Legg-Calve-Perthes disease

• Legg-Calve-Perthes disease (LCPD) is an idiopathic avascular necrosis of the capital femoral epiphysis in children, most often between 4 and 7 years of age (peak 5 to 6). ^[1]
• It declares itself with an afebrile, slowly worsening limp and hip, thigh or knee pain; the child does not look ill. ^[3]
• The key danger is not the pain (which usually settles within weeks) but progressive deformation of the soft femoral head while it is revascularising, a process that takes 2 to 4 years. ^[1]
• Plain radiographs evolve through the modified Waldenstrom stages; early disease may look normal, so MRI (perfusion MRI with gadolinium) is useful for early assessment. ^[1][2]
• Management is a race against time: protect the spherical femoral head in the younger, earlier stages (containment, limited weight-bearing), remodel severe hinge abduction later, and operate on late sequelae only when pain or function demand it. ^[1]

By the end of this page you should be able to:

1. Explain why an idiopathic vascular insult to the femoral head produces a self-limiting but potentially deforming disease in children aged 2 to 14 years. ^[1]
2. Recognise the typical presentation and distinguish it from transient synovitis and septic arthritis of the hip. ^[3]
3. Read the modified Waldenstrom stages on plain radiographs and name the classifications used to judge prognosis and femoral head involvement. ^[1][2]
4. Interpret plain radiographs, perfusion MRI and arthrography, stating what each adds. ^[2]
5. Plan management by stage: observation and containment in early disease, surgical containment or pelvic osteotomy when extrusion is developing, remedial and then salvage surgery later. ^[1]
6. State the principal prognostic factors, particularly age at onset, and give an honest appraisal of what the evidence supports. ^[1][2]

"Perthes disease", "coxa plana" and "osteochondritis deformans juvenilis" are all names for the same condition. Pathologically it is not an osteochondritis but a vascular event: occlusion of the lateral epiphyseal vessels (or the medial circumflex artery feeding them) renders part or all of the epiphysis avascular, and the same child can revascularise spontaneously a few weeks later. ^[1]

Although the blood supply returns on its own, the bone has been weakened. In the early stages the femoral head deforms simply under body weight, and in the fragmentation stage the weakened trabeculae crumble. The self-limiting nature of the disease is its defining characteristic; the variable amount of deformation it leaves behind is its clinical significance. ^[1]

LCPD may affect children between 2 and 14 years of age, with the peak age of onset around 5 to 6 years. ^[1] It is between 1 and 11 per 100,000 children in the populations described, and males are affected about five times more often than females. ^[3][1] About 10% to 24% of patients have bilateral disease, usually not synchronous. ^[3]

• Age. The highest incidence is at 4 to 7 years, when the femoral epiphysis is supplied solely by the lateral epiphyseal branches of the medial circumflex artery. ^[1] Occurrence in adolescence is uncommon and carries a worse prognosis because revascularisation is slower and often incomplete. ^[1]
• Ethnicity. The disease predominantly affects white children from Northern Europe; equatorial regions and children of African ancestry have a very low incidence. ^[1]
• Geography and social deprivation. Incidence varies from 0.2 to 19.1 per 100,000 children aged 0 to 14 per year, equating to a lifetime risk of about 1 in 400 to 1 in 35,000. ^[1] Marked gradients exist within countries (Scotland reported more than twice the rate of London; northern and western Norwegian counties differ several-fold). ^[1] The most socioeconomically deprived households have been reported to have about four times as much disease as the most affluent, and secular incidence is falling as living standards improve. ^[1] In Merseyside the rate halved from 14.2 to 7.4 per 100,000 children per year between 1976 and 2009. ^[1]
• Heritability. Familial clustering is weak. Parents of affected children have a 0.5% to 0.8% chance of having had the disease and siblings a 2% to 4% chance; a Danish twin study found no LCPD in the co-twin of affected identical twins, implicating a shared environment rather than a genetic susceptibility. ^[1]
• Associations. Low birth weight shows a dose-response relationship with risk, and children with the disease have consistently been found to be shorter than their peers ("rostral sparing" growth disturbance), with congenital anomalies, behavioural hyperactivity and asthma more commonly recorded than expected. ^[1] Parental smoking has been associated in a dose-response manner. ^[1] The etiological factor itself remains elusive; thrombophilia studies have not supported a major coagulation defect. ^[1]

The cause of LCPD is idiopathic. The favoured mechanistic model is an interruption of the epiphyseal blood supply, not a primary thrombophilia. ^[1] No modifiable single risk factor is established, although the strong, street-by-street socioeconomic gradient and the associations with low birth weight, growth restriction, passive smoke exposure and congenital malformations point to an early environmental determinant, probably prenatal or in the first years of life. ^[1]

The only confirmed "risk factors" are therefore epidemiological associations, and they explain why the disease clusters in deprived communities but do not allow prediction in an individual child. ^[1]

At 4 to 7 years the capital femoral epiphysis receives its blood almost entirely from the lateral epiphyseal branches of the medial circumflex artery. Occlusion of either the artery or these end-vessels produces necrosis of part, or all, of the epiphysis. ^[1] Angiography, radioisotope studies and perfusion MRI confirm spontaneous revascularisation, usually recanalising the occluded vessels from the lateral side. If the epiphysis collapses before revascularisation completes, these channels may be obliterated and recovery is delayed by months to years; revascularisation can remain incomplete when onset is in adolescence. ^[1]

The full cascade matters for understanding both the presentation and the treatment rationale:

1. Early, reversible deformation (modified Waldenstrom Stages Ia, Ib, IIa). Soon after infarction the mechanical properties of the epiphysis change. Weight-bearing produces an average 7.2% increase in femoral head width and a 12.4% decrease in height on standing, dimensions that return when the child lies down. ^[1] Epiphyseal cartilage hypertrophy and extrusion of the lateral part of the head initiate. ^[1]
2. Irreversible deformation (Stages IIb, IIIa). Osteoclastic resorption of dead bone races ahead of osteoblastic new bone formation. The trabeculae crumble, the head collapses to a "mushroom" or is uniformly flattened, and extrusion beyond the acetabulum increases abruptly once it exceeds 20% of the head. ^[1] The femoral head is most vulnerable in Stage IIb. ^[1]
3. Stabilisation (Stage IIIb onwards). The new woven bone remodels into lamellar bone aligned to resist load; no further passive deformation occurs. ^[1]

Severity also varies between clinical subtypes. A "discoid" epiphysis, completely sclerotic and flat before fragmentation, behaves like a whole-head infarct and has a uniformly poor outcome. ^[1] Three patterns of evolution are described in adolescents (late-onset, segmental collapse and destructive), all doing worse than classic childhood Perthes. ^[1]

The presentation is remarkably quiet: an afebrile child, not ill, develops a painless or mildly painful limp over weeks to months. ^[3] Pain is usually groin, lateral hip, thigh or medial knee (referred pain via the obturator nerve), and intensifies during and after activity. ^[3] There are no constitutional symptoms, which is the single most useful clinical pointer away from infection or malignancy. ^[3]

Functional clues include avoidance of running and jumping, difficulty putting on shoes, and a child who opts out of sport. In older children, hinge abduction pain appears later as the disease deforms.

• Gait: antalgic, often Trendelenburg (trunk shift over the affected hip) when the abductors are overpowered by leverage loss. ^[3]
• Hip range of motion: limitation of abduction and internal rotation is characteristic, and internal rotation loss is a sensitive marker of intra-articular hip pathology. ^[3]
• Pain at the end of the arc of motion rather than throughout (contrast septic arthritis). ^[3]
• Quadriceps and hip abductor weakness on resisted testing. ^[3]
• Leg-length inequality on the Galeazzi test, from early epiphyseal height loss; bilateral disease (10% to 24%) can mask it. ^[3]
• Early disease may be virtually painless with a subtle antalgic gait only.

• Hinge abduction. By Stage IIb or IIIa a "bump" on the lateral femoral head can no longer slide under the acetabular rim and impinges; abduction pain and restriction recur after an earlier lull. ^[1]
• Adolescent Perthes. Onset after about 11 years: slower revascularisation, poorer outcome. ^[1]
• Discoid epiphysis. Whole-head involvement, flat from the outset, poor regardless of treatment. ^[1]

The most important differential is the child with an acutely irritable hip. Transient synovitis (TS) and septic arthritis (SA) share the age range and the limp. LCPD is usually more indolent, afebrile and less locally tender, but cultures and aspiration must be used whenever the picture is unclear. ^[3]

The Kocher criteria are a clinical prediction tool that estimates the likelihood that a child with an acutely painful hip has septic arthritis rather than transient synovitis. ^[3] Each criterion present increases the probability; four out of four makes septic arthritis very likely, while a well child meeting none is more consistent with transient synovitis. ^[3]

A child with an acutely painful hip still needs broad consideration of trauma, osteomyelitis, Perthes disease, slipped capital femoral epiphysis, Lyme arthritis and juvenile idiopathic arthritis even when Kocher scores are low. ^[3]

Figure 1Roe-perthes

Enlarge

Long description

Hip radiograph of Legg-Calve-Perthes disease.

Figure 2LCPdisease2015

Enlarge

Long description

Radiograph of Legg-Calve-Perthes disease of the right hip.

Anteroposterior and frog-leg lateral films of the hips are the first investigations. In early disease they may be normal or show only a subtle sclerotic focus in the capital epiphysis. ^[3] With time they reveal the classic sequence: sclerosis, then fragmentation, then flattening and eventual re-ossification of the proximal femoral epiphysis. ^[3] If the radiograph is equivocal, repeat imaging after a few weeks often clarifies the picture. ^[3]

The modified Waldenstrom classification underpins both prognosis and treatment planning. ^[1]

The original (four-stage) Waldenstrom scheme was Stage I avascular necrosis, Stage II fragmentation, Stage III reconstitution, Stage IV healed. ^[1]

Three radiographic classifications are commonly used, and each measures something different (extent of epiphyseal infarction, lateral pillar height, and final shape).

Catterall classification is the oldest (1971), describing how much of the epiphysis is sclerotic and whether the metaphysis is involved. ^[2] It carries prognostic weight but has poor reproducibility. ^[2]

Herring lateral pillar classification (1992) divides the femoral head into lateral (15 to 30% of width), central (50%) and medial (20 to 35%) pillars and grades involvement of the lateral one. ^[2] It was devised specifically because Catterall's scheme reproduced poorly, and it has the best intra- and inter-observer reliability of the three. ^[2] A B/C border stage exists when the lateral pillar keeps more than half its height but is thin (2 to 3 mm), less dense, or lower than the central pillar. ^[2]

A useful caveat: the extent of collapse on which Herring's groups depend may not be visible until Stage IIb. ^[1]

Stulberg classification (1981) is applied at skeletal maturity and is strongly prognostic for the long-term fate of the hip. ^[2] Only Stulberg I and II (spherical or near-spherical heads) count as acceptable; Stulberg III gives mild to moderate arthritis and Stulberg IV and V progress to earlier degeneration. ^[1][2] The Mose method (spherical, crescent or non-spherical) is an alternative that similarly tracks long-term outcome. ^[2] A newer quantitative score, the Sphericity Deviation Score, regards a value of 10 or below as a spherical head. ^[1]

Perfusion MRI with gadolinium delineates the avascular (black, non-enhancing) portion of the epiphysis at presentation, before fragmentation, and a low residual perfusion index predicts greater later deformation and higher Herring group. ^[2] It also demonstrates extrusion, hinge abduction and the state of the articular cartilage, none of which plain films show. ^[2] Diffusion-weighted and ADC sequences have been correlated with Herring stage but not yet with final outcome. ^[2] Expect practical limitations: it is usually a research or specialist tool, not a first-line investigation, and no specific MRI prognostic sign is yet universally accepted. ^[2]

An arthrogram under anaesthesia assesses whether hinge abduction is reducible and whether the femoral head is congruent in abduction. The Laredo arthrographic classification has high prognostic value but is rarely used in routine practice. ^[2] Scintigraphy has no prognostic value. ^[2]

There is no blood test for LCPD. Tests are sent to exclude the mimics: full blood count, erythrocyte sedimentation rate, C-reactive protein, and joint aspiration when infection cannot be excluded (see Kocher criteria above). ^[3]

Management depends on three interacting judgements: the child's age (bone age more than chronological), how much of the epiphysis is involved and extruded, and which Waldenstrom stage the disease has reached. ^[1][2] Earlier intervention carries a much better chance of a spherical head: the odds of healing with an aspherical femoral head were reported to be 16.58 times greater when treatment began in Stage IIb or later. ^[1] Because containment attempted in Stage IIb no longer protects the femoral head from deformation, the therapeutic window is effectively Stages Ia to IIa. ^[1]

Early disease (Stages Ia, Ib, IIa): preventive intervention. The goal is to stop reversible deformation becoming permanent. Two approaches are combined:

• Containment. Keep the femoral head well seated in the acetabulum so that extruded portions are covered and the load is distributed. Non-operative containment (broomstick plaster, abduction casts or braces) attempts to reverse extrusion by hip abduction; surgical containment achieves the same aim with a proximal femoral varus osteotomy or a pelvic osteotomy (Salter innominate, or triple innominate). A shelf (labral support) or Chiari osteotomy covers the extruded head without reorienting the bones. ^[1]
• Weight-relief. Avoid weight-bearing until Stage IIIb (about 24 months from treatment start in children treated from onset), since loading during this window is what deforms the head irreversibly. ^[1] Prolonged non-weight-bearing combined with containment has reported 75% spherical heads at healing, versus 49% with only 6 months of protection. ^[1]

In a recent prospective study with surgical containment, prolonged abduction bracing was found to be ineffective and is not recommended; A-frame and Pogo stick braces have reported good results in smaller series, but robust comparative data are lacking. ^[1]

Surgical containment is generally favoured for the older child or whenever extrusion is present, provided the disease has not progressed beyond Stage IIa. ^[1] Pre-emptive containment is recommended in children older than 7 years at onset, even without visible extrusion, because extrusion invariably develops in them. ^[1] In younger children with limited involvement, observation with activity modification, physiotherapy for hip range of motion, and careful radiographic follow-up at regular intervals is a legitimate strategy. ^[1][2]

Trochanteric epiphysiodesis performed at the same time as containment may reduce greater-trochanteric overgrowth where capital physis arrest is a risk. ^[1]

Later presentation (Stages IIb, IIIa): remedial intervention. Containment no longer reliably prevents deformation, so the aims shift to relieving pain (often hinge abduction) and limiting further damage. Hinge abduction is distinguished as reducible (the head reduces into the acetabulum under anaesthesia) versus irreducible; a shelf acetabuloplasty suits the reducible form, while a proximal femoral valgus osteotomy repositions the head for the irreducible form. ^[1] The femoral head is seldom spherical by the end of the disease at this stage. ^[1]

Late and healed disease (Stage IIIb onwards): salvage. Whatever shape the head has at Stage IIIb is permanent. Surgery here addresses sequelae: femoral head reshaping via safe surgical dislocation, correction of intra- and extra-articular impingement, and management of the acetabular rim. ^[1] Long-term outcome data for hip-preservation surgery are not yet mature enough to promise delay of total hip replacement. ^[1]

Femoral varus osteotomy is the most widely practised surgical containment and has the appeal of simplicity and a single incision; most studies report restoration of the neck-shaft angle without late genu valgum. ^[1] In many North American centres innominate (Salter) osteotomy, or triple pelvic osteotomy, is preferred, and combined femoral and pelvic osteotomies are used by some surgeons, without clear evidence of superiority over single procedures. ^[1] A second operation to remove metalwork is unavoidable with femoral osteotomy. ^[1] The surgeon also needs accessory procedures (trochanteric epiphysiodesis, hip contracture release) decided case by case. ^[1]

• Abduction braces appear ineffective in a prospective study, despite their theoretical grounding; the evidence base for most non-operative strategies is poor. ^[1]
• There is no Level I (randomised controlled) evidence for preventive containment. One recent prospective study found no benefit from surgical containment, conflicting with earlier North American and Norwegian studies. ^[1]
• No method of surgical containment is proven superior to the others. ^[1]
• Bisphosphonates and bone morphogenetic protein have shown promising results in animal models (suppressing osteoclastic resorption and enhancing new bone formation) and clinical trials are underway; they are not yet standard treatment. ^[1]

• Femoral head deformation: the core complication, classified by Stulberg at maturity; irreversible from Stage IIb. ^[1]
• Hinge abduction with recurrent pain and reduced range of motion, late in the disease. ^[1]
• Limb-length discrepancy from capital physeal arrest and coxa brevis ("greater trochanteric overgrowth"). ^[1]
• Coxa magna, short femoral neck, oblique acetabulum (Stulberg II features), usually with preserved function. ^[2]
• Early degenerative arthritis in Stulberg III to V hips, often leading to total hip replacement in early or mid-adulthood. ^[1][2]
• Osteochondritis dissecans of the femoral head after incomplete revascularisation. ^[1]
• Adduction contracture and trochanteric bursitis from altered biomechanics; botulinum toxin to adductors is sometimes used adjunctively. ^[1]
• Psychological burden of prolonged weight-bearing restriction in children started on non-operative programmes. ^[1]

Prognosis is dominated by four factors, in rough order of importance: age at onset, extent of epiphyseal involvement, lateral extrusion, and (in practice) compliance with treatment. ^[1][2]

• Age. The border between good and poor prognosis oscillates between 6 and 9 years. ^[2] Herring (2004) proposed 8 years chronologically or 6 years of bone age as the worse-prognosis threshold. ^[2] Children presenting before 6 years do much better on average, and one series found excellent outcomes with conservative treatment in them. ^[2] The age at onset distribution is log-normal, centred on 5 to 6 years. ^[1]
• Gender. Female sex is a less consistent risk factor, in part because of earlier bone maturation and shorter remodelling time at the same chronological age. ^[2] The female excess in severe disease may partly be an age-maturation effect rather than biology alone. ^[2]
• Body habitus. Overweight is a risk factor for poor outcome. ^[2]
• Clinical signs. Hip stiffness, especially adduction contracture or loss of abduction, worsens prognosis. ^[2]
• Extent of necrosis. Catterall grades III and IV and Herring group C predict poor outcomes, while group A predicts an almost universally spherical head. ^[2]
• Extrusion. More than 20% of the femoral head protruding beyond the acetabular margin is associated with poor outcome. ^[1]
• Discoid epiphysis. Even with early treatment, only about 25% end with a spherical head. ^[1]
• Adolescent onset. Prognosis is poor, especially the destructive pattern with negligible revascularisation. ^[1]
• Her versus environmental determinants. The clinical message distilled in the literature is that a stiff hip in an overweight girl over 6 is likely to finish with a poor outcome. ^[2]

Honest limits: beyond these broad factors, there is no reliable way to predict outcome in an individual hip early in the disease; "head-at-risk" Catterall signs are seldom visible in the stages where treatment decisions are actually made; and no randomised data settle how much of the reported benefit of containment is the treatment rather than the disease's natural tendency to do better in younger children. ^[1][2]

• Younger children (under 6). Favourable natural history; observation with monitoring is reasonable unless progression occurs. ^[2]
• Older children (school-aged and above). Extrusion develops almost inevitably; pre-emptive containment is recommended. ^[1]
• Girls. Earlier skeletal maturity compresses the remodelling window; treat bone age, not calendar age. ^[2]
• Bilateral disease. Subtle because the asymmetry used for diagnosis is absent; compare both hips carefully on every film. ^[3]
• Adolescents. Different patterns, worse biology; management is closer to adult femoral head collapse than to childhood Perthes. ^[1]
• Discoid and whole-head infarcts. Expect a poor result regardless; manage expectations and symptoms rather than expecting sphericity. ^[1]

• Plain anteroposterior and frog-leg lateral hip radiographs are available almost everywhere and are sufficient to diagnose established disease. ^[3]
• Early perthes may look normal; a child with persistent groin or knee pain and limited internal rotation with a normal radiograph needs repeat imaging after a few weeks or referral, not reassurance. ^[3]
• Close clinical follow-up substituted for MRI: pain continuing or recurring after a brief lull suggests hinge abduction or progression, both of which change management. ^[1]
• Non-weight-bearing and physiotherapy for abduction can be taught to families; bracing equipment and even plaster may be unavailable, in which case activity modification and monitored protected weight-bearing is the practical alternative. ^[1]
• Complex osteotomies, shelf procedures, and perfusion MRI require a paediatric orthopaedic service; the clinician's job in a low-resource setting is to recognise, stage as far as possible, refer early, and avoid harmful delays from repeated "watching" of what is already clearly beyond observation. ^[1][3]

Coxa magna and hip remodelling after childhood avascular necrosis; slipped capital femoral epiphysis (a disease of adolescence that also presents with a limp); transient synovitis and septic arthritis of the hip; developmental dysplasia of the hip; osteomyelitis; bone scintigraphy and perfusion MRI in paediatric hip disease.