Supplementary Information: Multimodal imaging reveals multiscale mechanical interplay in

vertebral endplate microarchitecture during intervertebral disc loading

Supplementary Results

Full statistical analysis for the results presented in the heatmaps in Main Fig. 3 and 4 are given in
Table S1. Regression graphs for statistically significant correlations are given in Figs. 1 – 4, generated
using the rmcorr shiny app (Bakdash and Marusich, 2017). Full statistical analysis for the results
presented in the heatmap in Main Fig. 6 are given in Table S3, and regression graphs for statistically
significant results given in Figs. 6-8.
Table S1 – Statistical analysis used to generate the heatmaps in Main Fig. 3 and Fig. 4.

Parameters Repeated 95 % p-value Rank Alpha Level

Measures Confidence (Holm-
Correlation Interval Bonferroni)
Coefficient (0.05/(24-Rank))
Location MV/TV 0.76 0.678, 0.817 1.46e-29 1 2.17e-3
Shear TMD -0.56 -0.662, -0.443 4.11e-14 2 2.27e-3
Compression TMD -0.56 -0.638, -0.409 5.99e-14 3 2.38e-3
Tension TMD -0.53 -0.659, -0.439 1.25e-12 4 2.50e-3
Location DA 0.52 0.393, 0.626 6.11e-12 5 2.63e-3
Location Compression -0.48 -0.589, -0.342 5.50e-10 6 2.78e-3
Compression MV/TV -0.43 -0.547, -0.286 4.19e-8 7 2.94e-3
Location Shear -0.41 -0.534, -0.268 1.47e-7 8 3.13e-3
Tension ST.Th 0.36 0.207, 0.486 7.02e-6 9 3.33e-3
Shear MV/TV -0.34 -0.472, -0.19 1.87e-5 10 3.57e-3
Location Tension -0.29 -0.426, -0.135 3.18e-4 11 3.85e-3
Location Con/MV -0.28 -0.422, -0.129 4.10e-4 12 4.17e-3
Shear ST.Th 0.26 0.099, 0.397 2.00e-3 13 4.55e-3
Location ST.Th 0.23 0.071, 0.373 5.00e-3 14 5.00e-3
Tension MV/TV -0.19 -0.34, -0.034 0.018 15 5.56e-3
Compression ST.Th 0.17 0.01, 0.318 0.038 16 6.25e-3
Location TMD 0.16 0.002, 0.312 0.047 17 7.14e-3
Compression Con/MV 0.13 -0.026, 0.286 0.100 18 8.33e-3
Shear Con/MV 0.10 -0.063, 0.252 0.200 19 1.00e-2
Tension DA 0.10 -0.061, 0.253 0.226 20 1.25e-2
Tension Con/MV 0.04 -0.12, 0.197 0.600 21 1.67e-2
Compression DA -0.03 -0.186, 0.131 0.700 22 2.50e-2
Shear DA 0.03 -0.134, 0.183 0.700 23 0.05

1
Figure S1 - Repeated measures correlation regression graphs for a) compression (3rd Principal Strain), b) shear (Maximum
Shear Strain), c) Tension (1st Principal Strain) and d) Con/MV (Connectivity to mineralised tissue volume ratio) vs distance
from the endplate centre (Location).

2
Figure S2 - Repeated measures correlation regression graphs for a) ST.Th (Septal/ Trabecular Thickness), b) degree of
anisotropy, and c) MV/TV (Mineralised Tissue Volume Fraction) vs distance from the endplate centre (Location).

3
Figure S3 – Repeated measures correlation regression graphs for a) Tension (1st Principal Strain) vs TMD (greyscale value)
b) compression (3rd Principal Strain) vs TMD (greyscale value), c) shear (Maximum Shear Strain) vs TMD (greyscale value),
c) Tension (1st Principal Strain) and d) compression (3rd Principal Strain) vs MV/TV (Mineralised Tissue Volume Fraction).

4
Figure S4 - Repeated measures correlation regression graphs for a) shear (Maximum Shear Strain) vs MV/TV (Mineralised
Tissue Volume Fraction) b) tension (1st Principal Strain) vs ST.Th (septal/ trabecular thickness), c) shear (Maximum Shear
Strain) vs ST.Th (septal/ trabecular thickness).

5
Figure S5 - 3rd principal strain for the 4 regions shown in Main Fig. 4b.

Table S2 - Microstructure and strain values used to create the radar plots in Figure 4d.

Tens. Comp. Shear BMD MV/TV ST.Th (µm) DA Con/MV

(%) (%) (%) (greyscale
value)
Caudal 1.52 1.69 1.60 44021 0.259 13.7 0.565 3.41e-4
anterior
outer
Caudal 1.45 1.60 1.53 46624 0.314 13.3 0.261 6.06e-4
lateral
middle
Cranial 2.04 2.34 2.08 43630 0.426 38.7 0.312 4.27e-5
anterior
outer
Cranial 1.90 2.51 2.10 41241 0.034 12.6 0.034 4.63e-4
posterior
inner

Table S3 – Statistical analysis used to generate the heatmap in Main Fig.6.

Parameters Repeated 95 % p-value Rank Alpha Level

Measures Confidence (Holm-
Correlation Interval Bonferroni)
Coefficient (0.05/(19-Rank))
(002) d-period Width 0.44 0.317, 0.549 2.88e-10 1 0.0028
(002) d-period TMD 0.35 0.221, 0.473 6.80e-7 2 0.0029
(002) d-period Length 0.34 0.207, 0.461 1.90e-6 3 0.0031
Width TMD 0.33 0.194, 0.45 4.58e-6 4 0.0033
Width Con/MV 0.29 0.148, 0.412 7.44e-5 5 0.0036
Length TMD 0.27 0.135, 0.401 1.53e-4 6 0.0038
Width ST.Th -0.24 -0.373, -0.102 8.38e-4 7 0.0042
(002) d-period ST.Th -0.24 -0.368, -0.097 0.001 8 0.0045
(310) d-period TMD 0.24 0.097, 0.368 0.001 9 0.005
(002) d-period Con/MV 0.23 0.09, 0.362 0.002 10 0.0056
(002) d-period MV/TV -0.21 -0.34, -0.066 0.004 11 0.0063
Length MV/TV 0.18 0.035, 0.313 0.015 12 0.0071

6
Width MV/TV -0.18 -0.312, -0.034 0.015 13 0.0083
(310) d-period ST.Th 0.06 -0.087, 0.199 0.4 14 0.01
(310) d-period Con/MV 0.04 -0.099, 0.187 0.5 15 0.013
(310) d-period MV/TV 0.01 -0.13, 0.157 0.8 16 0.017
Length Con/MV 0.01 -0.138, 0.149 0.9 17 0.025
Length ST.Th 0 -0.147, 0.14 1 18 0.05

Figure S6 - Repeated measures correlation regression graphs for greyscale value (TMD) vs a) (002) d-period b) crystallite
length, c) (310) d-period, and d) crystallite width.

7
Figure S7 - Repeated measures correlation regression graphs for a) ST.Th (septal/ trabecular thickness), b) MV/TV
(Mineralised Tissue Volume Fraction), and c) Con.MV (connectivity to mineralised tissue volume ratio) vs (002) d-period.

8
Figure S8 - Repeated measures correlation regression graphs for a) ST.Th (septal/ trabecular thickness) and b) Con.MV
(connectivity to mineralised tissue volume ratio) vs crystallite width.

Table S4 – d-period, crystallite size, and microstructure values used to create the radar plots in Figure 6c.

(002) (310) Length Width (Å) TMD MV/TV ST.Th Con/MV

d- d- (Å) (greyscale (µm)
period period value)
(Å) (Å)
17565 ROI 3.433 2.275 117 11.7 37622 0.27 49 1.07e-5
10 (i)
17565 ROI 3.442 2.275 125 12.5 37815 0.12 26 1.88e-5
15 (ii)
17565 ROI 3.436 2.275 133 12.3 37579 0.38 65 3.57e-6
19 (iii)
17565 ROI 3.439 2.271 107 10.6 38150 0.25 37 1.75e-5
30 (iv)

9
Supplementary Methods
Table S5 - Sample information for whole IVD used for in situ sCT and DVC experiments.

Sample 3 5 7 8 10 12 13 14
ID:
Spinal L1 – L2 L2 – L3 L4 – L5 T13 – L1 T10 -T11 L1 – L2 L2 – L3 L3 – L4
Level:
Spine: 1 2 2 3 3 3 3 3

Table S6 - Sample information for correlative sCT and WAXD experiments.

Sample 5 12 13 16 17 20 21 19
ID:
1st Scan 17541 17557 17561 17565 17570 17593 17598 17583
Number:
Spinal L3 L3 L1 L3 L3 L3 L1 L1
Level:
Anterior/ Posterior Posterior Posterior Anterior Posterior Anterior Anterior Anterior
Posterior:
Cranial/ Caudal Cranial Cranial Caudal Cranial Caudal Cranial Cranial
Caudal:
Spine ID: 1 6 6 13 13 7 7 13

Figure S9 - Annotated photograph of the experimental set-up at I13-2.

10
The impact of carbon pillars on X-ray transmission

A custom open frame was designed for the Deben CT500 mechanical testing rig used for this
experiment, allowing for easy sample mounting, and maintaining hydration. The open frame pillars
used for the experiment are carbon rods with a diameter of 6 mm. The pillars appear in
approximately 18.5 % of the projections taken for each tomogram. The pillars reduce X-ray
transmission, causing a reduction in intensity at the detector. When one pillar covers the whole field
of view it results in a loss of image intensity of approximately 22 %, calculated from the drop in peak
frequency intensity shown in Fig. 10.

Figure S10 - ai) greyscale intensity projection image and ii) histogram of the projection intensity of a sample with no carbon
pillar in the field of view, showing a peak frequency intensity of 11800; bi) projection image and ii) intensity histogram of a
projection where one carbon pillar fully covers the field of view, showing a peak frequency intensity of 9200.

11
Radiation dose calculation

Radiation dose for each scan was calculated using the method described in Disney et al 2023. Dose
was estimated as
𝐽
𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒 𝑡𝑖𝑚𝑒 (𝑠) × 𝐴𝑏𝑠𝑜𝑟𝑏𝑒𝑑 𝑒𝑛𝑒𝑟𝑔𝑦 𝑓𝑙𝑢𝑥 ( 2 )
𝐷𝑜𝑠𝑒 (𝐺𝑦) = 𝑠𝑚
𝑘𝑔
𝑀𝑎𝑠𝑠 𝑝𝑒𝑟 𝑢𝑛𝑖𝑡 𝑎𝑟𝑒𝑎 ( 2 )
𝑚
Total exposure time is given by the number of projections multiplied by 0.1 s exposure time
(1801*0.1 = 180.1 s). Using an absorbed energy flux of 7.4x10-4, mass per unit area of 1 gcm3, and
thickness of 5 mm (Disney et al., 2023), total dose for one scan was calculated as 26.6 kG.

Figure S11 - a) displacement accuracy precision for components in x (u), y (v), and z (w) directions; b) accuracy and precision
for each strain component; c)resultant xy and d) yz images from subtraction of the second zero strain image generated
from the double projection method from the first, showing the noise between images.

12
Figure S12 – Photograph of experimental set-up at DIAD beamline.

13