Absorption of the Bone Fragment in Shoulders With Bony Bankart Lesions Caused by Recurrent Anterior Dislocations or Subluxations

Materials and Methods

Study participants were 145 patients (163 shoulders) with recurrent anterior dislocations or subluxations, in whom glenoid defects and bony Bankart lesions were prospectively evaluated by computed tomography (CT) at our hospital between 2007 and 2011. All had suffered an obvious traumatic episode. Excluded from this study were patients with only 1 traumatic episode including primary dislocation (60 shoulders), recurrent instability after a previous anterior stabilization (4 shoulders), multidirectional instability (2 shoulders), epilepsy (3 shoulders), rotator cuff tear without a Bankart lesion (5 shoulders), and insufficient data including no CT evaluation (5 shoulders). Institutional review board approval was obtained before study recruitment began, and all patients gave informed consent before participating in the study.

There were 17 women (19 shoulders) and 128 men (144 shoulders) included in the study. The mean patient age at the time of the primary traumatic episode was 18.4 years (range, 12-42 years), and the mean age at CT evaluation was 21.5 years (range, 14-55 years). The CT scan was usually taken during the patient’s first visit. The total number of dislocations and subluxations was 2 to 5 in 85 shoulders, 6 to 10 in 34 shoulders, and more than 11 in 44 shoulders. There were 130 athletes (148 shoulders), including 79 from collision sports (95 shoulders), 19 from contact sports (20 shoulders), and 23 from overhead sports (24 shoulders). Although none were professional athletes, 104 of them (122 shoulders) played at a competitive level, and most were highly active college or high school athletes. While their chief complaint was a sense of instability (151 shoulders), residual consistent pain after initial trauma was recognized in 12 shoulders.

Surgery was performed on 127 shoulders, all of which underwent arthroscopic Bankart repair. The mean period between CT evaluation and surgery was 1.5 months (range, 0-15 months). Twenty shoulders were operated on more than 3 months after CT evaluation. There was no open surgery and no bone grafting procedure during the same period of this series. The associated lesions were capsular tear in 15 shoulders, humeral avulsion of the glenohumeral ligament in 2, and severe partial rotator cuff tear on the articular side in 5. All of these lesions underwent arthroscopic repair simultaneously with the Bankart procedure.

To quantify the glenoid defects and bone fragment absorption, the inferior portion of the glenoid contour approximated a true circle on the en face 3-dimensional (3D) reconstructed CT image with elimination of the humeral head with a Toshiba whole-body X-ray CT scanner (spiral scan, 0.5-mm slice thickness, 0.3-mm reconstruction, 3D edit mode) (Aquilion 64, Toshiba Medical Systems Corp, Tochigi, Japan). Using a digital imaging and communications in medicine (DICOM) viewer of a picture archiving and communication system (PACS) (Mitsubishi Heavy Industries Mechatronics Systems, Kobe, Japan), the extent of the glenoid defect was calculated as a percentage with the equation B/A × 100%, where A is the diameter of the fitted circle and B is the width of the defect. Using the CT scan on which the articular surface of the fragment was clearest as the bone fragment was often rotated out of the plane of the glenoid, the width of the bone fragment was measured as C. Then, the extent of bone fragment absorption was calculated as a percentage with the equation (1 – C/B) × 100% (Figure 1).

OPEN IN VIEWER

Figure 1.

Quantification methods for glenoid defects and bone fragment absorption. (A) On the en face 3-dimensional reconstructed computed tomography (CT) scan with the humeral head eliminated, the inferior portion of the glenoid contour can be approximated to a true circle. The extent of the glenoid defect was calculated as a percentage with the equation B/A × 100%, where A is the diameter of the fitted circle and B is the width of the glenoid defect. (B) Using the CT scan on which the articular surface of the fragment was most clearly seen, the width of the fragment was measured as C. Then, the extent of bone absorption was calculated as a percentage with the equation (1 – C/B) × 100%.

In some patients, a bone fragment was not detected regardless of loss of the normal contour of the glenoid rim on CT and was either classified as a case of erosion or of complete bone fragment absorption depending on the appearance of the glenoid rim. For example, we judged the case shown in Figure 2A as erosion and the case shown in Figure 2B as complete bone fragment absorption. While in Figure 2B the anteroinferior portion of the glenoid appears straight and sharp on the en face view, suggesting a previous traumatic avulsion of the glenoid rim, in Figure 2A, the anteroinferior portion appears round and slightly compressed on the en face view, suggesting a compression fracture of the glenoid rim. Thus, shoulders with erosion were regarded as having no glenoid defect and no bone fragment, and shoulders with complete bone fragment absorption were regarded as having a glenoid defect and no bone fragment.

OPEN IN VIEWER

Figure 2.

Distinction between erosion and complete bone fragment absorption. (A) Erosion: Because the anteroinferior portion of the glenoid appeared round and slightly compressed on the en face view, suggesting a compression fracture of the glenoid rim, this case was judged as erosion (period: 1 year after the primary episode; glenoid defect: 11.6%). (B) Complete bone fragment absorption: Because the anteroinferior portion of the glenoid appeared straight and sharp on the en face view, suggesting a previous traumatic avulsion of the glenoid rim, this case was judged as complete bone fragment absorption accompanying a glenoid defect (period: 4 years after the primary episode; glenoid defect: 20.8%).

First, the glenoid rim contours in shoulders with recurrent instability were compared with those in 60 shoulders with primary instability evaluated by CT during the same period as the present series. The patient profiles were similar to those in patients with recurrent instability. Then, in cases of recurrent instability, the relation between the extent of the glenoid defect or bone fragment absorption with the period between the primary traumatic episode and CT evaluation was investigated. In the surgical cases, the presence of a bone fragment was investigated at arthroscopic examination, and the fragment was compared with that seen on CT scans. Especially for shoulders with erosion and shoulders defined as complete absorption of the bone fragment, characteristic findings including CT evaluation, the time after trauma, and arthroscopic findings were investigated in detail. Finally, bone union of a bone fragment and short-term clinical outcome after arthroscopic Bankart repair were evaluated in shoulders followed up for a minimum of 1 year.

Statistical analysis was performed with the χ² test and 1-factor analysis of variance (ANOVA), and statistical significance was accepted at P < .05.

Results

The glenoid rim characteristics in cases of recurrent instability are shown in Table 1. There were no glenoid defects in 55 shoulders (33.7%), erosions in 16 shoulders (9.8%), and glenoid defects in 92 shoulders (56.4%). Large glenoid defects (>20%) were seen in 33 shoulders (20.2%), and the mean extent of the glenoid defect was 10.1% ± 10.6% (range, 0%-41%). Absorption of the bone fragment was seen in all 92 shoulders with glenoid defects. The mean extent of bone fragment absorption was 63.8% ± 25.2% (range, 13%-100%), with the absorption being <50% in 32 shoulders, >50% in 45 shoulders, and 100% in 15 shoulders.

OPEN IN VIEWER

Table 1

Glenoid Rim Characteristics in Recurrent Instability and Primary Instability

	Recurrent	Primary	P Value
Normal glenoid, n (%)	55 (33.7)	48 (80)	<.0001
Glenoid erosion, n (%)	16 (9.8)	1 (1.7)	.042
Glenoid defect, n (%)	92 (56.4)	11 (18.3)	<.0001
0%-10%	15 (9.2)	4 (6.7)
10%-20%	44 (27.0)	4 (6.7)
20%-30%	26 (16.0)	1 (1.7)
30%-40%	6 (3.7)	1 (1.7)
40%-50%	1 (0.6)	1 (1.7)
Mean extent of glenoid defect, %	10.1	3.5	<.0001

In cases of primary instability, there were no glenoid defects in 48 shoulders (80%), erosion in 1 shoulder (1.7%), and glenoid defects in 11 shoulders (18.3%). As the mean extent of the glenoid defect was 3.5% ± 9.1% (mean, 0%-52%), the frequency and size of the glenoid defects were significantly different between recurrent instability and primary instability (Table 1). Complete absorption of the bone fragment in cases of primary instability was seen in only 1 shoulder with glenoid defects. This patient complained of persistent pain after the first dislocation and underwent CT evaluation 3 months later (glenoid defect: 11.6%; bone fragment absorption: 100%).

Thus, the effects of the time after trauma on glenoid defects and bone fragment absorption were investigated in cases of recurrent instability. The average extent of the glenoid defect was 7.9% at <1 year (49 shoulders), 10.7% between 1 and 2 years (36 shoulders), and 11.3% at >2 years (78 shoulders). There were no significant differences among the 3 groups (P = .219) (see Appendix 1, available at http://ajsm.sagepub.com/supplemental). Clinically significant large glenoid defects (>20%) were seen in 7 shoulders (14.3%) at <1 year, 7 shoulders (19.4%) between 1 and 2 years, and 19 shoulders (24.4%) at >2 years, with no statistical differences among the 3 groups (P = .385).

The average bone fragment absorption was 51.9% at <1 year (26 shoulders), 65.3% between 1 and 2 years (22 shoulders), and 70.0% at >2 years (44 shoulders). There were significant differences among these 3 groups (P = .013) (see Appendix 2, available online). In addition, severe bone fragment absorption (>50%) was seen in 11 shoulders (42.3%) at <1 year, 17 shoulders (77.3%) between 1 and 2 years, and 32 shoulders (72.7%) at >2 years, and there were also significant differences among the 3 groups (P = .014).

In shoulders with large glenoid defects (>20%), the average bone fragment absorption was 57.5% at <1 year (7 shoulders), 72.4% between 1 and 2 years (7 shoulders), and 73.0% at >2 years (19 shoulders), showing no statistically significant differences among the 3 groups (P = .224) (see Appendix 3, available online). In this same subgroup of large glenoid defects, severe bone fragment absorption (>50%) was found in 3 shoulders (42.9%) at <1 year, 6 shoulders (85.7%) between 1 and 2 years, and 15 shoulders (78.9%) at >2 years, showing no significant differences among the 3 groups (P = .128). A representative shoulder with a large glenoid defect and severe absorption of a bone fragment is shown in Figure 3.

OPEN IN VIEWER

Figure 3.

A shoulder showing a large glenoid defect and severe absorption of a bone fragment. A 20-year-old male American football player at 2 years after the primary traumatic episode (glenoid defect: 38.4%; bone fragment absorption: 50%).

It was more than 1 year since the primary traumatic episode in most of the 16 shoulders with glenoid erosion. The mean extent of the glenoid defect was 11.4%, and the defect was smaller than 20% in all cases. On the other hand, in the 15 shoulders defined as complete absorption of a bone fragment, more than 2 years had passed since the primary traumatic episode in most cases, the mean extent of the glenoid defect was 19.3%, and it was larger than 20% in 5 shoulders (see Appendix 4, available online).

In cases of recurrent instability, among 127 shoulders treated surgically, a bone fragment was found in 69 shoulders at surgery. A fragment was not found in 4 shoulders despite the presence of one on CT scans (Table 2). Among 20 shoulders treated surgically at more than 3 months after CT evaluation, a bone fragment was not found in 1 shoulder at surgery (4 months after CT), despite the presence of one on preoperative CT. On the other hand, a fragment was found in 12 shoulders that were thought to have no fragment on CT (Table 2). While the incidence of such fragments was 4 (10.8%) of 37 shoulders with no glenoid defect, the incidence was 3 (21.4%) of 14 shoulders with erosion and 5 of 15 shoulders (33.3%) with complete bone absorption accompanying a glenoid defect. Although there was no statistically significant difference (no glenoid defect vs erosion: P = .325; no glenoid defect vs complete absorption: P = .052), the incidence showed high frequency in the latter 2 groups (see Appendix 4).

OPEN IN VIEWER

Table 2

Presence of a Bone Fragment: Comparison Between Preoperative CT and Arthroscopic Findings ^a

	Fragment at Surgery, n
	Positive	Negative
Fragment on CT, n
Positive	57	4
Negative	12	54

a

CT, computed tomography.

Ninety-nine shoulders were followed up for a minimum of 1 year after arthroscopic Bankart repair. Among 57 shoulders with glenoid defects, 13 shoulders had no bone fragment, and 14 shoulders could not be evaluated by postoperative CT. Thus, bone union could be evaluated by postoperative CT in 30 shoulders. There was complete bone union in 24 shoulders (80%) and nonunion in 6 shoulders (20%). In most cases, bone union was confirmed between 4 months and 6 months after surgery. There was no shoulder showing bone absorption after the bone union.

Postoperative recurrence of instability occurred in 10 shoulders (10.1%), and apprehensive sensation remained in 5 shoulders (5.1%) (Table 3). While the rate of recurrences was 0% (0/42) in shoulders without a glenoid defect (including normal glenoid and erosion), it was 17.5% (10/57) in shoulders with a glenoid defect. There were statistically significant differences (P = .004). Moreover, among 43 shoulders with a glenoid defect evaluated by CT, the rate was 38.5% (5/13) in shoulders without a bone fragment, 8.3% (2/24) in shoulders with bone union, and 50% (3/6) in shoulders without bone union. Although there were no statistically significant differences regarding the presence of a bone fragment (P = .120), there were statistically significant differences between union and nonunion of the bone fragment (P = .014). In 2 shoulders with bone union of a bone fragment, a new bone fragment at another location was recognized by CT evaluated after the recurrence of instability.

OPEN IN VIEWER

Table 3

Patients With Postoperative Recurrence of Instability and With Apprehensive Sensation ^a

Patient	Age at Surgery, y	Sports	Dislocations and Subluxations, n	Extent of Glenoid Defect, %	Extent of Bone Fragment Absorption, %	Time Between CT and Surgery	Associated Lesions	Bone Fragment	Recurrence	Recurrence After Primary Bankart Repair, mo	Trauma
Postoperative recurrence
1	16	RFB	2	6.4	44.5	4 mo		Fracture	Subluxation	9	Major
2	19	VB	10	15.9	100	2 y		New fragment	Dislocation	29	Major
3	16	RFB	4	16.4	100	7 mo		No fragment	Subluxation	14	Major
4	16	RFB	4	17.8	61.9	1 y		Fracture	Dislocation	8	Major
5	17	RFB	21	19.3	52.5	1 y		Nonunion	Dislocation	6	Major
6	18	HB	60	20.8	100	4 y	Capsular tear	No fragment	Subluxation	2	Minor
7	18	AFB	12	23	100	1 y		New fragment	Subluxation	14	Major
8	18	HB	10	25.5	100	2 y	Capsular tear	No fragment	Subluxation	12	Minor
9	18	RFB	10	28.5	35.2	4 y	Free body	Nonunion	Subluxation	13	Major
10	29	—	30	31	88.6	11 y		Nonunion	Subluxation	12	Minor
Apprehensive sensation
1	35	—	12	19.4	100	8 y		No fragment
2	27	Judo	20	30.2	88.6	11 y		Nonunion
3	22	Martial arts	50	31.3	84	1 y		Bone union
4	19	Kung fu	45	31.5	100	4 y		No fragment
5	20	AFB	20	38.4	50	2 y		Bone union

a

AFB, American football; CT, computed tomography; fracture, fracture at another portion after bone union; HB, handball; new fragment, fracture with new bone fragment after Bankart repair without bone fragment; RFB, rugby football; VB, volleyball.

In cases of primary instability, among 8 shoulders with a glenoid defect and a bone fragment, 5 shoulders were evaluated by CT postoperatively, and bone union was recognized in all 5 shoulders. There was no postoperative recurrence of instability in those 5 shoulders.

Discussion

This study of patients with recurrent anterior dislocations or subluxations of the shoulder revealed a glenoid defect in more than half of them, and absorption of the bony Bankart lesion was seen to some extent in all of the shoulders with glenoid defects. The size of the glenoid defect showed no correlation with the time after the primary traumatic episode. In contrast, severe bone fragment absorption was noted with increasing frequency as the time from the initial trauma increased. Thus, when evaluating glenoid rim shape, it seems to be important to not only measure the size of the glenoid defect but also the size of the residual bone fragment.

Regarding morphological characteristics of the glenoid rim, Sugaya et al ¹¹ reported on 100 shoulders with recurrent anterior instability in detail. Although it is difficult to directly compare their study with ours because they did not mention bone fragment absorption and our method of measurement was different, the frequency of a glenoid defect was similar, but large bony defects (>20%) were more frequent in our series. Moreover, while they reported that the majority of such large glenoid defects almost always have an osseous fragment of variable size lying somewhere along the anteroinferior portion of the glenoid neck, ¹² in our series, severe absorption of the bone fragment was recognized in shoulders with large glenoid defects when more than 1 year had elapsed since the primary traumatic episode, although there was no statistically significant difference. On the other hand, the glenoid rim contour in cases of primary instability was also shown to be obviously different from that in cases of recurrent instability.

When the bone fragment was smaller than the size of the glenoid defect, we concluded that this was caused by absorption of the bone fragment. However, as Fujii et al ⁵ reported that most bone fragments were tightly connected to the labrum and maintained their blood supply, the occurrence of bone absorption was considered difficult. As one of the reasons for the mismatch in size, the possibility that the bone fragment was judged to be small because the glenoid defect had been enlarged by recurrent erosions from dislocations and subluxations cannot be denied. In fact, a tendency for glenoid defects to be larger over time was seen, although it was not statistically significant. Moreover, as compared with primary instability, in recurrent instability, glenoid defects were shown to be larger in this study (Table 1); the number of recurrences might influence the glenoid defects. However, the possibility that shoulders with large glenoid defects at primary trauma were more likely to develop into recurrent instability was also considered.

As another reason for the mismatch in size, the initial bone loss of the glenoid may have come from the crushing of the cancellous subchondral bone with comminution of the joint surface at the time of primary trauma. As most of our patients were young male athletes, we have seldom seen such crushing of the bone at CT evaluation and at arthroscopic surgery, although the possibility of this mechanism cannot be denied. To confirm our hypothesis of bone fragment absorption, primary trauma cases with bone fragments would need to be observed over time.

In the present study, a bone fragment was not detected regardless of the presence of a glenoid defect on CT in some patients, and we judged cases such as the one shown in Figure 2A as erosion and cases such as the one shown in Figure 2B as complete bone fragment absorption. However, the sole difference was the size of the glenoid defect, and there were many similarities such as the long period after primary trauma, so we could not clearly distinguish these 2 groups. Glenoid erosion was usually used to signify wear or a compression fracture, ¹¹ so such changes should be found frequently in cases of primary instability. However, only 1 case of erosion was seen in 60 shoulders with primary instability in this study (Table 1).

It was also interesting that we found some bone fragments at surgery that were not detected on preoperative CT scans. These were all small fragments consisting entirely of spongy bone and were not so useful for repairing glenoid defects; thus, the bone fragments were believed to be missed by CT because of severe absorption. Although there was no statistical significance, the incidence of such fragments showed a high frequency in shoulders with erosion and in shoulders with complete bone fragment absorption accompanying a glenoid defect. As more than 1 year had passed since the primary trauma in most of the shoulders with erosion and the glenoid defect was less than 20% in all of them, we hypothesized that they represent cases of complete absorption of the bone fragment with small glenoid defects. When small bone fragments were seen on CT, such as that shown in Figure 4, they were considered likely to disappear in the near future, leaving the shoulder to be classified as erosion. If our hypothesis regarding erosion is correct, glenoid defects and bone fragment absorption are more frequent in patients with recurrent anterior instability than were shown in the present study. To confirm our hypothesis, we would need to follow up primary trauma cases with small bone fragments very carefully.

OPEN IN VIEWER

Figure 4.

A very small bone fragment with a small glenoid defect. This will probably result in erosion after further bone fragment absorption (period: 1 year after the primary episode; glenoid defect: 12.1%).

The fact that the mismatch in size between the glenoid defect and the bone fragment increases markedly with time and that the bone fragment completely disappeared in 19% of shoulders in this study—if our hypothesis regarding erosion is correct—raises concern that such shoulders should not be left untreated. In fact, the postoperative recurrence rate of instability was high in shoulders with a glenoid defect and no fragment, and apprehensive sensation remained in some shoulders with a large glenoid defect and a small bone fragment (Table 3). Moreover, a bone fragment recognized on preoperative CT was not found at surgery in 4 shoulders. As a reason for the presence of no bone fragment at surgery despite the presence of one on preoperative CT, while malunion of the fragment on the glenoid might be considered, the possibility that a bone fragment was absorbed after CT evaluation could not be denied; we recommend early Bankart repair before complete absorption of the bone fragment.

At surgery, a bone fragment recognized on preoperative CT was found in all but 4 shoulders, and the bone fragment with the anterior glenoid labrum could be repaired by the arthroscopic bony Bankart repair technique in all shoulders. While recently bone grafting for large glenoid defects has been discussed, including the arthroscopic Latarjet procedure, Bristow procedure, and free iliac crest bone grafting among other methods,^1,7,13 we have observed that it is possible to fill the glenoid defect without grafting if a sufficiently large bone fragment is found, and the short-term clinical outcome was good after bone union (Figure 5) except that 2 shoulders had recurrences because of a fracture at another portion after bone union. Sugaya et al ¹² reported that excellent union of the fragment was established in all of their patients. Recently, Park et al ¹⁰ also reported that follow-up CT arthrographic evaluation showed that small bony Bankart fragments survived without resorption until 1 year postoperatively, even with fibrous union, and that the reattached bone fragment fixated to the anatomic position with the labrum could survive. While we have also never experienced resorption after bone union, in this study, the rate of bone union was 80%, and postoperative recurrence of instability occurred in 3 (50%) of 6 shoulders without bone union. As not only shoulders with a glenoid defect and no bone fragment but also shoulders without bone union were shown to have a significant risk of recurrence, a new device to improve the rate of bone union is needed.

OPEN IN VIEWER

Figure 5.

A 16-year-old female handball player showing complete union of a bone fragment with the glenoid (arrow) at 6 months after arthroscopic Bankart repair.

One of the main limitations of this study was our method for the quantification of glenoid defects and bone fragments. We used a simple and convenient method that assumes that the inferior portion of the glenoid rim can be approximated to a true circle on the en face 3D reconstructed CT image. However, there were some cases in which fitting a circle was difficult. If the circle is set too large, the size of the glenoid defect will be exaggerated, and the bone fragment will be considered too small, so that the bone fragment absorption would be too high. In the future, to evaluate glenoid defects and bone fragment absorption more precisely, it will be essential to employ other methods, such as comparison with the glenoid rim shape on the healthy opposite side. ⁴ Moreover, we calculated the sizes of the glenoid defects and bone fragments using their width, but it would have been better to use their area, although Sugaya et al ¹¹ used the area of glenoid defects in their first report and used the width as a simple method in their second report. ¹² Because of these limitations of our method and because comparison with the contralateral side becomes difficult in cases of bilateral shoulder instability (common in collision sports), it will be necessary to develop new measuring methods.

Conclusion

In patients with recurrent anterior shoulder dislocations or subluxations, bone fragment absorption was seen to some extent in all shoulders with bony Bankart lesions. We found that the time elapsed since the primary trauma correlates significantly with the severity of bone fragment absorption. Before arthroscopic Bankart repair, not only bony defects of the glenoid but also the size of the bone fragment of the bony Bankart lesion should be assessed by CT to determine whether the glenoid rim can be repaired using the residual bone fragment.