Evaluating proximal ulnar morphology in relation to the humeral flexion–extension axis

Abstract

Background

The study aimed to evaluate the 3D relationship between the plane of the posterior surface of the olecranon, a newly defined plane through the ridge of the greater sigmoid notch (GSN), and the ulno-humeral flexion–extension axis (FE axis), respectively.

Methods

Twenty-four healthy left elbows were computed tomography scanned and 3D-segmented. First, a Cartesian ulnar coordinate system (UCS) was determined. Next, several anatomical landmarks were identified. The UCS and landmarks were assessed for repeatability and reproducibility. The orientation of the posterior surface and the plane through the GSN was evaluated relative to the anatomical FE axis.

Results

Both the UCS and the landmarks were considered repeatable and reproducible. In the axial plane, the mean angle between the posterior surface and the FE axis was 3° (standard deviation: ±5°, 95% confidence interval [CI]: [−6°; 12°]) external rotation. The mean axial angle between the GSN and the FE axis was 88° (standard deviation: ±3°, 95% CI: [83°; 94°]) external rotation.

Conclusion

This study concludes that the angulation between the posterior plane surface and the FE axis is highly variable (95% CI range: 18°). The plane through the ridge of the GSN of a healthy proximal ulna could provide a more reliable anatomical landmark to estimate the position of the elbow FE axis compared to the posterior surface (95% CI range: 11°).

Daily living activities can be severely affected by the loss of elbow function due to the destruction of the elbow joint.^8,13,14,20 In carefully selected cases, total elbow replacement (TER) has become an increasingly popular option.^13,14 Although elbow prosthesis designs have greatly improved in the last decades, the complication rate after arthroplasty ranges from 20% to 45%, much greater than in other large joints such as knee or hip arthroplasty.²¹ According to literature evaluating the TER outcome, the etiology of complications can be categorized into biomechanical and nonbiomechanical causes (eg, infections, ulnar nerve deficit, etc.).^12,14,18,21 Mechanical failure is primarily caused by implant malpositioning, which can lead to subluxation, dislocation, early polyethylene wear, or aseptic loosening.^3,15,21

Therefore, understanding the humeroulnar joint biomechanics is essential, and restoring the flexion–extension axis (FE axis) is often considered crucial to achieving minimal mechanical failures and good functional outcomes of TER.^3,15,21 Although the FE axis describes a screw displacement axis, the average FE axis is most commonly represented by a line connecting the center of the capitulum and the center of the trochlea (humeral landmarks).^1,2,4,6,7,9, 10, 11^,16

Currently, ulnar implant positioning is performed on bony landmarks. Clinical practice suggests that the ulnar reamer is inserted parallel to the posterior plane of the olecranon.^16,19 However, only 1 study has evaluated the FE axis based on ulnar anatomical landmarks.³ But to our knowledge, no study has yet thoroughly assessed the angulation between the posterior plane of the ulna and the FE axis.

Therefore, this study aimed to analyze the angulation of the posterior plane about the FE axis. In addition, the relation between the plane through the greater sigmoid and the FE axis was evaluated.

Method

A group of 24 healthy individuals were recruited. The set consisted of 20 men and 4 women with a mean age of 28. There was no record of previous elbow pathologies. All elbows were computed tomography scanned in full extension and supination. The DICOM files were exported and uploaded into Mimics Research 21.0 (Materialise NV, Leuven, Belgium). The files were manually 3D segmented and exported to 3-Matic 13.0 (Materialise NV, Leuven, Belgium) for further analysis.

Landmark identification

Five researchers were enlisted to identify 7 distinct bony landmarks on a sample of 10 3D-generated ulnas chosen randomly from the previously described set of 24 ulnas using 3-Matic. The landmarks comprised of the following:

• 1.One point placed on the ulnar styloid process
• 2.Two circles fitted to the greater sigmoid notch (GSN) of the ulna and the medial border of the humeral trochlea
• 3.Three spheres fitted to the 2 concave articular surfaces of the ulnar trochlea and the humeral capitulum
• 4.One plane fitted to the posterior flat surface of the proximal ulna (see Fig. 1).

Figure 1.

Landmarks. Dark green circle = medial humeral trochlea; red circle = greater sigmoid notch; green sphere = medial trochlea; pink sphere = lateral trochlea; blue sphere = capitulum; orange plane = posterior surface.

The points were chosen by 1-click identification. The circles were fitted using a curve on the circumference of the ridge to find the best fitting circle. The spheres were constructed using a best-fitting technique on an area of voxels marked by the researchers. The plane was also constructed using a best-fitting method. To evaluate intraindividual reliability, each of these landmarks was fitted 4 times on the same 10 ulnas by a single researcher, with a minimum interval of 3 days between fittings.

One of the researchers then fitted all necessary landmarks on 24 ulnas.

Cartesian coordinate system

The next step in this study was determining a Cartesian ulnar coordinate system (UCS). This coordinate system was established using the medial and lateral spheres fitted on the concave sides of the ulnar trochlea and the point placed on the apex of the ulnar styloid process.

The origin of the UCS was defined as the midpoint of the sphere–sphere intersection circle. The axes were oriented as follows:

• •X-axis: medial, perpendicular to the intersection circle.
• •Y-axis: anterior, perpendicular to the line connecting the origin and the ulnar styloid process.
• •Z-axis: cranial, perpendicular to both X and Y-axes (see Fig. 2).

Figure 2.

Proximal ulna with newly defined ulnar coordinate system based on sphere–sphere intersection circle fitted on both sides of the ulnar trochlea. Green = x-axis; blue = y-axis; orange = z-axis.

Reliability assessment

To assess the repeatability and reproducibility of the UCS’s, a single Cartesian reference coordinate system per ulna was used with a similar orientation of the principal axes and the origin located at the tip of the coronoid process. Each constructed UCS’s translational and rotational errors were calculated by measuring the total translational error (TTE) and total rotational error (TRE) for each UCS compared to the reference coordinate system. This was done for both the interobserver and intraobserver sets.

The repeatability and reproducibility of the landmarks were then evaluated by calculating the mean for each landmark and subsequently measuring the distance/radius or angle relative to their mean landmark. The angles were measured in an XYZ fashion, providing a single angle measurement instead of measuring the angles in the 3 major planes, which would yield 3 different angle measurements, making interpretation more complex and challenging. Subsequently, means and standard deviations (SD) were calculated for each different measurement for all ulnas combined, allowing for easily interpretable and comparable parameters to clinically assess the repeatability and reproducibility of all landmarks used in this study.

FE axis orientation analysis

For all 24 ulnas, the orientation of the posterior surface and the plane through the GSN was assessed relative to the ulno-humeral FE axis. The FE axis was defined as a line connecting the center of the capitulum and the humeral trochlea, as described previously, thus pointing medially.⁵ The angles between both planes and the FE axis were measured in all three major planes using the newly constructed UCS. To achieve this, the normal of each plane was projected onto the plane of interest, and the angle between the resulting vectors was measured and transformed using goniometric functions, depending on the plane's orientation. The chosen method provided precise and valid angle measurements. However, not all angles generated by this method offer practical information to clinicians, and therefore displaying all calculated angles on 3D models would be irrelevant. As a result, only the angle in the transverse plane of both planes compared to the FE axis and the angle in the coronal plane of the plane through the GSN and the FE axis were demonstrated in Figure 3.

Figure 3.

Angles calculated between the humeral flexion–extension axis on the one hand and the posterior surface or the plane through the greater sigmoid notch on the other hand in 2 anatomical planes (left: XY or transverse plane; right: XZ or coronal plane). Red = humeral rotation axis; green = posterior surface; orange/yellow = greater sigmoid notch plane.

Lastly, the angles between the X-axis of the UCS and the FE axis were calculated in all three major planes.

Statistical analysis

The intraclass correlation coefficient (ICC) estimates and their 95% confidence intervals (CIs) of the UCS and the landmarks were calculated in MatLab R2018a (MathWorks, Natick, MA, USA) based on a single rating, absolute-agreement, 2-way mixed-effects model (ICC[A,1]).¹⁷ The various angle measurements for the FE axis analysis were exported to SPSS 27.0 (IBM Corp., Armonk, NY, USA) and descriptive statistics were automatically generated.

Results

Reliability assessment

The mean TTE of the UCS for the interindividual and intraindividual group was 0.3 ± 0.2 mm and 0.2 ± 0.1 mm, respectively. The mean TRE of the UCS for both groups, respectively, was 0.9 ± 0.9° and 0.5 ± 0.4°. This resulted in an ICC greater than 0.995 for both the TTE and the TRE regarding constructing this newly defined UCS.

The variability of the identified landmarks showed excellent results. In the interobserver group, a mean deviation from the central tendency of less than 0.9 mm was found, while in the intraobserver group, an even smaller mean deviation of less than 0.3 mm was demonstrated. The posterior plane and the GSN displayed an average deviation of 1.6 ± 1.3° and 2.5 ± 1.4°, in the interobserver group and 1.5 ± 1.2° and 1.7 ± 1.2° in the intraobserver group, respectively. An overview of all descriptives can be found in Table I.

Table I.

Means and SD of the position of all landmarks compared to the respective mean landmark.

Landmark variability
Landmarks	Inter		Intra
	Mean	SD	Mean	SD
Posterior plane (°)	1.61	1.30	1.52	1.16
GSN (°)	2.47	1.34	1.72	1.22
Humeral rotation axis (°)	0.91	0.77	0.31	0.18
Midpoint GSN (mm)	0.37	0.19	0.30	0.19
Midpoint medial trochlea (mm)	0.42	0.39	0.24	0.14
Midpoint capitellum (mm)	0.31	0.18	0.29	0.21
Radius GSN (mm)	0.47	0.49	0.09	0.04
Radius medial trochlea (mm)	0.37	0.19	0.30	0.19
Radius capitulum (mm)	0.42	0.39	0.24	0.14

SD, standard deviation; GSN, greater sigmoid notch.

All angles are calculated in XYZ, in 3-dimensional space.

FE axis orientation analysis

The average angle between the posterior plane and the ulno-humeral FE axis in the axial plane was 2.9 ± 4.6°, 95% CI: [−6.1°;11.9°] (convention: angles greater than 0° indicate external rotation of the plane in relation to the FE axis). The angle between the plane through the GSN and the FE axis resulted in a mean angle of 88.3 ± 2.8°, 95% CI: [82.9°; 93.9°] (convention: angles smaller than 90° indicate external rotation of the plane with respect to the FE axis). Figure 4 visually represents the distribution of all axial plane values for the 2 planes.

Figure 4.

Boxplot demonstrating the variable dispersion of the angle between the posterior plane and the FE axis compared to the angle between the greater sigmoid notch plane and the FE axis. The latter’s results were displayed as the perpendicular equivalent values for interpretation purposes. FE, flexion–extension.

Additionally, the mean angle in the coronal plane between the GSN and the rotation axis was 87.7° ± 1.7°, 95% CI: [84.4°; 91.0°] (convention: angles smaller than 90° indicate valgus of the plane with respect to the FE axis). The angle between the posterior surface and the FE axis in the coronal plane was 10.9° ± 62.4°. In the sagittal plane, both angles resulted in an angle of respectively −45.9° ± 74.3° and 40.8° ± 34.9° for the GSN and the posterior surface. Finally, the mean angles in the axial and coronal plane between the x-axis and the FE axis were measured, resulting respectively in −1.8 ± 1.7° and 0.3 ± 2.3°.

Discussion

This study aimed to assess the 3-dimensional correlation between the plane of the olecranon’s posterior surface and the ulno-humeral FE axis. Moreover, the study aimed to compare this correlation with a newly defined plane passing through the ridge of the GSN.

Therefore, a repeatable and reproducible UCS must be designed. This study validated a coordinate system using 2 spheres fitted on the ulnar trochlea and the styloid process. All parameters were evaluated for repeatability and reproducibility, and the results indicate excellent performance, leading to the establishment of a highly reliable Cartesian coordinate system. Therefore, this newly developed UCS was an excellent coordinate system for this study and future research on the biomechanics of the humeroulnar joint.

Additionally, the variability of the landmarks used was minimal. The humeral rotation axis based on the sphere fitted on the capitulum and the medial border of the trochlea had a mean deviation less than 1.0° in both the coronal and transverse planes. The mean deviations of the posterior plane and the GSN plane were relatively greater, with mean deviations of 1.5° and 2.5°, respectively. According to Wiggers et al,^22,23 a malalignment of 5-10 degrees can lead to a 3-7 times increase in motion resistance. The calculated mean deviations are still well below these values, and therefore the variation of these landmarks should not influence the results of this study.

When comparing the posterior plane and the GSN plane in the axial plane, it becomes evident that the GSN plane has a narrower distribution across all ulnas with respect to the humeral FE axis. The mean deviation values between the posterior surface and the GSN are constrained to 3° and 2°, respectively. However, the 95% CI reveals a substantial range for these measurements, with the posterior surface showing a considerably larger range of 18°, while the GSN has a notably narrower range of 11°.

In the coronal plane, the mean angle between the GSN plane and the FE axis was 88°, with a SD of 1.7°. Conversely, the mean angle between the posterior plane and the FE axis in the coronal plane was 11°, with a much higher SD of 62°. This variability in the angle can be attributed to the 3-dimensional relationship between the posterior surface and the FE axis in this plane. Similarly, in the sagittal plane, the angles of both objects compared to the FE axis showed a SD higher than 60°.

One should keep in mind that the GSN’s orientation is perpendicular to the posterior plane, which allows practical assessment of the FE axis’ position in the coronal and transverse planes. This evidently leads to a more accurate estimation of the FE axis’ position compared to using the posterior surface.

This study had several limitations that need to be acknowledged. Firstly, the dataset was relatively small, consisting of only 24 ulnas. Additionally, all ulnas were scanned in supination. Likewise, the position of the axes in pronation could not be assessed. Furthermore, active muscle activation and gravity can also not be evaluated in a static computed tomography scan. This is particularly important given the findings of Duck et al,⁶ reported a 2° greater valgus and 1.5° more internal rotation of the axis in active compared to passive supination. Moreover, the study did not screen for handedness, and it would be informative to investigate whether there are any differences in ulnar morphology between dominant and nondominant sides. Another limitation is the definition of the ulno-humeral FE axis based on the medial ridge of the trochlea, as opposed to the commonly used trochlear groove. However, according to Deschrijver et al⁵ our method appears to be accurate. Finally, this study only evaluated the GSN plane in normal ulnas, and the influence of degenerative or inflammatory pathology on the plane was not assessed.

Conclusions

These results provide valuable insights into the 3-dimensional relationship between the ulna and the already well-established humeral FE axis. The findings of this study indicate that the angle between the posterior plane surface and the elbow FE axis exhibits considerable variability with a 95% CI range of 18°. An alternative option is to use the plane that passes through the ridge of the GSN of a healthy proximal ulna as a more dependable anatomical landmark to estimate the position of the elbow FE axis. Compared to the posterior surface, this plane demonstrates a smaller variability with a 95% CI range of 11°.

Disclaimers:

Funding: No funding was disclosed by the authors.

Conflicts of interest: The authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.