Keywords
computed tomography - total hip replacement - hip - pelvis or acetabulum - dog
Introduction
Total hip replacement (THR) is a highly effective technique for the management of
hip dysplasia and other diseases of the coxofemoral joint in dogs.[1] However, several potentially challenging complications have been reported following
the procedure, including luxation, implant loosening, infection and femoral fracture.[2]
[3]
[4]
[5] Luxation has been described as the most common complication,[3]
[4]
[6]
[7]
[8] with a reported incidence of between 1.1 and 15.8%.[1]
[4]
[5]
[7]
A variety of patient-related and technical factors have been postulated as determinants
of the risk of luxation in dogs, including breed, conformation, implant sizing and
implant positioning.[4]
[6]
[9] Of these, the most attention has been paid to the role of acetabular cup positioning
(angles of lateral opening [ALO] and cup version).[4]
[6] Manufacturer's recommendations for acetabular positioning range from 35 to 45 degrees
for ALO[4] and 15 to 25 degrees for version angle.[10]
[11] However, these angles are not intended to be breed-specific. Given the variation
in pelvic size and conformation seen in dogs that are candidates for THR, it is unlikely
that population-based average values will be appropriate for all dogs. Additionally,
the published guidance does not take into consideration any pelvic pathology that
may affect acetabular conformation and orientation. Based on the current state of
knowledge, it is apparent that a more complete description of normal reference values
for acetabular orientation across different breeds and over a range of pathologies
could be helpful in better defining optimal implant position and reducing the risk
of postoperative complications following THR.
Two-dimensional measurements have been widely used to describe acetabular geometry.[12]
[13]
[14] However, plain radiographs do not account for pelvic rotation or tilting, and perfect
projections are necessary for accurate and repeatable measurements.[15] Measurements of acetabular morphometry utilizing computed tomography (CT) have also
been reported.[16] With CT becoming more accessible and accepted as a routine diagnostic modality in
veterinary medicine, there is potential to obtain more accurate and precise measurements.
Three-dimensional (3D), CT-derived bone models give detailed information on acetabular
morphometry and are relatively insensitive to variation in patient positioning during
the CT scan procedure.[17]
The specific aim of this study was to determine ALO and version angles of the native
canine acetabulum in a heterogeneous population of dogs to establish reference values
for dogs using 3D in-silico models derived from CT scans, and to validate this method by determining intra-observer
coefficient of variation. We hypothesized that measurements made on 3D models are
repeatable and that ALO and version angles are more variable than the range recommended
by implant manufacturers.
Materials and Methods
Study Population
This was a descriptive study of CT images from client-owned dogs. Dogs were included
if they were skeletally mature (as determined by closed growth plates), had a CT scan
of the entire pelvis and had no radiographic evidence of hip joint disease. The CT
scans of immature dogs or dogs with pelvic or hip pathology were excluded from the
study. Owners provided informed consent for the use of their dog's imaging data in
this study. Three-dimensional data (slice thickness <1 mm) were exported in Digital
Imaging and Communications in Medicine (DICOM) format to medical engineering software
(MIMICS version 24.0; Materialize, Leuven, Belgium) to build the in-silico pelvic models. The models were segmented on a bone algorithm, smoothed (2 cycles
at 0.4) and wrapped (smallest detail 1 mm, gap closing 0.5 mm) to minimize artifacts
from CT that could affect measurements. The pelvic models were exported further as
Standard Tessellation Language (STL) files to a mesh-based 3D measurement and design
software (3-Matic version 16.0; Materialise, Leuven, Belgium) for analysis.
Anatomical Measurements from CT Scans
For the purpose of measurement, anatomical pelvic landmarks and reference planes were
established. Four standardized landmarks were identified to define the alignment plane
of the pelvis – the cranial dorsal iliac spines on the left and right sides and the
ischial tuberosities on the left and right sides ([Fig. 1A]). The dorsal plane of the pelvis was defined by creating a datum plane that intersected
with three of the four landmark points ([Fig. 1B]). This plane ran along the ilioischial line and at right angles to the median plane.
The median plane of the pelvis was defined by creating a datum plane that bisected
the line between the two ischial landmarks ([Fig. 1C]). The third pelvic plane, the transverse plane, was defined by creating a datum
plane that intersected with the two ischial points and that was perpendicular to the
dorsal pelvic plane. This plane was set at right angles to both the median plane and
the dorsal pelvic plane ([Fig. 1D]).
Fig. 1 Anatomical landmarks and reference planes. (A) Four pelvic points were defined on
the left and right ilia, and the left and right ischia. These landmarks were then
used to define the three reference planes: dorsal pelvic plane (B), median plane (C)
and transverse plane (D).
The acetabulum was defined by marking triangles along the lunate surface of the acetabulum
([Fig. 2A]) and defining a best fit sphere ([Figs. 2B]). The centre of the acetabulum was identified by a point, representing the coordinates
of the centre of the best-fit sphere inside the acetabulum ([Figs. 2C]).
Fig. 2 Acetabular geometry was defined by marking triangles on the lunate surface of the
acetabulum (A), defining a best-fit sphere (B) and then calculating the centre of
the sphere (C).
The orientation of the ventral acetabular rim was defined by marking the triangles
that form the cranial and caudal rims of the ventral acetabulum ([Fig. 3A]). A plane – the ventral acetabular plane – was then defined by best fitting to these
highlighted triangles ([Fig. 3B]). The acetabular orientation plane was defined as a plane that was perpendicular
to both the ventral acetabular plane and the dorsal pelvic plane, and that passed
through the centre point of the acetabulum ([Fig. 3C]).
Fig. 3 Acetabular orientation was assessed by first marking triangles along the cranial
and caudal aspects of the ventral acetabulum (A) and then establishing a best-fit
plane to these voxels (B). The acetabular orientation plane was defined as a plane
that was perpendicular to both the ventral acetabular plane and the dorsal pelvic
plane, and that passed through the centre point of the acetabulum (C).
The version angle was measured as the angle formed between the acetabular orientation
plane and the transverse plane ([Fig. 4A]).
Fig. 4 The version angle was measured between the acetabular orientation plane and the transverse
plane (A). The angle of lateral opening was calculated by measuring the angle formed
between the best fit plane to the ventral acetabulum and the median plane (B), then
subtracting this from 90 degrees.
The ALO, the angle formed between the ventral acetabular plane and the median plane,
was measured in the transverse plane for the left and right acetabula ([Fig. 4B]).
Data Handling and Statistical Analysis
All data were collated and analysed using a commercial spreadsheet (Microsoft Excel
for Mac version 16.62; Microsoft Corporation, Seattle, Washington, United States).
The mean and standard deviation were calculated for ALO and version angles for each
hemipelvis. Left-right differences were evaluated using a paired t-test, with significance set at p less than 0.05, and with the symmetry index, according to the following formula:
Symmetry index = 100 * (Right-Left)/(0.5*(Right + Left))
For the determination of intra-observer repeatability, six hemipelves were each measured
three times and the coefficient of variation (%) calculated for both ALO and version.
For the determination of inter-observer reproducibility, six hemipelves were measured
independently by two investigators (MJA and ABF) and the coefficient of variation
(%) calculated.
Results
Twenty-seven dogs fulfilled the inclusion criteria. Breeds in this study included
Boerboel (n = 2), Leonberger, Staffordshire Bull Terrier, Lurcher, Greyhound (n = 3), Rhodesian Ridgeback (n = 2), Golden Retriever (n = 2), Great Dane, cross-breed (n = 3), Caucasian Shepherd dog, German Shepherd dog, Doberman, Bullmastiff, American
Bulldog, Pyrenean Mountain dog, Bernese Mountain dog, Weimaraner, Labrador Retriever
(n = 2) and Siberian Husky. There were 14 males (10 entire, 4 neutered) and 13 females
(5 entire and 8 neutered). The median age was 7 years, 5 months (range: 9 months to
12 years, 2 months) and the median body weight was 35.8 kg (range: 21–79 kg).
Complete data for ALO and version angles in the 27 pelves are presented in [Table 1]. The mean (±standard deviation) values for the ALO of left and right acetabula were
42.60 ± 4.15 degrees and 43.14 ± 3.92 degrees, respectively. Mean version angles for
the left and right acetabula were 27.51 ± 4.81 degrees and 26.85 ± 5.82 degrees. There
were no significant differences between left and right acetabula for ALO (p = 0.43) or version angle (p = 0.43) and the symmetry index was acceptable (6.8% for ALO, 11.1% for version angle).
The intra-observer coefficient of variation was 3.5% for ALO and 5.2% for version
angle ([Table 2]). The inter-observer coefficient of variation was 3.3% for ALO and 5.2% for version
angle.
Table 1
Angles of lateral opening (ALO) and version angles, means and standard deviation (SD)
for left and right hips of 27 dogs. p-Values for ALO and version angle are based on paired Student's t-test. Left-right symmetry is defined by symmetry index (see text for formula)
|
Dog
|
Angle of lateral opening
|
|
Version angle
|
|
|
Left (degree)
|
Right (degree)
|
Symmetry index (%)
|
Left (degree)
|
Right (degree)
|
Symmetry index (%)
|
|
1
|
39.9
|
44.7
|
11.3
|
28.1
|
25.6
|
9.3
|
|
2
|
42.1
|
45.5
|
7.8
|
26.1
|
24.7
|
5.5
|
|
3
|
34.9
|
32.7
|
6.5
|
38.7
|
39.9
|
3.1
|
|
4
|
46.0
|
45.2
|
1.8
|
26.8
|
25.8
|
3.8
|
|
5
|
43.6
|
47.1
|
7.7
|
26.3
|
19.8
|
28.2
|
|
6
|
45.3
|
41.1
|
9.7
|
28.2
|
29.5
|
4.5
|
|
7
|
49.7
|
51.5
|
3.6
|
22.6
|
18.4
|
20.5
|
|
8
|
41.9
|
43.0
|
2.6
|
25.8
|
28.8
|
11.0
|
|
9
|
36.2
|
41.1
|
12.7
|
35.8
|
31.8
|
11.8
|
|
10
|
43.5
|
43.2
|
0.7
|
23.8
|
24.9
|
4.5
|
|
11
|
44.5
|
43.1
|
3.2
|
29.5
|
12.6
|
80.3
|
|
12
|
48.6
|
47.6
|
2.1
|
20.7
|
20.1
|
2.9
|
|
13
|
48.1
|
47.5
|
1.3
|
23.4
|
24.1
|
2.9
|
|
14
|
41.7
|
36.4
|
13.6
|
29.4
|
34.8
|
16.8
|
|
15
|
47.3
|
43.6
|
8.1
|
25.1
|
25.8
|
2.8
|
|
16
|
44.6
|
40.6
|
9.4
|
28.5
|
33.1
|
14.9
|
|
17
|
42.4
|
44.6
|
5.1
|
30.0
|
29.8
|
0.7
|
|
18
|
42.2
|
42.2
|
0.0
|
23.5
|
24.7
|
5.0
|
|
19
|
39.8
|
42.9
|
7.5
|
26.6
|
26.6
|
0.0
|
|
20
|
47.4
|
42.8
|
10.2
|
28.1
|
28.8
|
2.5
|
|
21
|
36.4
|
37.9
|
4.0
|
32.5
|
33.7
|
3.6
|
|
22
|
43.4
|
47.2
|
8.4
|
29.3
|
24.5
|
17.8
|
|
23
|
37.9
|
45.4
|
18.0
|
15.0
|
19.5
|
26.1
|
|
24
|
40.8
|
42.1
|
3.1
|
29.5
|
28.4
|
3.8
|
|
25
|
38.7
|
37.1
|
4.2
|
31.5
|
32.0
|
1.6
|
|
26
|
36.2
|
43.5
|
18.3
|
34.1
|
31.6
|
7.6
|
|
27
|
47.3
|
45.4
|
4.1
|
23.7
|
25.4
|
6.9
|
|
Mean
|
42.6
|
43.1
|
6.8
|
27.5
|
26.9
|
11.1
|
|
SD
|
4.2
|
3.9
|
4.9
|
4.8
|
5.8
|
15.8
|
|
p-Value
|
0.43
|
|
0.44
|
|
Table 2
Intra-observer variability was calculated from 3 repeat measurements of 6 hemipelvis
specimens. All data are expressed in degrees
|
Trial 1
|
Trial 1
|
Trial 3
|
CV, %
|
|
ALO
|
|
|
|
|
|
Hemipelvis 1 (left)
|
37.62
|
39.99
|
39.01
|
3.06
|
|
Hemipelvis 2 (left)
|
43.82
|
40.31
|
45.26
|
5.90
|
|
Hemipelvis 3 (right)
|
45.62
|
45.78
|
44.03
|
2.14
|
|
Hemipelvis 4 (right)
|
42.13
|
44.3
|
43.54
|
2.54
|
|
Hemipelvis 5 (left)
|
46.88
|
47.95
|
45.72
|
2.38
|
|
Hemipelvis 6 (right)
|
47.93
|
44.01
|
47.03
|
4.43
|
|
|
|
Mean CV, %
|
3.48
|
|
Version angle
|
|
|
|
|
|
Hemipelvis 1 (left)
|
27.81
|
25.1
|
26.43
|
5.12
|
|
Hemipelvis 2 (left)
|
21.62
|
23.93
|
20.67
|
7.60
|
|
Hemipelvis 3 (right)
|
26.51
|
26.86
|
27.11
|
1.12
|
|
Hemipelvis 4 (right)
|
31.4
|
30.54
|
28.28
|
5.36
|
|
Hemipelvis 5 (left)
|
26.11
|
23
|
27.49
|
9.01
|
|
Hemipelvis 6 (right)
|
29.09
|
27.46
|
28.88
|
3.11
|
|
|
|
Mean CV, %
|
5.22
|
Abbreviations: ALO, angles of lateral opening; CV, coefficient of variation.
Discussion
The current study demonstrates that measurements of ALO and version angle using 3D
models based on CT data are repeatable and offer a practical approach to quantifying
the orientation of the acetabulum. The morphological data may be helpful in better
defining optimal acetabular cup orientation, which is crucial in preventing postoperative
luxation.[4] While the optimal cup position has been defined by BioMedtrix (Whippany, New Jersey,
United States) as an ALO of 45 degrees and version angle between 15 and 25 degrees,[10]
[11] this is a rather subjective assessment and in some dogs these angles may be imprecise
and contribute to hip luxation. Therefore, objective, patient-specific measurement
of native acetabular geometry may allow the surgeon to improve cup positioning and
reduce the overall risk for luxation.
In this study, the measurements of acetabular geometry on 3D models showed good repeatability
with a low intra-and inter-observer variability, and this allowed us to accept our
first hypothesis. Similar findings were noted by Leasure and colleagues[18] who confirmed the low variability in measurements of ALO and version angle when
CT images were used to measure acetabular cup position in dogs. Another human study,
by Park and colleagues, demonstrated that 3D measurements are reliable for evaluating
acetabular orientation and more consistent measurements were obtained using 3D bone
models.[19] Similarly, Sariali and colleagues reported that the use of CT scans for THR preoperative
planning results in greater accuracy than two-dimensional preoperative planning,[20] a finding that has since been also supported by results from other published studies.[21]
[22]
[23]
[24]
The results for ALO and version angle were similar to these obtained in a focused
study of 13 Labrador Retrievers by Wu and colleagues.[25] Additionally, measurements of left and right acetabula were not significantly different
in our study, which corroborates the findings from Wu's study. In the current study
the mean ALO was 42.6 degrees for the left acetabulum and 43.1 degrees for the right
acetabulum, compared with mean ALO of 40.5 degrees in the earlier publication. Our
mean version angles for the right and left acetabula were 27.51 and 26.85 degrees
respectively, which was similar to the 27.7 degrees reported by Wu and colleagues.[25] However, the recommended angles for the position of acetabular cup in commercial
THR system are slightly different – higher for ALO and lower for version angle.[10]
[26] Therefore, our second hypothesis was also supported. It has been reported that too
high an ALO increases the risk of hip luxation, so it is recommended to insert an
acetabular component at lower angle, since it may prevent luxation.[4] Some acetabula in our study, however, demonstrated more than 10 degrees difference
between the angles measured using this workflow (ALO and version angle) and those
recommended by manufacturers. These findings highlight the potential for angular mismatch
between the native acetabulum in dogs and the manufacturer's recommendations for acetabular
cup placement. This discrepancy needs to be considered when positioning the acetabular
component.
Different values between manufacturer's recommendation and those reported from the
current study may reflect our previous reliance on radiography rather than CT for
measurement of acetabular geometry and cup positioning. When using CT data and 3D
reconstructed pelvic models for measurement of acetabular cup position, pelvic rotation
and tilt are controlled by the operator,[27] while radiography does not account for the pelvic tilt and rotation. This may increase
variability between measurements.[28] For this reason, ideal positioning of the patient for radiographs is critical to
being able to obtain accurate angles and to avoid discrepancies in measurements.
Our study population consisted of 27 dogs of 18 different breeds, as compared with
the study of 13 dogs from a single breed (Labrador Retrievers) by Wu and colleagues.[25] The breed variability corresponds better with the real-life situation in which a
variety of pure- and cross-bred dogs are presented for THR. Although this larger and
more heterogenous sample of breeds improves the clinical relevance of the data, a
much larger study is needed to make definitive recommendations regarding the true
extent of variation in ALO and version angles in dogs. Notwithstanding the limitation
of sample size, this study demonstrates that although the mean values for acetabular
alignment were generally consistent with clinical THR guidelines, some dogs in this
study had more extreme values, and there was a wide range of angles across different
breeds. Using a standard set of recommended angles across all breeds of dogs may lead
to incorrect cup placement and an increased risk of postoperative complications such
as luxation.
The measurements reported in this study were based on the use of just four anatomical
landmarks – the cranial dorsal iliac spines and the ischial tuberosities, bilaterally.
We selected these specific landmarks because they are widely distributed across the
four corners of the ‘pelvic box’ and are palpable through the skin intraoperatively,
providing a realistic option for intraoperative surgical navigation. Similar observations
were made by Leasure and colleagues.[18]
Studies from human medicine suggest that there are some differences in hip morphometry
between ethnic groups.[29] In a veterinary setting, breed-related differences have also been described among
large-breed dogs. For example, St. Bernards and Bernese Mountain dogs have relatively
deep acetabula as compared with Labrador Retrievers and Boxers. In contrast, Labrador
Retrievers and Boxers had shallow and relatively open acetabula.[30] A similar comparison between two small-breed dogs, the Shih Tzu and the Maltese,
showed that the Shih Tzu acetabulum was deeper and wider than that of the Maltese.[31] Such variety in acetabular morphometry between breeds may have an impact on acetabular
measurements and surgical planning for THR, so further investigations are needed to
verify it.
In humans, differences have been demonstrated between male and female hip joints.
It has been reported that females have relatively greater acetabular depth, increased
acetabular version and smaller femoral heads,[32]
[33] while femoral offset is greater in males.[33] Interestingly, despite these anatomical differences, the same THR implant systems
are used successfully in both sexes.[34] Less is known about sex-related difference in acetabular geometry in dogs. In small-breed
dogs, sex was identified as a variable that impacted acetabular width and depth, but
acetabular index measurements were similar in the two sexes, suggesting that their
acetabula are shaped similarly.[31] Currently, it is unclear whether sex-related variation in canine acetabular morphometry
is sufficient to impact recommendations for optimum component positioning in THR.
The primary limitation of this study is that all the dogs included in this study had
normal hip joints without visible signs of pathology. Therefore, the results may vary
in dogs with dysplastic hip joints. We used visual (subjective) estimates of anatomical
landmarks, the identification of which will undoubtedly be subject to some intrinsic
error. Nevertheless, based on our results, the method of defining planes and angles
measurement appears feasible and repeatable. Further work is needed to compare the
outcome of acetabular component placement in dogs with normal hips and dogs with hip
disease. A much larger sample size will be needed to establish reliable reference
ranges and to allow for breed-to-breed comparisons of these measurements.
Conclusions
Measurements of the ALO and version angle on 3D in-silico models of the canine pelvis are feasible and repeatable. These data may be used to
better define the optimal placement of the acetabular component in THR surgery, leading
to a reduced risk of postoperative complications such as hip luxation. Patient-specific
morphometric data and the ability to obtain accurate and reproducible measurements
also establish the possibility of combining in-silico planning with intra-operative surgical navigation, further improving the surgeon's
ability to ensure correct placement of the acetabular components in dogs undergoing
THR surgery.
