A novel arthroscopic transosseous suture-button fixation technique for anterior glenoid fractures

The system utilized for this procedure (Smith&Nephew Co., London, UK) was developed originally for chronic instabilities by both Ettore Taverna for arthroscopic glenoid augmentation with tricortical iliac crest [26] and by Pascal Bouileau for arthroscopic coracoid transfer [3]. The respective arthroscopic surgery for glenoid fractures has been previously described by our group [27]. In summary, the procedure is performed in the beach-chair position with an en-face view on the glenoid with the arthroscope placed in the anterolateral portal. Several instruments are used through two anterior portals (in the rotator interval) for fracture reduction. Then, the anterior hook-end of the glenoid guide (Fig. 1) is inserted through the posterior portal and placed anteriorly parallel to the glenoid. It then captures the anterior edge of the anterior fragment under the hook. Two bullets are placed on the scapula neck over the glenoid guide through an additional posteromedial skin incision (Fig. 2). A 2.8-mm cannulated drill is placed through each bullet and advanced under power until exiting from the anterior aspect of the anterior fragment. Each drill will be 5 mm below the cortical edge of the glenoid surface. After removal of the glenoid guide, the inner part of the drills is removed posteriorly leaving the cannulated part in place (Fig. 3). Usually, only one pair of endobuttons was placed and therefore the cranial drill was removed in most cases. A flexible looped guide wire is then passed through the remaining cannulated drill from posterior to anterior and retrieved through the anterior portal. The drill can now also be removed, finally leaving only the looped guide wire in the lower drill tunnel.

With the guide wire, all four strands of the endobutton-suture-fixation device (Fig. 4a) are shuttled into the anterior portal, through the glenoid and out of the posteromedial incision. Thus, the anterior endobutton lies firmly upon the anterior surface of the anterior glenoid fragment (Fig. 4b). The posterior strands are knotted over a posterior endobutton, which is then advanced until it sits firmly against the posterior face of the scapular neck. The construct is locked with a suture tensioner and finally a standard knot pusher. The capsulolabral ring can be reattached additionally on the glenoid rim applying a standard arthroscopic soft-tissue repair technique.

According to the existing literature, surgery of glenoid fractures was indicated if fragment dislocation of at least 4 mm or a decentered humeral head on anteroposterior x‑rays was present. Patients were eligible for the novel transosseous technique if a solitary fracture was present or one large main fragment (Type Scheibel 1b or 1c; [21]).

From March 2017 to May 2021, 23 patients received the novel arthroscopic procedure in the two participating centers: Three patients had surgery at the “Klinik-am-Ring” Clinic and 20 patients at the Cologne-Merheim Medical Center (CMMC). There were initial plans for two other patients at CMMC to undergo the procedure, but in both cases intraoperative assessment had shown that the anterior fragment was not large enough for the endobutton to engage upon its anterior aspect. Thus, the implant was removed and a conventional arthroscopic suture anchor technique according to Sugaya [24] was performed.

Informed consent was given by all 23 patients. For the results of the clinical outcome, a minimum follow-up of 6 months was required. The study was approved by the local ethics committee of the University Witten/Herdecke (185/2019) and was conducted according to the Declaration of Helsinki.

Following surgery, patients were screened for complications related to the procedure (neurovascular injury, infection, impaired wound healing, re-operation) or events of instability or re-dislocation. At the final follow-up, all patients were assessed for clinical stability and rotator cuff strength. The range of motion of both shoulders was measured by applying a goniometer. The level of sports activities and return to work were recorded. If patients were already retired, they were asked to estimate the potential to return to their former occupation. Outcome scores were measured including the Constant Score (CS; [2, 6]), Rowe Score (RS; [19]), Melbourne Instability Shoulder Score (MISS; [28]), the Western Ontario Shoulder Instability Index (WOSI; [12]), and the Subjective Shoulder Value (SSV; [7]).

The size of the glenoid lesion was calculated according to Itoi et al. [11]. The calculation was performed by en-face three-dimensional (3D) computed tomography (CT). The fractures were classified according to the Scheibel classification, where acute lesions of the glenoid are categorized as “Type 1” (1a: Osteochondral avulsion lesion, 1b: Solitary glenoid rim fracture, 1c: Multifragmented glenoid rim fracture; [21]). Postoperatively, the patients underwent 3D CT of the scapula to assess fracture reduction and button placement (Fig. 5). These CT scans were examined for fracture reduction and the remained fracture step (in millimeters) by two radiologists. For radiological follow-up, x‑rays were obtained in three views (anteroposterior, y view, axial view) to assess fracture healing (Fig. 6) and onset or progression of glenohumeral osteoarthritis according to the Samilson–Prieto classification [20].

Statistical analysis with calculation of mean values and standard deviations was performed using Microsoft Excel 2019 (Microsoft, Redmond, WA, USA). Further statistical testing was performed applying SPSS (Version 21, IBM Statistics, IBM, Armonk, NY, USA): For clinical follow-up active range of motion of the affected shoulder was compared with the contralateral shoulder using the Wilcoxon test. For comparison of patients with versus without rotator cuff injuries, the Mann–Whitney U test was applied. A value of p < 0.05 was considered significant.

Concomitant injuries were found in 21 cases, mainly Hill–Sachs lesions (16 of 23). Furthermore, concomitant injuries requiring surgical treatment were found in 13 cases (Table 1): Four patients had injuries of the rotator cuff; two of these with a full-thickness tear of the supraspinatus tendon, one with full-thickness tears of the supraspinatus tendon plus partial tear of the infraspinatus tendon, and one patient with traumatic partial tears of both the infraspinatus and the subscapularis tendon plus a chronic retracted tear of the supraspinatus tendon. The supraspinatus tendon tears were all treated arthroscopically in a double-row technique, except for the chronic tear, which received debridement. One patient had a dislocated fracture of the greater tuberosity that was treated with a screw osteosynthesis via a mini-open delta approach. Undislocated greater tuberosity fractures in two patients were treated conservatively, since the fluoroscopy at the end of surgery did not show any displacement. Overall, 11 patients had a superior labral anterior–posterior type II lesion that was treated with tenotomy or tenodesis of the long head of the biceps. One of the patients with a full-thickness rotator cuff tear had a nondisplaced coracoid fracture that was treated conservatively. One patient showed a unique injury pattern with concomitant superior labral anterior–posterior type IV lesion, posteroinferior labral tear, acromioclavicular joint injury Rockwood type 5, and a nondisplaced acromion fracture. This patient received long head of the biceps tenodesis, posterior labral refixation, and minimally invasive acromioclavicular joint reconstruction.

Complications related to the surgery (neurovascular injury, infection, impaired wound healing) were not observed in any of the 23 cases. No patient required re-operation. After a minimum of 6 months and an average of 15.0 months (range: 6.0–34.5), 22 patients were available for clinical follow-up (three female, 19 male). One patient (with an additional supraspinatus tendon tear) had died from cancer. The average CS was 83.2 ± 16.7 points (93.4% ± 18.8% compared to the contralateral side), the RS was 90.7 ± 10.4 points, the MISS was 88.3 ± 14.5 points, the WOSI was 82.9% ± 16.7%, and the SSV was 86.9% ± 16.1%. The average range of motion was 171.4 ° ± 22.7 ° of flexion (contralateral side, 180 ° ± 0 °; p = 0.11), and 170.5 ° ± 23.6 ° of abduction (contralateral side, 179.6 ± 2.1; p = 0.07).

At the time of follow-up, no patient had suffered recurrent shoulder dislocation or a feeling of instability. We could not detect the presence of the apprehension sign in any of these patients. The evaluation of the subscapularis function revealed a negative internal rotation lag sign and belly-off sign in all patients. No patient showed clinically detectable differences in the lift-off test compared with the contralateral side. Of these patients, 21 reported return to work, and one patient had not returned to work 7 months after surgery. Furthermore, 16 patients reported that they had returned to sport at the same level, while six patients were at a slightly reduced level.

Four patients with concomitant rotator cuff tears or displaced greater tuberosity fracture were on average 7 years older than patients without these injuries to the rotator cuff mechanism (57.0 vs. 49.8 years; p = 0.37), and they scored significantly lower for the CS (70.5 vs. 98.4; p = 0.003), the SSV (65.0 vs. 91.8; p = 0.005), and the RS (77.5 vs. 93.6; p = 0.019). They also scored lower in the MISS (80.3 vs. 90.1; p = 0.34) and the WOSI (65.3% vs. 86.8%; p = 0.053); however, without statistical significance.

Several limitations of the present study have to be mentioned. We conducted a retrospective analysis, and there was no control group. We only presented a short-term follow-up. Only x‑rays with three views were obtained for the follow-up, as we believed that these were sufficiently valid for assessment and that CT scans would expose the patients to too much radiation. Finally, a main limitation of the surgical method presented here is that it is not applicable for smaller fragments or comminuted fractures: As described earlier, if the anterior fragment is too small (from medial to lateral) then the anterior button cannot engage upon the fragment, but will cut out medially. As for the two cases excluded from this study, the implant was just removed through the anterior portal and a standard arthroscopic method (e.g., the Sugaya technique) was applied.