Juvenile Osteochondritis Dissecans: Correlation Between Histopathology and MRI

Transcription

1 Pediatric Imaging Original Research Zbojniewicz et al. Histopathology-MRI Correlation in Juvenile OCD Pediatric Imaging Original Research Andrew M. Zbojniewicz 1 Keith F. Stringer 2 Tal Laor 1 Eric J. Wall 3 Zbojniewicz AM, Stringer KF, Laor T, Wall EJ Keywords: knee, MRI, osteochondritis dissecans, pathology, pediatrics DOI: /AJR Received August 4, 2014; accepted after revision October 26, E. J. Wall is a medical advisor for, has a royalty agreement with, and has equity in SpineForm. Based on a presentation at the Society of Skeletal Radiology 2013 annual meeting, San Antonio, TX. 1 Department of Radiology, Cincinnati Children s Hospital Medical Center, University of Cincinnati, 3333 Burnet Ave, MLC 5031, Cincinnati, OH Address correspondence to A. M. Zbojniewicz (andrew.zbojniewicz@cchmc.org). 2 Department of Pathology, Cincinnati Children s Hospital Medical Center, University of Cincinnati, Cincinnati, OH. 3 Department of Orthopedics, Cincinnati Children s Hospital Medical Center, University of Cincinnati, Cincinnati, OH. WEB This is a web exclusive article. AJR 2015; 205:W114 W X/15/2051 W114 American Roentgen Ray Society Juvenile Osteochondritis Dissecans: Correlation Between Histopathology and MRI OBJECTIVE. The objective of our study was to correlate specimens of juvenile osteochondritis dissecans (OCD) lesions of the knee to MRI examinations to elucidate the histopathologic basis of characteristic imaging features. MATERIALS AND METHODS. Five children (three boys and two girls; age range, years old) who underwent transarticular biopsy of juvenile OCD lesions of the knee were retrospectively included in this study. Two radiologists reviewed the MRI examinations and a pathologist reviewed the histopathologic specimens and recorded characteristic features. Digital specimen photographs were calibrated to the size of the respective MR image with the use of a reference scale. Photographs were rendered semitransparent and overlaid onto the MR image with the location chosen on the basis of the site of the prior biopsy. RESULTS. A total of seven biopsy specimens were included. On MRI, all lesions showed cystlike foci in the subchondral bone, bone marrow edema pattern on proton density or T2- weighted images, and relatively thick unossified epiphyseal cartilage. In four patients, a laminar signal intensity pattern was seen, and two patients had multiple breaks in the subchondral bone plate. Fibrovascular tissue was found at histopathology in all patients. Cleft spaces near the cartilage-bone interface and were seen in all patients while chondrocyte cloning was present in most cases. Focal bone necrosis and inflammation were infrequent MRI findings. Precise correlation of the MRI appearance to the histopathologic overlays consistently was found. CONCLUSION. A direct correlation exists between the histopathologic findings and the MRI features in patients with juvenile OCD. Additional studies are needed to correlate these MRI features with juvenile OCD healing success rates. J uvenile osteochondritis dissecans (OCD) is a frequent cause of knee pain in children. There has been a recent increase in the incidence of this condition in pediatric patients overall and in the percentage of children younger than 10 years old [1]. Although there is no consensus about the appropriate treatment, in general, articular cartilage instability at the OCD site is an indication for surgery, whereas stable lesions are often treated nonoperatively. Compared with adults with OCD, skeletally immature patients with juvenile OCD are more likely to have stable lesions and to have a better prognosis overall. However, many lesions that appear stable on imaging and at arthroscopy still will not respond to nonoperative management and it is difficult to predict in which patients nonoperative treatment will be successful [2 6]. The determination of which treatment option is best for a specific patient and of which patient is likely to respond is important because selection of the appropriate therapy will allow timely symptomatic relief and a return to normal activities. If a juvenile OCD lesion is left untreated or unhealed, there is a potential for early onset of degenerative joint disease. A previous study that evaluated the histopathology of juvenile OCD lesions indicated that deep instability at the OCD site was possible even if lesions were stable at arthroscopy [7]. This finding possibly accounts for the subset of patients with stable-appearing OCD lesions that do not respond to nonoperative management and suggests that early operative intervention may be better suited for some children [7]. Therefore, the identification of advanced imaging correlates to histopathologic findings is important to triage patients to the appropriate treatment. However, to date, there is no precise correlation of histopathologic findings to the MRI appearance of a ju- W114 AJR:205, July 2015

2 Histopathology-MRI Correlation in Juvenile OCD venile OCD lesion. Toward this end, the purpose of our study was to correlate specimens of juvenile OCD lesions of the knee to MRI examinations to elucidate the histopathologic basis of characteristic imaging features. Materials and Methods This HIPAA-compliant study was approved by our institutional review board, and the requirement for informed consent was waived. Five children (three boys and two girls; age range, years) who underwent transarticular biopsy of juvenile OCD lesions of the knee were retrospectively included in this study. Patients who underwent biopsy but had closed growth plates were excluded. All patients underwent both an MRI evaluation of the knee and a distal femoral biopsy at the site of the juvenile OCD lesion. The biopsies were performed by a single pediatric orthopedic surgeon at the time of clinically indicated surgical intervention. Surgery was performed in these patients because their symptoms persisted or no radiologic signs of healing were present after nonoperative treatment or prior unsuccessful surgical treatment. The patients medical records were reviewed to determine whether a previous surgical intervention for the juvenile OCD lesion had been performed before the biopsy and, if so, what type of intervention was performed. Additionally, the surgical report from the time of biopsy was reviewed to determine if the OCD lesion was stable or unstable at arthroscopy. MRI All MRI examinations were performed from 2010 to 2012 on 1.5-T clinical systems. Three examinations were performed at our institution using an 8-channel transmit-receive extremity coil. Each examination included at least a coronal and sagittal fat-suppressed T2-weighted fast spinecho (FSE) sequence (TR range/te range, /46 71; slice thickness, 3 mm; FOV, 14 cm; matrix, ; scanning time, approximately 2 minutes 30 seconds) and a sagittal fat-suppressed T1-weighted 3D spoiled gradient-recalled (SPGR) sequence (TR/TE range, 50/6 7; flip angle, 60 ; slice thickness, 2 mm; FOV, 16 cm; matrix, ; scanning time, approximately 3 minutes 35 seconds). Two examinations were performed at outside institutions. One of these examinations included proton density weighted images both with and without fat suppression (TR range/te range, /25 39; slice thickness, 4 mm; FOV, 16 cm; matrix, ) in the coronal and sagittal planes, and the other included proton density weighted fat-suppressed images ( /29 30; slice thickness, 3.5 mm; FOV, 16 cm; matrix, ) in both the coronal and sagittal planes. The time interval between the preoperative MRI examination and biopsy was recorded. MRI Analysis The MRI examinations were reviewed by two radiologists: one pediatric musculoskeletal radiologist with 21 years experience and one musculoskeletal radiologist with 4 years experience. The images were assessed for the side (right or left) of the affected knee, the side of juvenile OCD condylar involvement (lateral or medial), and the location within the femoral condyle (central, medial, or lateral third in the coronal plane; anterior, central, or posterior third in the sagittal plane). The MRI characteristics of each lesion on T2-weighted images were recorded. When T2- weighted images were unavailable, proton density weighted fat-suppressed images were used. The MRI characteristics included round cystlike foci in subchondral bone, adjacent bone marrow edema pattern, thickening or thinning of unossified epiphyseal cartilage (thicker or thinner than adjacent unaffected epiphyseal cartilage), fissures of articular cartilage, laminar configuration of hyperintense signal superficial to deeper linear hypointense signal intensity at the cartilage-bone interface, and breaks in the subchondral bone plate as described in prior reports [8 10]. When the laminar configuration was present, whether the hyperintense signal was equal to or less than fluid was noted. Histopathology Each child underwent a transarticular core biopsy. After harvest, all cores were fixed in formalin. Each core was then decalcified gently using an ethylenediaminetetraacetic acid solution. Once decalcified, the specimens were embedded in paraffin wax, sectioned onto a slide, and stained with H and E. An anatomic pathologist familiar with bone and cartilage pathologic entities and with 11 years experience independently reviewed all histopathologic specimens and documented the full length (perpendicular to the articular surface) of each core biopsy specimen (bone and cartilage), its cartilage portion, and its cellular characteristics. Correlation of MRI and Histopathology To correlate the histopathologic findings with the MR images, digital photographs of the biopsy specimens were obtained with a reference scale in millimeters placed adjacent to each sample (Fig. 1). The site of the biopsy was retrospectively determined on the preoperative MRI examination using follow-up MRI or CT examinations in three patients, and surgical reports, intraoperative photographs, or videos in two patients (Fig. 2). Both Fig. 1 Example of low-power photomicrograph of biopsy specimen from juvenile osteochondritis dissecans lesion placed next to scale (in millimeters) that indicates corresponding length of specimen. Additionally, colorcoded bars are placed next to scale to indicate cartilage (C) in blue, presence of fibrovascular tissue (F) in green, and bone without fibrivascular tissue (B) in yellow. Lesion shown is from 12-year-old girl (patient 1). T1-weighted SPGR images and either proton density weighted or T2-weighted FSE images were used for correlation of the MRI findings to the histopathologic findings. Once the site of the biopsy was identified on the MRI examination, a reference line was drawn (in millimeters) on the corresponding image using the line measurement tool from the PACS software, and the annotated image was saved. Specimen photographs were overlaid on the reference image from the MRI that showed the biopsy site and the photograph s size to correspond to the line measurement previously drawn. Specimen photographs were then rendered semitransparent using software (Photoshop, Adobe Systems). The pa- Fig. 2 Healed juvenile osteochondritis dissecans lesion of medial femoral condyle after osteochondral autograft procedure in 12-year-old girl (patient 1). Incorporated plugs (arrows) allow precise delineation of prior biopsy site. AJR:205, July 2015 W115

3 Zbojniewicz et al. TABLE 1: Demographic and Surgical Data Patient No. Age (y) Sex Side Location of OCD Lesion 1 12 F Left Medial femoral condyle 2 13 F Right Medial femoral condyle 3 13 M Left Medial femoral condyle 4 13 M Right Medial femoral condyle 5 12 M Right Lateral femoral condyle No. of Biopsies ICRS Classification Surgeon s Description of Stability in Arthroscopy Report Fig year-old girl with juvenile osteochondritis dissecans (OCD) and no previous intervention (patient 2). Scale is in millimeters. Color-coded bars next to scale indicate cartilage (C) in blue, fibrovascular tissue (F) in green, and bone (B) in yellow. A, Low-power photomicrograph on left shows thickening of cartilage and cleft space near base of cartilage. High-power view on right at chondroosseous junction shows abundant fibrovascular tissue (white arrows) that insinuates between trabeculae of bone. There also is abnormal cleft space near base of cartilage with hyperchromatic edge changes (solid black arrows), which indicates that formation of this cleft is not artifact of laboratory specimen preparation. Additionally, there is chondrocyte cloning (dashed black arrows), which indicates cartilage injury but is nonspecific finding seen with other pathologic conditions affecting cartilage. B, Sagittal proton density weighted image shows juvenile OCD lesion at central and posterior aspect of medial femoral condyle with abnormal thickening of epiphyseal cartilage (asterisk) overlying lesion, cystlike focus (dashed arrow) that measures 5 mm anteriorly in subchondral bone, and laminar appearance with linear margin of hyperintense signal superficial to deeper margin of hypointense signal (solid arrows). Biopsy overlay was sized appropriately relative to reference line before placement over site of biopsy. Vertical bar = reference line for histopathologic overlay. C, Corresponding MR image with semitransparent histopathologic specimen photograph placed over site of biopsy. This overlay shows excellent correlation between locations of fibrovascular tissue, cleft space formation, and thickened cartilage on histopathology with hyperintense signal foci in subchondral bone, laminar configuration of signal intensity abnormality at cartilage-bone interface, and thickened cartilage on fatsuppressed proton density weighted image, respectively. A Prior Intervention? 2 OCD I Intact articular surface, stable Y (retrograde drilling and bone grafting) 1 OCD I Focal areas of softness at the N otherwise intact articular surface 1 OCD II March 2011: slightly unstable: Articular surface smooth and intact, no mobility, but one site unhealed; biopsy of unhealed site 1 OCD I Contour abnormality with slight motion at site of indentation, but lesion as a whole did not move up and down. No articular cartilage fissure 2 OCD I Articular surface collapsed and flattened, but no unstable cartilage Note OCD = osteochondritis dissecans, ICRS = International Cartilage Repair Society, Y = yes, N = no, NA = not applicable. Y (retrograde drilling, bone grafting, and transarticular fixation) N Y (retrograde drilling and bone grafting) Time Between Biopsy and Previous Intervention 14 mo NA 4 mo NA 4 y B C W116 AJR:205, July 2015

4 Histopathology-MRI Correlation in Juvenile OCD thologist and a single radiologist chose, by consensus, the most appropriate position of the semitransparent biopsy images with respect to the site of biopsy on the MRI oriented to correspond with the contour of the affected femoral condyle. We subjectively noted which imaging sequence allowed best characterization of the different histopathologic findings. Results Seven biopsy specimens were obtained; in two children, the OCD lesion was biopsied in two locations (Table 1). Two patients had undergone no previous surgical intervention before biopsy; however, in one of these patients, the biopsy was obtained immediately after curettage, microfracture, and A bone graft placement. Three patients had undergone a previous surgical intervention, which consisted of retrograde drilling and bone grafting, that was performed between 4 months and 4 years before biopsy. All juvenile OCD lesions were largely stable at arthroscopy; the specific comments stated by the surgeon are reported in Table 1. C D Fig year-old girl with juvenile osteochondritis dissecans (OCD) previously treated with retrograde drilling and bone grafting (patient 1). In A and D, scale is in millimeters, and color-coded bars next to scale indicate cartilage (C) in blue, fibrovascular tissue (F) in green, and bone (B) in yellow. A, Low-power photomicrograph on left shows thickened cartilage with abundant fibrovascular tissue insinuating between bone trabeculae. Large area of fibrovascular tissue is round in configuration, resembling cyst (arrowhead). High-power view on right shows abundant fibrovascular tissue that insinuates between trabeculae (solid black arrows), hyperchromatic edge changes (white arrows) at site of cleft space, and chondrocyte cloning (dashed black arrows). B and C, Sagittal T2-weighted fat-suppressed (B) and 3D spoiled gradient-recalled (SPGR) (C) images show areas of hypointense signal (solid arrows) deep within subchondral bone, which are result of prior retrograde drilling and bone grafting. Abnormally thickened cartilage (asterisk, C) at juvenile OCD site is more conspicuous on SPGR image. Areas of fibrovascular change are represented by areas of increased signal intensity within subchondral bone (dashed arrows, C), which are also more defined on SPGR image than on T2-weighted fat-suppressed image. Vertical bar = reference line for histopathologic overlay. D, MR image with histopathologic overlay at site of biopsy shows excellent correlation between areas of fibrovascular tissue within subchondral bone as defined areas of increased signal intensity on SPGR image. Vertical bar = reference line for histopathologic overlay. B AJR:205, July 2015 W117

5 Zbojniewicz et al. To compare the level of severity of the OCD lesions with the findings of prior studies, the International Cartilage Repair Society (ICRS) classification was applied to each lesion [11]. All cases in our study corresponded to either ICRS OCD classification I or II lesions. These results are summarized in Table 1. The time interval between the preoperative MRI examination and biopsy ranged from 1 to 5 months. Four biopsies were performed less than 2 months from the time of the MRI examination. Of the three patients who had previous surgical intervention, two had undergone a repeat MRI examination before biopsy. In the one child with a prior surgical interven- C Fig year-old boy with juvenile osteochondritis dissecans and prior retrograde drilling, bone grafting, and transarticular fixation (patient 3). In A and D, scale is in millimeters, and color-coded bars next to scale indicate cartilage (C) in blue, fibrovascular tissue (F) in green, and bone (B) in yellow. A, Low-power photomicrograph on left shows thickened cartilage and abundant fibrovascular tissue (arrowhead). High-power image on right shows that fibrovascular tissue (solid arrows) is immediately adjacent to cartilage with no bone present to anchor cartilage. There are multiple foci of chondrocyte cloning (dashed arrows). B and C, Images obtained at same location highlight that distinction between thickened cartilage (asterisk, B) and underlying bone is clearer on spoiled gradient-recalled (SPGR) image (B) than on T2-weighted fat-suppressed image (C). In addition, SPGR image allows easier differentiation of diffuse pattern of fibrovascular tissue in subchondral bone (solid arrows, B) from subchondral bone marrow edema pattern. Both blend together on T2-weighted image. Also, hypointense bone or calcification (dashed arrow, B) can be differentiated from relatively hyperintense cartilage signal on SPGR image. However, multiple breaks in subchondral bone (arrows, C) are better seen on T2-weighted image. Vertical bar = reference line for histopathologic overlay. D, MR image with histopathologic overlay at site of biopsy shows excellent correlation between fibrovascular tissue, which extends deep within subchondral bone, and diffuse pattern of increased signal intensity on SPGR image. A D B W118 AJR:205, July 2015

6 Histopathology-MRI Correlation in Juvenile OCD tion but without a repeat MRI examination, 4 months elapsed the initial surgical intervention and biopsy performed at time of the salvage procedure. MRI Analysis Three OCD lesions were in the right knee and two, in the left knee. Four lesions involved the lateral third of the medial femoral condyle in the coronal plane (two were central posterior and two were central in the sagittal plane), and one lesion extended between the central and medial thirds of the lateral femoral condyle in the coronal plane (centrally in the sagittal plane). All lesions showed cystlike foci in the subchondral bone, bone marrow edema pattern on proton density or T2-weighted images, and thickening of unossified epiphyseal cartilage (Figs. 3 5). No cases showed an articular cartilage fissure. Four of five cases had a laminar configuration of hyperintense signal superficial to deeper linear hypointense signal intensity at the cartilage-bone interface on proton density or T2-weighted images with or without fat suppression. However, in all four cases, the hyperintense signal intensity layer was equal to fluid signal intensity only in small foci of the entire hyperintense area. Two cases showed multiple breaks in the subchondral bone; one was associated with the previously described area of laminar configuration and one was not. A summary of the MRI findings is shown in Table 2. Histopathology The total length of the core biopsy specimen ranged from 11 to 17 mm, whereas the cartilage length in isolation ranged from 1 to 7 mm. The cellular characteristics of the juvenile OCD lesions seen in the core biopsy specimens are shown in Table 3. Chondrocyte cloning was seen in three of five patients. Fibrovascular tissue was found in all patients and was present at TABLE 3: Histopathologic Findings Patient No. Biopsy Core Length (mm) Length of Cartilage (mm) TABLE 2: MRI Findings Patient No. Cystlike Foci Bone Marrow Edema Pattern the interface between cartilage and bone with extension into adjacent subchondral bone and ranged in thickness from 1 to 10 mm. Additionally, there were cleft spaces in the region of the cartilage-bone interface, typically in the deepest region of the cartilage and often leaving only scant microscopic fragments of thin and degenerated cartilage attached to the bone. Hyperchromatic changes were seen at the edges of the cartilage in the region of these cleft spaces. Focal bone necrosis, trabecular remodeling, and inflammation were infrequently seen. Histopathologic findings were similar in all specimens regardless of whether the patient had undergone a prior intervention. Correlation of MRI and Histopathologic Findings Precise correlation of the MRI appearance to the histopathologic overlays consistently was found. Abnormally thickened cartilage corresponded to cartilage signal intensity on proton density or T2-weighted FSE images and T1-weighted SPGR images in all cases. Similarly, the location and morphology of the fibrovascular tissue correlated with the cystlike foci seen within the subchondral bone on MRI in all cases (Figs. 3 7). The location of histopathologic cleft formation directly corresponded to the laminar configuration seen at the cartilage-bone interface on MRI. Chondrocyte Cloning Cartilage Thickening Cleft Space Near Cartilage-Bone Interface Articular Cartilage Fissure [8] Laminar Configuration ( Oreo Cookie Appearance) Discussion Juvenile OCD is an important cause of disability in young active children and may lead to premature osteoarthritis and permanent disability. Most OCD histopathologic evaluations have been performed on late-stage OCD lesions in skeletally mature patients, often through examination of arthroscopically retrieved loose bodies [12 17]. Recently, histopathologic findings of core biopsy specimens from stable in situ juvenile OCD lesions have been reported [7]. However, these prior studies have not offered direct correlation of the histologic specimens with the MRI features, which is a necessary step to contribute to the prospective evaluation and treatment of these patients. The MRI findings that were evaluated in our study group are those commonly encountered on examinations of skeletally immature patients diagnosed with juvenile OCD namely, cystlike foci, adjacent bone marrow edema pattern, overlying epiphyseal cartilage thickening, a laminar signal intensity pattern at the cartilage-bone interface, and multiple breaks in the subchondral bone [8 10]. Abundant fibrovascular tissue at the cartilage-bone interface and within the adjacent trabeculae of subchondral bone was seen in all patients of our study. Uozumi et al. [17] reported similar-appearing tissue in all patients in their Fibrovascular Tissue Multiple Breaks in Subchondral Bone [9] 1 Y Y Y N Y N 2 Y Y Y N Y N 3 Y Y Y N Y Y 4 Y Y Y N N Y 5 Y Y Y N Y N Note Y = yes, N = no. Length of Fibrovascular Tissue (mm) 1 (site 1) 16 5 N Y Y (site 2) 17 5 Y Y Y Y Y Y Y Y Y N Y Y (site 1) 16 3 N N N NA 5 (site 2) 15 4 N Y Y 4 7 Note N = no, Y = yes, NA = not applicable. AJR:205, July 2015 W119

7 Zbojniewicz et al. C Fig year-old boy with juvenile osteochondritis dissecans (OCD) and prior retrograde drilling and bone grafting (patient 5). In A and D, scale is in millimeters, and color-coded bars next to scale indicate cartilage (C) in blue, bone (B) in yellow. A, Low-power photomicrograph on left shows thin zone of cartilage (arrowhead) at base of OCD lesion. High-power view on right shows no fibrovascular tissue; instead, marrow space contains predominantly adipose tissue (arrows) with few lymphocytes (not shown) that may be inflammatory or normal developing blood cells. B, Sagittal T2-weighted fat-suppressed image shows findings of OCD at lateral femoral condyle. There is abnormal contour to articular surface (solid arrow) with bone marrow edema pattern present at anterior aspect of lesion (dashed arrow). Vertical bar = reference line for histopathologic overlay. C, Spoiled gradient-recalled (SPGR) image obtained at same level as B shows lack of hyperintense signal within subchondral bone anteriorly; this finding indicates that finding on T2-weighted image (B) is bone marrow edema pattern. This SPGR image also better differentiates bone from more hyperintense cartilage posteriorly (black arrow) and shows in better detail small foci of fibrovascular tissue within bone posteriorly (white arrows). Hypointense signal (arrowhead) is evidence of prior retrograde drilling. Vertical bar = reference line for histopathologic overlay. D, Corresponding MR image with histopathologic overlay at site of biopsy within anterior part of juvenile OCD lesion shows that absence of fibrovascular tissue at anterior aspect of lesion corresponds to area of bone marrow edema pattern on T2-weighted fat-suppressed image (B). There are no corresponding hyperintense signal foci in subchondral bone on SPGR image (C) to indicate presence of fibrovascular tissue. Vertical bar = reference line for histopathologic overlay. A study evaluating in situ OCD fragments. Yonetani et al. [7] also observed this finding in one of the two histopathologic patterns they described in patients with juvenile OCD. A direct imaging correlation for the fibrovascular tissue was present on all MRI examinations in our study. On MRI, the fibrovascular tissue appeared in two patterns: One pattern consisted of round cystlike foci, and the other consisted of a more linear or diffuse pattern of hyperintense signal intensity on T2-weighted and gradient-echo T1-weighted images. Delineation of the pattern of signal abnormality and differentiation of the hyperintense signal associated with fibrovascular tissue from that related to bone marrow edema pattern on proton density or T2-weighted images were, at times, difficult. However, D B W120 AJR:205, July 2015

8 Histopathology-MRI Correlation in Juvenile OCD on fat-suppressed T1-weighted gradient-echo images, the increased signal intensity in the areas of fibrovascular tissue contrasted to the hypointense signal of bone and fluid and allowed improved distinction (Figs. 4 7). The identification and accurate characterization of areas of fibrovascular tissue are important because prior studies suggest that the size of a single cystlike focus on MRI can be associated with arthroscopic instability and ultimately with patient prognosis [4, 8, 9]. The ability to accurately separate the bone marrow edema pattern from the pattern of fibrovascular tissue and to determine the extent of fibrovascular tissue on MRI may have prognostic and treatment implications. C Fig year-old boy with juvenile osteochondritis dissecans (OCD) and prior retrograde drilling and bone grafting (patient 5). These images are of same patient as Figure 6, but MRI slice shown here is more laterally located within lateral femoral condyle. In A and D, scale is in millimeters, and color-coded bars next to scale indicate cartilage (C) in blue, fibrovascular tissue (F) in green, and bone (B) in yellow. A, Photomicrograph shows abundant fibrovascular tissue (arrows), which is seen to better advantage on high-power view. B, Sagittal T2-weighted fat-suppressed image shows hyperintense signal within subchondral bone; it is difficult to distinguish between fibrovascular tissue and bone marrow edema pattern (solid arrow). Differentiation of bone from cartilage (dashed arrow) is also difficult. Vertical bar = reference line for histopathologic overlay. C, Sagittal spoiled gradient-recalled (SPGR) image distinguishes between bone marrow edema pattern and fibrovascular tissue, with latter being represented by discrete areas of hyperintense signal within subchondral bone (solid arrows). Thickened cartilage (dashed arrow) at site of OCD lesion is also more conspicuous. Decreased signal intensity (arrowhead) corresponds to site of prior retrograde drilling. Vertical bar = reference line for histopathologic overlay. D, Corresponding SPGR image with histopathologic overlay shows thick cartilage over juvenile OCD lesion and subchondral foci of hyperintense signal that correspond to fibrovascular tissue on biopsy specimen. Vertical bar = reference line for histopathologic overlay. A Fibrovascular tissue at the interface between cartilage and bone and within subchondral bone resembles findings found at the sites of fracture nonunion [17 19]. As suggested by Uozumi et al. [17], we believe that this tissue may require surgical débridement or fixation to incite healing, similar to treatment of fracture nonunion. Whether a patient with exten- D B AJR:205, July 2015 W121

9 Zbojniewicz et al. sive fibrovascular tissue (Fig. 5) can heal with nonoperative management alone is unknown. Additional studies are required to determine if there is a threshold for the amount of fibrovascular tissue that may heal with conservative treatment alone and to determine the amount that would require surgical intervention. No definite histopathologic correlate was seen to correspond to the bone marrow edema pattern detected either deeper to the areas of fibrovascular tissue or isolated to subchondral bone. This finding is expected because previous studies that evaluated the histopathologic correlation of bone marrow edema pattern on MRI in patients with osteoarthritis at the knee showed that most of these areas represented normal marrow spaces with minor components of bone marrow edema, necrosis, fibrosis, and bleeding [20]. Epiphyseal cartilage thickening and chondrocyte cloning on histopathology seen in our study population also have been described by Yonetani et al. [7]. Both their study [7] and our study included patients with early-stage disease (ICRS OCD grade I and II lesions). The epiphyseal cartilage thickening correlates to a prior study in the literature that describes this finding as a characteristic MRI finding in patients with juvenile OCD lesions [10]. The MRI correlate of thickened epiphyseal cartilage was seen more readily on the T1-weighted gradient-echo fat-suppressed images than on the T2-weighted fat-suppressed images (Figs. 4 7). Chondrocyte cloning is a nonspecific indicator of cartilage injury that is frequently found in osteoarthritis [21, 22]. The abnormal cleft spaces in the region of the cartilage-bone interface that we observed in our series indicate an abnormal structural relationship between cartilage and bone and correlate with the deep instability reported by Yonetani et al. [7]. Hyperchromatic edge changes found at these clefts indicate that these abnormal spaces are an in vivo phenomenon rather than one that results from histologic slide preparation. The laminar appearance of hyperintense signal on T2-weighted images with deeper and more superficial margins of hypointense signal, which we liken to the striated appearance of an Oreo cookie, is found at the same location as the cleft spaces seen histopathologically and may represent early separation or cleft space formation at the deep base of an OCD lesion that is, before any articular surface fissure or demarcation develops (Fig. 3). A similar appearance has been described by Kijowski et al. [9] as being specific for instability; however, the criteria proposed by those authors [9] required that the hyperintense middle layer be of signal intensity equal to that of fluid [9]. In all four children who showed this laminar imaging pattern in our study, the hyperintense layer was predominantly hypointense to fluid signal intensity. Given the relative articular surface stability of the lesions in our subset of patients, this finding may reflect an earlier stage of separation. The length of this Oreo cookie appearance may also contribute to an increased likelihood of gross instability at surgery. In this manner, this deep surface instability may eventually lead to surface cartilage fissuring and ultimately lead to detachment of the fragment. An analogous scenario is a pothole in a road that commonly develops after water erodes away the sand and gravel bed that supported the surface asphalt. A longitudinal study may determine if patients with these potentially high-risk imaging features fare worse with nonoperative treatment than those without these findings. Multiple breaks in the subchondral bone described previously on MRI as specific for instability when found in conjunction with the aforementioned laminar appearance may reflect a greater degree of disorganization at the cartilage-bone interface where there is extensive replacement by fibrovascular tissue (Fig. 5). Bone necrosis and inflammation were not prominent features in the specimens in our study; although these findings are potentially a cause of increased signal intensity within bone marrow, as described by Zanetti et al. [20], we could not confirm this to be the case in our study. The limitations for our study include the small number of biopsy specimens. However, despite this limited cohort, the histopathologic and radiologic imaging findings were remarkably consistent. Although an effort was made to determine the precise origin site of each biopsy specimen with respect to the MRI location, there was no method to evaluate precision, particularly when no follow-up imaging study was available to determine the location and orientation of the biopsy specimen. One patient (patient 3) did not undergo repeat MRI between the time of the first surgical intervention and the salvage treatment when the biopsy was obtained. Therefore, the findings on the initial MRI examination may not reflect the appearance at the time of the biopsy. In another patient (patient 4), the biopsy was performed immediately after bone grafting; however, the presence of fibrovascular tissue insinuating between bone trabeculae could not have been a result of surgical intervention. Two patients had outside MRI examinations that consisted of proton density weighted sequences rather than T2-weighted sequences and had no SPGR sequence, but the findings were still clearly identified. Finally, several of our patients had a previous surgical intervention at the juvenile OCD lesion sites. However, the histopathologic findings in patients without and with prior treatment with retrograde drilling and bone graft placement were identical. This similarity between untreated patients and those treated previously might be because the biopsy was not taken from the same site as the lesional drilling and bone grafting or that the biopsy was obtained in an area of recurrent lesion. In conclusion, we found direct correlations between the histopathologic findings and the MRI features in patients with juvenile OCD. Future studies are needed to correlate these MRI features with juvenile OCD healing success rates. These histopathologic and imaging findings may impact treatment decisions, healing rates, and ultimately prognosis in patients with juvenile OCD. References 1. Heywood CS, Benke MT, Brindle K, Fine KM. Correlation of magnetic resonance imaging to arthroscopic findings of stability in juvenile osteochondritis dissecans. Arthroscopy 2011; 27: Cahill BR, Phillips MR, Navarro R. The results of conservative management of juvenile osteochondritis dissecans using joint scintigraphy: a prospective study. Am J Sports Med 1989; 17: ; discussion, Hefti F, Beguiristain J, Krauspe R, et al. Osteochondritis dissecans: a multicenter study of the European Pediatric Orthopedic Society. J Pediatr Orthop B 1999; 8: Krause M, Hapfelmeier A, Moller M, Amling M, Bohndorf K, Meenen NM. Healing predictors of stable juvenile osteochondritis dissecans knee lesions after 6 and 12 months of nonoperative treatment. Am J Sports Med 2013; 41: Twyman RS, Desai K, Aichroth PM. Osteochondritis dissecans of the knee: a long-term study. J Bone Joint Surg Br 1991; 73: Wall EJ, Vourazeris J, Myer GD, et al. The healing potential of stable juvenile osteochondritis dissecans knee lesions. J Bone Joint Surg Am 2008; 90: Yonetani Y, Nakamura N, Natsuume T, Shiozaki Y, Tanaka Y, Horibe S. Histological evaluation of juvenile osteochondritis dissecans of the knee: a case series. Knee Surg Sports Traumatol Arthrosc 2010; 18: W122 AJR:205, July 2015

10 Histopathology-MRI Correlation in Juvenile OCD 8. De Smet AA, Ilahi OA, Graf BK. Reassessment of sected cartilage fragments in human osteo- knee. Am J Sports Med 2009; 37: the MR criteria for stability of osteochondritis chondritis dissecans of the femoral condyle. Arch 18. Mills LA, Simpson AH. In vivo models of bone dissecans in the knee and ankle. Skeletal Radiol Orthop Trauma Surg 2006; 126: repair. J Bone Joint Surg Br 2012; 94: ; 25: Barrie HJ. Hypertrophy and laminar calcification of 19. Volpon JB. Nonunion using a canine model. Arch 9. Kijowski R, Blankenbaker DG, Shinki K, Fine JP, cartilage in loose bodies as probable evidence of an Orthop Trauma Surg 1994; 113: Graf BK, De Smet AA. Juvenile versus adult osteochondritis dissecans of the knee: appropriate MR imaging criteria for instability. Radiology 2008; 248: Laor T, Zbojniewicz AM, Eismann EA, Wall EJ. Juvenile osteochondritis dissecans: is it a growth disturbance of the secondary physis of the epiphysis? AJR 2012; 199: International Cartilage Repair Society website. ICRS cartilage injury evaluation package. www. cartilage.org/_files/contentmanagement/icrs_ evaluation.pdf. Accessed June 29, Aurich M, Anders J, Trommer T, et al. Histological and cell biological characterization of dis- ossification abnormality. J Pathol 1980; 132: Koch S, Kampen WU, Laprell H. Cartilage and bone morphology in osteochondritis dissecans. Knee Surg Sports Traumatol Arthrosc 1997; 5: Lindén B, Telhag H. Osteochondritis dissecans: a histologic and autoradiographic study in man. Acta Orthop Scand 1977; 48: Milgram JW. Radiological and pathological manifestations of osteochondritis dissecans of the distal femur: a study of 50 cases. Radiology 1978; 126: Uozumi H, Sugita T, Aizawa T, Takahashi A, Ohnuma M, Itoi E. Histologic findings and possible causes of osteochondritis dissecans of the 20. Zanetti M, Bruder E, Romero J, Hodler J. Bone marrow edema pattern in osteoarthritic knees: correlation between MR imaging and histologic findings. Radiology 2000; 215: Okada A, Mochizuki S, Yatabe T, et al. ADAM- 12 (meltrin alpha) is involved in chondrocyte proliferation via cleavage of insulin-like growth factor binding protein 5 in osteoarthritic cartilage. Arthritis Rheum 2008; 58: Quintavalla J, Kumar C, Daouti S, Slosberg E, Uziel-Fusi S. Chondrocyte cluster formation in agarose cultures as a functional assay to identify genes expressed in osteoarthritis. J Cell Physiol 2005; 204: AJR:205, July 2015 W123