Dupuytren's disease & drug repurposing
Measuring hand function better
Why the current ways of measuring Dupuytren's make trials big and slow, and an idea I had for a better one using a phone camera.
Why it’s usefulEvery drug on this site is held up by the same measurement problem. Fix that, and every future trial gets cheaper.
How Dupuytren's is measured today
Finger angle
A clinician measures how far each finger is from straight with a protractor-like goniometer. It's direct, but it changes slowly, about 1.75–6.25° a year and only on the little-finger side PMID 35404338, and it's usually done once or twice a year in clinic.
Nodule hardness
A durometer presses on the lump. Softer means fewer active cells PMID 34617826. But nobody knows how much softening matters to patients, and 14% of readings couldn't be taken in the main trial PMID 35949922.
Questionnaires
PEM, URAM and QuickDASH ask how the hand affects daily life. They capture what patients feel, but they change slowly, which is why a new Dupuytren-specific questionnaire is being developed PMID 39547239.
The result is that a trial measuring nodule hardness needs about 136–160 patients but can't support a licence, and a trial measuring hand function needs about 708–826. There's no blood test or scan that could fill the gap PMID 38912961. Try the trial calculator
One fix for three problems
Trials are too big
A more sensitive measure picks up smaller changes, so fewer patients are needed.
Follow-up is too expensive
A measure people can do at home makes two to three years of follow-up affordable, including time off the drug.
An idea: a phone-camera hand screen Concept
Everyday hand use depends on two kinds of grip: precision grip (pinching a key or a coin) and power grip (wrapping the hand round a handle or a cup). Dupuytren's stops the ring and little fingers straightening, which changes how the hand opens and closes around things.
In my experience, the usual hand tests don't capture this well. The Purdue Pegboard measures dexterity with the fingertips, and the DASH questionnaire asks about the whole arm. Neither really shows how a curled little finger affects what the hand can do.
The idea is simple. Someone holds their hand up to a phone or laptop camera and copies a series of shapes on screen: a flat hand, spread fingers, a fist, a pinch. Google's free MediaPipe Hand Landmarker tracks 21 points on the hand from live video and can run inside a web browser. From those points you can work out:
- how far each finger joint straightens, joint by joint
- how closely the hand matches each target shape
- how quickly and smoothly it gets there
Because it's quick and free, people could do it every week or month at home. Many measurements over time give a much clearer picture of whether a hand is getting better or worse than one clinic visit a year.
What it would need to show
- It agrees with a clinician's goniometer, especially at the middle (PIP) joint.
- It agrees with existing measures of hand function, such as PEM, URAM and pegboard tests, so it can be validated against them.
- It picks up change sooner than current measures.
- People keep using it over two to three years.
- It works across different phones, lighting and skin tones.
- The video is processed on the device and no images of hands are stored.
This is an idea, not a tested tool. Phone-based joint-angle measurement has been studied in other settings, and that work would need reviewing first.
Why I'm interested
This came out of the project rather than going into it. My PhD looked at hand and arm function in older adults and after spinal cord injury, and along the way I built BATKAT, a tool for measuring grip function. So once the research kept pointing back to measurement as the real bottleneck, this was the part I recognised.
I probably won't build it myself, but I'd love to see it built. If you're a developer, a hand therapist, a surgeon or a researcher and this sounds interesting, get in touch (willshawphd@gmail.com).