The Segond fracture occurs at the site of lowest sub‐entheseal trabecular bone volume fraction on the tibial plateau

Abstract In a series of human cadaveric experiments, Dr. Paul Segond first described the avulsion injury occurring at the anterolateral tibial plateau that later took his name. The fracture is thought to arise as a consequence of excessive tibia internal rotation which often also elicits damage to other connective tissue of the knee. The exact mechanism behind the avulsion is, however, unclear. A number of ligamentous structures have been proposed in separate studies to insert into the Segond fragment. Suggestions include the iliotibial band (ITB), biceps femoris and the controversial ‘anterolateral ligament’ (ALL). Despite increasing knowledge of tibial plateau bony microarchitecture in both healthy and disease states, no studies have yet, to our knowledge, considered the role of tibial sub‐entheseal bone structure in pathogenesis of the Segond fracture. The goal of this study was thus to elucidate the differences in trabecular properties at regions across the tibial plateau in order to provide an explanation for the susceptibility of the anterolateral region to avulsion injury. Twenty human tibial plateaus from cadaveric donors were dissected and imaged using a Nikon‐XTH225‐μCT scanner with <80 μm isotropic voxel size. Scans were reconstructed using MicroView 3D Image Viewer and Analysis Tool. Subsequent virtual biopsy at ten anatomically defined regions of interest (ROI) generated estimates of bone volume fraction (‘bone volume divided by total volume’ (BV/TV)). The overall mean BV/TV value across all 20 tibiae and all 10 ROIs was 0.271. Univariate repeated‐measurements ANOVA demonstrated that BV/TV values differed between ROIs. BV/TV values at the Segond site (Sα, Sβ or Sγ) were lower than all other ROIs at 0.195, 0.192 and 0.193, respectively. This suggests that, notwithstanding inter‐ and intra‐specimen variation, the Segond site tends to have a lower trabecular bone volume fraction than entheseal sites elsewhere on the tibia. Since BV/TV correlates with tensile and torsional strength, the lower BV/TV at the Segond site could equate to a region of local weakness in certain individuals which predisposes them to an avulsion injury following the application of force from excessive internal rotation. The low BV/TV recorded at the Segond site also challenges the idea that the fracture occurs due to pull from a discrete ‘anterolateral ligament’, as the tension exerted focally would be expected to elicit a hypertrophic response in line with Frost's Mechanostat hypothesis. Our data would instead agree with the aforementioned reports of the fibrous band at the Segond site being part of a broader insertion of an ‘anterolateral complex’.


Abstract
In a series of human cadaveric experiments, Dr. Paul Segond first described the avulsion injury occurring at the anterolateral tibial plateau that later took his name.
The fracture is thought to arise as a consequence of excessive tibia internal rotation which often also elicits damage to other connective tissue of the knee. The exact mechanism behind the avulsion is, however, unclear. A number of ligamentous structures have been proposed in separate studies to insert into the Segond fragment.
Suggestions include the iliotibial band (ITB), biceps femoris and the controversial 'anterolateral ligament' (ALL). Despite increasing knowledge of tibial plateau bony microarchitecture in both healthy and disease states, no studies have yet, to our knowledge, considered the role of tibial sub-entheseal bone structure in pathogenesis of the Segond fracture. The goal of this study was thus to elucidate the differences in trabecular properties at regions across the tibial plateau in order to provide an explanation for the susceptibility of the anterolateral region to avulsion injury.
Twenty human tibial plateaus from cadaveric donors were dissected and imaged using a Nikon-XTH225-μCT scanner with <80 μm isotropic voxel size. Scans were reconstructed using MicroView 3D Image Viewer and Analysis Tool. Subsequent virtual biopsy at ten anatomically defined regions of interest (ROI) generated estimates of bone volume fraction ('bone volume divided by total volume' (BV/TV)). The overall mean BV/TV value across all 20 tibiae and all 10 ROIs was 0.271. Univariate repeated-measurements ANOVA demonstrated that BV/TV values differed between ROIs. BV/TV values at the Segond site (Sα, Sβ or Sγ) were lower than all other ROIs at 0.195, 0.192 and 0.193, respectively. This suggests that, notwithstanding inter-and intra-specimen variation, the Segond site tends to have a lower trabecular bone volume fraction than entheseal sites elsewhere on the tibia. Since BV/TV correlates with tensile and torsional strength, the lower BV/TV at the Segond site could equate to a region of local weakness in certain individuals which predisposes them to an avulsion injury following the application of force from excessive internal rotation. The low BV/ TV recorded at the Segond site also challenges the idea that the fracture occurs due to pull from a discrete 'anterolateral ligament', as the tension exerted focally would be 1 | INTRODUC TI ON

| The Segond fracture
In a series of human cadaveric experiments, Dr. Paul Segond (1879) first described the avulsion injury occurring at the anterolateral tibial plateau that later took his name. The Segond fracture commonly occurs alongside local soft tissue injury; studies have shown 75%-100% of injuries result in anterior cruciate ligament (ACL) tear and 66%-70% result in meniscal damage as a consequence of the trauma (Dietz et al. 1986;Goldman et al. 1988). It has been suggested that this is due to soft tissue structures of the anterolateral region being placed under a comparable degree of strain to intracapsular ligaments by tibial internal rotation (Dodds et al. 2014). Damage to these anterolateral soft tissue structures have been shown to elicit a positive pivot shift phenomenon-a finding classically indicative of ACL tear (Hughston et al. 1976;Bull et al. 1999)-suggesting this tissue is involved in maintaining similar axes of knee stability as the ACL. Moreover, anterolateral structures have been suggest to have a more important role in resisting internal rotation because of the larger moment arm they carry compared to the more centrally located ACL (Amis 2017).
Almost a century after Segond's original reports of a 'pearly, resistant, articular fibrous band' (Segond 1879, p. 14) that was placed under strain by the same internal rotation forces that are resisted by the ACL, Kaplan (1958) proposed that deep fibres of the iliotibial band (ITB) insert into the Segond fragment. Later groups suggested other insertions, including part of the short head of biceps femoris (Terry and LaPrade 1996) and an extension of the lateral capsular ligament (Johnson 1979;Woods et al. 1979). These observations underlie the classification of the injury as an avulsion, insofar as the mechanism involves a soft tissue structure pulling on the bone at the site of insertion. More detailed analysis later suggested that the fibres which insert into the Segond fragment may be considered-functionally or anatomically-part of a distinct ligamentous structure: the 'anterolateral ligament' (ALL). Comparison of radiological data from patients with a possible Segond fracture with cadaveric reports of the ALL's tibial insertion demonstrated that the avulsion occurred at the exact site of ALL insertion (Claes et al. 2014). Biomechanical analysis has shown that, like the ACL, the ALL is an important stabiliser of internal rotation, and may play a more important role in stability during knee flexion (Parsons et al. 2015).
There is a lack of consensus surrounding the incidence of the ALL in adult knees, with some authors arguing for inexistence whilst others claiming a presence in 100% of knees (Ariel de Lima et al. 2019). Arguments against include MRI data has suggested that the ALL is inseparable from neighbouring lateral collateral ligament (LCL) and ITB (Porrino et al. 2015). It follows that the Segond fragment could, therefore, receive insertion from one of several closely opposed ligamentous and capsular attachments within the 'anterolateral complex' (Shaikh et al. 2017). Knowledge of the anatomy of this region is important for characterising the mechanisms of traumatic internal structural derangement and to help guide anterolateral capsule repair-an intervention which has been shown to restore rotational stability and correct pivot shift (Ferretti et al. 2017b).

| MicroCT and virtual biopsy
Micro-computed tomography (μCT) is an imaging modality which is limited to scans of smaller scale specimens than a typical clinical CT scanner. It is, however, able to do so at a much higher resolution with a pixel size in the order of 10 s compared to 100 s of microns. The detail of the acquired image allows for repeated 'virtual biopsy' of a specimen in a non-destructive manner, while the richness of information also allows sampling of the image data volume at locations that can be selected from multiple orthogonal viewing planes.
μCT has advantages over dual-energy X-ray absorptiometry (DXA) for assessment of bone mineral density (BMD) due to its ability to incorporate three dimensions in the reconstructed imaging in addition to the superior resolution. It is therefore used as a means of conferring validity to novel DXA techniques attempting to mimic the resolution of other imaging modalities (Briggs et al. 2010). DXA remains the clinical modality of choice for assessment of BMD because of the lower radiation dose, relatively low cost and ability to scan larger specimens (Kleerekoper and Nelson 1997). μCT is instead currently restricted-in human imaging research-to analysis of ex vivo specimens.
Variations in apparent bone trabecular volume fraction are related by a power-law function to bone tensile and torsional strength (Kaplan et al. 1985;Sarin et al. 1999). Trabecular bone has been shown to have a significantly lower tensile strength compared to compression strength, explaining why the force of injury in avulsion fractures is typically much lower than that seen in other types of fracture (Kaplan et al. 1985). As a precedent, μCT has been used to quantify the correlation between trabecular bone volume fraction expected to elicit a hypertrophic response in line with Frost's Mechanostat hypothesis. Our data would instead agree with the aforementioned reports of the fibrous band at the Segond site being part of a broader insertion of an 'anterolateral complex'.
Trabecular bone volume fraction (bone volume over total volume, BV/TV, expressed in %) recorded using μCT has also been used as a means of predicting strength and stiffness in both normal and pathological trabecular bone (Nazarian et al. 2008).

| Gaps in the field and aims of the current study
Despite increasing knowledge of tibial plateau bony microarchitecture in both healthy and disease states, no studies have yet, to our knowledge, considered the role of tibial sub-entheseal bone structure in pathogenesis of the Segond fracture. The goal of this study was thus to elucidate the differences in trabecular properties according to μCT analysis at regions across the tibial plateau and quantify the relative bone densities underlying each enthesis.
When referencing entheseal sites, we intend to discuss each as a functional organ-including adjacent trabecular bone structure in addition to the cortex which receives the insertion. We hypothesised that BV/TV at the Segond site was lower than other entheses across the plateau, explaining the propensity for avulsion.

| Dissection
Lower limb specimens were randomly selected for dissection from human cadavers. All donors had provided written consent before decease for their bodies to be used for anatomical research, in compliance with the Human Tissue Act 2004. Specimens with evidence of overt knee trauma, surgery or degenerative joint disease were excluded. We further eliminated specimens whose records stated the cause of death was from breast, prostate or lung cancer, as these are the most common cancers which seed bony metastases (Mundy 2002;Svensson et al. 2017). Fifteen female (age range, 72-99; mean(SD) age, 87.2(8.4) years) and five male (age range, 82-93; mean(SD) age 87.4(5.3) years) donors passed the initial screening and were thereby included in the study. A standardised dissection procedure was used for each specimen, involving removal of skin and soft tissues, as well as disarticulation of the knee joint. The isolated tibial plateaus remained connected to the adjacent fibulae by their associated ligaments. Both were cut to around 10 cm in length such that they would fit within the apparatus for loading into the μCT scanner. Tendons were left in place to act as reference points for later virtual biopsy.

| MicroCT scanning
Specimens were packed individually into a polystyrene holding container such that they would stand upright independently and remain stationary during rotation of the platform within the scanner.
The holding containers were loaded into and imaged using a Nikon XTH225 μCT scanner (Nikon Metrology UK Ltd.), with <80 μm isotropic voxel size. The scan time for each sample was approximately 25 min. DICOM imaging output was exported for viewing and analysis. 10. Iliotibial band (ITB) insertion-same vertical depth as (7), centred 50% of the ML width of Gerdy's tubercle.

| Virtual biopsy
The relative locations of the ROIs detailed above are shown in Figure 2. The medial collateral ligament could not be included as its insertion varied between tibiae and thus could not be reliably found using the methodology above. The medial and lateral condyles represent compression zones: included as data to add insight into the relationship between compression force and bone volume. The Segond fracture has been stated as having an average length of 10 mm (Shaikh et al. 2017).
Due to the 5 mm diameter sphere which defined the portion being measured, 3 ROIs were used to sample the full diameter of the Segond site and to map any variations that might be across this region: Sα, Sβ and Sγ.
Using MicroView software, BV/TV was measured at each ROI.
The value represented the fraction of the ROI occupied by trabecular bone. A value of 0 would represent thin air, and a value towards 1 would be found in near-solid cortical bone. At each ROI, a total of five replicate measurements were taken by moving the measuring tool 1 mm from the calculated centre point in four opposing directions on the cortical axis-for example, at the ACL site, the repeat measurements were taken 1 mm anterior, posterior, medial and lateral from the calculated centre point. Crucially, cortical bone was avoided during repeat measurements by keeping depth in relation to the cortex constant for each repeat measurement. This was necessary so that the higher BV/TV of cortical bone did not inflate the measurement of bone volume fraction of the underlying trabecular bone.

| Statistical analysis
Five replicate BV/TV measurements were obtained using MicroView for each ROI in each specimen. The means of the replicates represent 10 ROI values (mean BV/TV) for each specimen. Descriptive statistics and analyses were performed on these values as described.

| RE SULTS
Mean BV/TV values for each ROI are shown in Table 1. In 15 of the 20 tibiae, the lowest intra-specimen mean BV/TV was found at the Segond site (range = 0.092-0.262, recorded at either Sα, Sβ or Sγ).
In the remaining 5 tibiae, the lowest intra-specimen mean BV/TV   Figure   S1a and Figure S1b shows the condensed SEG-substituted data.
A higher incidence of Segond fractures has been reported both in men (Claes et al. 2014) and in right-sided tibiae (Ferretti et al. 2017a). Our study was not designed to evaluate any effect of sex or side, but used an opportunity sample from cadavers available for anatomical dissection. Nevertheless, among the specimens, 13 were left-sided (10 female and 3 male) and 7 right-sided (5 female and 2 male). Secondary analyses to explore the effect of SEX or SIDE were performed using the repeated-measures ANOVA (with type II sum of squares) described above with incorporation of either SEX or SIDE as a between-subjects factor. Interpretation of the results must account for the opportunistic nature of the sample; nevertheless, our data provide no support for the existence of a difference in trabecular bone density between sexes (F 1,18 = 2.24, p = 0.152) or sides (F 1,18 = 0.345, p = 0.564). Neither was there evidence of an interaction between ROIs and either SEX (F 5.4,96.7

| D ISCUSS I ON
This study shows that BV/TV at the Segond site is significantly lower than other entheses across the tibial plateau. As mentioned previously, BV/TV has been shown to correlate with tensile and torsional strength-the forces putatively responsible for avulsion fractures (Kaplan et al. 1985;Sarin et al. 1999). The lower BV/TV at the Segond site could, therefore, equate to a region of local weakness in certain individuals which predisposes them to the avulsion injury following excessive internal rotation of the knee. This 'weakest link' hypothesis agrees with findings that the minimum BV/TV value for a specimen gave a far higher predictive power than the average specimen BV/TV in predicting the probability of mechanical failure of trabecular bone (Nazarian et al. 2006). Given the complex ligamentous and capsular arrangements around the knee joint, it is reasonable to assume that several structures would be placed under strain during internal rotation. The avulsion could, therefore, arise from the trabecular bone at the Segond site being a highly susceptible locus on the tibial plateau, with other injuries accumulating sequentially following progressive increases in internal rotational force. Since avulsions elsewhere on the tibial plateau occur at a much lower incidence (Bali et al. 2012;Edmonds et al. 2015;Caggiari et al. 2020), it may be the case that a higher entheseal BV/TV means the weakest link lies in the substance of the ligament, causing a mid-substance tear to be more likely than an avulsion fracture. Segond's original work described an extreme tractional force along a 'fibrous band' on the anterolateral aspect of the knee during tibial internal rotation (Segond 1879, p. 14). The low BV/TV we recorded at the Segond site challenges the hypothesis that this band represents a discrete ligament, such as the ALL, as the tension exerted focally by a single ligament would be expected to elicit a hypertrophic response in line with Frost's Mechanostat hypothesis (Frost 2000 however, not be possible using the current methodology since μCT is limited to imaging smaller, ex vivo specimens. A first step would be to mirror the data in human subjects using a technique such as DXA imaging; however, at present, the resolution of these alternate imaging methods falls short of μCT.

ACK N OWLED G EM ENTS
We would like to thank the Anatomical Society for granting the scholarship which was used in the pilot project involving five specimens; the donors whose generosity made this research possible; Mr

CO N FLI C T O F I NTE R E S T
Authors have no conflicts of interest to declare. approval of the article.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.