, Francois Lintz2, Cesar de Cesar Netto3, Alexej Barg4, Arne Burssens5 and Scott Ellis6
KeywordsMedializing calcaneal osteotomyComputed pre-postoperative assessmentWeight bearing CTAdult acquired flat foot deformity
A medializing calcaneal osteotomy (MCO) is a surgical procedure frequently performed to correct an adult acquired flat foot deformity (AAFD) [1–3]. This condition is characterized by a complex 3D deformity resulting in a loss of the medial longitudinal arch, valgus alignment of the hindfoot, and abduction of the midfoot [4–7]. The initial treatment in mild deformities is conservative (e.g., insoles), but, once progression of the deformity and impairment of functional activities occur, surgery is advised . In case a stage II AAFD, according to Johnson and Strom classification (containing a posterior tibial tendon dysfunction), does not respond to conservative treatment, a MCO is one of the surgical options available depending on the deformity characteristics . The main goal of the procedure is to correct hindfoot alignment (HA) . However, HA is markedly different among patients with stage II AAFD, and thus understanding how a MCO influences hindfoot alignment is essential. Chan et al.  have demonstrated a highly positive correlation between the change in the hindfoot moment arm and the amount of millimeters translated during a MCO. Although these findings have improved clinical practices, these measurements are still only performed on 2D radiographs and thus can be improved upon. Because 2D radiographs are a projection of a 3D deformity correction, 2D radiographs are prone to rotational errors and manual measurements. Laquinto et al.  addressed the limitations of 2D imaging but only for a 3D-simulated MCO model. Expanding upon this concept, Kido et al.  and subsequently Zhang et al.  utilized a 3D model in a clinical setting on patients with stage II AFFD. However, surgical corrections could not be assessed, and a CT scan was used to simulate weight bearing; the latter limitation can now be overcome by the use of a weight-bearing cone beam CT (WBCT) in foot and ankle pathologies . This novel device allows for standing position images at a high resolution and at a relatively low radiation dose [14–17]. The application of a WBCT combined with currently available computed measurement techniques would overcome the aforementioned shortcomings and be for the first time applied in assessing a surgical hindfoot correction in 3D.
Our aim is to analyze the pre- and postoperative hind- and midfoot alignment after a MCO using a WBCT and 3D computed measurement techniques. We hypothesize that there is a linear relationship between the amount of medial translation during a MCO and the correction of both the hind- and midfoot alignment.
Material and Methods
Study Population, Design, and Measurement Protocol
Patient demographics and concomitant procedures
Age (±) SD
41.8. ± 17.3 years
TMT 1 fusion
Spring ligament repair
Deltoid ligament reefing
A prospective pre-post study design was used: pre- and postoperative weight bearing CT scans were collected before surgery and 12 weeks after surgery. The local Institutional Review Board approved the study (EC15/49/537), and all patients gave informed consent.
A PedCAT® weight bearing cone beam CT was used (CurveBeam, Warrington, PA, USA) containing the following imaging protocol and settings: tube voltage, 96 kV; tube current, 7.5 mAs; CTDIvol 4.3, mGy; matrix, 160,160,130; pixel size, 0.4 mm; and slice interval, 0.4 mm. At the department of radiology, patients were asked to assume a natural stance with both feet parallel to each other at shoulder width apart.
In order to perform a 3D analysis, it is required to segment the CT slices based on their outer cortical surfaces. This was applied semiautomatic using the automatic bone segmentation tool in Mimics® 20.0 software (Materialise, Haasrode, Belgium) by manually appointing the cortical contour before Standard Triangulation Language (STL) files could be acquired. These were exported in 3-matic® software (Materialise, Haasrode, Belgium) to compute 3D goniometrical relationships.
A Cartesian coordinate system was acquired: the centroid of the talus was defined as the origin after being projected on the segmented base plate of the WBCT, to represent the ground floor . The z-axis was defined as running through the origin and perpendicular to the segmented base plate. The x-axis runs through the origin and the projected centroid of the head of the second metatarsal, which simulates the longitudinal axis of the foot. The Y-axis was defined as the cross product of the x– and z-axis. The coronal plane was defined as the YZ-plane, the sagittal plane as the XZ-plane, and the axial plane as the XY-plane (Fig 11.4a–c).
This coordinate system was incorporated in a custom build script, using Matlab® 2016b (The MathWorks, Inc., MA, USA), and aligned pre- and postoperatively, in order to reference the computed angles and distances.
In general, each 3D measurement was determined pre- as well as postoperatively and subsequently projected in the reference plane currently used in clinical practice, to allow comparison and to optimize interpretation of the results.
The landmarks and axes necessary to calculate the following parameters were automatically computed by using the three main functions called “Create point: center of gravity,” “Extrema analysis,” and “Create line: fit inertia axis” function of the 3-matic® software. These are based on goniometric functions to calculate respectively the centroid of a generated volume, the most outer point of a structure in the direction of a given axis and best fit centroidal axis on a 3D model.
The difference in the axial plane between the pre- and postoperative (TRx) allowed to determine the rotation of the tibia. Positive values equaled an external rotation and negative values an internal tibia rotation. The translations of the MCO were determined by reconstruction of the osteotomy plane (Fig. 11.2a). This allowed to divide the calcaneus in an anterior and posterior part (Fig. 11.2b). Based on the anterior part, the pre- and postoperative calcaneus could be matched on top of each other. The 3D distance between the computed centroid of the pre- and postoperative part of the posterior calcaneus was used to resemble the translation obtained after the MCO (Fig. 11.2c).
An additional deviation analysis was performed using CloudCompare® v 2.0 open source software (CloudCompare, Paris, France) by selecting the preoperative 3D model as a “mesh.” This was automatically used as the reference and the postoperative model as a “cloud.” The latter represented all vertices that form the 3D postoperative model. The distance between these vertices and the reference was determined by the CloudCompare®-software, and this analysis depicted the range of medial translation obtained after a MCO (Fig. 11.2d).
Measurements in the midfoot consisted of the navicular height (NH), navicular rotation (NR), and Méary angle (MA) .
The calcaneus was exposed through a lateral approach. The osteotomy was initiated with an oscillating saw blade 90° to the lateral calcaneal wall and inclined 45° in the sagittal plane according to Myerson et al. . A broad osteotome was used to complete the medial cortex. The amount of medial translation of the calcaneum was determined by the surgeon (TL), based on intraoperative assessment with a neutral heel according to the longitudinal tibia axis. No additional rotation of impaction was performed. The osteotomy was fixated with either a 5 mm, 7.5 mm, or 10 mm calcaneus Step Plate® (Arthrex, Naples, FS, USA) with locking screws or was fixated by the use of two 7 mm cannulated lag screws (Wright Medical, Memphis, TN, USA). Postoperatively, patients were treated consistently with 6–8 weeks of non-weight bearing in a removable boot. This was followed by progression to full weight bearing between weeks 8 and 10, depending on healing.
A Kolmogorov-Smirnov normal distribution test was performed for the hindfoot angle, rotation of the tibia, navicular height and rotation, and Méary angle. These demonstrated to be P > 0.05, indicating a normal distribution of the data and the use of further parametric testing. A paired Student’s t-test was conducted to compare the means of the preoperative versus the postoperative measurements of both the hindfoot (HA, TAx, TCx, and TRx) and midfoot alignment (NH, NR, and MA).
The correlation between the measured hindfoot/midfoot alignment and amount of medial translation after MCO was assessed by the Pearson coefficient (R). Linear regression analysis was demonstrated by the use of a corresponding scatter plot and calculation of the R2, when significant.
Inter- and intraobserver variability of the obtained midfoot measurements were analyzed using the interclass correlation coefficient ; the reliability of the hindfoot measurements was reported previously to be excellent .
Interpretations were as follows: ICC < 0.4, poor; 0.4 < ICC < 0.59, acceptable; 0.6 < ICC < 0.74, good; and ICC > 0.74, excellent .
The SPSS (release 22.0.0. standard version, SPSS, Inc., Chicago, IL, USA) statistical package was used to analyze the results. A probability level of p < 0.05 was considered significant.
An a priori statistical power analysis was performed using G∗Power (version 18.104.22.168; Dusseldorf University, Dusseldorf, Germany) . Previously reported data regarding regression analysis of the hind- and midfoot alignment were used for sample size estimation . Calculations have shown that a total sample size of N = 4 is needed for regression analysis of the hindfoot alignment and N = 47 for the midfoot alignment, to reach the respectively calculated effect size of (f = 0.96) and (f = 0.33) with a power level of 0.80 and a level of significance set at 0.05.