The CT-arthrography in the antero-inferior glenoid labral lesion: Pictorial presentation and diagnostic value

Objective: To present the Computed Tomography (CT)-Arthrography appearance of the most common types of anterior labral lesion and to assess the diagnostic value of this technique in the detection and classiﬁ cation of the antero-inferior labral tears in glenohumeral joint instability. Materials and Methods: The pre-operative CT-Arthrography records of 43 patients, who underwent surgery for anterior shoulder instability, were retrospectively evaluated independently by two radiologists. The data were compared with arthroscopic results and the diagnostic accuracy of CT-Arthrography was calculated to detect the labral lesion and the agreement between the CT-Arthrography lesions classiﬁ cation and the arthroscopy classiﬁ cation. Results: The CT-Arthrography sensitivity, speciﬁ city and accuracy were: 92% / 89% (reader 1/reader 2), 86% / 86% and 91% / 88% respectively. The CT-Arthrography classiﬁ cation was correct in 86% of cases. Conclusions: CT-Arthrography appears to be an accurate means for identiﬁ cation and classiﬁ cation of the anterior labral tears and, identifying the labral degeneration, this technique can be very helpful in the selection of patient for arthroscopic stabilization of the shoulder.

The shoulder is the most mobile joint in the human body due to the disproportion between the articular surface of the humeral head and the glenoid; the round surface of the humeral head is about twice the size of the oval and fl at glenoid fossa. The stability of the glenohumeral joint is provided by passive and active mechanisms. The active systems are represented by rotator cuff muscles and the tendon of the long head biceps. The passive mechanisms are composed of the capsulelabral structures and ligaments; lesions of one or more of these structures cause pain and instability sensation and are occasionally associated with recurrent dislocation.
Instabilities are more often anterior, occurring in 95% of all patients, in a few cases they are posterior (3%). The remaining 2% of patients have inferior, superior or multidirectional instability. [1] Fractures of the osseous glenoid and humeral head and tears of the labro-ligamentous complex are frequently associated with glenohumeral instability.
The classical labral tear, named "Bankart lesion", consists of antero-inferior capsule-labral detachments and it is the most common injury associated with instability; usually the labrum is detached from the glenoid rim, the inferior glenohumeral ligament (IGHL) and the capsule are damaged, as is the periosteum of the anterior neck of the scapula.
During recent years, with the development of the arthroscopy procedure, many variants of Bankart lesion have been described in the literature.
Anterior labro-ligamentous periosteal sleeve avulsion (ALPSA) lesion is an avulsion of the antero-inferior glenoid labrum with an intact scapular periosteum; it is characterized by the torn antero-inferior labrum being displaced infero-medially by the inferior glenohumeral ligament. [2] ALPSA differs from Bankart lesion in that the ALPSA lesion has an intact periosteum and the labrum is not detached but it remains attached to the scapula via an intact, but stripped, scapular periosteum.
The Perthes lesion represents a non-displaced avulsed anteroinferior labrum with medial stripping without disruption of the scapular periosteum. [3] In a small percentage of cases, a cartilage lesion of the glenoid surface is associated with labral tear; this lesion is named Glenoid Labrum Articular Disruption (GLAD). [4] A correct identifi cation of anterior labral tears is necessary for correct treatment and the pre-operative labral lesions categorization can be useful to the orthopedic surgeons. [3,5,6] Moreover many authors have suggested that, with precise patient selection, the shoulder arthroscopic stabilization can provide high levels of patient satisfaction with low rates of recurrence. The quality and quantity of capsulo-labral tissue and the quantity of glenoid and humeral bone loss are some of the parameters that can change the treatment decision. [7][8][9][10][11][12] To correctly diagnose and treat shoulder instability properly, many imaging techniques have been used. [1,13] The purpose of this study is to present the CT-Arthrography pattern of the different types of anterior labral lesion and to evaluate the accuracy of the technique in the identifi cation and categorization of the anterior glenoid labral lesions using arthroscopy as the reference standard. Many articles have been published in the literature about the accuracy of CT-Arthrography in the detection of labral lesions, but, to our knowledge, none of them have shown images of non-Bankart labral lesion and none of them have evaluated the sensitivity in classifying the anterior labral tears. [14][15][16][17][18]

Patients
We retrospectively examined the CT-Arthrography records of 43 consecutive patients referred for suspected instability of the shoulder. All patients fulfi lled the following criteria: 1) CT-Arthrography of the shoulder was performed at our institution (Ospedale Evangelico Villa Betania di Napoli), according to a standardized protocol, between March 2001 and June 2005; 2) all patients underwent surgery or arthroscopy performed by a single specialized shoulder surgeon; 3) the indication for surgery or arthroscopy was anterior joint instability; 4) surgery was performed within less than three months of CT-Arthrography; 5) None of the patients had undergone previous shoulder surgery.
Twenty-nine patients were men and 14 were women. Their ages were between 18 and 47 years.

CT-Arthrography imaging protocol
Arthrography was performed in a radiologic room under fl uoroscopic control. Ten to fi fteen millilitres of iodinated contrast agent were injected, with a 22G needle, into the shoulder joint, using the anterior approach. To obtain a homogeneous intra-articular diffusion, passive movements were performed, after removal of the needle. CT examination was performed with the arm in a neutral position in all patients; in four cases, additional scans with the arm in internal and external rotation was performed. Informed consent was obtained from the patients before CT-Arthrography. This method was approved by the ethics committee of the hospital.
CT study was performed using GE Pro-speed S Fast scanner (collimation 2 mm; pitch 1); reconstruction at 1 mm was obtained in axial, sagittal and coronal planes. Due to utilization of intra-articular contrast, images were recorded on film utilizing wide window width (2500-3000) and a window level of 300-800.

CT-Arthrography imaging analysis
CT images were re-analyzed independently by two radiologists (reader 1 and reader 2) expert in musculoskeletal radiology, blinded to patient history, arthroscopic results and the relative proportion of the various surgically fi ndings.
The criteria used to diagnose a labral tear were the following: the identifi cation of contrast material under the anterior labrum and out of the capsule for Bankart lesion; the identifi cation of contrast media under the anterior labrum with an intact capsule for Perthes lesion; the medially dislocation of anterior labrum with an intact capsule for ALPSA lesion; the identifi cation of contrast media within articular cartilage of anterior glenoid surface for GLAD; the absence of labrum or its inhomogeneous morphology and/or density was classifi ed by the radiologists as absent/fragmented labrum.

Arthroscopy
At arthroscopy the glenoid labrum was normal in 7 patients and abnormal in 36. Of the 36 pathologic labrum, the following types of lesion were found; 16 Bankart lesions; 10 ALPSA; 6 absent/fragmented labrum; 3 Perthes lesion; 1 GLAD of anterior labrum.

Data analysis
We compared the presence or absence of labral lesion at CT-Arthrography with the surgical fi ndings in each patient and determined the number of true-positive, true-negative, falsepositive and false-negative imaging results. The sensitivity, specifi city, positive predictive value, negative predictive value and accuracy of the imaging technique in identifi cation labral pathology were calculated.
The CT-Arthrography and arthroscopy classifi cations were compared evaluating the percentage of agreement and the sensitivities in detection and correctly categorization of the Bankart lesion, ALPSA lesion and absent/fragmented labrum were calculated.

Results are summarized in
Of 16 Bankart lesions identifi ed at arthroscopy; 14 / 14 were correctly diagnosed by CT-Arthrography [ Figure 1]; 1 / 1 lesion was identifi ed but it was categorized as absent/fragmented labrum; in one case the Bankart lesion was not identifi ed by both radiologists [ Figure 2]; one false-positive for Bankart lesion was found by reader 2.
The sensitivity in identifi cation of the Bankart lesion was of 93.7% and the sensitivity in categorization was of 87.5% for both radiologist.
Ten ALPSA lesions were identifi ed at arthroscopy and all of them were correctly identified and classified by CT-Arthrography with a sensitivity value of 100% [ Figures 3,4].
Of three Perthes lesions diagnosed by arthroscopy, only one was identifi ed at CT-Arthrography; the lesion was clearly diagnosed by the images of the scans with the arm in external rotation [ Figure 5]; in both cases in which the lesion was not identifi ed, additional scans with the arm in internal and external rotation were not performed; one false-positive for Perthes lesion was found by reader 1.
In six cases the surgeon described the labrum as degenerated; in fi ve of these cases the radiologists diagnosed the tears and    The CT-arthrography in the antero-inferior glenoid labral lesion correctly classifi ed them as fragmented/absent labrum [ Figure  6]; in the remaining case radiologist 1 identifi ed the lesion but categorised it as Bankart lesion and reader 2 classifi ed the labrum as normal diagnosing only the capsular tear. The sensitivity in identifi cation and classifi cation of the absent/fragmented labrum was 83.3% / 100% and 83.3% / 83.3% respectively.
The one GLAD lesion [ Figure 7] was correctly identifi ed and classifi ed.
The arthroscopic and CT-Arthrography classifi cations were in agreement in 86% of cases (37/43 cases).
Inter-observer agreement for classifi cation of labral lesions was 95.3 %.

DISCUSSION
The difference in size between the glenoid fossa and humeral head allows a wide range of motion in the shoulder joint. The peri-articular components help provide stability of this unstable structure.
In 1938, Bankart considered the labrum an essential component in the prevention of shoulder dislocation; [19] Turkel et al, in 1981, investigating the mechanism preventing joint dislocation, found that the inferior glenohumeral ligament labral complex is the  primary anterior stabilizer of the shoulder when the arm is at 90° of abduction and external rotation. [20] Successively, several authors have demonstrated that lesions of the glenoid labrum and its capsular attachment, can be associated with and cause shoulder instability and symptomatology. [21] To correctly diagnose and properly treat shoulder instability, many imaging techniques have been used such as routine radiography, ultrasound, Computed Tomography (CT), CT-Arthrography, conventional Magnetic Resonance (MR) and MR-Arthrography. [13,[22][23][24] Many authors indicate MR-Arthrography as superior to other imaging methods in evaluation of the glenohumeral joint, but the use of this technique is limited by several contraindications, such as claustrophobia, metallic objects inside the body or pacemakers. [1,25] Moreover, in our clinical practices, CT-Arthrography is required more than MR-Arthrography, probably because of the higher diagnostic confi dence of the orthopedic surgeons with CT images, the higher distribution of CT scanners in our region and the possibility of this method to evaluate the exact bone glenoid loss. [26] Several studies have described the diagnostic accuracy of the imaging methods in identifi cation of labral lesion but, to our knowledge, only Waldt et al, have been calculated the accuracy in the categorization of the labral lesions at MR-Arthrography. [27] Furthermore, no studies on CT-Arthrography anterior labral tear classifi cation have been published. [14][15][16][17][18]25] In this study the CT-Arthrography has demonstrated high values of sensitivity (91.6% and 88.8% respectively for each reader) in detection of anterior labral tears.
Previous CT-Arthrography studies show a large discordance of results, with a sensitivity varying from 73% to 92%. [14][15][16][17][18] Therefore, it is diffi cult to make a proper comparison with our results.
Instead, our results are comparable with those of previous investigations on the diagnostic accuracy of MR arthrography. [1,27] To correctly diagnose Bankart lesion, the CT-Arthrography technique has shown a high sensitivity, identifying and categorizing the lesions in 93% and 87% of cases respectively; the contrast agent creates a high density line between the glenoid rim and its detached labrum, showing the lesion clearly [ Figure 1].
In the older Bankart lesion, false negatives could be found; the presence of fi brous scar can prevent the contrast media passage and it may mimic an intact labrum.
Correct identifi cation and classifi cation of ALPSA lesion was established, in 100% of cases, with a sensitivity value higher than MR-Arthrography. [27] The intact capsule, allows a good contrast media concentration inside the joint, showing the medially dislocated labrum. Moreover, periosteal reaction of the anterior aspect of the glenoid, caused by periosteum stripping, could be considered a specifi c indirect sign of ALPSA lesion [ Figure 3].
CT-Arthrography has shown a very low sensitivity in detection and characterization of the Perthes lesion. The intact capsule and the undisplaced labrum create diffi culties in recognizing this type of lesion. Many authors have already suggested that scans with the arm in external rotation or in external abduction rotation (ABER) position are necessary to show the displacement of the labrum. [27,28] In our retrospective study, only a few patients were studied with additional scans, therefore no statistical results can be made.
In our study, in six cases the surgeon described the anterior labrum-ligamentous complex as degenerated, with a poor quality and quantity of tissue; therefore the arthroscopic stabilization procedure could not be performed. CT-Arthrography has shown a sensitivity of 83% / 100% in identifi cation of this abnormal condition and 83%, for both readers, in categorizing it as fragmented/absent labrum.
To our knowledge, the distinction of Bankart lesion from ALPSA, Perthes or GLAD may be useful but not necessary for a treatment decision, but, many several authors have suggested that a degeneration or absence of the antero-inferior capsulo-labral complex is an important criteria for treatment decision. [7][8][9][10][11][12] In this study the authors have demonstrated the high sensitivity value in diagnosing the labrum degenerative change; other authors have demonstrated the possibility of the CT to quantify the exact glenoid bone loss; [26] the authors suggest that the use of CT-Arthrography as a pre-operative imaging technique could be very helpful in treatment planning.
Here we mention some limitations in our study. First, although arthroscopy was the best reference standard achievable in this study setting, it is an operator-dependent technique and some labral lesions are diffi cult to see at this type of procedure. [28] The second limitation was the small number of cases with intact labrum and the absence of control subjects; the specifi city value in identifi cation of labral tears might not be correct.
Another limitation is the small number of cases of Perthes and GLAD lesions; again the sensitivity in detection of Perthes lesion must be studied with scans in ABER position and in our retrospective evaluations it was not possible. Therefore no signifi cant statistical conclusions about diagnostic sensitivity and inter-observer agreement for identifi cation of these types of lesions, can be made.
The CT-arthrography in the antero-inferior glenoid labral lesion The fourth point is that the patients with acute and chronic instability were not studied as different groups.
In conclusion, CT-Arthrography is a valid technique to identify and classify anterior labral lesion occurring in anterior shoulder instability. As it can detect the degenerative phenomena of the labrum and a quantify the glenoid bone loss, CT-Arthrography can be very helpful in the selection of patients for arthroscopic shoulder stabilization procedure.