Cortical and medullary morphology of the tibia

Abstract

Bone resorption caused by stress shielding and insufficient bone-implant contact continues to be problematic for orthopaedic endoprostheses that utilize osseointegration (OI) for skeletal fixation. Morphologic analyses have helped combat this issue by defining anatomic parameters to optimize endoprosthesis loading by maximizing bone-implant contact. These studies have not typically included diaphyseal medullary morphology, as this region is not pertinent to total joint replacement. To the contrary, percutaneous OI endoprostheses for prosthetic limb attachment are placed in the diaphysis of the long bone. This study examined the cortical and medullary morphology of 116 fresh-frozen human cadaveric tibia using computed tomography (CT). Anatomic landmarks were selected and custom Matlab scripts were used to analyze the cross-sectional cortical and medullary morphology normalized to biomechanical length (BML). BML measured the distance between the tibial plateau and the tibial plafond. Properties such as cortical thickness, medullary diameter, and circularity of the medullary canal were quantified. We tested the influence of sex and laterality on morphology, and examined variations along the length of the bone. Results showed that while both sex and laterality impacted the location of anatomic landmarks, only sex influenced cross-sectional morphology. Overall, morphology significantly affected shape along the length of the bone for all examined properties except medullary circularity. This analysis found that distal to 35% BML, the canal is conducive to a circular implant, with medullary diameter ranging from 13–32 mm between 20–80% BML. A large size range is necessary for sufficient implant contact in order to accommodate residual limb length after amputation.

Background

Bone remodeling around orthopaedic endoprostheses can either facilitate stability, or lead to loosening and failure. Osseointegration (OI) leverages positive bone remodeling through ingrowth into porous metals, promoting attachment of orthopaedic devices to the host bone. This concept was first introduced for dental implants (Laney et al., 1986), and has been successfully utilized in orthopaedic devices for cementless joint replacement. However, negative bone remodeling effects including bone resorption and cortical thinning from stress shielding and poor fit can cause aseptic loosening and lead to implant revision surgery (Sundfeldt et al., 2006; Nebergall et al., 2012). To combat these effects, implant designs have been modified to better distribute loads to the bone-implant interface (Bobyn et al., 1992; Lonner et al., 2001). To achieve positive bone remodeling and stabilizing bone ingrowth, endoprostheses must be designed specifically for the bone they are intended to fit.

The tibia is a common anatomic location for OI due to total knee and ankle arthroplasty. To inform and optimize the designs of these devices, several studies have examined proximal and distal morphology of the tibia (Yoshioka et al., 1989; Stiehl and Abbott, 1995; Stagni et al., 2005; Farfalli et al., 2009; Subburaj et al., 2009; Nakamura et al., 2015; Madadi et al., 2016). However, these studies did not examine the diaphyseal or medullary morphology of the bone because the devices were designed to replace the epi- and metaphyseal regions of the bone, which can achieve OI without primary fixation in the medullary canal.

Diaphyseal and medullary morphology is becoming important with the introduction of percutaneous OI for prosthetic limb attachment. Percutaneous OI has emerged as an alternative to standard socket fixation of prosthetic limbs, especially for those who are unable to use a socket suspension prosthesis (Hagberg and Branemark, 2009; Branemark et al., 2014; Al Muderis et al., 2017). These endoprostheses use a porous coated intramedullary stem that permanently passes through the skin to connect percutaneously to an external prosthesis.

Most morphologic research on percutaneous OI system design has been conducted for transfemoral devices (Nebergall et al., 2012; Webster et al., 2017) and transhumeral (Drew et al., 2019), ignoring the most frequent major extremity amputation: transtibial (Ziegler-Graham et al., 2008; Johannesson et al., 2009). Current tibial devices were originally designed for use in other anatomic locations. Though there are reported positive outcomes (Atallah et al., 2017), it is unclear whether translating other endoprosthetic designs to the tibia is the best solution for implant stability and long term success. The only study to date that has examined intramedullary morphology of the tibia quantified bowing, canal diameter, and medullary orientation at the narrowest point of the medullary canal (Tosun et al., 2003). Endoprosthesis placement depends on residual limb length, not simply the narrowest point of the medullary canal. Therefore, a comprehensive analysis is needed along the length of the bone in order to design for maximal bone-implant contact and to lessen the risk for stress shielding or fracture due to stress concentrations. This concept was highlighted by an animal model of percutaneous OI amputation which found bone resorption occurred when the distal osteotomy was not loaded by an end loading collar, resulting in zones of stress shielding (Jeyapalina et al., 2019). A similar phenomenon occurred when the proximal stem was tightly bound within the medullary canal, creating a stress concentration.

The present study provides an analysis of the cortical and medullary features along the length of the tibia, useful for design of tibial endoprostheses for percutaneous OI attachment. This was accomplished by measuring anatomic landmarks and cross-sectional morphology on 3D computer tomography (CT) models of cadaver tibia, then evaluating the effect of sex and laterality (right versus left side) on morphology. We hypothesized that sex would affect tibial morphology, while laterality would not.

Materials and Methods

This study was exempt per both University of Utah and Salt Lake City Veterans Affairs Medical Center IRB (#11755). Fifty eight pairs of fresh-frozen human cadaver tibia were obtained, consisting of 35 male donors and 23 female donors, age 16–71 (41±19 yrs) and 15–87 (53±21 yrs), respectively (Table 1). Three donors were African American, two Hispanic, one Asian, and 21 Caucasian. Ethnicity was not available for 31 pairs.

Table 1:

Cadaver specimen demographics

MEAN±SD	Age (yrs)	Height (m)	Weight (kgs)
Male (N=35)	41±19^	1.78±0.08^	88.0±19.1
Female (N=23)	53±21^	1.66±0.08^	79.5±24.2
All (N=58)	46±21	1.74±0.10	84.7±21.5

^{^}indicates significant differences between male and female populations.

CT Reconstruction

Axial CT scans were acquired using a Siemens SOMATOM Definition Flash scanner (120kVp, 100mAs, 512 x 512 acquisition matrix, 1-mm slice thickness). CT images were segmented via semi-automatic thresholding to identify cortical and medullary boundaries (MIMICS v21.0, Materialise, Leuven, Belgium). Three dimensional (3D) reconstructions of this segmentation were then decimated and smoothed to reduce aliasing artifacts.

Anatomic Coordinate System

Anatomic landmark identification and alignment to a common coordinate system were performed (3-Matic Research v13.0, Materialise, Leuven, Belgium). An anatomic coordinate system was adapted from the literature (Yoshioka et al., 1989; Fitzpatrick et al., 2007) utilizing only proximal tibial landmarks to allow its use in amputee populations (Figure 1). Biomechanical length (BML) was defined as the distance between the average heights of the proximal (tibial plateau, origin) and distal (tibial plafond) articular surfaces. This represents the functional biomechanical distance between the proximal and distal joints of the bone similar to how it was defined in the humerus (Drew et al., 2019). The Tibial Shaft Axis (TSA) was defined using the inertial axis of the medullary canal starting from one medial-lateral proximal plateau width down the shaft from the origin, terminating at 50% BML. This was done to exclude the cancellous-rich epiphyseal region. The 50% termination was selected since the shortest possible length was desired for translation of this anatomic coordinate system to amputees. Left bones were mirrored to rights for consistent axis definitions.

Figure 1: Anatomic coordinate system.

Origin height was defined as the average height of the six landmarks defining the borders of the medial and lateral tibial condyles. The Z-Axis defined the tibial shaft axis (TSA) and was determined from the inertial axis of the medullary canal starting from one medial-lateral proximal plateau width down the shaft from the origin, terminating at 50% biomechanical length (BML). The superficial center point of the tibial tuberosity projected from the TSA defined the anterior direction (Y-Axis), and the medial-lateral direction was defined by the cross-product between the TSA and anterior-posterior axis.

Anatomic Landmarks

Anatomic landmarks were used to measure medial, lateral, and total plateau width, height, and tilt (λ_m, λ_l, λ_t), margin varus angle (θ_M), and tuberosity lateral deviation (ν) (Figure 2A-C) (Yoshioka et al., 1989). Retroversion (θ_R) was defined as the angle between the medial-lateral axis and the line connecting the distal fibular notches with the center point of the medial malleolus (Figure 2D) (Madadi et al., 2016). This is the only approach that does not use the fibula to calculate retroversion angle of the tibia, which is applicable because the cadaver donors used did not have a fibula. Medial-lateral and anterior-posterior ankle offset (AO_ML and AO_AP) was calculated as the transverse medial-lateral and anterior-posterior distance between the origin and midpoint of the line connecting the distal fibular notches and the center point of the medial malleolus (Figure 2D). Plateau width to BML ratio, vertical tuberosity location along the TSA, and the tuberosity location to BML ratio were recorded.

Figure 2:

A) Anterior coronal view with ‘O’ marking the origin, dashed line representing the medial-lateral axis, biomechanical length (BML) and margin varus angle (θ_M). B) Lateral sagittal view with dashed line representing the anterior-posterior axis, medial, lateral, and total tilt (λ_m, λ_l, λ_t). C) Top-down transverse view with dashed line representing the anterior-posterior axis, tuberosity lateral deviation (ν) and all nine proximal landmarks used for alignment. D) Top-down transverse view where dashed lines are on the distal surface with retroversion (θ_R), medial-lateral and anterior-posterior ankle offset (AO_ML and AO_AP).

Cross-Sectional Morphology

Aligned 3D reconstructions were exported as stereolithography (STL) files to import surface node locations of the cortical and medullary boundaries into a custom Matlab script where they were divided into 100 transverse cross-sections perpendicular to the TSA, spaced at 1% increments of the BML (Figure 3). Proximal (10% BML) and distal (90% BML) termination points for cross-sectional analyses were established to exclude the cancellous rich epiphyseal regions where an intramedullary OI endoprosthesis would not be placed, leaving 80 total analyzed cross-sections. The following parameters were calculated at each resection level: cortical and medullary area, total area, ratio of cortical area to total convex area, cortical and medullary perimeter, medullary circularity, maximum, minimum, and average cortical thickness, first and second principal cortical and medullary diameter, first and second moment of inertia, polar moment of inertia (Tommasini et al., 2005), and cortical and medullary orientation (Ruff and Hayes, 1983). Only morphologic measurements with direct application to OI endoprosthetic design and fit are discussed hereafter: first principal medullary diameter, average cortical thickness, and medullary circularity. The first principal medullary diameter measured along the principal axis of the medullary canal. Average cortical thickness is measured radially out from the centroid of the medullary canal. Medullary circularity was determined by 4π*(Medullary Area/Medullary Perimeter²) where 1 is a perfect circle. All additional results comparing gender (Figure 4) and laterality (Figure 5) have been made available but will not be further discussed.

Figure 3: Cross-Sections.

3D reconstruction of the tibia cortex with transverse cross-sections of the cortex shown (right) at 20, 35, 50, 65 and 80% BML. Cross-sectional analysis was completed at 1% intervals along the length of the bone from 10–90% BML and statistical analysis was performed for 15% intervals to capture regional differences.

Figure 4: Male/Female cross-sectional cortical and medullary morphology.

Morphologic measurements indicate a scaled difference between male and female populations in cortical features. However, this is not seen for most medullary features where there is a scaled difference proximally and little to no difference distally. This behavior indicates that females have a scaled thinner cortex but overall bone morphology doesn’t necessarily follow this pattern.

Figure 5: Right/Left cross-sectional cortical and medullary morphology.

Morphologic measurements indicate there is little to no difference in laterality of the tibia. Cortical and medullary orientation were the only morphologic properties that did show differences between right and left sides, indicating there may be a difference in rotational loading between the sides, however, this does not translate to overall cortical and medullary morphologic differences over the length of the bone. This behavior highlights the symmetrical loading pattern of the lower extremity.

Statistical Analysis

An a priori power analysis was calculated in G*Power (v3.1, Heinrich Heine Universität Düsseldorf) based on a similar study of humeral morphology (Drew et al., 2019). A total of 14 comparisons were made to capture sex differences and intra-specimen variability. The comparison that yielded the most conservative sample size estimate was medullary diameter at 35% BML between male and female specimens (9.2±1.2 and 10.1±2.0 mm, respectively). This indicated that at least 96 specimens would provide 80% power to detect a difference in medullary diameter with an effect size of 0.6 and significance level of p≤0.05.

Proximal landmarks (Figure 2C) were tested for inter- and intra-observer reliability in their selection on the models. Intraclass correlation coefficient (ICC) values compared between 2 observers on 30 randomly selected tibia with a 95% confidence interval based on a 2-way random-effects model for absolute agreement. There was at least one week between point selection on the same specimen for each observer. The Euclidean distance between selected points on the same specimen were analyzed both before and after alignment to see the effect landmark selection variability would have on alignment orientation and distance between points selected. All ICC values were calculated in SPSS Statistics (v20, IBM Corp, Armonk, NY) and a value greater than 0.9 was considered excellent agreement (Koo and Li, 2016).

Independent t-tests compared male and female morphology (explicitly at 20, 35, 50, 65, and 80% BML). Paired t-tests compared right and left side morphology at the same levels. A one-way ANOVA test with Tukey’s Post Hoc analysis compared intra-specimen variability of cross-sectional morphology measurements at 20, 35, 50, 65, and 80% BML. Tests were conducted in Matlab (R2018b, MathWorks, Natick, MA). Significance was considered p≤0.05.

Results

Repeatability and Accuracy

Intra-observer repeatability of anatomic landmark selection showed excellent agreement (ICC≥0.968) with Euclidean distances between points of 1.3±1.0 mm. Inter-observer repeatability also showed excellent agreement between points (ICC≥0.968) and average distance between points of 3.0±1.8 mm.

Anatomic Landmarks

Overall, male specimens were larger than female (Table 2). This was especially pronounced in BML, plateau width, and plateau height measurements (p≤0.001). However, the ratio of plateau width to BML showed no statistical difference indicating a preserved scaling between sexes (p=0.162). Contradicting this, the ratio of tuberosity location to BML differed between the sexes (p=0.001). Angular measurements of lateral plateau tilt and margin varus angle were different between males and females (p≤0.049), but not for other angular measures.

Table 2:

Location of anatomic landmarks on the cortical surface of the tibia

MEAN+SD	BML (mm)	Tuberosity Location (mm)	Lateral Plateau Width (mm)	Medial Plateau Width (mm)	Total Plateau Width (mm)	Plateau/BML Ratio (%)	Lateral Plateau Height (mm)	Medial Plateau Height (mm)
All	369.5±32.6	31.9±3.7	26.7±3.4	25.4±2.9	64.0±5.8	17.4±1.3	27.5±4.1	34.3±3.8
Male	385.4±29.3^	32.4±3.9^	28.3±2.9^	26.9±2.4^	67.2±4.6^	17.5±1.3	28.8±4.4^	35.8±3.5^
Female	345.3±20.4^	31.0±13.2^	24.3±2.7^	23.1±2.0^	59.1±3.7^	17.2±1.2	25.4±12.7^	32.0±2.9^
Right	369.4±31.8	31.7±3.6	26.2±3.1*	25.2±2.5	63.3±5.3*	17.2±1.3*	26.5±3.5*	33.5±3.4*
Left	369.5±32.6	32.0±3.7	27.1±3.5*	25.6±3.3	64.7±6.1*	17.5±11.2*	28.5±4.3*	35.1±13.9*
	Lateral Plateau Tilt (deg)	Medial Plateau Tilt (deg)	Margin Varus Angle (deg)	Tuberosity Lateral Deviation (deg)	Tuberosity/BML Ratio (%)	Retroversi on (deg)	Ankle Offset-ML (mm)	Ankle Offset-AP (mm)
All	9.5±3.9	11.6±3.5	4.3±2.0	17.8±6.3	8.7±0.9	25.2±8.4	7.2±13.0	6.6±13.2
Male	10.1±3.6^	11.5±3.5	4.7±2.0^	17.5±6.8	8.4±1.0^	25.8±19.3	7.7±3.3^	6.8±13.2
Female	8.6±44.3^	11.7±3.5	3.8±1.9^	18.2±5.4	9.0±0.8^	24.3±6.7	6.6±2.3^	6.4±3.2
Right	10.3±4.2*	11.7±3.5	4.2±2.0	18.6±6.4	8.6±0.9	26.1±8.4	7.6±12.8*	6.9±13.2
Left	8.7±3.9*	11.4±3.5	4.4±2.0	17.0±5.5	8.7±0.9	24.3±7.7	6.9±13.0*	6.4±13.2

^{^}indicates significant difference between male and female populations.

^*indicates significant difference between right and left sides.

There was more overall agreement between anatomic landmarks on right and left sides (Table 2). Measurements associated with the tibial tuberosity, retroversion, and margin varus angle showed no statistical differences (p≥0.079). However, all measurements associated with the tibial plateau, except medial plateau width and tilt, yielded statistical differences between right and left sides (p≤0.018).

Cross-Sectional Morphology

The tibial medullary canal for the entire cadaver population had an hourglass shape with the largest volume proximally (Figure 3). On average for all specimens, 64% BML had the smallest diameter (13.4±2.3 mm first principal medullary diameter). Medullary diameter varied along the length of the bone (p≤0.003). Cortical thickness had an inverted relationship to medullary diameter, where the middle was thickest and the proximal and distal regions were the thinnest (Figure 3). On average, the thickest cortex occurred at 57% BML (5.1±0.9 mm). In cortical thickness we see mirroring about the mid-shaft, where the comparisons between 20 to 80% BML, 35 to 65% BML, and 50 to 65% BML did not differ due to the hourglass shape (p≥0.076). All other comparisons did result in statistical differences when reflected about the mid-shaft (p≤0.001). Medullary circularity differed between levels compared to the most proximal 20 and 35% BML (p≤0.001), but there were no differences in comparisons that did not involve 20 and 35% BML cross-sections.

Medullary circularity did not show differences between male and female specimens at 20, 35, 50, and 80% BML (p≥0.198, Figure 6). Average cortical thickness was statistically different between the sexes at all analyzed cross-sections (p≤0.001, Figure 6). First principal medullary diameter only showed differences between the sexes at 20, 35, and 50% BML (p≤0.020, Figure 6). There were no differences between right and left specimens for any of these comparisons.

Figure 6: Cross-Sectional Analysis Results.

Transverse cross-sectional cortical morphology normalized to percent biomechanical length (BML) where 10% is the most proximal. 1^st principal medullary diameter is the diameter measured along the principal axis of the medullary canal. Average cortical thickness is determined by the average cortical thickness measured radially out from the centroid of the medullary canal. Medullary circularity is determined by the equation 4π*(Medullary Area/Medullary Perimeter²).

* indicates significant difference between male and female populations. Statistical comparisons were made at 20, 25, 50, 65, and 80% BML.

Discussion

This investigation examined the tibia by two means: cortical anatomic landmarks and cross-sectional morphology along the length of the bone. This detailed analysis of the cortical and medullary morphology can be used to design orthopaedic endoprostheses, specifically those for transtibial percutaneous OI attachment of prosthetic limbs. We hypothesized that sex would influence morphology but laterality would not. Both sex and laterality impacted anatomic landmarks, but only sex influenced cross-sectional morphology. Cross-sectional morphology changed dramatically throughout the length of the bone for all examined properties except medullary circularity.

Comparisons between the sexes revealed that males have thicker cortices than females. Female average cortical thickness was 14% thinner than males. However, medullary diameter did not follow this pattern. At proximal locations there was a difference between males and females, but this did not remain true for distal levels. This indicates that females have a thinner cortex per medullary area proximally, possibly putting females at greater risk for periprosthetic fracture since they would have the same intramedullary endoprosthesis placed but less cortex to support the device.

There was also a scaling effect in the external cortical landmarks where males were, on average, larger than females. This agrees with previous research (Sherk et al., 2012). It is important to consider that male tibia is significantly larger in BML (p≤0.001) because this will inherently increase the loading of the implant due to increased body weight (Table 1) and increased moment arm for forces applied between the knee and ankle. A female with amputation at the same BML could experience lower bone-implant interface forces due to body habitus, but with a decreased amount of cortical bone to provide stable fixation there is a possibly of increasing fracture risk. However, the female bone may be more representative of amputees since amputees often experience disuse atrophy, osteopenia and osteoporosis (Gailey et al., 2008; Sherk et al., 2008). These conditions are also expected to increase risks for periprosthetic fracture. Future research of amputee residual limb morphology and the correlation between cortical thickness and fracture strength is necessary to better understand how these conditions would impact patient safety with use of percutaneous OI attachment.

To achieve OI fixation, there must be intimate contact between the endosteal bone and implant osseointegration surface. Medullary diameter dictates the minimum endoprosthesis size in the absence of cortical bone removal necessitated for canal preparation for the implant. Average medullary diameter varies dramatically from 13.4 mm at 64% BML to 53.8 mm at 10% BML, indicating that the size range necessary for a generic conical or cylindrical transtibial OI endoprosthesis is much larger than that for the humerus (Drew et al., 2019).

The tibia presents a challenge for percutaneous OI endoprosthetic design because the medullary canal has an hourglass shape with a proximal funnel and distal cone. This makes it difficult for one implant design to maintain intimate bone contact in all locations, so we look for design features from existing percutaneous OI endoprostheses. Two primary approaches have been taken to fit the inverted funnel shaped femoral medullary canal: a cylindrical (e.g. Osseoanchored Prosthesis for the Rehabilitation of Amputees (OPRA), Osseointegrated Prosthetic Limb (OPL), and Integral Leg Prosthesis (ILP)) and a conical (e.g. Percutaneous Osseointegrated Prosthetic (POP)) geometry (Zaid et al., 2019). These designs are intended to maximize bone-implant contact and facilitate bone ingrowth, but a cylinder will remove more bone proximally in an inverted funnel canal. Depending on the rate of endosteal convergence, the length of the endoprosthesis may compromise the region too much. Thus, a balance between endoprosthesis length, diameter, and depth of bone preparation must be struck to optimize implant fit and maximize contact area for bone ingrowth. Conical implants are the logical design choice in the inverted funnel medullary regions.

The funnel shape of the proximal tibia is more difficult to accommodate with existing cylindrical and conical implant designs. As the medullary canal diverges, intimate bone-implant contact is lost and the walls of the canal progress away from the axis of the implant stem. One approach used for short transhumeral residual limbs is to add proximal screws to enhance implant stability where there is not direct contact between the bone and implant (Drew et al., 2020). This approach has been utilized for transtibial amputees with short residual limbs, but in custom 3D printed endoprosthesis designs (Atallah et al., 2017; Frolke et al., 2017).

The present study suggests a combination of approaches or designs based on amputation levels may be needed to accommodate varying tibial medullary morphology. Conical and cylindrical endoprosthesis designs require a roughly circular medullary cross-section so that intramedullary shape does not need to be altered extensively. Other shapes could require extensive canal preparation and bias the implant axis, leaving little cortex to support an endoprosthesis in some areas. A nearly circular cross-section is present in the tibia from 35–90% BML. Proximal amputation levels appear to present the most challenge for a round intramedullary endoprosthesis. At the 20% BML location, cortex is thinnest (3.0±0.7 mm) with an accompanying large medullary diameter (31.9±4.2 mm). This relative lack of cortex leaves little to shape or support an implant. In addition, this location is the least circular (0.92±0.03 circularity) making it the most necessary region to prepare for a round implant design (Figure 6). In comparison, at 50% BML, the average cortical thickness (5.0±0.9 mm), medullary diameter (15.0±2.4 mm), and circularity (0.94±0.02) are much more balanced leaving room to shape the medullary canal to an endoprosthesis. These results indicate that a round endoprosthesis cross-section, like the majority of those currently in use (Zaid et al., 2019; Gerzina et al., 2020; Hagberg et al., 2020) may not be suitable for those with very short residual limbs. Finally, endoprosthesis diameters must vary dramatically, from 11–36 mm, to accommodate amputation between 20–80% BML. Percutaneous OI implant systems sometimes provide size options in millimeter increments. A tibial system could therefore require up to 25 variations, with size-matched instrumentation for each option. The cost of manufacturing alone could prove prohibitive unless alternate design strategies are evaluated.

It is also important to consider the tuberosity location in relationship to these cross-sectional features, because any amputation proximal to the patellar tendon insertion compromise knee function. The midpoint of the tibial tuberosity was found at roughly 9% BML (Table 2). To maintain the complete patellar tendon insertion (if intact), preparation would likely need to be distal to ~20% BML. Therefore, there is a threshold below which candidacy for a transtibial percutaneous OI device must be factored against lost knee function.

A limitation of the present study is that the sample population was comprised of intact tibia. In amputees, heterotopic ossification and other abnormal bone growth is common (Potter et al., 2007). Additionally, bone resorption and advanced osteoporosis are possible in amputees both with and without an intramedullary OI implant (Gailey et al., 2008; Sherk et al., 2008; Jeyapalina et al., 2012; Jeyapalina et al., 2014). These factors could compromise the shape of the distal medullary canal in amputees. For this reason, imaging and alternative surgical preparations should be available to plan for these circumstances preoperatively.

Though the focus of this study was to examine tibial morphology and how it relates to percutaneous OI device design, it also expands on the tibia morphology literature to provide a comprehensive characterization of anatomic landmarks and medullary morphology to further refine designs of the tibial component of total ankle and knee replacements. Intramedullary fixation is becoming more common as percutaneous OI systems have emerged as a method of prosthetic limb fixation. With this new technology comes a need to understand the morphology of the diaphyseal tibia. Previous transfemoral OI systems were often forced into the tibia without data to suggest the designs would translate (Thesleff et al., 2018). This study has provided detailed information regarding the diaphyseal morphology of the tibia that will translate to a more targeted approach to percutaneous OI attachment of transtibial prosthetic limbs for the largest amputee population.