Dupuytren's disease drug candidates: 12 existing medicines compared

Two scales that do not line up

Across: how far a drug has been tested in this disease, from no study (0) to a randomised trial (4).
Up: its standing in fibrosis elsewhere, from none (0) to approved for a fibrotic disease (3).

The drug with the best Dupuytren's evidence, adalimumab, has no fibrosis approval. The two drugs approved for fibrotic disease, pirfenidone and nintedanib, score 1 and 0 here. No candidate scores high on both.

Four of the five drugs at the top of the Dupuytren's ladder were tested only as add-ons to a procedure, and one of them (5-FU) failed.

Select a point to jump to its row in the table.

All twelve candidates Sortable

Repurposing candidates with two separate evidence scales. Select the arrow on a row to see the evidence detail, translation-risk note and sources.
Details					Route · safety

The three to watch

These are the three drugs I'd look at most closely. Pirfenidone and metformin are the leads; adalimumab is the one they have to beat. For each, the key question is whether it would change the disease for good or only hold it back while you keep taking it.

The closest to being ready for a trial. It's already approved for a scarring disease of the lungs, and an injectable version has been made specifically for Dupuytren's.

Mechanism

Inhibits TGF-β1-stimulated SMAD and non-SMAD signalling in Dupuytren's-derived fibroblasts PMID 27835939 PMID 30927912. It has no defined molecular target: "targets TGF-β" is a statement about the pathway, not about binding.

Evidence in Dupuytren's: in vitro (1 of 4)

Antifibrotic activity in patient-derived fibroblasts, replicated within the same group across two studies, plus an injectable formulation developed for this indication PMID 35129055. No animal model and no human study of any kind.

Fibrosis elsewhere: approved (3 of 3)

Approved for idiopathic pulmonary fibrosis: first US approval 15 October 2014 (2011 in the EU and Japan). Generic in the US, with 17 approved generics.

Safety and route

No boxed warning, but a demanding oral regimen (801 mg three times daily) with stomach upset, sun sensitivity, rash and liver monitoring. For a hand condition, IPF-level oral dosing is hard to justify, which is why the local injectable formulation decides whether this is a candidate at all.

Patent and cost

Off-patent in the US and cheap as an oral generic. UK and EU status is unverified, and a local preparation could be separately protected (open problem OP_05).

Durability: unknown, with grounds for concern

No human durability data. A 2026 study found that blocking TGF-β1 did not change new collagen synthesis in Dupuytren's tissue PMID 41588641, so pirfenidone might suppress the contractile cells while dosed and leave the accumulated matrix untouched. Plan for suppression, and design the trial to detect it.

Draft trial outline

Pre-registered lab validation in patient-derived fibroblasts and tissue, with 5-FU as the comparator that previously looked good and failed (OP_04).
Tolerability and pharmacokinetics of the intralesional formulation.
First efficacy trial: randomised, double-blind, placebo-controlled; patients with a palpable nodule and documented progression in the previous 6 months; primary endpoint nodule hardness at 12 months. A 5 AU difference on an SD of 8 AU needs about 136 participants (90% power, 20% attrition). Mandatory 12-month off-treatment follow-up to answer the durability question.

Two kinds of evidence agree. Lab work on patients' cells and large genetic studies of people both point the same way, which is true of no other candidate.

Mechanism

Prevents TGF-β-induced fibrotic marker expression and contraction in Dupuytren's cord fibroblasts and normal palmar fascia fibroblasts, via mitochondrial complex I / AMPK. It did not change baseline levels: it prevented their induction PMID 35427821.

Evidence in Dupuytren's: in vitro (1 of 4)

A single study in patient-derived cells, whose authors call for a trial as a surgical adjuvant to prevent recurrence. No animal studies and no trials in patients.

The independent argument

Mendelian randomisation in 379,708 UK Biobank participants: a 10 mmol/mol rise in genetically predicted HbA1c gives OR 1.17 (95% CI 1.01–1.35) for Dupuytren's, so long-term high blood sugar is a causal risk factor PMID 38205890. There are also two studies of people already taking metformin for diabetes, and they point in opposite directions. A 2004 UK study found Dupuytren's was more common in people prescribed metformin (adjusted odds ratio 3.56) PMID 15336742. A 2023 US database study found it less common in people taking metformin than in those taking insulin (RR 0.801) PMID 37443309. Both are tangled up with how severe each person's diabetes is, so neither tells us whether metformin helps. No trial has given it to people with Dupuytren's.

Safety and route

Oral and very well tolerated at standard doses, with three decades of population-scale use. Boxed warning for lactic acidosis, where kidney function is the main risk factor.

Patent and cost

One of the most generic drugs there is (47 US generic approvals) and negligible cost. UK/EU status still to be confirmed formally (OP_05).

Durability: prevention, not reversal

In the lab, metformin did not reverse the established fibrotic state; it prevented it from forming. The drug-addressable cells are concentrated in early nodules and largely gone once a cord is established PMID 19910144. So it should be tested in early disease or after a procedure to prevent recurrence, not expected to straighten an existing contracture. Continuous dosing fits the biology but commits patients to long-term treatment.

Draft trial outline

Adjuvant design first: patients having fasciectomy or needle fasciotomy randomised to metformin or placebo, started before the procedure and continued for 24 months; primary endpoint recurrence at 24 months using one pre-specified definition (OP_03). With a control recurrence rate of around 30–40% and a 40% relative reduction, roughly 470–710 participants. The control rate is the weakest input and is itself an open problem (OP_01).

The benchmark. It's the only drug ever tested on its own in a randomised trial in this disease, so it sets the standard the others have to meet.

Mechanism

TNF from M2 macrophages and mast cells acts through TNFR2 to drive fibroblast-to-myofibroblast conversion, via Wnt PMID 23431165. Myofibroblast IL-33 feeds back to stimulate more TNF, a self-sustaining loop PMID 31840071.

Evidence in Dupuytren's: randomised trial (4 of 4)

Phase 2a (n=28): the registered primary outcome was negative (no change in mRNA); protein secondary outcomes were positive at 40 mg PMID 29983350.
Phase 2b, RIDD (n=140, four injections over 12 months): nodule hardness −4.6 AU (95% CI −7.1 to −2.2); no related serious adverse events PMID 35949922.
Mechanistic limit: anti-TNF did not reduce type I collagen synthesis in Dupuytren's tissue PMID 40337865.

Safety, route and cost

Boxed warning for serious infections and malignancy. RIDD used four local injections three months apart, roughly an order of magnitude less drug than rheumatology dosing. A biologic, so copies are biosimilars rather than generics. Economics: £503,410 per QALY within the 12-month trial, but £14,593 per QALY over a lifetime if courses are repeated every three years PMID 36378072.

Durability: course-based, and that is an assumption

The favourable lifetime estimate assumes repeat courses every three years. That has never been tested: RIDD followed people to 18 months and the longer-term data are unpublished (OP_07). If adalimumab behaves like tamoxifen or triamcinolone, it is a suppressive treatment needing indefinite repeat injections: a different label, a different economic case and a different conversation with patients.

Why a phase 3 cannot be designed yet

Powering on nodule hardness needs roughly 136–160 participants, which is affordable, but hardness has no validated minimal important difference and has never been shown to predict contracture, so a positive result would not support a licence. Powering on hand function needs roughly 708–826. The rational next step is to validate the surrogate (OP_02) and report RIDD's longer-term functional data (OP_07). Any phase 3 must include at least 12 months off treatment (OP_08).

How the three compare

Comparison of adalimumab, pirfenidone and metformin
Attribute	Adalimumab	Pirfenidone	Metformin
Dupuytren's evidence	Randomised trial (4)	In vitro (1)	In vitro (1)
Fibrosis elsewhere	Trialled (2)	Approved (3)	Preclinical (1)
Independent human evidence	Trial data	None	Genetic studies support blood sugar as a cause; studies of metformin users disagree
Route	Injection into the nodule	Injection (formulation exists)	Oral
Boxed warning	Serious infection, malignancy	None	Lactic acidosis
Cost	Biosimilar biologic	Cheap generic	Negligible
Durability expectation	Suppression; repeat courses assumed	Suppression likely	Prevention; continuous dosing
Main unknown	Does the surrogate mean anything?	Does it work in humans at all?	Does it work in humans at all?

Adalimumab has the evidence and an unvalidated endpoint. Pirfenidone has the fibrosis credentials and no human data. Metformin has the cheapest, safest profile and the most independent lines of support, but the thinnest direct evidence. None is ready for a phase 3, and what holds all three back is shared: a validated endpoint and a known untreated progression rate. See Measuring outcomes.

Source: candidate_dossiers.md, with regulatory facts from Drugs@FDA (dataset dated 2026-10-29) and FDA labels. Full detail in the downloadable dossier.