Kienböck's disease: avascular necrosis of the lunate

Mechanical wrist pain, stiffness, grip strength that is falling. That is the picture of a sprain that drags on or of a tendinopathy, and it is also the picture of Kienböck's disease. In the early stages the plain radiograph is normal: it rules nothing out. Here is when to be wary of it, what the stage really changes for management, and where the physiotherapist's role stops.

A synthesis written for physiotherapists. Every reference carries its PubMed identifier or its DOI, checked one by one. The disease is rare: the series are small, often old, and only one randomised trial compares two treatments with each other, on about ten patients per arm. The level of evidence is flagged with each statement rather than hidden behind a confident tone.

What is Kienböck's disease, and why does it go unnoticed?

A silent bone necrosis, in a bone that no clinical test really palpates, in young patients whose complaint looks like every other wrist complaint. This chapter sets out what is established, and points out what is not.

A bone, its place, and its problem

The lunate, or semilunar bone, is the central bone of the proximal row of the carpus. It articulates proximally with the radius, distally with the capitate, laterally with the scaphoid and medially with the triquetrum. This position makes it the keystone of the central column of the wrist: whatever happens to it spreads mechanically to the whole carpus.

Kienböck's disease is the osteonecrosis of this bone, described by the Viennese radiologist Robert Kienböck in 1910. More than a century later, the landmark review in Hand Clinics sums up the state of knowledge as follows: “the precise aetiology, natural history and optimal treatment remain controversial”4. That sentence is not a piece of academic caution. It describes exactly the ground on which the clinician works. The synthesis published by Lutsky and Beredjiklian in the Journal of Hand Surgery makes the same point from another angle: the natural history is poorly understood, and clinical findings often diverge from radiographic findings1.

The disease exists without hurting, and hurts without being serious

This is the most disconcerting result in the recent literature, and it comes from a study that did what nobody had done: look for the disease in people who were not complaining of it. Van Leeuwen and his team reviewed the reports of all MRI, computed tomography and radiographic examinations including a wrist in 51,071 patients from a single institution over eleven years3.

They found 51 incidental cases (0.10 %) and 87 symptomatic cases (0.17 %). In other words, more than a third of the cases of Kienböck's disease visible on imaging caused no complaint that led to a consultation. And the distribution of stages between the two groups is striking: collapse of the lunate (stages III and IV) concerned 18 % of incidental cases against 51 % of symptomatic cases (odds ratio 0.21; 95 % confidence interval: 0.086 to 0.51; p less than 0.001).

Symptoms are not a good indicator of the severity or the prognosis of the disease, and osteonecrosis of the lunate can exist without complaint, or with complaints too slight to prompt a consultation.

This conclusion of the authors has two opposite consequences, and both must be held. The first: a wrist that hurts little does not rule out a disease that is progressing. The second, just as important: an alarming image does not necessarily predict a poor functional future. The clinician has no reliable way of telling, among visible necroses, which will collapse and which will stabilise.

The blood supply of the lunate, and why it is fragile

The vascular explanation is the oldest and the most intuitive. It has been refined by a three-dimensional micro-computed tomography study on fourteen cadaveric specimens, with injection of a lead-based contrast medium20. Measured results:

• twelve specimens out of fourteen had nutrient vessels both palmar and dorsal; two had no dorsal vessel at all ;
• on average 2.3 palmar vessels (mean diameter 118.1 micrometres) and 1.4 dorsal vessels (135.8 micrometres);
• three intraosseous vascular patterns described, called Y, I and X;
• the poorest perfusion in the dorsoradial, palmar-ulnar and proximal regions.

Two specimens out of fourteen with no dorsal supply is a proportion high enough for the anatomical variation to matter. A bone whose entire supply passes through a single surface, with vessels of the order of a hundred micrometres, has no redundancy: an injury that damages this supply has no plan B.

The mechanism proposed by Bain and Lichtman: a stress fracture, then ischaemia

The Australian school led by Gregory Bain, with David Lichtman, proposes a sequence that reverses the classical order. It is not ischaemia that produces the fracture, it is the fracture that produces the ischaemia21. The reasoning, as published:

1. in an active subject, repeated loading generates a stress fracture that begins in the proximal subchondral bone plate, which is a single layer at that point;
2. the fracture arises where the lunate overhangs the edge of the radius;
3. it ruptures the parallel veins of the subarticular venous plexus, which produces a localised venous hypertension, then ischaemia and oedema of the fatty marrow;
4. the rise in pressure within the bone compartment worsens the venous obstruction: necrosis sets in.

This text must be read for what it is: a postulate, the word is used by the authors themselves. It has the merit of explaining why some lesions heal on their own when the fracture stays localised, and why others collapse when it is comminuted or crosses the bone in the coronal plane. But it is not demonstrated.

What the large databases have added since 2020

Three registry studies have changed the view of risk factors, by bringing numbers that surgical series will never have.

A familial clustering. By cross-referencing the Utah Population Database, which links genealogies going back to the early nineteenth century with 31 million medical records, Kazmers and his team identified 394 affected people spread across 194 unrelated families with a high incidence. The relative risk is significantly raised in first-degree relatives, and the disease is significantly associated with alcohol consumption, corticosteroid therapy, smoking and a history of diabetes22.

A national registry. The Finnish study published in 2026 draws on health registries from 1996 to 2022. It finds a higher incidence in men (incidence rate ratio 1.24; 95 % confidence interval: 1.15 to 1.33) and above all a set of associations with systemic conditions: alcohol abuse (odds ratio 1.67), rheumatic diseases (1.81), coagulation disorders (1.59) and, foremost, a history of hand or wrist trauma (odds ratio 3.03)23. Among operated patients, type 1 diabetes stands out with an odds ratio of 2.43.

A surgical procedure database. The ten-year review of the SPARCS database compares 499 patients operated on for Kienböck's disease with 6,823 controls operated on for wrist osteoarthritis. The reoperation rate there is 12 % (59 patients), lower than that of the osteoarthritic controls at every time point24.

Why does a normal radiograph not rule out the diagnosis?

This is the point that governs all the rest. A wrist radiograph reported as “normal” is the most reassuring and the most misleading document in the file. Here is why, and what to do about it.

Because stage I is defined by a normal radiograph

This is not a limitation of the examination, it is the definition of the stage. The Lichtman classification describes stage I as follows: “the lunate retains normal architecture and density, which gives normal plain radiographs”2. The necrosis is there, the bone is suffering, and the film shows nothing because there is nothing to show on a film: mineralisation has not yet changed.

A negative radiograph at stage I is therefore not a false negative of the examination. It is a true radiographic negative on a real disease. No improvement in film technique, no additional view will correct that.

Because the radiograph does not see what it is supposed to see

The problem goes beyond stage I. A Munich team assessed in 2023 the classical radiographic indices used to document carpal collapse (carpal height, Ståhl and Nattrass indices, radioscaphoid angle) in 301 patients. Interobserver agreement was excellent: the measurements are reproducible. But their ability to decide between stage IIIA and stage IIIB, with the usual thresholds, was poor: sensitivity 0.60 to 0.95, specificity 0.09 to 0.69, and area under the curve 58 to 66 %10.

An area under the curve of 58 % for a decision that points towards either joint-preserving surgery or salvage surgery is barely better than chance. The authors conclude that these indices “do not reach sufficient accuracy to differentiate stages IIIA and IIIB”. They are reliable measurements of something other than what we want to know.

Because MRI does not say the same thing as the radiograph, even when both are abnormal

A Korean team classified the same 88 patients with the modified Lichtman classification and then with a classification based on MRI, built on three elements: partial marrow oedema, cortical integrity of the lunate, dorsal subluxation of the scaphoid11. The shift between the two systems is massive, and it is concentrated exactly where the decision is made:

A word of caution about reading this chart: the publications give the numbers per stage, not the path of each patient. So we know that stage IIIB loses 23 patients and that stage IIIA gains 23; we do not know, from a published source, that these are the same ones. The authors confine themselves to writing that the greatest shift between the two systems is seen at stages IIIA and IIIB11. That is enough for the point that matters: in a population of 88 patients, the choice of examination moves a number equivalent to a quarter of the series across the heaviest therapeutic boundary.

What each examination contributes, and at which stage

The position of radiologists has been consistent for ten years. Arnaiz and his team put it as follows in the American Journal of Roentgenology : MRI is useful for diagnosis and staging, and should be considered after conventional radiography in patients suspected of Kienböck's disease12. The same article stresses a point that is too little known: there are Kienböck pseudo-lesions, that is signal abnormalities of the lunate that are not the disease, the first of which is ulnocarpal impaction.

How can the picture be recognised in a session, and when should an MRI be requested?

There is no clinical test specific to Kienböck's disease. No manoeuvre, no pathognomonic sign. Screening therefore rests entirely on reasoning: recognising a profile, measuring a change over time, and knowing when the absence of response becomes a signal.

The picture, as it actually presents

Reviews describe a disarmingly ordinary presentation: “unilateral reduction of wrist motion with pain and swelling”2. In Wassef's series of 57 cases, the mean symptom duration was 33.3 months at the time of analysis, and a history of wrist trauma was found in 53 % of cases5. Twenty-seven patients out of 57, that is 47 %, were already at stage IIIB or beyond.

That figure is the real subject of this article. Almost half of patients reach the surgeon with a lunate that has already collapsed and a disorganised carpus. Between the first symptom and that consultation, nearly three years have passed on average, during which somebody saw that wrist. Often a physiotherapist.

The features that recur, without any of them being specific:

• pain dorsal and central in the wrist, of mechanical pattern, made worse by weight-bearing through the palm;
• tenderness to direct pressure over the lunate, in the dorsal dip just distal to Lister's tubercle;
• loss of extension more than of flexion, and sometimes global stiffness;
• reduced grip strength, often the symptom that worries the patient more than the pain;
• sometimes a slight dorsal swelling, without frank effusion;
• slow course, with no identifiable acute episode, or decompensation of an old injury judged to be trivial.

The differential diagnosis of wrist pain

The lunate is a central bone: its distress projects to the middle of the wrist, where several structures can hurt. The approach consists less in finding the argument for Kienböck than in seriously exhausting the more common causes, then not stopping when they fail to account for the picture.

The decision tree, from the physiotherapist's side

The physiotherapist does not make the diagnosis of Kienböck's disease. He produces the fact that makes it possible: a measured trajectory that is not going the right way.

Measuring, so that the doubt can be passed on

A letter saying “the patient is not improving” does not carry the same weight as a letter saying “extension has gone from 62 to 51 degrees in six weeks, grip strength from 34 to 27 kilograms, unaffected side at 46”. The second triggers imaging, the first rarely does.

• Range of motion with a goniometer, flexion and extension, recording both sides. Loss of extension is the most frequently reported feature.
• Grip strength on a dynamometer, three trials, mean value, comparison with the opposite side. This is the measurement most sensitive to change in this condition, and the one that every surgical series reports.
• A validated functional scale : the DASH questionnaire or the Patient-Rated Wrist Evaluation, whose measurement properties are compared by Smith et al.49. Series on Kienböck's disease report one or the other: the DASH in Watanabe and in Matsui3233, the DASH and the patient-rated wrist evaluation in Chim's review34.
• The date, at each measurement. Without it, comparison makes no sense.

What does the Lichtman classification say, and what does it change for management?

The Lichtman classification dates from 1977. It remains the shared grammar of the whole literature and of every surgical discussion. It has two properties that must be held together: it governs treatment, and it is less reliable than is believed precisely where it governs most.

The stages, and what they describe

The principle is radiographic and progressive: normality, then increased density, then collapse, then osteoarthritis. The definitions below are those of the modified version, as published2.

Three reliability measurements, three different results

The reproducibility of the classification has been measured three times, twenty years apart, with different protocols. The results do not agree, and the divergence is instructive.

How should this gap from 0.71 to 0.203 be read? The protocols differ, and that is exactly the point. Jafarnia and Goldfarb had films read by four observers, all involved in the study and all specialists. Aydemir had forty films read by thirty professionals of three levels of experience, which is much closer to the real world where a film is read by whoever is there. His conclusion is explicit: “the Lichtman classification alone is insufficient and must be supplemented by other imaging and measurement techniques”.

A classification that governs treatment and whose agreement between readers is 0.203 under real conditions does not say what should be done. It says what should be considered.

What the authors of the classification say about it themselves

David Lichtman has not defended his system as an absolute. With Bain and Pientka, he published in 2017 and then in 2022 two successive proposals that add two further classifications to his own : a classification based on contrast-enhanced MRI, which assesses perfusion and viability, and an arthroscopic classification based on the state of the articular cartilage, known as the Bain and Begg classification738.

The guiding idea of these two papers is that the radiograph is not enough because it answers only one question out of three. A lunate can be deformed and alive, or preserved in appearance and dead; it can carry intact cartilage on a collapsed bone. The authors therefore propose crossing three axes: bone, perfusion, cartilage. The title of the 2022 paper, Precision Medicine for Kienböck Disease, says clearly enough that radiographic staging on its own is regarded by its own authors as a crude tool. The review by the American Academy of Orthopaedic Surgeons published in 2020 goes so far as to describe the disease as “well known but poorly understood”, and the very abundance of treatment options it lists is the symptom of that uncertainty6.

The series published by MacLean and Bain in 2021 illustrates what a decision based on the state of the joint rather than on the radiograph alone produces: 27 patients, median follow-up of ten years, pain down from a median of 5 to 2 on the visual analogue scale, 18 of the 20 working patients back at the same level of work, and only one secondary recourse to a salvage procedure36. A case series, without a control group: the result is encouraging, it does not establish the superiority of the approach.

Is negative ulnar variance really a cause?

It is the most cited hypothesis in the whole literature on this disease, and the one that governs the most frequent surgical choice. It is solidly associated with the disease. It has not been demonstrated to be its cause, and the distinction has practical consequences.

What are we talking about

Ulnar variance is the difference in height, on a posteroanterior radiograph, between the distal articular surface of the ulna and that of the radius. It is said to be negative when the ulna is shorter than the radius. The mechanical reasoning is immediate: if the ulna does not carry its share, the load transmitted by the hand passes preferentially through the radius, therefore through the lunate, which receives it overhanging the radial edge.

This reasoning has a direct surgical consequence: shorten the radius to restore the distribution. That is the radial shortening osteotomy, the most commonly performed operation in this disease. In the survey of members of the American Society for Surgery of the Hand, 90 % of surgeons say they change their strategy according to ulnar variance28.

What the association is worth, and what it is not worth

The association exists and it is robust. Bonzar and his team compared the ulnar variance of 44 affected patients with that of 99 controls from a general clinic, on standardised radiographs. They found, in both groups, a relationship between age and variance, and confirmed the association between negative variance and the disease after adjustment for age17.

The decisive work came fifteen years later, from a team in Tübingen. Stahl and his co-workers did three things in the same paper: a case-control study of 81 patients against 212 healthy controls, a meta-analysis of six case-control studies, and a systematic confrontation with the nine Bradford Hill criteria of causality16.

The second result deserves a pause. If negative ulnar variance caused the disease, an affected patient should have an affected wrist more negatively varied than his healthy wrist. The opposite is observed in three cases out of four. Yet the two wrists of the same subject are exposed to the same genome, the same hormones and much the same life. This result does not cancel the association, but it moves the question: negative variance could be a marker of a predisposition, not the mechanism of injury.

The third point is methodological and it is severe. In all the included studies, MRI was performed only in the affected group. In other words, the controls were never explored with the examination capable of detecting early forms. Some of them could therefore have carried a stage I Kienböck's disease that nobody looked for. The bias works exactly in the direction that inflates the association.

The measurement itself depends on the position of the wrist

There is a more elementary problem, known since 1982 and regularly forgotten. Epner, Bowers and Guilford showed that the measured ulnar variance changes with the position of the forearm at the time of the film : supination increases the measurement of negative variance, pronation reduces it18. Wrist deviation and the orientation of the beam in the longitudinal plane also influence the result. The authors proposed a standardised radiographic protocol, with internal landmarks, and recommended applying it to all wrist films.

Forty years later, ulnar variance is still measured on films whose standardisation is not always documented. A parameter that decides on an osteotomy and that moves with forearm rotation deserves to be read with that in mind.

The other hypotheses fare no better

The systematic review by the same Tübingen team went through 220 references published since the original description of 1910. One hundred and fifty-two articles were retained, and 140 of them, that is 92 %, are level IV evidence, that is case series19.

This finding is worth knowing in France, where the question of occupational attribution comes up regularly for trades exposed to vibrating tools. The state of the literature allows neither confirmation nor refutation of the link: the authors say so in those terms, and that absence of a conclusion is itself the information.

A less well known morphological factor: radial inclination

Wassef's series of 57 cases, published in 2025, tested a set of radiographic parameters against the stage at the time of diagnosis. Two stand out: a reduced radial inclination is associated with stage IIIB or beyond at the first consultation, and a reduced ulnar variance is as well5. The mean values in the series were 21.9 degrees of radial inclination, 10.7 millimetres of radial height and an ulnar variance of minus 1.23 millimetres.

This is a retrospective single-centre level IV study, with an exploratory model. The authors accordingly phrase their conclusion conditionally: this correlation “suggests that it could be a risk factor for progression”. To be taken as a lead, not as an established fact.

Which treatment options according to stage, and what is the evidence worth?

We have to start with the conclusion, because it changes the reading of everything that follows: only one randomised trial has compared two treatments of Kienböck's disease with each other, on numbers of about ten patients per arm, and it finds no difference. Everything else is case series, pooled by systematic reviews. What is written below describes what is done and what has been observed, never what has been shown to be superior.

The level of evidence, first

A Europe PMC query restricted to the publication type “randomised controlled trial”, searching every name of the disease, returns only two records for more than a century of literature.

The first is the only trial comparing two treatments with each other. Mazhar and his team randomly allocated patients at stage II or III between radial shortening and capitate shortening, with clinical and radiological assessment at 6 and 12 months. Fifty-two patients were assessed for eligibility; the paper reports 12 patients in the radial shortening group and 17 still present at the end of the study. None of the outcome measures differed significantly between the groups : range of motion, grip strength, pinch strength. The authors conclude that capitate shortening, performed through a smaller incision and in less time, may be suitable for stages II and III whatever the ulnar variance31.

This result must be read for what it is worth. About ten patients per arm does not allow equivalence to be concluded: the absence of a significant difference on such numbers is the expected result even if a real difference exists. And the abstract does not clearly report the number finally analysed. This is the best comparative level of evidence available for this disease.

The second randomised trial does not compare strategies: it tests low-intensity pulsed ultrasound on healing at the osteotomy site after forearm shortening, in 27 patients, with a 27 % reduction in the time to cortical union40. Useful, but that is a question of an adjunct, not of treatment choice.

The systematic review by Wang et al. devoted to stages IIIA and IIIB screened 1,489 titles to retain 30 articles. Their quality assessment gives 3 low-quality studies and 27 of moderate quality, none of high quality. The authors' conclusion: all modalities provide pain relief and functional improvement, and motion-preserving procedures give clinical results comparable with those of salvage procedures at both stages29.

Immobilisation: what it does, what it does not do

It is the first step, at stage I as often at stage II, and it is the one the physiotherapist meets downstream. The reported protocol is a splint or a below-elbow cast for at least three months2. In the ASSH survey, 74 % of surgeons choose it as first line at stage I28.

The review in the Journal of Bone and Joint Surgery British volume states its scope unambiguously: many patients are improved by temporary immobilisation of the wrist, which does not stop the progression of carpal collapse25. Both halves of that sentence must be taken. Immobilisation relieves. It does not cure.

Radial shortening osteotomy, at ten years and beyond

This is the best documented comparison, and its result is more nuanced than what is usually remembered. Shin and his team gathered 17 studies following patients for at least ten years: 5 studies of non-operative treatment, 12 of radial osteotomy, with comparable ages and stage distributions at the outset26.

This dissociation is the central message of the treatment of Kienböck's disease. Surgery improves what the patient experiences without changing the radiographic fate of his bone. We do not operate to stop the disease, we operate to live better with it.

Two series with very long follow-up confirm this order of magnitude. Watanabe et al. followed 19 patients operated on with a radial shortening, of whom 13 answered the questionnaire, with a mean follow-up of 21 years: mean DASH score of 8 out of 100, motion at 81 % in flexion and 82 % in extension of the opposite side, grip strength at 88 %; but radiographic progression in 6 of the 12 assessable patients32. Matsui et al., on 11 wrists followed for a mean of 14.3 years, report a modified Mayo score of 92 and a DASH of 5, with no progression of stage33. Two series, two opposite radiographic results, on numbers of about ten patients: that is exactly the level of certainty available.

The other procedures, and what each one costs

The special case of the child

The childhood form is not the same disease, and it is important not to transpose. Irisarri, Kalb and Ribak report 13 cases and a review of the literature: in children under 12, the prognosis is very good after immobilisation alone. Persistent symptoms may indicate surgery, probably a radial shortening, in some older adolescents, with a result that appears better than in adults39. The authors indeed propose to reserve the term lunatomalacia for this form, to distinguish it explicitly from the adult disease.

This contrast says something useful: in a growing bone, with a potential for revascularisation, the natural history is favourable. In the adult it is not, and no rehabilitation modality restores that potential.

What can and what cannot the physiotherapist do?

This chapter is the most important in the article, and the shortest to sum up: rehabilitation does not revascularise a lunate. What follows sets out why, what that forbids, and what remains, which is not nothing.

What the literature says about rehabilitation in this disease

Nothing. It has to be put that way. A Europe PMC query crossing Kienböck's disease with physiotherapy, physical therapy or occupational therapy brings up articles where rehabilitation appears only as a passing mention within a surgical protocol : “proximal row carpectomy with rehabilitation”, “conservative treatment then surgery and rehabilitation”. No study has assessed a rehabilitation programme in this condition, neither as a treatment, nor as an adjunct, nor against a comparator.

This absence is not surprising for a disease whose incidence is estimated at 7 per 100,0002. It is simply the fact we have to work with. An article presenting a rehabilitation protocol for Kienböck's disease would be making it up.

Why rehabilitation cannot treat the necrosis

The reasoning rests on three observations drawn from the previous chapters.

First, the problem is vascular and structural. The lunate receives on average 2.3 palmar vessels and 1.4 dorsal vessels, of the order of 118 to 136 micrometres in diameter, and two cadaveric specimens out of fourteen have no dorsal supply at all20. No rehabilitation modality creates a nutrient vessel. That is precisely what surgery attempts with vascularised grafts, and it fails to prevent progression in 13 % of cases despite a transplanted arterial supply35.

Second, even unloading surgery does not stop radiographic progression. That is the result of Shin et al. at ten years26 and of Salmon et al. at medium term27 : radial shortening relieves pain without reversing or preventing carpal collapse. If mechanically removing part of the load with an osteotomy is not enough, it would be unreasonable to expect more from a programme of exercises.

Third, symptoms do not follow the disease. A patient who is doing better under rehabilitation is not a patient whose lunate is doing better. Van Leeuwen et al. established this in the most direct way: symptoms are not a good indicator of severity or of prognosis3. Clinical improvement under treatment can perfectly well accompany structural worsening. It is the most dangerous trap of this condition for a rehabilitation professional, because it rewards the wrong decision.

What the physiotherapist can do, and what counts

The physiotherapist is not powerless in the face of Kienböck's disease. He is powerful at one precise moment, before the diagnosis, and modest at all the others.

The occupational factor, which must not be neglected

Two results converge on the same piece of advice. The systematic review of proximal row carpectomy notes that long-term results are worse in patients doing heavy manual work34. And the analysis of the SPARCS database on 499 operated patients identifies workers' compensation status as a factor increasing the risk of reoperation24.

The practical implication is direct and it falls to the physiotherapist: analysis of the workstation, the search for movement adaptations, referral to the occupational physician. In a tiler, a mechanic or a hairdresser, that work is worth more than one extra mobilisation session.

A word about what immobilisation leaves behind

Three months minimum of below-elbow immobilisation is the protocol reported at stage I2. The patient who arrives in the practice after that period presents a perfectly classical picture, and for once entirely within the rehabilitation professional's field of competence:

• wrist stiffness in both planes, often predominantly in extension;
• limited pronation and supination, secondary to immobilisation of the elbow when it was included;
• wasting of the forearm and loss of overall strength in the limb;
• stiffness of the metacarpophalangeal joints and loss of fine dexterity;
• apprehension in use, and established avoidance of the affected hand;
• sometimes a shoulder or cervical spine problem, developed during the period of compensation.

On this ground, rehabilitation does exactly what it knows how to do. The difference from a post-fracture immobilisation lies in one point, and it must be kept in mind: the progression of loading must remain under the surgeon's control, because the bone that was being spared is still necrotic, and because resuming weight-bearing through the palm, carrying a load or a screwing movement does not mean the same thing as on a healed radius.

What do real clinical cases teach us?

The cases below are all published, all identified by their PMID. They were chosen because they show what pooled series erase: the path by which a patient reaches the diagnosis, and what he went through beforehand.

The case that sums the article up: a normal radiograph and six weeks in a cast on a wrong diagnosis

A 32-year-old right-handed woman, a neonatal intensive care nurse, attends a sports medicine clinic for left wrist pain of two months' duration. The mechanism is unremarkable: hypersupination of the wrist while catching her daughter as she fell41.

What follows is the typical scenario. Radiographs had been taken three weeks after the injury: they showed no fracture, no dislocation, no malalignment. She had nevertheless been told she had a stress fracture, and she had worn a cast for six weeks. After it was removed the pain persisted; she was using an over-the-counter thumb splint, for comfort.

It was MRI that settled it: diffuse low signal of the lunate on T1 sequences, high T2 signal on fluid-sensitive sequences, that is avascular necrosis. Lichtman stage I, the stage at which the radiograph shows nothing by definition. Management, conducted with a hand surgeon, combined a short below-elbow cast then an orthosis, occupational therapy rehabilitation and a home exercise programme, with resolution of the pain.

When the first consultation comes too late

A 41-year-old woman, with no history of trauma, attends for chronic left wrist pain with restricted movement, which had worsened over some twenty days. Blood tests are of little help: erythrocyte sedimentation rate of 14 millimetres per hour, slightly prolonged clotting time, the rest normal. Imaging shows collapse of the lunate, joint space narrowing, subchondral cysts and secondary osteoarthritis. Treatment was a proximal row carpectomy, with clear relief of pain and functional improvement. The authors entitled their paper “Diagnostic Hurdles”42.

A 29-year-old woman, with no trauma and no classical risk factor, attends for progressive left wrist pain with restricted motion, of six months' duration. Radiographs, CT and MRI confirm avascular necrosis of the lunate at stage IIIB. After proximal row carpectomy, the DASH score falls from 62 preoperatively to 8 at one year, and the Mayo score rises from 45 to 85. At one year the patient is pain-free at rest, with satisfactory motion and a return to activities of daily living44.

What these two cases have in common is that they contradict two received ideas. The disease is not confined to men doing heavy manual work: neither of these two patients had any trauma or identified occupational exposure. And the delay before consulting is nothing out of the ordinary, six months and twenty days of worsening, which did not prevent an already disorganised carpus from being found.

Three patients, three stages, three different courses of action

A series of three cases published in 2025 illustrates well how the stage governs management43.

Three months of symptoms for a stage IV in a 28-year-old woman: the delay declared by the patient does not tell us how long the disease has been developing. Pain that appears is often the moment when an already diseased bone decompensates, not the moment when the disease begins.

Two cases that move the usual limits

A 17-year-old adolescent, right-handed, presents with severe and progressive wrist pain, resistant to conservative treatment, with a stage IIIB comprising collapse of the lunate and fixed rotation of the scaphoid. A proximal row carpectomy was performed. At two years he is pain-free, with full motion and strength, and he has gone back to competitive basketball. The authors stress that this result questions the usual contraindications based on age45.

A 74-year-old man attends for numbness of the right hand of several years' standing, and an inability to extend the middle and ring fingers that appeared two months earlier. The diagnosis made is a subcutaneous rupture of the extensor tendons caused by stage IV Kienböck's disease complicated by carpal tunnel syndrome. Treatment combined endoscopic carpal tunnel release and tendon transfers, with improvement at two years46.

This second case deserves to be known by every physiotherapist who treats hands. Advanced Kienböck's disease is not only a painful wrist: bone fragments and remodelling can wear a tendon through to rupture, and the picture then presents as a loss of finger extension. See the site's synthesis on carpal tunnel syndrome for the neurological side of this association.

A case of mixed management, conservative then surgical

A 32-year-old woman attends for chronic right wrist pain and a fall in grip strength of two years' duration. Examination finds tenderness over the lunate and restricted wrist motion. Radiographs and MRI show sclerosis and avascular necrosis of the lunate, classified at stage II. Management was first conservative, combining non-steroidal anti-inflammatory drugs and rehabilitation, before a radial shortening osteotomy, with clear improvement in pain and motion after surgery and rehabilitation47.

Two years of falling grip strength before the diagnosis was made, in a 32-year-old patient. It is very likely that a health professional saw that wrist during those two years. That is exactly the window in which a physiotherapist can act.

How can this be applied concretely in practice?

What this article really changes in a day at the practice comes down to a few actions, none of them spectacular.

At the initial assessment of a painful wrist

• Ask three questions that cost nothing: how long for, was there any injury however minor, and has imaging been done and when.
• Measure before treating. Flexion and extension with a goniometer on both sides, grip strength on a dynamometer over three trials with comparison with the opposite side, a validated functional scale. Record the date.
• Locate the pain precisely. Central and dorsal pain, which does not move with tendon testing and which matches no tendon course, must be recorded as such in the notes.
• Note the background : corticosteroid therapy, diabetes, alcohol consumption, rheumatic disease, coagulation disorder. All are associated with the disease in registry studies2223.
• Record the occupation and the dominant occupational movement. Heavy manual work is associated with poorer results after surgery34.

At reassessment

• Repeat the same measurements, with the same equipment and the same protocol. A measurement that cannot be compared is of no use.
• Set the referral threshold in advance : four to six weeks with no objective progress on well-conducted management.
• Treat the absence of change as information, and not as a problem of adherence or of dosage.
• Write a letter with figures. The baseline values, the current values, the dates, and a sentence asking for an opinion. Not a diagnosis.

After immobilisation or surgery

• Ask for the operative report and the stage. Management after a radial shortening osteotomy has nothing in common with that after a total arthrodesis.
• Ask the surgeon for the loading instructions, in writing, and do not extrapolate them from a distal radius fracture protocol.
• Recover useful range of motion before strength. A wrist that lacks extension is more of a nuisance in daily life than a wrist that lacks strength.
• Treat the distant consequences : elbow, shoulder, cervical spine, finger dexterity, apprehension in use.
• Adapt the workstation rather than trying to restore a strength that surgery does not give back.

What to tell the patient, and how

The information available online about this disease is of poor quality: in an analysis of 38 websites, the mean quality score was 13.3 out of 30 and the reading level corresponded to a tenth grade of schooling, far above the level recommended for a patient with limited literacy48. In other words, the patient who searches alone will search badly.

Three messages hold up and deserve to be said:

• This is a disease of the bone, not of the tendon or the muscle. That explains why rehabilitation alone does not resolve it, and why a surgical opinion is necessary.
• Nobody can predict the course. Neither the surgeon nor the rehabilitation professional. Monitoring takes the place of prediction.
• Feeling better does not mean the bone is better. That is why review appointments are kept even when all is going well.

Frequently asked questions

Is a normal wrist radiograph enough to rule out Kienböck's disease?

No. Stage I of the Lichtman classification is defined precisely by a normal radiograph: the lunate there retains normal architecture and density2. Only MRI then shows the disease, as a diffuse low T1 signal with enhancement after injection. Faced with mechanical wrist pain that persists beyond four to six weeks of management, a normal radiograph is no longer an argument for exclusion.

Is there a clinical test for Kienböck's disease?

None. There is neither a specific manoeuvre nor a pathognomonic sign. The usual presentation combines unilateral reduction of wrist motion, pain and swelling2, which overlaps the picture of many other wrist conditions. The diagnosis is made on imaging. The clinician's role is to recognise an abnormal trajectory and to document it.

Can rehabilitation cure Kienböck's disease?

No, and there is no evidence suggesting otherwise. The problem is a bone necrosis linked to a defect of blood supply, and no rehabilitation modality restores a vascular supply. The vascularised bone graft, which literally transplants an artery, itself fails to prevent progression in 13 % of patients35. Rehabilitation has a real role, but it lies before the diagnosis (screening) and after treatment (functional recovery).

Can grip strength be strengthened in an affected patient?

Not without knowing the stage and without instructions from the surgeon. Gripping loads the central column of the carpus, therefore the lunate, which is what all the surgery of this disease seeks to unload: joint-levelling osteotomies, capitate shortening, metaphyseal core decompression738. Grip strength remains an excellent indicator for follow-up, measured on a dynamometer at regular intervals. It is not a training goal as long as the diagnosis and the stage are not established.

Does immobilisation stop the disease from progressing?

No. The synthesis published in the Journal of Bone and Joint Surgery British volume is explicit: many patients are improved by temporary immobilisation of the wrist, which does not stop the progression of carpal collapse25. Immobilisation nevertheless remains the first step at stage I, for at least three months2, and 74 % of the surgeons surveyed choose it as first line at that stage28.

Should Kienböck's disease be operated on?

The question has no general answer. No randomised trial has compared surgery with no surgery, and only one has compared two techniques with each other: radial shortening against capitate shortening, with no difference found in range of motion or in strength31. What is known, at more than ten years of follow-up, comes down to a systematic review of 17 studies: moderate to severe pain persists in 5.7 % of patients operated on with a radial osteotomy against 23.2 % of patients not operated on, and the arc of motion is 107.4 degrees against 88.8, but radiographic worsening is comparable in both groups26. The decision is taken case by case, with a hand surgeon.

Does negative ulnar variance cause the disease?

It is associated with it, with an odds ratio of 3.58 in meta-analysis, but causality is not established: six of the nine Bradford Hill criteria do not support it, and in 75 % of cases the negative variance is identical or less marked on the unaffected side of the same patient16. The measurement itself varies with the position of the forearm at the time of the film18. That does not stop 90 % of surgeons from basing their strategy on this parameter28.

Is occupational vibration recognised as a cause?

The state of the literature allows neither confirmation nor refutation. The systematic review that examined this question across 220 references concludes that none of the nine Bradford Hill criteria was supported for the relationship between hand-transmitted vibration and Kienböck's disease, and that the disease therefore does not meet the International Labour Organization criteria defining an occupational disease. The authors explicitly call for studies of a better level of evidence19.

What is the prognosis in children?

It is good, and in no way comparable with that in adults. Irisarri, Kalb and Ribak report that in children under 12 the prognosis is very good after immobilisation alone; in older adolescents surgery may be necessary but the result appears better than in adults39. The authors propose to reserve the term lunatomalacia for this form, precisely in order to avoid transposing one prognosis to the other.

Does Kienböck's disease affect both wrists?

Rarely. Bilateral forms account for about 4 % of cases2. Bilateral involvement should prompt a search for a systemic cause, and registry studies note associations with corticosteroid therapy, diabetes, alcohol, coagulation disorders and rheumatic diseases2223.

What should be written in the letter to the doctor?

Figures, not a diagnostic hypothesis. Flexion and extension range measured with a goniometer on both sides, grip strength on a dynamometer with its comparison with the opposite side, the score on a validated functional scale49, each with its baseline date and its current date, and a sentence stating the absence of progress under well-conducted management. The referring doctor remains free to decide on referral; he has the fact that justifies it.

How can it be told apart from ulnocarpal impaction?

Topography points the way first: Kienböck sits centrally and dorsally, ulnocarpal impaction on the ulnar side, with pain reproduced on ulnar deviation and loaded pronation, and an ulnar variance that is readily positive13. The confusion is above all a trap on imaging: ulnocarpal impaction produces lunate signal abnormalities described as Kienböck pseudo-lesions12. So it is the reading of the sequences, and not the mere presence of a signal abnormality, that separates them: reviews of imaging of the painful wrist set out the criteria to weigh up15.

Why does the patient find contradictory information online?

Because the quality of the information available is poor, and this has been measured. Across 38 websites analysed, the mean quality score was 13.3 out of 30, the accuracy score 10.4 out of 12, and the reading level corresponded to a tenth grade of schooling, far above the level recommended for a patient with limited literacy48. The websites that were hardest to read were also the least accurate. The authors recommend directing patients to specialty society websites and referring back to the doctor in case of doubt.