Does the triceps-on approach affect alignment in total elbow arthroplasty? A cadaveric study

Abstract

Background

The triceps-on approach for total elbow arthroplasty has gained popularity due to the theoretical benefit of preserving the extensor mechanism. However, there is concern that the exposure may be reduced in comparison to a triceps-off approach and may affect the implant alignment achieved.

Method

Total elbow arthroplasties were implanted in 18 randomised, paired cadaveric elbows using the triceps-on or triceps-off approach. The bones were dissected out and the position of the implants measured relative to anatomical landmarks. The flexion/extension and varus/valgus angles, and the distance of centre of rotation from the anatomic centre of rotation in the sagittal plane for both components were obtained as well as the humeral component rotation relative to the transepicondylar axis.

Results

All humeral components were positioned in external rotation and all ulna components were placed in flexion. Seven components were positioned greater than 5° away from the ideal in one measurement, with no significant difference between the two approach groups.

Discussion

This unique study showed no significant difference in the alignment of the implants between the two approaches. These results support the theory that the triceps-on approach does not result in larger alignment errors in component positioning when performing total elbow arthroplasty.

Background

The common indications for total elbow arthroplasty (TEA) are changing. There is a reduction in the use of arthroplasty in the inflammatory arthropathies due to successful medical intervention.¹ However, there is an increase in its use for elderly trauma patients, either as the primary treatment or for delayed complications such as pseudo-arthrosis.² The complication rate for TEA is not insignificant with aseptic loosening, infection, skin breakdown, and triceps failure being among the most common.³

TEA can be performed through a variety of approaches with many involving reflecting the triceps from its insertion.⁴ More recently, a triceps-preserving approach has been advocated due to the benefit of preserving the extensor mechanism and allowing faster post-operative rehabilitation.^5,6 There is a suggestion that the intra-operative exposure may be reduced and could lead to difficulties in terms of implant positioning.⁷ It has been suggested that components with malalignment greater than 5° are linked to higher failure rates of the implant.⁸ However, other considerations, such as maintaining the integrity of the extensor mechanism, must be taken into account as triceps failure can have a catastrophic functional impact on patients.⁹

Therefore, a balance must be achieved between adequate exposure to allow correct implantation and preservation of as much anatomy as is possible. The aim of this study was to randomise matched cadaveric elbows and establish whether there is any difference in alignment between the two approaches, with the null hypothesis being that there is no difference.

Method

Eighteen paired elbow specimens were prepared for implantation of total elbow prostheses. Local ethical approval for the study was gained. The specimens were fresh frozen and thawed overnight prior to the study. They were then transected at the proximal humerus and distal forearm and were held in a vice at the humeral end in the lateral decubitus position. The triceps muscle belly was held within the vice to ensure the normal tension throughout the triceps tendon. Total elbow replacements (Coonrad–Morrey, Zimmer Biomet, Switzerland) were implanted into all specimens using either the ‘triceps-on’ or ‘triceps-off ’ approach. The ‘triceps-off ’ approach used was the Shahane approach.¹⁰ This splits the triceps with ¾ lateral and ¼ medial, the lateral ¾ being reflected from the olecranon along with anconeus, the ulnar nerve being protected beneath the remaining medial triceps. Repair of the triceps is therefore necessary with transosseous sutures. The ‘triceps-on’ approach, or Alonso-Llames approach, involves elevating the triceps from the intermuscular septum.^10,11 The ulnar nerve is released, with a cuff of fascia to allow subsequent repair, from the cubital tunnel and protected. In both approaches the collateral ligaments were sacrificed, as is routine when performing linked elbow joint replacement surgery.

Two experienced consultant elbow surgeons, who have clinically performed both exposures, undertook the surgical approaches in all cadavers. The specimens were randomised to surgeon, laterality, and approach using a random number generator.

Trial components were implanted with impaction to allow distal humeral and proximal ulna hold during analysis. A full range of sizes was available for use during implantation and good hold was achieved in all specimens without compromising on the position of the implant. Once implanted, the specimens were stripped of soft tissues and measurement of component positioning was undertaken by a single reviewer, blinded to surgeon and approach used. Centre of rotation (COR) of the ulna component was measured using photography. Each specimen was digitally photographed using a static, high resolution 12 megapixel camera (Fujifilm Fuji x10, Tokyo, Japan). The images were uploaded to Orthoview (Materialise, Plymouth, USA) and the distance of the COR of the ulna component was measured from the centre of a circle drawn around the remaining articular surfaces of the ulna (Figure 1).

Figure 1.

Determining the COR of the ulna component.

A vector was also recorded to establish the relative positions of the component COR from the anatomical COR. Rotation of the humeral component was measured relative to the transepicondylar axis using a k-wire passed through the epicondyles and another passed through the humeral component.^2,12–14 The angle of rotation was then measured using a goniometer. The position of the humeral component COR was then calculated by measuring the height and the anterior offset of the humeral component relative to the transepicondylar axis using an electronic calliper (Powerfix Profi Digital Calliper: Measuring range: 0–150 mm; resolution: 0.01 mm; accuracy 0–100 mm ± 0.02 mm, 100–150 mm ± 0.03 mm; maximum measuring speed 1.5 m/s) and a vector was calculated to represent the distance from the anatomical COR.

Varus/valgus and flexion/extension angles of the component shafts within the respective bones were measured by resecting a dorsolateral portion of the cortex proximally in the humerus and a dorsolateral portion in the ulna distally. This allowed visualisation of the prostheses whilst retaining fixation within the bone. The angle between the axis of the component and the axis of the shaft of the bone could then be measured using a goniometer.

All statistical analysis was undertaken using STATA 14 (StataCorp. 2015. Stata Statistical Software: Release 14. College Station, TX: StataCorp LP). The Mann–Whitney U test was used to detect significant differences in this unpaired, nonparametric data set. A Kruskal–Wallis calculation was used to detect differences in the two groups accounting for both approach used and surgeon performing the procedure. A significance level of 95% was used.

Results

No significant differences were detected in any of the measurements taken for either the ulna or the humeral component. This includes analysis for all variables including surgeon and approach.

All humeral components were placed in external rotation relative to the transepicondylar axis with no significant differences between the two groups (Table 1). The degree of external representation of the components is represented in Figure 2.

Table 1.

Summary of Mann–Whitney U tests.

No.	Variable	p value	Decision
1	Distance of humeral component centre of rotation from the anatomical centre of rotation distance	0.489	Retain null hypothesis
2	Vector of humeral component centre of rotation from anatomical centre of rotation	0.666	Retain null hypothesis
3	Humeral component alignment in the sagittal plane (flexion/extension axis)	0.73	Retain null hypothesis
4	Humeral component alignment in the coronal plane (varus/valgus axis)	0.863	Retain null hypothesis
5	Humeral component external rotation compared to the transepicondylar axis	0.666	Retain null hypothesis
6	Distance of the ulna component centre of rotation from the anatomical centre of rotation	0.161	Retain null hypothesis
7	Vector of ulna component centre of rotation from anatomical centre of rotation	0.931	Retain null hypothesis
8	Ulna component alignment in the sagittal plane (flexion/extension axis)	0.258	Retain null hypothesis
9	Ulna component alignment in the coronal plane (varus/valgus axis)	0.190	Retain null hypothesis

Figure 2.

Representation of the external rotation of the humeral component relative to the transepicondylar axis.

All ulna components were placed in flexion relative to the anatomical axis of the ulna; however, there were no significant differences between the two approach groups. Two ulna components were placed in greater than 5° flexion in each group. There were no significant differences when measuring varus/valgus positioning relative to the anatomical axis (Table 1).

All humeral components were implanted with the COR distal to the transepicondylar axis except for one. All implants were positioned with anterior offset. The average distance of the prosthesis COR from the anatomical COR was 4.4 mm. Implant COR achieved for all humeral components can be seen in Figure 3. There were no significant differences when comparing surgeon or approach (Table 1).

Figure 3.

COR of the humeral component relative to the transepicondylar axis in the lateral plane. Blue: triceps off. Red: triceps on.

When analysing ulna component COR all components were placed with dorsal offset with seven components placed proximal to the anatomical COR and 11 placed distal to the anatomical COR. The average distance from the anatomical COR was 6 mm. The ulna implant COR achieved is represented in Figure 4. Again, there were no significant differences when comparing surgeon or approach (Table 1).

Figure 4.

COR of the ulna component relative to the anatomical COR in the lateral plane. Blue: triceps off. Red: triceps on.

Discussion

The results of this study show no significant difference in the deviation away from known anatomical landmarks of the elbow joint between the two approaches. This study is one of the largest looking at TER prosthesis alignment in the literature and supports the use of the triceps-on approach when performing TEA.

We have taken the centre of the transepicondylar axis as a reproducible anatomical landmark¹⁵ and centre of a circle formed by completing the curvature of the ulna articular surface as the COR of the anatomical elbow joint.¹⁶ The humeral components were all placed in external rotation when compared to the native transepicondylar axis. This reflects the referencing from the posterior humeral surface. Whilst there are studies that show that the anatomical joint axis of the elbow joint relates to the transepicondylar axis, specifically being 3–8° internally rotated to this axis^12–15,17 there are no data demonstrating the ideal positioning of the implants in this plane. The flexion–extension axis of the ulna has been defined using a variety of methods, some based on anatomical landmarks and some using motion tracking technology; however, to our knowledge no CT-guided or patient-specific instrumentation has been used for placement of the ulna component more accurately,^18,19 although one study has demonstrated improved alignment of the humeral component.²⁰

This study demonstrated the positioning of the COR of the implant in comparison to the native COR. The COR of the humeral component was always anterior to that of the transepicondylar axis, which is expected by the design. The ulna component COR was placed posterior to the anatomical COR with a mixture of proximal and distal locations (Figure 4). The combination of these two findings indicates that the COR of the prosthesis is commonly placed anteriorly with attendant anterior translation of the ulna relative to the humerus. This could potentially shorten the triceps lever arm and therefore defunction triceps.^7,21

Misalignment of components of greater than 5° in the coronal or sagittal planes has been shown to be associated with higher failure rates and poor functional outcomes.^8,21 Four ulna components were placed in greater than 5° of flexion, two from each approach group. This increased flexed position has been shown to lead to higher rates of peri-prosthetic fracture due to stress riser formation.⁷ This demonstrates that accessing the ulna canal is challenging when using either approach. Passing a transolecranon wire to locate and cannulate the ulna canal has been advocated to aid in better positioning of the ulna implant.⁷ Alternatively, using the flat spot of the dorsal ulna to reference the position of the ulna component has been advocated as a method of avoiding this problem.²² Specific jigs to guide positioning of the components have been used in a series of 18 elbows; however, the mean deviation of the ulna component from the jig targeted angle was 8.6°.⁶ In fracture surgery, the proximal ulna dorsal angulation has been used to reference accurate reduction of the proximal ulna.²³ This measurement is similar in nature to the flat spot of the ulna described by Duggal et al.²² Our technique used the ulna shaft to reference the position of the ulna component and this measurement took place distal to the angulated portion of the proximal ulna, as the use of a complete forearm cadaver allowed us to do this. Whichever technique is used, it is clear that care must be taken to accurately site the ulna component along the long axis of the ulna. When excising the olecranon tip, exposure of the proximal ulna down to the insertion of the triceps tendon can help to maximise the amount of proximal ulna resected and therefore give better access to the ulna canal.

Aseptic loosening is significant problem in TEA.⁸ A review of TER revision surgeries revealed an association with anterior impingement and loosening,²⁴ with the hypothesised mechanism being pistoning of the ulna component. We were unable to measure the post-implantation range of motion data during our study, but the position of the COR post-operatively will clearly affect the degree to which the remaining structures can impinge in deep flexion. However, in clinical practice, we do assess the post-implantation range of motion to ensure no impingement.

Much of the work on alignment in TEA is based on radiographic analysis. There are no direct comparative cadaveric studies. We feel the direct measurement in the cadaveric situation is a distinct advantage over the previous radiographic assessment. Our technique for measuring alignment has not specifically been used in this setting previously. However, there is good evidence that goniometer measurements can replicate radiographic measurements of elbow range of motion^25,26 and clinical photography with a goniometer has also been validated for this purpose.²⁷

Many approaches have been described in order to perform TEA. For an approach to be adequate it needs to be safe, reproducible, achieve adequate exposure to perform the intended procedure, and be minimally destructive to anatomical structures.²⁸ The two approaches presented in this study demonstrate no difference in positioning of the implants. With the hypothetical benefit of preservation of the extensor mechanism and earlier mobilisation using the triceps-on approach is a valid option from these results. Clinical comparisons between approaches are still needed to define whether the perceived benefits are reproducible.

Author note

This article is not based upon any previous correspondence with the British Elbow and Shoulder Society.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.