Pre-surgical CT-assessment of neurogenic myositis ossificans of the hip and risk factors of recurrence: a series of 101 consecutive patients

CT-scans were performed with a 16-section scanner (MX IDT 8000, Philips Medical Systems, Best, the Netherlands) and a double-piston power injector (Medtron, Saarbruecken, Germany). Patients installation was crucial for the success of the exam. It was often difficult because of the deformities and required various methods especially the use of cushions. We placed tourniquets around thighs and calves in order to obtain better opacification of the main vessels, not of superficial vessels. We generally catheterized a vein of the elbow using a 20G cannula and performed a biphasic injection using a high-concentration iodinated solution (iomeprol, Iomeron 400 mg/mL, Bracco Imaging, France) including a first injection of 120 mL at the rate of 1.5 mL/s followed by an injection of 80 mL at the rate of 3 mL/s. CT acquisition was performed with a collimation of 16 × 0.75 mm and was triggered 135 s after the start of injection. Voltage was 140kV and amperage 300 mAs/slice. Rotation time was 0,75 s/rotation. Axial reconstructions were obtained 2 mm in thickness every 1 mm. Multi-planar reconstructions were performed as well as volume rendering 3D images with colour encoding according to tissue densities so as the bone appeared white, veins appeared blue and arteries appeared red.

The analysis of CT imaging was performed by investigators RC and DS, both of whom have 21 years of experience with this technique. Results are detailed in the Table 1. For each lesion, we determined the implantation base, the location and borders, the presence of maturation zones defined by hypodense non-mineralized tissue (Fig. 1), fragmentation, pseudoarthrosis or intra-lesional neo-joint. We studied the relations to the vessels and classified the either as a slight displacement, a gutter (i.e., a covering of the vessel less than 180° in circumference), a tunnel (engainement of the vessel), a compression or a thrombosis (Fig. 2). Anterior osteomas presented risks for femoral and anterior circumflex vessels while posterior osteomas presented risks for posterior circumflex and gluteal vessels. Apart from hemorrhages, surgical lesions of the circumflex arteries could result in femoral head necrosis. Relationships with nerves were depicted likewise by the following terms: contact, gutter or tunnel. Risks of nervous complications concerned posterior osteomas and sciatic nerves (Fig. 3). Relationships to joint capsules were depicted either as no involvement, touches joint capsule or disrupts joint capsule. Additionally, we reported intra-articular calcifications or ossifications (Fig. 4). Epiphyseal bone mineralization was depicted and classified into four categories: normal (M1), mild demineralization (M2), significant demineralization with a risk of fracture (M3) and severe demineralization (M4), with a replacement of osseous tissue by fatty tissue (Fig. 5). Finally, the joint space was depicted as preserved (S1), narrowed (S2) or ankylosed (S3) [1].

Among 101 patients, 80 had surgery within 30 days after the CT scan. We studied all surgical reports and collected the following data: time between accident and surgery, indication for surgery (ankylosis, pain, deformities with flessum, joint limitation, recurrence, sciatica), presence and type of nerves/vessels relationship, joints relations and surgical complications (fractures, hemorrhages). All the patients were seen every six months for follow-up examination to look especially for infections, fractures or recurrence.

Chi2-test was used to comparequalitative variables with normal distribution and Fischer’s test was used for non-normal qualitative values with. Quantitative values were analyzed with a one factor analysis of variance (ANOVA) test. A p value < 0.05 was considered significant. Agreement between pre-surgical CT and surgical observations was evaluated using a Cohen’s kappa test. A coefficient < 0.20 was considered weak agreement, from 0.21 to 0.40 a moderate agreement, from 0.41 to 0.60 a good agreement, from 0.61 to 0.80 a strong agreement and from 0.81 to 1 .00 an excellent agreement.

A hundred and one patients were included in the study. 83 patients were male (82%) and 18 were female (18%). NPOA were bilateral in 31 patients (30%), representing a total of 132 hips. Mean age was 46 years (16-73 years). Etiologies were: brain trauma (n = 48 (46.5%)), prolonged resuscitation (n = 25 (21.7%)), stroke (n = 21 (19.8%)), spinal cord injury (n = 8 (7.9%)) and one case of progressive ossifying fibrodysplasia (1%). Two patients (2%) suffered from both spinal cord and brain trauma. Among 132 hips, 86 had surgery (65%). Patients who did not have surgery had either little functional impact or contra-indications (risks of infections and hemorrhages). Main indication for surgery was ankylosis, defined as a single bony bridge, and concerned 33 cases (38.4% of 86). Joint limitation with deformities was the second main indication (20 cases (23.2%)). Five patients were treated for recurrence of NMO. Median time from initial injury to surgery was 21 months (6-285 months). Median time to surgery was 22 months after brain trauma and 20 months after spinal cord injury.

Regarding CT-scans, mean total radiation dose per patient was 555 mGy.cm +/-261 SD and mean effective dose was 8,3 mSv +/- 3,9 SD.

Osteomas were anteriorly located in 64 hips (48.5% of 132 hips), posterior in 52 hips (39.4%) and circumferential in 16 hips (12.1%). The osteomas had clear margins in 101 cases (76.5% of 132 hips) and irregular infiltrating margins in 31 cases (23.5%). We observed maturation zones in 30 cases. Osteomas were monofragmentary in 49 cases (37.1%) and polyfragmentary in 83 cases (62.8%). Pseudoarthrosis was observed in 56 cases with appearance of a new joint (42.4%).

Contacts with vessels were observed in 56 cases (42.4% of 132 hips), 39 consisted solely of displacements, 6 took the form of gutters and 8 were tunnels. There were 3 cases of common femoral vein thrombosis (Fig. 2d).

All cases of contact with nerves concerned the sciatic nerves. Osteomas were in contact with sciatic nerves in 36 cases (37.2% of 132 hips) including 12 displacements, 23 gutters and 3 tunnels (Fig. 3).

The joint capsule was respected in 48 cases (36.3%). It was touched by the osteomas in 61 cases (46.2%) and disrupted in 24 cases (18.2%).

In most of the cases, bone mineralization was normal (M1) (40.9%). We observed mild demineralization (M2) in 51 patients (38.6%), significant demineralization (M3) in 18 patients (13.6%) and severe demineralization (M4) in 9 patients (6.8%).

Hip joint space was normal in 82 (62%) of 132 hips, narrowed in 33 (25%) of 132 and ankylosed in 17 cases (13%).

Relationships between osteomas and vessels were well depicted in surgery reports. Seven displacements, 4 gutters and 6 tunnels were reported. Agreement with pre-surgical CT was excellent (0.82). Contacts with sciatic nerves were observed by the surgeons in 30 hips (12 displacements, 15 gutters and 3 tunnels). There was a good agreement with pre-surgical CT (0.62). Nine gutters depicted on CT were qualified solely as displacements by the surgeon.

A close contact to joint capsule was reported by the surgeon in 16 hips. Agreement with CT was good (0.68).

There were little complications regarding the difficulty of the surgery:

Excision of osteoma was incomplete in one patient due to the risk of vessel injury. Patients with M3 and M4 demineralizations presented a high risk of fracture. They had an appropriate management with gentle per- and post-surgical movements. Two patients had a surgical ablation of femoral head and two had osseous grafts on their femoral head. No fracture occurred per- or post-procedure. Mean follow-up after surgery was 10 months. Twelve patients were considered lost for follow-up.

NMO recurrence was observed in 7 patients.

The statistical analysis found an association between three characteristics and NMO recurrence: osteomas touching or disrupting the joint capsule (p = 0.005), joint space narrowing (p = 0.007) and demineralization (p < 0.001). Though not significant, there was a trend towards recurrence when time to surgery was longer.

Our study was retrospective. A future prospective study will certainly allow to confirm our data.

Second, the surgeons’ experience should certainly be taken into account in the good results aside from the imaging work-up. Furthermore, measurement of surgical blood loss would be an interesting parameter to evaluate in further studies.

Finally, relations to tendons should also be studied in future studies as contacts between heterotopic ossifications and adjacent tendons of psoas, gluteus medius and piriformis muscles were reported by the surgeons in 9 cases.

Though difficult to perform in all patients, MRI is undoubtedly the best imaging modality to assess relationships with tendons and tendons' integrity. It might thus be beneficial to conduct MRI following CT whenever the relationships between tendons and osteomas seem problematic for surgery and if MRI is technically possible.