549

Osteochondritis Dissecans of the Knee: Value of MR Imaging in Determining Lesion Stability and the Presence of Articular Cartilage Defects

Arthur A. De Smet1 David R. Fisher1

Ben K. Graf

Osteochondntis dissecans is a lesion of articular surfaces that is of uncertain These lesions are seen on radiographs as a bony defect or fragmentation

etiology. of the

subchondral

bone. A bony defect

may be an actual

surface

hole or the defect

may be

Richard

H. Lange2

filled with fibrous tissue be only partially attached

or fibrocartilage. Similarly, the apparent bone fragments may so they are unstable and prone to displacement or they may be firmly attached with fibrous tissue. Knowledge of fragment stability and the presence of an articular cartilage defect is useful in deciding on treatment. This information cannot be determined on plain films or clinical examination. We correlated MR examinations with arthroscopic findings in 21 patients with osteochondntis dissecans of the knee to see if MR imaging could be used to predict lesion stability and articular cartilage defects. A high-signal interface between the lesion and the femur was used as evidence of lesion instability and was found in 15 lesions. One of these lesions was questionably stable at surgery; the remainder were unstable and partially attached. The other six
patients had displaced fragments with large articular defects that were clearly visualized

on the MR examinations. We conclude that MR imaging is useful in evaluating lesion stability in patients with osteochondntis dissecans.

articular

surface

defects

and

AJR

155:549-553,

September

1990

Osteochondritis dissecans (OCD) of a femoral condyle is a common articular lesion in adolescents and young adults. Although most lesions are thought to be traumatic in origin, other proposed causes include mechanical stress, familial dysplasia, peripheral avascular necrosis, and fat emboli [1 , 2]. In some patients with OCD, the decision of whether to operate immediately or manage conservatively can be difficult [3, 4]. The plain films may show either a subchondral bone defect or fragmentation of the subchondral bone. If there are partially attached cartilage or bone fragments, the fragments should be fixed or drilled to help healing, or they should be debnded [5-7]. Plain films and clinical examination cannot determine whether fragments are

attached firmly or only partially. Accordingly, a number of techniques including radionuclide bone scanning [4, 8-1 0], computed arthrotomography [1 1], and MR
imaging
Received April 2, 1990; accepted May 7, 1990.
I

[1 0, 1 2-1 4] have been used cartilage defects

to determine with distal

the status

of OCD OCD

lesions.

after revision

We report our experience

articular

in using MR imaging
in 21 patients

to evaluate
femoral

lesion stability
in whom

and

Department of Radiology, University of Wiscon-

arthro-

sin Clinical Science Center, 600 Highland Ave., Madison, WI 53792. Address reprint requests to A. A. De Smet. 2 Division of Orthopedic Surgery, University of

scopic correlation
Materials and

of the findings

was available.

Methods

Wisconsin
53792.

Clinical Science Center,

Madison,

WI

0361 -803x/90/1 553-0549 C American Roentgen Ray Society

MR scans of the knees were obtained in 47 patients with OCD of one of the femoral condyles. The 21 patients who also had arthroscopic evaluation and treatment of OCD form the basis of this report. There were 1 8 men and three women with an average age of 19 years (range, 1 2-35). The adjacent growth plate was closed in 1 7 of the 21 patients. All the

550

DE

SMET

ET AL.

AJR:155, September

1990

patients
5 and

had chronic
7 months,

knee pain. Two

the

patients

had had knee pain for

had had pain for 2-15

displaced

fragments

within

the

joint

(Fig.

4D),

and

(5) defects

in the findpartially findings,

respectively;

remainder

years. Three of the patients who had a history of mild knee pain were shown to have acutely displaced bone fragments on radiographs.
The classification of plain film findings devised by Hughston et al.

articular cartilage (Fig. 4C). The MR findings were then compared

ings to see and were fragments if there displaced partially were differences attached, lesions. attached

with the arthroscopic

between the stable, of arthroscopic cartilage lesions (Fig. with

On the basis with intact attached

[2] was used to determine

condyle.

the location of the lesions on the femoral

the lesions
1), three

were divided with

(Fig.

into four groups. cartilage

4). Partially

One lesion was stable

(Fig.

On the anteroposterior films, one lesion was in the femoral notch; 1 5 involved the notch and a portion of the central surface of the condyle; and five were localized to the central, nonmeniscal portion of the condyle. On the lateral films, 1 9 lesions were directly distal to a line projected down the posterior cortex of the femur, that is, on the weight-bearing surface while the patient was standing. Two
lesions flexion, were such posterior as when to this climbing line in a position of loading with knee stairs.

2), 1 i were intact with carti-

partially
placed was

attached those
the

fractures with
at in

(Fig. 3), and six had disprobing,

lesions and of

lage were
intact fractures,

that were
fractures was

ballotable
In partially seen found

but the cartilage

cartilage lesions the the

at arthroscopy.

attached arthroscopy

were often

were

ballotable
tissue

or could

be elevated

by probing.
the bases

Loose

fibrous
partially

granulation

All examinations were performed on a 1 .5-T General Electric MR unit (Milwaukee, WI) with a dedicated transmit-and-receive
tremity mm. two coil. The excitations

Signa exx 128; Ti

-

attached fragments that did not show

no cartilage with probing. fracture,

that could be elevated. A stable lesion bony union on plain films but at surgery
no cartilage softening, and was not

was one showed

ballotable

Slice thickness
of view were was used.

was 3 mm with an interslice

1 2-1 6 cm Spin-echo with a matrix imaging was

gap of 1.5
used.

field

of 256

weighted sequences, sequences, 2000/20,

T2-weighted, The first five respectively. patients

600/20 (TR/TE), were used, as were two-echo 90, which were proton-density-weighted and
had coronal and sagittal Ti-weighted images

Results
Intensity of signal within the lesions did not differ between the stable or two categories of partially attached lesions. On Ti -weighted images, the lesions had inhomogeneous signal ranging from grades 0 to -2 (Fig. 36). Fragment appearance for the stable or partially attached lesions on T2-weighted images was more variable. Four lesions were homogeneous with -2 absent signal. The remainder had an inhomogeneous pattern with focal areas ranging from +2 high signal to -2 absent signal with marked variation between the lesions (Figs. iB, 2C, and 3C). Lesions with displaced fragments had only bone defects, so no partially attached fragment could be assessed for central signal or high signal at the adjacent bone interface. All six patients with displaced fragments had bony defects filled with joint fluid that was of intermediate signal on Ti -weighted images and of high signal on T2-weighted images (Figs. 4B and 4C). All patients with partially attached fragments had a highsignal line at the interface between the lesion and the femur

followed by a T2-weighted sequence in the plane in which the lesion was best visualized. The last 16 patients had both coronal and sagittal
T2-weighted images and Ti-weighted images only in the plane in

which

visualization

of the lesion was optimal.

This protocol

was used

because the T2-weighted images best defined the joint surface and the reactive interface between the fragment and the femur.
Each MR scan was evaluated by one observer without knowledge of the arthroscopic findings for fragment signal intensity, articular
cartilage defects, and was evidence of instability as listed in the following

sections.
weighted

Signal
images

intensity femoral markedly

within medullary decreased

the fragment

on both
the fragment

Ti-

and T2with

graded

by comparing

signal

the surrounding in which -2 =

moderately marrow, increased or absence i
=

space by using a grading system or absent signal, -i = mildly or

intensity fluid. equal to that of femoral
=

decreased
signal equal

signal, 0
to that

moderately images
signal

increased
of joint

signal,

and

+2

markedly

The T2-weighted
of high

were also evaluated

at the fragment-femur

for (1) the presence

interface (Figs. 1 B,

2B, 2C, and 3C), (2) disruption of the subchondral bone plate (Figs. 2B, 2C, 3C, 4B, and 4C), (3) adjacent focal cystic areas (Fig. 3D), (4)

Fig. 1.-Stable osteochondritis dissecans diagnosed at surgery. A, Preoperative plain film shows large medial condyle lesion. B, Sagittal MR image, 2000/90, shows highsignal line (arrowheads) above fragment; this resulted in false-positive MR diagnosis of unstable fragment. Two years after drilling, lesion appeared healed on plain film and MR (not shown).

AJR:155, September

1990

MR OF OSTEOCHONDRITIS

DISSECANS

OF THE

KNEE

551

Fig. 2.-Partially attached osteochondritis dissecans with intact cartilage. A, Plain film shows medial condyle defect with central fragment. B and C, Sagittal (B) and coronal (C) MR images, 2000/90. Bone-fragment Signal within fragment is nonspecific, being mostly -2 absent signal but with (arrows), although cartilage was intact at arthroscopy.

interface shows high signal (arrowheads), linear -1 and +2 signal within it. Subchondral

indicating unstable fragment. bone plate appears disrupted

3.-Partially attached osteochondritis with cartilage fracture. A, Plain film shows lateral condyle defect. B-D, High signal on MR Images at fragmentbone Interface and cyst indicate fragment is unstable. MR reveals that large plain film defect is filled with fibrous tissue or fibrocartilage. On coronal MR image, 600/20 (B), fragment has central homogeneous -1 signaL On coronal MR image, 2000/90 (C), fragment has nonspecific mixed -Ilow (arrows) and -2 absent signal. On sagittal MR image, 2000/90 (0), subchondral plate is disrupted (arrow) with adjacent cyst (arrowhead).

Fig.

dissecans

552

DE SMET

ET AL.

AJR:155,

September

1990

Fig. 4.-Osteochondritis dissecans with displaced fragment. A, Plain film shows bony defect in medial femoral condyle (arrow). B-D, MR reveals that small plain film defect is actually a large surface cavity. On coronal MR image, 600/20 (B), defect is filled with intermediate-signal fluid. On sagittal MR image, 2000/ 90 (C), defect is filled with high-signal fluid. On sagittal MR image, 2000/90 (D), displaced fragment (arrow) is embedded in fat pad with overlying fluid.

:k
. . . ., :- . .

-.

:..4.

..

P,
mine lesion stability has been reported only for radionuclide imaging [4, 9, 1 0] and MR imaging [1 0]. With radionuclide scanning, findings between stable and unstable fragments overlap. Radionuclide scanning also provides no anatomic information on articular surface deformity. We expected that MR imaging would be ideal for evaluation of OCD because of its noninvasive nature, absence of ionizing radiation, excellent anatomic detail, and soft-tissue contrast allowing cartilage visualization. We had hoped that, in addition to detecting surface defects, MR could accurately distinguish between stable and unstable fragments. Our observation that one patient with a stable lesion had MR findings similar to the unstable lesions was disappointing but consistent with the literature. Mesgarzadeh et al. [1 0] found that MR imaging

(Figs. 2B, 2C, and 3C). Even the stable lesion had a definite high-signal line at the fragment-femur interface (Fig. 1 B). Nine of the 20 patients with partially attached or displaced fragments had round, cystlike areas in the femur adjacent to the OCD (Fig. 3D). These cystic areas were of intermediate signal intensity on Ti -weighted images and of very high signal intensity on T2-weighted images. The stable lesion did not have an adjacent cyst. The subchondral bone plate was disrupted on both Ti and
-

T2-weighted

images

in all 2i patients

(Figs. 2B, 3D, 4B, and

4C). Loose bodies were found on MR imaging in all six patients with displaced fragments. These loose bodies were visualized as low-signal filling defects completely or partially surrounded by high-signal joint fluid on T2-weighted images (Fig. 4D). If there were multiple loose bodies, it was difficult to find all of the loose bodies on the MR scan. Correlation with the plain films was needed to confidently identify several fragments embedded in the synovium. Two loose bodies were cartilaginous and were seen on MR images but not seen on the plain films.

was 92% sensitive

and 90% specific in separating

nine stable

from i 2 loose lesions. They as unstable using disruption

misdiagnosed one healed lesion of the cartilage surface and the

presence of a high-signal line as criteria for lesion instability. The most dramatic role for MR imaging is in visualizing the
presence cation of useful in about the Because bility, we and size of articular cartilage defects and identificartilaginous loose bodies. This information was preoperative planning and in advising the patient poorer prognosis with a large cartilage defect. of our one false-positive diagnosis of lesion instaare using the presence of a high-signal line beneath

Discussion
OCD

Although many imaging methods have been used to study fragments, imaging of large series of patients to deter-

AJR:155,

September

1990

MR OF OSTEOCHONDRITIS

DISSECANS

OF THE

KNEE

553

the lesion as a suggestive but not absolute indicator of an unstable lesion. Although our one false-positive case did not meet our arthroscopic criteria for an unstable lesion, we think but cannot prove that the lesion was still unstable. Although the lesion appeared stable at surgery, it had an overlying dimple and the arthroscopist drilled the lesion to induce healing. The lesion began healing immediately and was completely healed in 2 years. We think that although the lesion was stable to probing, microscopic fragment motion occurred under greater forces generated by weight bearing. This patients knee had been painful with activity for 2 years before we began caring for him. The patient was followed up for i

lesions

will be needed

to prove

that

cystic

areas

are never

seen in stable lesions.

In summary, MR imaging of OCD provides mation by detecting cartilaginous loose bodies unique inforand articular

surface defects. High signal at the fragment interlace was seen in all unstable lesions but also in one questionably stable lesion. Cystic lesions beneath the fragments were seen only
in unstable lesions. On the basis of these findings, we believe that MR imaging is useful in evaluating lesion stability and articular cartilage defects in patients with femoral OCD.

year with a conservative

program

including

3 months

of knee

REFERENCES
1 . Cahill B. Treatment of juvenile osteochondritis dissecans and osteochondritis dissecans of the knee. Clin Sports Med 1985;4(2):367-384 2. Hughston JC, Hergenroeder PT, Courtenay BG. Osteochondritis dissecans of the femoral condyles. J Bone Joint Surg [Am] 1984;66-A(9): 1340-1348 3. Steiner ME, Grana WA. The young athletes knee: recent advances. Gun Sports Med 1988;7(3):527-546 4. Cahill BR, Phillips MA, Navarro A. The results of conservative management ofjuvenile osteochondritis dissecans using joint scintigraphy: a prospective study. Am J Sports Med 1989;17(5):601-606 5. Denoncourt PM, Patel D, Dimakopoulos P. Arthroscopy update #1 . Treatment of osteochondritis dissecans of the knee by arthroscoplc curettage. follow-up study. Orthop Rev 1986;15(10):652-657 6. Ewing JW, Voto SJ. Arthroscoplc surgical management of osteochondritis dissecans of the knee. Arthroscopy 1988;4(1):37-40 7. Thomson NL. Osteochondritis dissecans and osteochondral fragments managed by Herbert compression screw fixation. C!in Orthop 1987;

bracing and the use of crutches. The patient had knee pain with ambulation throughout the conservative management period but ran without knee pain within 7 months after surgery.

Mesgarzadeh

et al. [1 0] found that eight of their nine stable

lesions had no high signal at the fragment interface. We also noted in a previous study [i 5] that the absence of a highsignal line in talar OCD was a reliable sign of a healed OCD. On the basis of these two studies, we have been assuming that patients with no high signal at the interface have healed lesions. Large series of patients with MR examinations and treating patients conservatively until bony union occurs will

be needed
reliable

to see if absence
lesion.

of a high-signal

interface

is a

sign of a stable

224:71-78

The reason patients

with stable or partially attached

lesions

and intact cartilage had apparently disrupted cartilage on MR imaging in the presence of intact cartilage at arthroscopy is unknown. Mesgarzadeh et al. [1 0] also noted apparent carti-

lage discontinuity

on MR imaging

in i i of their unstable

lesions, but only five had disrupted cartilage at surgery. Presumably the cartilage suffers an initial injury with either reparative tissue or old hemorrhage persisting at that site. We cannot explain why this tissue in the articular cartilage and subchondral bone plate region would have persistent signal characteristics resembling fluid or granulation tissue. The high-signal interface at the junction between the fragment and the femur is not as puzzling. Histologic studies after MR imaging of avascular necrosis of the femoral head have found that granulation tissue at the margins of the necrotic

8. Litchman HM, McCullough AW, Gandsman EJ, Schatz SL. Computerized blood flow analysis for decision making in the treatment of osteochondritis dissecans. J Pediatr Orthop 1988;8(2):208-212 9. McCullough RW, Gandsman EJ, Litchman H, Schatz SL, Deutsch SD. Computerized blood-flow analysis in osteochondritis dissecans. C!in Nuc! Med 1986;11(7):511-513 10. Mesgarzadeh M, Sapega AA, Bonakdarpour A, et al. Osteochondritis dissecans: analysis of mechanical stability with radiography, scintigraphy, and MA imaging. Radiology 1987;165:755-780 1 1 . Paille P. Ouesnel C, Baunin C, Railhac JJ. Computed arthrography: its role in the screening of joint diseases in pediatric radiology. Pediatr Radio!

1988;18:386-390
12. Hartzman 5, Reicher Gold RH. MR imaging MA, Bassett of the knee. LW, Duckwiler Part II. Chronic GA, Mandelbaum B, disorders. Radiology

1987;162:553-557
13. Lutten C, Thomas W. NMR criteria of osteochondritis dissecans of the knee joint, the trochlea of the talus, patella and capitulum of the humerus: preoperative situation and postoperative follow-up. E!ectromedica

bone can present

as a high-signal

line [i 6, 17]. In agreement

1988;56:118-1 25
14. Sims RE, Genant HK. Magnetic resonance imaging of joint disease. Radio! din North Am 1986;24(2): 179-188 15. De Smet AA, Fisher DR, Bumstein Ml, Graf BK, Lange RH. Value of MR imaging in staging osteochondral lesions of the talus (osteochondritis dissecans): results in 14 patients. AiR 1990;154:555-558 16. Bassett LW, Mirra JM, Cracchiolo A, Gold RH. lschemic necrosis of the femoral head. C!in Orthop 1987;223:181-187 17. Mitchell DG, Steinberg ME, Dalinka MK, Rao VM, Fallen M, Kressel HY. Magnetic resonance imaging of the hip: alterations within the osteonecrotic, viable, and reactive zones. C!in Orthop 1989;244:60-77

with these femoral head studies, we noted at surgery that partially attached fragments often had fibrous granulation tissue in the base of the defect. Focal cystic areas beneath the fragments were seen only

with unstable fragments. We previously showed that in talar OCD such cysts were seen only with unstable fragments
[i 5]. These
lation tissue

cysts were found to be cavities filled with granu[15]. Further study of patients with proven stable