Abstract
Background The Tolat sigmoid notch classification is a commonly used classification to characterize the distal radioulnar joint (DRUJ). This classification was based on a limited assessment of the entire joint, which may lead to inaccuracies in sigmoid notch evaluation.
Questions/Purposes The purpose of this study is to assess the reliability of the Tolat classification for sigmoid notch characterization.
Methods The sigmoid notch of 52 models of cadaveric forearms was assessed by applying the Tolat classification to the three-dimensional (3D) modeled notch and then slices at the start of the notch (0 mm) and 4 mm more proximal. The inter- and intrarater agreement was assessed using Cohen's and Fleiss' kappa statistic.
Results Agreement between iterations regardless of slices or surgeons/radiologists was moderate. Intrarater agreement between pairs of slices (0 vs 4 mm, 0 mm vs 3D, 4 mm vs 3D) was moderate, whereas agreement between all slices was slight. Agreement between surgeons and between radiologists was moderate, while agreement across all raters and slices was fair. Models described as “other” were more consistent in 3D classifications and were commonly classified as a reverse ski slope.
Conclusions Classification using the Tolat scheme is fair to moderate at best. Classification of the sigmoid notch using an axial view of the distal radius may not accurately reflect the anatomy throughout the notch.
Clinical Relevance The Tolat classification supplies a limited analysis of the sigmoid notch, and does not represent a comprehensive evaluation of the entire joint. Future classification systems should characterize the entire sigmoid notch.
Keywords: sigmoid notch, Tolat, distal radioulnar joint
The distal radioulnar joint (DRUJ) is a complex joint that relies on a combination of soft tissue and muscle stabilizers, along with stability from the bony articulation. 1 Injuries that involve the DRUJ can lead to significant biomechanical sequelae. 2 3 This in turn can lead to detrimental outcomes in terms of function and quality of life, especially if the injury is localized to the dominant extremity. 2 3 Considering the importance of the DRUJ from a functional standpoint, knowledge and understanding of this joint would be of great importance to treat DRUJ issues.
The Tolat classification system characterizes the large degree of variation in the bony anatomy of the sigmoid notch. 4 Based on visual interpretation in the transverse plane, this classification system characterizes the sigmoid notch into one of four types: C-shaped (30%), S-shaped (14%), ski slope (14%), and flat face (42%). As this study bases the classification on gross visualization of the DRUJ on an axial view, it does not account for the variability that may prevail throughout the entire length of the joint.
A previous study investigated the three-dimensional (3D) anatomy of the sigmoid notch and determined conservation of symmetry between bilateral limbs but considerable variability of anatomy between limbs. 5 Assessing 3D models of the sigmoid notch allows the entire joint to be analyzed in order to accurately characterize the joint as a whole. Given this information, it is questionable whether the Tolat classification can accurately characterize the sigmoid notch, considering that it only accounts for a limited portion of the notch. The purpose of this study is to determine the reliability of the Tolat sigmoid notch classification by comparing it with classification using 3D models.
Materials and Methods
Computed tomography (CT) images of 52 forearms from the elbow joint to the carpus were used. Specimens with notable arthritic changes or previous bony injury were excluded in this study.
Using MIMICS Medical Imaging Software (Materialise, Ann Arbor, MI), 3D models of the entire radius were generated from the CT DICOM data. These models were created by thresholding the CT data to only include bone and exclude soft tissue, as has been done in previous studies. 6 7 Using the 3D models of the radius, landmark points were established on each bone to create a reproducible coordinate system.
The radius was analyzed in ParaView (Kitware, Clifton Park, NY) to obtain the appropriate sections for evaluation and quantification. Assessment of the sigmoid notch was performed by classifying the entire 3D-modelled sigmoid notch on an axial view and then axial slices at the start of the sigmoid notch (0 mm) and 4 mm more proximal ( Fig. 1 ). The assessment of the 3D model on the axial view is meant to re-create the type of assessment that was done by Tolat et al using cadavers 4 ( Fig. 2 ). These three views (3D, 0 mm, 4 mm) were classified according to the system as described by Tolat et al. 4 A total of 156 images were included and their classification was randomized such that each reviewer would look at each image in isolation and in a random order. The classification was performed by two fellowship-trained hand and upper extremity surgeons (Surgeons A and B), along with two radiologists (Radiologists A and B). This classification was repeated at least 2 weeks later in order to determine intrarater agreement. The inter- and intrarater agreement was assessed using Cohen's and Fleiss' kappa statistic. Sample size was calculated using previously derived formulas. 8 Based on this formula, a sample size of at least 52 cases was sufficient.
Fig. 1.
ParaView model of ( A ) axial cut, along with the level of cut as shown on the 3D model ( B ).
Fig. 2.
3D model of axial view of distal radius.
Results
The distribution of classifications in this study based on the original classification by Tolat et al 4 was as follows: flat face (24%), ski slope (15%), C-shaped (41%), and S-shaped (20%). The images that could not be classified according to Tolat's system constituted 12% of cases.
Intrarater Reliability between Iterations
Intrarater reliability between iterations for the 3D assessment and slices at 0 and 4 mm is described in Table 1 .
Table 1. Intrarater reliability between iterations.
| Surgeon A | Surgeon B | Radiologist A | Radiologist B | |
|---|---|---|---|---|
| 3D | 0.6 | 0.56 | 0.58 | 0.52 |
| 0-mm slice | 0.56 | 0.55 | 0.54 | 0.51 |
| 4-mm slice | 0.47 | 0.5 | 0.54 | 0.41 |
Intrarater Reliability between Pairs of Slices
Intrarater reliability between 0 mm versus 4 mm slices, 0 mm versus 3D slices, and 4 mm versus 3D slices for the first iteration is described in Table 2 .
Table 2. Intrarater reliability between pairs of slices.
| Surgeon A | Surgeon B | Radiologist A | Radiologist B | |
|---|---|---|---|---|
| 0 vs 4 mm | 0.57 | 0.58 | 0.62 | 0.53 |
| 0 mm vs 3D | 0.6 | 0.57 | 0.61 | 0.53 |
| 4 mm vs 3D | 0.62 | 0.55 | 0.59 | 0.51 |
Intrarater Reliability between All Slices
Intrarater reliability between 3D slices versus 0 mm versus 4 mm slices for the first iteration is described in Table 3 .
Table 3. Intrarater reliability between all slices.
| Surgeon A | Surgeon B | Radiologist A | Radiologist B | |
|---|---|---|---|---|
| 3D vs 0 mm vs 4 mm | 0.14 | 0.26 | 0.15 | 0.22 |
Reliability between Surgeons and Radiologists
Interrater reliability between Surgeon A and B, and between Radiologist A and B for the 3D assessment and slices at 0 and 4 mm is described in Table 4 .
Table 4. Reliability between surgeons and radiologists.
| Surgeon A vs B | Radiologist A vs B | |
|---|---|---|
| 3D | 0.61 | 0.58 |
| 0-mm slice | 0.58 | 0.57 |
| 4-mm slice | 0.52 | 0.48 |
Reliability across All Raters
Interrater reliability for the 3D assessment and slices at 0 and 4mm for all four raters was 0.35, 0.38, and 0.46, respectively.
Classification of “Other”
With respect to the 3D classifications, a total of 10 images were described as “other,” with agreement across iterations ( Fig. 3 ). Four images in the Surgeon group had agreement between raters in this category, compared to two in the Radiologist group. Of all these agreements, only one image had agreement across iterations and raters and was classified by all four raters as “other.” This image was described as a reverse ski slope. With respect to the 0-mm classifications, a total of nine images were described as “other,” with agreement across iterations. Two images in the Surgeon group had agreement between raters in this category. This image was described as a reverse ski slope. There were no agreements between raters in the Radiologist group for images in this classification. With respect to the 4-mm classifications, there were at total of seven images described as “other” with agreements across iterations. No images in the Surgeon or the Radiologist group had agreements between raters.
Fig. 3.
Axial view of slice through sigmoid notch that did not correspond to any classification as described by Tolat et al.
Discussion
The DRUJ is an important structure from a functional standpoint, as injuries to this joint can lead to significant detrimental outcomes if the anatomy is not restored. 2 3 Considering this point, it is important to accurately describe the anatomy of the DRUJ. The Tolat classification classifies the sigmoid notch based on gross visualization of the DRUJ on an axial view. 4 A drawback of this classification is that it does not account for the variability that may prevail throughout the entire length of the joint. In this study, we sought to assess the reliability of the Tolat classification for characterization of the sigmoid notch.
There were some limitations of this study. First, on one hand, there are many advantages to using 3D models for joint assessment. The use of 3D models allows for an accurate representation of anatomy and can highlight subtle deformities throughout the anatomical structure. On the other hand, the disadvantage of 3D bony model assessments include the inability to include articular cartilage in the assessment. The assumption in this case is that the bony anatomy of the sigmoid notch represents the same shape and contour as cartilage that covers the bone. A study by Johnson and coworkers 9 demonstrated the accuracy of using CT 3D reconstructions in air to characterize osseous and cartilaginous anatomy, showing mean errors consistently less than 0.5 mm. Therefore, it is likely that the 3D models used in this study accurately reflect the true anatomy of the articular surface of the joint. Second, there could also be the influence of a learning effect that could have had an influence on the results. We did not include a training period in this study, which may have made this study susceptible to this type of bias. Despite these limitations, rater agreements remained in the “moderate” category overall.
The main finding of the study was that there was fair agreement between all raters for the 3D model and slices at 0 mm, but more moderate agreement at a level 4 mm below the start of the sigmoid notch. There was no substantial agreement between any comparison of raters, slices, or iterations. There was only slight to fair agreement when comparing all slices within the same rater. There was only moderate agreement between surgeons and between radiologists, indicating that medical specialty does not necessarily favor more agreement using the current Tolat classification. At best, the agreement is moderate when comparing agreement between ratings at different levels around the sigmoid notch and a 3D axial view of the notch. Given this information, it is fair to say that anatomy as assessed on an axial view of a cadaver may not be consistent with the anatomy as assessed at 0 mm and slices more proximal ( Fig. 4 ). This stresses the variability of the anatomy of the sigmoid going more proximal, and portends less credence to the Tolat classification, as this only assesses the sigmoid notch based on an axial view of the sigmoid notch. Further classification systems should emphasize characterization of the entire joint rather than a single view. Given the utility of CT 3D reconstructions, this format can be used to develop a classification of this joint that could potentially be more useful for anatomic and surgical considerations in the future.
Fig. 4.
Serial axial cuts from the same 3D model indicating cuts at 0 mm ( A ) and 4 mm proximal ( B ). Note that the contour of the sigmoid notch is different between cuts.
Our data demonstrated a distribution of classifications similar to what was presented in the paper by Tolat et al. 4 Our study did note a higher proportion of “C-shaped” versus “flat face” as opposed to Tolat's paper, but overall these classifications both showed a higher proportion compared to the “S-shaped” and “ski slope” classifications, which is consistent with Tolat's paper. The differences in proportions can be explained by the lack of agreement seen when applying the Tolat classification, as shown in this study. A subset of the analysis could not be adequately characterized by the available Tolat classification and was described as “other.” It is likely that these slices were more adequately characterized by a variation of what was described in the original Tolat classification and leaves grounds for the expansion of the original system. It is unlikely that arthritis was a cause for variability in the classification as this was an exclusion criterion for cadaveric forearms. A potential expansion of the current classification would include a reverse ski slope pattern, as this was the most prevalent description of the notch when it could not be classified by the current system. A previous study has characterized the 3D anatomy of the sigmoid notch along with the variability of specific characteristics of the notch such as length, radius of curvature, and slope. 5 The current study has demonstrated that the sigmoid notch can be classified differently based on which level of the joint is visualized. As a result, future classifications should be based on 3D anatomy using CT, along with magnetic resonance imaging (MRI), to incorporate the shape of the cartilage in the sigmoid notch.
In conclusion, the lack of concordance between raters and assessments, and the number of models that could not be classified suggest that the Tolat classification does not characterize the variety of possible patterns that the sigmoid notch may fall under. This classification only supplies a limited analysis of the sigmoid notch, and does not represent a consistent, comprehensive evaluation of the entire joint. To our knowledge, this is the only study examining the reliability of this classification. Future classification systems should emphasize characterization of the entire notch, rather than rely on a single view. Given the lack of reliability of this classification, its use may misrepresent the anatomy of this complex joint and may lead radiographic and functional issues when repair or reconstruction procedures are planned. In addition, use of this classification, as it stands, may lead to inaccurate predictions regarding stability of this joint. The incorporation of 3D CT modelling and MRI may mitigate the limitations of characterizing the complex anatomy of the sigmoid notch. Understanding the shape of the sigmoid notch will help understand patterns of instability across the DRUJ, assist with arthroplasty design, and allow surgeons to understand potential risks of altering the joint, as in cases where ulnar shortening osteotomies are considered.
Conflict of Interest None declared.
This work was completed at the Department of Orthopaedics, University of British Columbia, Vancouver, BC.
References
- 1.Huang J I, Hanel D P. Anatomy and biomechanics of the distal radioulnar joint. Hand Clin. 2012;28(02):157–163. doi: 10.1016/j.hcl.2012.03.002. [DOI] [PubMed] [Google Scholar]
- 2.Kihara H, Palmer A K, Werner F W, Short W H, Fortino M D. The effect of dorsally angulated distal radius fractures on distal radioulnar joint congruency and forearm rotation. J Hand Surg Am. 1996;21(01):40–47. doi: 10.1016/S0363-5023(96)80152-3. [DOI] [PubMed] [Google Scholar]
- 3.Nishiwaki M, Welsh M F, Gammon B, Ferreira L M, Johnson J A, King G J. Effect of volarly angulated distal radius fractures on forearm rotation and distal radioulnar joint kinematics. J Hand Surg Am. 2015;40(11):2236–2242. doi: 10.1016/j.jhsa.2015.07.034. [DOI] [PubMed] [Google Scholar]
- 4.Tolat A R, Stanley J K, Trail I A. A cadaveric study of the anatomy and stability of the distal radioulnar joint in the coronal and transverse planes. J Hand Surg [Br] 1996;21(05):587–594. doi: 10.1016/s0266-7681(96)80136-7. [DOI] [PubMed] [Google Scholar]
- 5.Daneshvar P, Willing R, Pahuta M, Grewal R, King G J. Osseous anatomy of the distal radioulnar joint: an assessment using 3-dimensional modeling and clinical implications. J Hand Surg Am. 2016;41(11):1071–1079. doi: 10.1016/j.jhsa.2016.08.012. [DOI] [PubMed] [Google Scholar]
- 6.Katchky R N, King G J, Johnson J A, Athwal G S. The bicipital tuberosity and distal radius are unreliable landmarks for radial head implant alignment. J Shoulder Elbow Surg. 2013;22(09):1242–1247. doi: 10.1016/j.jse.2013.02.013. [DOI] [PubMed] [Google Scholar]
- 7.Sabo M T, Athwal G S, King G J. Landmarks for rotational alignment of the humeral component during elbow arthroplasty. J Bone Joint Surg Am. 2012;94(19):1794–1800. doi: 10.2106/JBJS.J.01740. [DOI] [PubMed] [Google Scholar]
- 8.Shoukri M, Asyali M, Donner A. Sample size requirements for the design of reliability study: review and new results. Stat Methods Med Res. 2004;13(04):251–271. [Google Scholar]
- 9.Lalone E A, Willing R T, Shannon H L, King G J, Johnson J A. Accuracy assessment of 3D bone reconstructions using CT: an intro comparison. Med Eng Phys. 2015;37(08):729–738. doi: 10.1016/j.medengphy.2015.04.010. [DOI] [PubMed] [Google Scholar]