Legg-Calvé-Perthes Disease

Legg-Calvé-Perthes disease, also called Perthes disease or coxa plana, is idiopathic avascular necrosis of the femoral head in a child. The blood supply to the capital femoral epiphysis is interrupted; the epiphysis dies, fragments, revascularises and remodels, and the whole process runs its course in 2-4 years whether or not anyone intervenes. What treatment can change is the shape the head has when it finishes, and that depends on keeping it within the acetabulum while it heals. Younger children have excellent potential for recovery because they remodel so well.

Why "idiopathic". The other causes of femoral head necrosis in children (trauma, sickle cell disease, steroid use) have an identifiable cause; Perthes has none. Repetitive trauma, a coagulopathy and an anatomical vascular watershed have all been proposed, and none is proven.

Who. A child aged 4-8 years, peak 5-6, and usually a boy: the male to female ratio is roughly 3-5:1 (3.1:1 in Swedish national data, Johansson 2016). Incidence is higher in children of Nordic, Caucasian and Asian background and lower in those of African ancestry. Affected children are often shorter than their peers, with a delayed bone age.

Associations.

• Low birth weight (below 2.5 kg)
• Delayed skeletal maturation (bone age)
• Lower socioeconomic status
• Second-hand smoke exposure
• Not associated with activity level

Blood supply. The paediatric femoral head is fed by three routes, and only one of them counts:

• Lateral epiphyseal vessels, branches of the medial femoral circumflex artery: the primary supply
• Foveal vessels from the ligamentum teres: minimal in young children
• Metaphyseal vessels: blocked by the growth plate

Unlike the adult, the child has no metaphyseal contribution across the physis, so before the age of 7-8 the lateral epiphyseal vessels are the only significant blood supply to the head. Interrupt them and the head is ischaemic; that single route is what makes the paediatric femoral head vulnerable.

What interrupts them. Four theories, none proven:

• Vascular occlusion: thrombosis of the lateral epiphyseal vessels, possibly from a coagulopathy
• Repetitive microtrauma: mechanical stress on vulnerable vessels
• Anatomical watershed: terminal vessels predisposed to ischaemia
• Venous outflow obstruction: raised intraosseous pressure

The four stages. Waldenström described the radiographic course in four stages.

The containment principle. During reossification the femoral head is soft and mouldable, and the acetabulum acts as the mould. Held within the socket, the head reossifies spherical and the long-term outcome is good; allowed to extrude laterally, it reossifies in a non-spherical, mushroom shape, the deformity is permanent, and an aspherical head means early osteoarthritis. Containment therefore has to be in place through fragmentation, when the head is most vulnerable and loss of containment leads to extrusion, and maintained while it reossifies.

The window. Because the head deforms during fragmentation, the value of containment depends on timing. It is most effective in the late-necrosis to early-fragmentation window, around stage IIa, before significant lateral pillar collapse: surgery during necrosis is premature, and once late fragmentation or reossification has set the head shape, containment can no longer restore sphericity and the role shifts to salvage. The modified Waldenström staging subdivides the classic stages to time that decision; it is the staging used in current outcome studies and showed good intra- and interobserver reliability (Nakamura 2023, PMID 36374570).

Several systems exist and they answer different questions. Herring's lateral pillar classification is the most commonly used and the most reliable for prognosis and treatment decisions, and is the one to reach for; Catterall is historical but still referenced; Salter-Thompson grades the subchondral fracture; Stulberg describes the result at skeletal maturity, not the disease.

Herring (Lateral Pillar)Catterall ClassificationSalter-ThompsonStulberg Classification

View larger

How to measure. The lateral pillar is the lateral 15-30% of the epiphysis on the AP radiograph. Compare the affected side with the normal side, and grade at maximal fragmentation, the peak of stage II. A hip that sits exactly at 50% height loss is classified as B/C border and treated as high-risk.

A

Lateral Pillar Height

No loss of height (100%)

Prognosis

Excellent

Treatment (age dependent)

Observation at all ages

B

Lateral Pillar Height

Less than 50% height loss

Prognosis

Good if under 8; guarded if over 8

Treatment (age dependent)

Observation under 8; consider containment over 8

B/C Border

Lateral Pillar Height

Exactly 50% height loss (borderline)

Prognosis

Intermediate (acts like C in practice)

Treatment (age dependent)

Containment recommended age over 6

C

Lateral Pillar Height

More than 50% height loss or total collapse

Prognosis

Poor

Treatment (age dependent)

Salvage procedures or palliative treatment

Group	Lateral Pillar Height	Prognosis	Treatment (age dependent)
A	No loss of height (100%)	Excellent	Observation at all ages
B	Less than 50% height loss	Good if under 8; guarded if over 8	Observation under 8; consider containment over 8
B/C Border	Exactly 50% height loss (borderline)	Intermediate (acts like C in practice)	Containment recommended age over 6
C	More than 50% height loss or total collapse	Poor	Salvage procedures or palliative treatment

View larger

Catterall grades the extent of epiphyseal involvement. Inter-observer agreement is lower than Herring's, but the groups are still quoted.

I

Description

Anterior epiphysis only

Involvement

Less than 25%

II

Description

More extensive, sequestrum present

Involvement

25-50%

III

Description

Most of head involved, only lateral portion intact

Involvement

50-75%

IV

Description

Entire head involved

Involvement

75-100%

Group	Description	Involvement
I	Anterior epiphysis only	Less than 25%
II	More extensive, sequestrum present	25-50%
III	Most of head involved, only lateral portion intact	50-75%
IV	Entire head involved	75-100%

Why Herring predicts better. A Catterall IV head, with the whole epiphysis involved, can have a different outcome depending on the lateral pillar: if the pillar maintains its height (Herring A or B) the prognosis is better, which is why Herring is more predictive than Catterall.

The head-at-risk signs. Five signs mark a hip likely to do badly regardless of its group. It is two or more together, or their persistence into reossification, that changes management. A single sign can be transient during the active phase, so do not escalate on one: document each sign explicitly and reassess on serial radiographs every 3 to 6 months while the disease is active.

Lateral subluxation

What it looks like

Lateral displacement of the head with widening of the medial joint space (Waldenström)

Why it matters

The head is breaking out of containment; deformity will follow

Gage sign

What it looks like

Small radiolucent V-shaped wedge at the lateral epiphyseal margin with speckled calcification

Why it matters

Lateral epiphyseal collapse; predicts aspherical healing

Diffuse metaphyseal reaction

What it looks like

Wide, ill-defined lucency or cyst in the metaphysis next to the physis

Why it matters

Severe physeal involvement; reduced remodelling potential

Horizontal growth plate

What it looks like

Physis lying horizontally rather than at its normal slope

Why it matters

Metaphyseal/physeal damage; disturbed further growth

Lateral calcification / loss of teardrop

What it looks like

Calcification lateral to the epiphysis, obliteration of the radiologic teardrop

Why it matters

The lateral buttress has failed; head uncovered and deforming

The five Catterall head-at-risk signs

Sign	What it looks like	Why it matters
Lateral subluxation	Lateral displacement of the head with widening of the medial joint space (Waldenström)	The head is breaking out of containment; deformity will follow
Gage sign	Small radiolucent V-shaped wedge at the lateral epiphyseal margin with speckled calcification	Lateral epiphyseal collapse; predicts aspherical healing
Diffuse metaphyseal reaction	Wide, ill-defined lucency or cyst in the metaphysis next to the physis	Severe physeal involvement; reduced remodelling potential
Horizontal growth plate	Physis lying horizontally rather than at its normal slope	Metaphyseal/physeal damage; disturbed further growth
Lateral calcification / loss of teardrop	Calcification lateral to the epiphysis, obliteration of the radiologic teardrop	The lateral buttress has failed; head uncovered and deforming

Grade only in the fragmentation phase

Catterall is valid in the fragmentation (active) phase only. A hip looks like Group I in the early avascular phase and appears to "improve" in reossification, so stating a group on the wrong phase is a classic error, and part of why the system's inter-observer reliability is only moderate.

Salter-Thompson grades the hip on the frog-leg lateral by how much of the head the subchondral fracture, the crescent sign, involves. It maps roughly onto Catterall: Group A corresponds to Catterall I-II and Group B to Catterall III-IV.

A

Description

Subchondral fracture involves less than 50% of head

B

Description

Subchondral fracture involves more than 50% of head

Group	Description
A	Subchondral fracture involves less than 50% of head
B	Subchondral fracture involves more than 50% of head

Stulberg is not a prognostic classification during the disease. It describes the final shape of the joint at skeletal maturity, and from that shape the risk of osteoarthritis.

View larger

I

Description

Normal spherical head

Risk of OA

None

II

Description

Spherical head, coxa magna, short neck

Risk of OA

Low

III

Description

Aspherical (ovoid) head, preserved concentric

Risk of OA

Moderate

IV

Description

Aspherical, flat head (coxa plana) with a flat acetabulum, still congruent

Risk of OA

High

V

Description

Flat head with a round acetabulum, incongruent

Risk of OA

Very high

Class	Description	Risk of OA
I	Normal spherical head	None
II	Spherical head, coxa magna, short neck	Low
III	Aspherical (ovoid) head, preserved concentric	Moderate
IV	Aspherical, flat head (coxa plana) with a flat acetabulum, still congruent	High
V	Flat head with a round acetabulum, incongruent	Very high

The two principles the classes encode. Quote these rather than reciting the table. Sphericity determines contact area: a spherical head spreads load over the maximum articular surface and minimises peak contact stress, while losing sphericity concentrates load and accelerates wear. Congruency determines whether the surfaces match: a congruent but aspherical joint redistributes load imperfectly yet symmetrically, whereas an incongruent joint edge-loads.

Why Class V is worse than Class IV. Both are non-spherical, so the difference is congruency. A flat head articulating with a still-spherical acetabulum point-loads at the edges of contact; a flat-on-flat congruent joint shares load across the whole mismatched surface and degenerates more slowly despite looking equally deformed.

Why Class II does well despite coxa magna. The head stays spherical, just enlarged, and the acetabulum remodels during growth to match it, preserving congruency. Long-term osteoarthritis risk approaches that of the general population.

History. A child of 4-8, peak 5-6, with an insidious onset over weeks to months of a painless limp or mild pain in the hip, thigh or knee. The pain is worse with activity and better with rest, and symptoms have usually been present for weeks before the family presents.

Examination. An antalgic limp, a positive Trendelenburg sign from pain-inhibited abductors, apparent shortening and, in the chronic presentation, thigh atrophy. The findings, in order of sensitivity, are:

1. Limited abduction, the most sensitive early sign
2. Limited internal rotation, especially in flexion
3. Protective muscle spasm in the hip flexors and adductors
4. Leg length discrepancy, an apparent shortening from muscle spasm
5. Trendelenburg gait from abductor dysfunction

Both hips, every time. Perthes is bilateral in 10-15% of cases, which is why both hips are radiographed. The pattern of bilateral involvement decides whether it is Perthes at all. True bilateral Perthes is almost always asynchronous: the two ischaemic events happen at different times, so the two hips are at different stages on the same film. Bilateral symmetric changes, both hips at the same stage and mirror images of each other, are not Perthes and should trigger a search for a generalised cause. Treating a symmetric dysplasia as Perthes is a classic error.

View larger

Faced with symmetric changes, order a skeletal survey, thyroid function tests and haemoglobin electrophoresis, ask about stature and family history, and remember Meyer dysplasia, a benign ossification delay in 2-4 year-olds that resolves on its own.

Radiographs. An AP pelvis and a frog-leg lateral of both hips. The AP is read for symmetry, lateral pillar height and extrusion; the frog-leg lateral for the crescent sign and the extent of involvement.

The crescent sign. A subchondral lucency on the frog-leg lateral representing a fracture through the necrotic epiphysis. It appears early, often before fragmentation is visible on the AP, it is the feature Salter-Thompson grades, and it indicates significant involvement.

View larger

What each stage looks like. The stages are radiographic, so the film tells you where the hip is:

View larger

Measurements. Three measurements are made against the normal side and the acetabulum:

• Lateral pillar height, compared with the normal side, for the Herring group
• Extrusion index: the percentage of the femoral head lateral to Perkin's line, normal less than 20%
• Epiphyseal quotient: width of the affected epiphysis divided by the normal side, a measure of coxa magna

MRI. The gold standard for early diagnosis, before any X-ray change, and the test to order when the radiograph is normal but symptoms persist beyond 3 weeks.

• T1: low signal in the necrotic epiphysis
• T2: variable signal, may show hyperaemia
• Gadolinium: lack of enhancement in the necrotic area

Bone scan or SPECT. Historically the early-diagnosis test, now largely replaced by MRI. It shows a photopenic cold spot in the femoral head with less anatomical detail than MRI gives.

Arthrography. Rarely needed, and then intraoperatively, to assess containment when surgery is being considered. It shows the cartilaginous contour of the head and whether the head is contained in abduction and internal rotation.

Principles. Treatment depends on age at onset and on the extent of involvement, graded by Herring. The goal is a spherical, congruent femoral head at skeletal maturity; the means is containment during reossification and a hip that keeps its range of motion, to prevent stiffness. The disease itself lasts 2-4 years whatever is done, so treatment changes the shape the head ends with, not how long it takes to get there.

The protocol depends on the operation, but the principles do not: keep the hip moving, watch containment on serial films, return the child to age-appropriate activity, and follow to skeletal maturity.

Follow-up schedule.

• Surgical cases: 2 weeks (wound), 6 weeks (healing), 3 months, 6 months, then every 6 months until reossification is complete
• Non-operative cases: every 3-4 months until reossification is complete
• Long-term: annually until skeletal maturity, to assign the final Stulberg class
• Transition to adult services for ongoing osteoarthritis risk monitoring

Prognostic factors. Age at onset is the most important; the others are read from the radiograph and the examination.

Favourable:

• Age at onset under 6 years
• Herring A or B, with less than 50% lateral pillar collapse
• Minimal extrusion (less than 20%)
• Male sex
• Maintained range of motion

Unfavourable:

• Age at onset over 8 years
• Herring B/C or C, with more than 50% lateral pillar collapse
• Significant extrusion (more than 20%)
• Female sex, worse at the same age and classification (in Herring's cohort, significantly worse than boys when over 8 at onset)
• Restricted range of motion, especially abduction

How the cohorts rank them. The studies on this page do not rank the factors identically, and a candidate should know that. Herring's 1992 series found the lateral pillar group a stronger predictor of final outcome than age at onset; his 2004 cohort found lateral pillar and age both strong independent factors; Wiig found femoral head involvement of more or less than 50% the strongest predictor (OR 7.76), ahead of age at diagnosis and lateral pillar group. Age at onset is nonetheless the most important factor, and with the Herring group it is what the treatment algorithm turns on.

Natural history without treatment. Under 6, most reach a spherically or aspherically congruent head (Stulberg I-III) without surgery. Between 6 and 8 the outcome is variable and depends on extent: B/C and C hips often progress to Stulberg IV-V. Over 8 a poor outcome is common, especially in Herring B/C or C.

What the Stulberg classes mean for the adult. Stulberg I-II give excellent long-term function with a low risk of osteoarthritis. Stulberg III carries a moderate risk, typically after age 50-60. Stulberg IV-V carry a high risk of early osteoarthritis, often requiring total hip arthroplasty by age 40-50, which is why a paediatric diagnosis has adult consequences.

Be careful calling Stulberg I-III a "good" outcome; only I and II are. Stulberg himself defined classes III and IV as aspherical congruency, in which mild-to-moderate arthritis develops in late adulthood, and Larson's 20-year data put numbers on it: Stulberg III hips were poor or fair in 61% on the Iowa Hip Score and 72% on the Nonarthritic Hip Score. Spherical congruency (I-II) is the only genuinely reassuring result.

What containment surgery can change. Done appropriately it may improve the outcome by one Stulberg class (IV to III). Where that benefit lies depends on which cohort is read: Herring located it only in lateral pillar B and B/C border hips over 8.0 years, and Wiig in children over 6 with more than 50% head involvement. The benefit is limited if the child is too young (under 6) or too old (over 10).

Perthes disease has no single international clinical-practice guideline; management is guided by the prospective cohort evidence above and by paediatric-orthopaedic registries. The picture below is deliberately global rather than tied to any one health system.

Global epidemiology (population-based):

Reported incidence therefore varies several-fold worldwide (single figures to mid-teens per 100,000), with a consistent male predominance (roughly 3-5:1) and a robust association with socioeconomic deprivation. A Swedish national-register study also found increased rates of obesity, hypertension and other osteochondroses in affected children, supporting a possible systemic component (Hailer 2018, DOI).

Where management genuinely differs (and where it agrees):