, and all four trials were supported by the manufacturer.

The 2019 pooling found a mean pain reduction of 9.64 mm on a visual analogue scale (95% CI −17.43 to −1.84, p = 0.02) with I² = 92%. Split by joint, ASU reduced both pain and Lequesne index in the knee and neither in the hip. Adverse events were indistinguishable from placebo (RR 1.02, 0.83–1.25).

The gap that has been open since 2014

Put the two halves together and the shape of the problem is obvious. The symptomatic benefit is concentrated in the knee. Both structural trials were run in the hip. And the three-year hip trial found no clinical benefit whatsoever — which sits awkwardly beside six-month meta-analyses reporting one.

Nobody has run a structural trial of ASU in the knee. That is the study this topic has needed for over a decade, and its absence is why an entry with more long-term randomized data than most of this section still sits at promising.

What it is proposed to do

In chondrocyte culture, ASU inhibits IL-1 and the metalloproteinases downstream of it, blocks the cartilage breakdown that follows, and stimulates aggrecan synthesis — anticatabolic and anabolic in the same preparation, which is an unusually complete mechanistic story for a plant extract. It is also largely in vitro, and the review assembling it was written in part by authors affiliated with a company marketing an ASU product.

Practical notes

Every trial used 300 mg a day of the same preparation, ASU-Expanscience (Piascledine), which makes the dose question unusually simple. Safety is the strongest part of the record: adverse events matched placebo across the pooled analyses and across three years in ERADIAS. Availability and regulatory status differ by country.

What would change this entry

A three-year structural trial in the knee. ERADIAS proved that a trial of that length and rigour can be funded and completed for this compound; running the same design in the joint where the symptom data actually point would resolve in one study what two hip trials and two meta-analyses have circled for twenty years.