Juvenile Osteochondritis Dissecans of the Knee: Trainee Guide

Definitions

OCD is an acquired, idiopathic lesion of subchondral bone that can lead to instability of the overlying articular cartilage and separation of an osteochondral fragment. König coined the term in the late 1880s, describing it as an inflammation of the bone-cartilage interface, a view later abandoned. [Masquijo 2019]

The AAOS 2010 guideline uses "skeletally immature" and "skeletally mature" rather than juvenile and adult, and gives separate recommendations for each.

Epidemiology

The true incidence is unknown. Estimates range from 2.3 to 31.6 per 100,000 and vary with age, sex and ethnicity. [Masquijo 2019]

• Age: rare before 6. Most present at 13 to 21. Incidence rose from 6.8 per 100,000 at ages 6 to 11, to 11.2 per 100,000 at ages 12 to 16. [Kessler 2014]
• Sex: male to female ratio 2:1 to 4:1. Female incidence may be rising with sports participation.
• Ethnicity: highest in non-Hispanic Black patients (31.6 per 100,000), lowest in Asian patients (4.7 per 100,000), in one US cohort.
• Bilateral: 2.7% to 30% (typically 14% to 30%). Up to 40% of bilateral cases are asymptomatic on the second side. No side preference.

Lesion location

The classic site is the lateral aspect of the medial femoral condyle. [Hefti 1999, multicentre series; figures as quoted in Masquijo 2019, which sum to over 100%]

ROCK prospective cohort (2022)

The largest prospective cohort so far: 1,004 knees in 903 patients from 17 US centres. 68.9% were male, median age 13.1 years (range 6.3 to 25.4). [Nissen 2022]

• Trochlear lesions were far more common than in older series (9.5% vs 1%).
• 68.1% played more than one sport. The main sports were basketball for boys (27.3%) and soccer for girls (27.6%).
• Initial treatment was surgical in 55.4% and activity restriction in 44.0%.
• At surgery, 48.1% of lesions were judged stable.

Aetiology

The cause is unknown and probably multifactorial. A widely held view is repetitive microtrauma to a vulnerable osteochondral area at a vascular watershed, in a genetically susceptible individual. [Masquijo 2019]

The ossification theory may explain why some lesions heal and others do not. With temporary arrest, normal ossification can resume.

Presentation and examination

There are no pathognomonic symptoms or signs. The typical picture is poorly localised, activity-related knee pain in a sporty child. [Masquijo 2019]

History

• Pain with activity, often vague in location
• Swelling
• Later: crepitus, catching or locking, suggesting instability or a loose body
• Some children are asymptomatic, with an incidental finding on imaging

Examination

• Inspection, point tenderness, effusion, range of motion
• Limb alignment
• Ligaments and menisci, for associated injury
• Always examine the hip. SCFE and Perthes disease commonly refer pain to the knee.

Wilson's test (pain on internal rotation of the tibia between 30 and 90 degrees of flexion, relieved by external rotation) has limited diagnostic value. It may help track clinical progress.

Differential diagnosis

• Normal ossification variants of the femoral condyles (ages 6 to 8): irregular, spiculated or fragmented ossification. Usually in the non-weight-bearing part of the condyle, without marrow oedema.
• Epiphyseal dysplasia
• Meniscal tear or symptomatic discoid meniscus
• Acute osteochondral fracture
• Patellofemoral pain
• Symptomatic medial plica
• Hip pathology: SCFE, Perthes disease

Imaging

Radiographs make the diagnosis. MRI characterises the lesion but cannot reliably determine stability on its own. [Masquijo 2019]

Radiographs

• Views: AP, lateral, tunnel (notch) and skyline (sunrise or Merchant). AAOS lists these as an option (Limited, unchanged in 2023).
• Condylar lesions are often posterior, so the tunnel view shows them better than the AP.
• Typical appearance: a well-circumscribed area of subchondral bone separated by a crescent-shaped, sclerotic, radiolucent outline.
• Contralateral films are advised in the review because of bilateral disease. AAOS 2010 was Inconclusive on this.
• Serial radiographs track healing in non-operative care.

MRI

• Sensitivity close to 100% for detecting JOCD.
• Shows overlying cartilage, subchondral bone, marrow oedema and associated injury (meniscus, ACL, cartilage).
• Useful for monitoring healing or progression.
• Poor at predicting instability in children. Specificity for fragment instability was 15% in one study. [Heywood 2011] MRI and arthroscopic morphology disagreed in 59.6% of femoral condyle lesions in another. [Roßbach 2016]
• Opinion is split on the reference standard. The 2019 review advises against using MRI alone to decide stability. A 2023 review argues MRI, read with juvenile criteria, should replace arthroscopy as the reference. [Accadbled 2023]

MRI instability criteria

Adult criteria over-call instability in children. With open physes, add the secondary signs. [Accadbled 2023, summarising De Smet and Kijowski]

• Closed physes: primary signs were 100% sensitive and specific against arthroscopy.
• Open physes: primary signs were 100% sensitive but only 11% specific. Specificity reached 100% with the secondary criteria.
• Age: instability was found in 3% of patients under 13 (2 of 71), but in all patients over 17 (7 of 7). [Siegall, via Accadbled 2023]

Reliability of MRI features

Ten ROCK surgeons rated 31 MRI features on 42 lesions. [Fabricant 2020]

The features most used to judge stability, the interface and the fragment, were the least reliable.

Classification

No system is universally accepted. All share one logic: progression from an intact surface to a loose body. [Masquijo 2019]

Hefti MRI classification

Commonly used for JOCD:

1. Small signal change without clear fragment margins
2. Osteochondral fragment with clear margins, no fluid between fragment and bed
3. Fluid partly visible between fragment and bed
4. Fluid completely surrounds the fragment, still in situ
5. Completely detached, displaced fragment (loose body)

ROCK arthroscopic classification

Developed by the Research in OsteoChondritis of the Knee (ROCK) group, with excellent intra- and inter-observer reliability. [Carey 2016]

A 2022 systematic review found 33 OCD classifications: 11 radiographic, 13 MRI and 9 arthroscopic. Across 193 clinical studies, the ICRS arthroscopic grade was used most. Only the ROCK arthroscopic system had published reliability data. [Andriolo 2022]

The ROCK group has also published a radiographic feature classification. [Wall 2015]

Management algorithm

This is one group's preferred pathway, not a validated guideline. Unstable lesions go to arthroscopy, which decides between drilling, fixation and salvage. It predates the 2023 ROCK RCT, which found transarticular drilling healed faster than the retroarticular drilling shown here.

Non-operative management

Non-operative care is first line for stable JOCD. About 50% to 67% of lesions heal within 6 to 12 months. [Masquijo 2019] Most authors give at least 3 to 6 months before considering surgery. No regimen has been shown to be better than another; AAOS 2010 could not recommend any specific one.

Options include immobilisation (cast, brace, splint, unloader brace), limited weight-bearing and activity restriction. Duration and timing are debated.

Example protocol: three phases [Kocher 2006]

Example protocol: casting [Wall 2008]

• 6 weeks weight-bearing cylinder cast.
• If radiographs showed re-ossification, cast removed. If not, 3 to 7 days out of cast for motion, then 4 to 6 more weeks in cast.
• Unloader brace, no running, jumping or sport. Radiographs every 6 to 8 weeks.
• Return to full activity only after complete re-ossification.
• At 6 months, 31 of 47 lesions (66%) were progressing to healing and pain-free.

Predictors of non-healing

• Larger lesion size
• Atypical location: non-weight-bearing lateral femoral condyle, patellofemoral lesions
• Effusion or mechanical symptoms at presentation
• Sclerosis around the lesion on radiographs (presence and extent)
• Closing or closed physes

If these are present, the review's algorithm moves earlier to drilling (see the management algorithm).

Newer evidence (2019 to 2026)

• Healing rates vary widely. In a 2025 review of 13 studies (710 patients, 783 knees, ages 8 to 18), non-operative healing ranged from 40.3% to 87.5%. [Muchintala 2025]
• ROCK prospective cohort, machine-learning analysis. In the subset analysed, 24 of 64 patients (37.5%) healed without surgery. [ROCK Group 2024]
  • Wider lesions, relative to condylar width, were less likely to heal (OR 1.41 per 1% increase).
  • Posterior lesions on sagittal MRI were more likely to heal (OR 0.08 for failure), as were lesions at the medial-most or lateral-most edge on coronal MRI (OR 0.05).
  • The model predicted healing with an AUC of 0.89 and 83.3% accuracy. It has not been validated externally.
• Return to sport. Among lesions that healed without surgery, 84.7% to 100% returned to sport. [Muchintala 2025]

Lesions on the main weight-bearing surface, and wider lesions, are the ones to watch most closely.

Operative management

Surgery is indicated for stable lesions that fail non-operative care, and for unstable or displaced lesions. Every attempt should be made to retain the native fragment. [Masquijo 2019]

Drilling (stable lesions, intact cartilage)

Drilling breaches the sclerotic margin, bringing in vascular ingrowth and growth factors from healthy cancellous bone.

In the review's summary of 17 level IV series (1981 to 2015), healing ranged from 70.6% to 100%. Where reported, mean time to radiographic healing was 3.4 to 11.8 months. No complications were reported in any series that recorded them.

ROCK randomised trial (2023). 91 skeletally immature patients with stable medial femoral condyle lesions that had failed at least 3 months of non-operative care. Transarticular (n = 51) vs retroarticular (n = 40) drilling. [Heyworth 2023]

This is the first level I evidence on JOCD drilling. It supports transarticular drilling for stable medial condyle lesions, and conflicts with the review's stated preference for retroarticular drilling.

Retroarticular technique (as described in the review):

1. Diagnostic arthroscopy confirms intact cartilage.
2. With fluoroscopy (AP, tunnel and true lateral views), pass a 1.6 mm K-wire percutaneously from below the physis to the centre of the lesion, stopping just under the cartilage.
3. Make 10 to 12 parallel perforations around the guide wire.
4. Pre-operative planning on MRI and lateral radiographs can reduce intra-operative radiation.

Internal fixation (unstable, salvageable)

Indicated whenever there is doubt about stability. Fixation is added to drilling.

• Debride fibrous tissue from the base and fragment (curette or shaver). Bone graft any void from the proximal tibia or iliac crest.
• Use 2 to 4 devices for compression and rotational stability. Choose lengths that avoid the physis (often 20 or 25 mm).
• Metal implants (cannulated or Herbert screws, staples): MRI artefact, often need removal, risk of migration, breakage and loosening.
• Bioabsorbable implants: PGA degrades in about 3 months with frequent foreign body reactions. PLA can take up to 6 years and may damage opposing cartilage. Newer PGA/PLA copolymers aim to balance this.
• Biological fixation: autologous osteochondral plugs provide graft and fixation together, with promising early results.
• Hybrid fixation: screws plus osteochondral plugs, mainly described in adults.

Smart Nail (PLA) fixation in 24 unstable lesions: 22 (91.7%) good to excellent, 2 needed reoperation. [Tabaddor 2010] Bioabsorbable pins in a Japanese series: 97% radiographic healing at 3.3 years. [Adachi 2015]

Does physeal status matter after fixation? In 87 unstable lesions fixed at one centre, 76% healed at 2 years or more. Healing was similar with open and closed physes (RR 0.68, 95% CI 0.29 to 1.72). Lateral femoral condyle location independently predicted failure (HR 4.25). [Wu 2018]

Return to sport after surgery for unstable lesions. A 2024 review (6 studies, 197 patients aged 15 to 25) reported return to the previous level of sport after: [Coladonato 2024]

Return took 6.5 to 9.7 months. Revision was more frequent after arthroscopic than open fixation (40% vs 5.9%). Previous surgery predicted worse outcomes.

Salvage (unsalvageable fragment)

Rare in children, with scarce outcome data.

Outcomes after ACI were better with symptoms under 12 months and fewer previous operations.

Long-term Swedish ACI data (median follow-up 19 years) showed graft failure in 43.3% of OCD lesions at 10 years, compared with 19.1% for traumatic cartilage lesions. [via Crawford 2026] Take this into account when counselling about ACI for OCD.

Rehabilitation and return to sport

Return to play requires radiographic healing and quadriceps strength back within normal limits. [Masquijo 2019]

After retroarticular drilling (review authors' protocol)

After fixation or salvage procedures, protocols vary by technique and are generally slower.

Return-to-sport timing in recent reviews

Sources: Muchintala 2025 (stable); Coladonato 2024 (unstable); timings as summarised in Crawford 2026. Protocols differ widely, so these are ranges, not targets.

Guideline position

• Post-operative physiotherapy should be offered (AAOS 2023, Consensus).
• AAOS 2010 could not recommend for or against physiotherapy in non-operative care (Inconclusive).
• Patients who remain symptomatic after treatment should have history, examination, and radiographs and/or MRI to assess healing (AAOS 2023, Consensus).

AAOS guideline (2023 rapid update)

AAOS approved a rapid update on 1 December 2023. It keeps 6 recommendations and upgrades 3. The 10 Inconclusive recommendations from 2010 were not revisited and moved to an appendix.

Still not addressed

The 2023 update did not revisit:

• contralateral knee X-rays
• non-operative treatment, for any group
• drilling of stable lesions, or drilling technique
• choice of cartilage repair for unsalvageable lesions
• repeat MRI in asymptomatic mature patients
• physiotherapy in non-operative care
• counselling on activity and weight to prevent osteoarthritis

The 2023 wording for surgery says "unstable or displaced" lesions. The 2010 wording said "salvageable unstable or displaced". The update added 42 supporting studies. The ROCK drilling RCT is not reflected in any recommendation, because inconclusive items are not revisited.