Bone Marrow Edema Injury Patterns in the Pediatric Knee: An MRI Study

Abstract

Background:

Symptomatic pediatric patients referred for magnetic resonance imaging (MRI) commonly present with traumatic bone marrow edema (BME) patterns.

Purpose:

We sought to associate discrete MRI patterns of BME with specific injury mechanisms in pediatric knee injuries to classify injury patterns by anatomical location of the BME. We aimed to group these into 6 patterns: patellar dislocation, extensor mechanism overload, hyperextension, single compartment impaction, ligament avulsion/translation, and direct contusion.

Methods:

We retrospectively reviewed 314 MRIs performed with a standard protocol on symptomatic patients aged 3 to 18 years at 1 institution. Our analysis included images, reports, and traumatic BME patterns. A musculoskeletal radiologist and orthopedic surgeon independently assigned 1 of the 6 injury patterns to each scan.

Results:

After exclusion criteria were applied to the 314 MRIs, 62 (19.7%) remained, 40 boys and 22 girls. The average age was of 12.2 years. The most frequent injury patterns were patellar dislocation (n = 22, 35%) and extensor mechanism overload (n = 14, 22%). κ value associated with pattern determination was .766, indicating substantial concordance. Bone marrow edema signal intensity on fat-suppressed sequences was classified as severe in 92% of cases.

Conclusions:

The strength of pediatric knee ligaments and tendons relative to epiphyseal bone may contribute to a high rate of BME injury patterns seen on MRI in symptomatic pediatric patients. We found that pediatric BME could be classified into 6 specific injury patterns, which might be useful to clinicians in recognizing mechanisms of injury. Further clinical studies are needed to assess the clinical differences in both short-term and long-term outcomes of the BME patterns described.

Introduction

Knee injuries are common across all age groups, many resulting from accidents or sports-related events. Causes may include hyperextension, hyperflexion, varus, valgus, and anteroposterior translation. Contusions or externally applied forces also give rise to internal derangements of the knee. It is widely understood that specific injuries result from specific forces or combinations of forces. Although a precise history of the mechanism of injury is important, pediatric patients cannot always precisely recall their injuries, which can render impossible an accurate physical examination after an acute injury.

Magnetic resonance imaging (MRI) is an excellent tool in evaluating the musculoskeletal system for the presence of soft-tissue and bony abnormalities in the acute setting [11]. Bone marrow edema (BME) patterns are frequently identified with MRI after an acute musculoskeletal injury. They represent radiographically occult osseous injuries and appear as areas of poorly marginated signal intensity alteration in the cancellous bone marrow. Edema secondary to micro-trabecular fractures or injury is thought to result in these areas of increased intensity on short tau inversion recovery (STIR) and T2-weighted sequences [6,16]. Bone marrow edema patterns seen on MRI are thought to be a static representation of the impact that occurred at the time of injury—similar to a “footprint” left behind—and may be used to gain insight into the mechanism of injury [4,14]. Inferring a mechanism of injury enables the physician to predict with greater accuracy associated soft-tissue injuries that are likely present, as well as to plan appropriate treatment.

Pediatric knee ligaments and tendons exhibit high strength relative to epiphyseal bone, which may contribute to the high rate of BME patterns seen on MRI in symptomatic pediatric patients. We therefore sought to better characterize the frequency and patterns of bone bruises seen on MRI in the symptomatic pediatric knee, with the aim of classifying injury patterns by anatomical location of the bone bruise. We hypothesized that most traumatic pediatric BME patterns would follow 6 specific patterns: patellar dislocation, extensor mechanism overload, hyperextension, single compartment impaction (varus or valgus load), ligament avulsion/translation, and direct contusion. We also aimed to associate discrete MRI patterns of BME with specific injury mechanisms in pediatric knee injuries.

Methods

After obtaining Institutional Review Board approval, we retrospectively reviewed MRIs of patients aged 3 to 18 years with bone bruises. Of 314 MRIs performed with a standard protocol at 1 institution, there were 243 in skeletally immature patients. Patients with closed physes as well as BME secondary to osteochondral lesion and discernible fracture line on plain radiographs were excluded. Images and reports were reviewed, and traumatic BME patterns were analyzed. A traumatic BME pattern was identified in 95/314 (30.3%) of the studies.

Of the original 314 MRIs, 62 (19.7%) remained after exclusion criteria were applied: 34 left knees and 28 right knees. There were 40 boys and 22 girls, and the average age was 12.2 years. After review of all MRIs, an orthopedic surgeon and a musculoskeletal radiology fellow independently assigned 1 of the 6 mechanistic patterns to each scan upon review of the sagittal STIR sequences and based solely upon the BME patterns in the remaining 62 patients (19.7%). Each MRI was reviewed twice, first, blinded and second, independently.

Scanning was performed on either a 1.5-Tesla or 3-Tesla MRI scanner using a quadrature or an 8-channel phased-array knee coil with the knee positioned at near full extension. All examinations included at least the following sequences performed at 1.5 T: sagittal inversion recovery (IR) (repetition time [TR] 4000–6000, echo time [TE] 16 ms, inversion time [TI] 150 ms, echo train length [ETL] 12–16, matrix 288 × 192–288, field of view [FOV] 150–180 mm, number of excitations [NEX] 2) or sagittal T2 fat-suppressed (FS) (TR 3500–5000, TE 42–45 ms, ETL 13–16, matrix 288 × 224–288, FOV 150–180 mm, NEX 2). For studies performed on a 3-T scanner, the following parameters were used: sagittal IR (TR 4000–6000, TE 14–17, TI 150 ms, ETL 10–12, matrix 288 × 256–288, FOV 150–180 mm, NEX 1) or sagittal T2 FS (TR 4000–6000, TE 35–60 ms, ETL 12–16, matrix 288 × 256–288, FOV 150–180 mm, NEX 1). The section thickness in all scans ranged from 2.4 to 3.5 mm on sagittal images with no interspace gap.

Data analysis was performed using SPSS. A κ value representing the agreement between observers was calculated and analyzed according to previously described criteria [7]. Values of .81 to 1.00 indicate excellent concordance, .61 to .80 substantial concordance, .41 to .60 moderate concordance, .21 to .40 fair concordance, and 0 to .20 poor concordance.

Results

The most frequent patterns of BME identified on MRI were patellar dislocation (N = 22, 35.5%; Fig. 1) and extensor mechanism overload (N = 14, 22.6%; Fig. 2). We classified 9 (14.5%) knee injuries in the hyperextension group (Fig. 3), 7 (11.3%) in the varus/valgus load group (Fig. 4), and 5 each (8.1%) in the ligament avulsion/translation and direct contusion groups (Fig. 6). Out of the 7 in the single compartment impaction injury group, there were 5 valgus injuries and 2 varus. κ value associated with pattern determination for the 2 readers was 0.766, indicating substantial concordance. Bone marrow edema signal intensity was classified as severe in 92% of cases [19].

Fig. 1.

Patellar dislocation injury pattern. (a, b) Edema in the lateral femoral condyle in the area of impaction and (c, d) Edema pattern in the medial patellar facet in the area of impaction.

Fig. 2.

(a-c) High signal in the tibial tubercle related to stress reaction, extensor mechanism overload pattern of injury.

Fig. 3.

Hyperextension injury. Sagittal short tau inversion recovery magnetic resonance images show focally intense bone marrow edema pattern (a) over the anteromedial margins of the lateral femoral condyle and lateral tibial plateau and (b) over the anteromedial margins of the medial femoral condyle and medial tibial plateau.

Fig. 4.

Valgus injury pattern. There is extensive reactive signal (a) in the lateral tibial plateau and (b) in the lateral femoral condyle.

Fig. 6.

(a, b) Demonstration of the effects of a direct contusion injury with condensation of trabeculae and bone marrow edema pattern along the medial margin of the medial femoral condyle and (c, d) Accompanying overlying soft-tissue edema confirms contusion mechanism of injury.

Discussion

The majority of literature describing BME patterns in the knee has evaluated knee MRIs in skeletally mature patients after anterior cruciate ligament (ACL) injury [4,17,19]. In our study, 25.5% (62/243) of symptomatic skeletally immature patients who underwent MRI demonstrated a BME pattern that was unrelated to osteochondritis dissecans (OCD) or discernible fracture on plain radiographs. The corresponding percentage of the magnetic resonance (MR) studies performed in the skeletally mature patients in our series was 18.3% (13/71). We speculate that BME patterns are a common yet under-recognized finding associated with acute knee pain in the active child. The strength of pediatric knee ligaments and tendons relative to epiphyseal bone may contribute to a relatively high frequency of BME patterns in the skeletally immature population.

Our study has limitations. First, it was completed at a single center. In addition, it was limited by its retrospective design and lack of clinical follow-up. Although its aim was to classify injury pattern according to the anatomical location of the bone bruise, we did not know the exact injury mechanism. Nor did we have follow-up or patient-reported outcomes, which prevented us from reporting long-term outcomes. Further clinical studies are needed to assess the clinical differences in both short-term and long-term outcomes of the BME patterns described.

Patellar dislocation is a relatively common traumatic injury in the pediatric and adolescent population, representing 9% to 16% of injuries found on surgical exploration of the knee for traumatic hemarthrosis [12,15]. Bone marrow edema patterns are created at the time of reduction of the patella into the trochlear groove, resulting in increased signal in the anterolateral portion of the lateral femoral condyle and the medial patella (Fig. 1). This is commonly accompanied by injury to the medial patellofemoral ligament and medial retinaculum. Associated chondral and osteochondral injuries of the medial patella are also not infrequent [3]. Nearly, half of the cases with osteochondral defects have the classic finding of a concave impaction deformity of the inferomedial patella. Similar findings are seen on the lateral femoral condyle. Moreover, chondral fragments as intra-articular loose bodies may be present [3].

Extensor mechanism overload in pediatric patients is common and encompasses such injuries as patellar sleeve fractures, tibial tubercle stress reactions, and jumper’s knee with stress reaction in the inferior pole of the patella (Fig. 2). Small avulsions of the tibial tubercle may also be diagnosed on MR examination.

Hyperextension injuries to knee result most commonly when a direct force is applied to the anterior tibia with the foot statically planted. At the point of maximum hyperextension, the anterior tibial plateau and anterior aspect of the femoral condyle come in contact with one another producing a BME pattern referred to as a “kissing” lesion (Fig. 3). This can be located in the medial or lateral compartments, when the hyperextension force is accompanied by a respective varus or valgus force, respectively. It can also be present in both compartments if a pure hyperextension mechanism occurs. Varying amounts of hyperextension forces will result in associated soft-tissue abnormalities from posterior capsular stretch, meniscal injury, posterior cruciate ligament (PCL) avulsion, and ACL injury to frank knee dislocation [1,2,18,20]. Moreover, injuries to the posterolateral complex may be associated with hyperextension injury [10].

Single compartment impaction injury patterns are divided into 2 types: injuries to the medial tibiofemoral compartments occur when a varus load is imparted to the knee joint, while injuries to the lateral tibiofemoral compartments occur when a valgus load is imparted to the knee joint. With a varus load, the pediatric knee tends to exhibit BME patterns within the medial articular distal femur and medial articular proximal tibia. This mechanism is analogous to that in the adult which would result in a lateral collateral ligament injury or posterolateral corner injury without the frank ligamentous injury. After a valgus load, BME patterns are observed in the lateral distal articular femur and the lateral proximal articular tibia (Fig. 4). In the adult patient, this is commonly accompanied by injury to the medial collateral ligament (MCL), whereas in the pediatric population it would be more likely to observe increased BME signal at the femoral or tibial insertions of the MCL without frank disruption of the ligament.

Ligament avulsion and translational events of the tibiofemoral joint result in a spectrum of BME patterns and injury in the skeletally immature patient. The classic translational pattern is observed in the skeletally mature and immature populations, involving edema of the posterior aspect of the lateral tibial plateau and the midportion of the lateral femoral condyle near the sulcus terminalis (Fig. 5). In the patient near skeletal maturity, this is more commonly combined with rupture through the midsubstance of the ACL. The skeletally immature patient exhibits relatively high strength of the ACL compared with that of the tibial epiphyseal bone. This results in phenomena unique to the pediatric population: tibial spine edema pattern and BME evidence of a translational event without ligament disruption (Fig. 5). In the tibial spine edema pattern, the edema is observed in the proximal tibia in a focal area around the anterior medial tibial spine, which we believe to be the MR precursor to a type I tibial spine fracture seen on plain radiographs as described by Myers and McKeever [9,8]. Avulsions of the femoral insertion of the ACL are also seen, although less frequently.

Fig. 5.

Ligament avulsion/translation. Bone marrow edema pattern is seen at the sulcus terminalis of the lateral femoral condyle and posterior aspect of the lateral tibial plateau related to a (a) recent pivot shift injury and (b) intact anterior cruciate ligament.

Contusions result from any external force to the knee and are commonly secondary to patellar impaction with a fall onto the knee. The defining aspect of this type of injury pattern is the presence of soft-tissue edema directly overlying the bony edema (Fig. 6). Any focal bone bruise not associated with other patterns described should raise suspicion of such an injury mechanism.

We do not propose that the MR patterns we describe be used as a formal orthopedic classification. Rather, we believe these 6 groupings may be a useful tool to help clinicians to categorize BME patterns seen on MRI of young patients with knee pain. Understanding the mechanism of injury allows the physician to predict, with greater accuracy, associated soft-tissue injuries and to perform a focused physical examination. Furthermore, accurate diagnosis will allow for the appropriate direction of post-injury rehabilitation. Such a mechanism-based approach to describing BME patterns in the pediatric knee has not previously been reported.

Clinically, we would expect the patella dislocation group and, to a lesser extent, the extensor mechanism overload and ligament avulsion/translation groups would require a longer period of conservative treatment. We would expect that symptoms would resolve within 4 weeks of conservative treatment in a majority of patients in the hyperextension, varus or valgus load, and direct contusion groups.

In a mixed cohort of 761 patients, without evidence of soft-tissue injury or of fracture on plain radiographs, Oeppen et al [13] showed that 19% (143/618) presented with a BME pattern. The researchers also confirmed that younger patients may be more likely to present with a BME pattern in isolation versus with a concomitant soft-tissue injury.

Bone marrow edema patterns seen on MRI are thought to be a static representation of the impact that occurred at the time of injury, similar to a “footprint” left behind, and may be used to gain insight into the mechanism of injury [4,14]. This is particularly useful when the patient cannot provide a complete history of an injury, which is common in children.

Conclusion

In conclusion, we found that each traumatic pediatric BME pattern followed 1 of 6 specific patterns. Whereas we observed the patellar dislocation pattern as the most common BME pattern in our cohort of pediatric patients, in a different cohort of 100 skeletally mature patients, the majority of the patients (46%) presented with BME patterns consistent with a flexion valgus and external rotation mechanism [5]. In our study, 25.5% (62/243) of symptomatic skeletally immature patients who underwent MRI demonstrated a BME pattern that was unrelated to an OCD or discernible fracture on plain radiographs. These findings suggest that subtle acute bone injuries (as seen on MRI) are a common but not well-characterized cause of acute knee pain in the active child. More rigorous study is warranted.