ABSTRACT

Currently, much is debated on the optimal treatment of borderline hips, being in the continuum between stable and unstable hips. The diagnosis of stability is often difficult but is a prerequisite for further treatment. Analysis includes a variety of radiographic parameters. We observed that unstable hips often had a crescent-like gadolinium collection in the postero-inferior joint space. We therefore questioned if the ‘crescent sign’ could be an indicator for hip instability? A retrospective comparative study was conducted including 56 hips in the instability group (treated with PAO) and 70 hips with femoroacetabular impingement (FAI) as control group. Based on standard radiographic parameters and magnetic resonance imaging (MRI), the association between hip instability and the ‘crescent sign’ was analyzed. For univariate group comparisons, the non-parametric Wilcoxon two sample test was used. Association between discrete variables was examined by means of chi-square tests. To examine predictive variables, logistic regression models were carried out. Most hips with a crescent sign belong to the instability group. A crescent sign has a sensitivity of 73.3% and specificity of 93% for instability. Based on our results, the crescent sign is a factor that is more prevalent in unstable hips. However, its absence does not exclude instability of the hip. If present, the specificity speaks strongly in favor for instability of the hip.

INTRODUCTION

The natural history of hip dysplasia is well documented [1, 2] and is associated with the early onset of osteoarthritis. Reduced coverage results in excessive loading of the acetabular rim leading to joint degeneration [3]. Surgical correction of this unfavorable morphology can be successfully achieved by reorienting the acetabulum, most commonly through the use of the Bernese periacetabular osteotomy (PAO) [4]. Improved coverage of the femoral head reduces shear stress and leads to a more physiologic distribution of the joint reaction forces. Long term survivorship and good clinical results support the use of the PAO in hip dysplasia [5]. While the indications for acetabular reorientation in hips with clear radiographic signs of dysplasia with a reduced lateral center edge angle (LCEA) < 20° [6] and acetabular index (AI) of > 10–14° [7–9] are well established, decision-making becomes much more difficult in hips with borderline acetabular coverage radiologically defined with an LCE of 20–25° [6, 10]. The acetabular coverage is only slightly reduced, and it may be difficult to determine the stability of the hip. Unstable, symptomatic borderline dysplastic hips should be treated with a PAO, whereas stable hips are treated according to their underlying pathology.

Classifying a borderline hip as stable or unstable can be challenging [11]. Several radiographic parameters are associated with instability including a low LCEA, a high AI, a high neck shaft angle, increased femoral torsion (FT), upsloping lateral sourcil [12] or a positive Femoro-Epiphyseal Acetabular Roof (FEAR) index [13]. A break in Shenton’s line is indicative of hip subluxation and instability [14]. In magnetic resonance arthrography (MRA) hip instability can be associated with chondral and labral disorders. Hypertrophy of the labrum [15–17] or femoroacetabular hypertrophy of the cartilage has been identified as a sign of hip dysplasia. Further an increased size of iliocapsularis muscle [18, 19], a fovea alta that represents an abnormal superior position of the fovea capitis [20] or tears of the ligamentum teres [21] can be visualized in MR imaging as indirect signs of instabiliy of the hip joint. Some other signs are visible during arthroscopy and give further intraoperative information as for hip dysplasia like the ‘inside-out’ chondrolabral flap or chondrolabral lesions Outerbridge Grade III or IV in the weightbearing area in borderline dysplastic hips [22].

During clinical practice, we observed that unstable hips often had a gadolinium collection in the postero-inferior joint space (‘crescent sign’). Based on that observation we questioned: (i) if the so-called crescent sign is a predictor for hip instability and (ii) if its predictive power can be increased in combination with other established radiographic parameters to identify hip instability?

MATERIAL AND METHODS

This study was approved by the local ethics committee (Ethikkommission Nordwest- und Zentralschweiz, 2020-00587). A retrospective, comparative study was conducted using the database of hip preserving surgeries performed at our institution between January 2010 and December 2017 (881 hips). All patients were treated by a single orthopedic surgeon (M.B.). One hundred and ten patients (129 hips) had a PAO and 442 patients (482 hips) underwent arthroscopic FAI correction. Of these, all hips who had a PAO for symptomatic hip instability with an LCE angle < 20° and AI > 10° or unstable symptomatic borderline dysplasia with an LCE between 20 and 25° were located into the instability group. In our daily practice, the FEAR-index and presence of a hypertrophic labrum were used as additional radiological factors. Typical clinical signs of overload of abductor muscles and a positive apprehension sign were used for the decision-making process. The control group consisted of stable hips treated for FAI. The number of hips in the control group was adjusted to the number of hips in the instability group. Cases were reviewed chronologically from the most recent to the most distant until an equal number of cases was reached. 110 patients (129 hip) with a PAO and 127 patients (156 hips) with FAI represented the base of the study. Exclusion criteria for both groups included previous surgery of the hip (instability group 4 hips, control group 34 hips), inadequate imaging e.g. missing radial sequence, insufficient image quality (instability group 46 hips, control group 48 hips), slipped capital femoral epiphysis (control group 1 hip), Legg-Calvé-Perthes disease (instability group 9 hips), treatment of traumatic lesions (control group 3 hips) and acetabular retroversion (instability group 14 hips). The remaining cases for analysis consisted of 56 hips (51 patients) in the instability group and 70 hips (66 patients) in the control group (Fig. 1).

Fig. 1.

Flowchart of selected patients.

Fig. 1.

Flowchart of selected patients.

In both study groups, we evaluated preoperative radiographs and MRA assessing following parameters [15]: Joint degeneration, using Tönnis’ classification [23] and FEAR-index [13], was assessed on standardized anteroposterior (AP) pelvic radiographs. LCEA, AI, extrusion index (EI), retroversion index, anterior coverage, posterior coverage and total coverage were measured on the preoperative standardized AP pelvic radiographs using the validated software Hip2Norm (University of Bern, Bern, Switzerland) [24–26]. Based on the MR imaging, FT was measured according to the method of Murphy et al. [27] and the alpha angle to detect the femoral neck offset according to the method of Nötzli et al. [28].

MRA followed a standardized protocol as described previously [15]. Briefly, the protocol for MRA included a transverse T1-weighted turbo spin-echo sequence, a coronal and sagittal intermediate-weighted turbo spin-echo sequence, a radial intermediate-weighted turbo spin-echo sequence, a transverse THRIVE (T1-weighted high-resolution isotropic volume examination) or DESS (dual-echo steady state) sequence and a rapid transverse T1-weighted sequence over the proximal and distal femur for anteversion measurement.

The measurement of the crescent sign was carried out in a standardized way by two observers (J.M. and V.K.) independently. Following four imaging planes were used: axial, two different sagittal and radial. The imaging planes were linked and the slices at the level of the center of femoral head (CFH) were used in all planes for measurement. A crescent sign was assessed as present whenever contrast agent could be seen between femoral head and acetabulum. The planes were defined such that they all are going through the center of the posterior horn. The technique was as follows:

The four planes were linked and the slice through the CFH identified on each. The resulting axial and sagittal images were used for the first three measurements (Fig. 2a–c).

Axial plane: The center of the posterior wall (CPW) is identified as the midpoint between the edge of the fossa and the rim of the posterior wall (Fig. 3a and b).

Sagittal plane: Two measurements were done here—the first at the center of the femoral head CFH. The second at the CPW, to account for possible lateral migration of the femoral head (Fig. 4a–c). A line is drawn from the anterior to the posterior acetabular rim. Then, a perpendicular line is drawn at mid-distance to divide the femoral head. From the intersection a line is drawn at 22.5°, reaching the center of the posterior horn (Fig. 5a and b, Fig. 6a and b), where we see the ‘crescent sign’ mostly.

Radial plane: With the three other linked images on the CFH, the radial slice which intersects the CPW is then selected (Fig. 7a and b).

Fig. 2.

Femoral head center is detected in the three planes: (A) coronal, (B) axial, (C) sagittal. Based on that, the further analysis is performed.

Fig. 2.

Femoral head center is detected in the three planes: (A) coronal, (B) axial, (C) sagittal. Based on that, the further analysis is performed.

Fig. 3.

A/B: Axial plane. (A) Example of a dysplastic hip. Posterior horn (1) at the level of the femoral head center. Gadolinium in between the posterior horn and femoral head (2), representing a positive crescent sign. (B) Control hip without a crescent sign.

Fig. 3.

A/B: Axial plane. (A) Example of a dysplastic hip. Posterior horn (1) at the level of the femoral head center. Gadolinium in between the posterior horn and femoral head (2), representing a positive crescent sign. (B) Control hip without a crescent sign.

Fig. 4.

A–C: Sagittal plane at the level of the center of the posterior wall (CPW). The crescent sign was assessed on the sagittal plane (A) going through the CFH and the center of posterior wall (C).

Fig. 4.

A–C: Sagittal plane at the level of the center of the posterior wall (CPW). The crescent sign was assessed on the sagittal plane (A) going through the CFH and the center of posterior wall (C).

Fig. 5.

A/B: Sagittal plane measurement at CFH. (A) Instable, borderline dysplastic hip. A line is drawn between the edge of the anterior and posterior wall (1). At mid-distance a perpendicular line is drawn (2). From the intersection an angle of 22.5° (postero-inferior quarter) in the direction to the posterior wall is made (3). (B) Sagittal plane at CFH in a control hip.

Fig. 5.

A/B: Sagittal plane measurement at CFH. (A) Instable, borderline dysplastic hip. A line is drawn between the edge of the anterior and posterior wall (1). At mid-distance a perpendicular line is drawn (2). From the intersection an angle of 22.5° (postero-inferior quarter) in the direction to the posterior wall is made (3). (B) Sagittal plane at CFH in a control hip.

Fig. 6.

A/B: (A) Instable, borderline dysplastic hip. Crescent sign on the sagittal slice through CPW. Identification of location to assess the presence of a crescent sign is performed like in Fig. 5 but at the level of CPW. Number (4) shows the crescent sign in this plane in with a black triangle. (B) Absence of crescent sign in a control hip.

Fig. 6.

A/B: (A) Instable, borderline dysplastic hip. Crescent sign on the sagittal slice through CPW. Identification of location to assess the presence of a crescent sign is performed like in Fig. 5 but at the level of CPW. Number (4) shows the crescent sign in this plane in with a black triangle. (B) Absence of crescent sign in a control hip.

Fig. 7.

A/B: Radial plane. (A) Instable, borderline dysplastic hip. A line between anterior and posterior wall is defined (1). Then a perpendicular (2) is drawn at the mid-distance is placed. From the intersection a line is made at an angle of 22.5° towards the posterior horn (3). Positive crescent sign (4). (B) Radial plane without a crescent sign in a control hip.

Fig. 7.

A/B: Radial plane. (A) Instable, borderline dysplastic hip. A line between anterior and posterior wall is defined (1). Then a perpendicular (2) is drawn at the mid-distance is placed. From the intersection a line is made at an angle of 22.5° towards the posterior horn (3). Positive crescent sign (4). (B) Radial plane without a crescent sign in a control hip.

Inter-observer agreement was analyzed by adopting the method according to Shrout and Fleiss using the intra-class correlation coefficient (ICC) as a principle measure of reliability [29, 30]. Agreement of the intra-class correlation values was interpreted as: greater than 0.75 = excellent, 0.40–0.75 = fair to good and less than 0.40 = poor [29]. The inter-rater agreement is demonstrated by a graphical technique suggested by Bland and Altman [31].

Univariate group comparisons were carried out using the non-parametric Wilcoxon Two Sample test. Test on normality was carried out by means of the Shapiro–Wilk test. Association between discrete variables was examined by means of chi-square tests. To examine predictive variables with respect to group classification (unstable vs. control), logistic regression models were carried based on a stepwise variable selection.

For all statistical analyses the SAS software, version 9.4 was used (SAS Institute, Cary, NC).

RESULTS

There were no significant differences according to age, gender or affected side between the two study groups (Table I). Joint degeneration of a maximum of Tönnis grade I was visible without a significant difference between the groups.

Parameter . Instability group . Control group . P-value . Number of patients 51 66 – Number of hips 56 70 – Age (years) 29 ± 10 (15–63) 32 ± 12 (14–54) 0.141 Gender (% male of all hips) 71 56 0.070 Side (% right of all hips) 64 49 0.056 Parameter . Instability group . Control group . P-value . Number of patients 51 66 – Number of hips 56 70 – Age (years) 29 ± 10 (15–63) 32 ± 12 (14–54) 0.141 Gender (% male of all hips) 71 56 0.070 Side (% right of all hips) 64 49 0.056 Parameter . Instability group . Control group . P-value . Number of patients 51 66 – Number of hips 56 70 – Age (years) 29 ± 10 (15–63) 32 ± 12 (14–54) 0.141 Gender (% male of all hips) 71 56 0.070 Side (% right of all hips) 64 49 0.056 Parameter . Instability group . Control group . P-value . Number of patients 51 66 – Number of hips 56 70 – Age (years) 29 ± 10 (15–63) 32 ± 12 (14–54) 0.141 Gender (% male of all hips) 71 56 0.070 Side (% right of all hips) 64 49 0.056 

The ICC for all analyzed parameters demonstrated excellent agreement (ICC 0.85–0.99, excellent >0.75), with the highest level for the crescent sign demonstrated in the axial plane with an ICC of 0.99.

There was no significant difference in degree of arthritis between groups using the Tönnis score. As expected, the instability group was found to have reduced LCEA and acetabular coverage, elevated EI, AI and a pathological FEAR-index > 2 compared to the control group with stable hips. With exception of the FT and the retroversion index, which were equal and normal in the two study groups, all parameters between the two groups showed a significant difference (Table II). Both groups had 13 hips with borderline cover (LCEA ≥ 20 and <25°), one hip in the control group had an LCEA of 19°.

Table II.

Hip parameters of the two study groups

Parameter . Normal value . Instability group . Control group . P-value . x-ray-based measurements LCEA (degrees) [6] 25–33 16 ± 7 (3–29) 31 ± 6 (19–47) <0.0001 AI (degrees) [35] 0–10 19 ± 7 (4–39) 6 ± 4 (−5–17) <0.0001 EI (degrees) [7] 17–27 33 ± 7 (18–48) 20 ± 6 (6–34) <0.0001 Retroversion index (percent) [36, 37] <4 8 ± 15 (0–47) 12 ± 15 (0–51) 0.170 Anterior coverage (percent) [7] 15–26 13 ± 7 (0–33) 24 ± 9 (6–61) <0.0001 Posterior coverage (percent) [7] 36–47 34 ± 11 (7–57) 44 ± 9 (27–72) <0.0001 Total coverage (percent) [7] 70–83 63 ± 9 (43–79) 80 ± 7 (63–93) <0.0001 FEAR-index (degrees) [13] <2 4 ± 11 (−28–27) −16 ± 8 (−41–6) <0.0001 Tönnis score (percent) [23]    0.414 Grade 0  89 (50 hips) 87 (61 hips)  Grade 1  11 (6 hips) 13(9 hips)  Grade 2  0 0  Grade 3  0 0  MR-based measurements Alpha angle (degrees) [28] <50 50 ± 12 (31–90) 63 ± 12 (34–84) <0.0001 FT (degrees) [23, 27] 10–25 24 ± 12 (−4–50) 20 ± 9 (−1–42) 0.052 Parameter . Normal value . Instability group . Control group . P-value . x-ray-based measurements LCEA (degrees) [6] 25–33 16 ± 7 (3–29) 31 ± 6 (19–47) <0.0001 AI (degrees) [35] 0–10 19 ± 7 (4–39) 6 ± 4 (−5–17) <0.0001 EI (degrees) [7] 17–27 33 ± 7 (18–48) 20 ± 6 (6–34) <0.0001 Retroversion index (percent) [36, 37] <4 8 ± 15 (0–47) 12 ± 15 (0–51) 0.170 Anterior coverage (percent) [7] 15–26 13 ± 7 (0–33) 24 ± 9 (6–61) <0.0001 Posterior coverage (percent) [7] 36–47 34 ± 11 (7–57) 44 ± 9 (27–72) <0.0001 Total coverage (percent) [7] 70–83 63 ± 9 (43–79) 80 ± 7 (63–93) <0.0001 FEAR-index (degrees) [13] <2 4 ± 11 (−28–27) −16 ± 8 (−41–6) <0.0001 Tönnis score (percent) [23]    0.414 Grade 0  89 (50 hips) 87 (61 hips)  Grade 1  11 (6 hips) 13(9 hips)  Grade 2  0 0  Grade 3  0 0  MR-based measurements Alpha angle (degrees) [28] <50 50 ± 12 (31–90) 63 ± 12 (34–84) <0.0001 FT (degrees) [23, 27] 10–25 24 ± 12 (−4–50) 20 ± 9 (−1–42) 0.052 Table II.

Hip parameters of the two study groups

Parameter . Normal value . Instability group . Control group . P-value . x-ray-based measurements LCEA (degrees) [6] 25–33 16 ± 7 (3–29) 31 ± 6 (19–47) <0.0001 AI (degrees) [35] 0–10 19 ± 7 (4–39) 6 ± 4 (−5–17) <0.0001 EI (degrees) [7] 17–27 33 ± 7 (18–48) 20 ± 6 (6–34) <0.0001 Retroversion index (percent) [36, 37] <4 8 ± 15 (0–47) 12 ± 15 (0–51) 0.170 Anterior coverage (percent) [7] 15–26 13 ± 7 (0–33) 24 ± 9 (6–61) <0.0001 Posterior coverage (percent) [7] 36–47 34 ± 11 (7–57) 44 ± 9 (27–72) <0.0001 Total coverage (percent) [7] 70–83 63 ± 9 (43–79) 80 ± 7 (63–93) <0.0001 FEAR-index (degrees) [13] <2 4 ± 11 (−28–27) −16 ± 8 (−41–6) <0.0001 Tönnis score (percent) [23]    0.414 Grade 0  89 (50 hips) 87 (61 hips)  Grade 1  11 (6 hips) 13(9 hips)  Grade 2  0 0  Grade 3  0 0  MR-based measurements Alpha angle (degrees) [28] <50 50 ± 12 (31–90) 63 ± 12 (34–84) <0.0001 FT (degrees) [23, 27] 10–25 24 ± 12 (−4–50) 20 ± 9 (−1–42) 0.052 Parameter . Normal value . Instability group . Control group . P-value . x-ray-based measurements LCEA (degrees) [6] 25–33 16 ± 7 (3–29) 31 ± 6 (19–47) <0.0001 AI (degrees) [35] 0–10 19 ± 7 (4–39) 6 ± 4 (−5–17) <0.0001 EI (degrees) [7] 17–27 33 ± 7 (18–48) 20 ± 6 (6–34) <0.0001 Retroversion index (percent) [36, 37] <4 8 ± 15 (0–47) 12 ± 15 (0–51) 0.170 Anterior coverage (percent) [7] 15–26 13 ± 7 (0–33) 24 ± 9 (6–61) <0.0001 Posterior coverage (percent) [7] 36–47 34 ± 11 (7–57) 44 ± 9 (27–72) <0.0001 Total coverage (percent) [7] 70–83 63 ± 9 (43–79) 80 ± 7 (63–93) <0.0001 FEAR-index (degrees) [13] <2 4 ± 11 (−28–27) −16 ± 8 (−41–6) <0.0001 Tönnis score (percent) [23]    0.414 Grade 0  89 (50 hips) 87 (61 hips)  Grade 1  11 (6 hips) 13(9 hips)  Grade 2  0 0  Grade 3  0 0  MR-based measurements Alpha angle (degrees) [28] <50 50 ± 12 (31–90) 63 ± 12 (34–84) <0.0001 FT (degrees) [23, 27] 10–25 24 ± 12 (−4–50) 20 ± 9 (−1–42) 0.052 

The majority of patients with a positive crescent sign belong to the group with unstable hips. In 37 hips (29.4%, 33 hips instability group, 4 hips control group), the sign was present. When a crescent sign is present in all planes, an acceptable sensitivity of 73.3% and good specificity with 93% for the presence of instability was found (Table III). Of all measurements of the newly introduced crescent sign (radial, sagittal CFH/CPW, axial), the most sensitive individual measurement was the axial [67.9%, negative predictive value 77% (CI 95% 67 to 85)] (Table IV), whereas the most specific was the sagittal at the CFH [94.3%, positive predictive value 84% (CI 95% 64 to 95)] (Table V). The strongest association of an instability was seen when the crescent sign was present in all slices (Table VI).

Table III.

Summary of statistical measures of FEAR-index and crescent signs parameters

.  . Crescent signs . Parameter . FEAR-Index . Axial . Radial . Sagittal (CFH) . Sagittal (CPW) . Signs present in all planes . Sensitivity (%) 65.5 67.9 66.1 39.3 66.1 73.3 Specificity (%) 95.7 88.6 84.3 94.3 87.1 93  .  . Crescent signs . Parameter . FEAR-Index . Axial . Radial . Sagittal (CFH) . Sagittal (CPW) . Signs present in all planes . Sensitivity (%) 65.5 67.9 66.1 39.3 66.1 73.3 Specificity (%) 95.7 88.6 84.3 94.3 87.1 93 Table III.

Summary of statistical measures of FEAR-index and crescent signs parameters

.  . Crescent signs . Parameter . FEAR-Index . Axial . Radial . Sagittal (CFH) . Sagittal (CPW) . Signs present in all planes