| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Departments of Radiology (D.G.D., E.R., D.A.M.), Orthopaedic Surgery (D.G.D., J.S.W.), and Biomedical Engineering (J.S.W.), Medical College of Virginia of Virginia Commonwealth University, Main Hospital, 3rd Floor, 401 N 12th St, Richmond, VA 23298-0615, and the Department of Radiology, Yale University School of Medicine (T.R.M.). From the 1999 RSNA scientific assembly. Received July 20, 1999; revision requested August 17; revision received September 2; accepted September 16. Supported in part by ATL Ultrasound, Bothell, Washington. Address correspondence to D.G.D.
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: US was used to evaluate 175 porcine and 16 human knee surfaces with a linear 5–12-MHz transducer. Porcine defects of varying diameter and depth were surgically created. Each porcine surface was independently assessed in blinded fashion by two radiologists for the presence and severity of defects. Accuracy of detection, interobserver reliability, and concordance between US and surgical grades were determined. Human specimens were retrieved from knees of patients who underwent joint arthroplasty. Defects in human knees detected with US were correlated with defects seen at direct surface visualization.
RESULTS: Sensitivities for detection of porcine defects were 94% and 93% for readers 1 and 2, respectively; specificities were 90% and 77%, respectively; positive predictive values were 98% and 95%, respectively; and negative predictive values were 78% and 73%, respectively. Interobserver agreement was high (weighted 
= 0.80), and concordance between US and surgical grades for both readers was high (weighted = 0.90 and 0.78). In human cartilage, the distribution of cartilage denudation determined at US was the same as that determined at direct visualization.
CONCLUSION: High-frequency US was accurate and reliable for detection and grading of knee articular cartilage defects.
Index terms: Animals • Knee, injuries, 452.40, 452.485, 452.86 • Knee, ligaments, menisci, and cartilage, 452.40, 452.70 • Knee, US, 452.12981, 452.12989 • Specimens • Ultrasound (US), experimental studies, 452.12981, 452.12989
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We hypothesized that articular cartilage defects in the knee could be accurately and reliably assessed with high-frequency US. We studied a range of abnormalities in porcine and human knee articular cartilage specimens to determine the ability to detect and grade cartilage defects by using US.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
For the first part of the study, nine of the 175 surfaces were randomly selected. On three surfaces (medial patellar, medial femoral, medial tibial surfaces), no defects were created. These surfaces were classified as having a grade 0 defect. The other six surfaces were used for the creation of articular defects. A grade 1 defect was a surface abrasion of minimal depth that was created with a scoring device. A grade 2 defect consisted of a partial-thickness defect with a depth of less than 100% of the thickness of articular cartilage, in which the subchondral bone was not exposed. A grade 3 defect consisted of a full-thickness defect with exposed but otherwise intact subchondral bone. A grade 4 defect consisted of an osteochondral defect that included the subchondral layer of bone.
On each of the six surfaces, four cylindric defects (grades 1, 2, 3, and 4) of the same diameter were created. The diameter of the defect on each surface varied from 2 to 12 mm in 2-mm increments. All defects were created in the articular cartilage by using graduated-diameter, hollow, cylindric scoring devices. A single investigator (E.R.), who was not one of the interpreting radiologists, created all defects.
All nine surfaces were then studied with real-time US by using a state-of-the-art US device with a linear 5–12-MHz transducer (HDI 5000 with 12-5 MHz transducer; ATL Ultrasound, Bothell, Wash) and settings optimized by the manufacturer for small-parts imaging and two-dimensional optimization-selected resolution. Other pertinent imaging parameters included a 150-dB dynamic range, a medium frame rate, a depth of 3.0 cm with two focal zones, and constant time gain compensation. The articular surfaces were studied by two experienced musculoskeletal radiologists (D.G.D., D.A.M.) who worked in consensus. The two reviewers were not blinded to defect location, size, or depth. The two reviewers determined the US appearance of cartilage surfaces and established a baseline level of interpretive skills for the subsequent blinded evaluation of the remaining 166 surfaces, as described subsequently. These specimens were also studied to determine whether limitations of spatial resolution would prevent detection for this range of lesion sizes.
For the second part of the study, the remaining 166 surfaces were used to create groups of defects categorized into the five grades for blinded US evaluation. Of the 166 surfaces, 31 were without a surgically created defect (grade 0), 32 had a grade 1 defect, 34 had a grade 2 defect, 35 had a grade 3 defect, and 34 had a grade 4 defect. Only one defect per surface was created, although the location of the defect on the surface was randomly selected and unknown to the interpreting radiologist. Defect size and depth were evenly distributed among the seven surface locations (medial and lateral patellar, medial and lateral femoral, medial and lateral tibial, and trochlear surfaces). The same investigator (E.R.) who created the defects in the first part of the study created all defects for this part of the study by using 4-, 8-, or 12-mm-diameter cylindric hollow scoring devices. The 135 modified surfaces were randomly sorted with the 31 normal (unmodified) surfaces.
Each of the 166 surfaces was then evaluated separately and independently by the same interpreting radiologists (D.G.D., D.A.M.) and with the same US device, transducer, and technical settings as described for the first part of the study. Each of the two interpreting radiologists was blinded with regard to the surfaces studied and the presence, location, and morphologic features of the defect. To accomplish this task, neither of the interpreting radiologists handled the specimens; rather, a third investigator (E.R.) held the specimen in a water bath in a darkened room while the interpreting radiologist handled the US transducer. The articular surface of each specimen was oriented toward the US probe, so that the articular surface could be interrogated. The interpreting radiologist controlled the position of the US transducer. The orientation of the specimen did not have to be strictly controlled for each investigator, because the entire surface of each specimen was scanned in search of a potential defect, the location of which on the surface was unknown to interpreting radiologist. A fourth investigator managed the technical settings of the US device, including depth, focus, and overall gain, and recorded the images on magneto-optical disk.
All assessments were performed during real-time scanning and were based on the interpreting radiologist's findings. For each surface, the interpreter determined whether an articular cartilage defect was present and recorded his level of confidence on a scale of one to five as follows: 1, definitely no defect; 2, probably no defect; 3, indeterminate; 4, probably a defect; and 5, definitely a defect. If a defect was believed to be present, the interpreter also graded the depth of the defect as grade 0–4, with these US grades identical to the surgical grades described earlier. Defect size was not analyzed separately.
Receiver operator characteristic curves could not be calculated because the raw data for the confidence scores were degenerate (confidence scores were categorized mostly as level 1 or 5, with a nonuniform distribution among levels of confidence). Therefore, 2x2 contingency tables were calculated for each interpreter to determine sensitivity, specificity, and positive and negative predictive values by using confidence levels of 4 and 5 as positive for a defect, and confidence levels of 1–3 as negative for a defect. This classification was chosen so that the results would favor a lower sensitivity for the detection of articular defects. Concordance of the US grades, as determined by each interpreter, with the surgical grades was assessed by using the weighted statistic. In addition, interobserver agreement for grades was assessed with the weighted statistic. A score of less than 0.40 was considered to indicate poor agreement; that of 0.40–0.59, fair agreement; that of 0.60–0.74, good agreement; and that of 0.75–1.00, excellent agreement (16).
Human Knee Evaluation
To determine whether naturally occurring articular defects could be detected and graded with US, we studied 16 human knee specimens retrieved from patients who underwent total knee arthroplasty. This patient population is known to have severe articular cartilage damage and, thus, would be ideal for the comparison of naturally occurring and surgically created defects. Because, as was expected, most of the surfaces had numerous areas with full-thickness loss of articular cartilage, the surfaces were studied by the two interpreters in consensus and unblinded with regard to the surfaces and defects under study.
The human knee articular specimens were not further modified before study. US and direct visual assessment of the specimens were performed by the same two radiologists (D.G.D., D.A.M.) to evaluate for the presence and extent of cartilage defects. The same US device, transducer, and technical settings as for the porcine knee evaluation were used. Defect depth was graded 0 to 4, as described earlier.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
|
|
Sensitivities for the detection of articular cartilage defects were 94% and 93% for readers 1 and 2, respectively (Table 1). The specificities were 90% and 77% for readers 1 and 2, respectively; the positive predictive values, 98% and 95%, respectively; and the negative predictive values, 78% and 73%, respectively.
|
|
|
values for interpreter concordance with surgical grades indicated excellent agreement for both readers, with values of 0.90 for reader 1 and 0.78 for reader 2. Interobserver agreement for readers 1 and 2 also was excellent, with a weighted value of 0.80.
Human Knee Evaluation
In each of the 16 human knee specimens, a sharp hyperechoic interface between cartilage and the water bath, similar to that in the porcine specimens, was shown in areas of remnant cartilage, except where the cartilage showed surface fibrillation or loss of thickness. In these latter cases, the interface between articular cartilage and the water bath was ill defined, and, occasionally, evaginations of hyperechoic material were shown at the articular surface. A smooth transition was always demonstrated between articular surfaces where cartilage was completely denuded and adjacent surfaces where cartilage was still present (Fig 2a). The distribution of morphologic cartilage loss as determined at US was the same as that determined at direct visualization (Fig 2b).
|
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The ability to detect articular cartilage defects is important for a number of reasons. First, detection of articular cartilage defects may provide an explanation for a patient's symptoms, because articular defects can occur as isolated findings in joints. Second, imaging will allow determination of which patients may benefit from new therapies, such as autologous chondrocyte implantation (1), mosaicplasty (2,3), and treatment with chondroprotective drugs, which are directed at repairing focal articular cartilage defects or slowing the progression of disease. Because such therapies are currently undergoing clinical investigation, accurate and reliable imaging methods would be useful to determine treatment efficacy. Thus, imaging will play an important role in guiding and helping determine the success of articular cartilage therapies as they evolve.
There are many advantages to the use of US for evaluation of articular cartilage. First, high-frequency transducers permit high resolution and, thus, an increased potential for the detection and characterization of small-diameter or shallow lesions. Second, US permits high soft-tissue contrast. Of importance for the purpose of articular cartilage assessment, a readily recognizable echogenic interface at the surface of cartilage and the presence of a strong reflector at the interface between cartilage and subchondral bone allow detection of subtle lesions such as abrasions and partial-thickness defects and determination of the relative depth of articular cartilage. Because contrast at US is determined by properties that differ from those of MR imaging, US may have advantages over MR imaging in terms of the ability to help evaluate lesions. For example, MR imaging for postoperative evaluation of cartilage repair sites can be limited by the presence of microscopic metallic artifacts. However, we did not compare US with MR imaging; such studies await further investigation.
There were several limitations to our study. First, we were unable to determine any internal characteristic of the articular cartilage, as might be expected because cartilage is highly anisotropic and because laminations in cartilage have been shown with MR imaging (17,18). It is not clear whether the lack of detection of such internal order is related to an intrinsic limitation of US or to the need for resolution and contrast higher than those currently available commercially. Although our study was performed with a transducer capable of the highest frequency available for routine clinical use, previous results (14) with 50-MHz transducers suggest that layers in immature cartilage can be detected. We are aware of no study in which laminations were shown in mature articular cartilage.
Second, although our study was an evaluation of morphologic derangement of articular cartilage, we did not assess whether US could be used to detect early degeneration of cartilage in which fine surface fibrillation or changes in proteoglycan content occur (12,13,15). This was beyond the scope of our initial investigation.
Third, we did not assess for intrinsic differences in the echotexture of articular cartilage between normal porcine specimens and human specimens with severe osteoarthritis. We did not specifically evaluate whether US could help discriminate between hyaline cartilage and fibrocartilage, which would have enormous implications for postoperative assessment after cartilage repair therapy. This is because the restoration of hyaline cartilage rather than fibrocartilage after repair is highly desirable, since hyaline cartilage is biomechanically superior to fibrocartilage for normal joint lubrication and function. In addition, lack of a fluid interface with articular cartilage might be expected to reduce sonographic accuracy. We simulated the presence of joint fluid, which optimized the likelihood of the detection of defects.
Fourth, our study was performed ex vivo, and naturally occurring defects may be more difficult to detect than surgically created defects because of differences in the morphology of the defects and in the constitution of adjacent cartilage, which may contain internal changes due to degeneration. Therefore, the results we report may be more favorable than those that may be achieved in the clinical setting. Further study is necessary in this area.
Whereas some articular surfaces of the knee, such as the trochlear groove, the posterior femoral condyles in extension, and the weight-bearing femoral condyles in extreme flexion, are accessible with an external US probe, patellar and tibial surfaces cannot be evaluated in this manner. This is currently the major limitation of the application of US for articular cartilage assessment. Future studies with intact knees would be necessary to determine the value of US for articular cartilage assessment. To establish the feasibility of US of joint cartilage, intraarticular probes may prove to be useful to achieve sufficient imaging access to articular surfaces. Although use of an intraarticular probe would change articular evaluation from a noninvasive to an invasive procedure, this may have advantages because the entire joint cartilage could be assessed and, unlike in arthroscopy, determination of features of cartilage subsurface integrity, such as cartilage depth, would be possible.
Practical application: We showed that real-time US with a high-frequency transducer could help detect and grade articular cartilage defects of the knee in porcine specimens, with high accuracy and reliability. Findings in human surgical specimens were correlated with those in porcine specimens. Further studies await the determination of whether US will be useful in the clinical setting and whether it will have any potential application in the evaluation of nonmorphologic derangement or the discrimination between hyaline cartilage and fibrocartilage.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Author contributions: Guarantor of integrity of entire study, D.G.D.; study concepts, D.G.D., D.A.M., J.S.W.; study design, all authors; definition of intellectual content, D.G.D.; literature research, D.G.D.; experimental studies, D.G.D., E.R., D.A.M., J.S.W.; data acquisition, D.G.D., E.R., D.A.M., J.S.W.; data analysis, D.G.D., E.R., T.R.M.; statistical analysis, T.R.M.; manuscript preparation, D.G.D.; manuscript editing and review, all authors.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Kuroki, Y. Nakagawa, K. Mori, K. Ikeuchi, and T. Nakamura Mechanical Effects of Autogenous Osteochondral Surgical Grafting Procedures and Instrumentation on Grafts of Articular Cartilage Am. J. Sports Med., April 1, 2004; 32(3): 612 - 620. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. L. Cameron, K. K. Briggs, and J. R. Steadman Reproducibility and Reliability of the Outerbridge Classification for Grading Chondral Lesions of the Knee Arthroscopically Am. J. Sports Med., January 1, 2003; 31(1): 83 - 86. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
