Arthroscopy of the hip has been used with increasing frequency for the diagnosis and treatment of intra-articular hip disorders.1,3,4,7,9,10-14,18,21 Several studies have identified various etiologies of persistent hip pain including acetabular labral tears, loose bodies, mild osteoarthritis, osteochondral defects, chondral flaps, and synovitis.1,3,7,9,10,11,14,18,21 These hip disorders can be associated with normal radiographs or minor radiographic abnormalities and therefore can be difficult to diagnose.1,2,3,8,16,19,21,28,29 Hip arthroscopy is an excellent diagnostic tool, but its relative cost and invasive nature have prompted the search for alternative methods to establish a diagnosis.12,13,14,18,19,26 Furthermore, obtaining an accurate preoperative diagnosis enables the surgeon to provide individual patients with reasonable prognostic expectations and to optimize selection of arthroscopic versus open surgical techniques.
Conventional magnetic resonance imaging (MRI) is a common, noninvasive imaging technique for evaluating bone and soft tissue pathologies; however it is limited in detecting important intra-articular entities including loose bodies, labral tears, and cartilage defects.1,5,8,9 Traditional arthrograms similarly have notable shortcomings, particularly with identifying soft tissue pathology.9,10,23 The combination of these imaging modalities in magnetic resonance arthrography (MRA) has improved the capability to detect a spectrum of intra-articular hip pathologies, specifically allowing improved detection of acetabular labral tears.5,23 Many orthopaedic surgeons and musculoskeletal radiologists have accepted MRA as the best diagnostic tool for evaluating occult hip disorders. Reports in the radiology literature have also suggested that MRA has excellent sensitivity and specificity with regard to the evaluation of intra-articular pathology.27
At our institution, hip arthroscopy has been offered to patients with a clinical diagnosis of acetabular labral tear after failure to improve with an appropriate trial of conservative treatment consisting of activity modification, nonsteroidal anti-inflammatory medications, and physical therapy. Magnetic resonance arthrography has been obtained for all patients with signs and symptoms suggestive of an acetabular labral tear before proceeding with hip arthroscopy. The purpose of obtaining the MR arthrogram was twofold: to confirm the clinical diagnosis of acetabular labral tear; and to exclude other occult hip pathologies (such as femoral neck stress fracture, osteonecrosis, neoplasm) as potential sources of hip pain.
Based on the experience of the senior author, we hypothesized that MRA would underestimate the incidence of acetabular labral pathology and articular cartilage defects identified during arthroscopy. To address this hypothesis, we reviewed a large series of patients who were evaluated with MRA before having hip arthroscopy. A retrospective review of MRA findings was compared with surgeon-documented intra-articular hip pathology to assess the effectiveness of MRA in identifying the presence, size and location of acetabular labral tears and in recognizing associated articular cartilage abnormalities.
MATERIALS AND METHODS
One hundred one consecutive patients (102 hips) had hip arthroscopy between September 1999 and August 2003 to evaluate and treat persistent hip pain consistent with the clinical diagnosis of an acetabular labral tear. The clinical diagnosis of a labral tear was made on the basis of anterior inguinal pain, positive provocative testing for acetabular labral tear (impingement test),17 absence of external tendon pathology, and minimal degenerative or dysplastic findings on plain radiographs. Typically in patients with acetabular labral tears, pain, activity limitation and physical examination findings are disproportionate with findings on plain radiographs. Diagnostic injection can be done for patients with either equivocal findings on physical examination or concurrent extra-articular pathology, but is not used routinely in our practice as a determinant for surgical intervention. In general, patients with plain radiograph findings of considerable acetabular dysplasia (center edge angle < 12°), moderate osteoarthritis ( > 50% joint space loss), or other major hip deformity were not offered arthroscopic treatment because other surgical procedures were though to be more appropriate in these clinical situations.
Right and left hips were involved with approximately the same incidence, with a higher percentage of women (69.6%) noted. Average time from onset of symptoms until surgical referral was 21.6 months (range, 3-120 months). Retrospective review of preoperative radiographs from this patient population identified two patients with mild dysplasia (CEA 13 and 14°), 14 patients with femoral head-neck offset smaller than 9 mm, five patients with femoral head-neck offset smaller than 7 mm, and seven patients with acetabular retroversion. Radiographic osteoarthritis was classified as Tonnis 0 for 76 patients, Tonnis 1 for 20 patients and Tonnis 2 for six patients. Fifty-three patients (52%) had normal radiographs.
Preoperative MRA was done in all patients to assess for an acetabular labral tear and concomitant intra-articular pathology. All procedures were done by one surgeon (JCC) using the same operative technique. Patients having hip arthroscopy for other indications were excluded from this study. Patients having simultaneous open surgery (osteotomy or femoral head-neck junction debridement) also were excluded. This study population (Table 1) was reviewed in accordance with an IRB approved protocol.
Magnetic resonance arthrography was done at our institution in 96 (95 patients) of 102 hips (101 patients). Six MR arthrograms were obtained at other hospitals and were not repeated because no quality issues were identified. At our institution, one of four musculoskeletal radiologists completed the preoperative arthrogram with an injection of 15 cc of 1:200 dilution of gadodiamide (Omniscan, Amersham Health, Princeton, NJ) in sterile saline and iodinated contrast. Magnetic resonance imaging was done using a 1.5-T magnet (GE Medical Systems, Milwaukee WI; or Siemens Medical Systems, Isssaguah, WA) and multicoil array. High-resolution T1-weighted images of the affected hip were obtained in transverse, sagittal, and coronal planes. Fat suppression was applied in at least two imaging planes and T2-weighted, fast-spin echo images were also obtained in one or two of the imaging planes.
The radiographic diagnosis of acetabular labral tear was made for findings of abnormal labral shape, detachment of labrum from the underlying acetabular rim, abnormal signal within the labrum on T2-weighted image, presence of gadolinium within the labrum, or presence of an associated labral cyst. Confirmation was noted by the presence of any of these findings in adjacent frames or other imaging planes. Radiologists also evaluated for coexisting pathology including osteoarthritis, chondromalacia, articular cartilage flaps, osteochondritis dissecans, intra-articular loose bodies, or soft tissue abnormalities. All MRA studies were reviewed by subspecialty radiologists with experience in musculoskeletal MRI interpretation and the results were available to the operative surgeon at the time of surgery.
Hip arthroscopy was done under general anesthesia with the patient positioned supine on a fracture table. The surgically treated hip was placed in neutral flexion or extension and slight abduction. Approximately 6-8 mm of joint distraction was confirmed by C-arm fluoroscopy and maintained during the procedure. Observation of intra-articular structures was done using 30° and 70° (4.5-mm cannula) arthroscopes (Smith and Nephew, Andover, MA). A systematic evaluation of the hip was done with prospective documentation recorded on photographic images and in the operative report. Labral tears were described by location and appearance. Tears were identified as anterior, superolateral, or posterior. 20 Some tears were observed in more than one quadrant and were referenced as multiregional. Common tear patterns were described as longitudinal tears (isolated avulsion) or complex degenerative tears (hypertrophy and fraying with or without avulsion). Chondromalacia, when present, was described according to the classification reported by Outerbridge. 22
The combination of labral tear and articular pathology led to prognostic staging as described by McCarthy et al20 (Table 2). Stage 0 refers to a labral contusion with associated synovitis. Stage 1 represents a labral-free margin tear with intact acetabular and femoral articular cartilage. Stage 2 labral tears are associated with articular cartilage abnormalities confined to the femoral head. Stage 3A tears are associated with acetabular cartilage abnormalities smaller than 1 cm. Stage 3B tears are associated with acetabular cartilage abnormalities larger than 1 cm. Stage 4 disease signifies diffuse osteoarthritis with an associated labral tear. In this study, patients with low-grade chondromalacia involving the femoral head and associated articular cartilage defects were included with Stage 3 lesions and classified on the basis of the size of the acetabular articular defect.
An independent retrospective review of radiographs, MRA reports, operative reports, and arthroscopic images was done by surgeons not directly involved in the care of the study patients (JK, MP). Size of tear and articular cartilage abnormalities were concluded from the senior author’s description in the operative report and from review of arthroscopic images for all patients. The size of the arthroscopy probe (4 mm) was used as a reference for assessing the tear length. Tears were delineated on the basis of size (< 10 mm, between 10 and 20 mm, or > 20 mm).
Our retrospective review of MRA reports was remarkable for the identification of acetabular labral pathology in 71 of 102 hips (70 patients, 69.6%). Location of the labral tear was described as anterior in 24 hips (29.6%), superolateral in 33 hips (40.7%), posterior in 4 hips (4.9%), and multiregional in 10 hips (14.1%). Tear size was not consistently quantified on the basis of MRA.
Magnetic-resonance-arthrography-detected acetabular articular abnormalities were observed in 24 hips (23.5%) in 24 patients and included moderate osteoarthritis (8 hips; 8 patients), mild osteoarthritis (6 hips; 6 patients), chondral defect (3 hips; 3 patients), and chondral flap tear (2 hips; 2 patients). Other findings noted on the MRA included labral cyst (3 hips; 3 patients), acetabular degenerative cyst (1 hip; 1 patient), loose body (1 hip; 1 patient), ligamentum teres rupture (1 hip; 1 patient), capsular defect (1 hip; 1 patient), soft tissue calcification (1 hip; 1 patient), and synovitis (1 hip; 1 patient). Femoral abnormalities were observed on MRA images in 10 hips in 10 patients (9.8%), and included advanced osteoarthritis (3 hips; 3 patients), mild osteoarthritis (2 hips; 2 patients), degenerative cyst (3 hips; 3 patients), femoral head deformity secondary to old Perthes disease (1 hip; 1 patient), and osteochondritis dissecans (1 hip; 1 patient).
Our retrospective review of hip arthroscopy reports and images identified 93 hips in 92 patients with labral tears (91.2%). Tears were noted as anterior in 56 hips (56 patients; 60.2%), superolateral in 15 hips (15 patients; 16.1%), posterior in 5 hips (5 patients; 5.4%), and multiregional in 17 hips 17 patients; 18.4%). The tear pattern was complex or degenerative in 49 patients (49 hips; 52.7%) and horizontal or an avulsion pattern in 44 patients (44 hips; 47.3%). Tear size was smaller than 10 mm in nine hips (9 patients; 9.7%), 10-20 mm in 62 hips (62 patients; 66.7%) and greater than 2 cm in 22 hips (22 patients; 23.7%). Nine hips (9 patients) did not have acetabular labral tears at arthroscopy. Five of these patients (55.6%) had labral tears predicted by MR arthrogram. In spite of the absence of labral pathology, seven of these nine patients had conditions identified at arthroscopy that had potential clinical significance. Six hips had synovitis: three associated with a chondral defect; one associated with a hypertrophic labrum; and 2 hips had synovitis without an associated articular abnormality. One patient had an unstable chondral flap without synovitis.
During arthroscopy, acetabular cartilage abnormalities were noted in 46 hips in 45 patients (45.1%) and femoral cartilage abnormalities were observed in 21 hips in 20 patients (20.6%). Intra-articular findings at arthroscopy included Grade 1 or 2 acetabular chondromalacia (24 hips; 24 patients), Grade 3 or 4 acetabular chondromalacia (27 hips; 26 patients), acetabular chondral flap tears (13 hips; 13 patients), Grade 1 or 2 femoral chondromalacia (five hips; five patients), Grade 3 or 4 femoral chondromalacia (18 hips; 18 patients), intra-articular loose bodies (3 hips; 3 patients), and ligamentum teres rupture (1 hip; 1 patient). Of the Grades 3 and 4 acetabular cartilage defects, 12 were greater than 1 cm in diameter and 15 were smaller than 1 cm in diameter. Although the clinical significance is not clear for many of the articular cartilage abnormalities, they may provide prognostic significance (Table 2).20 Only two of 102 hips (2%) had no appreciable abnormality noted during arthroscopic evaluation.
Comparison of the MRA and hip arthroscopy findings noted above showed MRA to have a sensitivity of 71%, positive predictive value of 93%, negative predictive value of 12.9%, and accuracy of 69% with respect to the identification of labral pathology (Table 3). The specificity of MRA for labral pathology (calculated at 44% based on this study population) cannot be accurately determined from this study because hip arthroscopy was not done in the setting of a normal labrum or clinically asymptomatic hip. The ability of the MRA to accurately detect an intact labrum is not reliably assessed. In addition, 27 hips in 27 patients (27%) were identified with arthroscopically determined acetabular labral tears in the setting of an MR arthrogram that was interpreted as normal. Six of these patients (22.2%) had labral tears smaller than 10 mm, 16 patients (59.3%) had labral tears between 10 and 20 mm, and 5 patients (18.5%) had tears greater than 20 mm. Fourteen of the tears (51.9%) were degenerative tears whereas the other 13 tears (48.1%) were horizontal intersubstance or avulsion tears.
Comparison of MRA with arthroscopy with respect to articular cartilage abnormalities highlights the underrecognition of articular cartilage abnormalities in 32 hips (31.7%). Potentially notable articular cartilage abnormalities (full thickness defects greater than 1 cm diameter) were underestimated in 9 hips (9 patients; 8.9%), and focal cartilage defects (< 1 cm) and low-grade (Grades 1 or 2) cartilage lesions were missed in an additional 23 hips in 23 patients (22.8%). Complete agreement with regard to the presence or absence of articular cartilage abnormality was noted in 63 hips (63 patients; 62.4%), and the degree of degenerative articular change was overestimated in six patients (5.9%). Overall, when compared with arthroscopically documented findings in the assessment of articular cartilage pathology, MRA showed a sensitivity of 47%, a specificity of 89%, a positive predictive value of 84%, a negative predictive value of 59%, and an accuracy of 67% (Table 4).
Although the ability to accurately define articular hip pathology before proceeding with hip arthroscopy is desirable, the experience of the senior author in this study suggested that labral and articular cartilage conditions are underrecognized by MRA. To our knowledge, this is currently the largest series comparing gadolinium MRA of the hip with arthroscopic findings. Previous studies in the radiologic literature have evaluated the effectiveness of MRA in detecting acetabular labral tears with smaller sample sizes5,6,23,24,27 (Table 5).
Petersilge et al23 used MRA in 24 hips, 10 of which had subsequent surgery (six hips had arthroscopy, three had open, and one had combined). The senior author’s interpretation of the MR arthrograms was correct in 8 of 10 patients. Czerny et al5 found MRA was 90% sensitive and 91% accurate in diagnosing labral tears confirmed by open arthrotomy. In this study, three musculoskeletal radiologists staged labral tears identified on MRA by consensus. They correctly identified 20 of 22 labral tears. Thirty-five patients in the study group were treated nonoperatively based on MR arthrogram results and lack of severe symptoms. In a subsequent study, Czerny et al6 reported MRA as 91% sensitive and 71% specific in diagnosing labral lesions confirmed by open arthrotomy. Two musculoskeletal radiologists staged labral tears by consensus and correctly staged 35 of 40 lesions. Two false positive and three false negative MR arthrograms were noted.
Schmid et al27 reported on the effectiveness of MRA in detecting cartilage defects. Forty-two hips in 40 patients were assessed independently by two musculoskeletal radiologists with experience in MRI interpretation. Sensitivity and specificity were reported as 77% and 79%, respectively, by one reviewer, with 50% and 84% sensitivity and specificity, respectively, reported by the second reviewer. The radiologists were most consistent in detecting femoral cartilage abnormalities, but had greater difficulty identifying acetabular defects. The authors also noted that the presence of degenerative findings, such as osteophytes, cysts, and subchondral sclerosis, potentially resulted in radiologists overestimating the severity of articular cartilage abnormalities.
In our study, the comparison of MRA with hip arthroscopy findings resulted in a specificity and sensitivity of 44% and 71%, respectively, in the diagnosis of acetabular labral tears. Whereas the positive predictive value was 93%, a negative predictive value of only 12.9% was noted. The assessment of specificity of MRA in the evaluation of acetabular labral tears is limited because of the small number of patients without acetabular labral tears in this study (n = 9). The use of MRA in evaluating nonlabral articular cartilage pathology resulted in a sensitivity of 47%, a specificity of 89%, a positive predictive value of 84%, a negative predictive value of 59%, and an accuracy of 67%.
Although the purpose of this study was to assess the reliability of MRA reporting with actual findings at arthroscopy, we recognize that the reliability of MRA assessment is dependent on the experience of the individual radiologist doing the study. In light of this observation, we have retrospectively reviewed our data to assess the sensitivity and specificity for the radiologists doing our MRA studies (Table 6). It is important to note that, although there are differences in the sensitivity and specificity data among individual reviewers, these radiologists were not evaluating the same MR arthrograms. Because the prevalence of labral tears and articular cartilage abnormalities was not consistent, this assessment may not accurately reflect interobserver or intraobserver variance. Nonetheless, it is appreciated that surgeons doing hip arthroscopy need to understand the experience of their radiology consultants when considering surgical intervention in light of positive and negative MRA reports.
Although precise assessment of interobserver and intraobserver reliability for individual radiologists may not be obtained from this study, it is important to appreciate the potential variability in reporting on the basis of experience. Schmid et al27 reported significant variability in the sensitivity and specificity between two observers (Kappa < 0.2) in their study on the assessment of articular cartilage abnormalities with MRA. Difficulty in consistently identifying articular cartilage abnormalities was noted, particularly in the posterior acetabulum, where the articular cartilage is normally thinner in appearance.
The radiologists involved in the assessment of the MR arthrograms in our study all have extensive experience as subspecialists in musculoskeletal MRI. Although there is likely some variability between each of them in their ability to detect either acetabular labral tears or other articular cartilage abnormalities, these individuals represent a reasonable standard in MR interpretation from a tertiary referral, high-volume center. The clinical relevance to the orthopedic surgeon doing hip arthroscopy is in recognizing whether this procedure may be warranted in the absence of hard findings on MRA. Although the majority of articular cartilage abnormalities that were not detected by MRA were thought to be minor (Table 4) from a treatment standpoint, the underrecognition of 23 out of 93 acetabular labral tears (23.7%) is likely to be clinically significant.
In our experience, MRA provides a useful adjunctive tool to assist in the evaluation of intra-articular pathology of the hip. The detection of acetabular labral tears and other intra-articular pathology may assist the surgeon and patient in confirming the decision to proceed with arthroscopic management of articular hip pain. The positive predictive value for acetabular labral tears with MRA (93%) provides reasonable assurance of the indication for hip arthroscopy. However, as noted in this study, a substantial number of symptomatic labral tears and additional intra-articular abnormalities were not appreciated with MRA. This partially may be explained by a learning curve in interpretation that may improve as clinical experience increases. Additional viewing planes also have been described, which may optimize detection of tears.25 These were not part of existing MRA protocols at our institution, but should be evaluated in future studies.
In addition to detecting labral tears and articular cartilage pathology, MRA may have an expanding role in the evaluation of mild osseous abnormalities of the hip. In some cases, variations of acetabular retroversion and decreased femoral head-neck offset can be difficult to assess on standard radiographic images. For example, radiographs taken slightly out of rotation from the plane of reference may be suboptimal in diagnosing mild hip deformity.29 These may be more clearly defined on MR images because of the ability to assess for deformity in multiple planes of reference. In our experience, MRA can be a valuable tool in assisting the evaluation of these osseous sources of femoroacetabular impingement and secondary labral pathology.15 The decision to proceed with an open surgical procedure rather than arthroscopic labral debridement may be strongly influenced by the identification of sources of femoroacetabular impingement on the MR arthrogram.
The findings of our study confirm the requirement for a high clinical index of suspicion to diagnose symptomatic acetabular labral pathology. Although MRA is a good adjunctive study providing important diagnostic information, the importance of a careful patient history and physical examination cannot be overemphasized. Arthroscopy continues to serve an important diagnostic and therapeutic role in the treatment of patients with symptomatic articular hip pain. Additionally, we note that MRA arthrography provides important information when evaluating occult hip pathology. Specifically, alternative sources of hip pain, including stress fracture, osteonecrosis, and neoplasm, can be excluded. Useful information regarding the integrity of the labrum and articular cartilage is obtained in most cases and subtle osseous abnormalities, including acetabular retroversion and reduced femoral neck offset, may be characterized further before proceeding with surgical intervention. Collectively, this information is valuable in selecting appropriate surgical management approaches and optimizing treatment of patients with intra-articular sources of hip pain.
1. Baber YF, Robinson AHN, Villar RN: Is diagnostic arthroscopy of the hip worthwhile? A prospective review of 328 adults investigated for hip pain. J Bone Joint Surg 81B:600-603, 1999.
2. Beck M, Leunig M, Parvizi J, et al: Anterior femoroacetabular impingement: Part II: Midterm results of surgical treatment. Clin Orthop 418: 67-73.
3. Byrd JW: Labral lesions: An elusive source of hip pain case reports and literature review. Arthroscopy 12:603-612, 1996.
4. Byrd JW, Jones KS: Prospective analysis of hip arthroscopy with 2-year follow-up. Arthroscopy 16:578-587, 2000.
5. Czerny C, Hofmann S, Neuhold A, et al: Lesions of the acetabular labrum: accuracy of MR imaging and MR arthrography in detection and staging. Radiology 200:225-230, 1996.
6. Czerny C, Hofmann S, Urban M, et al: MR arthrography of the adult acetabular capsular-labral complex: Correlation with surgery and anatomy. AJR Am J Roentgenol 173:345-349, 1999.
7. Dorfmann H, Boyer T: Arthroscopy of the hip: 12 years of experience. Arthroscopy 15:67-72, 1999.
8. Edwards DJ, Lomas D, Villar RN: Diagnosis of the painful hip by magnetic resonance imaging and arthroscopy. J Bone Joint Surg 77B:374-376, 1995.
9. Farjo LA, Glick JM, Sampson TG: Hip arthroscopy for acetabular labral tears. Arthroscopy 15:123-137, 1999.
10. Fitzgerald Jr RH: Acetabular labral tears. Diagnosis and treatment. Clin Orthop 311:60-68, 1995.
11. Frich LH, Lauritzen J, Juhl M: Arthroscopy in diagnosis and treatment of hip disorders. Orthopedics 12:389-392, 1989.
12. Funke EL, Munzinger U: Complications in hip arthroscopy. Arthroscopy 12:156-159, 1996.
13. Glick JM, Sampson TG, Gordon RB, et al: Hip arthroscopy by the lateral approach. Arthroscopy 3:4-12, 1987.
14. Ide T, Akamatsu N, Nakajima I: Arthroscopic surgery of the hip joint. Arthroscopy 7:204-211, 1991.
15. Ito K, Minka MA, Leunig M, et al: Femoroacetabular impingement and the cam-effect: A MRI-based quantitative anatomical study of the femoral head-neck offset. J Bone Joint Surg 83B:171-176, 2001.
16. Lavigne M, Parvizi J, Beck M, et al: Anterior femoroacetabular impingement: Part I. Techniques of joint preserving surgery. Clin Orthop 418:61-66, 2004.
17. MacDonald SJ, Garbuz D, Ganz R: Clinical evaluation of the symptomatic young adult hip. Semin Arthroplasty 8:3-9, 1997.
18. McCarthy JC, Day B, Busconi B: Hip arthroscopy: Application and technique. J Am Acad Orthop Surg 3:115-122, 1995.
19. McCarthy JC, Lee JA: Acetabular dysplasia: A paradigm of arthroscopic examination of chondral injuries. Clin Orthop 405:122-128, 2002.
20. McCarthy J, Noble P, Aluisio FV, et al: Anatomy, Pathologic features, and treatment of acetabular labral tears. Clin Orthop 406:38-47, 2003.
21. O’Leary JA, Berend K, Vail TP: The relationship between diagnosis and outcome in arthroscopy of the hip. Arthroscopy 17:181-188, 2001.
22. Outerbridge RE: The etiology of chondromalacia patellae. J Bone Joint Surg 43B:752-757, 1961.
23. Petersilge CA, Haque MA, Petersilge WJ, et al: Acetabular labral tears: evaluation with MR arthrography. Radiology 200:231-235, 1996.
24. Plotz GMJ, Brossmann J, Schunke M, et al: Magnetic resonance arthrography of the acetabular labrum. Macroscopic and histological correlation in 20 cadavers. J Bone Joint Surg 82B:426-432, 2000.
25. Plotz GM, Brossmann J, von Knoch M, et al: Magnetic resonance arthrography of the acetabular labrum: Value of radial reconstructions. Arch Orthop Trauma Surg 121:450-457, 2001.
26. Sampson TG: Complications of hip arthroscopy. Clin Sports Med 20:831-835, 2001.
27. Schmid MR, Notzli HP, Zanetti M, et al: Cartilage lesions in the hip: diagnostic effectiveness of MR Arthrography. Radiology 226:382-386, 2003.
28. Siebenrock KA, Schoeniger R, Ganz R: Anterior femoro-acetabular impingement due to acetabular retroversion. Treatment with periacetabular osteotomy. J Bone Joint Surg 85-A:278-286, 2003.
29. Siebenrock KA, Wahab KH, Werlen S, et al: Abnormal extension of the femoral head epiphysis as a cause of cam impingement. Clin Orthop 418:54-60, 2004.