Talonavicular Joint Coverage and Bone Morphology between Different Foot Types Philip K. Louie,1,2 Bruce J. Sangeorzan,1,3 Michael J. Fassbind,1 William R. Ledoux1,3,4 1 RR&D Center of Excellence for Limb Loss Prevention and Prosthetic Engineering, VA Puget Sound, Seattle, Washington 98108, 2School of Medicine, University of Washington, Seattle, Washington 98195, 3Department of Orthopaedic Surgery and Sports Medicine, University of Washington, Seattle, Washington 98195, 4Department of Mechanical Engineering, University of Washington, Seattle, Washington 98195
Received 8 March 2013; accepted 18 February 2014 Published online 9 April 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.22612
ABSTRACT: This study explored three dimensional (3D) talonavicular joint (TNJ) coverage/orientation and bone morphology to reveal parameters that could classify and identify predispositions to cavus and planus feet. 3D models of 65 feet from 40 subjects were generated from computed tomography images classified as pes cavus, neutrally aligned, or asymptomatic/symptomatic pes planus. We calculated the talar and navicular overlap (TNJ coverage). We also measured orientation of the navicular, morphological parameters of the talus and navicular, and angular position of the talar head to body. Pes cavus showed significantly less talonavicular coverage (58 2% talus and 86 2% navicular) compared to asymptomatic pes planus (63 2% and 95 2%) and neutrally aligned feet (98 2% navicular), and significantly more navicular dorsiflexion and adduction relative to the talus (p < 0.0083). The talar head in cavus feet was inverted relative to the body compared to planus feet (p < 0.0083). For symptomatic pes planus, significant abduction was measured for the navicular relative to the talus and the talar head was plantar flexed relative to the body (p < 0.0083). The talar head in planus feet was everted relative to the body compared to neutrally aligned feet. Both intrinsic (bone morphology) and extrinsic (bone position) differences exist in groups of feet described as cavus and planus. ß 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 32:958–966, 2014. Keywords: pes planus; pes cavus; talonavicular joint; talus; navicular
Human feet are highly variable in shape and are often characterized as cavus (high arch), neutrally aligned (normal arch), or planus (low arch). The differences in structure can result from factors such as bone shape, altered position of the bones, soft tissue injury or failure, and muscle imbalances or injuries.1,2 The association between foot shape and clinical disorders is poorly defined. In the general community, population studies showed the prevalence of pes cavus and pes planus to be 5–10% and 3–19%, respectively.3–5 Both pes cavus and pes planus are associated with longterm pain and lead to deficits in balance and mobility.1,6 Pes planus, characterized by a depressed longitudinal arch and hindfoot valgus and/or forefoot abduction,7 can occur from improper development,8 trauma,9 or degenerative processes of aging, hypertension, diabetes, and obesity that contribute to soft tissue rupture.10 Pes cavus is characterized by increased calcaneal pitch angle and varus, and excessive forefoot plantar flexion.6 This is often caused by a neuromuscular imbalance,11 but can also be subtle and without neuromuscular alteration. Cavus and planus feet can present asymptomatically or symptomatically over a wide range of severity and discomfort; some asymptomatic feet with high or low arches could be considered part of the continuum of “normal”––this is especially true for flat feet. As such, planus feet were divided into two groups:
Conflict of interest: None. Grant sponsor: University of Washington, School of Medicine, Medical School Research Training Program (MSRTP); Grant sponsor: VA RR&D Grant; Grant number: A4843C. Corresponding to: Bruce J. Sangeorzan (T: þ1-206-764-2991; F: þ1-206-764-2127; E-mail: [email protected]) # 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.
958
JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
asymptomatic feet that were flat, but pain free, and symptomatic flat arch feet that presented with pain, often in the medial malleolus and medial arch.1 Why symptoms are present or not is not understood. Perhaps the deformity is more severe in symptomatic feet, or they may just be subtly different in key mechanisms. Furthermore, of misalignments that can often result in symptomatic feet, the deformities of the talonavicular joint (TNJ) are present in children and adults suffering from symptoms of a foot arch deformity.12,13 Past studies showed that the TNJ is critical for deformity correction,7,13,14 and a recent study demonstrated altered talar and navicular bone morphology in non-weight bearing pes planus.15 Thus, the talonavicular relationship might be a crucial endpoint in the development of a pathologic foot type, and surgical reconstruction should be focused on returning this joint to normal. Importantly, little information exists in the literature to help the surgeon understand what a “normal” or “abnormal” talonavicular relationship should look like. However, we do know that both extrinsic (e.g., weight bearing subtalar subluxation in planus feet)16 and intrinsic (e.g., non-weight bearing independent talar and navicular morphology)15 parameters are associated with flat feet; thus, similar parameters could exist at the TNJ to differentiate between foot types. Therefore, we employed 3D bone shapes extracted from partially weight bearing computed tomography (CT) scans of cavus, neutrally aligned, and planus feet to observe differences in the TNJ coverage and orientation, and talar and navicular morphology between the foot groups to begin identifying levels of deformity that may predispose to a symptomatic foot. Our measures included joint coverage (i.e., the percentage of bone joint surface that overlaps with the other bone
TALONAVICULAR JOINT COVERAGE
joint surface); as such, we quantified talar and navicular coverage independently, but collectively referred to this as TNJ coverage. We also examined the position of the navicular relative to the talus. In addition, we analyzed variations in talar and navicular morphology. Lastly, the angular position of the talar head to the talar body was examined. We hypothesized that symptomatic pes planus and pes cavus (i.e., painful feet) have less talar and navicular coverage and significantly different navicular to talus orientation compared to neutrally aligned or asymptomatic pes planus (i.e., non-painful feet). We also hypothesized that the talar heads in the symptomatic pes planus group are more dorsiflexed (superior), everted, and abducted than all other foot types. On the other end of the spectrum, we expect the talar heads from the cavus group will be more plantar flexed (inferior), inverted, and adducted than other foot types. Lastly, we hypothesized that differences exist between the shape of the talus and navicular in low, high, and neutral arch feet.
METHODS Sixty-five feet (28 female, 37 male) from 40 individuals (age 47 10 years, BMI 27.3 5.3) were used for this IRBapproved study. The feet were categorized by arch height, hindfoot varus/valgus, and forefoot adduction/abduction as demonstrated on physical examination by an orthopedic surgeon and verified with medial/lateral and AP radiographic views.17 Subjects were grouped as pes cavus (10 subjects/14 feet), neutrally aligned (10/20), asymptomatic pes planus (10/18), or symptomatic pes planus (10/13). Using a custom designed two-piece acrylic frame, partial weight bearing (20% body weight) CT scans were collected. The subjects were supine with their feet at shoulder width while a foot plate was pulled towards their torso; 20% BW was tolerated by the subjects and led to noticeable changes in arch shape from the non-weight bearing condition. A GE Lightspeed Ultra Scanner (GE Healthcare Technologies, Waukesha, WI; 0.5-s helical scans, 1.25-mm slice thickness, 1.25 slice interval, 0.625:1 pitch) was used; scans were reformatted to produce 250 coronal view slices spaced 1 mm apart.17 The bones of the foot were segmented into 3D computer models using Multi-Rigid, a custom software package (Fig. 1).18 The TNJ surface on both the talus and navicular was selected from the CT data with Blender, an open source 3D graphics program. Since cartilage is invisible on CT, we
959
manually selected the relevant bone surfaces. In Blender, a partially transparent CT stack was overlaid with a rendering of the segmented talus and navicular. Working on one bone at a time, the joint surface was selected on the 3D rendering based on the detailed shape of the bone in the scan. The manual joint surface selection was achieved by rotating in the sagittal plane in 15˚ increments while viewing across the joint surface in multiple sagittal slices. One person performed all the analyses, using visual clues from the curvature of the bone and anatomic knowledge of the talus and navicular to select the joint surfaces. Each surface was selected at least three times in a blinded fashion, and the surfaces were carefully compared and adjusted for discrepancies. An orthopedic surgeon verified the joint surface selection. The result was a 3D model including the talus, talar joint surface, navicular, and navicular joint surface visibly delineated and anatomically scaled (Fig. 2). Several coordinate systems were objectively defined. For the TNJ surface, AP axes (from the anterior to posterior direction) were created for the talus and navicular. After the joint surfaces were selected, each surface was represented as a uniformly distributed point cloud by using each of the vertices of the triangles describing the surface. Next a sphere was fit to the talar and navicular point clouds. Then axes were defined from the center of the sphere to each point in the respective cloud. Finally, these lines were averaged to give a resulting axis (Fig. 3). The medial/lateral and inferior/superior axes for the talar and navicular TNJ surface were determined from fitting an ellipse to the joint surfaces collapsed along their respective AP axis. This was done by fitting a plane to the all the joint surface points, then projecting all the points back to the plane resulting in a roughly elliptical shape. The joint surface was now essentially 2D, and thus an ellipse could be fit to all the flattened joint surface points; a least squares algorithm was used to fit the data resulting in an ellipse that mostly circumscribed the raw data. The major and minor ellipse axes served as the medial/lateral and inferior/superior axis, respectively. To determine a coordinate system for the talar head, the talus was separated into a head and body using an objectively defined split plane (Fig. 4). Determining this plane was a two-step process. First, a rough separation of the head and body was performed by slicing the talus at 1-voxel intervals along the first principal axis of the entire talus. The area of each slice was calculated and plotted. The calculated minimum surface area was used to give the initial separation of the head and body. A new axis was defined from the centers of the initially separated head and body volumes; using this
Figure 1. (A) Partial weight bearing CT scan of a neutrally aligned foot, (B) simulated volumetric data, (C) bones of each foot segmented into 3D models: Red ¼ talus, blue ¼ navicular. Note that (A) and (B) are a different scale than (C). JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
960
LOUIE ET AL.
Figure 2. (A) Visualization of bones within simulated volumetric data, (B) talocrural joint surface and talonavicular joint (TNJ) surface on the talus selected from the CT data, gray ¼ talocrural joint surface (previously determined),19 green ¼ TNJ surface, and (C) posterior aspect of the navicular, gold ¼ TNJ surface. Note that (A) is a different scale than (B) and (C).
Figure 3. (A) Talonavicular joint (TNJ) coverage, transparent red ¼ talus, green ¼ talar TNJ surface, green arrow ¼ talar TNJ anterior/posterior axis, gold ¼ navicular TNJ surface, gold arrow ¼ navicular TNJ anterior/posterior axis, transparent blue ¼ navicular, (B) TNJ coverage with bones removed.
axis, the talar slice area minimization routine described above (i.e., looking for the plane with the minimized surface area) was repeated to create a more robust head and body separation. Once the separation was complete, the AP axis
Figure 4. Sample talus separated into a head and body using an objectively defined split plane, red ¼ talar body, black ¼ talar head, gray ¼ split plane. JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
defined from the talar joint surface was used to collapse all the talar head and, as above, an ellipse fit to the resulting 2D data generated the inferior/superior and medial/lateral axes (Fig. 5). The talar body coordinate system was obtained from a third articular surface, the talocrural joint, which was previously determined for the same feet19 using the methodology described above (Figs. 1 and 2). A quadric surface (hyperbolic or elliptic paraboloid) was fit to this joint surface to create a set of axes (Fig. 6). Lastly, fitting the entire navicular to an ellipsoid derived the navicular coordinate system (Fig. 5). In summary, coordinate systems were generated for the talar and navicular TNJ surfaces, for the talar head and body, and for the navicular. We used these five embedded systems to describe bone-to-bone position (navicular to talus), joint coverage (navicular and talus), and bone morphology (talar head to body). The axes were AP, inferior/superior, and medial/lateral, and the position or orientation was described as inversion/eversion, adduction/ abduction, and dorsiflexion/plantar flexion. To measure TNJ coverage, the navicular joint surface was moved along its AP axis towards the talar joint surface until they visually had the greatest intersection. Overlapped joint surfaces were delineated in the plane normal to navicular axis. Again, the best-fit joint surface selections were performed three times by a single individual, then reviewed and confirmed by a single orthopedic surgeon for accuracy. Percent joint coverage (covered joint surface/total joint surface) of the talar and navicular joint surfaces was calculated. Orientation of the navicular relative to the talus
TALONAVICULAR JOINT COVERAGE
961
Figure 5. Coordinate systems of the talar head and navicular shown on a neutrally aligned (NA) foot. Gray arrows are approximately in the coronal plane, green arrows are approximately in the transverse plane, and yellow arrows are approximately in the sagittal plane. (A) Sagittal, lateral view, (B) coronal, anterior view, and (C) transverse, superior view.
Figure 6. Coordinate systems of the talar body and talar head shown on a neutrally aligned (NA) foot. Gray arrows are approximately in the coronal plane, green arrows are approximately in the transverse plane, and yellow arrows are approximately in the sagittal plane. (A) Sagittal, lateral view, (B) coronal, anterior view, and (C) transverse, superior view.
was calculated using projection angles (transverse, coronal, and sagittal). Morphological measures included: Overall talar length, talar body height and width, talar head height and width, and navicular height and width (Fig. 7). For measurement of
the overall talar length and talar body width, the talus was rotated about the plane defined by the inferior/superior talar body axis until the AP talar body axis was positioned vertically. Overall talar length was defined as the perpendicular distance from the most anterior to the most posterior
Figure 7. Morphological measures of the talus and navicular, (A) overall talar length and talar body width, (B) talar body height, (C) talar head height and width, (D) navicular height and width. The boxes represent the planes used for measurement. JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
962
LOUIE ET AL.
Table 1. Age, Gender and BMI by Foot Type: Mean [SE] or Frequency [%]
Age BMI Female
Pes Cavus (n ¼ 10/14 )
Neutrally Aligned (n ¼ 10/20 )
Asymptomatic Pes Planus (n ¼ 10/18 )
Symptomatic Pes Planus (n ¼ 10/13 )
p
39.8 [3.2] 28.7 [1.9] 3 (30%)
48.9 [2.0] 24.7 [1.1] 5 (50%)
46.8 [2.5] 27.5 [1.9] 5 (50%)
53.9 [4.3] 28.4 [1.7] 4 (40%)
0.0181 0.33 0.89
Number of subjects/feet. Linear regression of age or BMI on foot type or chi square test for gender and foot type. 1Significant difference between PC and SPP (p < 0.05, Tukey’s test).
point, while talar body width was defined as perpendicular distance from the medial trochlear to the most lateral point of the talar lateral process. To measure talar body height, the talus was rotated to a lateral view until the medial and lateral trochlear arcs were aligned (normal to the medial/ lateral talar body axis, with the AP talar body axis positioned horizontally). The talar body height was the perpendicular distance from the most superior point of the talar dome to the most inferior point of the lateral process. To measure the talar head height and width, the talar head was rotated to a coronal plane view normal to the AP talar head axis, with the medial/lateral talar head axis positioned horizontally. The navicular height and width were measured by a comparable method, with the navicular also rotated to view at a similar coronal plane using the navicular axes. All the height and width measures were made using the perpendicular distances from the most superior and inferior points (for height) and the most medial and lateral points (for width). To account for potential differences due to foot size, overall talar length was normalized by foot length; all height and width measurements were presented as ratios. Foot length was measured by having each subject stand on an acrylic frame over a desktop scanner. Lastly, the talar head to body angle was calculated using the coordinate systems defined above (Fig. 6). Statistical analyses were conducted using R20 and lme4 package21 for the linear mixed effects regression. The regression was used to determine if mean talar/navicular measures (dependent variables) differed by foot type (independent fixed effect) with the study participant as the random effect to account for two feet from the same individuals. Hypothesis testing of ratio variables was conducted using the numerator as the dependent variable and the denominator as a fixed independent covariate. Significance was set at 0.05. If significant differences by foot type were found, then pairwise comparisons among foot type groups were carried out with significance set at 0.05 or 0.0083 using Bonferroni’s correction. All analyses were conducted with and without adjusting for age, gender, and BMI.
RESULTS BMI and gender were similar across all foot types, but cavus subjects were significantly younger than symp-
tomatic pes planus subjects (Table 1). When adjusting for age, gender, and BMI, the only difference in the results was in joint coverage (Table 2); therefore, only the unadjusted data are presented for all other results. An overall significant difference was found across foot types for both talar (Table 2, p ¼ 0.036/0.018 [unadjusted/adjusted]) and navicular (0.0004/0.0019) coverage. Talar coverage was significantly decreased in pes cavus (58 2%, mean SE) compared to asymptomatic pes planus (63 2%, p < 0.0083). Furthermore, the pes cavus (86 2%) had decreased navicular coverage compared to the neutrally aligned (98 2%, p < 0.0083) and asymptomatic pes planus (95 2%, p < 0.0083) feet. Thus, the navicular in the pes cavus had less contact with the talus. The differences among foot types were more pronounced when considering the orientation of the navicular to the talus, with an overall significant difference in the sagittal (p < 0.0001) and transverse (p < 0.0001) planes (Fig. 8 and Table 3). The navicular of pes cavus subjects were significantly more dorsiflexed (i.e., superior) and adducted (i.e., medial) than all other foot types (p < 0.0083). Navicular orientation in the neutrally aligned foot group was significantly more adducted (i.e., medial) compared to the symptomatic pes planus. No other significant post-hoc comparisons were found for navicular orientation. Overall significant differences existed for talar body height/width and foot length (Table 4, p ¼ 0.032 and 0.0036), but the only post-hoc difference was that cavus feet were shorter than other foot types (p < 0.0083). However, foot length, which was only measured to normalize talar length, was not a primary variable of interest. Lastly, the talar head to body angular relationships had overall significance about the medial/lateral axis (p ¼ 0.0012) and the AP axis (p < 0.0001) planes (Table 5). Symptomatic pes planus were significantly more plantar flexed (i.e., inferior) than all other foot
Table 2. Talus and Navicular Coverage (% [SE]) for the Different Foot Types
Talar coverage Navicular coverage
Pes Cavus (n ¼ 10/14 )
Neutrally Aligned (n ¼ 10/20 )
Asymptomatic Pes Planus (n ¼ 10/18 )
Symptomatic Pes Planus (n ¼ 10/13 )
p
58 [2] 86 [2]
62 [2] 98 [2]
63 [2] 95 [2]
56 [2] 92 [2]
0.036/0.018†,1 0.0004/0.0019†,1,2
Number of subjects/feet. Likelihood ratio test for the association between talus–navicular measure and foot type from linear mixed effects regression/adjusted for age, gender, and BMI.†Numerator adjusted for denominator. 1APP versus PC significant p < 0.0083. 2NA versus PC significant p < 0.0083.
JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
963
TALONAVICULAR JOINT COVERAGE
Figure 8. Navicular orientation relative to the talus at the talonavicular joint (TNJ), representations of each foot type are displayed in the transverse plane (superior view) and coronal plane (anterior view), red ¼ talus, green ¼ talar TNJ surface, gold ¼ navicular TNJ surface, blue ¼ navicular. (A) Pes cavus, (B) neutrally aligned, (C) asymptomatic pes planus, and (D) symptomatic pes planus. Severe feet were selected for cavus and planus examples.
types (p < 0.0083). Pes cavus were also significantly more inverted than all the asymptomatic and symptomatic pes planus groups (p < 0.0083). Neutrally aligned feet were also significantly inverted compared to the asymptomatic and symptomatic pes planus (p < 0.0083).
DISCUSSION Cavus and planus feet can present with a wide range of pain from asymptomatic to extreme discomfort, leading to deficits in balance and mobility.1,6,22 It is
not clear why symptoms are present or not in abnormal feet. Past studies showed the TNJ to be an important joint for achieving correction in deformed feet.7,13,14 The talus, navicular, and TNJ are also involved in supporting the body’s weight during movement. Flat feet, the more common foot deformity, have been the subject of multiple studies showing that position of the navicular around the talus is the most sensitive measure of deformity in painful flat feet.7 Some abnormalities of talar and navicular bone morphology in planus feet were also measured on non-
Table 3. The Orientation (˚ [SE]) of the Navicular Relative to the Talus
Dorsiflexion (þ)/plantar flexion () Inversion (þ)/eversion () Adduction (þ)/abduction ()
Pes Cavus (n ¼ 10/14 )
Neutrally Aligned (n ¼ 10/20 )
Asymptomatic Pes Planus (n ¼ 10/18 )
Symptomatic Pes Planus (n ¼ 10/13 )
p
8.2 [2.0] 15.5 [3.3] 25.3 [3.1]
4.6 [1.9] 4.3 [3.1] 9.7 [3.0]
2.4 [1.9] 4.6 [3.1] 14.0 [3.0]
5.3 [2.0] 4.7 [3.3] 22.2 [3.1]
Received 8 March 2013; accepted 18 February 2014 Published online 9 April 2014 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/jor.22612
ABSTRACT: This study explored three dimensional (3D) talonavicular joint (TNJ) coverage/orientation and bone morphology to reveal parameters that could classify and identify predispositions to cavus and planus feet. 3D models of 65 feet from 40 subjects were generated from computed tomography images classified as pes cavus, neutrally aligned, or asymptomatic/symptomatic pes planus. We calculated the talar and navicular overlap (TNJ coverage). We also measured orientation of the navicular, morphological parameters of the talus and navicular, and angular position of the talar head to body. Pes cavus showed significantly less talonavicular coverage (58 2% talus and 86 2% navicular) compared to asymptomatic pes planus (63 2% and 95 2%) and neutrally aligned feet (98 2% navicular), and significantly more navicular dorsiflexion and adduction relative to the talus (p < 0.0083). The talar head in cavus feet was inverted relative to the body compared to planus feet (p < 0.0083). For symptomatic pes planus, significant abduction was measured for the navicular relative to the talus and the talar head was plantar flexed relative to the body (p < 0.0083). The talar head in planus feet was everted relative to the body compared to neutrally aligned feet. Both intrinsic (bone morphology) and extrinsic (bone position) differences exist in groups of feet described as cavus and planus. ß 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res 32:958–966, 2014. Keywords: pes planus; pes cavus; talonavicular joint; talus; navicular
Human feet are highly variable in shape and are often characterized as cavus (high arch), neutrally aligned (normal arch), or planus (low arch). The differences in structure can result from factors such as bone shape, altered position of the bones, soft tissue injury or failure, and muscle imbalances or injuries.1,2 The association between foot shape and clinical disorders is poorly defined. In the general community, population studies showed the prevalence of pes cavus and pes planus to be 5–10% and 3–19%, respectively.3–5 Both pes cavus and pes planus are associated with longterm pain and lead to deficits in balance and mobility.1,6 Pes planus, characterized by a depressed longitudinal arch and hindfoot valgus and/or forefoot abduction,7 can occur from improper development,8 trauma,9 or degenerative processes of aging, hypertension, diabetes, and obesity that contribute to soft tissue rupture.10 Pes cavus is characterized by increased calcaneal pitch angle and varus, and excessive forefoot plantar flexion.6 This is often caused by a neuromuscular imbalance,11 but can also be subtle and without neuromuscular alteration. Cavus and planus feet can present asymptomatically or symptomatically over a wide range of severity and discomfort; some asymptomatic feet with high or low arches could be considered part of the continuum of “normal”––this is especially true for flat feet. As such, planus feet were divided into two groups:
Conflict of interest: None. Grant sponsor: University of Washington, School of Medicine, Medical School Research Training Program (MSRTP); Grant sponsor: VA RR&D Grant; Grant number: A4843C. Corresponding to: Bruce J. Sangeorzan (T: þ1-206-764-2991; F: þ1-206-764-2127; E-mail: [email protected]) # 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc.
958
JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
asymptomatic feet that were flat, but pain free, and symptomatic flat arch feet that presented with pain, often in the medial malleolus and medial arch.1 Why symptoms are present or not is not understood. Perhaps the deformity is more severe in symptomatic feet, or they may just be subtly different in key mechanisms. Furthermore, of misalignments that can often result in symptomatic feet, the deformities of the talonavicular joint (TNJ) are present in children and adults suffering from symptoms of a foot arch deformity.12,13 Past studies showed that the TNJ is critical for deformity correction,7,13,14 and a recent study demonstrated altered talar and navicular bone morphology in non-weight bearing pes planus.15 Thus, the talonavicular relationship might be a crucial endpoint in the development of a pathologic foot type, and surgical reconstruction should be focused on returning this joint to normal. Importantly, little information exists in the literature to help the surgeon understand what a “normal” or “abnormal” talonavicular relationship should look like. However, we do know that both extrinsic (e.g., weight bearing subtalar subluxation in planus feet)16 and intrinsic (e.g., non-weight bearing independent talar and navicular morphology)15 parameters are associated with flat feet; thus, similar parameters could exist at the TNJ to differentiate between foot types. Therefore, we employed 3D bone shapes extracted from partially weight bearing computed tomography (CT) scans of cavus, neutrally aligned, and planus feet to observe differences in the TNJ coverage and orientation, and talar and navicular morphology between the foot groups to begin identifying levels of deformity that may predispose to a symptomatic foot. Our measures included joint coverage (i.e., the percentage of bone joint surface that overlaps with the other bone
TALONAVICULAR JOINT COVERAGE
joint surface); as such, we quantified talar and navicular coverage independently, but collectively referred to this as TNJ coverage. We also examined the position of the navicular relative to the talus. In addition, we analyzed variations in talar and navicular morphology. Lastly, the angular position of the talar head to the talar body was examined. We hypothesized that symptomatic pes planus and pes cavus (i.e., painful feet) have less talar and navicular coverage and significantly different navicular to talus orientation compared to neutrally aligned or asymptomatic pes planus (i.e., non-painful feet). We also hypothesized that the talar heads in the symptomatic pes planus group are more dorsiflexed (superior), everted, and abducted than all other foot types. On the other end of the spectrum, we expect the talar heads from the cavus group will be more plantar flexed (inferior), inverted, and adducted than other foot types. Lastly, we hypothesized that differences exist between the shape of the talus and navicular in low, high, and neutral arch feet.
METHODS Sixty-five feet (28 female, 37 male) from 40 individuals (age 47 10 years, BMI 27.3 5.3) were used for this IRBapproved study. The feet were categorized by arch height, hindfoot varus/valgus, and forefoot adduction/abduction as demonstrated on physical examination by an orthopedic surgeon and verified with medial/lateral and AP radiographic views.17 Subjects were grouped as pes cavus (10 subjects/14 feet), neutrally aligned (10/20), asymptomatic pes planus (10/18), or symptomatic pes planus (10/13). Using a custom designed two-piece acrylic frame, partial weight bearing (20% body weight) CT scans were collected. The subjects were supine with their feet at shoulder width while a foot plate was pulled towards their torso; 20% BW was tolerated by the subjects and led to noticeable changes in arch shape from the non-weight bearing condition. A GE Lightspeed Ultra Scanner (GE Healthcare Technologies, Waukesha, WI; 0.5-s helical scans, 1.25-mm slice thickness, 1.25 slice interval, 0.625:1 pitch) was used; scans were reformatted to produce 250 coronal view slices spaced 1 mm apart.17 The bones of the foot were segmented into 3D computer models using Multi-Rigid, a custom software package (Fig. 1).18 The TNJ surface on both the talus and navicular was selected from the CT data with Blender, an open source 3D graphics program. Since cartilage is invisible on CT, we
959
manually selected the relevant bone surfaces. In Blender, a partially transparent CT stack was overlaid with a rendering of the segmented talus and navicular. Working on one bone at a time, the joint surface was selected on the 3D rendering based on the detailed shape of the bone in the scan. The manual joint surface selection was achieved by rotating in the sagittal plane in 15˚ increments while viewing across the joint surface in multiple sagittal slices. One person performed all the analyses, using visual clues from the curvature of the bone and anatomic knowledge of the talus and navicular to select the joint surfaces. Each surface was selected at least three times in a blinded fashion, and the surfaces were carefully compared and adjusted for discrepancies. An orthopedic surgeon verified the joint surface selection. The result was a 3D model including the talus, talar joint surface, navicular, and navicular joint surface visibly delineated and anatomically scaled (Fig. 2). Several coordinate systems were objectively defined. For the TNJ surface, AP axes (from the anterior to posterior direction) were created for the talus and navicular. After the joint surfaces were selected, each surface was represented as a uniformly distributed point cloud by using each of the vertices of the triangles describing the surface. Next a sphere was fit to the talar and navicular point clouds. Then axes were defined from the center of the sphere to each point in the respective cloud. Finally, these lines were averaged to give a resulting axis (Fig. 3). The medial/lateral and inferior/superior axes for the talar and navicular TNJ surface were determined from fitting an ellipse to the joint surfaces collapsed along their respective AP axis. This was done by fitting a plane to the all the joint surface points, then projecting all the points back to the plane resulting in a roughly elliptical shape. The joint surface was now essentially 2D, and thus an ellipse could be fit to all the flattened joint surface points; a least squares algorithm was used to fit the data resulting in an ellipse that mostly circumscribed the raw data. The major and minor ellipse axes served as the medial/lateral and inferior/superior axis, respectively. To determine a coordinate system for the talar head, the talus was separated into a head and body using an objectively defined split plane (Fig. 4). Determining this plane was a two-step process. First, a rough separation of the head and body was performed by slicing the talus at 1-voxel intervals along the first principal axis of the entire talus. The area of each slice was calculated and plotted. The calculated minimum surface area was used to give the initial separation of the head and body. A new axis was defined from the centers of the initially separated head and body volumes; using this
Figure 1. (A) Partial weight bearing CT scan of a neutrally aligned foot, (B) simulated volumetric data, (C) bones of each foot segmented into 3D models: Red ¼ talus, blue ¼ navicular. Note that (A) and (B) are a different scale than (C). JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
960
LOUIE ET AL.
Figure 2. (A) Visualization of bones within simulated volumetric data, (B) talocrural joint surface and talonavicular joint (TNJ) surface on the talus selected from the CT data, gray ¼ talocrural joint surface (previously determined),19 green ¼ TNJ surface, and (C) posterior aspect of the navicular, gold ¼ TNJ surface. Note that (A) is a different scale than (B) and (C).
Figure 3. (A) Talonavicular joint (TNJ) coverage, transparent red ¼ talus, green ¼ talar TNJ surface, green arrow ¼ talar TNJ anterior/posterior axis, gold ¼ navicular TNJ surface, gold arrow ¼ navicular TNJ anterior/posterior axis, transparent blue ¼ navicular, (B) TNJ coverage with bones removed.
axis, the talar slice area minimization routine described above (i.e., looking for the plane with the minimized surface area) was repeated to create a more robust head and body separation. Once the separation was complete, the AP axis
Figure 4. Sample talus separated into a head and body using an objectively defined split plane, red ¼ talar body, black ¼ talar head, gray ¼ split plane. JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
defined from the talar joint surface was used to collapse all the talar head and, as above, an ellipse fit to the resulting 2D data generated the inferior/superior and medial/lateral axes (Fig. 5). The talar body coordinate system was obtained from a third articular surface, the talocrural joint, which was previously determined for the same feet19 using the methodology described above (Figs. 1 and 2). A quadric surface (hyperbolic or elliptic paraboloid) was fit to this joint surface to create a set of axes (Fig. 6). Lastly, fitting the entire navicular to an ellipsoid derived the navicular coordinate system (Fig. 5). In summary, coordinate systems were generated for the talar and navicular TNJ surfaces, for the talar head and body, and for the navicular. We used these five embedded systems to describe bone-to-bone position (navicular to talus), joint coverage (navicular and talus), and bone morphology (talar head to body). The axes were AP, inferior/superior, and medial/lateral, and the position or orientation was described as inversion/eversion, adduction/ abduction, and dorsiflexion/plantar flexion. To measure TNJ coverage, the navicular joint surface was moved along its AP axis towards the talar joint surface until they visually had the greatest intersection. Overlapped joint surfaces were delineated in the plane normal to navicular axis. Again, the best-fit joint surface selections were performed three times by a single individual, then reviewed and confirmed by a single orthopedic surgeon for accuracy. Percent joint coverage (covered joint surface/total joint surface) of the talar and navicular joint surfaces was calculated. Orientation of the navicular relative to the talus
TALONAVICULAR JOINT COVERAGE
961
Figure 5. Coordinate systems of the talar head and navicular shown on a neutrally aligned (NA) foot. Gray arrows are approximately in the coronal plane, green arrows are approximately in the transverse plane, and yellow arrows are approximately in the sagittal plane. (A) Sagittal, lateral view, (B) coronal, anterior view, and (C) transverse, superior view.
Figure 6. Coordinate systems of the talar body and talar head shown on a neutrally aligned (NA) foot. Gray arrows are approximately in the coronal plane, green arrows are approximately in the transverse plane, and yellow arrows are approximately in the sagittal plane. (A) Sagittal, lateral view, (B) coronal, anterior view, and (C) transverse, superior view.
was calculated using projection angles (transverse, coronal, and sagittal). Morphological measures included: Overall talar length, talar body height and width, talar head height and width, and navicular height and width (Fig. 7). For measurement of
the overall talar length and talar body width, the talus was rotated about the plane defined by the inferior/superior talar body axis until the AP talar body axis was positioned vertically. Overall talar length was defined as the perpendicular distance from the most anterior to the most posterior
Figure 7. Morphological measures of the talus and navicular, (A) overall talar length and talar body width, (B) talar body height, (C) talar head height and width, (D) navicular height and width. The boxes represent the planes used for measurement. JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
962
LOUIE ET AL.
Table 1. Age, Gender and BMI by Foot Type: Mean [SE] or Frequency [%]
Age BMI Female
Pes Cavus (n ¼ 10/14 )
Neutrally Aligned (n ¼ 10/20 )
Asymptomatic Pes Planus (n ¼ 10/18 )
Symptomatic Pes Planus (n ¼ 10/13 )
p
39.8 [3.2] 28.7 [1.9] 3 (30%)
48.9 [2.0] 24.7 [1.1] 5 (50%)
46.8 [2.5] 27.5 [1.9] 5 (50%)
53.9 [4.3] 28.4 [1.7] 4 (40%)
0.0181 0.33 0.89
Number of subjects/feet. Linear regression of age or BMI on foot type or chi square test for gender and foot type. 1Significant difference between PC and SPP (p < 0.05, Tukey’s test).
point, while talar body width was defined as perpendicular distance from the medial trochlear to the most lateral point of the talar lateral process. To measure talar body height, the talus was rotated to a lateral view until the medial and lateral trochlear arcs were aligned (normal to the medial/ lateral talar body axis, with the AP talar body axis positioned horizontally). The talar body height was the perpendicular distance from the most superior point of the talar dome to the most inferior point of the lateral process. To measure the talar head height and width, the talar head was rotated to a coronal plane view normal to the AP talar head axis, with the medial/lateral talar head axis positioned horizontally. The navicular height and width were measured by a comparable method, with the navicular also rotated to view at a similar coronal plane using the navicular axes. All the height and width measures were made using the perpendicular distances from the most superior and inferior points (for height) and the most medial and lateral points (for width). To account for potential differences due to foot size, overall talar length was normalized by foot length; all height and width measurements were presented as ratios. Foot length was measured by having each subject stand on an acrylic frame over a desktop scanner. Lastly, the talar head to body angle was calculated using the coordinate systems defined above (Fig. 6). Statistical analyses were conducted using R20 and lme4 package21 for the linear mixed effects regression. The regression was used to determine if mean talar/navicular measures (dependent variables) differed by foot type (independent fixed effect) with the study participant as the random effect to account for two feet from the same individuals. Hypothesis testing of ratio variables was conducted using the numerator as the dependent variable and the denominator as a fixed independent covariate. Significance was set at 0.05. If significant differences by foot type were found, then pairwise comparisons among foot type groups were carried out with significance set at 0.05 or 0.0083 using Bonferroni’s correction. All analyses were conducted with and without adjusting for age, gender, and BMI.
RESULTS BMI and gender were similar across all foot types, but cavus subjects were significantly younger than symp-
tomatic pes planus subjects (Table 1). When adjusting for age, gender, and BMI, the only difference in the results was in joint coverage (Table 2); therefore, only the unadjusted data are presented for all other results. An overall significant difference was found across foot types for both talar (Table 2, p ¼ 0.036/0.018 [unadjusted/adjusted]) and navicular (0.0004/0.0019) coverage. Talar coverage was significantly decreased in pes cavus (58 2%, mean SE) compared to asymptomatic pes planus (63 2%, p < 0.0083). Furthermore, the pes cavus (86 2%) had decreased navicular coverage compared to the neutrally aligned (98 2%, p < 0.0083) and asymptomatic pes planus (95 2%, p < 0.0083) feet. Thus, the navicular in the pes cavus had less contact with the talus. The differences among foot types were more pronounced when considering the orientation of the navicular to the talus, with an overall significant difference in the sagittal (p < 0.0001) and transverse (p < 0.0001) planes (Fig. 8 and Table 3). The navicular of pes cavus subjects were significantly more dorsiflexed (i.e., superior) and adducted (i.e., medial) than all other foot types (p < 0.0083). Navicular orientation in the neutrally aligned foot group was significantly more adducted (i.e., medial) compared to the symptomatic pes planus. No other significant post-hoc comparisons were found for navicular orientation. Overall significant differences existed for talar body height/width and foot length (Table 4, p ¼ 0.032 and 0.0036), but the only post-hoc difference was that cavus feet were shorter than other foot types (p < 0.0083). However, foot length, which was only measured to normalize talar length, was not a primary variable of interest. Lastly, the talar head to body angular relationships had overall significance about the medial/lateral axis (p ¼ 0.0012) and the AP axis (p < 0.0001) planes (Table 5). Symptomatic pes planus were significantly more plantar flexed (i.e., inferior) than all other foot
Table 2. Talus and Navicular Coverage (% [SE]) for the Different Foot Types
Talar coverage Navicular coverage
Pes Cavus (n ¼ 10/14 )
Neutrally Aligned (n ¼ 10/20 )
Asymptomatic Pes Planus (n ¼ 10/18 )
Symptomatic Pes Planus (n ¼ 10/13 )
p
58 [2] 86 [2]
62 [2] 98 [2]
63 [2] 95 [2]
56 [2] 92 [2]
0.036/0.018†,1 0.0004/0.0019†,1,2
Number of subjects/feet. Likelihood ratio test for the association between talus–navicular measure and foot type from linear mixed effects regression/adjusted for age, gender, and BMI.†Numerator adjusted for denominator. 1APP versus PC significant p < 0.0083. 2NA versus PC significant p < 0.0083.
JOURNAL OF ORTHOPAEDIC RESEARCH JULY 2014
963
TALONAVICULAR JOINT COVERAGE
Figure 8. Navicular orientation relative to the talus at the talonavicular joint (TNJ), representations of each foot type are displayed in the transverse plane (superior view) and coronal plane (anterior view), red ¼ talus, green ¼ talar TNJ surface, gold ¼ navicular TNJ surface, blue ¼ navicular. (A) Pes cavus, (B) neutrally aligned, (C) asymptomatic pes planus, and (D) symptomatic pes planus. Severe feet were selected for cavus and planus examples.
types (p < 0.0083). Pes cavus were also significantly more inverted than all the asymptomatic and symptomatic pes planus groups (p < 0.0083). Neutrally aligned feet were also significantly inverted compared to the asymptomatic and symptomatic pes planus (p < 0.0083).
DISCUSSION Cavus and planus feet can present with a wide range of pain from asymptomatic to extreme discomfort, leading to deficits in balance and mobility.1,6,22 It is
not clear why symptoms are present or not in abnormal feet. Past studies showed the TNJ to be an important joint for achieving correction in deformed feet.7,13,14 The talus, navicular, and TNJ are also involved in supporting the body’s weight during movement. Flat feet, the more common foot deformity, have been the subject of multiple studies showing that position of the navicular around the talus is the most sensitive measure of deformity in painful flat feet.7 Some abnormalities of talar and navicular bone morphology in planus feet were also measured on non-
Table 3. The Orientation (˚ [SE]) of the Navicular Relative to the Talus
Dorsiflexion (þ)/plantar flexion () Inversion (þ)/eversion () Adduction (þ)/abduction ()
Pes Cavus (n ¼ 10/14 )
Neutrally Aligned (n ¼ 10/20 )
Asymptomatic Pes Planus (n ¼ 10/18 )
Symptomatic Pes Planus (n ¼ 10/13 )
p
8.2 [2.0] 15.5 [3.3] 25.3 [3.1]
4.6 [1.9] 4.3 [3.1] 9.7 [3.0]
2.4 [1.9] 4.6 [3.1] 14.0 [3.0]
5.3 [2.0] 4.7 [3.3] 22.2 [3.1]
