Introduction

Because of its subcutaneous situation associated with its lack of soft tissue protection, the acromioclavicular (AC) joint is predisposed to be affected by direct and/or indirect traumatisms; hence, AC disjunctions are accountable for 3.2% of shoulder girdle injuries, and they are even more common among athletes practicing contact sports [1]. To categorize these lesions, Rockwood described six different types of AC injury based on radiographic displacement and clinical examination, the rising grade correlating with the increasing severity [2]. Although non-operative treatment of Rockwood type I and II lesions is generally consensual, as well as surgical management is for Rockwood type IV–VI dislocations, a certain debate is still ongoing regarding type III injuries [3, 4]. Regarding the surgical management of high-grade acute injuries, numerous techniques have been proposed to reduce displacement and allow for coracoclavicular (CC) ligaments healing; these include AC joint fixation, hook plates, CC fixation and CC ligaments reconstruction [6]. If no technique has properly demonstrated its superiority yet due to high-level evidence literature paucity [1], biomechanical and clinical outcomes tend to be superior following anatomical CC ligaments reconstruction, including for type III separations [1, 6,7,8,9,10,11,12,13].

During the last decade, endoscopic techniques and devices have been widely developed, aiming for an anatomical, minimally invasive and non-rigid reconstruction of the CC ligaments [14,15,16,17,18,19]. The TightRope® (TR) system and the Dog Bone® (DB) technology (Arthrex, Naples, FL, USA) are two double-button fixation systems allowing for such a repair. The TR implant consists of a round clavicular button and an oblong coracoid button, joined by a continuous double loop of #5 FiberWire® (#5FW, 4-sutures bundle) [20]. The DB device is made of two bone-shaped buttons, linked by two loops of 2-mm FiberTape® (2FT, 4-sutures bundle) [21]. The latter was specifically designed for acute high-grade AC dislocations in the aim of providing stronger fixation than the TR implant.

The purpose of this retrospective study was to compare clinical and radiological outcomes between these two endoscopically assisted double-button techniques. Based on the specific design of the Dog Bone®, we hypothesized that the DB reconstruction technique would provide superior radioclinical outcomes.

Method

Population criteria

This single-center retrospective study was conducted in patients operated on consecutively from January 2009 to May 2014. All patients who had isolated acute AC joint dislocation Rockwood types III to V, and who benefited primarily from an endoscopically assisted surgical management with a double-button fixation system (i.e., TR or DB), were included. Exclusion criteria were prior condition of the injured shoulder girdle, separations older than 15 days, other degrees of injury, other types of treatment (e.g., AC joint fixation) and follow-up lesser than 1 year. Diagnosis of acute AC joint dislocation was based on interrogation and radioclinical evaluations of both shoulder girdles, conducted by the operating surgeon. Physical examination included testing for anteroposterior and/or superoinferior instability, as well as the injured AC joint reducibility. Initial X-rays included standard anteroposterior (AP), scapular-Y, axillary lateral and Zanca views of both shoulders. The latter were obtained with the patient standing erect with the arms hanging; CC distances were assessed on both sides between the superior surface of the coracoid process and the opposing inferior cortex of the clavicle shaft (Fig. 1). Based on this radioclinical information, the lesion was categorized according to Rockwood’s classification system [2].

Fig. 1
figure 1

Posttraumatic bilateral Zanca views of a 29-year-old male, demonstrating on the right side a coracoclavicular distance (cc’, red arrow) increased by more than 100% compared to the physiologic contralateral shoulder (cc, white arrow); a Rockwood type V injury was thus identified. This radiograph was taken with the patient in the standing position, with both arms hanging on the sides of the body

Full size image

Surgical techniques

All patients were operated on endoscopically in the beach chair position, with intraoperative fluoroscopic monitoring, by the same senior surgeon. Two different double-button techniques were then used, in two consecutive series. The first cohort was treated with the TR system. Under fluoroscopic and endoscopic control, a guide pin was inserted from the superior clavicular surface to the inferior aspect of the coracoid base, 3 cm medial to the AC joint. A 4-mm tunnel was drilled over the pin, which allowed the operator to insert vertically the oblong button down through the osseous tunnels; the latter was then flipped to a horizontal position underneath the coracoid base. With the scope placed into the subacromial bursa, the operator controlled the AC joint reduction while pulling on the TR sutures to apply the superior button against the clavicular surface. Once the desired reduction was obtained, the fixation was secured by knotting the sutures over the clavicular button. Final assessment included direct visualization of the superior button applied onto the clavicular cortex, endoscopic control of AC reduction and coracoid button horizontal placement, and AP fluoroscopic control of the reconstruction. The second series benefited from the DB button. First steps were similar to the TR technique, except for the 3-mm drill: Since only the tapes tails were pulled up through the osseous tunnels, this tunnel width was sufficient in all cases. Once the concavity of the inferior button was seated against the coracoid base, the superior button was positioned onto the tapes exiting the clavicle, with its concavity applied to its superior surface. Reduction was achieved by external maneuvers and maintained until the first tape was tied over the clavicular button. Final assessment was then conducted as previously described, and the second tape was tied afterward (Fig. 2).

Fig. 2
figure 2

Endoscopic view of the inferior Dog Bone button during final assessment, once seated against the coracoid basis (black star); note the laser mark (white arrow) in line with the coracoid arch; lateral portal, right shoulder

Full size image

Postoperative care

Radioclinical evaluation was conducted the day after surgery to rule out any major complication (nerve damage, coracoid and/or clavicle fractures) and assesses AC joint reduction and implant position (Fig. 3). All patients followed the same postoperative protocol, including immobilization in a sling and gentle passive pendulum arm motion for 6 weeks; then, the sling was discontinued and active mobilization was started. Any types of activities involving heavy lifting were prohibited for 3 months and contact sports for 6 months. Systematic surgical follow-up was conducted at 6 weeks and then at 3, 6 and 12 months.

Fig. 3
figure 3

Postoperative bilateral Zanca views of a 24-year-old male who presented a Rockwood type V lesion, right shoulder; he was treated with the Dog Bone® device, allowing for satisfactory acromioclavicular joint reduction and restoration of a coracoclavicular distance similar to the contralateral side (white arrows). This X-ray was taken the day after surgery, with the injured shoulder protected in a sling shoulder

Full size image

Data collection

Investigations were performed according to the 1964 Declaration of Helsinki ethical standards, and all patients gave written informed consent. An independent observer (first author) conducted final assessment, which consisted of a chart review and a radioclinical examination. Functional evaluation was completed with the Constant score (CS), along with patient self-evaluation through the Quick-DASH (QD) questionnaire. Pain was rated on a 1–10 visual analog scale (VAS). Radiographic examination was the same as preoperatively. On bilateral Zanca views, CC distances were measured and the difference between the injured and the physiologic shoulders was calculated, which allowed assessing the posttraumatic vertical displacement. A lack of reduction was defined by a 5-mm or greater displacement the day after surgery; a loss of reduction was defined by the same difference, but identified on last follow-up X-rays. On AP views, the AC joints were assessed in order to document recurrent dislocations and potential subsequent degenerative changes. All radiographic data were digitally assessed using IMPAX version 6 (Agfa-Gevaert, Mortsel, Belgium).

Statistical analysis

The Shapiro–Wilk test was used on all data and excluded their normal distribution; nonparametric tests were subsequently performed. Chi-squared tests were used for qualitative data (e.g., lack and loss of reduction rates), whereas quantitative data (e.g., preoperative, postoperative and follow-up displacements) were treated with Mann–Whitney U tests. Results are presented as the mean ± standard deviation, unless otherwise stated. The level of significance was defined as P < 0.05, for all tests. Computerized statistical analysis was performed using SPSS version 22 (IBM, Armonk, NY, USA).

Results

Series characteristics

Forty-one patients were operated on at our institution; twenty-two received the TR device and nineteen the DB button. One patient from the DB series was excluded from the study: He showed a complete recurrence of the dislocation the day after surgery, due to an excentric coracoid tunnel with subsequent monocortical medial coracoid fracture. Surgical revision included open reduction and AC internal fixation with K-wires. Otherwise, the two cohorts were comparable (Table 1), with a high-velocity direct blow to the shoulder identified in all patients, due to recreational sport activities (n = 26), road traffic accidents (n = 9) or work-related injuries (n = 6). No type IV dislocation was encountered during the study period.

Table 1 Cohorts criteria
Full size table

Clinical outcomes

Functional results were good to excellent in both the TR and the DB series without significant difference (Table 2), CS and QD averaged, respectively, 94.3 ± 4.4 and 2.0 ± 2.6 in the TR series, whereas they averaged, respectively, 95 ± 6.1 and 3.4 ± 3.3 in the DB series after a mean follow-up of 24.1 ± 5 months (P CS = 0.16, P QDS = 0.08). Nonetheless, six patients (3 TR, 3 DB) had not resumed their usual activities at final evaluation, including two construction workers who suffered from a work-related injury and did not resume their previous professional occupation; the four remaining patients were competitor athletes (i.e., two judo players, one tennis player and one soccer player) who had returned to competition, but not to their original level. No patient needed pain medication after 3-month follow-up.

Table 2 Clinical and radiological outcomes
Full size table

Radiological outcomes

No statistical differences were noted between the two cohorts in terms of mean displacement, either preoperatively, the day after surgery or at last follow-up (Table 2). However, two significant differences were outlined. Indeed, a lack of reduction was found in four cases the day after surgery in the DB population, whereas only one case was demonstrated in the TR series (P = 0.0005). Also, final X-rays showed a loss of reduction in more than half of the DB population, whereas less than a quarter of the TR group presented such a displacement (P < 0.0001). In three patients of each series, the loss of reduction was associated with a complete recurrent dislocation on the AP views, without major implant migration or fracture (Fig. 4a, b). In three cases (2 TR, 1 DB), this recurrence was related to a new trauma; if no traumatic event was noted in the three remaining patients, the recurrence was clearly identified by two of them as a painless snap during daily living activities (1TR, 1DB). Mild AC joint degenerative changes were noted in the TR patient; since all patients were asymptomatic, no additional surgery was performed.

Fig. 4
figure 4

Immediate postoperative (a) and 1-year follow-up (b) anteroposterior X-rays of a 50-year-old male with initial left acromioclavicular Rockwood type V separation, who benefited from a Dog Bone® device. If the implant position remained unchanged, complete recurrence of the dislocation occurred (white lines), without any significant traumatic event reported by the patient

Full size image

Complications

Two patients treated with the DB button required surgical revision, in addition to the excluded one with a monocortical coracoid fracture. They complained about material-related discomfort due to a bulky clavicular knot; removal of the device was performed after 6 months, and no recurrent dislocation was demonstrated after removal. No intraoperative complications, such as neurovascular injuries, thoracic lesions, coracoid and/or clavicle fractures, were reported; likewise, no postoperative infection was noted.

Discussion

This retrospective study aimed to compare clinical and radiological results between the TightRope® and the Dog Bone® devices, two endoscopically assisted double-button techniques. Despite its design specifically developed to address the challenges of acute AC joint high-grade dislocations, the DB button did not provide superior clinical results, and radiological outcomes were significantly inferior.

The TR fixation is a well-known procedure, leading to adequate biomechanical performances as well as excellent clinical results in acute high-grade AC dislocations [22,23,24]. Moreover, its compact design and the non-rigid fixation theoretically prevent from systematic surgical revision for implant removal, which is mandatory with Bosworth screws, hook plates or temporary K-wire AC fixations [12, 13, 25]. Nonetheless, different flaws have been described in the TR technique, with primarily the postoperative partial loss of reduction due to cortical osteolysis under the clavicular button, with implant migration rates of up to 92% [26]. Another shortcoming may be the large tunnel through the coracoid base to pass the inferior flip button, accountable for early failures subsequent to monocortical coracoid fractures [14, 26]. Finally, the initial stability provided by a single TR would not be sufficient, as advocated by Scheibel and colleagues who described two cases with early failures and complete redislocations among their five first cases [27].

The DB design, specifically studied for AC dislocations, was meant to overcome these TR weaknesses. Superior and inferior DB buttons are similar, curved and wider, with a surface of approximately 80 versus 33.2 mm2 for the first-generation TR clavicular flat button [20, 21]. Allowing for a better tensile forces distribution, this characteristic theoretically prevents caudal migration of the superior button due to osteolysis [5]. In addition, two loops of 2FT tapes are used, with the same composition and structure as #5FW sutures (braided composite polyethylene and polyester sutures) but with a larger cross-sectional area (CSA), and consequently a greater ultimate load to failure [28]. Finally, since only the tapes are passed through the tunnels, the latter may be drilled smaller (between 2.4 and 3 mm). Despite these promising features, the DB device was found to be significantly inferior in our study regarding the lack of initial reduction, the loss of reduction during follow-up and the need for surgical revision. Since no osteolysis or fractures were noted in our DB population, our primary hypothesis to explain the radiological results is the use of tapes. Indeed, the increased CSA of a suture is accounting for lower pliability, which may generate partial loss of reduction because of knot slippage [29]. Surgical revisions due to material-related pain may also be related to the DB design. Indeed, larger buttons and sutures on the superior aspect of the distal clavicle, an area lacking of soft tissue protection, may be held accountable for this type of shortcoming and subsequently lead to an increased material removal rate.

If a closer look is taken to the different points upon which the DB button development had been based, one could argue that the proposed solutions were not relevant. Caudal migrations of TR clavicular buttons were observed with the 6.5-mm-diameter flat buttons (first-generation implant). The new generation of TR implants includes a 10-mm-diameter curved clavicular button that provides a contact area of approximately 78.5 mm2, thus very similar to the DB buttons. This performance difference between the first and second generations of TR implants was identified in several series and herein, since no significant bone lysis was noted in our TR series [27, 30]. Regarding coracoid-related complete failures, they may be related to the inherent technique learning curve, as reported by Salzmann [14]; with experience, the operator was more capable of centering the coracoid tunnel and avoids subsequent monocortical fractures. In addition, since the only similar case we encountered was with the DB system, our study shows that the latter does not prevent this type of complication. Regarding suture breakages, Mazzocca et al. demonstrated that there was no difference between the vertical ultimate loads to failure of native and #5FW-reconstructed AC joints, suggesting that clinical failure of a TR device will not occur because of tensile loading, especially without any trauma history [31]. When Motta and colleagues reported four cases of suture breakage following the use of a single TR, they hypothesized those failures were caused by the cutting edges of the bone tunnels; joint hyperlaxity causing excessive anteroposterior movements in their patients, shear and friction forces were exerted against the sutures (i.e., windshield-wiper effect) [30]. Since replacing #5FW sutures (TR) by 2FT tapes (DB) increases the vertical ultimate load to failure of the device without improving the stability of the reconstruction, the sutures are still subjected to the same windshield-wiper effect. We noted two cases of non-traumatic complete failures in our DB series that may be explained by that phenomenon, with a sudden painless non-traumatic recurrence of the preoperative deformity and no fracture or button mobilization on the X-rays. Scheibel and colleagues proposed another approach, by switching to the double TR technique after facing their two cases of early failures with a single TR. Therefore, with a double-bundle reconstruction, not only the tensile load of the construction is increased but also the joint stability, restraining these horizontal shear forces [27].

Our study suffers from different flaws. First, the data distribution compelled us to use nonparametric tests, decreasing the statistical power of our results. Second, the study design may be questionable, since two successive case series were compared retrospectively as a result of a single-operator consecutive recruitment using one device after the other. Finally, the main shortcoming is the limitation of the clinical results interpretability, due to the relative small number of patients that may have caused a type II error associated with the short-term follow-up that may have prevented to identify a clinical deterioration over time, especially with such a high rate of loss of reduction in the DB population. In fact, long-term follow-up in a larger population would be mandatory to ascertain the absence of clinical difference between those devices.

Conclusions

This comparative study outlines that both the TR and DB devices provide good to excellent short-term clinical outcomes. Nonetheless, considering the early radiological superiority of the TR implant over the DB and the greater need for surgical revision following the use of the latter, we would prefer the TR button for the operative management of high-grade acute AC joint dislocations.