Validation of Scoring Hip Osteoarthritis with MRI (SHOMRI) Scores using Hip Arthroscopy as a Standard of Reference

Abstract

Purpose:

To validate SHOMRI gradings in pre-operative hip magnetic resonance imaging (MRI) with intra-arthroscopic evaluation of intraarticular hip abnormalities.

Methods:

Pre-operative non-arthrographic 3.0T MRI of 40 hips in 39 patients (one patient with bilateral hip surgery) with femoroacetabular impingement (FAI) syndrome (mean age, 34.7 years ± 9.0; n=16 women), refractory to conservative measures, that underwent hip arthroscopy were retrospectively assessed by two radiologists for chondrolabral abnormalities and compared to intra-arthroscopic findings as the standard of reference. Arthroscopically accessible regions were compared to the corresponding SHOMRI subregions and assessed for presence and grade of cartilaginous pathologies in the acetabulum and femoral head. The acetabular labrum was assessed for presence or absence of labral tears. For the statistical analysis sensitivity and specificity were calculated as well as intraclass correlation (ICC) for interobserver agreement.

Results:

With respect to chondral abnormalities, 58.8% of the surgical cases showed chondral defects. SHOMRI scoring showed a sensitivity of 95.7% and a specificity of 84.8% in detecting cartilage lesions. Moreover, all cases with full-thickness defects (n=9) were identified correctly and in n=6 cases (out of n=36 with partial-thickness defects) the defected cartilage was identified but the actual depth overestimated. Labral tears were present in all cases and the MR readers identified 92.5% correctly. ICC showed a good interobserver agreement with 86.3% (95% CI 80.0, 90.6%)

Conclusion:

Using arthroscopic correlation, SHOMRI grading of the hip proves to be a reliable and precise method to assess chondrolabral hip joint abnormalities.

INTRODUCTION

Osteoarthritis (OA) is a well-known global challenge for today’s health care system with far-reaching economic burden to society and consequences to each individual who is affected by this disease [1, 2]. Moreover, there is no known cure and no proven pharmacologic intervention available to treat OA [3, 4]. Because the progressive process of joint degeneration is directly associated with the aging population [5, 6], the prevalence of OA is anticipated to continually increase, resulting in OA as the leading cause of musculoskeletal disability in the elderly [7]. Hip OA in particular demonstrates rising incidence and prevalence with 7.4% of older individuals, aged 60 to 90 years, being affected [8]. Nonetheless, hip OA is also found in younger patients, whereas, especially femoroacetabular impingement (FAI) is a frequent source for hip joint degeneration [9]. Along with the OA related clinical features of pain and immobility, the crux lies in its irreversible progressive nature: gradual cartilage loss, accompanied by degeneration of other joint tissues, ultimately results in altered bone structure and joint destruction. Knowing this, it is not surprising that hip replacement procedures have doubled over the last decades, with 306,600 hip arthroplasties performed in 2011 in the United States alone [10]. To prevent late stage hip OA, precise imaging tools are necessary to assess morphological changes within the joints and to identify patients that are at risk of developing severe OA and to prevent further damage to the joint.

Magnetic Resonance Imaging (MRI) plays a critical role in assessing disease burden and in monitoring progression of joint abnormalities in the setting of OA. Therefore, MRI has become indispensable for clinical detection of joint alterations. However, the interpretation of hip MR images is challenging. Semi-quantitative MR-based hip OA evaluation systems have been introduced, such as the Scoring hip osteoarthritis with MRI (SHOMRI) grading system [11]. SHOMRI grades tissues and abnormalities involved in the degenerative process of hip OA, and it has been shown to significantly correlate with radiographic and clinical scores [11]. However, to date, SHOMRI has not been validated using an intra-arthroscopic evaluation as a standard of reference, which would provide information on the accuracy of this MR imaging-based grading in comparison with intra-operative-proven hip defects. We hypothesize, that SHOMRI grading of the hip is an accurate method for scoring intra-articular abnormalities of the hip using non-arthrographic 3.0T MRI.

Therefore, the purpose of this study was to correlate SHOMRI gradings in pre-operative hip MRI with intra-arthroscopic evaluation of hip abnormalities. Using a hybrid map that overlays the geographical zones used in hip arthroscopy on top of the MR-defined subregions provided by SHOMRI, we assessed the presence and grade of cartilaginous pathologies as well as labral abnormalities.

METHODS

Participants

All patients in our study cohort were prospectively enrolled from the hip preservation clinic at our institution. Throughout the recruitment, participants were initially examined by a sports medicine fellowship trained orthopedic surgeon and the data (including patient demographics, questionnaires and image acquisitions) were collected prior to hip arthroscopy. In total, forty hips in thirty-nine patients (one patient underwent surgery on both hips) with clinical and morphological signs of FAI were eligible for a surgical treatment and underwent hip arthroscopy. Patients were eligible for surgery if they were between the ages of 18 to 50 years, had a BMI less than 30 kg/m², and presented with functionally limiting hip pain due to cam-type or cam-predominant mixed-type FAI that was refractory to conservative measures including physical therapy and/or corticosteroid injections. Pre-operative radiographs were obtained in all subjects. Cam morphologies were defined as alpha angle greater than or equal to 55 degrees on a 45-degree Dunn lateral radiograph [12, 13]. Only patients with no or mild signs of osteoarthritis (grade 0–1; Tönnis classification [14] of OA by radiographic changes) were included. Exclusion criteria included evidence of moderate to severe arthritic changes (Tönnis grade 2 or higher) on pre-operative radiographs or prior history of hip surgery. Furthermore, the initial planning MRI for each patient was required to be within 6 weeks prior to the actual surgery.

This study was approved by the institutional Committee on Human Research and all patients provided written consent.

Image acquisition

Pre-operative imaging of the hip undergoing arthroscopic surgery was performed in all 39 patients. 3.0T MRI (GE Discovery MR750) were acquired using a 16-Channel Large Flex Array coil. All patients were scanned in the same position, with both feet taped together and rotated inwards. First, a three-plane localizer which was centered on the greater trochanter was obtained. The MRI protocol included triplanar 2D fat-suppressed fast spin-echo (echo time 60ms, repetition time 2400 to 3700ms, FOV 18.0cm, slice thickness 4mm) and 3D FOCUS fat-suppressed fast spin-echo, reformatted in in coronal, sagittal, and oblique axial FOCUS (TE 20.0ms, TR 1200.0ms, FOV 15.3cm, Slice Thickness 4.0mm) sequences.

Scoring chondrolabral lesions of the hip using MRI and arthroscopy

Using the SHOMRI scoring system, the MR images were retrospectively reviewed, independently by two musculoskeletal radiologists (BJS, JN) both blinded to the clinical and arthroscopic information. Each patient received a complete SHOMRI analysis [11], whereas, the current analysis focused on cartilage and labrum abnormalities since other features (bone marrow edema, subarticular cysts) are due to the nature of the abnormality not suitable for an arthroscopic correlation. Labral abnormalities were scored on a simplified, 2-point SHOMRI rating scale: no labral abnormalities or labral tear present. Articular cartilage lesions were graded on a 3-point rating scale: no cartilage lesions, partial thickness defects, or full-thickness defects. Acetabular and femoral arthroscopically-accessible cartilage regions (femur: anterior, superomedial and superolateral; acetabulum: anterior, superomedial and superolateral) were scored separately for each patient.

Diagnostic arthroscopy served as the standard of reference for the evaluation of intraoperative hip abnormalities. A 2-portal access was used for dynamic arthroscopic examination. Arthroscopic evaluation was assessed by a sports medicine fellowship trained orthopedic surgeon with a focus on hip arthroscopy surgery (ALZ) and graded the femoral and acetabular cartilage lesions based on the Outerbridge and Beck classifications, respectively [15] (Table 1).

Table 1: Arthroscopic evaluation of chondrolabral lesions.

Tables shows the arthroscopic evaluation of chondrolabral lesions in the hip.

Grading of the acetabular cartilage
SHOMRI 0 - No damage to the cartilage surface
Beck grade 0	Macroscopically sound cartilage with no morphological defects
Beck grade 1	Softening/malacia of the cartilage
Beck grade 2*	Debonding with loss of fixation of the cartilage to the acetabular bone
SHOMRI 1 - Partial-thickness defect of the articular surface
Beck grade 3	Partial thickness defects with fraying and tearing of the chondral surface
SHOMRI 2 - Full-thickness defect of the articular surface
Beck grade 4	Full-thickness defects to the subchondral bone
Grading of the femoral head cartilage
SHOMRI 0 - No damage to the cartilage surface
Outerbridge grade 0	Macroscopically sound cartilage with no morphologically defects
Outerbridge grade 1	Softening/malacia of the cartilage
SHOMRI 1 - Partial-thickness defect of the articular surface
Outerbridge grade 2	Partial thickness defects with fragmentation or tearing of the articular surface <0.5 inches
Outerbridge grade 3	Partial thickness defects with fragmentation or tearing of the articular surface >0.5 inches
SHOMRI 2 - Full-thickness defect of the articular surface
Outerbridge grade 4	Full-thickness defect featuring erosion of the cartilage down to the bone
Grading of the chondro-labral junction
No labral tear	Inspection and probing of the chondro-labral junction shows no abnormalities
Labral tear	Inspection and probing of the chondro-labral junction shows presence of a labral tear

^*Since acetabular delamination (Beck grade 2) is not included in the SHOMRI grading, these findings were not part of the analysis.

Evaluation of combined geographic zones of SHOMRI and arthroscopy

Both SHOMRI and arthroscopy use analogous geographic zones for evaluating intra-articular hip pathologies. SHOMRI uses a sagittal and coronal sequence to divide the femoral head and the acetabular articular surface into 6 and 4 subregions, respectively. Arthroscopic measurement of hip joint abnormalities relies on a similar zone-based system, recently published by Ilizaliturri et al [16]. To define the areas of agreement between the arthroscopic zone-based system and SHOMRI subregions, five training cases were retrospectively reviewed by an experienced orthopedic surgeon (ALZ) and an experienced musculoskeletal radiologist (TML). The training cases included arthroscopic videos, obtained during surgery, and pre-operative MRIs. Through the training cases, we created a hybrid map of surgical and MRI-based zones to define the matching regions (Figure 1). To compare both modalities we combined cartilage lesions that were located in the SHOMRI subregions: acetabulum anterior (AA), acetabulum superolateral (ASL), and acetabulum superomedial (ASM); and the arthroscopic zones: anterior-inferior region (zone 1), anterior-superior region (zone 2), and mid-superior region (zone 3). The same system was applied to the femoral head, where we combined the SHOMRI subregions femur anterior (FA), femur superolateral (FSL), and femur superomedial (FSM); and the arthroscopic zones: anterior-inferior region (zone 1), anterior-superior region (zone 2), and mid-superior region (zone 3). Since access with arthroscopic instrumentation is limited in certain areas, the posterior regions of the acetabulum and femoral head (zones 4 and 5) were excluded from the analysis due to limited access. The mid-inferior region (zone 6, acetabular fossa) was also excluded since this area is not covered by cartilage [17]. For the case that several lesions were visible in the same region, the more severe lesions were used for correlation. Furthermore, if cartilage lesions detected by the MRI were unable to be clearly assigned to the corresponding arthroscopic zone, the cases were reviewed, using intraoperative arthroscopy video recordings, to identify the corresponding regions and to improve clarity.

Figure 1: Evaluation of Combined Geographic Zones of SHOMRI and Arthroscopy.

Illustration of the left hemipelvis (A), anterior (B) and posterior (C) view of the femoral head. Arabic numerals within in dotted white lines refer to the different geographical zones as introduced by Ilizaliturri, whereas, colored areas refer to the subregions as used in SHOMRI scoring. (D) and (E) show the Arabic numerals and colored SHOMRI regions both overlaid on a coronal and sagittal 2D T2 fat-suppressed FSE sequence. (D) shows the areas of agreement for the acetabulum, whereas, (E) shows the areas of agreement for the femoral head. Please note, using SHOMRI, cartilage lesions in the anterior and posterior region are assessed on a sagittal view and lesions in the center region are assessed in the coronal view. AA = Acetabulum anterior, AP = Acetabulum posterior, ASL = Acetabulum superior lateral, ASM = Acetabulum superior medial, FA = femoral head anterior, FP = femoral head posterior, FSL = femoral head superior lateral, FSM = femoral head superior medial, FSI = femoral head superior inferior.

Agreement of chondrolabral lesions using SHOMRI and arthroscopy

For both modalities, in a first step cartilage defects were either graded as present or not present and the agreement between MRI and arthroscopy for this dichotomous categorization was assessed. In a second step, if cartilage defects were present, they were either graded as (1) partial-thickness or (2) full-thickness defects, and the agreement for this categorization was assessed separately. This was done separately for all acetabular and femoral subregions included in this analysis. If cartilage defects were correctly detected by MRI but the grade (partial- versus full-thickness defects) was scored differently during arthroscopy, the results were reported as (3) a cartilage lesion overestimated by MRI (a partial-thickness defect that was described as a full-thickness defect) or (4) a lesion underestimated by MRI (a full-thickness defect that was described as a partial-thickness defect). The interpretation of labral defects were scored in agreement if labral tears were present or absent. For all categories and comparisons of agreement, the lesions documented in the surgical reports served as standard of reference.

Statistical analysis

Statistical analysis was performed using SPSS statistical package version 23 (IBM; Armonk, NY, USA). The number of true-positive (TP), false-positive (FP), true-negative (TN), and false-negative (FN) test results was calculated for each category (chondral and labral lesions) using two-by-two tables. Using this data, sensitivity and specificity were calculated against intra-arthroscopic evaluation as the reference standard. Inter- and intraclass correlation (ICC) estimates and their 95% confidence intervals were calculated based on a mean-rating (k = 2), absolute agreement, 2-way mixed-effects model, where ICC values less than 0.5 indicated poor reliability, values between 0.5 and 0.75 indicated moderate reliability, values between 0.75 and 0.9 indicated good reliability, and values greater than 0.90 indicated excellent reliability [18].

RESULTS

Patient demographics

Thirty-nine patients with clinical and morphological signs of FAI were included in our study. Of these, one patient had bilateral hip MR imaging, resulting in a total of n=40 hip cases. For each case, the cartilage of the acetabulum and femur were scored separately, resulting in a total of n=80 regions with n=40 acetabular and n=40 femoral regions. The average age of all study participants was 34.7 years (standard deviation (SD) 9.0). Our study cohort consisted of n=16 women (41.0%) and n=23 men (59.0%). The average time between the pre-operative MR imaging study and hip surgery was 7.8 days (SD 8.2). The detailed patient characteristics are summarized in Table 2.

Table 2: Patient demographics.

In total 39 patients were included and 40 hips were scanned since 1 patient had surgery on both hips. Continues data are expressed as mean ± SD. Categorical data are presented in numbers of participants with percentage in parentheses.

Demographics
Mean age (years) of all patients	34.7 ± 9.0
BMI (kg/m2) of all patients	24.1 ± 3.1
Females [n (%)]	16 (41.0%)
Males [n (%)]	23 (59.0%)
Surgery left hip [n (%)]	22 (55.0%)
Surgery right hip [n (%)]	18 (45.0%)
Average time between imaging and surgery (days)	7.8 ± 8.2

Inter-/intrareader reliability

Interreader ICC between the readers for all chondrolabral readings showed a good agreement with 0.863 and a 95% confidence interval from 0.800 to 0.906 (p<.001). Subcategorized into the readings for chondral abnormalities of the femoral head, the acetabulum, and labral tears, the ICC ranged from 0.740 to 0.859. The average ICC for cartilage lesions in the acetabulum was 0.859 with a 95% confidence interval from 0.735 to 0.925 (p<.001) and 0.777 in the femoral head with a 95% confidence interval from 0.627 to 0.866 (p<.001). For labral tears the ICC was 0.905 with a 95% confidence interval from 0.800 to 0.955 (p<.001).

Intrareader reliability for both readers were good to excellent. The ICCs for intrareader agreement for reader 1 were as follows: All chondrolabral readings - 0.957 with a confidence interval from 0.936 to 0.970 (p<0.001); chondral abnormalities of the femoral head - 0.803 with a confidence interval from 0.667 to 0.883 (p<0.001); chondral abnormalities of the acetabulum – 0.852 with a confidence interval from 0.720 to 0.922 (p<0.001); labral tears - 0.979 with a confidence interval from 0.957 to 0.990 (p<0.001). The ICCs for intrareader agreement for reader 2 were as follows: All chondrolabral readings - 0.923 with a confidence interval from 0.891 to 0.945 (p<0.001); chondral abnormalities of the femoral head - 0.852 with a confidence interval from 0.753 to 0.911 (p<0.001); chondral abnormalities of the acetabulum - 0.753 with a confidence interval from 0.535 to 0.869 (p<0.001); labral tears - 0.890 with a confidence interval from 0.641 to 0.956 (p<0.001).

Surgical hip evaluation

Out of the 80 identified hip regions, arthroscopic evaluation revealed cartilage defects in 47 (58.8%) cases, with 9 full-thickness defects (11.3%) and 38 partial-thickness defects (47.5%). Thirty-three regions (41.2%) showed no cartilage defects. Interestingly, 8 out of 9 full-thickness defects were found in at the acetabular region (Figure 2 and 3). Partial-thickness defects were also more prevalent in the acetabular cartilage (Figure 4) than at the femoral head, with 23 and 15 cases of partial-thickness, respectively. With respect to labral abnormalities, all 40 hips showed labral tears.

Figure 2: Full-thickness defect of the acetabular cartilage.

MRI (coronal 2D T2 fat-suppressed FSE) of the left hip in a 40-year old female patient (A) shows a 9.4 mm full-thickness defect in the center of the superior acetabulum and a labral tear superior-lateral (white arrowhead). The arthroscopy image (B), 9 days after the pre-operative hip MRI, confirmed the full thickness defect of the articular cartilage.

Figure 3: Full-thickness defect of the superior lateral femoral head.

Hip MRI (coronal 2D T2 fat-suppressed FSE) of a 53-year old female patient with advanced degeneration of the left hip. The labrum was torn with associated paralabral cysts (A – white arrowhead). Moreover, the patient presented a full-thickness defect in the superior lateral region of the femoral head (A – white brackets). The femoral full-thickness cartilage defect is nicely seen in the arthroscopy (B) 5 days after the MRI with erosions of the cartilage to the bone (B - black arrowheads).

Figure 4: Torn labrum and partial-thickness loss of the acetabular cartilage.

Right hip MRI (coronal 2D T2 fat-suppressed FSE) of a 24-year old male patient (A) with intra-arthroscopic evaluation (B) of the same hip 4 days after the MRI. The labrum was torn at the chondrolabral junction (A – white arrow head) with a partial-thickness defect in the anterior superolateral region of the acetabulum (A – white bracket). Arthroscopy (B) shows the partial-thickness defect with the probe on the labrum (white arrow) and the partial-thickness defect of the acetabulum (B – black arrowhead).

SHOMRI evaluation

SHOMRI scoring showed a sensitivity of 95.7% and a specificity of 84.8% in determining if cartilage lesions were present in all scored regions (Table 3). Positive and negative predictive values (PPV, NPV) also showed good results (90% and 93%, respectively). Out of 47 arthroscopically-proven cartilage defects, 45 cases were correctly detected by the MR readers, and 28 cases (out of 33) were correctly identified as having no chondral defects.

Table 3: Results for evaluation of cartilage and labral lesions using SHOMRI readings with intra-arthroscopic standard of reference.

Data shows the correlation of SHOMRI readings with intra-arthroscopic evaluation of chondral and labral lesions in the hip. Combined analysis of chondral defects at the acetabulum and femoral head shows the combined results for any cartilage lesion present in the acetabulum and/or femoral head. Separate analysis of chondral lesions at the acetabulum/femoral head each show results for any cartilage lesions present only in the acetabulum or femoral head.

TP	TN	FN	FP	Sens.	Spec.	PPV	NPV	Acc.	Total cases
	Combined analysis of chondral defects at the acetabulum and femoral head
45	28	2	5	0.957	0.848	0.90	0.933	0.913	80
	Separate analysis of chondral lesions at the acetabulum
31	7	0	2	0.100	0.778	0.939	0.10	0.950	40
	Separate analysis of chondral lesions at the femoral head
14	21	2	3	0.875	0.875	0.824	0.913	0.875	40
	Analysis of labral tears
37	#n/a	3	0	0.925	#n/a	0.100	#n/a	#n/a	40

TP = True positive, TN = True negative, FN = False negative, FP = False positive, Sens. = Sensitivity, Spec. = Specificity, PPV = Positive predictive value, NPV = Negative predictive value, Acc. = Accuracy, Total readings = total readings that were included in the analysis.

When comparing the results for the acetabulum and the femoral head separately, the sensitivity was higher at the acetabulum with 100%, and a PPV of 93.9%, than in the femoral head with 87.5%, and a PPV of 82.4%. However, specificity was higher in the femoral head than the acetabular region, showing 87.5% and 77.8% respectively.

Within the 45 cases with cartilage defects correctly identified by the MR readers, all cases with arthroscopically-proven full-thickness defects (n=9, 20%) were correctly described as full-thickness defects using SHOMRI. Furthermore, 30 (66.67%) cases were correctly described as partial-thickness defects. The remaining 6 (13.34%) cases showed also arthroscopically proven partial-thickness defect, whereas, the MR readers overestimated the actual depth of the lesions and scored them as full-thickness defects (Table 4). Only 2 out of the 47 arthroscopically-proven cartilage defects were missed by the radiologists. In 5 cases, readers described partial-thickness defects that were not confirmed during arthroscopy, leaving a total of n = 7 cases that were identified as false positive (n=5) or false negative (n=2) readings. All patients in this study presented with intra-operative labral tears, and the radiologists only missed 3 tears (sensitivity = 92.5%). Table 4 summarizes the results of comparisons between SHOMRI gradings and arthroscopic evaluation.

Table 4: Cross table performance of SHOMRI readings with intra-arthroscopic evaluation.

5 × 5 cross table performance of all SHOMRI readings with intra-arthroscopic evaluation of chondral and labral lesions in the hip. Data in the first three columns/rows shows the agreement of both modalities with respect to grading of cartilage lesions, whereas, the least two columns/rows show the agreement of both modalities with respect to the labrum. Grey cells show lesions correctly identified using SHOMRI.

		Arthroscopy
		No chondral lesion	Full- thickness defect	Partial- thickness defect	Labral tear	No labral tear
SHOMRI	No chondral lesions	28	0	2
	Full-thickness defect	0	9	6
	Partial- thickness defect	5	0	30
	Labral tear				37	0
	No labral tear				3	0

DISCUSSION

This is the first study validating the semi-quantitative MRI hip osteoarthritis SHOMRI grading concerning intra-articular hip abnormalities using arthroscopic correlation. We validated SHOMRI scores using intra-operative evaluation as the standard of reference and were able to show both high sensitivity and specificity in detecting chondral lesions, and high sensitivity in detecting labral tears. Based on our findings, SHOMRI scoring of the hip is a precise and reliable method to evaluate hip cartilage and labral abnormalities.

MRI has been shown to be highly sensitive and specific for characterization and assessment of musculoskeletal disorders [19–23] and especially in the setting of joint pathologies several MR-based scoring methods have been introduced [11, 19, 24–28], all with the common purpose to provide practice guidelines in describing and quantifying radiologic findings and to minimize subjectivity. Moreover, the accurate and systematic classification of MRI using a standardized terminology of reporting creates reproducibility that is ideally reflected in high intra- and interrater reliability of each method.

The complex anatomy and mechanical conditions of the hip joint present a unique challenge for a standardized and reproducible image assessment. The closely apposed articular surfaces of the femoral head and the acetabulum, in an already small intra-articular volume, require a well-elaborated imaging protocol for optimal visualization of the hip joint. To better differentiate the cartilage layers Schmaranzer et al used traction MR arthrography in 73 patients who underwent arthroscopy afterwards [29]. While this technique improves the visualization of articular cartilage layers as two distinct layers, and as shown by the authors of this study also allows an accurate detection of chondrolabral abnormalities, in the clinical setting it assumes a certain level of comfort and might carry the risk of patient discomfort and eventually development of pain. Furthermore, as pointed out by the authors, a direct comparison of conventional MR arthrography and traction MR arthrography is still needed to determine if either of these techniques is superior to the other.

So far, only a few studies have focused on the evaluation of MR-based hip scoring methods [24– 26] and, moreover, only one of these studies used the complete study cohort for correlation with surgical reports [26]. Overall, these studies showed a good performance evaluation hip MRI with surgical correlation. However, the complete imaging protocol either required an MRI arthrogram [24, 26] or administration of intravenous contrast [25]. In contrast, by using SHORMI, the assessment of hip joint abnormalities in our investigation was performed with a non-arthrographic 3.0T system. Still, we were also able to show high sensitivity and specificity for the assessment hip joints by means of MRI.

Generally, in an effort to optimize image quality while maintaining scan time efficiency, the use of MR arthrography challenges the conventional, non-enhanced, MRI. However, although several studies have reported a higher accuracy for MR arthrography in the detection of chondrolabral lesions [17, 30, 31], the use of intraarticular contrast might not be logistically feasible in a larger study trial and/or in clinical routine [32] and, more importantly, for the majority of these studies a 1.5T MRI systems was used. In this context, recent studies have shown that the use of a high resolution, non-arthrographic 3T MRI is capable of detecting chondral and labral abnormalities with high accuracy [20, 33, 34], and therefore, may obviate the need for intraarticular contrast.

One of the primary characteristics of our study design was the feasibility of SHOMRI to detect cartilage lesions with surgical reports as the standard of reference. Naturally, other findings in the adjacent bone, that are also involved in the process of joint degeneration (bone marrow edema patterns or subarticular cysts), are not visible on macroscopic examination during surgical interventions [24, 26] and were therefore not part of our investigation. Cartilage deterioration, as the main characteristic of joint degeneration, is usually at some point accompanied by changes in the subchondral bone and therefore both can be seen as a continuously challenged functional unit when involved in degeneration [35]. However, studies have indicated a possible protective effect of the cartilage layer on the functional and structural changes in the subjacent bone, such as bone marrow pattern and subarticular cysts [36–38]. Therefore, especially the accurate identification of cartilage lesions will be of great benefit in the setting of a pre-operative hip evaluation using MRI.

We acknowledge that our study has several limitations. Firstly, all of our study participants showed an acetabular labral tear and, therefore, no specificity was computed during analysis of the SHOMRI scores. However, because FAI is the predominant cause of labral tears in nondysplastic hips [39, 40] and all of our patients were eligible for surgery, it is expected to find a high prevalence of labral tears in such a cohort. Secondly, our readings were only performed on patients with FAI syndrome and no patients without clinical or morphological signs of FAI were included. Due to the purpose of our study, evaluation of hip abnormalities with intraoperative correlation, a surgical report was needed and thus, to extend this study design to including controls without indication for surgical treatment is not feasible. Finally, although arthroscopy is the choice of treatment for patients with FAI and no to mild radiographic changes (Tönnis 0 – 1) [41] the access of the arthroscopic instrumentation is limited especially in the posterior regions of the hip joint and thus, we did not include the posterior cartilage regions in our investigation.

In conclusion, the semi-quantitative MR-based scoring system, SHOMRI, is a useful evaluation method for hip abnormalities that with good to excellent reliability [11]. In the current study, we evaluated SHOMRI scores with arthroscopic correlation as a standard of reference. We were able to show a high sensitivity and specificity for chondral assessment at the acetabulum and femoral head in patients with FAI syndrome. Additionally, our results also showed a high sensitivity in detecting labral tears. Within the arthroscopically accessible regions, SHOMRI grading proved to be a reliable and precise method to assess chondrolabral abnormalities in the hip joint.

Key points:

1. Assessment of hip abnormalities using MRI with surgical correlation.

2. Comparing surgery and MRI by creation of a hybrid anatomic map that covers both modalities.

3. Non-arthrographic use of 3.0T MRI provides detailed information on cartilage and labral abnormalities in hip joints.

Abbreviations:

FAI

Femoroacetabular impingement

ICC

Intraclass correlation

MRI

Magnetic Resonance Imaging

NPV

Negative predictive value

OA

Osteoarthritis

PPV

Positive predictive value

SD

Standard deviation

SHOMRI

Scoring hip osteoarthritis with MRI