Auricular surface morphology and surface area does not influence subchondral bone density distribution in the dysfunctional sacroiliac joint

The subchondral lamella of the sacroiliac auricular surface is morphologically inconsistent. Its morpho‐mechanical relationship with dysfunction (SIJD) remains unstudied. Here, the iliac and sacral subchondral bone mineralization is compared between morphological subtypes and in large and small surfaces, in SIJD joints and controls. CT datasets from 29 patients with bilateral or unilateral SIJD were subjected to CT‐osteoabsorptiometry. Surface areas and posterior angles were calculated and surfaces were classified by size: small (<15 cm3) and large (≥15 cm3), and morphological types: 1 (>160°), 2 (130°–160°), and 3 (<130°). Mineralization patterns were identified: two marginal (M1 and M2) and two non‐marginal (N1 and N2). Each sacral and iliac surface was subsequently classified. Dysfunctional cohort area averaged 15.0 ± 2.4 cm2 (males 16.2 ± 2.5 cm2, females 13.7 ± 1.6 cm2). No age correlations with surface area were found nor mean Hounsfield Unit differences when comparing sizes, sexes or morphology‐type. Controls and dysfunctional cohort comparison revealed differences in female sacra (p = 0.02) and small sacra (p = 0.03). There was low‐conformity in marginal and non‐marginal patterns, 26% for contralateral non‐dysfunctional joints, and 46% for dysfunctional joints. The majority of painful joints was of type 2 morphology (59%), equally distributed between small (49%) and large joints (51%). Larger joints had the highest frequency of dysfunctional joints (72%). Auricular surface morphology seems to have little impact on pain‐related subchondral lamella adaptation in SIJD. Larger joints may be predisposed to the onset of pain due to the weakening of the extracapsular structures. Dysfunctional joints reflect common conformity patterns of sacral‐apex mineralization with corresponding superior corner iliac mineralization.


| INTRODUCTION
Sacroiliac joint dysfunction (SIJD) is increasingly recognized as a cause of pelvic girdle pain (Sembrano & Polly, 2009;Vleeming et al., 2008). It is defined as an abnormal condition of the sacroiliac joint (SIJ), accompanied by altered motion, instability, asymmetry and pain usually in the groin, leg and buttocks or a combination of these areas (Freburger & Riddle, 2001). In the painless joint state under bipedal loading conditions, the nutation of the sacrum relative to both ilia, tightens the posterior sacroiliac joint ligaments resulting in an inwards rotation of the iliac wings to the sacrum in a sagittal axis. This draws the ilia and sacrum firmly together via the "forceclosure" and "form-closure" phenomena (Booth & Morris, 2019;Dontigny, 1985;Jordan, 2006;Snijders et al., 1993;Vleeming et al., 2012;Zlomislic & Garfin, 2019). In the dysfunctional state, this homeostasis is affected: the sacrum tends to "counter-nutate," with innominate anterior rotation impacting the overall stability of the pelvic girdle (Hungerford & Gilleard, 2007;Ito et al., 2020;Lee, 2007).
Sacroiliac dysfunction can arise from several mechanisms associated with a current or a past pregnancy, etiological conditions, predisposed conditions but may also arise secondary to trauma (Booth & Morris, 2019;Chou et al., 2004;Forst et al., 2006;Vleeming et al., 2012). Because the SIJ's innervation is highly complex and variable, pain can be somatically referred from other adjacent nociceptors in the buttocks, groin and leg which underlines the involvement of ligaments as pain generators (Hammer et al., 2013;McGregor & Cassidy, 1983;Spiker et al., 2012).
It has been hypothesized that morphological variation in the shape of the auricular surfaces may play a part in the spontaneous onset of dysfunction (Jesse et al., 2017). The morphology of the auricular surfaces of the SIJ has been studied in relation to the changes occurring in response to pelvic sexual dysmorphism and the natural degeneration that the joint endures with time (Anastasiou & Chamberlain, 2013;Brunner et al., 1991;Nishi et al., 2017;Nishi et al., 2018;Nishi et al., 2019;Rmoutilova et al., 2017;Valojerdy & Hogg, 1989). Some studies have looked into the influence of the morphological differences and the placement of the endplates in relation to the bony anatomy (Brunner et al., 1991;Rana et al., 2015;Waldrop et al., 1993) to understand the role of the auricular endplate in joint mobility and bipedal force transmission. But, only a handful of papers have looked into whether the general shape and size of the SIJ auricular surface is associated with painful conditions of the SIJ, and what this might signify in relation to the detection and management of these conditions (Beal, 1982;Jesse et al., 2017;Nishi et al., 2019;Ou-Yang et al., 2017).
The auricular surface (size and morphology) is variable, varying inter-individually, as well as side dependently even within the same person. Three dimensionally, the auricular surface is not flat but composed of ridges and grooves (Forst et al., 2006;Puhakka et al., 2004;Snijders et al., 1993) allowing the sacra and ilia to firmly interlock.
A previous article (Poilliot, Doyle, et al., 2021) reports variation in size and shape of the auricular surface in a control cohort of SIJs and look into potential differences in subchondral bone density. Results showed that size and morphology have little impact on the subchondral bone density patterns, however, larger joints had higher mineralization only seen on the iliac side (Poilliot, Doyle, et al., 2021).
Applying the same methods, this study aims to compare the mineralization distribution between the three morphological types in a cohort of SIJ dysfunctional joints. This will also be looked at when comparing joint size to see if that has an influence on mineralization. The densitograms would allow the visualization of the mineralization pattern across the surface, which corresponds to the continuous loading conditions the joint endures. This will highlight the areas of high/low stress within the joint and thus provide information on the load dissipation within the painful state of the SIJ. It was hypothesized that joints with dysfunction will reflect differences in patterns compared to non-dysfunctional joints in the superior and anterior parts of the auricular surface.

| Specimens
Twenty-seven patient cases diagnosed with unilateral or bilateral SIJD (13 females; 14 males; range 26-79 years) were collected between 2009 and 2018 the JCHO Sendai Hospital, Sendai, Japan. These patients have been used in previous studies (Poilliot, Doyle, et al., 2021;. All patients identified the posterior superior iliac spine as the painful zone via a one-finger test (Murakami et al., 2008) and following examination were considered to having SIJ pain. Definitive diagnosis of SIJD in these patients was confirmed by more than 70% pain relief after SIJ peri-and intra-articular injections under fluoroscopic guidance (Murakami et al., 2007;Murakami et al., 2018). Patients with a history of infection, tumors in the lumbopelvic area, recent lumbar spine and pelvic fractures, and seronegative spondylarthropathy were excluded. All included patients had a history of other injections including selective nerve root infiltration and/or lumbar disc nerve block that were negative. All patients were diagnosed as having severe chronic SIJ pain for a minimum of 6 months.
In this study, the SIJD cohort was separated into the dysfunctional joint cohort (all bilateral cases and the unilateral painful sides) and the non-dysfunctional joints (those with dysfunction on the contralateral side) (Figure 1). In addition, data corresponding to a group of control specimens (with no known SIJ pain or pathology), were used for the final comparison of this study comparing the SIJD patients with the controls. This cohort was made up of body donors and patient CT scans ( Figure 1) representing a general 'pain-free state' of the joint. These were analyzed radiographically by experienced radiographers to rule out SIJ-related conditions.

| CT osteoabsorptiometry (CT-OAM)
Conventional clinical CT (SOMATOM as64 open, Siemens, Munich, Germany) data sets from patients were deployed as in previous studies (Poilliot et al., 2020). Slice thickness was 0.6 mm. Datasets were evaluated using an image analysis software (Analyze, v7.4, Biomedical Imaging Resource, Mayo Foundation, Rochester, NY, USA). Ilia and sacra were first manually segmented within the CT scans to reconstruct a three-dimensional lateral reconstruction of both bones before the subchondral endplate of both the sacrum and the ilium of each specimen was manually isolated. The maximum intensity projection revealed the Hounsfield Unit (HU) of each pixel of the endplate as a colormap (Johnston et al., 2010) so that for every image point, each maximum density value of the underlying bone plate was projected onto the surface. Threshold values were ≤200 to ≥1200 HU chosen according to previous studies and displayed as color-coded densitograms (Leumann et al., 2015).

| Classification into morphology types
Using ANALYZE, three-dimensional densitograms of the bones were oriented laterally to face the auricular surfaces. Inkscape v1.0.2-2 (The Inkscape Project, NY, USA; https://inkscape.org) was then used to calculate the alpha-angle for the classification of the surface into morphology T1, T2 and T3 based on the method developed by Jesse et al. (2017). The three morphologies were as follows: T1 has a wide posterior angle (>160 ), T3 narrow posterior angle (<130 ) and T2 is in between (130 and 160 ). The process involved making two lines through the centre of each limb and a third through the cross over point of the first two lines and the "centre" of the angle at the posterior border. The mean angle between the two limbs was then measured via the three points ( Figure 2). This process was repeated anew three times for each surface and an average of the three measurements was calcuated for the final angle classification.

| Area classification
Classification of "large" and "small" joints was arbitrarily categorized based on the mean area of all the joints and the controls. "Small" surfaces had an area lower than the mean area, 'large' joint surfaces had an area higher than the mean area.

| Pattern classification methodology
Patterns were identified belonging to two main groups: the marginal (M) and non-marginal groups (N). This assessment was made based on a semi-quantitative analysis of the entire surface region colourmap of each joint surface. A pattern was marginal if 60%-70% of the surface was less mineralized compared to the highly dense regions constituting the remaining 30%-40%. The classification of the non-marginal patterns used the opposite criteria.
F I G U R E 2 Schematic representation of the method for the posterior angle calculation using Inkscape. Yellow dotted lines: Perpendicular through the Centre of both limbs (points 1 and 3), blue dashed line: Through the crossover point of the yellow lines and Centre of the posterior angle (point 2). The posterior angle is measured between points 1, 2, and 3. A, anterior; I, inferior; P: posterior; S, superior F I G U R E 1 Schematic representation of the materials used for the study. *None of these cases had a current or past history of low back pain, sacroiliac joint-related pathology or abnormalities on previous medical records and visible sacroiliac joint conditions visible on the CT (computed tomography) scans. This cohort was previously used in Poilliot, Doyle, et al. (2021) and  2.6 | Statistical analysis Prism (version 9.0.2, GraphPad, San Diego, CA, USA) was used for the statistical analyses. Normal distribution of the data was determined using the Shapiro-Wilk test. Based on the distribution, an ANOVA or a Kruskal-Wallis test was used to compare size differences and HU values between left and right sides as well as iliac and sacral sides between sexes. Age correlations were determined using a Pearson-r test. Dunn's post-hoc correction was not applied. p-values of 0.05 or less were considered statistically significant. Values are given as mean values ± standard deviations with 95% confidence intervals (CI). Correlation was defined as follows: strong ≥0.7, moderate 0.7 > r ≥ 0.5, weak 0.5 > r ≥ 0.1. For the semi-quantitative analyses, the patterns were classified into the four groups and percentages were formulated based on the amount in each category. Moreover, association between sub-patterns in each category was quantified using Cramér's Φ. Strong association (Cohen, 1988).
Intra-and inter-observer reliability for the pattern analysis was determined and assessed by two examiners (A.P. and M.G.). A.P repeated the classification 4 months after the first classification. Both judges were blinded to each other's' measurements. Using Cronbach's α and a two-way, mixed intra-class correlation coefficient (ICC), a good result was set at ≥0.70, and ≥0.90 being excellent. For this study, Cronbach's α intra-rater reliability result was 0.92, and an interrater reliability ICC result of 0.90 was found.
In unilaterally dysfunctional joints, a significant size difference was observed between sexes (p < 0.01 sacra and ilia) ( Figure 3C) 3.3 | Unilateral SIJD yielded no size, sex or morphology-related mineralization differences to nondysfunctional joints nor bilateral SIJD Comparison of cases with unilateral dysfunction with their contralateral non-dysfunctional side yielded no significant size differences ( Figure 4A-C). Mineralization was indifferent when comparing sexes, sizes and morphology T1 and T2. No significant difference was found between bilateral and unilateral dysfunctional joints when comparing mean HU mineralization between sex, size and morphology T1 and T2 ( Figure 4A-C).
3.4 | Sacrum mineralization in females with SIJD is different when compared to the SIJ of healthy controls Significant difference in mineralization was found between controls and the dysfunctional joints from the SIJD cohort. These were found only in females (p < 0.01), small joints (p < 0.02), and T2 joints (p < 0.01) only on the sacral side ( Figure 5). Furthermore, correlations between mean surface mineralization and surface area reveal weak positive correlations on the iliac side in both males and females in the control cohort ( Figure 6A) and bilaterally dysfunctional joints ( Figure 6B).

| Semi-quantitative assessment of all sacroiliac joint surfaces revealed four main patterns for the purpose of semclassification
Two of these were marginal patterns: M1 had mineralization located around the anterior border, which could include the superior corner and/or apex and pattern M2 had mineralization scat-

Sacra Ilia
Type 1 Type 2 (A) (C) (B) F I G U R E 4 Box plot comparison between the unilaterally dysfunctional joints, bilaterally dysfunctional joints and the contralateral nondysfunctional side (no pain) from the SIJD cohort. (A) Size comparison, (B) sex comparison, and (C) morphology comparison between types 1 and 2 (too few values were available for an accurate type 3 comparison. The outlines of the boxes indicate the 25-and 75-percentile, the solid black horizontal line, the median. Whiskers indicate the minima and maxima. The dotted lines separate sub-groups (size, sex, and type). The thicker black line separates the sacra from the ilia values. Significance is considered as p < 0.05.
3.6 | Dysfunctional joints exhibit higher pattern conformity in corresponding sacra and ilia than those without pain A joint is "conforming" when the sacral and iliac articulating surfaces reflect the same pattern. For the dysfunctional joints, there was 46% conformity between patterns M1 to N2, and 6% with the M1 sub-patterns. Conformity was higher compared to the controls (M1 to N2: 26%; M1 sub-pattern added: 5%). The majority of dysfunctional joints was of T2 morphology (59%), and was equally distributed between small (49%) and large joints (51%). The majority of non-dysfunctional joints was also of T2 morphology (53%) and had smaller surface areas (63%) ( Table 1). From a different perspective, within each morphology type, T1 and T2 had a majority of dysfunctional joints (both >60%), with T3 having close to equal dysfunctional and non-dysfunctional joints.
Larger joints had the highest frequency of dysfunctional joints (72%).
Regarding the prominence of sub-patterns, M1C was more prominent in dysfunctional joints (39%), whilst pattern M1A was more frequent in the controls (31%) ( Table 2). When comparing ilia and sacra in dysfunctional joints, M1B sub-pattern was more common in the ilia (>55%) whilst M1C was most common in the sacra (>60%). Cramér's Φ revealed strong associations between patterns M1A and M1C in all dysfunctional joints and associated sub cohorts as well as in the controls (Table 2). Unilaterally and bilaterally dysfunctional joints were similarly associated with all three sub-patterns (Table 2).

| DISCUSSION
This study semi-quantitatively analyzed the mineralization patterns of the sacral and iliac auricular surfaces of the SIJ in SIJD-affected patients. It provides insights into the influence of surface size and shape in morpho-mechanical differences between individuals. The results presented here are likely representative of the long-term loading conditions of the SIJ, which demonstrate the biomechanical stresses applied to the joint when it suffers from dysfunction.

| Dysfunctional joints have more commonly larger SIJ surface areas
Joint pain was previously associated more with the T3 (acute angle) morphology. Authors speculate that a relative decrease in surface area F I G U R E 5 Box plot comparison between the control cohort, dysfunctional joints from SIJD cohort and contralateral nondysfunctional side (no pain) also from the SIJD cohort. The outlines of the boxes indicate the 25-and 75-percentile, the solid black horizontal line, the median. Whiskers indicate the minima and maxima. The dotted lines separate categories (females, small, and type 2). Significance is considered as p < 0.05, with brackets illustrating this.  available for dissipation was present in T3 when compared to the more obtuse T1 and T2 (more surface for force dissipation) (Jesse et al., 2017). Therefore, pain could be caused by increased force transmission to the ligaments and periarticular structures. T1 is more ovoid and might allow more transmission over the surface preventing excess forces from reaching the ligamentous structures, thus causing less pain (Jesse et al., 2017). Our study here however, does not produce conclusive results for dysfunctional joints with T3 morphology as there were too few to reliably assess the mineralization pattern.
When looking at the patterns corresponding to dysfunctional joints in the SIJD cohort, the majority of large surfaces were dysfunctional.
These findings suggest that larger joints are predisposed to develop pain, which may be due to the dysfunction of extracapsular joint structures, which take active part in pelvic load distribution (i.e., ligaments, cartilage, fat etc.) (Hammer et al., 2013). As there is more surface to dissipate loads in 'large joints' this would require less dissipation via extracapsular structures, which may result in a certain weakness of these compared to those in "smaller" joints. This may be why larger joints are predisposed to pain. In addition, larger bones and joints often belong to patients with greater height and males, which may then result in an increase of stresses to be dissipated within the SIJ with motion (Jesse et al., 2017;Nishi et al., 2019).
In addition, comparisons between the controls and the SIJD cohort revealed significant differences in total surface mineralization density in smaller joints and in females only on the sacral side. This may be caused by the sexual dimorphism where females retain more mobility in their joints for childbearing and, in consequence, have smaller joints for that purpose (Derry, 1912). Previous studies have demonstrated the potential for the sacrum to become more stressed in dysfunctional cases (Poilliot, Doyle, et al., 2021).

| Conformity may be an indicator of dysfunctional sacroiliac joints
When comparing the conformity of patterns in corresponding surfaces, this reflected higher conformity in dysfunctional joints than in the non-dysfunctional state. This was expected, as a previous study has shown that the inferior region of the SIJ shows similar mineralization density in the dysfunctional state (Poilliot, Doyle, et al., 2021).

| Dysfunctional joints reflect variable mineralization patterns compared to controls
Previous results show that morphology likely does not affect mineralization patterns at the SIJ in the 'healthy' state, however, size may be a factor. In fact, it was shown that larger joints reflected higher mineralization patterns than smaller joints only on the iliac side. It is hypothesized that this may be due to cartilage thinning on the iliac side, which may account for a mineralization "compensation mechanism" for the loss of chondral tissue. In this study, it was found that mean mineralization in the dysfunctional joint cohort was higher in females/ small joints on the sacral side when compared to the controls. As female joints tend to be smaller than male joints, they require more dampening of forces and are therefore more prone to higher densities across the surface. Furthermore, mean mineralization positive correlations with size were revealed in the controls and bilaterally dysfunctional joints on the iliac side in both males and females, and a negative correlation on the sacral side in females in the dysfunctional joint cohort, specifically in unilaterally dysfunctional joints. These results reflect previous results where larger joints reflect higher mineralization patterns than smaller joints only on the iliac side, as do bilaterally dysfunctional joints. This may reflect the similar walking mechanisms in the dysfunction-less state as the bilaterally dysfunctional state.
In the control cohort, when assessing the mineralization maxima zone differences using the sub-patterns, all sub-groups (albeit T3 sacra) showed a M1A (anterior surface) sub-pattern majority. This suggests that mineralization maxima zones are not affected by size, shape, sex nor bone. However, when assessing the SIJD joints both dysfunctional and non-dysfunctional sides, a variability in patterns was observed compared to the control cohort. The SIJD cohort revealed that morphology did not seem to affect the patterns as all sacra had a majority of pattern M1C and the ilia M1B. Respectively the dysfunctional and non-dysfunctional state did not play a role.
However, size may have had an effect with larger joints having a majority of anterior border mineralization (M1A) and smaller joints having a majority of apex mineralization (M1C). This again may reflect the need for larger joints to compensate for the extracapsular structures and show a broader subchondral mineralization across the surface than 'smaller' joints. Non-dysfunctional joints had variable patterns compared to the control cohort but these were similar to those found in the dysfunctional state. In fact, Cramér's Φ revealed significant associations between surface sub-patterns and a particular cohort where M1A and M1C were the most frequent sub-patterns in the dysfunctional joints, controls and non-dysfunctional joints. Bilaterally and unilaterally dysfunctional joints revealed close to equal frequencies of all three sub-patterns. This suggests that sub-patterns are likely not indicative of the dysfunctional state of the joint nor type of dysfunction (unilateral/bilateral). Therefore, our hypothesis stating that joints with dysfunction would reflect pattern differences than those of the non-dysfunctional group can only be partly accepted.
This similarity with the dysfunctional state suggests that the subchondral bone mineralization evolves in a similar way on both sides independent of pain, suggesting that pain is not directly related to bone mineralization density.
Regarding the limitations of the study, the manual calculation of the posterior angle was performed by only one investigator. T3 morphology had too few values to reliably compare with the other two morphologies so was often excluded from comparisons. Anatomical variants were not accounted for and may have had some influence on joint area (Ziegeler et al., 2021). The study does not account for population difference and potential variables between the two cohorts compared: controls and SIJD patients. The assessment of patterns remains qualitative, although agreement between three authors was sought for the final classification.

| CONCLUSIONS
Load distribution related to auricular surface morphological differences seems to have little impact on pain-related subchondral bone adaptation in cases of sacroiliac joint dysfunction. Larger joints may be predisposed to develop dysfunction due to the weakening of the extracapsular structures directly related to the surface size. Painful joints reflect common conformity patterns of sacral apex mineralization with corresponding superior corner iliac mineralization, which is a sign of abnormal load transfer within the joint.