Minimally invasive fixation with modified palm tree technique for proximal humerus fractures: Outcomes in a series of 132 patients

Abstract

Introduction

Various surgical techniques and implants are available for surgical treatment of significantly displaced proximal humerus fractures. We describe a minimally invasive technique using 3 curved wires, inserted in a retrograde fashion into the humeral head. These are aimed to diverge within the humeral head to provide three separate 3-point fixations to achieve good stability. We present the results of proximal humerus fractures managed with the modified palm tree technique.

Methods

A retrospective analysis of data collected prospectively including demographics, radiographs, clinical outcomes, complications and revision surgery for patients treated with the palm tree technique was performed.

Results

Between 1998 and 2017, 132 patients underwent fixation with this technique. Average age was 61.8 years. Fifty-Eight fractures were 2 part, 46 were three part and 28 were four part. Average follow up was 26 months. In three to four part fractures, a bone graft substitute block was used behind the humeral head fragment for structural support. There were 11 early revisions (8.3%). In 7 cases the fixation failed early and was revised to other implants like angular & locking plates (4), hemiarthroplasties (2) and reverse arthroplasty (1). In 4 cases the construct was revised to achieve better positioning of the wires. From the 125 remaining patients, 120 achieved union(96.8%). There were 3 painless fibrous non-unions and 2 painful non-unions requiring revision. The mean final Constant score was 75.5 and subjective shoulder value was 7.8/10. Nine patients (6.8%) developed avascular necrosis of the humeral head of which three patients were revised later to an arthroplasty.

Conclusion

This technique is a simple, minimally invasive technique which can be used for two, three and four part fractures with good functional outcomes and high union rates. No metalwork remains in the proximal humerus should another procedure like arthroplasty be required in case of avascular necrosis of the humeral head or fracture sequela.

1. Introduction

Proximal humerus fractures are common injuries seen predominantly in the elderly population.^1,2 Undisplaced and minimally displaced fractures can be treated conservatively with good outcomes.³ With significantly displaced fractures, non-operative management could lead to complications like mal-union, stiffness and reduced range of motion compromising the function of the shoulder.⁴ The commonly employed techniques and implants for treatment of proximal humerus fractures include locking plates,^5,6 intramedullary nails,⁷ percutaneous pinning⁸ and flexible intramedullary nails.⁹ The disadvantages of techniques with open reduction include extensive soft tissue dissection, periosteal stripping and associated risks of osteonecrosis.¹⁰ On the other hand, closed reduction techniques do preserve the soft tissue envelope around the shoulder, however they have their own set of complications like suboptimal reduction, loss of reduction and pin migration.^11,12

We describe the clinical and radiological outcomes of a simple, reproducible and minimally invasive percutaneous procedure which a modification of a Kapandji’s minimally invasive technique for the fixation proximal humerus fractures.¹³ Our technique envisages to circumvent the disadvantages of percutaneous fixation like wire migration and loss of reduction and provide a stable fixation.

2. Methods

A retrospective analysis of data collected in a computerised database for patients treated with the modified palm tree technique between 1998 and 2017 was done. The study was performed in accordance of the ethical standards laid down by the 1964 Declaration of Helsinki and after approval of the local institutional review board. Demographic data like age, gender and side operated was noted. The modified palm tree technique¹⁴ was used for treatment of all displaced two part and 3 to 4 part fractures and was slightly different for the two groups. This series was a consecutive series of surgically treated proximal humerus fractures predominantly by two surgeons and their trainees in a shoulder and elbow unit. All proximal humeral fractures that were sufficiently displaced so as to require surgical management (including both 2 part and 3–4 part fractures) were treated with this technique. Conservatively treated fractures that displaced on successive radiographs such that they needed surgery were also included in the study. This technique was not used for fracture-dislocations and comminuted multi-fragmentary fractures in the elderly which were managed by open reduction and internal fixation, hemiarthroplasty or reverse shoulder arthroplasty.

The range of motion at the final clinical follow up including forward elevation, abduction, external rotation and internal rotation was compiled. The functional clinical outcome was assessed using the Constant-Murley Score¹⁵ and the Subjective shoulder value at the last visit. The radiographs at the last visit were analysed by an independent examiner. Early complications such as failure of fixation, wire migration and revision surgery were noted. Delayed sequelae like avascular necrosis of the humeral head, malunion and revision surgery if any in the subsequent years were noted. All data was recorded in a computerised database (Excel Sheet- Microsoft Corp Redmond, WA) and mean ranges of motion and clinical scores along with standard deviations were calculated.

2.1. Surgical technique

2.1.1. Closed reduction and percutaneous pinning for two part fractures

The surgery is performed in the standard beach chair position. The arm is draped free to allow full movement and a narrow arm board is used to provide support. The reduction of the fracture is done under fluoroscopic guidance by reducing the humeral shaft (abduction and external rotation) on to the displaced and rotated humeral head. (Fig. 1). Once the fracture is reduced, an incision is made just at the level of the insertion of the deltoid muscle. This level is identified on the lateral aspect of the arm where the ‘V’ shaped convergence of the deltoid muscle is noted (Fig. 2). This area is a ‘safe window’ of about 4 cm between the axillary nerve above and the radial nerve below (Fig. 2). The surgeon must avoid performing the incision further distal or too proximal from this entry point. The lateral cortex is approached by blunt dissection. A 4.5 mm drill hole is made in the lateral cortex which is angled obliquely superior. Three 1.8 mm wires (usually obtained from cannulated screws set) (Fig. 3) are pre-bent into a gentle curve (Fig. 4) and mounted on a T handle on a Jacob’s chuck. The wires are introduced into the medullary cavity through the drill hole and passed in a retrograde direction across the fracture and towards the humeral head, where they are aimed to diverge away from each other (similar to a ‘palm tree’). Rotation of the T handle helps control the direction in which the wire is headed. The wires are then impacted into the subchondral bone. We thus achieve a three point fixation for each of the wires. The three points of fixation include the entry point at the lateral cortex, medial cortex where the wires bounce off and the subchondral bone in the humeral head. (Fig. 5, Fig. 6). The wires are bent, cut and buried under the skin but remain palpable.¹⁴

Fig. 1.

Demonstration of patient positioning and closed reduction under image intensifier control.

Fig. 2.

Incision for insertion of wires at the insertion point of the deltoid muscle.

Fig. 3.

The simple surgical kit.

Fig. 4.

Pre-bending of the wires.

Fig. 5.

Common fracture patterns in the series; a. Varus displaced fracture; b. Displaced surgical neck fracture c. Three part fracture with displaced greater tuberosity.

Fig. 6.

Antero-posterior radiograph of the palm tree wire insertion (for Fig. 5c fracture).

2.2. Surgical technique

Mini-open reduction with use of bone graft substitute block and palm tree fixation for 3 to 4 part fractures.

In case of displaced 3 to 4 part fractures, the fracture is approached through minimal antero-superior approach (Neviaser- Mackenzie). When needed, the anterior fibres of the deltoid were elevated off the anterior margin of the acromion to improve exposure. The distal limit of the incision was proximal to the distal margin of the bursa thereby staying proximal to the axillary nerve. The greater tuberosity and the lesser tuberosity are released and held with stay sutures through the tendon insertion into the bone. The humeral head fragment is elevated and a bone graft substitute (BGS) block is inserted to provide structural support (Fig. 7). The most common bone block used was 12.5 x 12.5 × 10 mm. Sometimes a bone graft wedge was used instead when there was eccentric metaphyseal comminution and head collapse. An approximation of the greater and lesser tuberosities is performed by tying the stay sutures in order to ‘close the book’ behind the humeral head fragment and the BGS block, and hence convert the fracture to a 2 part fracture. Now, the palm tree fixation is performed through a separate incision at the deltoid insertion level as described above (Figs. 2, Figs. 7 and 8).¹⁴

Fig. 7.

Palm tree technique in 3–4 part fracture; a. Mini-open anterosuperior (Neviaser-Mackenzie) approach and reduction of the greater tuberosity; b. Insertion of bone substitute block under fluoroscopy; c. Insertion of the wires through the ‘safe window’; d. Final fixation with the modified palm tree technique and ‘closing the book’- suture fixation of the tuberosities.

Fig. 8.

Palm tree technique in a 3 part fracture; a. Three part fracture with displaced greater tuberosity; b. Open reduction and use of bone graft substitute block followed by palm tree fixation; c. and d. Radiographs after removal of wires showing union.

2.3. Post-operative protocol and implant removal

Post-operatively, the arm is placed in a sling for 3 weeks. Passive mobilisation is started if there is any sign of callus formation at this stage.(Fig. 9). If there is no callus then immobilization is continued for a further three weeks. The wires are removed under a short general anaesthetic at six weeks. (Fig. 8, Fig. 9).

Fig. 9.

Radiologic outcome of the fracture with good union & anatomical result (fracture Fig. 5c); a. Union of fracture with wires in situ; b. Radiograph after removal of wires.

3. Results

A total of 132 patients underwent fixation with the modified palm tree technique for proximal humerus fractures between 1998 and 2017. Of these 96 (72.7%) were females while 36 were male. The mean age at the time of fixation was 61.8 years (range 16–92 years). Most patients were operated within two weeks of their injury. However, there were a few (8 patients) who were operated in the 3rd week due to various reasons including delayed presentation and patient having other injuries needing prioritization. These also included fractures that were initially managed conservatively but the fracture displaced into varus or valgus on follow-up radiographs. Fifty Six percent of the fractures were on the right side. All fractures were classified according to the Neer Classifcation into displaced two part, three part or four part fractures.¹⁶ Each part had at least 1 cm displacement or 45° angulation as described by Neer. There were 58 fractures that were two part (44%) and 74 (56%) three to four part fractures. The average clinical follow up was 26 months (6–156 months). Patients were usually discharged after two years of follow up. They were clinically reviewed until 2 years post-operatively to rule out development of avascular necrosis of the humeral head.

There were a total of 11 (8.3%) early revision surgeries. Of these, four cases that were operated initially by a trainee required revision to a new palm tree construct to achieve better positioning of wires. The reason for failure of the construct in two patients was bunching of the wires in humeral head thereby making them act as a single eccentric wire rather than a well distributed supporting construct. In the other two patients the positioning of the wires led to early cutout through the fracture site necessitating revision.

Further, an arthroplasty (2 hemiarthroplasties and one reverse shoulder arthroplasty) was performed for three patients when the under-estimation of the fracture comminution led to failure of the construct and unacceptable displacement of the humeral head. Additionally, two were revised to an angular plate and two were fixed using a locking plate (combination of factors including collapse of the fracture in unacceptable varus/valgus, fracture comminution, poor wire hold and wire cutout). 125 patients remained with modified ‘Palm tree’ fixation. Of them, 120 achieved union (96.8%).

There were three cases of pin irritation and adventitious bursitis which settled down after the pins were removed. Three patients had superficial infection and erythema at the proximal suture line with settled down with oral antibiotics. There were no cases of deep-seated infections needing washout. Four patients had axillary nerve neuropraxia pre-operatively which recovered in the succeeding follow-up reviews.

There were 3 painless fibrous non-unions and 2 painful non-unions requiring revision to a plate fixation and to a reverse shoulder arthroplasty. Nine patients (6.8%) developed osteonecrosis of the humeral head on succeeding follow-up visits between 3 and 18 months after the index surgery. All of them occurred in 3 and 4 part fractures. Of these, 5 remained with a reasonable function, but the remaining 4 patients were offered surgery. Of these, 2 underwent hemiarthroplasty, one underwent reverse shoulder replacement and one patient declined further surgical intervention [Table 1].

Table 1.

Outcomes of modified palm tree fixation.

The mean final Constant score was 75.5 with the mean deltoid strength component being 19. The mean subjective shoulder value at the last clinical visit was 7.8/10. The mean range of active forward elevation was 130° and 125° of abduction. The average range of active external rotation was 48°. The median range of active internal rotation was palm to L3 [Table 2].

Table 2.

Clinical Results with modified palm tree fixation.

	Clinical parameter	Result
1	Mean Forward elevation	130°
2	Mean Abduction	125°
3	Mean External rotation	48°
4	Mean Constant score	75.5
5	Mean subjective shoulder value	7.8

4. Discussion

Locking plates are a popular choice in treatment of proximal humerus fractures. They offer rigid angular fixation. However in an osteoporotic bone, screw cut out and loss of fixation may be seen.¹⁷ The use of a locking plate also means extensive soft tissue dissection for exposure, reduction and fixation, thereby, increasing the risk of avascular necrosis of the humeral head.^17,18 A systematic review of 514 patients treated with locking plates reported an overall complication rate of 49% with the use of locking plates, with an eight percent rate of screw cut out, and 14% rate of reoperation and 10% risk of avascular necrosis¹⁹ Brunner et al. reported in a prospective multicentre analysis involving eight trauma units (158 proximal humerus fractures in 157 patients treated with locking plates), an overall complication rate of 44% and re-operation rate of 25%.²⁰ It has also been noted that early mobilization is not necessarily associated with improved outcomes.^21,22

These problems are offset to an extent by the minimally invasive, less rigid fixation techniques.²³ These fixation methods like the palm-tree technique work with limited exposure, minimal periosteal stripping and preserving the medial soft tissue hinge thereby providing an optimal environment for fracture healing. Furthermore as these implants have more axial flexibility allowing the implant to comply with the osteoporotic bone without cutting out through the soft bone.²⁴ As the wires are removed at 6 weeks, there is no metalwork left inside the bone. In case of development of avascular necrosis or fracture sequelae arthropathy, arthroplasty can be easily performed.

Kapandji first presented this technique in 1974 calling it the ‘palm tree’ technique and it involved the use of three divergent k wires for fixation of proximal humerus fractures.¹³ He based his technique on anchorage of three divergent pins in the cancellous bone. Results using this technique have been variable. A series by Le Bellec et al. noted 71% satisfactory results and a complication rate was 48.4% with the use of Kapandji’s technique. The complications included fixation failure, reflex sympathetic dystrophy, radial nerve palsy, osteonecrosis and fracture.²⁵

We have modified our technique in a few aspects compared to the Kapandji technique. In our technique, all the wires are introduced through a single drill hole of 4.5 mm in the lateral cortex allowing for a good interference fit. The first two wires are introduced with their blunt end first to allow them to slide on the medial cortex and fixation is obtained in the subchondral bone. The third wire is introduced with the sharp end first and it slides off the other wires.¹⁴ The construct achieved is a 3 x ‘3 point fixation’ where each wire is fixed (locked) in the lateral cortex entry drill hole, the medial cortex and the subchondral bone. This relatively rigid construct offers the advantage of achieving angular and rotational stability to the fracture, yet preserving axial flexibility. Such stability is achieved by the divergence of the wires in the proximal fragment and by the three-point fixation of the wires (subchondral bone, medial cortex and tight fitting lateral cortex).^26,27 Fixation is not dependant entirely on the purchase of the wires in the osteoporotic subchondral bone. The minimal hardware is routinely removed with minimal risk, leaving no metalwork in the shoulder. This, we believe, eliminates the risk of adhesions around the metalwork and, therefore, less stiffness, despite our conservative postoperative regime.

The divergent nature of the curved wires allows for better distribution of the wires in the head of the humerus. Moreover, the risks such as pin migration into the joint are lower with curved wires as compared to straight wires.²⁴ Straight wires also tend to bunch up in the medial part of the proximal fragment providing less stability.²⁸ Biomechanical studies have noted that the divergent configuration is provides excellent rotational stability as opposed to other wire constructs.28, 29, 30

El-Alfy et al. (2010) used a slightly different modified palm-tree technique for fixation of displaced proximal humerus fractures in eighteen patients. At an average follow-up of 22 months they noted 100% healing without any loss of fixation or neurovascular injury. Each wire was inserted from a separate drill hole. They used a percutaneous k wires for fixation of the greater tuberosity. They also cut the wires outside the skin. They noted two cases of pin tract infection and three cases of skin necrosis around the wires. The mean Constant score was 73.²⁴ Levy et al. (2006) published a series of 26 proximal humerus fractures treated with the modified palm tree technique. The mean Constant score was 75.3 at an average follow-up of 28 months. In the event of a displaced greater tuberosity, they performed an open reduction using the anterosuperior approach.¹⁴

This study is the largest study with the use of this technique. Our results are comparable to other minimally invasive techniques in the literature [Table 3]. Due to the minimal fixation, this technique should be regarded for rehabilitation purposes as.

Table 3.

Outcomes with percutaneous fixation techniques of proximal humerus fractures.

Study	Technique	Number of patients (n)	Mean follow-up	Outcome measures	Additional comments
Ogawa et al.31 (2006)	Retrograde intramedullary multiple pinning through deltoid ‘V’	n = 10	Minimum 12 month follow-up	One collapse and one Avascular necrosis. Mean constant score within 92% of the uninjured side for other eight patients	Used for valgus impacted fractures only
Bogner et al.32 (2016)	K wire fixation with humerus block device	n = 129	23 months	Constant Score 2 part-67.7 3 part-67.9 4 part-43.0	Loss of reduction-7.8% Nonunion- 3.5% Avascular necrosis- 6.2%
Levy et al.14 (2006)	Modified palm tree technique	n = 26	28 months	Mean Constant Score- 75.3	Displaced greater tuberosities reduced by mini-open anterosuperior approach.
El Alfy et al.24 (2011)	Modified palm tree technique	n = 18	22 months	Mean Constant score- 73	Three separate drill holes proximal to the deltoid tuberosity. No non-unions. Two pin tract infections and three episodes of skin necrosis.
Calvo et al.33 (2007)	Closed reduction and percutaneous pinning with minimum 3 wires	n = 73	13.6 months	Mean Constant score- 65.8	Worst results in four part fractures
Blonna et al.34 (2009)	Percutaneous pinning for varus displaced fractures	n = 32	30 months	Mean Constant score-72	Mean modified Constant score- 88 Quick DASH score-15
Fenichel et al.11 (2006)	Percutaneous pinning with 3–5 k wires for 2 to 3 part proximal humerus fractures	n = 50	30 months	Mean Constant score-81	Loss of fixation- 14%
Our Study	Modified Palm tree technique	n = 132	26 months	Mean Constant score – 75.5 Mean subjective shoulder value - 7.8/10	Elevation of the humeral head fragment and augmentation with bone graft substitute block in 3–4 part fractures with ‘closing the book’ – reduction and fixation of greater and lesser tuberosities.

“augmented conservative treatment” with immobilization for 3–6 weeks.

There is a risk of pin migration and loss of fixation. However, the risk of pin migration is reduced by the locking of the 3 wires in the lateral cortex drill hole. Only 11 patients (8.3%) required a revision operation due to early loss of fixation. Of these four were due to poor surgical technique in the first operation and were managed by re-doing the palm tree fixation. We believe that the use of bone graft substitute block, in 3–4 part fractures, to fill the void created by the collapsed osteoporotic bone and support the ‘elevated’ humeral head fragment from collapsing into the void, along with fixation of tuberosities together plays a key role. Risk of neurological injury can be averted by careful selection of the pin insertion site in the ‘safe window’ between the axillary and radial nerve.

We have not been able to compare this technique directly to others methods of fixation in this study. Although most proximal humerus fractures needing surgery were treated with this technique, we do not have data on those that were excluded and received other forms of treatment. Although this is a single centre study we do believe that the results are reproducible as many surgeons have used this technique over almost twenty years.

5. Conclusion

The palm tree technique is a minimally invasive procedure for fixation of displaced fractures of the proximal humerus. It does not affect the integrity of the rotator cuff and the deltoid muscle. The technique has provided good functional outcomes in this study. Careful patient selection, analysis of fracture pattern and fracture reduction remain keys to the success of this technique.

Funding

No funds were received in support of this study

Declaration of competing interest

There are no conflicts of interest to declare