Incidence and Imaging Findings of Costal Cartilage Fractures in Patients with Blunt Chest Trauma: A Retrospective Review of 1461 Consecutive Whole-Body CT Examinations for Trauma

Introduction

After head injuries, blunt chest trauma is the second leading cause of morbidity and mortality in patients with trauma (1). Although radiologic findings of chest trauma and chest wall disorders have been widely described in the literature (2–9), costal cartilage (CC) fractures are rarely mentioned (10–14). Contact sports–related costochondral injuries are well known and usually discovered at magnetic resonance (MR) imaging (15–20). As of today, the true incidence, clinical significance, and long-term effects of these injuries in patients with blunt polytrauma remain unknown. CC fractures contribute to rib cage instability and may manifest clinically as late as weeks or months after the acute trauma (16). The data on the healing process of CC fractures remains sparse (21).

Conventional radiography cannot reveal CC fractures, but computed tomography (CT), MR imaging, and ultrasonography have proven to be useful in their depiction (10–12,22,23). Previous publications of CC injuries have focused on posttraumatic evaluation weeks or months after the initial trauma (10,11). Imaging has been used to help determine severe posttraumatic parasternal pain unexplained by radiographic findings, or because of a painful parasternal mass with clinical suspicion of tumor in patients without obvious recent trauma (10). Several case reports of contact sports–related costochondral injuries in rugby and American football players have also been published (15–20). Recent retrospective research over a period of 10 years found a total of 44 patients with CC injuries seen at CT. However, this review was based on a retrospective keyword search in the written reports, thereby excluding any missed injuries, and no incidence was reported (24).

Because patients with polytrauma have multiple severe and potentially life-threatening concomitant injuries by definition, CC fractures add to the trauma burden by increasing thoracic instability. Thus, it is important to pay attention to these injuries. To our knowledge, there are no previously published reports of the incidence of CC fractures in a large cohort of patients with blunt polytrauma. The purpose of our study was to assess the incidence of CC fractures in whole-body CT examinations for blunt trauma and to evaluate distribution of CC fractures, concomitant injuries, mechanism of injury, accuracy of reporting, and the effect on 30-day mortality.

Materials and Methods

Patients

Institutional review board approval for this retrospective study was obtained from the ethics committee of the Department of Surgery. All acute-phase whole-body CT examinations for blunt trauma (1461) from the emergency unit of a single large level I trauma center during a period of 36 months (January 1, 2013–December 31, 2015) were consecutively included in the study. All patients aged 18 years and older with a history of blunt trauma and who underwent a contrast material–enhanced chest and abdominopelvic CT examination for trauma were included regardless of their injury severity score. Patients for whom emergency imaging indication was other than blunt trauma—search for primary malignancy, metastatic disease, source of fever, or focus of infection—were excluded from our study (6.7%, 105 of 1566). Patients with penetrating trauma of the torso are referred and transferred to another departmental unit in this catchment area and are not treated in the level I trauma center in question.

At the time of imaging, all examinations were primarily reviewed by either an on-call resident or specialist of radiology, followed by a second reading on the following day by an attending board-certified trauma radiologist with more than 5 years of experience.

CT Imaging and Image Analysis

Imaging was performed by using a 64-section CT scanner (Discovery CT750 HD; GE Medical Systems, Milwaukee, Wis) with the following image sets available in picture archiving and communication systems (IMPAX 6.0; Agfa HealthCare, Mortsel, Belgium) and used for review: axial section thickness of 2.5 mm and standard reformatted images at 2 mm sagittal, 4 mm coronal, and (starting from May 2015) 3 mm in both sagittal and coronal reformated images. A board-certified radiologist (M.N., with 5 years of experience) was blinded to the initial reports and reviewed all whole-body CT examinations by using picture archiving and communication systems. In the first phase of this retrospective reading, all whole-body CT examinations were classified by reader 1 (M.N.) into subgroups of chest trauma, nonchest trauma, and no signs of trauma based on the findings. Special attention was paid to the CCs (Fig 1) and ribs. In the second phase of our study, all whole-body CT examinations with signs of chest trauma were independently reviewed by a second reader (F.B., a board-certified trauma radiologist with 12 years of experience) to determine the interobserver agreement for cartilage fracture detection. Reader 2 was blinded to initial reports and to the opinion of reader 1. Any disagreements (9.9%, 57 of 574) between readers 1 and 2 were subsequently resolved by consensus. Interobserver agreement was calculated with weighted kappa values.

Figure 1:

Download as PowerPointOpen in Image Viewer

Patients with signs of chest trauma (any of the following: CC fracture, rib fracture, sternal fracture, pneumothorax, hemothorax, pulmonary contusions, pulmonary lacerations, mediastinal hematoma, pneumomediastinum, aortic injury, chest wall hematoma) as categorized by reader 1 were further categorized by both readers 1 and 2 into two subgroups: patients with CC fractures and patients without CC fractures. Any isolated fracture of the thoracic spine, clavicle, or scapula was not included in the chest trauma group.

A nondisplaced CC fracture was identified as a hypoattenuating fracture line across the cartilage seen in more than one image and most commonly oriented perpendicular to the rib margins. Cartilages were assessed on axial and coronal images. A step-off on the even surface of the cartilage and often overlapping cartilage fragments reveal the dislocated fracture on axial images. Also, hematoma surrounding the dislocated costochondral fracture can facilitate detection.

Variables

Cartilage fracture location, cartilage calcification, and the presence of concomitant intrathoracic, skeletal, and intra-abdominal injuries were recorded. Depending on their location, CC fractures were categorized as costochondral, midchondral, or chondrosternal. Costochondral separation (ie, costosternal dissociation) was categorized as chondrosternal fracture. Costochondral fractures that were agreed on in the consensus reading were subsequently compared with the original report to determine the initial detection rate of CC fractures.

Measuring the detection rate of other variables was beyond the main focus of this article. However, reporting of intrathoracic injuries, sternal fractures, and rib fractures was screened by reader 1. Overall, the presence of concomitant intrathoracic, skeletal, and intra-abdominal injuries were recorded on the basis of the original reports and the reading of all whole-body CT examinations by reader 1.

The mechanism of injury and additional clinical information, along with the 30-day mortality rate, were obtained from patients’ medical records. Any available follow-up CT study covering the area of injured cartilages was also reviewed by readers 1 and 2 to evaluate posttraumatic changes at the fracture site. All of these follow-up examinations were performed for different indications than CC fractures—for example, for the evaluation of pulmonary infection (pneumonia, empyema), size of pneumothorax or hemothorax, retained pleural fluid, pulmonary embolism, abdominal emergency, follow-up after blunt cerebrovascular injury, or a successive event of blunt trauma.

Statistical Analysis

χ² test and odds ratios (ORs) with 95% confidence intervals (CIs) of different mechanisms of trauma, associated injuries, and 30-day mortality between patients with and patients without CC fractures were calculated. Two-tailed t test for independent samples was used to assess the age difference between sexes. Logistic regression models for dichotomous outcome variable were used to find out whether costochondral fractures are an independent predictor for 30-day mortality. Statistical significance was set at .05. Interobserver agreement was calculated with linearly weighted kappa values. Magnitude for interobserver agreement was interpreted as follows: values less than 0, no agreement; 0–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; and 0.81–1, almost perfect agreement (25). We used a commercial software package (SAS/STAT, version 9.3; SAS Institute, Cary, NC).

Results

Over a period of 36 months, a total of 1461 consecutive whole-body CT examinations met the inclusion criteria. Radiologic signs of chest trauma were found in 574 of 1461 (39%) examinations (mean age, 47.2 years; range, 18–97 years; men, 74.0% [425 of 574]; mean age, 46.6 years; range, 18–91 years; women, 26.0% [149 of 574]; mean age, 48.9 years; range, 18–97 years). The age difference between men and women was not statistically significant (P = .213).

Among the 574 patients with chest trauma, a total of 114 patients had 221 CC fractures (mean age, 48.2 years; range, 18–84 years; men, 86.8% [99 of 114]; mean age, 48.6 years; range, 18–84 years; women, 13.2% [15 of 114]; mean age, 45.1 years; range, 20–74 years). The male-to-female OR for having a CC fracture was 2.71 (95% CI: 1.52, 4.84; P = .0005).

Incidence

The incidence of CC fractures was 7.8% (114 of 1461) in all whole-body CT examinations and 19.9% (114 of 574) in patients with chest trauma. Altogether, 221 CC fractures were found in 114 patients and categorized as either midchondral (52.9%, 117 of 221), costochondral (41.2%, 91 of 221), or chondrosternal (5.9%, 13 of 221) (Figs 2–4). The CCs of rib 7 (21.3%, 47 of 221), rib 6 (14.9%, 33 of 221) (Fig 2), and rib 1 (17.2%, 38 of 221) (Fig 3) were most commonly injured (Table 1).

Figure 2a:

Download as PowerPointOpen in Image Viewer

Figure 2b:

Download as PowerPointOpen in Image Viewer

Figure 3:

Download as PowerPointOpen in Image Viewer

Figure 4:

Download as PowerPointOpen in Image Viewer

Table 1 Distribution of CC Fractures by Location per Rib

Note.—Unless otherwise indicated, data are the number of fractures; data in parentheses are percentages.

Fourteen patients of 114 (12.3%) had a CC fracture without osseous rib fractures. Multiple CC fractures were found in 60 of 114 patients (52.6%) (Figs 2a, 5a) and 17 of 114 patients (14.9%) had bilateral fractures (Fig 5a). Only two of 114 patients (1.8%) had an isolated CC fracture without any concomitant injuries. Pseudoarthrosis-type calcifications and the cleftlike appearance of the first rib costochondral joint were occasionally mistaken for signs of fracture (Fig 6). In one patient, posterior dislocation of the second costosternal joint was associated with cardiac contusion (Fig 4). Available follow-up CT examinations of 17 patients showed posttraumatic calcifications adjacent to or inside the fracture cleft (Fig 5b) starting from 27 days after trauma. The average length of duration to first follow-up CT (for any indication) that showed the calcifications was 160 days (range, 27–595 days).

Figure 5a:

Download as PowerPointOpen in Image Viewer

Figure 5b:

Download as PowerPointOpen in Image Viewer

Figure 6:

Download as PowerPointOpen in Image Viewer

Concomitant Injuries

Associated intrathoracic injuries were pneumothorax (65.8%, 75 of 114), pulmonary contusions (52.6%, 60 of 114), hemothorax (51.8%, 59 of 114), and pulmonary lacerations (17.5%, 20 of 114). Pneumomediastinum was found in 7.9% (nine of 114) (Table 2). No internal mammary or subclavian artery injuries were observed (Figs 3, 5a). Blunt traumatic aortic injury was rare (3.5%, four of 114). Associated osseous injuries were fractures of the thoracic spine (24.6%, 28 of 114), sternum (24.6%, 28 of 114), clavicle (21.9%, 25 of 114), and scapula (10.5%, 12 of 114). A four-column fracture of the thoracic spine, including fractured sternum and associated rib and costochondral fractures, was found in 9.6% (11 of 114). Multiple consecutive rib fractures occurred in 92 of 114 patients (80.7%), of which 41 fractures were bilateral. Intra-abdominal injuries were seen in 22.8% of patients (26 of 114), including the liver (n = 15), kidneys (n = 11), and spleen (n = 7) (Table 2).

Table 2 Additional Injuries

Note.—Unless otherwise indicated, data are the number of patients, with percentages in parentheses.

^*Patients with radiologic signs of chest trauma but no CC fractures.

^†Data in parentheses are 95% CIs.

Pneumothorax (OR, 2.88; 95% CI: 1.88, 4.43), hemothorax (OR, 3.37; 95% CI: 2.21, 5.16), and aortic injury (n = 4 vs n = 0; OR, unavailable) were significantly more common in patients with CC fractures than in patients without CC fractures (P < .0001). The same finding applied to bilateral multiple consecutive rib fractures (OR, 3.48; 95% CI: 2.18, 5.53; P < .0001), sternal fractures (OR, 2.50; 95% CI: 1.50, 4.18; P = .0003), clavicle fractures (OR, 2.03; 95% CI: 1.20, 3.42; P = .0074), and thoracic spine fractures (OR, 1.75; 95% CI: 1.07, 2.88; P = .0248). Injuries to the liver (OR, 3.72; 95% CI: 1.81, 7.64; P = .0001) were also more frequent in the CC fracture group, but splenic injuries were equally common in both groups (OR, 1.01; 95% CI: 0.43, 2.37; P = .9830) (Table 2).

Mechanism of Injury, Detection Rate, and Effect on Mortality

Most common trauma mechanisms were motor vehicle crashes (34.2%, 39 of 114) and falls (28.1%, 32 of 114) (Table 3). Isolated CC fractures were caused by either direct blunt impact or momentary compression force to the upper abdomen and/or lower thorax area. No significant difference was found in the prevalence of different mechanisms of injury between the two chest trauma groups, although fall was a less common trauma mechanism in patients with costochondral fractures than in patients with chest trauma without costochondral fractures. Costochondral fractures were documented in 39.5% (45 of 114) of initial reports, when the result of the consensus reading was compared with the initial reports. Nondislocated fractures were most commonly missed. Nondislocated osseous rib fractures were either missed or only partially mentioned in 45 of 413 (10.9%) patients with multiple rib fractures, and sternum fracture were missed in nine cases of 81 (11.1%). Only one small-sized pneumothorax was left undetected (0.04%, 1 of 259). Calculated with χ² test, no statistically significant difference was found in the 30-day mortality in patients with CC fractures (7.02%, 8 of 114) compared with patients without CC fractures (4.78%, 22 of 460) (OR, 1.50; 95% CI: 0.65, 3.47; P = .3371), and CC fracture was not an independent predictor for 30-day mortality (logistic regression analysis; P = .3886).

Table 3 Mechanism of Injury

Note.—Unless otherwise indicated, data are the number of patients, with percentages in parentheses.

^*Patients with radiologic signs of chest trauma but no CC fractures.

^†Data in parentheses are 95% CIs.

Interobserver Agreement

All chest trauma examinations (574) that were categorized by reader 1 were reviewed by a second reader (reader 2) to determine interobserver agreement on detection of CC injuries. All differences were reassessed and settled by using consensus. Both reviewers initially agreed on 517 of 574 examinations (90.1%); these included 436 of 460 (94.8%) negative findings and 81 of 114 (71.1%) positive findings for CC fractures. These detection rates were not separately compared with initial reports; only the final result of the consensus reading was compared with the initial reports. Linearly weighted kappa value for interobserver agreement was 0.56 (95% CI: 0.48, 0.64; moderate agreement). Evidence of a learning curve was observed in detecting costochondral fractures, because frequency of disagreement between the two readers decreased from the early to the late phase of the study. Nondislocated fractures were more difficult to both detect and assess in the acute phase of injury. Some of the fractures were more easily identified at follow-up imaging as a result of increased dislocation. Streak artifact from intravenous contrast agents or dense cortical bone and subtle hypodense irregularities inside the cartilage can mimic a nondislocated fracture. Also, the large variance in the appearance of the first rib synchondrosis makes the detection of nondislocated fractures in this particular area challenging.

Discussion

CC fractures are common in blunt chest trauma. However, the incidence of CC fractures in patients with blunt trauma has not been previously reported. Most commonly fractured cartilages in our study cohort were in ribs 1, 6, and 7, located at the upper and lower margins of the anterior chest wall.

Fracture of the first rib costochondral junction was commonly associated with posterior first rib fractures. The first rib differs anatomically from the others by having a short radius and by forming a synchondrosis with the manubrium of the sternum. Different stages of calcification and/or ossification in the first rib synchondrosis and the often cleftlike appearance of the costosternal joint complicates the evaluation of acute signs of trauma in this location (5,8,26–28). Fractures of the first rib have been reported to indicate high trauma energy (6,29,30), a finding that our study cohort confirms, and which we suggest to be applicable also to the CC fracture of the first rib.

The absence of vascular injuries to the subclavian vessels or the internal mammary arteries in our study implies that these structures are not at immediate risk. Fractures of the third and fourth CCs and/or ribs were associated with traumatic aortic injury, all of which originated at the most common site for these injuries, the aortic isthmus (31). In one patient, posterior dislocation of the second costosternal joint was associated with cardiac contusion (32). These vascular injuries most likely indicate thoracic deformation during the initial trauma that far exceeds the extent of deformation seen at the time of imaging.

Midchondral fractures of CCs 6 and 7 were associated with cases of multiple consecutive rib fractures and concomitant abdominal injuries. On CT images, abdominal parenchymal injuries may indicate a CC injury and vice versa. Injuries to the subcostal angle (cartilages of ribs 6 and 7) were overall most common and sometimes presented as isolated injuries as a result of momentary local compression force to the lower thorax.

In general, CC fractures are highly associated with multiple consecutive rib fractures and severe intrathoracic injuries such as pneumothorax, hemothorax, and pulmonary contusions. Half of all patients had multiple cartilage fractures. Multiple consecutive posterior rib fractures together with CC fractures increase thoracic instability and may form a large flail chest segment, thus impairing the patient’s respiratory function. Fractures of the sternum, as an adjacent structure, were significantly more common in the costochondral fracture group than in patients without cartilage injuries. Also, unstable hyperflexion fractures of the thoracic spine were accompanied with anterior chest wall (ie, costochondral and sternal) fractures in this cohort, further emphasizing that the thoracic cage—with all of its components—should be viewed as a single functional unit.

CC fractures commonly occur in high-energy trauma, with the most common mechanisms of injury being motor vehicle crashes and falls. Trauma mechanisms did not differ significantly between patients with chest trauma with or without costochondral injuries. However, the higher incidence of concomitant injuries in patients with CC fractures indicates its connection to higher trauma energies.

Fractures of the first rib synchondrosis were accurately identified in most of the on-call reports, whereas CC fractures in the lower parts of the anterior chest wall often remained undetected. The radiologic signs of CC fractures may not be as familiar as the imaging findings of osseous rib fractures. Thus, noticing these injuries seems to be more of a matter of perception than detection. Because fractured cartilage ends are usually smooth and well delineated, it is difficult to detect the fracture. Also, the abundance of other injuries in patients with polytrauma may cause satisfaction of search, whereby the finding of one abnormality on an image results in a second abnormality being overlooked (33). We found coronal reformatted images to be useful in the detection of CC fractures. This finding is in accordance with a study on evaluating missed rib fractures in multidetector CT examinations, in which the use of coronal reformatted images was recommended for detecting the commonly missed buckled anterior rib fractures (34).

Even though the majority of thoracic injuries are treated conservatively, the knowledge of CC injuries may explain persistent pain in patients with chest trauma and may guide pain relief protocol and ventilation support analogous to the management of flail chest and multiple consecutive rib fractures. Whether these fractures truly heal to form a stable union is not entirely known, although dense calcifications are frequently seen at the fracture site in follow-up CT examinations, which suggests a solid union. Particularly concerning dislocated fractures, cartilage healing process and the restoration of chest wall stability over time still call for further investigation (21).

Limitations of our study include that it was a retrospective review. The data were collected over a 3-year period at a single large level I trauma center covering a population of 1.6 million people, providing a cohesive and representative sample. Patients were not selected based on injury severity score; all patients older than age 18 years who were assigned to and underwent a whole-body CT examination for blunt trauma were included in our study. The effect of CC fracture on 30-day mortality would require a more comprehensive analysis of other injuries to calculate the well-known prognostic indexes of Trauma and Injury Severity Score and Revised Injury Severity Score. All whole-body CT examinations were reviewed twice during the acute phase; first by an on-call resident or a board- certified radiologist in the acute phase, and reread by an experienced trauma radiologist within 24 hours.

Another limitation was the prescreening of the whole-body CT examinations that was performed by reader 1. Reader 2 was given a prescreened subset of 574 chest trauma examinations, from the cohort of all 1461 whole-body CTs, to determine the interobserver agreement on costochondral fracture detection. This setting might create a somewhat biased assessment for the detection of CC fractures; however, this method was chosen because the main purpose of our study was to focus on the incidence of CC fractures (not on all signs of chest trauma, which are generally well known). Overall, this retrospective study constitutes the third reading of all whole-body CTs and the fourth reading of the whole-body CTs with findings of chest trauma.

In conclusion, CC fractures are common in high-energy blunt chest trauma and often occur with multiple consecutive rib fractures. Patients with costochondral fractures are at higher risk of associated intrathoracic and liver injuries. CC fractures often remain overlooked. In the presence of multiple rib fractures, cartilage fractures should be suspected and actively searched for when evaluating chest wall instability.

Author Contributions

Author contributions: Guarantors of integrity of entire study, M.T.N., F.V.B.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; approval of final version of submitted manuscript, all authors; agrees to ensure any questions related to the work are appropriately resolved, all authors; literature research, all authors; clinical studies, T.T.P., S.K.K.; statistical analysis, M.T.N., F.V.B., S.K.K.; and manuscript editing, all authors