JBJS | Magnetic Resonance Imaging of Articular Cartilage in the Knee. An Evaluation with Use of Fast-Spin-Echo Imaging*

The Journal of Bone & Joint Surgery, Volume 80, Issue 9

Articles | September 01, 1998

Magnetic Resonance Imaging of Articular Cartilage in the Knee. An Evaluation with Use of Fast-Spin-Echo Imaging*

HOLLIS G. POTTER, M.D.†; JAMES M. LINKLATER, M.D.†; ANSWORTH A. ALLEN, M.D.†; JO A. HANNAFIN, M.D., PH.D.†; STEVEN B. HAAS, M.D., M.P.H.†, NEW YORK, N.Y.

View Disclosures and Other Information

Investigation performed at the Departments of Radiology and Orthopaedics, The Hospital for Special Surgery, New York City

The Journal of Bone & Joint Surgery. 1998; 80:1276-1284

5 Recommendations (Recommend) | 3 Comments | Saved by 3 Users Save Case

text A A A

Abstract

The purpose of this study was to demonstrate that specialized magnetic resonance imaging provides an accurate assessment of lesions of the articular cartilage of the knee. Arthroscopy was used as the comparative standard.

Eighty-eight patients who had an average age of thirty-eight years were evaluated with magnetic resonance imaging and subsequent arthroscopy because of a suspected meniscal or ligamentous injury. The magnetic resonance imaging was performed with a specialized sequence in the sagittal, coronal, and axial planes. Seven articular surfaces (the patellar facets, the trochlea, the femoral condyles, and the tibial plateaus) were graded prospectively on the magnetic resonance images by two independent readers with use of the 5-point classification system of Outerbridge, which was also used at arthroscopy.

Six hundred and sixteen articular surfaces were assessed, and 248 lesions were identified at arthroscopy. Eighty-two surfaces had chondral softening; seventy-five, mild ulceration; fifty-three, deep ulceration, fibrillation, or a flap without exposure of subchondral bone; and thirty-eight, full-thickness wear. To simplify the statistical analysis, grades 0 and 1 were regarded as disease-negative status and grades 2, 3, and 4 were regarded as disease-positive status. When the grades that had been assigned by reader 1 were used for the analysis, magnetic resonance imaging had a sensitivity of 87 per cent (144 of 166), a specificity of 94 per cent (424 of 450), an accuracy of 92 per cent (568 of 616), a positive predictive value of 85 per cent (144 of 170), and a negative predictive value of 95 per cent (424 of 446) for the detection of a chondral lesion. Interobserver variability was minimum, as indicated by a weighted kappa statistic of 0.93 (almost perfect agreement).

With use of this readily available modified magnetic resonance imaging sequence, it is possible to assess all articular surfaces of the knee accurately and thereby identify lesions that are amenable to arthroscopic treatment.

Figures in this Article

Introduction

Introduction | Materials and Methods | Results | Discussion | References

The articular or hyaline cartilage of the knee covers the articulating surfaces of the patella, femur, and tibia. The thickness of the cartilage is variable. The articular cartilage transmits and distributes load, thereby minimizing stress on the underlying subchondral bone, and provides a smooth articulating surface as well as lubrication, thereby reducing friction in the joint. Most chondral injuries are secondary to trauma or degeneration of the joint. Although chondral injuries usually occur in association with other intra-articular abnormalities, such as ligamentous and meniscal injuries, isolated chondral lesions can occur, with a reported prevalence of approximately 4 per cent (twelve of 312)²⁸. Patients are often seen with non-specific pain in the knee, low-grade effusion, and symptoms that mimic those of a meniscal tear, such as pain along the joint line and locking^11,20. Traditional techniques for the operative treatment of acute and chronic chondral lesions include débridement of the chondral flap; perforation of the subchondral bone by drilling¹⁵, abrasion^3,10,12, or microfracture^15,24; and insertion of a perichondral graft^6,17,18,25. More recently, the implantation of autologous cartilage² as well as mosaicplasty⁹ have stimulated new interest in the treatment of chondral injuries. This interest has resulted in an increasing demand for an accurate, reproducible, and, ideally, non-invasive method for the assessment of lesions of articular cartilage, both for the initial diagnosis and for subsequent monitoring after operative treatment. Magnetic resonance imaging, by virtue of its superior soft-tissue contrast, lack of ionizing radiation, and multiplanar capabilities, is superior to more conventional techniques for the evaluation of articular cartilage^14,22. The particular magnetic resonance imaging sequence most suited for the detection of chondral abnormalities is controversial. There has been some consensus in the radiological literature that a fat-suppressed volumetric (three-dimensional) gradient-echo sequence may be the best technique⁴. However, this sequence is limited by a long acquisition time^4,5 and typically relies on subsequent reconstruction of images derived from a single plane. The imaging time is prolonged further by the need for additional sequences, as the fat-suppressed gradient-echo sequence is not optimum for the assessment of ligaments, menisci, or subchondral bone. Previous authors have suggested that multiplanar reconstructions derived from the initial sagittal volumetric sequence do not offer images of sufficient quality to allow for the accurate diagnosis of chondral lesions in areas best assessed in the axial or coronal plane²³. Our hypothesis was that a specialized proton-density-weighted, high-resolution, fast-spin-echo sequence, which was previously shown to provide an accurate assessment of other structures in the knee²¹, would provide a suitable assessment of articular cartilage in the knee without substantially prolonging the total imaging time. The purpose of the present investigation was to assess the accuracy of magnetic resonance imaging in the detection, grading, and localization of chondral lesions in the knee and to compare the findings with those of direct arthroscopic inspection.

*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.

†Departments of Radiology (H. G. P. and J. M. L.) and Orthopaedics (A. A. A., J. A. H., and S. B. H.), The Hospital for Special Surgery, 535 East 70th Street, New York, N.Y. 10021.

*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.

†Departments of Radiology (H. G. P. and J. M. L.) and Orthopaedics (A. A. A., J. A. H., and S. B. H.), The Hospital for Special Surgery, 535 East 70th Street, New York, N.Y. 10021.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG1-A:: Figs. 1-A and 1-B: A thirty-year-old man who had grade-0 (intact) articular cartilage. Fig. 1-A: Axial fast-spin-echo magnetic resonance image demonstrating uniform thickness of the articular cartilage over the patellar facets. The sharp interface between the articular surface and the joint fluid (arrow) indicates an absence of fibrillation of the surface.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG1-B:: Fig. 1-B Coronal fast-spin-echo magnetic resonance image demonstrating uniform thickness of the cartilage and a homogeneous appearance over the condyles and plateaus.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG2:: Fig. 2 Axial fast-spin-echo magnetic resonance image through the knee of a thirty-year-old woman who had an arthroscopically confirmed grade-2 lesion of the lateral patellar facet. There is a focal blister in the articular surface as well as a lack of a sharp interface between the articular surface and the joint fluid (arrow). Hyperintensity of the signal in the superficial layers of cartilage suggested intrasubstance softening, or a grade-1 lesion.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG3-A:: Figs. 3-A, 3-B, and 3-C: A thirty-four-year-old man who had symptoms suggestive of a meniscal tear, including intermittent locking of the knee. Fig. 3-A: Sagittal fast-spin-echo magnetic resonance image demonstrating a discrete full-thickness chondral defect (white arrow) over the posterior margin of the lateral femoral condyle, with an adjacent chondral flap (black arrow).

View Large | Download Slide(.ppt)
Add to My JBJS

FIG3-B:: Fig. 3-B Axial fast-spin-echo magnetic resonance image demonstrating a loose body in the medial aspect of the suprapatellar pouch (curved arrow). There is severe fibrillation (straight arrow) of the surface over the lateral patellar facet, which was interpreted as grade 3 and focal grade 4. There is also a relatively intact rind of cartilage over the trochlea, with mild intrasubstance softening over the medial patellar facet. The findings on the magnetic resonance images were confirmed at arthroscopy.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG3-C:: Fig. 3-C Arthroscopic image of a chondral flap (arrow) over the lateral femoral condyle. (Reprinted, with permission, from: Linklater, J., and Potter, H. G.: Imaging of chondral defects. Op. Tech. Orthop., 7: 281, 1997.)

View Large | Download Slide(.ppt)
Add to My JBJS

FIG4-A:: Figs. 4-A, 4-B, and 4-C: A thirty-three-year-old man who had an arthroscopically confirmed grade-4 lesion over the medial femoral condyle. Fig. 4-A: Sagittal fast-spin-echo magnetic resonance image demonstrating the discrete chondral defect (arrow).

View Large | Download Slide(.ppt)
Add to My JBJS

FIG4-B:: Fig. 4-B Coronal fast-spin-echo magnetic resonance image confirming the full-thickness defect (arrow).

View Large | Download Slide(.ppt)
Add to My JBJS

FIG4-C:: Fig. 4-C Arthroscopic photograph confirming a full-thickness chondral defect over the posterior aspect of the medial femoral condyle. The defect measured eight by twenty millimeters.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG5:: Fig. 5 Sagittal fast-spin-echo magnetic resonance image through the knee of a fifty-two-year-old man, demonstrating a smoothly marginated but full-thickness chondral defect over the posterior margin of the lateral tibial plateau (arrow) beneath the posterior horn of the lateral meniscus. This lesion was not confirmed at arthroscopy.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG6:: Fig. 6 Sagittal fast-spin-echo magnetic resonance image through the knee of a sixty-one-year-old woman, demonstrating a discrete osteochondral injury over the lateral femoral condyle (arrow) with a full-thickness chondral defect. This was not seen at arthroscopy, possibly because of its eccentric location.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG7:: Fig. 7 Coronal fast-spin-echo magnetic resonance image through the knee of a thirty-seven-year-old man, demonstrating a small full-thickness chondral defect over the medial femoral condyle (arrow). The lesion was confirmed at arthroscopy. With use of appropriate pulse sequences, the lack of effusion in the joint does not limit prospective evaluation of important chondral lesions.

View Larger

TABLE I: ARTHROSCOPIC GRADING OF THE SIX HUNDRED AND SIXTEEN ARTICULAR SURFACES*

Surface	Grade 0	Grade 1	Grade 2	Grade 3	Grade 4
Medial	47	10	18	9	4
patellar
facet
Lateral	49	14	13	8	4
patellar
facet
Trochlea	50	14	7	10	7
Medial	34	8	13	19	14
femoral
condyle
Lateral	71	9	4	2	2
femoral
condyle
Medial	56	12	13	2	5
tibial
plateau
Lateral	61	15	7	3	2
tibial
plateau
Total	368	82	75	53	38

*The values are given as the number of surfaces.

Add to My JBJS

TABLE I: ARTHROSCOPIC GRADING OF THE SIX HUNDRED AND SIXTEEN ARTICULAR SURFACES*

Surface	Grade 0	Grade 1	Grade 2	Grade 3	Grade 4
Medial	47	10	18	9	4
patellar
facet
Lateral	49	14	13	8	4
patellar
facet
Trochlea	50	14	7	10	7
Medial	34	8	13	19	14
femoral
condyle
Lateral	71	9	4	2	2
femoral
condyle
Medial	56	12	13	2	5
tibial
plateau
Lateral	61	15	7	3	2
tibial
plateau
Total	368	82	75	53	38

*The values are given as the number of surfaces.

View Larger

TABLE II: AGREEMENT BETWEEN THE FINDINGS OF THE INDEPENDENT READERS OF THE MAGNETIC RESONANCE IMAGES AND THE FINDINGS AT ARTHROSCOPY

Surface	Exact Agreement (per cent)		Difference of One Grade (per cent)		Difference of Two Grades (per cent)
Reader 1	Reader 2	Reader 1	Reader 2	Reader 1	Reader 2
Medial patellar facet	66	70	30	29	4	1
Lateral patellar facet	71	76	27	20	2	3
Trochlea*	75	74	25	25	0	2
Medial femoral condyle	67	65	28	32	5	2
Lateral femoral condyle	80	78	19	21	1	1
Medial tibial plateau†	70	79	29	20	1	1
Lateral tibial plateau	77	74	18	22	5	4
All articular surfaces	72	73	25	24	3	2

*The arthroscopic assessment differed by three grades from the magnetic resonance assessment by reader 2 for one surface.†The arthroscopic assessment differed by three grades from the magnetic resonance assessment by reader 1 and reader 2 for one surface each.

Add to My JBJS

TABLE II: AGREEMENT BETWEEN THE FINDINGS OF THE INDEPENDENT READERS OF THE MAGNETIC RESONANCE IMAGES AND THE FINDINGS AT ARTHROSCOPY

Surface	Exact Agreement (per cent)		Difference of One Grade (per cent)		Difference of Two Grades (per cent)
Reader 1	Reader 2	Reader 1	Reader 2	Reader 1	Reader 2
Medial patellar facet	66	70	30	29	4	1
Lateral patellar facet	71	76	27	20	2	3
Trochlea*	75	74	25	25	0	2
Medial femoral condyle	67	65	28	32	5	2
Lateral femoral condyle	80	78	19	21	1	1
Medial tibial plateau†	70	79	29	20	1	1
Lateral tibial plateau	77	74	18	22	5	4
All articular surfaces	72	73	25	24	3	2

View Larger

TABLE III: WEIGHTED KAPPA STATISTICS

Comparison	Weighted Kappa Statistic
Reader 1 and arthroscopy	0.88
Reader 2 and arthroscopy	0.89
Reader 1 and reader 2	0.93

Add to My JBJS

TABLE III: WEIGHTED KAPPA STATISTICS

Comparison	Weighted Kappa Statistic
Reader 1 and arthroscopy	0.88
Reader 2 and arthroscopy	0.89
Reader 1 and reader 2	0.93

View Larger

TABLE IV: RELATIVE SENSITIVITY, SPECIFICITY, AND ACCURACY FOR EACH ARTICULAR SURFACE FOR THE FINDINGS OF READER 1 COMPARED WITH THOSE AT ARTHROSCOPY*

Surface	Findings on Magnetic Resonance Imaging				Disease- Positive Status at Arthroscopy†	Sensitivity	Specificity
True- Positive†	True- Negative†	False- Positive†	False- Negative†
						(per cent)	(per cent)
Medial patellar facet	30	51	6	1	31	97	89
Lateral patellar facet	23	59	4	2	25	92	94
Trochlea	22	61	3	2	24	92	95
Medial femoral condyle	38	40	2	8	46	83	95
Lateral femoral condyle	8	77	3	0	8	100	96
Medial tibial plateau	16	65	3	4	20	80	96
Lateral tibial plateau	7	71	5	5	12	58	93
Total	144	424	26	22	166	87	94

*Grades 0 and 1 were considered disease-negative status, and grades 2, 3, and 4 were considered disease-positive status. †The values indicate the number of surfaces associated with each finding.

Add to My JBJS

TABLE IV: RELATIVE SENSITIVITY, SPECIFICITY, AND ACCURACY FOR EACH ARTICULAR SURFACE FOR THE FINDINGS OF READER 1 COMPARED WITH THOSE AT ARTHROSCOPY*

Surface	Findings on Magnetic Resonance Imaging				Disease- Positive Status at Arthroscopy†	Sensitivity	Specificity
True- Positive†	True- Negative†	False- Positive†	False- Negative†
						(per cent)	(per cent)
Medial patellar facet	30	51	6	1	31	97	89
Lateral patellar facet	23	59	4	2	25	92	94
Trochlea	22	61	3	2	24	92	95
Medial femoral condyle	38	40	2	8	46	83	95
Lateral femoral condyle	8	77	3	0	8	100	96
Medial tibial plateau	16	65	3	4	20	80	96
Lateral tibial plateau	7	71	5	5	12	58	93
Total	144	424	26	22	166	87	94

*Grades 0 and 1 were considered disease-negative status, and grades 2, 3, and 4 were considered disease-positive status. †The values indicate the number of surfaces associated with each finding.

Materials and Methods

Introduction | Materials and Methods | Results | Discussion | References

Patient Selection

All patients who had had magnetic resonance imaging and subsequent arthroscopy of the knee at our institution from January 1996 to August 1997 and for whom the operating surgeon had completed a chondral injury data sheet were eligible for the present study. A total of ninety patients were enrolled. Two patients were excluded from the study because of severe degradation of the images, caused by motion of the patient, that precluded confirmation of chondral lesions in two planes. Another two patients were included despite the presence of metallic hardware in the knee in association with reconstruction of the anterior cruciate ligament and the placement of an interference screw. The average age of the eighty-eight patients who met the criteria for inclusion was thirty-eight years (range, twenty-three to eighty-two years). There were forty-eight men and forty women. The average interval between magnetic resonance imaging and arthroscopy was twenty-seven days (range, four to 377 days). Seventy-five patients (85 per cent) had arthroscopy within two months after magnetic resonance imaging; a review of the charts revealed that no injuries had been documented during this interval. The study was approved by our Institutional Review Board.

Magnetic Resonance Imaging

Magnetic resonance imaging of the knee was performed with a 1.5-tesla magnet (Signa, Horizon; General Electric Medical Systems, Milwaukee, Wisconsin) and a conventional send-receive extremity coil. Three sequences were used to assess the articular cartilage in the sagittal, axial, and coronal planes. Each sequence consisted of fast-spin-echo images that were performed with a long repetition time (3500 to 5000 milliseconds), a short effective-echo time (thirty to thirty-four milliseconds), an echo train length of eight to ten, a bandwidth of 20.8 to 31.3 kilohertz, two excitations, and a slice thickness of 3.5 to 4.0 millimeters with no interslice gap. Sagittal images were performed with a 512-by-256 matrix and a field of view of fifteen to sixteen centimeters (scan time, 4.5 minutes), axial images were performed with a 512-by-256 matrix and a field of view of fourteen to fifteen centimeters (scan time, 4.0 minutes), and coronal images were performed with a 256-by-256 matrix and a field of view of twelve to thirteen centimeters (scan time, 4.0 minutes). As part of our routine assessment of the knee, we performed an additional sagittal fast-spin-echo sequence (repetition time, 4000 milliseconds; effective-echo time, thirty-four milliseconds) with frequency-selective fat suppression (Chemsat; General Electric Medical Systems) and an individual scan time of 3.5 minutes. The total scanning time was approximately twenty minutes. The magnetic resonance images were prospectively analyzed by two blinded, independent readers. One of the readers (H. G. P.) was a musculoskeletal magnetic resonance imaging radiologist, and the other (J. M. L.) was a musculoskeletal magnetic resonance imaging fellow. Seven articular surfaces were assessed: the medial and lateral patellar facets, the trochlea, the medial and lateral femoral condyles, and the medial and lateral tibial plateaus. The articular cartilage was graded on the magnetic resonance images and at arthroscopy with a modification of the classification system of Outerbridge¹⁹. Grade 0 indicated intact cartilage; grade 1, chondral softening or blistering with an intact surface; grade 2, shallow superficial ulceration, fibrillation, or fissuring involving less than 50 per cent of the depth of the articular surface; grade 3, deep ulceration, fibrillation, fissuring, or a chondral flap involving 50 per cent or more of the depth of the articular cartilage without exposure of subchondral bone; and grade 4, full-thickness chondral wear with exposure of subchondral bone. All chondral lesions were confirmed in at least two separate planes. On magnetic resonance images, the intermediate signal intensity of the articular cartilage contrasted with the lower signal intensity of the subchondral plate and the higher signal intensity of the joint fluid. The cartilage was considered to be intact if the band of intermediate signal intensity had a uniform thickness (Figs. 1-A and 1-B).

Arthroscopic Technique

Arthroscopy was performed by one of three surgeons (S. B. H., A. A. A., or J. A. H.). The indications for arthroscopy included a meniscal tear, disruption of the anterior cruciate ligament, and chondral injury. Arthroscopy was performed with use of standard anterolateral and anteromedial portals. A superolateral or superomedial portal was used for inflow. The typical examination included a thorough inspection of the undersurface of the patella from both the anterolateral and the anteromedial portal. The surface of the cartilage was probed from the contralateral portal. The femoral trochlea was viewed from the anterolateral portal. The knee was flexed and extended to assess tracking of the patella and to document the presence of so-called kissing lesions involving the femoral trochlea and the patella. The medial femoral condyle and the medial tibial plateau were viewed through the anterolateral portal. The posterior aspect of the medial femoral condyle was viewed through the intercondylar notch. The lateral compartment was viewed with the lower limb in the figure-four position, and the anteromedial portal was used as necessary to visualize the posterior aspect of the lateral femoral condyle. The articular surfaces were probed, and arthroscopic photographs of the involved joint surfaces were made. The arthroscopic assessment of the articular cartilage was recorded at the time of arthroscopy by the operating surgeon on a standardized data-entry form. The magnetic resonance images and related reports were available to the surgeon at the time of arthroscopy. However, the reports did not include the comprehensive grading sheet that had been used to evaluate the individual articular surfaces, and the images primarily were used to assess the ligaments and menisci of the knee.

Statistical Analysis

Weighted kappa statistics were calculated to assess the degree of interobserver agreement as well as the degree of agreement between the findings on magnetic resonance imaging and those on arthroscopy. The weightings were calculated with the formula 1 - (|i - j|/[k - 1]), where i and j index the rows and columns of the ratings assigned by the two readers and k is the maximum number of possible ratings (Release 4, Statistical Analysis Package; Stata, College Station, Texas). If the grades agreed exactly, the weighting was 1.00; if they differed by one, the weighting was 0.75; if they differed by two, the weighting was 0.50; if they differed by three, the weighting was 0.25; and if they differed by four, the weighting was 0.00. A weighted kappa value of less than 0.00 indicates poor agreement, a value of 0.00 to 0.20 indicates slight agreement, a value of 0.21 to 0.40 indicates fair agreement, a value of 0.41 to 0.60 indicates moderate agreement, a value of 0.61 to 0.80 indicates substantial agreement, a value of 0.81 to less than 1.00 indicates almost perfect agreement, and a value of 1.00 indicates perfect agreement¹³. To facilitate comparison with previous reports, the sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of magnetic resonance imaging were calculated for each reader and each articular surface by collapsing all grades into disease-positive status (grades 2, 3, and 4) and disease-negative status (grades 0 and 1)⁵. Grade 1 was considered disease-negative status because of its relatively limited clinical importance and a suspected higher subjectivity of establishing its presence at arthroscopy.

Results

Introduction | Materials and Methods | Results | Discussion | References

A total of eighty-eight patients met the criteria for inclusion in the study, and a total of 616 articular surfaces were assessed (Table I). At arthroscopy, 368 articular surfaces were classified as normal (grade 0); eighty-two, as grade 1; seventy-five, as grade 2 (Fig. 2); fifty-three, as grade 3 (Figs. 3-A, 3-B, and 3-C); and thirty-eight, as grade 4. Grade-4 lesions were most commonly seen over the medial femoral condyle (Figs. 4-A, 4-B and 4-C) and the trochlea but were uncommon in the lateral femorotibial compartment. Grade-3 lesions were also most commonly seen over the medial femoral condyle and the trochlea. When the classifications that were assigned at arthroscopy were compared with those that had been assigned by each independent reader on the basis of magnetic resonance imaging, there was exact agreement for 444 articular surfaces (72 per cent) as assessed by reader 1 and for 453 articular surfaces (74 per cent) as assessed by reader 2, a difference of one grade for 155 surfaces (25 per cent) as assessed by reader 1 and for 147 surfaces (24 per cent) as assessed by reader 2, a difference of two grades for sixteen surfaces (3 per cent) as assessed by reader 1 and for fourteen surfaces (2 per cent) as assessed by reader 2, and a difference of three grades for one surface (0.2 per cent) as assessed by reader 1 and for two surfaces (0.3 per cent) as assessed by reader 2 (Table II). When discrepancies between grade 1 (disease-negative status) and grade 2 (disease-positive status) were analyzed, with use of the arthroscopic classification as the standard, it was found that reader 1 had classified nineteen grade-1 surfaces as grade 2 and thirteen grade-2 surfaces as grade 1. Similarly, reader 2 had classified fifteen grade-1 surfaces as grade 2 and thirteen grade-2 surfaces as grade 1. When the seven different articular surfaces were analyzed separately, the rate of exact agreement between arthroscopy and magnetic resonance imaging ranged from 66 to 80 per cent for reader 1 and from 65 to 79 per cent for reader 2 (Table II). The weighted kappa statistic for interobserver agreement between readers 1 and 2 was 0.93, indicating almost perfect agreement¹³. The weighted kappa statistic for agreement between magnetic resonance imaging and arthroscopy was 0.88 for reader 1 and 0.89 for reader 2, also indicating almost perfect agreement (Table III). The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of magnetic resonance imaging were calculated for each reader by collapsing all grades into disease-positive status (grades 2, 3, and 4) and disease-negative status (grades 0 and 1). A review of the grades that had been assigned by reader 1 revealed 144 true-positive findings, 424 true-negative findings, twenty-six false-positive findings, and twenty-two false-negative findings; these data yielded a sensitivity of 87 per cent (144 of 166), a specificity of 94 per cent (424 of 450), a positive predictive value of 85 per cent (144 of 170), a negative predictive value of 95 per cent (424 of 446), and an accuracy of 92 per cent (568 of 616). When the grades that had been assigned by reader 2 were used for the analysis, the sensitivity was 87 per cent (144 of 166), the specificity was 95 per cent (429 of 450), and the accuracy was 93 per cent (573 of 616). The relative sensitivity, specificity, and accuracy of magnetic resonance imaging also were calculated for each articular surface (Table IV). Once again, grades 0 and 1 were considered disease-negative status and grades 2, 3, and 4 were considered disease-positive status. The lateral tibial plateau was associated with the poorest sensitivity, with proportionately more false-negative readings. In contrast, the patellar facets (Fig. 1-A) and the trochlea were associated with relatively high sensitivity and specificity, possibly because of the relative thickness of the cartilage in this region. False-positive results (observed when a surface that had been prospectively classified as grade 2, 3, or 4 on the basis of magnetic resonance images was classified as grade 0 or 1 at arthroscopy) typically were associated with smoothly marginated ulcerations without fibrillation of the adjacent surface (Fig. 5) and with very posterior condylar lesions (Fig. 6). Reader 1 had prospectively classified five grade-0 surfaces (as determined arthroscopically) as grade 2, one grade-0 surface as grade 3, nineteen grade-1 surfaces as grade 2, and one grade-1 surface as grade 3, for a total of twenty-six false-positive results. Reader 2 had prospectively classified fifteen grade-1 surfaces as grade 2, one grade-1 surface as grade 3, four grade-0 surfaces as grade 2, and one grade-0 surface as grade 3, for a total of twenty-one false-positive results. None of the surfaces that were classified as grade 0 or 1 at arthroscopy had been prospectively classified as grade 4 by either reader.

Discussion

Introduction | Materials and Methods | Results | Discussion | References

The results of our study suggest that the use of a specialized fast-spin-echo magnetic resonance imaging sequence with a high-resolution matrix allows for an accurate assessment of articular cartilage in the knee with little interobserver variability. Our data compare well with those in the study by Disler et al.⁵, in which a fat-suppressed three-dimensional T1-weighted gradient-echo technique was used to assess chondral defects in the knee and exact agreement was found for forty-three (63 per cent) of sixty-eight true-positive interpretations. Recht et al.²³ reported a sensitivity of 81 per cent (thirty of thirty-seven), a specificity of 97 per cent (283 of 291), and an accuracy of 95 per cent (313 of 328) in a study in which a similar gradient-echo technique was used to grade 328 articular surfaces in forty-one patients. Arthroscopy was used as the comparative standard. In that study, evaluation was limited to the patellofemoral joint, arthroscopic grading was performed retrospectively by viewing a videotape, and the magnetic resonance images were graded by consensus review. The pulse sequence described in the present study offers a slight advantage in that it allows adjacent menisci, ligaments, and subchondral bone to be evaluated. In addition, after arthroscopic treatment, metallic instrumentation (such as suture anchors and compression screws) and residual metallic debris from the arthroscope may cause local disturbance in the magnetic field. Fast-spin-echo sequences are superior to gradient-echo techniques in diminishing the susceptibility artifact caused by metallic instrumentation, which may obscure the overlying articular cartilage²¹. Previous authors have reported that magnetic resonance imaging has a lower sensitivity with regard to the detection of partial-thickness chondral lesions. Ochi et al., in a study of sixty-three patients, reported that magnetic resonance imaging had a sensitivity of 38 per cent (nine of twenty-four) for the detection of chondral injuries on prospective evaluation and a sensitivity of 75 per cent (eighteen of twenty-four) on retrospective evaluation. However, the pulse sequence that was used to evaluate the cartilage, a conventional spin-echo multiecho acquisition with a relatively thick (five-millimeter) slice and a 256-by-256 matrix, was different from ours and yielded poorer spatial resolution. Similarly, Blackburn et al. reported that arthroscopy was more sensitive than both plain radiography and magnetic resonance imaging for the evaluation of chondral abnormalities in patients who had osteoarthrosis of the knee. Those authors conducted a consensus review of the magnetic resonance images that had been performed for sixteen of their thirty-three patients. The images in that study were performed with T1-weighted fat-suppressed spin-echo sequences as well as gradient-recalled sequences; however, the maximum matrix was 256 by 256, which limited the spatial resolution. Finally, Spiers et al., in a prospective study of fifty-eight patients, reported that three-dimensional gradient-echo magnetic resonance images that were performed with a 256-by-256 matrix had a high specificity (100 per cent) but a low sensitivity (18 per cent) for the detection of lesions in the articular cartilage of the knee. We believe that proper technique is crucial to the ability to detect subtle chondral injuries with the use of magnetic resonance imaging. Higher resolution allows for the detection of early morphological changes in the articular surface, such as fibrillation, and places less diagnostic reliance on changes in signal intensity, which may be subject to imaging artifacts^7,8. Even in the absence of substantial intra-articular joint fluid, the pulse sequence that we used permitted differentiation among articular cartilage, subchondral bone, and meniscus (Fig. 7). This obviates the need for an intra-articular contrast agent and preserves the advantage of magnetic resonance imaging as a non-invasive tool. Poor contrast between joint fluid and the superficial layers of cartilage has been implicated as a cause of relatively poor sensitivity when magnetic resonance imaging has been used prospectively for the detection of chondral injuries²⁶. In the present study, confirmation of chondral injuries in two planes helped to avoid partial volume effects, particularly in the region where the condyles curve toward the intercondylar notch. Because of the obliquity of the patellar facets, axial images were deemed superior for the visualization of the articular cartilage of the patella. Magnetic resonance imaging demonstrated relatively poor sensitivity (seven of twelve) with regard to the detection of severe (grade-2, 3, or 4) chondral lesions of the lateral tibial plateau; this finding is in part reflective of the relatively small number of chondral lesions that were noted in this region at arthroscopy. It should be noted, however, that the lateral tibial plateau also was described by Disler et al.⁵ as a difficult region in which to detect chondral lesions. In part, this difficulty may be a function of the convex surface of the plateau, which, when subjected to sectioning into tomographic coronal and sagittal images, may impart more partial volume effects and imaging artifacts. In contrast, magnetic resonance imaging demonstrated superior sensitivity (92 and 96 per cent) with regard to the detection of defects involving the patellar facets, which are relatively thick and straight; this finding also is consistent with those of other studies²³. When the results of magnetic resonance imaging and arthroscopy were compared, there appeared to be a tendency for the readers of the magnetic resonance images to overdiagnose chondral softening, or grade-1 lesions. It is unclear if this finding suggests that magnetic resonance imaging has superior sensitivity with regard to the detection of edema in the cartilage, which may not be apparent as palpable softening at arthroscopy, or if it represents an imaging artifact. The reported truncation artifact⁷ that could account for a factitious laminar appearance on magnetic resonance pulse sequences is minimized with our technique by virtue of the relatively small pixel size and the high-resolution matrix. In addition, chemical shift misregistration, which may account for misregistration of the interfaces between the fat contained in the subchondral marrow and the fluid contained in the cartilage, is minimized with the additional fat-suppression technique. Because of its reliance on morphological changes rather than changes in signal intensity, our magnetic resonance pulse sequence can be used in conjunction with most standardized arthroscopic classification systems, which are designed to assess morphological changes in the chondral surface. Of note, given the superior visualization of subtle chondral lesions with the use of a smaller pixel size, we subsequently modified our imaging parameters to include a matrix of 512 by 384 in the sagittal plane and a matrix of 512 by 256 in the coronal plane. The present study had a number of limitations. First, there was a selection bias because of the high prevalence of chondral injuries observed at our tertiary-care center. Second, the orthopaedic surgeons had access to the magnetic resonance images and reports at the time of arthroscopy. However, the magnetic resonance images primarily were used to assess menisci and ligaments and the surgeons did not have access to the grading sheets. Finally, although our patients ranged in age from twenty-three to eighty-two years, the average age of thirty-eight years is relatively young, indicating that a greater number of chondral lesions were due to trauma as opposed to osteoarthrosis. Additional studies of older patients are necessary to determine the value of this pulse sequence in the detection of lesions due to osteoarthrosis. Despite these limitations, we believe that this magnetic resonance pulse sequence provides a non-invasive, accurate, and reproducible method with which to diagnose traumatic chondral lesions of the knee. This is the largest study to date, of which we are aware, in which all of the articular surfaces of the knee were assessed by independent readers with the use of magnetic resonance imaging. With the use of readily available software on all high-field-strength magnetic resonance units, this sequence should provide a reasonable method with which to study patients after operative intervention for lesions in the cartilage. NOTE: The authors acknowledge the contributions of George A. Paletta, M.D., Riley Williams, M.D., and Margaret Peterson, Ph.D.

References

Introduction | Materials and Methods | Results | Discussion | References

Blackburn, W. D., Jr., Bernreuter, W. K., Rominger, M.,Loose, L. L. Arthroscopic evaluation of knee articular cartilage: a comparison with plain radiographs and magnetic resonance imaging. J. Rheumatol.,21: 675-679. 1994;21675 1994 [PubMed]

Brittberg, M., Lindahl, A., Nilsson, A., Ohlsson, C., Isaksson, O.,Peterson, L. Treatment of deep cartilage defects in the knee with autologous chondrocyte transplantation. New England J. Med.,331: 889-895. 1994;331889 1994

Convery, F. R., Akeson, W. H.,Keown, G. H. The repair of large osteochondral defects. An experimental study in horses. Clin. Orthop.,82: 253-262. 1972;82253 1972 [PubMed]

Disler, D. G., McCauley, T. R., Wirth, C. R.,Fuchs, M. D. Detection of knee hyaline cartilage defects using fat-suppressed three-dimensional spoiled gradient-echo MR imaging: comparison with standard MR imaging and correlation with arthroscopy. AJR: Am. J. Roentgenol.,165: 377-382. 1995;165377 1995 [PubMed]

Disler, D. G., McCauley, T. R., Kelman, C. G., Fuchs, M. D., Ratner, L. M., Wirth, C. R.,Hospodar, P. P. Fat-suppressed three-dimensional spoiled gradient-echo MR imaging of hyaline cartilage defects in the knee: comparison with standard MR imaging and arthroscopy. AJR: Am. J. Roentgenol.,167: 127-132. 1996;167127 1996 [PubMed]

Engkvist, O.,Ohlsen, L. Reconstruction of articular cartilage with free autologous perichondrial grafts. An experimental study in rabbits. Scandinavian J. Plast. and Reconstr. Surg.,13: 269-274. 1979;13269 1979

Erickson, S. J., Waldschmidt, J. G., Czervionke, L. F.,Prost, R. W. Hyaline cartilage: truncation artifact as a cause of trilaminar appearance with fat-suppressed three-dimensional spoiled gradient-recalled sequences. Radiology,201: 260-264. 1996;201260 1996 [PubMed]

Frank, L. R., Brossmann, J., Buxton, R. B.,Resnick, R. MR imaging truncation artifacts can create a false laminar appearance in cartilage. AJR: Am. J. Roentgenol.,168: 547-554. 1997;168547 1997 [PubMed]

Hangody, L.,Karpati, Z. A new surgical treatment of localized cartilaginous defects of the knee. Hungarian J. Orthop. Trauma,37: 237-242. 1994;37237 1994

Johnson, L. L. Arthroscopic abrasion arthroplasty historical and pathologic perspective: present status. Arthroscopy,2: 54-69. 1986;254 1986 [PubMed]

Johnson-Nurse, C.,Dandy, D. J. Fracture-separation of articular cartilage in the adult knee. J. Bone and Joint Surg.,67-B(1): 42-43. 1985;67-B(1)42 1985

Kim, H. K. W., Moran, M. E.,Salter, R. B. The potential for regeneration of articular cartilage in defects created by chondral shaving and subchondral abrasions. An experimental investigation in rabbits. J. Bone and Joint Surg.,73-A: 1301-1315. Oct. 1991;73-A1301 1991

Landis, J. R.,Koch, G. G. The measurement of observer agreement for categorical data. Biometrics,33: 159-174. 1977;33159 1977 [PubMed]

Linklater, J.,Potter, H. G. Imaging of chondral defects. Op. Tech. Orthop.,7: 279-288. 1997;7279 1997

Mitchell, N.,Shepard, N. The resurfacing of adult rabbit articular cartilage by multiple perforations through the subchondral bone. J. Bone and Joint Surg.,58-A: 230-233. March 1976;58-A230 1976

Ochi, M., Sumen, Y., Kanda, T., Ikuta, Y.,Itoh, K. The diagnostic value and limitation of magnetic resonance imaging on chondral lesions in the knee joint. J. Arthroscopy,10: 176-183. 1994;10176 1994

O'Driscoll, S. W.,Salter, R. B. The repair of major osteochondral defects in joint surfaces by neochondrogenesis with autogenous osteoperiosteal grafts stimulated by continuous passive motion. An experimental investigation in the rabbit. Clin. Orthop.,208: 131-140. 1986;208131 1986 [PubMed]

O'Driscoll, S. W., Keeley, F. W.,Salter, R. B. Durability of regenerated articular cartilage produced by free autogenous periosteal grafts in major full-thickness defects in joint surfaces under the influence of continuous passive motion. A follow-up report at one year. J. Bone and Joint Surg.,70-A: 595-606. April 1988;70-A595 1988

Outerbridge, R. E. The etiology of chondromalacia patellae. J. Bone and Joint Surg.,43-B(4): 752-757. 1961;43-B(4)752 1961

Outerbridge, R. E.,Dunlop, J. A. Y. The problem of chondromalacia patellae. Clin. Orthop.,110: 177-196. 1975;110177 1975 [PubMed]

Potter, H. G., Rodeo, S. A., Wickiewicz, T. L.,Warren, R. F. MR imaging of meniscal allografts: correlation with clinical and arthroscopic outcomes. Radiology,198: 509-514. 1996;198509 1996 [PubMed]

Recht, M. P.,Resnick, D. MR imaging of articular cartilage: current status and future directions. AJR: Am. J. Roentgenol.,163: 283-290. 1994;163283 1994 [PubMed]

Recht, M. P., Piraino, D. W., Paletta, G. A., Schils, J. P.,Belhobeck, G. H. Accuracy of fat-suppressed three-dimensional spoiled gradient-echo FLASH MR imaging in the detection of patellofemoral articular cartilage abnormalities. Radiology,198: 209-212. 1996;198209 1996 [PubMed]

Rodrigo, J. J.; Steadman, J. R.; and Silliman, J. F.: Osteoarticular injuries of the knee. In Operative Orthopaedics, edited by M. W. Chapman. Ed. 2, vol. 3, pp. 2077-2082. Philadelphia, J. B. Lippincott, 1993.

Rubak, J. M. Reconstruction of articular cartilage defects with free periosteal grafts. An experimental study. Acta Orthop. Scandinavica,53: 175-180. 1982;53175 1982

Speer, K. P., Spritzer, C. E., Goldner, J. L.,Garrett, W. E., Jr. Magnetic resonance imaging of traumatic knee articular cartilage injuries. Am. J. Sports Med.,19: 396-402. 1991;19396 1991 [PubMed]

Spiers, A. S. D., Meagher, T., Ostlere, S. J., Wilson, D. J.,Dodd, C. A. F. Can MRI of the knee affect arthroscopic practice? A prospective study of 58 patients. J. Bone and Joint Surg.,75-B(1): 49-52. 1993;75-B(1)49 1993

Terry, G. C., Flandry, F., Manen, J. W. V.,Norwood, L. A. Isolated chondral fractures of the knee. Clin. Orthop.,234: 170-177. 1988;234170 1988 [PubMed]

Topics

View Large | Download Slide(.ppt)
Add to My JBJS

FIG1-B:: Fig. 1-B Coronal fast-spin-echo magnetic resonance image demonstrating uniform thickness of the cartilage and a homogeneous appearance over the condyles and plateaus.

View Large | Download Slide(.ppt)
Add to My JBJS

FIG4-B:: Fig. 4-B Coronal fast-spin-echo magnetic resonance image confirming the full-thickness defect (arrow).

View Large | Download Slide(.ppt)
Add to My JBJS

FIG4-C:: Fig. 4-C Arthroscopic photograph confirming a full-thickness chondral defect over the posterior aspect of the medial femoral condyle. The defect measured eight by twenty millimeters.

View Large | Download Slide(.ppt)
Add to My JBJS

View Larger

TABLE I: ARTHROSCOPIC GRADING OF THE SIX HUNDRED AND SIXTEEN ARTICULAR SURFACES*

Surface	Grade 0	Grade 1	Grade 2	Grade 3	Grade 4
Medial	47	10	18	9	4
patellar
facet
Lateral	49	14	13	8	4
patellar
facet
Trochlea	50	14	7	10	7
Medial	34	8	13	19	14
femoral
condyle
Lateral	71	9	4	2	2
femoral
condyle
Medial	56	12	13	2	5
tibial
plateau
Lateral	61	15	7	3	2
tibial
plateau
Total	368	82	75	53	38

*The values are given as the number of surfaces.

Add to My JBJS

TABLE I: ARTHROSCOPIC GRADING OF THE SIX HUNDRED AND SIXTEEN ARTICULAR SURFACES*

Surface	Grade 0	Grade 1	Grade 2	Grade 3	Grade 4
Medial	47	10	18	9	4
patellar
facet
Lateral	49	14	13	8	4
patellar
facet
Trochlea	50	14	7	10	7
Medial	34	8	13	19	14
femoral
condyle
Lateral	71	9	4	2	2
femoral
condyle
Medial	56	12	13	2	5
tibial
plateau
Lateral	61	15	7	3	2
tibial
plateau
Total	368	82	75	53	38

*The values are given as the number of surfaces.

View Larger

TABLE II: AGREEMENT BETWEEN THE FINDINGS OF THE INDEPENDENT READERS OF THE MAGNETIC RESONANCE IMAGES AND THE FINDINGS AT ARTHROSCOPY

Surface	Exact Agreement (per cent)		Difference of One Grade (per cent)		Difference of Two Grades (per cent)
Reader 1	Reader 2	Reader 1	Reader 2	Reader 1	Reader 2
Medial patellar facet	66	70	30	29	4	1
Lateral patellar facet	71	76	27	20	2	3
Trochlea*	75	74	25	25	0	2
Medial femoral condyle	67	65	28	32	5	2
Lateral femoral condyle	80	78	19	21	1	1
Medial tibial plateau†	70	79	29	20	1	1
Lateral tibial plateau	77	74	18	22	5	4
All articular surfaces	72	73	25	24	3	2

Add to My JBJS

TABLE II: AGREEMENT BETWEEN THE FINDINGS OF THE INDEPENDENT READERS OF THE MAGNETIC RESONANCE IMAGES AND THE FINDINGS AT ARTHROSCOPY

Surface	Exact Agreement (per cent)		Difference of One Grade (per cent)		Difference of Two Grades (per cent)
Reader 1	Reader 2	Reader 1	Reader 2	Reader 1	Reader 2
Medial patellar facet	66	70	30	29	4	1
Lateral patellar facet	71	76	27	20	2	3
Trochlea*	75	74	25	25	0	2
Medial femoral condyle	67	65	28	32	5	2
Lateral femoral condyle	80	78	19	21	1	1
Medial tibial plateau†	70	79	29	20	1	1
Lateral tibial plateau	77	74	18	22	5	4
All articular surfaces	72	73	25	24	3	2

View Larger

TABLE III: WEIGHTED KAPPA STATISTICS

Comparison	Weighted Kappa Statistic
Reader 1 and arthroscopy	0.88
Reader 2 and arthroscopy	0.89
Reader 1 and reader 2	0.93

Add to My JBJS

TABLE III: WEIGHTED KAPPA STATISTICS

Comparison	Weighted Kappa Statistic
Reader 1 and arthroscopy	0.88
Reader 2 and arthroscopy	0.89
Reader 1 and reader 2	0.93

View Larger

TABLE IV: RELATIVE SENSITIVITY, SPECIFICITY, AND ACCURACY FOR EACH ARTICULAR SURFACE FOR THE FINDINGS OF READER 1 COMPARED WITH THOSE AT ARTHROSCOPY*

Surface	Findings on Magnetic Resonance Imaging				Disease- Positive Status at Arthroscopy†	Sensitivity	Specificity
True- Positive†	True- Negative†	False- Positive†	False- Negative†
						(per cent)	(per cent)
Medial patellar facet	30	51	6	1	31	97	89
Lateral patellar facet	23	59	4	2	25	92	94
Trochlea	22	61	3	2	24	92	95
Medial femoral condyle	38	40	2	8	46	83	95
Lateral femoral condyle	8	77	3	0	8	100	96
Medial tibial plateau	16	65	3	4	20	80	96
Lateral tibial plateau	7	71	5	5	12	58	93
Total	144	424	26	22	166	87	94

*Grades 0 and 1 were considered disease-negative status, and grades 2, 3, and 4 were considered disease-positive status. †The values indicate the number of surfaces associated with each finding.

Add to My JBJS

TABLE IV: RELATIVE SENSITIVITY, SPECIFICITY, AND ACCURACY FOR EACH ARTICULAR SURFACE FOR THE FINDINGS OF READER 1 COMPARED WITH THOSE AT ARTHROSCOPY*

Surface	Findings on Magnetic Resonance Imaging				Disease- Positive Status at Arthroscopy†	Sensitivity	Specificity
True- Positive†	True- Negative†	False- Positive†	False- Negative†
						(per cent)	(per cent)
Medial patellar facet	30	51	6	1	31	97	89
Lateral patellar facet	23	59	4	2	25	92	94
Trochlea	22	61	3	2	24	92	95
Medial femoral condyle	38	40	2	8	46	83	95
Lateral femoral condyle	8	77	3	0	8	100	96
Medial tibial plateau	16	65	3	4	20	80	96
Lateral tibial plateau	7	71	5	5	12	58	93
Total	144	424	26	22	166	87	94