Introduction

The surgical treatment of complex proximal humeral fractures remains controversial particularly in the elderly [1–13]. Although hemiarthroplasty has long been considered the gold standard treatment, particularly in patients older than 65 years [1, 2, 4, 9, 12, 14–16], the functional outcome of endoprosthesis is not always satisfactory especially in the elderly due to poor bone stock quality, and in cases of insufficient cuff repair, muscle fatty degeneration, tuberosity malunion or non-union, loss of tuberosity reduction and consequent tuberosity resorption [2, 4, 6, 14, 17, 18]. Reverse prosthesis was proposed as a means to prevent complications due to tuberosity malunion or non-union and secondary cuff insufficiency [3, 10, 19–23]. Although this procedure results in adequate pain relief, the range of motion, particularly in terms of rotations, is limited and tuberosity healing is not always achieved.

Our hypothesis was that a novel surgical technique using a bone graft called “bCAT” (bone Collar And Tie), which fills the void resulting from bone loss under the tuberosity, would (1) enable the tuberosity to be placed more laterally and would stretch the residual cuff; and (2) help to promote tuberosity healing thereby preventing resorption from bone loss. The hypothesis stems from the observation that when the greater tuberosity is multifragmented, resorption is more frequent and is often visible on radiograms as a thin cortex layer early as 3 months after surgery.

The purpose of this study was to evaluate the clinical outcome of the bCAT technique versus a technique that does not involve a bone graft, namely the puzzle piece reconstruction (PPR), a procedure inspired by technique described by Brems for the treatment of fractures with the hemiarthroplasty [24].

Methods

Between January 2005 and December 2011, we implanted a reverse shoulder prosthesis in 46 patients affected by a complex-displaced proximal humeral fracture. Inclusion criteria for reverse shoulder prosthesis were subjects aged 69 years or above affected by complex four part-fractures and irreparable fracture dislocations. Exclusion criteria include circumflex nerve lesion and neurological disorders (chorea and epilepsy). Fractures were classified according to Neer in 4-part fracture and 4-part fracture/dislocation using preoperative radiographs (true antero-posterior view); computerized tomography (CT) scanning was used to evaluate the morphology of the fractures and the state of the glenoid. According to the newest Orthopaedic Trauma Association’s (OTA), classification fracture was 11-C [25]. Additionally, patients with a complex 4-part anterior fracture dislocation underwent Doppler examination to check the function of the axillary artery. In 44 cases we used the SMR Modular Shoulder System (Lima-LTO, San Daniele del Friuli, Udine, Italy) and in 2 patients we implanted a cemented Delta CTA™ reverse shoulder prosthesis (DePuy Orthopaedics Inc., Warsaw, IN, USA). The SMR prosthesis was cemented in two cases, and uncemented in the remaining 42 cases. In all cases, we used a concentric glenosphere measuring 36 mm. Average age was 73.5 years (range 69–95 years). In 22 patients we used the PPR technique to reconstruct the tuberosity fragments around the prosthesis, whereas in 24 patients we implanted a reverse shoulder prosthesis using the bCAT technique in which, in addition to tuberosity reconstruction using the PPR technique, the fractured humeral head is modelled into a “collar” shape and placed in the void under the tuberosities and around the prosthesis like a “tie”. From 2007 we start to use the bCAT surgical procedure and we still use it for the satisfactory results achieved. To obtain a homogeneous group, we compared the cases in which was used the same prosthesis with similar follow-up.

The decision regarding which procedure to use since 2007 has been established according to a random method.

Here we report the surgical technique and outcome exclusively of the patients (n = 40) who were treated with an uncemented SMR prosthesis and had at least 24 months of follow-up (mean 45.8 months, range 24–84 months). Twenty patients were treated with the PPR technique (group A) and 20 patients were treated with the bCAT technique (group B). All patients reviewed underwent an X-ray examination (see in the paragraph Assessment) and CT scanning to determine the healing and position of the tuberosities, and were evaluated with the absolute and age-adjusted Constant-Murley score [26].

Assessment

Patients were evaluated clinically and radiologically at 1, 3, 6 and 12 months after surgery. Radiographic follow-up assessment included standard antero-posterior and axillary lateral X-rays. On X-rays, with the shoulder in neutral rotation, we measured the distance between the metallic line of the socket and the most lateral aspect of the tuberosity (tuberosity thickness index: TTI). CT was performed in all patients around the third month post-surgery in order to investigate position and healing process of tuberosity, position of glenosphere and screws used to stabilize the metal back. Outcome was evaluated using the absolute and age-adjusted Constant-Murley score.

Statistical analysis

Statistical analysis was carried out using the Statistical Package of Social Sciences (SPSS, Windows release 21.0; Chicago, IL, USA). Results are presented as mean ± standard deviation, with statistical significance set at p ≤ 0.05. The independent sample t test was used to compare means, while the Fisher exact test was used to compare proportions.

Surgical technique

Surgery was performed at a mean of 11 ± 8.9 days after trauma, and average operation time was 75.7 ± 18.11 min. All operations were performed by the senior author (RR). In all cases, a standard delto-pectoralis approach was used with the patient in the beach-chair position. After exposure of the fracture site and evaluation of the fracture, two high strength sutures (n° 2) are passed through the tendon-bone junction of the tuberosities. The long head of the biceps tendon is cut at the superior part of the glenoid through the open rotator interval, after which the humeral head is easily identified and removed using a Kocher clamp. The tuberosities are mobilized and opened by pulling them up with four sutures, and the morphology of the fracture is evaluated in terms of entity of bone loss of each tuberosity and number of fragments.

Glenoid preparation

We place the metal-back glenoid implant slightly below the anatomical centre of the glenoid so that the glenosphere is flush to or only a few millimetres below the lower part of the glenoid. We try to implant the glenosphere at 0° retroversion and with an inferior tilt in the frontal plane a few degrees less than 8°–10° to prevent scapular notching. The central peg of the glenoid baseplate is secured and primary stability is enhanced with two 6.5-mm screws, one proximal and one distal (SMR implant) or four screws (DePuy implant). Finally, the definitive 36-mm glenosphere is screwed into place.

Humeral preparation

After exposure of the fracture site, the greater and lesser tuberosities are identified and tagged with n. 5 Ethibond non-absorbable sutures while preserving soft tissue attachments as much as possible. We use intramedullary stem trials (from 12 to 24 mm in diameter) to achieve the correct fit and primary stability. The length of the prosthesis is determined in relation to calcar length. If the calcar is fragmented or if no calcar remains, the prosthesis length is determined by reduction of the tuberosity fragments relative to the diaphysis. Once the humeral stem size has been selected, the assembled trial prosthesis (stem and humeral body) is inserted with 0° of retrotorsion, and the reduction of the greater and lesser tuberosities on the shaft fracture lines is checked. What remains of the supraspinatus tendon is removed starting from the footprint up to the muscle junction.

Puzzle pieces reconstruction

The PPR technique requires the precise identification of the two tuberosities and other aspects of the fragmented meta-humeral epiphysis. In this procedure, the fragments are identified and reconstructed like a puzzle to obtain the correct relationship between tuberosity height and the cephalic cap of the implant [24]. To achieve stable synthesis of the tuberosities, we perform osteo-sutures with non-absorbable wires in a fashion inspired by the Boileu technique [17]. This technique, adapted to reverse prosthesis, consists in implanting two parallel wires, one above and one below the tuberosities; the wires are placed around the neck of the prosthesis and tied together horizontally so as to tighten the tuberosities around the prosthesis. Then three pairs of wires are attached vertically to the pair of parallel wires, and passed through holes in the diaphysis. The greater tuberosity is cleared of the supraspinatus tendon, while the lesser tuberosity is detached from the coraco humeral ligament to obtain satisfactory exposure. Once all the fracture lines of the tuberosities are identified, the fragments are reduced taking care not to remove the periosteum.

The bone collar and tie procedure

We prepare the bCAT by removing cartilage from the humeral head using an oscillating saw or a burr and a Luer Rongeur (Fig. 1a; head without cartilage). Three-quarters of the head are cut perpendicularly to the axial plan of the cephalic cup so as to form a crescent. With a burr or with a Luer Rongeur, the central part of the cancellous bone is modelled to create a “collar” (Fig. 1b, c). The bCAT technique consists in placing the collar graft between the prosthetic humeral body and the greater tuberosity to fill the void in the lateral-posterior aspect of the prosthetic component. The space filled with the bCAT graft results both from the spongious bone loss and from the 0° of retrotorsion of the implanted humeral body (Fig. 2). The graft can be rotated around the proximal portion of the trial humeral stem like a tie and placed so as to fill the space between the greater tuberosity and the humeral stem (Fig. 3). Generally, the positioning of the bCAT on the humeral stem is done transversally from the posterior part of the residual calcar to the postero-superior insertion of the teres minor and the infraspinatus tendon. The graft, which can move freely around the humeral body, provides a stable support on which to tie the tuberosities thereby avoiding their collapse, and consequently obtaining the correct rotator cuff tension, particularly the posterior tension (Fig. 4).

Fig. 1
figure 1

Preparation of the bCAT graft: a fractured humeral head. b Modelling of the humeral head into a collar shape. c The bCAT graft on the reverse prosthesis

Full size image
Fig. 2
figure 2

Example of the relationship between the bCAT graft and the tuberosity a difference of space below tuberosity in 30° of retroversion and 0° of retroversion of humeral component. Grey circle indicates 30° of retrotorsion, darker grey circle indicates 0° of retrotorsion b bCAT inserted between the humeral prosthesis and tuberosity

Full size image
Fig. 3
figure 3

The bCAT graft can be inserted around the prosthetic stem and rotated as required to reach the best position in the back and lateral aspect of the prosthesis

Full size image
Fig. 4
figure 4

Post-operative X-ray of left shoulder. Arrows indicate the bCAT under the greater tuberosity

Full size image

Rehabilitation

After surgery, the patient wears a shoulder sling for 4–6 weeks. The patient is instructed to practice pendulum exercises, and passive and active mobilization of the elbow, wrist and hand. Twenty-one days after surgery, the patient starts mobilization in warm water according to the Liotard et al. [27]. Active movements are allowed 6 weeks after surgery.

Complications

Axillary nerve damage due to the fracture and trauma was identified in seven cases by an electromyographic (EMG) study (2.10 %), and in one case, it was associated with ulnar nerve compression. The damage regressed within 6–8 months after surgery and without complications associated with prosthesis stability. No complications, re-operations and no dislocations were observed.

Results

Forty patients (32 females, 8 males) with a mean age of 73.8 ± 4.73 years were included in the study and divided into two groups, according to the different surgery procedures: group A was represented by 20 patients who underwent the PPR technique and group B by 20 patients who underwent the bCAT technique. The two groups did not differ in terms of age or sex (Table 1). After a mean follow-up of 45.8 months, shoulder function was compared between the two groups, the clinical results are shown in Table 1; elevation, abduction and extrarotation were significantly higher in group B than group A. The differences in articular function are significant (p < 0.001, Table 1) for the three variables: elevation, abduction and extrarotation (Fig. 5).

Table 1 Statistical Evaluation of two groups
Full size table
Fig. 5
figure 5

Clinical results 5 months after surgery. The right shoulder shows excellent function particularly regarding rotations. a X-ray: complex 4 part fracture; b X-ray post-surgery; c X-ray of right shoulder 45 days after surgery; d extrarotation, e flexion, f intrarotation

Full size image

Mean absolute and age-adjusted Constant-Murley scores were, respectively, 55 and 67.85 % in group A, and 70.8 and 83.85 % in group B.

The radiographic study showed resorption and dislocation of the greater tuberosity in 40 % (8/20) patients of group A, whereas resorption occurred in only 15 % (3/20) patients of group B; however, this difference was not significant (p = 0.155) due to the low sample. The average TTI was 8 mm (5–10 mm) in group A and 14 mm (12–20 mm) in group B. We observed peri-prosthetic calcification in 5 (12.5 %) cases, but without any clinical sign, as pain or limitation of motion.

Discussion

The functional outcome of hemiarthroplasty in elderly patients with 3- or 4-part displaced humeral fractures is unpredictable because tuberosity non-unions and malunions can occur in 51 % and because of secondary cuff tears or insufficiency [6, 14, 17, 18]. Reverse shoulder prosthesis was initially proposed as an option to treat such cases [3, 10, 19–23].

Gallinet et al. [20] reported a comparative study of 40 patients treated with hemiarthroplasty implants or reverse shoulder implants and concluded that the latter are better in terms of predictability of function and rapid recovery. The same authors observed that reverse shoulder implants seem to be indicated for patients over 70 years of age, can impair rotation, and glenoid notching impacts on long-term implant durability. Martin and Iannotti [28] recommended limiting the implant of reverse shoulder prosthesis to selected cases because of the technical difficulty of the operation, and suggested that modification of implant design could improve functional outcome in the elderly. Gerber et al. [29] observed that proper patient selection and attention to technical details are required in order to reduce the high complication rate associated with reverse arthroplasty, which is three times that of conventional prosthesis.

Cazeneuve et al. [30] reported a very low Constant score (53 absolute and 67 % age-adjusted) in cases of reverse shoulder arthroplasty without tuberosity healing, which demonstrates the importance of tuberosity healing for a satisfactory functional outcome. They also reported 11 % of prosthesis dislocations, and radiologic signs of loosening in 63 % of patients. In another study of 35 patients with a follow-up of one to 16 years, the same authors reported a reoperation rate of 17 %, and an increase in scapular notching over time [31]. Bufquin et al. [3] reported a mean Constant score of 44 (66 % age-adjusted) and complications after reverse shoulder arthroplasty performed in elderly patients, i.e. secondary tuberosity displacement in 53 % of patients and scapular notching in 25 % at a follow-up of 22 months. In a more recent paper, Gallinet et al. [32] compared patients treated with arthroplasty who achieved tuberosity healing with those who did not achieve tuberosity healing, and concluded that clinical outcomes and rotational ability were better when tuberosities were repaired around the stem. Our assumption is that satisfactory tuberosity healing and position is crucial to achieve a satisfactory clinical outcome of reverse shoulder arthroplasty.

We started to use the bCAT technique in 2007 and have reported the preliminary results in international congresses in 2009, 2010 and 2011 [33, 34]. In 2011, Levy and Badman [35] reported a similar technique called “horse shoe” procedure in seven patients. The technique consists in using a graft sutured to the tuberosities, the shaft and prosthetic stem according to the Frankle technique [36]. The advantage of the bCAT technique is that the crescent-shaped graft is free to rotate as required to reach the best position in the back lateral aspect of the prosthesis so as to fill the void resulting from bone loss and to restore the tension of the teres minor. This procedure fulfils two aims, it fosters the healing process and enables the tuberosity to be placed more laterally so as to improve the residual cuff tension.

To evaluate the effectiveness of the bCAT technique we have compared its outcome in patients with a 4-part proximal humeral fracture with an age-matched control group in whom the graft was not used. Function, particularly rotation, was better in the case group (group B). In fact, the average anterior elevation was 160° and the mean external rotation was 43° in group B versus 125° and 15°, respectively, in group A. The radiological findings showed a low rate of resorption of tuberosity (40 %) and an increased thickness of the greater tuberosity (mean TTI: 12 mm) in group B. Therefore, the bCAT technique can improve healing and increase the thickness of the greater tuberosity. This, in turn improves the lever arm of the tendon cuff thereby increasing abduction and external rotation without changing the reverse prosthesis design. This better function, particularly as regards extrarotation, greatly improves the patient’s quality of life especially elderly patients. In addition, tuberosity healing favours prosthesis stability, thereby reducing the possibility of dislocation and scapular notching; however, a longer follow-up is necessary to verify this.

Conclusions

The key to a satisfactory outcome, in terms of stability and function, of reverse prosthesis surgery in patients affected by complex proximal fractures, is correct tuberosity positioning and healing. Both the PPR and bCAT techniques restore tuberosity position and cuff tension. We believe that the bCAT technique can promote a better healing and correct positioning of the tuberosities in elderly patients, thereby resulting in good functioning of the residual cuff without changing the reverse prosthesis design. Based on our findings, the bCAT technique can improve the clinical outcome of reverse prosthesis surgery without complications in this difficult-to-treat subset of patients. The clinical improvement in our patients probably resulted from the good tuberosity healing and the increased tension of the residual cuff. Larger studies and longer follow-ups are necessary to verify the benefits, and eventually any limits of this technique.