Pregnancy-related transverse arch deformation: a subject-specific finite element analysis of the contributions of body weight and tissue stiffness Authors: Xin Li, Zhenghui Lu, Yufei Fang, Gusztáv Fekete, Dong Sun, Yaodong Gu Published: 2026 ABSTRACT BackgroundPregnancy is accompanied by physiological changes that alter foot structure and loading conditions, which may contribute to foot pain, instability, and an increased risk of falls. Gestational weight gain and pregnancy-related changes in passive soft-tissue compliance occur simultaneously, making it difficult for clinical studies to distinguish their individual contributions to transverse arch deformation. The purpose of this study was to independently quantify the effects of increased body weight and reduced soft tissue stiffness on transverse foot arch geometry during pregnancy and to clarify their potential mechanical relevance.MethodsA computed tomography (CT) -based subject-specific finite element (FE) model and a full factorial design were used to independently quantify the effects of body weight (W) and tissue stiffness (K) on the transverse arch. The reference-condition FE prediction was benchmarked against in-shoe plantar pressure measurements from the same participant, showing close agreement for forefoot and heel peak pressures (< 3% difference). A 4 × 4 factorial analysis was then performed across sixteen conditions, including the reference condition and the 1st, 2nd, and 3rd trimester pregnancy states.ResultsUnder the prescribed W–K schedule, the greatest model-predicted flattening occurred in the second trimester, corresponding to the stage with the largest prescribed reduction in tissue stiffness. Regression analysis showed that increased load was the primary driver of transverse widening of the midfoot, whereas modelled reductions in passive tissue stiffness predominantly governed loss of arch height and curvature. Interaction effects between load and tissue stiffness were minimal, indicating largely independent mechanical pathways. Model-predicted stresses in the medial cuneiform, intermediate cuneiform, and cuboid were highest in the second trimester.ConclusionsWithin this single-subject static FE framework, gestational weight gain (W) and modelled reductions in passive tissue stiffness (K) influenced the tarsal transverse arch through distinct mechanical pathways. Increased W was more strongly associated with transverse widening, whereas lower modelled K was more closely related to vertical flattening and loss of arch curvature. Among the simulated conditions, the most pronounced model-predicted flattening occurred in the second trimester, consistent with the prescribed second-trimester minimum in K. These findings provide a mechanical interpretation of pregnancy-related transverse arch deformation and may help guide future studies of supportive exercise and footwear strategies.Trial registrationNot applicable.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12884-026-09438-x. Background Pregnancy is associated with physiological adaptations that alter mechanical loading and tissue properties, which may compromise biomechanical stability and increase fall risk. Falls have been reported in up to 25% of pregnant women [1], highlighting the importance of foot mechanics in maintaining postural control. As the primary load-bearing interface with the ground, the foot must accommodate these pregnancy-related changes [2, 3]. While deformation of the medial longitudinal arch during pregnancy has been extensively documented, the tarsal transverse arch (TTA) remains comparatively understudied [3–7]. The TTA contributes to mediolateral stability by increasing midfoot stiffness through a mechanical locking mechanism, implying that disruption of its geometry may impair balance control and elevate fall risk [8, 9]. However, while the broad spectrum of pregnancy-related foot alterations, including medial arch collapse, hindfoot eversion, and dimensional enlargement, is extensively documented [2, 4], the primary mechanical determinant remains a subject of controversy. Previous studies have reported pregnancy-related increases in peripheral joint laxity and have proposed that hormonal changes, particularly relaxin-related effects, may contribute to altered connective-tissue compliance [10, 11]. However, although peripheral joint laxity has been reported to increase during pregnancy, several investigations have failed to demonstrate a clear association between circulating relaxin levels and peripheral joint laxity [12, 13]. Body mass accumulation may also contribute to pregnancy-related foot structural changes [14, 15], and BMI has been associated with plantar fascia thickness in healthy adults [16]. Mechanically, the TTA functions as a rigid beam utilizing a geometric locking mechanism to resist deformation and maintain mediolateral stability [17]. During pregnancy, however, this structural integrity may be influenced by concurrent mechanical drivers: increased body mass (W) amplifies vertical bending moments, whereas pregnancy-related reductions in passive soft-tissue stiffness (K) may diminish ligamentous restraint. Although experimental evidence suggests that foot structures may be sensitive to both increased loading and pregnancy-related connective-tissue adaptations [18–20], the exact in vivo interplay of these factors remains unclear. Consequently, as clinical assessments inherently capture only the aggregated effect of these simultaneous adaptations, it remains unresolved whether TTA deformation is primarily driven by the magnitude of load, the extent of tissue laxity, or their interaction. This mechanistic ambiguity is further exacerbated by limitations of conventional clinical assessments, as pregnancy-related soft tissue swelling and adipose accumulation reduce the reliability of surface-based measurements of arch height and curvature [21, 22]. To overcome these limitations, the present study employs finite element (FE) analysis to independently examine the effects of load and tissue stiffness on TTA mechanics. A subject-specific foot model was used in a full factorial design in which W and K were systematically varied within physiologically relevant ranges. This framework allows the relative geometric and mechanical effects of increased loading and ligamentous laxity to be examined separately within a controlled FE setting, while also assessing their interaction on midfoot stress distribution. We hypothesized that although both factors contribute to pregnancy-related TTA deformation, increased load and reduced tissue stiffness would be associated with different mechanical responses within the FE framework. Methods Finite element modelling To quantitatively assess the effects of pregnancy-related weight gain and soft tissue laxity on the mechanical behaviour of the midfoot and transverse arch, a FE model of the participant’s dominant right foot was constructed from computed tomography (CT) data of a healthy 25-year-old female volunteer (height 163 cm and weight 57.3 kg). The participant had a normal medial longitudinal arch and no history of musculoskeletal disorders. CT scanning was performed with the subject in the supine position and the ankle and knee in neutral alignment, yielding 1,079 axial slices with a slice thickness of 0.5 mm and a reconstruction interval of 0.25 mm. Biplanar standing X-ray images were then used for 2D–3D registration to adjust the bone alignment before FE analysis [23]. The posture-corrected model was used as the initial configuration for the standing simulations (Fig. 1A). All study procedures were approved by the Medical Ethics Committee of Ningbo University (TY2025070) and conducted in accordance with the Declaration of Helsinki; written informed consent was obtained from the participant.Fig. 1Workflow for finite element (FE) model development and geometrical definition of the tarsal transverse arch. A CT segmentation, 3D reconstruction, biplanar standing X-ray registration, and FE model construction. B Generation of the full skeletal model, including mesh construction and assembly of bone components. C Finite element model incorporating bones, ligaments, cartilage layers, and the plantar fascia. D Geometrical definition of the tarsal transverse arch and calculation of structural indicators. The arch is defined using anatomical landmarks on the medial (C1), intermediate (C2), and lateral (C3) cuneiforms and the cuboid (Cu). Chord length L spans the cuneiforms, effective rise height \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${S}_{\mathrm{e}\mathrm{f}\mathrm{f}}$$\end{document} is measured orthogonally from the chord to the apex, and curvature is derived from the radius R fitted to the osseous arc Workflow for finite element (FE) model development and geometrical definition of the tarsal transverse arch. A CT segmentation, 3D reconstruction, biplanar standing X-ray registration, and FE model construction. B Generation of the full skeletal model, including mesh construction and assembly of bone components. C Finite element model incorporating bones, ligaments, cartilage layers, and the plantar fascia. D Geometrical definition of the tarsal transverse arch and calculation of structural indicators. The arch is defined using anatomical landmarks on the medial (C1), intermediate (C2), and lateral (C3) cuneiforms and the cuboid (Cu). Chord length L spans the cuneiforms, effective rise height \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${S}_{\mathrm{e}\mathrm{f}\mathrm{f}}$$\end{document} is measured orthogonally from the chord to the apex, and curvature is derived from the radius R fitted to the osseous arc Bone and soft tissues were segmented in Mimics 21.0 (Materialise, Leuven, Belgium) using a grey-value thresholding approach. Surface smoothing was performed in Geomagic Studio 2014 (Geomagic Inc. NC, USA), and three-dimensional (3D) assembly was completed in SolidWorks 2021 (Dassault Systèmes, France) (Fig. 1B). The model comprised 28 foot bones and 51 major ligaments, which were defined based on anatomical literature [24–26] and subsequently refined by adjusting ligament attachment sites to the subject-specific CT geometry (Fig. 1C). Articular cartilage and joint capsules were represented as 2-mm solid layers interposed between adjacent bony surfaces. The plantar fascia was modelled using truss elements extending from the medial process of the calcaneal tuberosity to the proximal phalanges. FE meshing was performed in ANSYS Workbench (ANSYS Inc., PA, USA) using four-node tetrahedral elements for all structures. Mesh convergence was assessed using six progressively refined meshes (from coarse to fine), generated by reducing the global element size by approximately 10% at each refinement level. For each mesh, the peak plantar contact pressure (overall/heel/forefoot) and the displacements of representative midfoot bones (the three cuneiforms and the cuboid) were compared. Convergence was considered satisfactory when the change between two consecutive mesh levels was < 5% for these key outputs, and the converged mesh was adopted for all subsequent simulations (Supplementary Text S1 and Figure S1). Unless otherwise specified, all tissues were modelled as homogeneous, isotropic and linearly elastic. Cortical bone was assigned a Young’s modulus (E) of 17,000 MPa and a Poisson’s ratio (ν) of 0.3, consistent with its predominantly cortical composition [27, 28]. Articular cartilage was defined with E = 1 MPa and ν = 0.4 [29]. Encapsulated soft tissue was prescribed a low stiffness (E = 0.15 MPa) and a high Poisson’s ratio (ν = 0.45) in order to approximate its compliant and nearly incompressible behaviour [30]. Ligaments were represented using two node truss elements that could bear tension but not compression, which prevents non-physiological resistance in the compressive direction. This ligament modelling scheme and the associated cross-sectional area range were adopted from a previously established foot–ankle FE modelling framework and are consistent with our previous foot–ankle FE models [23, 24]. Nine primary ligaments that mainly sustain load were modelled with hyperelastic material properties. These ligaments included the anterior talofibular, anterior tibiofibular, calcaneofibular, posterior talofibular, tibionavicular, spring, tibiocalcaneal, anterior tibiotalar and posterior tibiotalar ligaments. Their cross sectional areas ranged from 1.5 to 5.91 mm2, based on anatomical measurements [31]. Secondary ligaments that mainly contribute to stability, including the cervical ligament, the talonavicular ligament, the interosseous membrane and the posterior interosseous ligament, were modelled as linearly elastic truss elements with E = 264.8 MPa and a cross sectional area of 1.784 mm2 [27, 28]. All ligament elements were assigned a uniform density of 1,000 kg/m3. The plantar fascia was represented using hyperelastic truss elements that could carry tension but not compression, spanning from the medial calcaneal tubercle to the proximal phalanges. Its stiffness parameters were defined according to the range reported in previous studies (0–700 MPa) [27, 28, 32]. The plantar surface interacted with a rigid floor, which was defined with E = 14,200 MPa, ν = 0.1. A complete summary of all material models and parameters is provided in Table 1.Table 1Material properties used in the FE modelComponentElement TypeYoung’s modulus (MPa)Density (kg/m3)Poisson’s ratio (ν)Element Size (mm)Bone3D tetrahedral1700019900.33.24Cartilage3D tetrahedral1N/A0.40.81Soft tissue3D tetrahedral0.159500.453.24Floor3D tetrahedral1420010000.18.10Primary ligamentsTruss (tension only)Hyperelastic; Area: 1.5–5.91 mm21000N/AN/AOther ligamentsTruss (tension only)Linear elastic (E = 264.8 MPa); Area: 1.784 mm21000N/AN/APlantar fasciaTruss (tension only)Hyperelastic (0–700 MPa); Area: 290.7 mm21000N/AN/A Table 1Material properties used in the FE modelComponentElement TypeYoung’s modulus (MPa)Density (kg/m3)Poisson’s ratio (ν)Element Size (mm)Bone3D tetrahedral1700019900.33.24Cartilage3D tetrahedral1N/A0.40.81Soft tissue3D tetrahedral0.159500.453.24Floor3D tetrahedral1420010000.18.10Primary ligamentsTruss (tension only)Hyperelastic; Area: 1.5–5.91 mm21000N/AN/AOther ligamentsTruss (tension only)Linear elastic (E = 264.8 MPa); Area: 1.784 mm21000N/AN/APlantar fasciaTruss (tension only)Hyperelastic (0–700 MPa); Area: 290.7 mm21000N/AN/A Material properties used in the FE model Simulation conditions Four physiological conditions were defined on a single foot FE geometry using literature-informed trajectories of body mass and soft tissue mechanics from the nonpregnant state to late pregnancy. These conditions represented the nonpregnant reference condition and gestational ages of 9.1, 20.9, and 32.3 weeks, and shared identical bone morphology and mesh discretisation. Pregnancy-related biomechanical variation was represented by two model inputs: the single-foot vertical load (W) and a prescribed passive soft-tissue stiffness-scaling parameter (K), which was applied to the Young’s modulus of the plantar ligaments and plantar fascia. The nonpregnant reference condition was based on the measured body mass of the modelled participant, 57.3 kg, corresponding to a single-foot vertical load of 281 N under an idealized symmetric bilateral stance assumption. The pregnancy-stage body masses were not direct longitudinal measurements from this participant, but were derived from the trimester-specific BMI trajectory reported by Cherni et al. [18]. Cherni et al. reported mean BMI values of approximately 24, 25, and 27 kg/m2 in the first, second, and third trimesters, respectively. In the present study, this BMI trajectory was used as a literature-based reference to define pregnancy-stage body-mass inputs and the corresponding single-foot vertical loads under the same idealized symmetric bilateral stance assumption. This resulted in prescribed pregnancy-stage body masses of 65.3, 68.1, and 73.5 kg, corresponding to single-foot vertical loads of 320, 334, and 361 N, respectively. This load was applied vertically to the superior aspect of the talus to represent gravitational input during standing. To represent pregnancy-related increases in passive soft-tissue compliance, the Young’s modulus of the plantar ligaments and plantar fascia was scaled by 5.3%, 15.2%, and 14.4% across the three gestational stages [11, 33]. These reductions were implemented as relative stiffness-scaling factors rather than as direct measurements of plantar-tissue material properties. In this study, the percentage reductions in K were derived from published data describing changes in peripheral joint laxity during pregnancy. Although these data were not obtained from direct mechanical testing of the plantar ligaments or plantar fascia, they provide a biologically informed approximation for representing changes in the compliance of passive connective tissues in distal peripheral joints during pregnancy. Because longitudinal, pregnancy-stage-specific material property data for the plantar ligaments and plantar fascia are currently unavailable for direct finite element implementation, this scaling approach was adopted as a modelling approximation. The same relative K scaling factor was applied uniformly to all plantar ligament and plantar fascia truss elements at each gestational stage. W and K were each divided into four levels within the prescribed W–K parameter space, producing sixteen full-factorial simulation conditions. The reference condition and the three pregnancy-stage conditions corresponded to four physiologically defined points within this parameter space. To simulate static bipedal standing, the ground was modeled as a rigid support with its base fully fixed to constrain all translational and rotational degrees of freedom (Fig. 2B). A frictional contact interface was defined between the plantar surface and the ground, with a coefficient of friction of 0.5 to represent foot–ground interaction during standing [23]. A vertical downward concentrated load was applied to the superior surface of the talus to represent 50% of body weight under the same idealized symmetric bilateral stance assumption. Bilateral load sharing was not directly measured in this study. The Achilles tendon action was represented by applying a vertical upward traction over the posterosuperior calcaneal insertion area, with a magnitude equal to one half of the applied downward load, reflecting the contribution of the triceps surae–Achilles tendon complex to static force equilibrium [34]. For internal interactions, bonded contact was assigned between the bones and the surrounding soft tissue, as well as between ligament insertion sites and the corresponding bony surfaces, to prevent relative slip. Frictionless contact was defined between adjacent articular surfaces to permit relative joint motion.Fig. 2Reference-condition plantar-pressure agreement and simulated plantar contact pressure distributions. A Pedar measurements and finite element predicted peak pressures in the forefoot and heel regions in the nonpregnant reference condition. B Finite element predicted plantar contact pressure for the reference condition and for 9.1, 20.9, and 32.3 weeks of gestation. All 4 panels in (B) were generated from the same subject-specific plantar surface and display region; the silhouettes are shown only as a common visual reference for pressure distribution and do not represent condition-specific plantar-surface deformation Reference-condition plantar-pressure agreement and simulated plantar contact pressure distributions. A Pedar measurements and finite element predicted peak pressures in the forefoot and heel regions in the nonpregnant reference condition. B Finite element predicted plantar contact pressure for the reference condition and for 9.1, 20.9, and 32.3 weeks of gestation. All 4 panels in (B) were generated from the same subject-specific plantar surface and display region; the silhouettes are shown only as a common visual reference for pressure distribution and do not represent condition-specific plantar-surface deformation Key geometrical and mechanical indicators of the transverse arch were extracted from the FE results to characterise its responses under different W–K combinations. Geometrical indicators were computed according to the structural definition of the transverse arch (Fig. 1D) [35], including chord length, effective rise height, curvature, and the spatial position of the arch apex. In this study, the transverse arch was defined at the tarsal level using the three cuneiforms and the cuboid, reflecting the focus on midfoot/tarsal deformation. Mechanical outputs included plantar contact pressure (kPa) at the plantar soft tissue–ground interface, the forefoot-to-heel peak contact pressure ratio, and peak von Mises stress within the transverse arch bones. Plantar contact pressure was extracted as the normal contact stress at the plantar soft tissue–ground interface and reported in kPa, representing the reaction force per unit contact area generated by foot–ground interaction. The forefoot-to-heel peak contact pressure ratio was calculated as a derived regional index to characterise relative anterior–posterior load sharing. Transverse arch collapse was characterised primarily using structural descriptors, including chord length, effective rise height, curvature, and apex displacement. Peak von Mises stress was used as a model-predicted indicator of regional osseous stress concentration associated with transverse arch deformation. All simulations used identical geometry, material definitions, and boundary conditions, ensuring that differences in model response were attributable solely to variations in W and K [36]. A standardised linear regression model was then constructed with all transverse arch indicators as dependent variables to quantify the relative contributions of increased loading and soft tissue relaxation. Regression results were presented as forest plots to illustrate the direction and magnitude of effects associated with each predictor. All statistical analyses were conducted in R and Python, with a significance level of α = 0.05. Reference-condition plantar-pressure benchmarking To provide a reference-condition external check for the FE model, the forefoot and heel peak plantar contact pressures predicted under the reference condition were compared with Pedar in-shoe pressure data obtained from the same participant’s dominant right foot during static standing [23, 24]. Agreement was quantified by calculating the relative differences in peak contact pressure within the forefoot and heel regions. Regions were defined by foot-length percent masks (heel: proximal 30%; forefoot: distal 40%), and peak contact pressure was extracted within each region. Bilateral asymmetry was not assessed in this study [37]. No independent plantar-pressure measurements were available for the pregnancy-stage conditions. Therefore, the pregnancy-related predictions inherit the confidence level and limitations of this single-condition external check. Results Reference-condition plantar-pressure agreement Under the nonpregnant reference condition, the model predicted peak contact pressures of 93.71 kPa in the forefoot and 103.12 kPa in the heel. The corresponding peak pressures measured with Pedar were 95.13 kPa and 106.24 kPa, which gave relative errors of 1.49% and 2.94%, respectively (Fig. 2A). These differences were below 5%, indicating close agreement between FE-predicted and Pedar-measured regional peak plantar pressures under the nonpregnant reference static standing condition. This comparison was limited to the reference condition and should not be interpreted as independent validation of the pregnancy-stage simulations. FE-simulated plantar contact pressure During static standing, peak plantar contact pressure increased overall across the four simulation conditions from the nonpregnant reference condition to late pregnancy (Fig. 2B). Forefoot peak plantar contact pressure increased from 93.71 kPa at reference condition to 106.98 kPa in the first trimester, 109.82 kPa in the second trimester, and 125.18 kPa in the third trimester. Heel peak plantar contact pressure increased from 103.12 kPa at reference condition to 119.23 kPa in the first trimester, decreased to 106.31 kPa in the second trimester, and then increased to 139.56 kPa in the third trimester. Accordingly, the forefoot-to-heel peak pressure ratio was 0.91 at reference condition, 0.90 in the first trimester, 1.03 in the second trimester, and 0.90 in the third trimester, indicating a transient anterior redistribution of plantar contact pressure during the second trimester. The complete results for all sixteen W–K combinations are provided in Supplementary Table S1. To further examine this pattern, all 16 W–K combinations in the 4 × 4 full factorial design were additionally post-processed. Across these combinations, increasing load elevated peak plantar contact pressure in both the forefoot and heel, whereas modelled reductions in passive tissue stiffness were associated with a relatively greater reduction in heel peak pressure, thereby increasing the forefoot-to-heel peak pressure ratio. Standardized linear regression analysis further showed that forefoot peak plantar contact pressure was primarily associated with W, whereas heel peak plantar contact pressure was influenced by both W and K. In contrast, the forefoot-to-heel peak pressure ratio was significantly associated with K, whereas W and the interaction term were not significant. Detailed plantar contact pressure results for all 16 W–K combinations and the corresponding regression coefficients are provided in Supplementary Tables S1 and S2. Peak von mises stress in the transverse arch bones In the medial cuneiform (C1), peak stress increased from 1.36 MPa under the reference condition to 2.78 MPa in the second trimester, before decreasing to 1.77 MPa in the third trimester. The intermediate cuneiform (C2) exhibited a similar pattern, increasing from 0.94 MPa to 1.59 MPa and showing a slight reduction in the third trimester. The lateral cuneiform (C3) maintained relatively low stress levels, with limited variation across conditions (0.41–0.58 MPa). Unlike C1, C2, and Cu, C3 reached its maximum stress in the third trimester, although the absolute stress level remained low. In contrast, the cuboid (Cu) showed a more pronounced change, with peak stress increasing from 1.23 MPa at reference condition to 2.48 MPa in the second trimester and declining to 1.61 MPa in the third trimester (Fig. 3A). The FE-predicted stress maps showed a more concentrated stress distribution in the second trimester than in the other simulated conditions (Fig. 3B).Fig. 3Model-predicted von Mises stress in the transverse arch region across the reference condition and the three pregnancy stages. A Peak von Mises stress at the medial cuneiform (C1), intermediate cuneiform (C2), lateral cuneiform (C3), and cuboid (Cu) across reference condition and the three pregnancy stages. B FE-predicted transverse-arch stress distributions across pregnancy. All four stress maps in panel B are displayed using the same color scale and identical numerical limits Model-predicted von Mises stress in the transverse arch region across the reference condition and the three pregnancy stages. A Peak von Mises stress at the medial cuneiform (C1), intermediate cuneiform (C2), lateral cuneiform (C3), and cuboid (Cu) across reference condition and the three pregnancy stages. B FE-predicted transverse-arch stress distributions across pregnancy. All four stress maps in panel B are displayed using the same color scale and identical numerical limits Geometrical changes of the transverse arch during pregnancy The transverse arch geometry exhibited systematic, non-linear deformation patterns throughout the simulated gestational period. While the overall arch profile demonstrated progressive inferior displacement, the underlying geometric responses diverged temporally. This decoupling of deformation modes is illustrated by the spatial trajectory of the arch apex (C2) (Fig. 4A). Mediolateral displacement (Δx) peaked in the second trimester (+ 0.19 mm) before returning toward reference condition. In contrast, vertical displacement (Δy) descended progressively throughout gestation, reaching its greatest absolute magnitude (− 1.30 mm) in the third trimester.Fig. 4Geometric Evolution of the Tarsal Transverse Arch Across Pregnancy. A Spatial trajectory of the arch apex relative to reference condition, illustrating stage-specific mediolateral and vertical displacement. B Dual-axis plot showing changes in chord length (L) and sagitta height (s) across gestation. C Evolution of transverse arch profiles across the four simulated stages. Deformation vectors (magnified 5 ×) highlight the stage-specific directions of landmark displacement relative to the reference profile Geometric Evolution of the Tarsal Transverse Arch Across Pregnancy. A Spatial trajectory of the arch apex relative to reference condition, illustrating stage-specific mediolateral and vertical displacement. B Dual-axis plot showing changes in chord length (L) and sagitta height (s) across gestation. C Evolution of transverse arch profiles across the four simulated stages. Deformation vectors (magnified 5 ×) highlight the stage-specific directions of landmark displacement relative to the reference profile This structural behavior is further quantified (Fig. 4B), which shows the relationship between arch width and height. Chord length (L) increased monotonically from 32.69 mm at reference condition to 32.76 mm in the third trimester, consistent with progressive transverse splay under increasing vertical loading. In contrast, sagitta height (s) decreased from reference condition to the second trimester and then increased slightly in the third trimester, with its minimum value occurring in the second trimester (approximately 10.00 mm). Consistent with the reduction in sagitta height, the normalized curvature (k) of the transverse arch decreased to 99.74% in the first trimester, reached its minimum value of 99.06% in the second trimester, and partially recovered to 99.51% in the third trimester. These findings indicate that, under the prescribed W–K schedule, the most pronounced geometric flattening occurred in the second trimester rather than at the point of peak load. This response corresponded to the stage at which the prescribed reduction in K was greatest, and therefore should be interpreted as a stiffness-linked model response rather than as an entirely independent temporal prediction. Finally, the deformation vectors illustrate this stage-specific pattern. Under the prescribed W–K schedule, the second trimester showed the most apparent loss of arch shape and mediolateral shift, whereas the greatest absolute inferior displacement occurred in the third trimester (Fig. 4C). These associations explain why the stage-specific minima in effective arch height and curvature mapped onto the second-trimester minimum in K, whereas load-sensitive widening continued to follow the increasing W trajectory. Deformed mesh visualizations of the reference condition and pregnancy states are provided in Supplementary Figure S2 to allow direct comparison of the global deformation pattern across conditions. Regression analysis Standardised regression analysis showed consistent associations between body weight (W), soft tissue stiffness (K), and the geometric descriptors of the transverse arch (Fig. 5). For tarsal height measures (C1, C2, C3, and Cu), W showed negative coefficients (β ≈ − 0.50 to − 0.70), indicating lower height under increased loading (Fig. 5A). K showed positive coefficients (β ≈ + 0.45 to + 0.65), indicating higher height values at greater stiffness levels. Transverse widths demonstrated regional variation (Fig. 5B). The C1–C2 width was primarily associated with K (β_K ≈ + 0.55), whereas the C2–C3 and C3–Cu widths were more strongly associated with W (β_W ≈ + 0.50 and + 0.95).Fig. 5Standardised effects of load and stiffness on transverse arch geometry. Forest plots illustrate the standardised regression coefficients (β) for vertical positions (A), mediolateral widths (B), and geometric shape parameters (C). Circles indicate coefficient estimates for load, stiffness, and their interaction Standardised effects of load and stiffness on transverse arch geometry. Forest plots illustrate the standardised regression coefficients (β) for vertical positions (A), mediolateral widths (B), and geometric shape parameters (C). Circles indicate coefficient estimates for load, stiffness, and their interaction For the aligned transverse arch geometry (Fig. 5C), arch length was almost entirely associated with W (β_W ≈ + 0.95), with minimal contribution from K (|β_K|< 0.10). Effective arch height was more strongly associated with K (β_K ≈ + 0.55), while W showed smaller negative coefficients (β_W ≈ − 0.25). Arch curvature showed opposing associations with the two predictors: W was associated with flatter curvature (β_W ≈ − 0.54), whereas K showed positive associations (β_K ≈ + 0.47). Apex displacement demonstrated combined influences of W and K. Vertically, W was associated with downward displacement (β_W ≈ − 0.40), while K showed positive associations (β_K ≈ + 0.43). In the mediolateral direction, W was associated with lateral displacement (β_W ≈ + 0.73), whereas K showed small negative coefficients (β_K ≈ − 0.26). Across all geometric measures, both predictors demonstrated moderate-to-large effect sizes (|β| ≈ 0.40–0.95). Interaction terms were generally small (|β_int|< 0.10), although moderate interactions were observed for apex displacement (β_int ≈ ± 0.15–0.20). The full set of standardized regression coefficients is provided in Supplementary Table S3. Discussion A fundamental challenge in pregnancy biomechanics is that gestational weight gain (W) and pregnancy-related changes in passive soft-tissue compliance occur concurrently and cannot be independently isolated using conventional in vivo assessments. As a result, their individual contributions to pregnancy-related arch deformation have remained unresolved and, in some cases, controversial. By employing a subject-specific FE model combined with a full factorial design, the present study mechanically separated these two factors and quantified their independent mechanical effects on transverse tarsal arch (TTA) geometry. Within this model framework, the interaction between W and K was small, suggesting that increased loading and modelled reductions in passive tissue stiffness influenced the transverse arch through largely distinct mechanical pathways. Within the prescribed W–K framework, weight gain was more strongly associated with transverse widening, whereas modelled reductions in passive tissue stiffness were more closely associated with vertical arch flattening and curvature loss. Therefore, the main mechanistic implication is not the timing of the second-trimester response itself, but the decoupling between load-sensitive widening and stiffness-sensitive flattening. These findings provide a mechanical interpretation of how concurrent changes in loading and tissue properties may contribute to pregnancy-related transverse arch deformation. They also suggest that pregnancy-related forefoot symptoms may not be explained solely by postural adaptation, but may also be associated with structurally mediated changes in transverse arch function. Mechanisms of forefoot overloading The plantar contact pressure results provide an externally interpretable context for the structural changes predicted by the model. Under the reference benchmarking condition, FE-predicted forefoot and heel peak plantar contact pressures were in close agreement with the Pedar measurements, supporting the model’s ability to reproduce regional peak plantar loading during static standing. Across the four simulated physiological conditions, peak plantar contact pressure increased in both the forefoot and heel, indicating an overall increase in plantar loading demand. However, the relationship between these regions was not monotonic. In particular, the forefoot-to-heel peak pressure ratio increased from 0.91 at reference condition and 0.90 in the first trimester to 1.03 in the second trimester, before returning to 0.90 in the third trimester. This pattern suggests a temporary change in forefoot-versus-heel load sharing rather than a simple progressive forward shift with advancing pregnancy. Elevated forefoot loading during pregnancy has been reported previously and is commonly attributed to compensatory postural adaptations and anterior displacement of the whole-body center of mass as abdominal mass increases [38, 39]. In the present model, however, loading was applied only as a vertical force under a controlled static standing condition, without explicitly representing trunk posture or forward body lean. The transient forefoot-dominant loading pattern predicted in the second trimester therefore may not be explained solely by global postural compensation and suggests that internal structural changes within the foot also contribute to regional plantar load redistribution. This interpretation was further supported by additional post-processing of all 16 W–K combinations in the 4 × 4 factorial design. Across these combinations, increasing load elevated peak plantar contact pressure in both the forefoot and heel, whereas reduced tissue stiffness was associated with a relatively greater reduction in heel peak pressure, thereby increasing the forefoot-to-heel peak pressure ratio. Regression analysis further showed that forefoot peak plantar contact pressure was primarily associated with W, whereas heel peak plantar contact pressure was influenced by both W and K. In contrast, the forefoot-to-heel peak pressure ratio was significantly associated with K, whereas W and the interaction term were not significant. Together, these findings indicate that the transient increase in the forefoot-to-heel peak pressure ratio observed around the second trimester was more closely associated with reduced tissue stiffness than with load increase alone. Such an interpretation is also consistent with current physiological and biomechanical evidence, although the causal pathway should be interpreted cautiously. Pregnancy-related hormonal and connective-tissue adaptations have been proposed as possible contributors to changes in passive soft-tissue compliance during pregnancy. However, evidence linking circulating relaxin or oestrogen levels to peripheral joint laxity and plantar-tissue stiffness remains inconsistent [12, 13]. Therefore, in the present study, K was not interpreted as a hormone-specific material parameter, but rather as a modelling proxy for examining the mechanical consequences of reduced passive soft-tissue support. In parallel, recent FE work has shown that reduced transverse arch stiffness alters plantar load transfer and increases mechanical demand in the forefoot [39]. Within the prescribed K range, reduced tissue stiffness was also the dominant determinant of effective arch height loss and curvature flattening [39, 40]. As the transverse arch loses its geometric locking capability, its ability to maintain efficient three-dimensional load transfer may be diminished, which may increase the relative loading borne by the forefoot. Within the present subject-specific model, the second-trimester increase in the forefoot-to-heel peak pressure ratio is therefore more consistent with a structurally mediated redistribution of plantar loading than with a simple forward shift of external body mass. The role of the transverse arch in foot stability The TTA behaves mechanically as a shallow shell, in which load-bearing capacity depends strongly on geometric curvature rather than on material stiffness alone [9, 41]. Although this concept was originally formulated at the metatarsal-base level, the present study applies the same general geometric principle to the cuneiform/cuboid-level tarsal transverse arch. In such systems, even modest reductions in curvature can produce disproportionate losses of structural stiffness and resistance to deformation. It should be noted that the absolute magnitude of the normalized curvature change observed in the present model was small, approximately 1% relative to the reference condition. Therefore, this value should be interpreted primarily as a model-based directional indicator accompanying the reduction in effective arch height, and the magnitude of this change may be difficult to detect in clinical or physiological measurements. Under the prescribed W–K schedule, the geometric indicators of flattening were most compromised during the second trimester, when the prescribed K reduction was greatest and effective arch height and curvature reached their minimum values. Therefore, the second-trimester flattening response should be interpreted as a mechanical consequence of the prescribed stiffness trajectory rather than as an entirely independent temporal prediction. Importantly, this does not mean that all deformation metrics peaked at this stage, as the greatest absolute inferior displacement of the arch apex occurred in the third trimester. Rather, the deformation pattern suggests a stage-specific response: the second trimester was characterized more prominently by loss of arch shape and greater mediolateral displacement, whereas the third trimester was characterized by greater downward settlement under load. This interpretation is supported by the regression analysis, which showed that effective arch height and curvature were more strongly associated with tissue stiffness, whereas arch length and apex displacement were more strongly associated with body weight. Together, these findings suggest that flattening of the transverse arch and inferior displacement of the apex were not governed by the same mechanical pathway. As the locking mechanism of the transverse arch weakens, passive stabilizers approach their functional limits, which is consistent with the higher model-predicted von Mises stresses observed in the medial cuneiform and cuboid [9, 42]. Loss of transverse arch integrity may reduce the capacity of the midfoot to resist mediolateral deformation, increasing reliance on ligamentous restraint and the medial column to maintain alignment [43, 44]. In this context, the increase in the forefoot-to-heel peak pressure ratio and the increased stress concentration in the medial and central midfoot appear to represent different manifestations of the same mechanically compromised state. Within the limitations of this subject-specific static model, these findings support the interpretation that reduced transverse arch may alter internal load sharing and reduce mechanical resistance to mediolateral deformation. More importantly, they indicate that stiffness-sensitive flattening and load-sensitive widening followed different input pathways and therefore did not necessarily co-localise temporally. This pattern also has cautious practical implications. From a mechanical perspective, reducing excessive pregnancy-related loading may be more relevant to limiting transverse widening and forefoot load concentration, whereas improving passive support or providing additional external mechanical support may be more relevant to limiting arch flattening. However, these implications remain hypothesis-generating and require prospective validation. Limitations This study was based on a single subject-specific FE model, and therefore the findings should be interpreted as subject-specific mechanical responses rather than population-level predictions. The single-foot vertical load was defined as 50% of body weight under an idealised symmetric bilateral stance assumption. Because bilateral load sharing was not measured, this assumption may not fully represent the participant’s actual right–left loading pattern during static standing and may influence the absolute plantar-pressure values. However, the same load-sharing assumption was applied consistently across all W–K conditions, allowing relative comparisons between loading and stiffness conditions within a controlled framework. To reduce kinematic variability and examine the passive mechanical interaction between pregnancy-related loading and tissue laxity, the simulations were restricted to bilateral static standing and did not include gait or dynamic loading. In addition, the Achilles tendon traction was idealized as an upward force equal to half of the applied downward load, and active muscle activation was not considered. Therefore, the model represents mechanical responses under passive static equilibrium rather than dynamic neuromuscular control during gait. The model was developed within a solid-mechanics framework and did not account for fluid-dependent tissue behavior or fluid–structure interaction. Accordingly, the findings should be interpreted within the assumptions of this simplified model. The external comparison with Pedar plantar-pressure data was limited to the nonpregnant reference condition and to forefoot and heel peak pressures from the same participant. Nevertheless, because the pregnancy-stage simulations were performed using the same subject-specific geometry, contact definitions, and modelling framework, with only W and K varied according to the prescribed conditions, the close agreement under the reference condition provides a reasonable basis for cautious extrapolation to the simulated pregnancy conditions. A further limitation of this study concerns the parameterization of pregnancy-related tissue stiffness. As noted above, the reductions in K were inferred from reported changes in peripheral joint laxity rather than from direct material testing of the plantar ligaments or plantar fascia during pregnancy. Although distal peripheral joint laxity provides a biologically plausible reference for representing systemic changes in passive connective-tissue compliance, the results should still be interpreted with caution. Conclusions Within this single-subject static finite element framework, gestational weight gain (W) and prescribed reductions in passive tissue stiffness (K) contributed differently to transverse arch deformation. Increased W was more strongly associated with transverse widening, whereas reduced passive tissue stiffness was more closely related to loss of arch height and curvature. These findings suggest that reducing excessive mechanical loading may be more relevant to limiting transverse widening and forefoot load concentration, whereas improving passive support may be more relevant to limiting arch flattening. Across the prescribed W–K conditions, the second trimester showed the most pronounced model-predicted flattening response, which corresponded to the greatest prescribed reduction in K. More importantly, the factorial analysis indicated that transverse widening and vertical flattening were associated with different mechanical inputs and therefore did not necessarily co-localize temporally. Together with the accompanying plantar pressure redistribution and model-predicted stress concentration, these findings provide a subject-specific mechanical interpretation of pregnancy-related transverse arch deformation and may help inform future studies of supportive exercise and footwear strategies. Supplementary Information Supplementary Material 1. Supplementary Material 1.