Multiplanar reconstruction: a new method for the diagnosis of tracheobronchial rupture?

Abstract

To compare multiplanar reconstruction with operative techniques (bronchoscopy, surgery and/or autopsy) for the diagnosis of tracheobronchial rupture.

Prospective, observational study.

Surgical intensive care unit.

Tracheobronchial rupture was suspected on clinical grounds and from radiological findings.

An initial helical computed tomography scan was performed on all patients meeting the inclusion criteria, and operative techniques were then performed. Multiplanar reconstructions were reformatted and reviewed by two independent radiologists.

Twenty-four consecutive patients met the inclusion criteria. Tracheobronchial rupture was diagnosed in 13 patients by at least one operative technique. Multiplanar reconstructions were positive in 15 patients. The diagnostic sensitivity and specificity of multiplanar reconstructions were 100% (95%CI, 85-100) and 82% (95%CI, 64-82), respectively. The positive and negative predictive values were 87% (95%CI, 74-87) and 100% (95%CI, 78-100), respectively. For tracheobronchial rupture, the positive and negative likelihood ratios were 5.5 (95%CI, 2.35-5.5) and 0 (95%CI, 0-0.24), respectively. The Kappa coefficients were 0.83 (95%CI, 0.6-1.06) for agreement between operative techniques and multiplanar reconstruction, and 0.91 (95%CI, 0.59-0.91) for agreement between the two radiologists.

Multiplanar reconstruction appears to be a sensitive technique for the identification of tracheobronchial rupture because of its excellent negative likelihood ratio. In clinical practice, negative multiplanar reconstruction can exclude a diagnosis of tracheobronchial rupture, making bronchoscopy unnecessary. When multiplanar reconstruction is positive, tracheobronchial rupture should be confirmed by bronchoscopy.

Trauma. [PUBLICATION ABSTRACT]

Intensive Care Med (2007) 33:21732178DOI 10.1007/s00134-007-0830-9 BRIEF REPORT

Multiplanar reconstruction: a new method for the diagnosis of tracheobronchial rupture?

Alexandre Faure Bernard Floccard Frank Pilleul Frdric Faure Bruno Badinand Nicolas Mennesson Thierry Ould Christian Guillaume Albrice Levrat Farida Benatir Bernard Allaouchiche

Abstract Objective: To compare multiplanar reconstruction with operative techniques (bronchoscopy, surgery and/or autopsy) for the diagnosis of tracheobronchial rupture. Design: Prospective, observational study. Setting: Surgical intensive care unit. Patients and participants: Tracheobronchial rupture was suspected on clinical grounds and from radiological ndings. Interventions: An initial helical computed tomography scan was performed on all patients meeting the inclusion criteria, and operative techniques were then performed. Multiplanar reconstructions were reformatted and reviewed by two independent radiologists. Measurements and results: Twenty-four consecutive patients met the inclusion criteria. Tracheo-bronchial rupture was diagnosed in 13 patients by at least one operative technique. Multiplanar reconstructions were positive in 15 patients. The diagnostic sensitivity and specicity of multiplanar reconstructions were 100% (95%CI, 85100) and 82% (95%CI, 6482), respectively. The positive and negative predictive values were 87% (95%CI, 7487)

and 100% (95%CI, 78100), respectively. For tracheobronchial rupture, the positive and negative likelihood ratios were 5.5 (95%CI, 2.355.5) and 0 (95%CI, 00.24), respectively. The Kappa coefcients were 0.83 (95%CI, 0.61.06) for agreement between operative techniques and multiplanar reconstruction, and 0.91 (95%CI, 0.590.91) for agreement between the two radiologists. Conclusions: Multiplanar reconstruction appears to be a sensitive technique for the identication of tracheobronchial rupture because of its excellent negative likelihood ratio. In clinical practice, negative multiplanar reconstruction can exclude a diagnosis of tracheobronchial rupture, making bronchoscopy unnecessary. When multiplanar reconstruction is positive, tracheobronchial rupture should be conrmed by bronchoscopy. Descriptor: Trauma.

Keywords Diagnostic techniques and procedures Multiplanar reconstruction Wounds and injuries Airway rupture Helical CT Bronchoscopy

Received: 31 January 2007 Accepted: 19 July 2007 Published online: 8 August 2007 Springer-Verlag 2007

Support was provided only by institutional sources.

A. Faure B. Floccard B. Badinand T. Ould C. Guillaume A. Levrat F. Benatir B. Allaouchiche ( ) Hpital Edouard Herriot, Dpartement dAnesthsie-ranimation, Hospices Civils de Lyon,Place dArsonval, 69437 Lyon Cedex 03, Francee-mail: [email protected] Tel.: [phone]Fax: [phone]

F. Pilleul N. Mennesson Hpital Edouard Herriot, Fdration de Radiologie, Hospices Civils de Lyon, Place dArsonval, 69437 Lyon Cedex 03, France

F. FaureHpital Edouard Herriot, Service dORL, Hospices Civils de Lyon, Place dArsonval, 69437 Lyon Cedex 03, France

Introduction Tracheobronchial rupture (TBR) is a rare but potentially life-threatening event [1]. The diagnosis of TBR is not always straightforward and can often be missed in trauma

patients [2, 3]. This delay in diagnosis can have a signi-cant effect on outcome [4].

There is no gold standard for the diagnosis of TBR and each technique can miss injury. Bronchoscopy is usually used to diagnose and plan the management of TBR [2, 57]. However, this invasive procedure has distinct prac-

2174

tical limitations and several potential risks in polytraumatized patients [5, 812].

Multidetector computed tomography (CT) scans do not provide enough evidence to diagnose TBR [1315]. Nevertheless, a recent computed-based method of data reformation allowing 2-D multiplanar reconstruction (MPR) represents a signicant advance for the diagnosis of TBR [14]. Some case reports and two retrospective studies have described the successful use of this technique in the management of TBR [10, 1519]. However, to date there have been no prospective studies evaluating this technique in the diagnosis of TBR.

The aim of this prospective study was to assess the predictive value of MPR for the diagnosis of TBR.

Materials and Methods Study sample

This was a single center, prospective, observational study approved by the hospital ethics committee. All suspected TBR patients admitted to our intensive care unit (ICU) between March 2002 and August 2004 were included. TBR was suspected if the patient fullled at least one criterion from each of the lists of clinical and radiological ndings showninTable1[11].

Data collection

Age, sex, APACHE II score, Injury Severity Score (ISS), diagnostic methods, CT scan ndings, cause and location of injury and outcome were recorded for each patient.

First, initial CT images from the neck to the diaphragm were obtained using a Somatom scanner (Somatom Plus 10; Siemens, Erlangen, Germany), with collimation of 3 mm, table speed of 22 mm/s, pitch of 5.5, and reviewed

Findings n (%)

Clinical ndings

Cervicothoracic subcutaneous emphysema 12 (50) Dyspnea or airway distress 8 (33.3) Hemoptysis 6 (25) Air leak from penetrating wound or thoracic tube 3 (12.5) Difcult intubation 2 (8.3) Cervical ecchymosis 1 (4.2) Radiological ndings

Subcutaneous emphysema 3 (12.5) Unilateral or bilateral pneumothorax resistant to thoracic tube 5 (20.8) Pneumomediastinum 4 (16.7) Paratracheal or parabronchial air 1 (4.2) Tracheal distortion, extraluminal location of endotracheal tube, 1 (4.2) herniation or overdistension of endotracheal tube balloon First rib fractures 4 (16.7)

Patients were enrolled if they fullled at least one criterion from each list

on a workstation (Leonardo; Siemens, Erlangen, Germany). Multidetector CT images were studied for direct visualization or indirect signs of TBR.

Airway endoscopy was then performed by a trained physician. If the patient underwent immediate surgical exploration before bronchoscopy, the surgical ndings were reported. If the patient died before bronchoscopy or surgical exploration, an autopsy was performed. The denitive diagnosis of TBR was considered to be direct visualization of injury during at least one of the operative techniques (OT) (bronchoscopy, surgery or autopsy).

Second, MPRs were reformatted from the initial axial CT data set using specialized software (Leonardo). MPR images were displayed in the coronal and sagittal planes, orthogonal to a point of reference, or in a curved fashion along the axis of the airway, and reviewed by two independent radiologists for direct visualization of TBR. Radiologists were trained in the use of the software for 2 hours. MPR criteria for the diagnosis of TBR were discontinuity or a defect of the tracheal/bronchial wall (positive MPR) (Fig. 1 and 2). When there was no agreement between radiologists, MPR was considered to be positive.

An expert committee (two surgeons, two radiologists and two intensivists) reviewed each report in a blinded fashion to conrm the positive diagnosis and anatomic location of TBR.

Statistical analysis

All statistical tests were performed with JMP software version 5.1 (SAS, Cary, USA). Assuming that the sensitivities of OT and MPR for the positive diagnosis of TBR are similar (99%), with a maximal difference between the two techniques of 10% (alpha risk 5%, beta risk 20%, power 80%), we calculated that a sample size of 24 patients was required to demonstrate equivalence for the diagnosis of TBR.

Table 1 Clinical and radiological ndings used as inclusion criteria

2175

Fig. 2 Sagittal 2-D multiplanar reconstructions with anterior tracheal air (white arrow) after gunshot injury

Table 2 Summary of patient data

Case Sex APACHE II Mechanism Site of injury Operative technique MPR number of injury

1 Male 30 Blunt Right main bronchus Endoscopy P2 Male 9 Blunt None Endoscopy P3 Female 14 Iatrogenic None Endoscopy N4 Female 25 Iatrogenic Anterior wall cervical trachea Endoscopy P5 Male 7 Penetrating None Endoscopy and surgery N6 Male 28 Blunt Postero-lateral wall thoracic trachea Endoscopy P7 Male 40 Blunt Right lateral wall thoracic trachea Endoscopy P8 Male 18 Blunt Anterior wall thoracic trachea Autopsy P9 Male 21 Penetrating Anterior wall cervical trachea Surgery P10 Male 8 Penetrating Posterior wall cervical trachea Endoscopy P11 Male 18 Penetrating Anterior wall cervical trachea Surgery P12 Male 21 Blunt Posterior wall thoracic trachea Endoscopy P and left main bronchus13 Male 27 Blunt None Surgery and autopsy P14 Female 9 Penetrating Anterior wall cervical trachea Endoscopy and surgery P15 Male 7 Blunt Postero-lateral wall thoracic trachea Endoscopy P and right main bronchus16 Female 34 Blunt None Endoscopy N17 Female 32 Iatrogenic Right lateral wall thoracic trachea Endoscopy P and right main bronchus18 Male 3 Penetrating None Endoscopy and surgery N19 Male 36 Blunt Right main bronchus Endoscopy P20 Female 23 Blunt None Endoscopy N21 Male 36 Blunt None Autopsy N22 Male 24 Blunt None Endoscopy N23 Female 34 Iatrogenic None Endoscopy N24 Male 51 Blunt None Endoscopy N

APACHE II, Acute Physiology and Chronic Health Evaluation II score; MPR, multiplanar reconstruction; P, positive; N,negative

Fig. 1 Coronal 2D-multiplanar reconstructions with right laterotracheal air (white arrow), left pneumothorax (hatched black and white arrow) and pneumomediastinum (black arrow)

2176

Sensitivity, specicity, positive and negative predictive values, positive and negative likelihood ratios and 95% condence intervals (CI) were calculated. The overall agreement between OT results and MPR, and between the two radiologists, was calculated using Cohens kappa. Results are expressed as counts (%) and median values (interquartile range).

Results Patient characteristics

During the investigation period, 1346 consecutive patients were assessed for entry into the study. Of these, 24 patients had suspected TBR: median age 36.5 years (2465.5), median ISS 38.5 (2363.8) and median APACHE II score23.5 (10.333.5). The mechanisms of injury are summarized in Table 2.

Endotracheal intubation was performed before CT scans in 62.5% of patients (the median duration of mechanical ventilation was 3 days [111.8]). The median length of ICU stay was 3 days (1.313.5) and the mortality rate was 54.2%.

All trauma patients presented with associated injuries.

Clinical and radiological ndings

The symptoms and radiological ndings are listed in Table 1.

Diagnostic methods and location of injury

TBR was diagnosed in 13 patients (54.2%) by at least one OT. The diagnostic methods used and injury location are summarized for each patient in Table 2. MPR was positive in 15 patients. There were no false-negative MPR results. Two false-positive MPR results were reported. The diagnostic sensitivity and specicity of MPR was 100% (95%CI, 85100) and 82% (95%CI, 6482), respectively. The positive and negative predictive values were 87% (95%CI, 7487) and 100% (95%CI, 78100), respectively. For TBR, the positive and negative likelihood ratios were 5.5 (95%CI, 2.355.5) and 0 (95%CI, 00.24), respectively. The Kappa coefcients were 0.83 (95%CI,0.61.06) for agreement between OT and MPR, and0.91 (95%CI, 0.590.91) for agreement between the two radiologists.

DiscussionIn this study, patients were enrolled if airway injury was suspected on clinical or radiological grounds [11]. MPR

was found to have excellent sensitivity and high specicity for the diagnosis of TBR. The excellent negative predictive value of MPR readily identies patients without TBR and the negative likelihood ratio contributes to the exclusion of a diagnosis of TBR. In other words, when MPR is negative it is not necessary to perform endoscopy to exclude TBR.

This is the rst study to assess the predictive value of MPR for the diagnosis of TBR. This study was designed as a real-life study and, in contrast to case reports of MPR, provides a comparison of MPR with OT. Although MPR criteria for the diagnosis of TBR were more restrictive in our study, all cases of TBR were diagnosed by MPR [10, 1519].

There are some limitations to this study however. Although the sample size required was calculated before the study it is nevertheless small, reecting the rare nature of this condition. Other prospective studies are therefore necessary to conrm our results. Our results may only apply to patients who fullled at least one inclusion criterion from each data list and may not be valid in patients with minor TBR. The exact prevalence of TBR is unknown because bronchoscopy is not performed systematically in blunt trauma patients. Although MPR had an excellent negative predictive value in our study, the overall results of this test may vary with the size of the study cohort. The condence interval around the negative predictive value was 78100% reecting the small cohort size rather than the performance of the test. However, further studies are necessary to conrm this. The patient cohort included in this preliminary study was heterogeneous and it would be interesting to perform other studies with subgroups consisting of blunt and penetrating trauma patients only. At present CT scans are performed on all trauma patients. MPR could also be carried out on all of these patients after evaluation of the risks of transport [20]. When TBR does not require surgical management, bronchoscopy should be unnecessary. We propose a diagnosis algorithm for trauma patients (Fig. 3).

Fig. 3 Diagnosis algorithm for tracheobronchial rupture in trauma patients

2177

The specicity of MPR was 82% because two false-positive results were obtained. The rst patient (patient no. 2) had moderate dyspnea. Multidetector CT ndings included pneumothorax, pneumomediastinum and paratracheal air. A posterior tracheal injury near the carina was suspected on MPR but was not conrmed by bronchoscopy. Epelman et al. diagnosed a complete bronchial transection by MPR and surgery in a patient with normal bronchoscopy [10]. Because of the design of this study, we must consider our result to be a false-positive MPR. The second patient (patient no.13) presented with hemoptysis, bilateral pulmonary contusions, pneumomediastinum, hemopneumothorax and rst rib fractures. Thoracotomy and autopsy were performed after cardiac arrest and no TBR was diagnosed. A second MPR was reviewed and TBR was diagnosed in the posterior tracheal wall. In critical thoracic trauma, TBR may be misdiagnosed by surgery and autopsy. The main pitfall is that the reformation introduces some distortion to the true anatomical relationship between thoracic structures and may even generate artefactual lesions. Such reformations therefore need to be inter-

preted together with the axial images. Nevertheless, one of the most signicant advances is the possibility of obtaining reformation quickly in any arbitrary plane with the same resolution plane as the native axial scan.

Some cases of virtual bronchoscopy have been reported for the diagnosis of TBR. Like MPR, virtual bronchoscopy cannot visualize the structure of the mucous membrane and, in trauma patients, artifacts (secretions, blood) can lead to misdiagnosis. So we did not assess this technique.

ConclusionsIn summary, our results demonstrate that MPR is a very sensitive method for the diagnosis of TBR. MPR seems to be a discriminating technique because of its excellent negative likelihood ratio. In clinical practice, negative MPR can rule out a diagnosis of TBR without the need for bronchoscopy. Because of false-positive MPR results, however, it is necessary to perform bronchoscopy to diagnose TBR denitively.

References

1. Huh J, Milliken JC, Chen JC (1997) Management of tracheobronchial injuries following blunt and penetrating trauma. Am Surg 63:896899

2. Cicala RS, Kudsk KA, Butts A, Nguyen H, Fabian TC (1991) Initial evaluation and management of upper airway injuries in trauma patients. J Clin Anesth 3:9198

3. Rossbach MM, Johnson SB, Gomez MA, Sako EY, Miller OL, Calhoon JH (1998) Management of major tracheobronchial injuries: a 28-year experience. Ann Thorac Surg 65:182186

4. Cassada DC, Munyikwa MP, Moniz MP, Dieter RA, Schuchmann GF, Enderson BL (2000) Acute injuries of the trachea and major bronchi: importance of early diagnosis. Ann Thorac Surg 69:15631567

5. Devitt JH, Boulanger BE (1996) Lower airway injuries and anaesthesia. Can J Anaesth 43:148159

6. Velly JF, Martigne C, Moreau JM, Dubrez J, Kerdi S, Couraud L (1991) Post traumatic tracheobronchial lesions. A follow-up study of 47 cases. Eur J Cardiothorac Surg 5:352355

7. Hara KS, Prakash UB (1989) Fiberoptic bronchoscopy in the evaluation of acute chest and upper airway trauma. Chest 96:627630

8. Barmada H, Gibbons JR (1994) Tracheobronchial injury in blunt and penetrating chest trauma. Chest 106:7478

9. Triaridis S, Konstantinidis I, Noussios G, Karagiannidis K, Permekerlis A (2002) A diagnostic challenge: investigating suspected tracheobronchial and pharyngoesophageal tears. Are there gold standards? Acta Otorhinolaryngol Belg 56:331334

10. Epelman M, Ofer A, Klein Y, Best LH, Guralnik L, Bentur L, Traubici J (2002) CT diagnosis of traumatic bronchial rupture in children. Pediatr Radiol 32:888891

11. Nishiumi N, Maitani F, Yamada S, Kaga K, Iwasaki M, Inokuchi S, Inoue H (2002) Chest radiography assessment of tracheobronchial disruption associated with blunt chest trauma. J Trauma 53:372377

12. Jolliet P, Chevrolet JC (1992) Bronchoscopy in the intensive care unit. Intensive Care Med 18:160169

13. Poli-Merol ML, Belouadah M, Parvy F, Chauvet P, Egreteau L, Daoud S (2003) Tracheobronchial injury by blunt trauma in children: is emergency tracheobronchoscopy always necessary? Eur J Pediatr Surg 13:398402

14. Boiselle PM, Ernst A (2002) Recent advances in central airway imaging. Chest 121:16511660

15. Chen JD, Shanmuganathan K, Mirvis SE, Killeen KL, Dutton RP (2001) Using CT to diagnose tracheal rupture. AJR Am J Roentgenol 176:12731280

16. Scaglione M, Romano S, Pinto A, Sparano A, Scialpi M, Rotondo A (2006) Acute tracheobronchial injuries: Impact of imaging on diagnosis and management implications. Eur J Radiol 59:336343

17. Nakamori Y, Hayakata T, Fujimi S, Satou K, Tanaka C, Ogura H, Nishino M, Tanaka H, Shimazu T, Sugimoto H (2002) Tracheal rupture diagnosed with virtual bronchoscopy and managed nonoperatively: a case report. J Trauma 53:369371

2178

18. Norwood SH, McAuley CE, Vallina VL, Berne JD, Moore WL (2001) Complete cervical tracheal transection from blunt trauma. J Trauma 51:568571

19. Kuhne CA, Kaiser GM, Flohe S, Beiderlinden M, Kuehl H, Stavrou GA, Waydhas C, Lendemanns S, Paffrath T, Nast-Kolb D (2005) Nonoperative management of tracheobronchial injuries in severely injured patients. Surg Today 35:518523

20. Beckmann U, Gillies DM, Berenholtz SM, Wu AW, Pronovost P (2004) Incidents relating to the intra-hospital transfer of critically ill patients. An analysis of the reports submitted to the Australian Incident Monitoring Study in Intensive Care. Intensive Care Med 30:15791585