No drug has been approved that slows osteoarthritis, and the near-misses show why

Nothing licensed anywhere slows the progression of osteoarthritis. Every approved treatment addresses symptoms; no disease-modifying osteoarthritis drug has reached approval, and the reasons given in the field are consistent — single-pathway targeting, insufficient exposure inside the joint, and a disease that is not one disease [s4]. The candidates that got furthest are worth reading closely, because each failed in a different and instructive way.

What "disease-modifying" would have to mean

Osteoarthritis is described in this literature as a whole-joint disorder in which cartilage, synovium, subchondral bone, meniscus and the infrapatellar fat pad deteriorate under chronic inflammatory and mechanical stress [s4]. A disease-modifying drug would have to change that trajectory, not merely reduce the pain it produces — which means demonstrating a structural outcome, usually joint space width on X-ray or cartilage thickness on MRI, alongside symptom benefit.

That requirement is what makes the trials so hard. Structural change is slow, the imaging endpoints are noisy, and the relationship between what a joint looks like and how much it hurts is weak enough that a drug can succeed on one and fail on the other.

Lorecivivint: the symptom half without the structure half

Lorecivivint is an intra-articular inhibitor of CLK and DYRK kinases that modulates the Wnt pathway, developed explicitly as a disease-modifying candidate for knee osteoarthritis. A systematic review and meta-analysis pooled six randomised controlled trials including 2,179 patients, comparing lorecivivint with placebo, searched to January 2026 [s1].

The drug produced modest improvements in WOMAC pain scores at weeks 12 and 24 [s1]. It did not significantly improve functional outcomes, Pain Numeric Rating Scale scores, patient global assessment, or structural changes at any assessed time point [s1]. Subgroup analyses found no consistent dose–response relationship, and no meaningful improvement at the 0.07 mg dose [s1]. Safety outcomes were comparable to placebo [s1].

A modest effect on one pain instrument but not another, no dose–response, and no structural signal is the profile of a drug that has not demonstrated what it was designed to demonstrate. The reviewers' own conclusion asks for long-term studies to validate the clinical importance of the pain reduction that was seen [s1].

Sprifermin: the structure half without the benefit half

Sprifermin took the opposite path. It is a recombinant fibroblast growth factor 18 given intra-articularly, and in the phase 2b FORWARD trial — a five-year randomised controlled trial in 549 patients with knee osteoarthritis — it did what almost no osteoarthritis drug has done: it increased cartilage thickness [s2].

A post-hoc reanalysis addressed a methodological problem in the original readings, which had been performed without blinding to the order in which the MRIs were acquired. The reanalysis read the year-2 and year-5 images blinded to both temporal order and treatment group, in 337 participants: 57 on placebo injections every six months, 69 on 30 μg every twelve months, 67 on 30 μg every six months, 73 on 100 μg every twelve months and 71 on 100 μg every six months, all dosed between baseline and month 18 [s2].

Between year 2 and year 5 — after treatment had stopped — total femorotibial joint cartilage thickness changed by −26 μm across the cohort (SD 64; 95% CI −32 to −19), with no statistically significant difference between the sprifermin arms and placebo (P = 0.80) [s2]. Every group lost cartilage [s2]. The cartilage gained by year 2 with 100 μg sprifermin was maintained to year 5, and annualised cartilage change in placebo participants was −8.2 μm (SD 21; 95% CI −14 to −2.5) during years 2 to 5 against −5.4 μm (SD 27; −13 to 1.8) from baseline to year 2, with no significant difference between those periods [s2].

So the drug grew cartilage, the extra cartilage persisted, and after the drug stopped everyone decayed at the same rate. That is a real biological effect. Whether it is a clinical one is a separate question this analysis was not designed to answer, and the honest reading is that cartilage thickness has not been shown to be a surrogate for anything a patient experiences.

Borrowing from rheumatoid arthritis

A third approach has been to repurpose the drugs that work in inflammatory arthritis. A network meta-analysis of hand osteoarthritis assembled 12 randomised controlled trials covering 1,317 participants and seven active therapies: methotrexate and hydroxychloroquine among conventional synthetic disease-modifying antirheumatic drugs, and adalimumab, etanercept, tocilizumab, lutikizumab and otilimab among the biologics [s3]. No eligible trial of a targeted synthetic agent existed [s3].

Etanercept and tocilizumab were associated with lower long-term visual analogue scale pain than placebo [s3]. No intervention showed consistent improvement in hand function [s3]. Tocilizumab reduced painful and swollen joint counts in the short term, and adalimumab reduced swollen joint counts in some analyses [s3]. The reviewers characterise these findings as outcome-specific, time-dependent and largely based on sparse, placebo-centred networks — meaning most comparisons rest on indirect evidence rather than head-to-head trials [s3]. No clear differences appeared in overall treatment-emergent or serious adverse events, though the safety evidence was limited [s3].

Why it keeps failing

The field's own diagnosis has three parts [s4]. Drugs have targeted single pathways in a disease that involves several tissues at once. Getting and keeping a drug inside a joint at useful concentrations is genuinely difficult — which is why so much current work is about delivery systems rather than new molecules. And "osteoarthritis" covers heterogeneous phenotypes that may not respond to the same intervention, so a trial enrolling everyone with a radiographic diagnosis may dilute a real effect in a subgroup into nothing overall.

What to watch

Whether any candidate produces a structural and a symptomatic result in the same trial, in the same patients, over a long enough period to matter. Lorecivivint managed a symptom signal without a structural one [s1]; sprifermin managed a structural one whose consequences are unknown [s2]. Until one drug does both, "slows osteoarthritis" remains a claim no product can make.

This article is informational and is not medical advice. Doses named here are trial parameters.

Sources

• Efficacy and safety of lorecivivint for the treatment of knee osteoarthritis: A systematic review and meta-analysis of randomized controlled trials — Journal of Orthopaedic Surgery, 2026-07-05
• Unbiased analysis of knee cartilage thickness change over three years after sprifermin vs. placebo treatment — a post-hoc analysis from the phase 2B FORWARD study — Osteoarthritis and Cartilage Open, 2024-08-23
• Disease-modifying antirheumatic drugs for hand osteoarthritis: A systematic review and network meta-analysis — Hand Surgery and Rehabilitation, 2026-08-22
• Delivery of potential natural disease-modifying osteoarthritis drugs: Beyond polyphenols and flavonoids — Biomedicine & Pharmacotherapy, 2026-03-27