Herring Lateral Pillar Classification of Perthes

What Perthes is. Legg-Calvé-Perthes disease is idiopathic osteonecrosis (avascular necrosis) of the femoral capital epiphysis in a child. The epiphysis has a precarious blood supply, predominantly the medial femoral circumflex artery via the lateral epiphyseal/retinacular vessels; interruption of that supply is followed by revascularisation, structural weakening and repair. The classification assumes the disease, and the examiner will ask you to frame it.

Who gets it. Boys about 4-5 times more often than girls, with peak onset at 4-8 years (range about 2-14). Associations include short stature or delayed bone age, low birth weight and passive smoking. Disease is bilateral in roughly 10-15%, but asynchronous, with the two hips at different stages; simultaneous, same-stage bilateral disease should prompt a search for a skeletal dysplasia or another cause.

How it presents. An insidious, often painless limp with intermittent groin or thigh pain or knee pain referred along the obturator nerve, loss of hip abduction and internal rotation, a possible Trendelenburg gait and thigh atrophy. The child is well and afebrile, which distinguishes it from septic arthritis.

Waldenström radiographic stages. The disease passes through four stages, and the Herring grade belongs to the second:

1. Initial (necrotic): a smaller, denser, sclerotic epiphysis, widened joint space and a possible subchondral crescent
2. Fragmentation: the stage in which the Herring grade is assigned
3. Reossification (healing)
4. Remodelled (residual)

The bilateral-symmetric differential. When both hips are at the same stage, the conditions to search for are multiple epiphyseal dysplasia, spondyloepiphyseal dysplasia, hypothyroidism, sickle cell disease, Gaucher disease and steroid-induced avascular necrosis.

The Herring lateral pillar classification grades the height of the lateral pillar, the lateral one-third of the femoral capital epiphysis, on a true anteroposterior pelvic radiograph taken during the fragmentation stage of Perthes disease. It is the most widely used and validated radiographic prognostic tool in Perthes. Examiners expect you to classify a case from an AP radiograph, state the prognosis by age group, and explain how the group guides containment.

View larger

Finding the pillar. The lateral pillar is the lateral one-third of the epiphysis, bordered medially by a vertical line from the junction of the middle and lateral thirds. Its height is judged against the original height of the pillar; the central and medial pillars are not assessed.

Why the lateral pillar. It is the last part of the epiphysis to collapse in Perthes disease because it receives the most robust residual blood supply, from the lateral circumflex artery anastomoses. An intact lateral pillar mechanically supports the central and medial segments during reossification, which is why its preservation correlates so strongly with a good outcome: the lower the pillar, the worse the result.

The B-C border group. The original classification had three groups; the border group was added afterwards (Herring 2004, Part I) for the pillar that is thin or poorly ossified while retaining at least 50 percent of its height, or that has lost exactly 50 percent of its height. Head-at-risk signs such as lateral calcification or a Gage sign are not part of the definition. The Gage sign is a radiolucent V-shaped defect at the lateral epiphysis/metaphysis, an early sign of Perthes that may accompany lateral pillar involvement; on its own it does not define the Herring group but supports the diagnosis when seen alongside pillar changes. Adding the border group did not improve reliability or prognostic value on independent testing (Huhnstock): the interobserver study found it worsened both agreement and prognostic value compared with the original three groups.

Several classification systems exist for Perthes disease. The Herring lateral pillar system has largely superseded the earlier ones in clinical practice because of its superior inter-observer reliability and prognostic accuracy, and it has the strongest correlation with sphericity and the Stulberg outcome.

In the exam. If asked to classify Perthes, reach for Herring first. Mention Catterall to show breadth, but do not confuse the two: Catterall assesses involvement of the whole head in four groups, and Herring is more reproducible and prognostically superior.

Catterall's head-at-risk signs. These are markers of a poorer prognosis that support containment. The examiner expects you to define them, not just name them:

• Gage sign, defined above
• Calcification lateral to the epiphysis
• Lateral subluxation or extrusion of the femoral head
• A horizontal physis
• Metaphyseal cysts or reaction

View larger

The Stulberg classification. This grades the sphericity and congruence of the healed head at skeletal maturity, and is the outcome measure that predicts late arthritis. The governing principle is that aspherical incongruency (class V) predicts early osteoarthritis, whereas congruency (classes I to IV) is protective, which is exactly the spherical outcome that lateral-pillar-guided containment aims to preserve.

• Class I: a normal spherical head
• Class II: spherical, but with coxa magna, a short neck or a steep acetabulum. Classes I and II are "spherical congruency": good, with a low arthritis risk
• Class III: an ovoid, non-spherical (not flat) head
• Class IV: a flat head with a congruent flat acetabulum. Classes III and IV are "aspherical congruency": intermediate
• Class V: a flat head with a normal (incongruent) acetabulum, "aspherical incongruency": the worst, with premature osteoarthritis

The principle. Containment keeps the femoral head covered by the acetabulum during the fragmentation and reossification phases so that it heals spherically, and the lateral pillar classification directly guides the decision to pursue it. The group, read with the age at onset, sorts hips as follows:

• Group A: no containment needed. The lateral pillar is intact and the prognosis is excellent with observation and activity modification alone.
• Group B, chronological age 8 or under (or skeletal age 6 or under): observation. These hips did equally well with and without surgery in the prospective trial, so containment surgery adds risk without measured benefit.
• Group B, over 8 at onset: containment surgery is recommended. This is the one group with a demonstrated treatment benefit.
• Group B/C border, over 8 at onset: as for group B over 8, surgery outperformed non-operative treatment. Below that age the same caution applies as for group B, and the category itself is contested.
• Group C: the honest position is that no treatment has been shown to change the outcome. Group C hips had the least favourable results, and in the prospective trial there was no difference between operative and non-operative management at any age. Surgery is still performed in selected group C hips, for containment of a subluxing head, for hinge abduction, or where deformity is progressing, but it should be offered as an attempt to improve congruency in a specific mechanical problem, not as a general recommendation supported by outcome data, and the family should be told the prognosis is guarded either way.

The methods. Non-operative containment means Petrie casts, broomstick casts or abduction bracing; operative containment means femoral varus derotational osteotomy, Salter innominate osteotomy, double-level osteotomy or shelf acetabuloplasty. The principle is the same for all of them: maintain femoral head coverage within the acetabulum during the vulnerable reossification phase.

What the trial did to that list. It is not a menu of equals, and the evidence cuts it in two. In Herring's prospective multicentre study, bracing and range-of-motion therapy produced no better outcome than no treatment at all, so abduction bracing should not be offered as though it were containment; it commits a child to months in an orthosis for no measured benefit. Femoral varus and innominate osteotomy, by contrast, performed identically to each other, and both beat non-operative care in the one group where any treatment helped, lateral pillar B or B/C in a child over 8 at onset. The real decision is whether to contain surgically, not which operation to pick: choose between the two on concomitant acetabular dysplasia, range of motion and surgeon expertise rather than on any expected difference in outcome.

View larger

View larger

Accurate lateral pillar classification depends on the right film at the right time.

A true AP pelvis. The radiograph must be a true anteroposterior view of the pelvis, not a frog lateral or an oblique. Internal or external rotation of the limb distorts the apparent pillar height.

Timing. The classification is valid only during the fragmentation phase, typically 6 to 12 months after onset. Applied too early, in the initial avascular phase, it underestimates involvement; applied too late, in reossification, it overestimates recovery.

Serial films. Pillar height can change, so the group is a moving target during fragmentation. Reassess it on sequential AP pelvic radiographs at 3 to 4-month intervals and re-grade before committing to treatment.

The frog lateral. The lateral pillar is assessed on the AP view only; the frog lateral does not provide the pillar measurement. It is complementary, for the head-at-risk signs and for hinge abduction.

View larger

View larger

Age modifies the grade. The interaction between age at onset and lateral pillar group is the critical prognostic determinant: the same group carries a very different prognosis in a four-year-old and a ten-year-old. In the prospective multicentre trial the age threshold has two limbs, chronological 8.0 years and skeletal 6.0 years, and treatment had no significant effect in children at or under either at onset. Reading the skeletal figure as a chronological one would send six- to eight-year-olds to theatre for nothing.

View larger

Two courses, side by side. The figures below follow a group C hip and a group B hip from fragmentation to skeletal maturity.

View larger

View larger

• No single international guideline governs Perthes management. Treatment decisions are individualised based on age, lateral pillar group, range of motion, and surgeon preference. The evidence base consists primarily of retrospective cohort studies and expert consensus.
• POSNA (Pediatric Orthopaedic Society of North America) and EPOS (European Paediatric Orthopaedic Society) broadly endorse the lateral pillar classification as the preferred radiographic prognostic tool. Containment decisions follow the age-and-pillar framework described above.
• Global practice variation: North American surgeons more commonly use femoral varus osteotomy for containment. European surgeons, particularly in the UK and Scandinavia, more often favour Salter innominate osteotomy or non-operative containment (Petrie casts). Japanese centres have reported good results with prolonged bed rest and traction in younger children. These differences reflect training tradition rather than definitive evidence of superiority.
• The lateral pillar classification has been validated across multiple ethnic populations and healthcare settings, supporting its use as a global prognostic tool.
• Long-term follow-up to skeletal maturity is universally recommended regardless of treatment approach, because the Stulberg outcome cannot be reliably predicted before growth plate closure.