Osteochondritis Dissecans of the Knee Abstract Osteochondritis dissecans is a condition of the joints that appears to affect subchondral bone primarily, with secondary effects on articular cartilage. With progression, this pathology may present clinically with symptoms related to the integrity of the articular cartilage. Early signs, associated with intact cartilage, may be related to a softening phenomenon and alteration in the mechanical properties of cartilage. Later stages, because of the lack of underlying support of the cartilage, can present with signs of articular cartilage separation, cartilage flaps, loose bodies, inflammatory synovitis, persistent or intermittent joint effusion, and, in severe cases, secondary joint degeneration. Selecting and recommending a surgical intervention require balancing application of nonsurgical interventions with assessment of the degree of articular cartilage stability and the potential for spontaneous recovery. T he etiology of osteochondritis dissecans (OCD), in contrast to its etymology, remains unclear. Al- though also described by Pare and Paget, the disease was named by Konig 1 in 1888, who by his nomen- clature indicated an inflammatory basis to explain a phenomenon of loose bodies in the joint. 1,2 No theory regarding the cause of OCD is uni- versally accepted, even though, as Konig later recognized, an inflamma- tory origin is unlikely. Repetitive mi- crotrauma, secondary effects associ- ated with vascular insufficiency, and potentially inherited factors remain important areas for investigation and clarification. Classification of OCD in the knee involves identification of a specific location, potential frag- mentation and/or displacement, and the status of the growth plate (Table 1). Skeletal age at onset of symptoms appears to be the most important de- terminant of prognosis and remains an essential factor , directing the tim- ing and nature of treatment deci- sions. Confusion regarding the etiol- ogy, treatment, and natural history o f these lesions is compounded by the common practice of referring to both osteochondritis dissecans and osteo- chondral defects (which can be sec- ondary to osteochondritis dissecans or to traumatic osteochondral frac- ture) as OCD. Incidence of OCD has been esti- mated at between 0.02% and 0.03%, based on a survey of knee radio- graphs, and at 1.2%, based on knee arthroscopy. 3,4 The highest rates ap- pear among patients aged between 10 and 15 years. Male-to-female ra- tio historically is approximately 2:1. Bilateral lesions, typically in differ- ent phases of development, are re- ported in 15% to 30% of cases, man- dating assessment of both knees in Dennis C. Crawford, MD, PhD Marc R. Safran, MD Dr. Crawford is Assistant Professor, Sports Orthopaedic and Arthroscopic Surgery, Department of Orthopaedics and Rehabilitation, Oregon Health & Science University, Portland, OR. Dr. Safran is Associate Professor and Director, Sports Medicine, Department of Orthopaedic Surgery, University of California, San Francisco, San Francisco, CA. None of the following authors or the departments with which they are affiliated has received anything of value from or owns stock in a commercial company or institution related directly or indirectly to the subject of this article: Dr. Crawford and Dr. Safran. Reprint requests: Dr. Safran, Department of Orthopaedic Surgery, University of California, 500 Parnassus Avenue, MU 320W, San Francisco, CA 94143-0728. J Am Acad Orthop Surg 2006;14:90- 100 Copyright 2006 by the American Academy of Orthopaedic Surgeons. 90 Journal of the American Academy of Orthopaedic Surgeons all patients presenting with this di- agnosis. 5 Etiology Lesions described as OCD can be caused by several factors, the signif- icance of which may vary depending on the area of the knee affected. 1,6 The essential mechanisms responsi- ble are divided into constitutional or hereditary, vascular, and traumatic. Constitutional Factors Ribbing 7 suggested that OCD may represent a variation or sub- group of epiphyseal dysplasia and thus may display a similar inheri- tance pattern. One report of familial predisposition to OCD-type lesions supports this idea; 8 however, Petrie, 9 in a study showing minimal trans- mittance to first-degree relatives, found limited evidence for a genetic pattern and suggested that the usual presentation is not familial. Despite this determination, an association of OCD has been found with a variety of inherited conditions, including dwarfism (described only as “short stature”), tibia vara, Legg-Calvé- Perthes disease, and Stickler’s syn- drome. 5 The relationship between the na- ture of developing OCD lesions, pos- sible hereditary factors, and the po- tential for abnormal ossification of the growth plate remains uncertain. Abnormalities of epiphyseal matura- tion are common and typically re- solve without long-term sequelae. Distinguishing normal ossification centers of the distal femur is critical in evaluating the young patient with knee pain. Caffey et al 10 describe the presence of irregularities of ossifica- tion in the distal femoral growth plates as the rule and explain them as an imbalance between rapid carti- lage proliferation and ossification. These areas are typically benign, re- solve without sequelae, and should not be confused with OCD lesions. Some investigators, however, have proposed that accessory nuclei might separate from these epiphys- eal areas and subsequently act as the precursors for OCD lesions. 9,11 More recent data from equine OCD stud- ies suggest a role for elevated matrix- metalloproteinase activity in sub- chondral bone. 12,13 Whether a predisposition to this condition is manifest in genetic inheritance is likely to prove to be multifactorial. Vascular Factors Analogous pathophysiology be- tween osteonecrosis and OCD con- stituted a popular theory for etiolo- gy among many early investigators (eg, Paget, Ficat, Enneking). 2 En- neking specifically championed a theory centered on poor end-arterial cascades in the distal femur and a predisposition for this bone to devel- op and behave in a manner analo- gous to a sequestrum. Often cited as evidence against this hypothesis is an anatomic study of 200 adult, 16 newborn, and 4 “juvenile” femurs that indicated extensive vascular anastomosis. 14 Further proof is found in another study that exam- ined six detached lesions from pa- tients with diagnosed osteochondri- tis with no histopathologic evidence of osteonecrosis. 11 Despite these in- vestigations, several recent reports have suggested that the cause is a paucity of vascular supply to the me- dial femoral condyle subjacent to the posterior cruciate ligament inser- tion, an area most commonly associ- ated with “classic” lesions of OCD. 15,16 Similarly, Linden and Tel- hag 17 demonstrated limited uptake of tetracycline and radionucleotide in 14 adults with OCD lesions and so concluded that the reparative pro- cess of subchondral bone was arrest- ed at a fibrocartilage stage, possibly the result of poor blood supply. Traumatic Factors A history of injury is reported in as many as 40% of patients with OCD, although some studies suggest a far more limited role for direct in- jury. 18 Cahill and colleagues 19,20 found no specific history of direct trauma among 204 patients in whom they attributed the pathology to a stress fracture. This theory is based on the unproven hypothesis that a series of pathologic reactions within articular cartilage and sub- chondral bone occur secondary to re- petitive microtrauma and yield a chronic osteochondral injury that manifests as an OCD lesion. Shear forces particular to the lat- eral aspect of the medial femoral condyle may be a contributing fac- tor. Fairbank 21 described repetitive impingement from the tibial spine as causal for OCD of the lateral as- pect of the medial femoral condyle. In this hypothesis, supported by bio- mechanical studies later performed by Nambu et al, 22 shear forces caused by impingement are generat- ed as the knee rotates medially with Table 1 Imaging and Arthroscopic Criteria for Classifying Osteochondritis Dissecans Lesions in the Knee Radiographic Magnetic Resonance Imaging* Arthroscopic Open vs closed physis Location of lesion Size of lesion Presence of loose bodies Low signal between the osseous fragments Low signal breaching the cartilage Focal defect ≥5mm Stable: Cartilage softening, cartilage breach Unstable: Cartilage flag tears, osteochondral loose body Osteochondral defects * T2-weighted (fluid-weighted) sequence Dennis C. Crawford, MD, PhD, and Marc R. Safran, MD Volume 14, Number 2, February 2006 91 loading in flexion. Smillie 23 also fa- vored this hypothesis, citing factors that could increase contact forces, including meniscectomy, instability, genu recurvatum, and condylar flat- tening. Several investigators subse- quently have shown an association between discoid lateral meniscus and the less common lateral femoral condylar lesion. 24-26 Specifically, the amount of direct microtrauma or macrotrauma necessary to produce an osteochondritis cannot be ascer- tained from the literature. What is clear is that distinguishing osteo- chondral fractures that fail to unite from lesions of OCD based on static radiographic and histologic evidence has proved to be difficult and re- mains controversial. Smillie 23 distinguished two essen- tial forms of OCD, juvenile and adult, and suggested unique etiolo- gies. In the variety manifested in the skeletally immature individual, there may be a fundamental disturbance of the epiphyseal development, with re- sultant formation of small accessory areas of subchondral bone that sepa- rate from the principal ossification center of the epiphyseal plate. Min- imal trauma, whether repetitive mi- crotrauma or direct macrotrauma, then may cause osteonecrosis within this region, as separation of the frag- ments disturbs the balance of oxygen tension necessary for ossification. In contrast to this type of developmen- tal etiology, Smillie postulated a more direct traumatic causation for the adult form. Clinical Presentation Cahill 19 and Mubarak and Carroll 8 emphasized a distinction between the juvenile and adult types of OCD, based on the osseous age of the pa- tient at the time of symptom onset. Those with open physes are consid- ered to have juvenile-onset OCD, whereas those with skeletal maturity are considered to have the adult form. Cahill 19 reported cases of adult- onset OCD in which radiographs taken during childhood did not reveal OCD. Later, during adulthood, the patients did have radiographically ap- parent OCD. However, the patients presented by Cahill likely had child- hood OCD that may not have been apparent on the plain radiographs, possibly because of the position of the lesion relative to the angle of the knee during radiography. It has clearly been shown that OCD (in the “classic” location, the lateral aspect of the medial femoral condyle) may be missed on posteroanterior radio- graphs with the knee in full exten- sion, yet may be visualized on flexed- knee views. 26 Adult-onset OCD may simply be a delayed onset of previ- ously asymptomatic juvenile OCD that failed to heal and presents later with loosening and joint degenera- tion. Early presentation often encom- passes poorly defined complaints. Pain is generalized to the anterior knee, with variable amounts of swelling that is typically intermit- tent. Anecdotal but consistent re- ports suggest an association between periods of increased activity and ep- isodes of swelling and effusion. An effusion may be found in association with joint synovitis and does not necessarily reflect a loose osteocarti- laginous fragment. The true source of this synovitis and/or effusion is elusive. In patients with more ad- vanced OCD, persistent swelling or effusion may be accompanied by catching, locking, or giving way. In late-stage disease, the sensation of a loose body is often described. Physical findings may be correlat- ed with the area of the lesion. Wil- son 27 describes an external rotation of the tibia during gait as signifying compensation for impingement of the tibial eminence on an OCD le- sion of the medial femoral condyle. Wilson’s test involves reproduction of pain on examination by internal- ly rotating the tibia during extension of the knee between 90° and 30°, then relieving the pain with tibial external rotation. The presumption is that, in internal rotation and ex- tension, the tibial eminence imping- es on the OCD lesion, causing pain, and that external rotation moves the eminence away from the lesion, re- lieving the pain. A recent case series has shown a poor predictive value of this maneuver with radiographically proven OCD. 28 However, the same authors suggest use of this maneu- ver, when it is initially positive, as a tool for following disease resolution. Standard techniques for testing sta- bility and joint palpation are neces- sary to identify concurrent patholo- gy, including loose bodies, associated meniscal tears, malalignment, and ligamentous injury. Imaging Studies Characterizing the lesion type and assessing growth plate status typically begins by making standard weight-bearing anteroposterior and lateral radiographs of both knees. Lateral radiographs allow recogni- tion of a relatively anteroposterior lesion location and identification of normal, benign accessory ossifica- tion centers in the skeletally imma- ture knee, as described by Caffey et al. 10 An axial view can be added when lesions of the patella or troch- lea are suspected. In addition, the ra- diographic “notch view,” taken with the knees bent 30° to 50°, may help identify the lesions in the posterior condyles. Plain radiographs provide initial data to determine lesion size, pres- ence or absence of sclerosis, poten- tial dissection, and assignment to one of several classification systems. Cahill and Berg 29 describe a method of localizing lesions by dividing the knee into 15 distinct alphanumeric zones (Figure 1). From medial to lat- eral, five zones numbered 1 through 5 are divided centrally by the notch; each compartment is then divided in half. The lateral radiograph uses Blu- mensaat’s line anteriorly and the posterior cortical line to divide zone A (anterior) from B (central) and C (posterior). This alphanumeric sys- Osteochondritis Dissecans of the Knee 92 Journal of the American Academy of Orthopaedic Surgeons tem provides standardization for re- search purposes, although it has found limited application to date. 6,26 Cahill and Berg 29 also describe a classification system for juvenile OCD based on technetium Tc 99m phosphate scintigraphy findings. Grading is based on the relative de- gree of scintigraphic activity in rela- tion to plain radiographs. Stage 0 is normal in both. Stage 1 demon- strates a defect on plain radiographs but no increased activity on the bone scan. Stage 2 shows increased uptake in the lesion but not in the adjacent femoral condyle. Stage 3 indicates isotope uptake in both the lesion and the adjacent condyle. Finally, stage 4 demonstrates increased isotope up- take in both the lesion and adjacent tibial surface. Patients with stage 3 or 4 disease were described as having symptomatic OCD. Cahill et al 20 lat- er reported limited correlation be- tween this staging system and pre- diction of lesion stability or the need for subsequent surgery. However, Paletta et al 30 suggested a role for this imaging technique that distin- guishes between results in juveniles and those in adults. They reported that four of four patients with open physeal plates and increased activity on bone scan healed with nonsurgi- cal treatment, whereas the two pa- tients without increased activity did not heal. In contrast, among patients with closed growth plates, only 33% (2/6) healed despite having similar increased activity within the le- sion. Magnetic resonance imaging (MRI) has proved to be particularly valuable in assessing osteochondral lesions. Several investigators have attempted to characterize the stabil- ity of the OCD lesion with findings on MRI. Dipaola et al 31 classified le- sions according to appearance on MRI and associated specific findings with the potential for fragment de- tachment. They described lesions containing fluid behind the joint as partially or completely detached, as evidenced by high signal intensity on T2-weighted images when a breach of the cartilage surface was detected. They distinguished carti- lage breach with an attached frag- ment by interpreting interposed low signal intensity on the rim as fibrous tissue (Figure 2). Others have added criteria for fragment instability to include the following: an area of increased ho- mogenous signal ≥5 mm in diameter beneath the lesion; a focal defect ≥5 mm in the articular surface; and a high signal line traversing the sub- chondral plate into the lesion. 32 In cases of limited joint effusion, Kramer et al 33 expressed a high level of confidence for predicting lesion stability using intra-articular gado- linium Gd 153 contrast material. More recent advances in MRI tech- nology (eg, cartilage-specific se- quences) may eliminate the necessi- ty of intra-articular injections and allow distinction between areas of interposed synovial fluid, fibrocarti- lage, and degenerated or lytic sub- chondral bone. Classification and Characterization Distribution of OCD lesions in the knee are most commonly associ- ated with the lateral aspect of the medial femoral condyle. Aichroth 26 described this as the classic location and confirmed it in 69% (72) of 105 knees (Figure 3). The patella was in- volved in five patients (5%); the re- mainder of the lesions involved the lateral femoral condyle (15% [16]) and the medial femoral condyle (69% [72]). In a large multicenter ret- Figure 1 Anteroposterior (A) and lateral (B) views of the knee, demonstrating the 15 alphanumeric radiographic regions described by Cahill and Berg. 29 The five numbered zones on the anteroposterior view are divided centrally by the notch (zone 3). The lettered zones on the lateral view are divided by Blumensaat’s line anteriorly and the posterior cortical line. The half-moon–shaped shaded area in each view of the distal femur represents an old lesion. (Adapted with permission from Cahill BR, Berg BC: 99m-Technetium phosphate compound joint scintigraphy in the management of juvenile osteochondritis dissecans of the femoral condyles. Am J Sports Med 1988;11:329-335.) Dennis C. Crawford, MD, PhD, and Marc R. Safran, MD Volume 14, Number 2, February 2006 93 rospective study of 713 patients and 798 knees, Hefti et al 5 described a slightly different distribution. The medial femoral condyle was typical- ly affected, with the majority of le- sions involving the lateral aspect (51%), 19% the central, and 7% the medial aspect. Involvement of the lateral condyle in all areas encom- passed 17% of lesions; those of the patella, 7%; and 0.2% (one lesion), the tibial plateau. Lateral condylar lesions are more commonly associ- ated with discoid meniscus or with occurrence after meniscal sur- gery. 26,34 Knee radiographs provide not only the initial basis for distinction of growth plate maturation but also as- sessment of lesion location and sta- bility (ie, free or loose bodies). Berndt and Harty 35 described four stages of chondral lesions based on plain ra- diographs of the talus; this system has been widely applied to lesions about the knee: stage I, involvement of a small area of compression of the subchondral bone; stage II, partially detached osteochondral fragment; stage III, completely detached frag- ment that remains in the underlying crater; and stage IV, complete detachment/loose body. Other crite- ria, such as lesion size, have been used to assess the potential for heal- ing with nonsurgical intervention. Several authors 20,32,36 have thought that patients could be successfully treated nonsurgically when the mean area was smaller than between 194 and 424 mm 2 . In contrast, le- sions larger than between 436 to 815 mm 2 were associated with poor out- comes. Others have suggested the presence of “marked sclerosis” as a poor predictor of successful nonsur- gical management. 37-39 Understanding and characterizing the spectrum of OCD lesions as sta- ble or unstable is often considered central to the treatment plan. How- ever, this characterization has proved to be difficult to determine prior to surgical intervention and of- ten remains a clinical judgment. MRI criteria have proved to be rea- sonably accurate compared with the gold standard of arthroscopic find- ings in predicting lesion integrity. Strict adherence to the MRI criteria of Dipaola was shown in one study 40 to have an 85% correlation with arthroscopic findings when ap- plying Guhl’s arthroscopic staging system. Guhl’s intraoperative classi- fication is defined by cartilage integ- rity and fragment stability. 37 Type I Figure 2 Sagittal MRI scans of unstable osteochondritis lesions of the distal femur. T2- weighted (fluid-weighted) images of osteochondral separation are indicated by high signal line between the osseous components (A) and extending from the intraosseous portion to the joint surface, “breaching” the cartilage (B). Figure 3 Anteroposterior radiographs demonstrating an OCD lesion on the lateral aspect of the medial femoral condyle before (A) and after (B) displacement. Osteochondritis Dissecans of the Knee 94 Journal of the American Academy of Orthopaedic Surgeons represents softening of cartilage but no breach; type II, breached cartilage that is stable; type III, a definable fragment that remains partially at- tached (flap lesion); and type IV, a loose body and osteochondral defect at the donor site. Further surgical characterization of OCD lesions, however, should not be limited to this system. Assess- ment of the size and number of loose fragments, the presence of bone as- sociated with each chondral frag- ment and its potential reparability, and the quality and character of the underlying subchondral bone (pres- ence of fibrocartilage or cystic de- generative material) are important factors in characterizing and surgi- cally treating this heterogenous pa- thology. Natural History and Prognosis No randomized, controlled clinical trials exist for either surgical or non- surgical interventions for OCD of the knee. In general, physeal maturi- ty, dissection of the lesion from the adjacent subchondral bone (stabili- ty), size and location of lesions, and integrity of the cartilage surface have been used as predictive criteria for surgical intervention. Historical- ly, data from case reports, case se- ries, and retrospective reviews have directed care of patients with OCD of the knee. Thus, caution should be exercised in proceeding with recom- mendations from these studies be- cause they provide only a limited ability to predict the natural history of OCD; that is, they are level IV and V evidence-based studies. A large, recent multicenter re- view of the European Paediatric Or- thopedic Society study (509 knees [318 juvenile, 191 adult] in 452 pa- tients) has provided notable data. 5 The authors made several important distinctions and reached a number of conclusions. (1) When there are no signs of dissection (defined as a sta- ble fragment), the prognosis is mark- edly better than it is with signs of dissection. (2) Pain and swelling are not good indicators of dissection. (3) Plain radiography and computed to- mography are not useful in predict- ing dissection. (4) Sclerosis on plain radiography predicts poor response to drilling. (5) Lesions >2 cm in di- ameter have a worse prognosis than smaller lesions. (6) When there is ev- idence of dissection, surgical results are better than those of nonsurgical treatment. (7) Lesions in the classic location had a better prognosis. (8) Although patients with adult-onset symptoms had a higher proportion of abnormal findings on radiographs after the treatment period (42%), more than one in five of those with open epiphyseal plates (22%) had ab- normal knee radiographs an average of 3 years after starting treatment. Pill et al 6 recently compared the value of MRI and clinical criteria for predicting the success of nonsurgical treatment of OCD. Their retrospec- tive review correlated outcomes to radiographic measures using the MRI criteria of DeSmet et al 32 and the radiographic criteria of Cahill and Berg. 29 Although they found no single factor to be uniformly predic- tive of successful nonsurgical treat- ment, several important associa- tions were found. Older patients with one or more signs of chondral disruption by MRI failed nonsurgical treatment most often. Similarly, larger lesions and lesions deemed to be within the weight-bearing area, as indicated by the lateral radiograph, also were most likely to fail nonsur- gical treatment. Younger patients with no MRI criteria for instability were most likely to recover with nonsurgical treatment. Management and Outcomes An algorithm for management deci- sions is given in Figure 4. The goal of nonsurgical treatment is to promote healing of lesions in situ and prevent lesion displacement. The preferred surgical goals are salvage of native cartilage, when possible, followed by restoration procedures. Initial discussion in early OCD involves patient, family, and physi- cian education. Understanding the nature of this disease, the potential long-term implications, and the timeline for management are crucial early strategies to help both the pa- tient and surgeon avoid frustration. Symptoms that are exacerbated by activity, particularly episodes of trauma and athletic and weight- bearing activities, should be identi- fied. Significantly limiting sports and high-impact activities is univer- sally recommended. Nonsurgical treatment is primari- ly mitigated through activity modi- fication. It may include a wide spec- trum of approaches that historically have included crutches (for limited weight bearing) as well as braces or even casts for noncompliant pa- tients. The efficacy of limiting activ- ity compared with limited or non– weight-bearing activity has not been studied in a controlled trial. Sales de Gauzy et al 38 followed a group of 30 children (mean age, 11 years 4 months) to complete resolution of symptoms by discontinuing sports activities; the authors recommended no surgical intervention because symptoms resolved with discontinu- ation of sports activities. In patients without marked sports participa- tion, prescribing a non–weight- bearing status and range-of-motion knee exercises may be beneficial to cartilage and may help avoid the po- tential disaster of “cast disease” (eg, joint stiffness, muscle atrophy, os- teoporosis) and arthrofibrosis associ- ated with cast immobilization. Symptom relief may be gained with nonsteroidal anti-inflammatory drugs (NSAIDs), but doing so may in- terfere with monitoring disease pro- gression. In young patients with ra- diographically and clinically stable lesions, we prefer to control pain with a non–anti-inflammatory med- ication (eg, acetaminophen). This Dennis C. Crawford, MD, PhD, and Marc R. Safran, MD Volume 14, Number 2, February 2006 95 avoids the theoretical negative influ- ence of NSAIDs on bone healing. In this strategy, we record days with swelling and those without and plot trends versus compliance with peri- ods of inactivity. This semiobjective, patient-reliant practice helps support recommendations for reduced activ- ity over a potentially long period of limited intervention. For patients with persistent pain or continued ep- isodes of swelling/effusions, deci- sions about surgical intervention may be supported by the data. In adults without evidence of loose fragments or unstable lesions, we employ a strategy similar to that used for patients with early, focal os- teoarthritis. Pain medication, activ- ity modification, strengthening, and weight control are the central tenets. Concurrent pathology (ie, malalign- ment, instability, osteoarthritis) is more likely in older patients and is an important consideration for both surgical and nonsurgical manage- ment of OCD. Choosing surgical intervention for OCD, and selecting a strategy for repair versus reconstruction or re- moval of osteochondral lesions, es- sentially depend on the stability of the inciting lesion and the integrity of the overlying cartilage. With fail- ure of the nonsurgical approach, sur- gical management often begins with arthroscopy. Classification using ar- throscopic findings is based on a de- scription of the lesion using two es- sential criteria: the integrity of the overlying articular cartilage and the stability of the lesion to distinguish three categories (intact, not intact but stable, not intact and unsta- ble). 19,37 Lesions with intact cartilage are considered either stable or unsta- ble. Lesions with damaged cartilage surfaces comprise a mixture of ad- vanced lesions and may be either disrupted or macerated, and by Ca- hill’s definition are unstable. Ca- hill 19 further subdivided the unstable-cartilage, injured lesion as predetached, hinged, or loose, attrib- uting characteristics of temporal symptoms and reduction congruity to each. Using this system, he rec- ommended a treatment algorithm, based on arthroscopic findings and scintigraphic data. We employ a sim- ilar approach using symptoms, ra- diographic and MRI findings, and ar- throscopic observations to apply treatments based on skeletal maturi- ty, osteochondral stability and re- ducibility, and articular cartilage in- tegrity (Figure 4). Surgical treatment for stable le- sions with normal articular cartilage involves drilling the subchondral bone with the intention of stimulat- Figure 4 Juvenile (open physis) Radiographs Adult (closed physis) Stable Physical examination Stable MRI Stable Bone scan Activity restriction (3 mos) • Impending physeal closure • Clinical signs of instability • Expanding lesion on plain films Arthroscopy Stable Physical examination MRI Stable Malalignment Yes Treat symptomatically Stable Unstable Osteotomy Stable Unstable reducible Fixation graft Fixation and graft chondrocyte transplant, or osteochondral graft Not positive No Positive Unstable with fragmentation or osteolysis Unstable and chondral damage Fixation and grafts, Transchondral drilling Treatment algorithm for knee osteochondritis dissecans. Osteochondritis Dissecans of the Knee 96 Journal of the American Academy of Orthopaedic Surgeons ing vascular ingrowth and subchon- dral bone healing. Retrograde tech- niques (defined as methods that avoid articular cartilage disruption using a transosseous approach) have given way to arthroscopically assist- ed methods that have proved to be highly efficacious in skeletally im- mature patients. Anderson and col- leagues 39,41 described success with this technique in 24 patients fol- lowed for an average of 5 years. Av- erage time to healing was 4 months. Twenty-two patients had good or ex- cellent results based on the Hugh- ston rating scale; in two of four skel- etally mature patients, the lesion did not heal. Kocher et al 42 treated 30 knees in 23 skeletally immature patients who had failed 6 months of nonsurgical therapy (average age, 12.3 years). Us- ing arthroscopically directed ante- grade (transchondral) drilling of sta- ble lesions, the authors reported radiographic healing in all 30 knees at an average of 4.4 months. They advocated this treatment in patients with intact articular surfaces who had failed nonsurgical treatment. Our experience has been the same, that drilling works better in skeletal- ly immature patients than in older patients, although it is still worth at- tempting in all patients with a per- sistently symptomatic lesion and in- tact articular cartilage. Surgical treatment for unstable lesions typically is attributed to Smillie 23 because he developed a nail for open reduction and internal fixa- tion of displaced and unstable le- sions. Surgical intervention to en- hance union has included Kirschner wires, cannulated screws, Herbert screws, and bone pegs. These typi- cally require a second surgery to re- move the device and have been asso- ciated with several complications, including wire migration, adjacent cartilage damage, and implant frac- ture. Biodegradable implants (ie, pins and screws) have the potential ad- vantage of not requiring removal; however, some of these devices have been associated with sterile abscess formation, synovitis, and loss of fix- ation. 43 Compressive devices provide the possibility of loading the osseous components, a technical advantage that may facilitate healing (Figure 5, E). When indicated, hardware re- moval often can be done arthroscop- ically, with low morbidity, and can provide an opportunity to directly assess healing and cartilage integrity. Simple removal of a loose or de- tached fragment is rarely considered to be an effective treatment, aside from cases in which the fragment is macerated and irreparable. Wright et al 44 reported only 35% good and ex- cellent results with fragment exci- sion at an average of 9 years after surgery. In cases in which simple trans- chondral drilling is unsuccessful, or when the lesion is hinged, loose, or displaced, the objective is to restore articular congruency by stimulating subchondral bone repair via com- pression and bone grafting, when necessary. In this manner, the os- seous portion of the fragment may heal and allow stabilization of the overlying articular surface. This strategy also would provide protec- tion to the adjacent uninjured car ti- lage that, after fragment excision, could be subject to increased contact stress and shear forces secondary to surface irregularity, step-off, and re- sultant edge loading. Green 18 used a similar argument to suggest that the technical difficulty of repairing loose fragments and subsequent malreduc- tion could have similar consequenc- es. His advice, which remains a te- net of joint surgery, was to replace larger fragments and remove those that were essentially too small to be fixed anatomically. Cahill 19 similar- ly advocated fixation whenever pos- sible because the results of excision usually are ineffective. Discrepancy in the size of a lesion as a result of overgrowth of displaced fragments, loss of fragment substance because of mechanical damage, or craterization of the donor site all have been described and provide technical challenges. Several strategies have been described to address these poten- tial issues. Johnson et al 45 treated 35 knees via an arthroscopically assisted technique that involved fragment fix- ation using cannulated AO-type screws. Results, comparable with other in situ methods, were good or excellent in 90% of cases. When poor congruency of the fragment-donor interface exists, a technique similar to that described by Anderson et al 41 may be used. In this method, the lesion is evaluated arthroscopically, followed by ante- grade open curettage, grafting, reduc- tion, and fixation (Figure 5, C and D). This method is done by reflecting a partially attached fragment, or re- moving it temporarily, to allow in- spection of the osseous surfaces and removal of fibrocartilage from the op- posing subchondral inter face. The ensuing fragment-crater mismatch then can be grafted with autogenous bone (tibial metaphysis) before com- pression screw fixation. We have used cannulated Acutrak (Acumed, Hillsboro, OR) headless screws or a 4.0-mm headed screw countersunk 1 to 2 mm below the cartilage surface to avoid causing articular car tilage le- sions on the opposing surface. Fol- lowing a period of strict non–weight- bearing and range-of-motion exercise, hardware is removed arthroscopically after a minimum of 6 weeks to as long as 12 weeks. Several techniques for salvage of full-thickness defects of articular cartilage, including autologous chondrocyte implantation, mosaic- plasty, and osteochondral allograft, have been advocated. Browne and Branch 46 have presented an algo- rithm for approaching these types of injuries. The efficacy of their tech- niques for addressing the symptoms of advanced OCD has been mixed. Some have advocated fixing loose osteochondral fragments with autol- ogous osteochondral autografts or by using autologous osteochondral au- tografts for filling empty craters to Dennis C. Crawford, MD, PhD, and Marc R. Safran, MD Volume 14, Number 2, February 2006 97 decrease edge loading. Case reports have indicated generally favorable results for these procedures but have limited follow-ups. Yoshizumi et al 47 describe successful union by 6 months in three cases of adult OCD using a modification of a method de- scribed by Berlet et al. 48 In the Berlet technique, the OCD lesion is essen- tially fixed in situ by applying pe- ripheral autologous osteochondral plugs. However, Yoshizumi et al 47 describe a technique using one cen- tral plug to fix the lesion. Others have advocated using either tech- nique to reduce edge loading. Anoth- er method for unloading cartilage for adult OCD patients was described by Slawski. 49 He performed seven valgus osteotomies for medial femo- ral condyle OCD and, at 2 years, de- scribed an average improvement in the Lysholm score from 39 to 89, with an average postoperative knee angle of 9º valgus. Use of autologous chondrocyte transplantation has been discussed and advocated by several authors. Peterson et al 50 reviewed their expe- rience with autologous chondrocyte Figure 5 Surgical reduction and fixation of an unstable osteochondritis dissecans injury. A, Preoperative anteroposterior radiograph demonstrating a loose and fragmenting chondritis dissecans of the medial femoral condyle. B, Intraoperative photograph indicating the margin (dotted area) of the extent of the loose/unattached articular cartilage. C, The osteochondral fragment is elevated. The fibrocartilage has been débrided from the interval between the osseous component prior to bone grafting. D, Fixation with two compression screws and absorbable pins. E, Postoperative anteroposterior radiograph. Osteochondritis Dissecans of the Knee 98 Journal of the American Academy of Orthopaedic Surgeons implantation in 94 patients at a min- imum of 2 years; 18 patients (19%) had chondral defects secondary to OCD. Defects in this group were characterized as particularly recalci- trant to previous surgery and often required a longer period to mature, compared with the other articular le- sions studied. Interestingly, 89% of patients (16/18) improved, with a similar distribution of excellent, good, and fair results compared with posttraumatic isolated femoral condyle lesions. Subsequently, these authors 51 re- ported results of 58 patients with a variety of OCD lesions. Results at a mean of 5.6 years were similar to those at a minimum of 2 years in the previous study. 50 Interestingly, in this larger study, a small group of pa- tients received bone graft for sub- chondral defects to provide a chon- drocyte bed prior to transplantation. Results of autologous osteochondral and allograft transplantation are gen- erally difficult to interpret; indica- tions in most studies for such proce- dures involve a variety of conditions, including osteonecrosis, osteoarthri- tis, trauma, osteochondral fracture, and OCD. Summary Understanding of the origins and natural history of OCD of the knee continues to progress. Tw o principle factors, skeletal maturity at symp- tom onset and contiguity of the sub- chondral and bone-cartilage surface, remain the most important determi- nants in choosing treatment. The challenge is identifying those deter- minants within the spectrum of dis- ease (eg, who may benefit from longer periods of nonsurgical man- agement versus earlier surgical treat- ment). OCD is not a benign condi- tion, even in the skeletally immature knee. Despite the fact that many patients are asymptomatic, the potential for arthrosis and degen- eration, demonstrated radiographi- cally, remains a problem, the exact consequences of which remain un- certain. Additional significant chal- lenges include dissecting and defin- ing the different subtypes of OCD, determining the potential of each for spontaneous healing or progression, and improving opportunities and techniques for intervention to main- tain and restore joint integrity. References Evidence-based Medicine: The au- thors note that there are no Level I or Level II evidence-based studies. Citation numbers printed in bold type indicate references published within the past 5 years. 1. Konig F: Ueber freie Korper in den Glenken. Zeiteschr Chir 1888;27:90- 109. 2. Schenck RC Jr, Goodnight JM: Osteo- chondritis dissecans. J Bone Joint Surg Am 1996;78:439-453. 3. Lindén B: Theincidence ofosteochon- dritis dissecans in the condyles of the femur. Acta Orthop Scand 1976;47: 664-667. 4. Bradley J, Dandy DJ: Osteochondritis dissecans and other lesions ofthe fem- oral condyles. J Bone Joint Surg Br 1989;71:518-522. 5. Hefti F, Beguiristain J, Krauspe R, et al: Osteochondritis dissecans: A mul- ticenter study of the European Pediat- ric Orthopedic Society. J Pediatric Orthop B 1999;8:231-245. 6. Pill SG, Ganley TJ, Milam RA, et al: Role of magnetic resonance imaging and clinical criteriain prediciting suc- cessful nonoperative treatment of os- teochondritis dissecans in children. J Pediatr Orthop 2003;23:102-108. 7. Ribbing S: The hereditary multiple epiphyseal disturbance and its conse- quences for the etiologies of local malacia-particularly the osteochon- dritis dissecans. Acta Orthop Scand 1955;24:286-998. 8. Mubarak SJ, Carroll NC: Familial os- teochondritis dissecans of the knee. Clin Orthop Relat Res 1979;140:131- 136. 9. Petrie PW: Etiology of osteochondritis dissecans: Failure to establish a famil- ial background. J Bone Joint Surg Br 1977;59:366-367. 10. Caffey J, Madell SH, Royer C, Morales P: Ossification of the distal femoral epiphysis. J Bone Joint Surg Am 1958;40:647-654. 11. Chiroff RT, Cooke CP III: Osteochon- dritis dissecans: A histologic and mi- croscopic analysis of surgically ex- cised lesions. J Trauma 1975;15:689- 696. 12. Laverty S, Okouneff S, Ionescu M, et al: Excessive degradation of type II collagen in articular cartilage in equine osteochondrosis. J Orthop Res 2002;20:1282-1289. 13. Al-Hizab F, Clegg PD, Thompson CC, Carter SD: Microscopic localization of active gelatinases in equine osteo- chondritis dissecans (OCD) cartilage. Osteoarthritis Cartilage 2002;10: 653-661. 14. Rogers WM, Gladstone H: Vascular foramina and arterial supply of the distal end of the femur. J Bone Joint Surg Am 1950;32:867-875. 15. Reddy AS, Frederick RW: Evaluation of the intraosseous and extraosseous blood supply to the distal femoral condyles. Am J Sports Med 1998;26: 415-419. 16. Lankes M, Petersen W, Hassenpflug J: Arterial supplies of the femoral condyle [German]. Z Orthop Ihre Grenzgeb 2000;138:174-180. 17. Linden B, Telhag H: Osteochondritis dissecans: A histologic and autoradio- graphic study in man. Acta Orthop Scand 1977;48:682-694. 18. Green JP: Osteochondritis dissecans of the knee. J Bone Joint Surg Br 1966;48:82-91. 19. Cahill BR: Osteochondritis dissecans of the knee: Treatment of juvenile and adult forms. J Am Acad Orthop Surg 1995;3:237-247. 20. Cahill BR, Phillips MR, Navarro R: The results of conservative manage- ment of juvenile osteochondritis dis- secans using jointscintigraphy: A pro- spective study. Am J Sports Med 1989;17:601-605. 21. Fairbank HAT: Osteochondritis disse- cans. J Bone Joint Surg Br 1933;21:67- 73. 22. Nambu T, Gasser B, Schneider E, Ban- di W, Perren SM: Deformation of the distal femur: A contribution towards the pathogenesis of osteochondritis dissecans in the knee joint. J Biomech 1991;24:421-433. 23. Smillie IS: Osteochondritis Disse- cans: Loose Body in Joints. Edin- burgh, UK: Churchill Livingstone, 1960. 24. Aichroth PM, Patel DV, Marx CL: Congenital discoid lateral meniscus in children: A follow-up study and evolution of management. J Bone Joint Surg Br 1991;73:932-936. 25. Stanitski CL, Bee J: Juvenile osteo- Dennis C. Crawford, MD, PhD, and Marc R. Safran, MD Volume 14, Number 2, February 2006 99