Progressive distance mapping of the plantar fat pad in hallux valgus reveals subgroup-specific morphologic subtypes

Abstract

Background/Objectives

Hallux valgus (HV) is a prevalent three-dimensional foot deformity, yet it is primarily assessed in the anteroposterior plane radiographs. The role of the sagittal plane and specifically the plantar fat pad remains poorly understood. This study aimed to quantitatively explore the relationship between plantar fat-pad geometry and HV progression.

Methods

In this retrospective study, we analysed 274 feet from 144 patients who underwent HV corrective surgery at Hasharon Hospital, Israel (2014–2024). Using custom Python-based software, we annotated 22 anatomical landmarks on preoperative, weight-bearing lateral radiographs (3 plantar fat pads and 19 plantar bony points). For each fat pad, 19 Distance features were generated (total 57), comprising 19 pairwise standardised distances. Associations between these distances and standard HV angles (HVA, IMA, DMAA, HIA) were assessed using Pearson and Spearman correlations, A subgroup analysis was carried out to compare patients aged below and above 40 years, as well as between male and female patients.

Results

In the full cohort, no distance feature correlated with standard HV angles at the prespecified threshold ( or , two-sided ). Stratified analyses revealed subgroup-specific patterns: females showed three significant relationships (mainly involving the calcaneal fat pad), whereas males showed 29 predominantly negative forefoot–hindfoot correlations (e.g., base of 5th metatarsal fat pad calcaneal fat pad vs. HIA: , , both ). The  y group exhibited six significant features (including sesamoid-based positives), while the  y group showed two first-ray–focused associations. Several effects were stronger on Spearman than Pearson, consistent with nonlinearity, and many between-group differences were significant by Fisher’s -to- (). Progressive Distance Mapping (PDM) summarized these effects into composite, radiograph-overlaid maps.

Conclusions

Progressive Distance Mapping (PDM) may suggest subgroup-dependent, progressive associations between plantar fat–pad–anchored distance geometry and hallux valgus (HV) severity. Despite no universal cohort-level correlations, PDM identifies four distinct morphologic subtypes within each subgroup (stratified by age and sex), this may suggest differing pathogenesis and patterns of plantar fat-pad involvement. These subtype-specific signatures are often nonlinear;

Introduction

Hallux valgus (HV) is a common forefoot deformity, affecting up to 30% of adults, with higher prevalence among women and older individuals [1]. Defined by lateral deviation of the hallux and medial deviation of the first metatarsal, HV often results in pain, impaired function, and difficulty with footwear [2]. Although increasingly recognized as a triplanar deformity [1, 3–5], severity is still primarily assessed in the transverse plane using standard radiographic measures such as the hallux valgus angle (HVA), intermetatarsal angle (IMA), distal metatarsal articular angle (DMAA), and hallux interphalangeal angle (HIA) [2]. These two-dimensional (2D) metrics, however, cannot fully capture the complex three-dimensional (3D) nature of the deformity. Hallux valgus presenting before age 40 is typically hereditary, often associated with a positive family history, anatomical variants (e.g., curved metatarsal head, increased DMAA, long first metatarsal), and a tendency toward bilaterality and faster progression, particularly in males where deformities are more severe and less related to footwear influences [2, 6–8]. Conversely, HV developing after age 40 is more commonly acquired, linked to age-related structural changes, prolonged footwear pressure, and comorbidities such as pes planus. With aging, its prevalence and severity increase, often accompanied by reduced joint mobility, lower arch height, and calf tightness [2, 6–11].

The plantar fat pad plays an important role in foot biomechanics and has been implicated in multiple conditions, including diabetic foot pathology [12, 13], predislocation syndrome [14], systemic sclerosis [15], plantar heel pain [16], and plantar fasciitis [17, 18]. Therapeutic strategies targeting fat-pad atrophy such as fillers, heel cups, and taping [19], or allograft adipose matrix [20, 21] highlight its clinical relevance and the potential for therapeutic benefit. In the context of HV, abnormal pressure distribution and callosity formation suggest that the plantar fat pad may be secondarily compromised, potentially amplifying pain and functional impairment [22–24]. Yet, despite this plausible link, the literature is sparse regarding its relationship to sagittal-plane radiographic changes.

Soft-tissue alterations beyond the fat pad have also been documented in HV. Ultrasound studies report reduced plantar soft-tissue thickness beneath the first metatarsal head compared with controls [25]. Moreover, displacement of the sesamoid complex and attenuation of the plantar ligaments compromise joint mechanics, with lateral sesamoid shift strongly associated with callosity formation and symptomatic pain [23, 26]. Collectively, these findings suggest that soft-tissue dysfunction is both a driver and a consequence of HV, undermining joint stability, altering load distribution, and exacerbating symptoms [27, 28]. Muscle remodeling further reinforces this pattern, with reductions in the cross-sectional area and thickness of the abductor hallucis and flexor hallucis brevis accompanied by compensatory thickening of the plantar fascia [27].

Recent sagittal-plane studies in HV have demonstrated that first metatarsal inclination decreases as transverse-plane deformity progresses [29]. However, the interplay between plantar soft-tissue pathology, particularly fat pad alterations, and sagittal-plane parameters remains poorly defined. This gap limits our understanding of how soft-tissue dysfunction contributes to the multiplanar progression of HV and underscores the need for integrative investigations that bridge radiographic and soft-tissue perspectives.

We hypothesize that the geometry of the plantar foot cushion including the heights of the first metatarsal head, fifth metatarsal base, and calcaneal heel fat pads; their distances from bony landmarks plays a pathological role in HV and deteriorates with disease progression. In this study, we investigate a broad set of plantar distance features using a framework developed by the authors (AGSV: Annotate, compute Geometric features, perform Statistics, and Visualize) to identify which features correlate with the progression of HV.

Materials and methods

Study population

This retrospective study included patients who underwent hallux valgus corrective surgery at Hasharon Hospital, Rabin Medical Center, Petach Tikva, Israel, between 2014 and 2024. A total of 457 patients were initially identified. Of these, 292 were excluded due to the absence of standing weightbearing anteroposterior (AP or LAT) foot radiographs. An additional 21 feet were excluded due to prior surgical interventions that altered foot morphology. The final cohort comprised 274 feet from 144 patients with adequate preoperative standing AP radiographs suitable for morphometric analysis (see Fig. 1). The age of 40 years was chosen as the cutoff to distinguish hereditary, early-onset hallux valgus typically observed in both sexes but often more severe in younger males from late-onset, acquired deformities that are more prevalent in females and influenced by age-related physiological changes, cumulative footwear exposure, and soft-tissue degeneration.

Fig. 1.

Flowchart of patient inclusion and exclusion. A total of 457 patients who underwent hallux valgus corrective surgery between 2014 and 2024 were screened. Of these, 292 patients were excluded due to the absence of weightbearing foot radiographs. Among the remaining 165 patients (295 feet), 21 feet were excluded due to previous surgery. The final analysis included 274 feet from 144 patients

AGSV workflow (Annotation, Geometry, Statistics, Visualize)

We employed a novel analytical framework developed by the authors. The AGSV workflow aims to evaluate anatomical geometry. The approach proceeds through four stages: (1) Annotation of anatomical landmarks; (2) Geometry extraction by converting landmark coordinates into comprehensive descriptors (e.g., pairwise distances, triangular areas, and angles); (3) Statistics to test associations and quantify significant geometric alterations, including nonlinear trends; and (4) Visualization to summarize and communicate morphometric patterns. This end-to-end pipeline enables a holistic, multi-scale assessment of morphological change.

As an application, Progressive Distance Mapping (PDM) operationalizes AGSV by (i) computing inter-landmark distances from the annotated anatomy, (ii) performing correlation analyses between each distance and deformity metrics (HVA, IMA, DMAA, HIA) using Pearson/Spearman with nonlinear checks and Fisher’s -to- for subgroup contrasts, and (iii) visualizing significant effects as composite maps and overlays on representative weight-bearing radiographs. Practically, PDM orders features along the clinical severity axis and encodes effect direction and magnitude via color and line thickness, yielding an intuitive spatial summary of morphometric change.

Data acquisition and retrieval

Standardized clinical imaging protocols were used to ensure consistent radiographic quality across all patients. Digital radiographs were retrieved from the institutional Picture Archiving and Communication System (PACS, CareSystem VueMotion).

Landmark annotation and feature extraction

Step 1: anatomical point annotation

A custom annotation application was developed by the authors in Python 3.0 using Visual Studio Code v1.1 (Microsoft Corp.), with libraries including matplotlib, pandas, pydicom, and PyQt5. Three orthopedic residents independently annotated each AP and Lateral radiograph with a standardized set of anatomical points. Each point was saved as coordinates. Final point locations were derived by averaging the coordinates across all annotators. A total of 12 anatomical landmarks were annotated per foot on anteroposterior (AP) radiographs, and were used to calculate standard Hallux Valgus angles. 22 anatomical landmarks were annotated per foot on Lateral (LAT) radiographs and were used to calculate Fat pad geometrical features (Table 1; Fig. 2).

Table 1.

Summary of the 12 anatomical landmarks annotated on AP foot radiographs

Structure	Landmarks
1st Metatarsal	5 points: Head center; medial, lateral, and midpoint of head surface; middle base point
2nd Metatarsal	3 points: Head center; midpoint of base; head articular surface midpoint
Proximal Phalanx	2 points: Head midpoint; base midpoint
Distal Phalanx	2 points: Tip and base midpoint

Landmarks were distributed across the first and second metatarsals, proximal phalanx, and distal phalanx. Each structure includes points relevant for quantifying Spatial alignment and deformity in hallux valgus

Fig. 2.

(A) Annotated anatomical landmarks of the foot on the Anteroposterior radiograph: red point (first metatarsal), green points (proximal phalanx), purple point (distal phalanx), and blue points (second metatarsal), see (Table 1). (B) Annotated anatomical landmarks on the Lateral radiograph, see (Table 2)

On the AP radiograph the annotations included 5 points on the first metatarsal, 2 points on the proximal phalanx, 2 points on the distal phalanx, and 3 points on the second metatarsal (Fig. 2A).

On the Lateral radiograph the annotations included 3 Fat Pad points (Calcaneal fat pad, 5th metatarsus base fat pad and 1st metatarsal head fat pad), and 19 points on the plantar side of foot bones: 5 points on the first metatarsal, 2 points on the proximal phalanx, 2 points on the distal phalanx, 2 points in each sesamoid, 2 points on the navicular, 2 points on the cuneiform, 2 points on the cuboid and 2 points on the calcaneus (Fig. 2B). The Python code is available per request from the corresponding author.

Table 2.

Summary of anatomical landmarks annotated on lateral foot radiographs

Structure	Landmarks
Fat pads	Calcaneal fat pad, base of the 5th metatarsal fat pad, 1st metatarsal head fat pad
1st Metatarsal	5 points: Head center; apex of anterior articular surface; base of anterior articular surface; apex of inferior shaft bowing, inferior border of the metatarsal base
Proximal Phalanx	2 points: Head anterior articular surface; inferior border of the phalanx base
Sesamoids	2 points for each sesamoid: center of the sesamoid bone, and lower border of that sesamoid bone
Cuboid	2 points: anterior lower border point, posterior lower border point
Navicular bone	2 points: anterior lower border point, posterior lower border point
Cuneiform	2 points: anterior lower border point, posterior lower border point
Calcaneus	2 points: anterior lower border point, posterior lower border point

Step 2: standard HV angles calculation

Standard HV angles calculation and Progressive distance analysis and visualization were conducted using custom code written in Python 3.0 (JupyterLab v4.2.5). Libraries used included pandas, numpy, scipy, seaborn, cv2, and pydicom.

Standard hallux valgus angle calculation

The following standard radiographic angles were calculated using vector geometry between the appropriate anatomical landmarks on AP radiographs:

• Hallux Valgus Angle (HVA).

• Distal Metatarsal Articular Angle (DMAA).

• Intermetatarsal Angle (IMA).

• Hallux Interphalangeal Angle (HIA).

Angles were computed by defining vectors between landmarks and measuring the angle at their intersection.

Step 3: geometrical features calculation

Distance measurements

All pairwise distances between each fat pad point and the 19 plantar anatomical points were calculated using the Euclidean distance formula.

To facilitate inter-subject comparison, all distances were normalized to the length of the first metatarsal midshaft (measured between the apices of the superior and inferior shaft bowing). This scaling length itself showed no correlation with any standard HV angle (HVA, IMA, DMAA, HIA), indicating that the normalization did not bias the analyses.

The number of unique distances from the single fat pad point to efficacy of shoes modification and orthotics in hallux valgus deformity: a comprehensive review of literature. r point was equal to the number of plantar points (19 points). Thus, a total of 57 unique distances were computed for each foot.

Step 4: correlation analysis

We computed Pearson’s and Spearman’s to quantify linear and monotonic associations, respectively, between the 57 distance features and the standard HV angles (HVA, IMA, DMAA, HIA). Features were flagged as noteworthy if they met both a magnitude threshold ( or ) and statistical significance (two-sided ). Between-group differences in correlation strength were formally evaluated with Fisher’s -to- tests for Pearson an Spearman coefficients.

Step 5: visualization and clinical relevance analysis

To aid interpretation, for each subgroup, all distance features that met the significance criteria were:

1. summarized in a single composite schematic and illustrated with a sample example, and.

2. overlaid on two representative weight-bearing radiographs:

•  one depicting a foot with mild hallux valgus (HV), and.

• one depicting a foot with severe HV.

These visualizations clarify the progression of deformity and highlight clinically relevant spatial relationships.

Results

Cohort characteristics

The study cohort consisted of 274 feet from 144 unique patients. Female patients constituted the majority, with 107 females contributing 200 feet, while 37 males contributed 74 feet. The mean age of the cohort was years (range: 17–85). Female patients had a mean age of years (range: 18–85), whereas male patients were younger on average ( years, range: 17–78). A detailed summary of demographics is provided in Table 3.

Table 3.

Patient demographics of the study cohort

Sex	Patients (n)	Feet (n)	Age (meanSD)	Age range
F	107	200		18–85
M	37	74		17–78
Total	144	274		17–85

The table summarizes the distribution of patients and feet by sex, along with corresponding age statistics. Female patients constituted most of the cohort, contributing 200 out of 274 analysed feet

Standard radiographic angles

Radiographic severity varied across the four standard hallux valgus angles. HVA classified most feet as moderate (172; 62.8%), followed by mild (70; 25.5%) and severe (32; 11.7%). IMA was predominantly mild (142; 51.8%), with moderate (98; 35.8%) and severe (34; 12.4%) comprising the remainder. DMAA showed a mild majority (141; 51.5%) but a comparatively higher severe proportion (78; 28.5%), with moderate in 55 feet (20.1%). HIA was more evenly distributed, with moderate (89; 32.5%) and severe (79; 28.8%) cases alongside a large mild group (106; 38.7%). These distributions are summarized in Table 4 and illustrated in Fig. 3.

Table 4.

Distribution of radiographic severity levels across standard hallux valgus angles

Severity	HVA	IMA	DMAA	HIA
Mild	70 (25.5%)	142 (51.8%)	141 (51.5%)	106 (38.7%)
Moderate	172 (62.8%)	98 (35.8%)	55 (20.1%)	89 (32.5%)
Severe	32 (11.7%)	34 (12.4%)	78 (28.5%)	79 (28.8%)

The table presents the number and percentage of feet classified as mild, moderate, or severe based on hallux valgus angle (HVA), intermetatarsal angle (IMA), distal metatarsal articular angle (DMAA), and hallux interphalangeal angle (HIA). Classification thresholds were applied independently for each angle, highlighting variation in deformity presentation

Fig. 3.

Distribution of standard hallux valgus angles (HVA, IMA, DMAA, HIA) across the study cohort. HVA exhibits the widest variability, while IMA shows the most compact distribution

Distribution of standard hallux valgus angles (HVA, IMA, DMAA, HIA) across the study cohort. HVA exhibits the widest variability, while IMA shows the most compact distribution.

Distance-Based morphometric analysis

A total of 19 pairwise distances were computed between each fat pad and plantar bony anatomical landmarks annotated on each lateral foot radiograph (a total of 57 distance features). No significant Pearson or Spearman moderate correlation ( or , ) was identified between those distances and HV angle severity.

Subgroup analyses

We performed stratified analyses by sex (female vs. male) and age ( y vs.  y). Within each stratum, we recomputed Pearson’s and Spearman’s between all geometric features and the standard HV angles (HVA, IMA, DMAA, HIA), using the same thresholds as the primary analysis (magnitude criterion or and two-sided ). Between-group differences in correlation strength were formally evaluated with Fisher’s -to- tests for Pearson an Spearman coefficients.

By sex

In females () and males (), distance-based associations revealed two distinct sets of correlations.

For females, only few significant relationships were observed (three features met or with ), primarily involving the positive correlations with calcaneal fat pad and single negative correlation with first metatarsal fat pad (Table 5; Fig. 4a). These included: calcaneus inferior facet (posterior) fat pad at base of the 5th metatarsal vs. HIA (, ; , ); calcaneus inferior facet (posterior) calcaneal fat pad vs. DMAA (, ); and fat pad at the 1st metatarsal head proximal phalanx base (inferior) vs. HVA (, ).

Table 5.

Female subgroup: distance features associated with standard HV angles

Distance	Angle	n	Pearson r (p)	Spearman ρ (p)	Between-sex p
Calcaneus inferior facet (posterior) → fat pad at base of 5th metatarsal	HIA	188	0.25 (< 0.01)	0.22 (< 0.01)	< 0.01
Calcaneus inferior facet (posterior) → calcaneal fat pad	DMAA	188	0.22 (< 0.01)	0.25 (< 0.01)	0.02
Fat pad at 1st metatarsal head → proximal phalanx base (inferior)	HVA	188	-0.22 (< 0.01)	-0.25 (< 0.01)	0.56

Rows included when either |r| or |ρ| ≥ 0.25 with p < 0.05, irrespective of the fisher between-sex test. Bolded coefficients Meet the magnitude threshold

Fig. 4.

Lateral radiographs illustrating distance features that were significant in the sex stratified analysis. (a.) 31-year-old female with overlaid distances meeting the inclusion criterion (|r| or |ρ| ≥ 0.25 with p < 0.05) of female subgroup. (b.) 29-year-old male with overlaid distances significant in the male subgroup. Color bars at right encode the correlation coefficient: red indicates positive Pearson r, blue indicates negative Pearson r

The first two associations were significantly different from the male subgroup, with Fisher’s -test -values of and , respectively.

By contrast, males exhibited many stronger significant associations (29 features), spanning the plantar cushion including the 1st metatarsal fat pad, base of the 5th metatarsal fat pad, and calcaneal fat pad (Table 6; Fig. 4b). These relationships were predominantly negative correlations with HIA and DMAA, and to a lesser extent with IMA and HVA. Many involved hindfoot–forefoot fat-pad constructs (e.g., base of the 5th metatarsal fat pad calcaneal fat pad vs. HIA: , , both ).

Table 6.

Male subgroup (B): all distance features with |r| or |ρ| ≥ 0.25 and p < 0.05

Distance	Angle	Pearson r (p)	Spearman ρ (p)	Between-sex p
Fat pad at base of 5th metatarsal → calcaneal fat pad	HIA	-0.44 (< 0.01)	-0.39 (< 0.01)	< 0.01
Calcaneus inferior facet (posterior) → fat pad at 1st metatarsal head	HIA	-0.41 (< 0.01)	-0.37 (< 0.01)	< 0.01
Calcaneus inferior facet (posterior) → fat pad at base of 5th metatarsal	HIA	-0.40 (< 0.01)	-0.35 (< 0.01)	< 0.01
Fat pad at 1st metatarsal head → calcaneal fat pad	HIA	-0.38 (< 0.01)	-0.34 (< 0.01)	< 0.01
1st sesamoid center → calcaneal fat pad	HIA	-0.36 (< 0.01)	-0.31 (< 0.01)	< 0.01
Fat pad at 1st metatarsal head → navicular (posterior–inferior)	HIA	-0.36 (< 0.01)	-0.34 (< 0.01)	< 0.01
Fat pad at 1st metatarsal head → navicular (anterior–inferior)	HIA	-0.35 (< 0.01)	-0.35 (< 0.01)	< 0.01
Fat pad at 1st metatarsal head → 1st metatarsal inferior articular surface	HVA	-0.35 (< 0.01)	-0.42 (< 0.01)	0.09
Fat pad at 1st metatarsal head → proximal phalanx base (inferior)	IMA	-0.34 (< 0.01)	-0.40 (< 0.01)	0.05
1st sesamoid inferior → calcaneal fat pad	HIA	-0.34 (< 0.01)	-0.27 (0.02)	< 0.01
Calcaneal fat pad → 1st metatarsal shaft (inferior bowing apex)	HIA	-0.34 (< 0.01)	-0.31 (< 0.01)	< 0.01
Calcaneal fat pad → proximal phalanx base (inferior)	HIA	-0.33 (< 0.01)	-0.29 (0.01)	< 0.01
Calcaneal fat pad → 1st metatarsal inferior articular surface	HIA	-0.33 (< 0.01)	-0.30 (< 0.01)	< 0.01
Calcaneal fat pad → 1st metatarsal base (inferior)	HIA	-0.33 (< 0.01)	-0.30 (< 0.01)	< 0.01
2nd sesamoid centre → 1st metatarsal head fat pad	HVA	-0.32 (< 0.01)	-0.36 (< 0.01)	0.17
Calcaneus inferior facet (anterior) → fat pad at 1st metatarsal head	HIA	-0.32 (< 0.01)	-0.29 (0.02)	< 0.01
Calcaneus inferior facet (anterior) → calcaneal fat pad	HIA	-0.32 (< 0.01)	-0.29 (0.01)	< 0.01
1st sesamoid centre → 1st metatarsal head fat pad	HVA	-0.29 (0.02)	-0.30 (0.02)	0.32
1st sesamoid centre → 1st metatarsal head fat pad	DMAA	-0.29 (0.02)	-0.26 (0.04)	0.01
Fat pad at 1st metatarsal head → 1st metatarsal base (inferior)	HIA	-0.29 (0.01)	-0.29 (0.01)	0.02
2nd sesamoid centre → calcaneal fat pad	HIA	-0.29 (0.01)	-0.25 (0.04)	0.06
Fat pad at 1st metatarsal head → 1st metatarsal bowing apex	HIA	-0.29 (0.01)	-0.29 (0.01)	< 0.01
Fat pad at 1st metatarsal head → 1st metatarsal head articular surface	DMAA	-0.28 (0.02)	-0.25 (0.04)	0.01
Calcaneal fat pad → 1st metatarsal inferior articular surface	DMAA	0.28 (0.02)	0.30 (0.01)	0.14
Fat pad at 1st metatarsal head → proximal phalanx base (inferior)	DMAA	-0.28 (0.02)	-0.32 (< 0.01)	0.02
Fat pad at 1st metatarsal head → proximal phalanx base (inferior)	HVA	-0.27 (0.03)	-0.33 (< 0.01)	0.56
2nd sesamoid inferior → calcaneal fat pad	HIA	-0.27 (0.02)	-0.22 (0.06)	0.01
Fat pad at base of 5th metatarsal → 1st metatarsal inferior articular surface	DMAA	0.26 (0.04)	0.25 (0.04)	0.06
Fat pad at 1st metatarsal head → 1st metatarsal shaft (inferior bowing apex)	HVA	-0.26 (0.04)	-0.18 (0.14)	0.91

Bolded coefficients Meet the threshold

Between-sex Fisher’s -tests indicated sex-specific differences for most features (e.g., calcaneus inferior facet (posterior) fat pad at base of the 5th metatarsal vs. HIA: ; calcaneus inferior facet (posterior) calcaneal fat pad vs. DMAA: ). However, some features showed no material differences (e.g., 1st metatarsal head fat pad proximal phalanx base vs. HVA: ).

As illustrated in Fig. 5, the distance between the 1st metatarsal fat pad and the inferior metatarsal head articular surface progressively decreases with increasing DMAA severity, exemplifying the inverse relationship characteristic of the male subgroup.

Fig. 5.

Anteroposterior and lateral radiographs of two patients from the male subgroup illustrating the relationship between the distance separating the 1st metatarsal fat pad and the inferior metatarsal head articular surface with the severity of the distal metatarsal articular angle (DMAA). (a) Radiographs of a 20-year-old male with severe deformity (DMAA = 21°, HVA = 28°, IMA = 13°) showing a shorter distance (0.8 standardized units). (b) Radiographs of a 36-year-old male with mild deformity (DMAA = 6.3°, HVA = 22°, IMA = 12°) showing an increased distance (1.65 standardized units), consistent with the observed negative correlation between this metric and DMAA severity

By age

In participants  years () and  years (), distance-based associations were more numerous in the younger subgroup. The group yielded six features meeting or with (Table 7., Fig. 6a), including forefoot–hindfoot pairs and sesamoid relations (e.g., sesamoid inferior1st metatarsal head fat pad vs. IMA: , ; and ; this correlation was the only positive correlation in this subgroup; and calcaneus anterior–inferiorcalcaneal fat pad vs. HIA: , ; both ).

Table 7.

Age ≥ 40 subgroup: distance features with |r| or |ρ| ≥ 0.25 and p < 0.05

Distance	Angle	Pearson r (p)	Spearman ρ (p)	Between-age p
Fat pad at 1st metatarsal head → 1st metatarsal inferior articular surface	HVA	-0.23 (< 0.01)	-0.26 (< 0.01)	0.80
Fat pad at 1st metatarsal head → 1st metatarsal shaft (inferior bowing apex)	HVA	-0.25 (< 0.01)	-0.25 (< 0.01)	0.32

Bolded coefficients Meet the threshold

Fig. 6.

Lateral radiographs illustrating distance features that were significant in the age-stratified analysis. (a) 29-year-old male ( years) with overlaid distances meeting the inclusion criterion ( or with ) of subgroup below 40 years of age. (b) 70-year-old female ( years) with overlaid significant distances of subgroup above 40 years of age. Color bars at right encode the correlation coefficient (Pearson ): red indicates positive correlations, blue indicates negative correlations

By contrast, the group (Table 8., Fig. 6b) showed two HVA-focused associations centered on first-ray geometry (1st metatarsal head fat pad1st metatarsal inferior articular surface: , ; and 1st metatarsal head fat pad1st metatarsal shaft bowing apex: , , both ). Between-age Fisher tests indicated significant differences for selected features, and only in the younger subgroup (e.g., IMA: 1st sesamoid inferior1st metatarsal head fat pad, ; DMAA: 1st metatarsal head fat padproximal phalanx base, ; HIA: calcaneus anterior–inferiorcalcaneal fat pad, ), while others showed no material age effect (e.g., HVA: 1st metatarsal head fat pad1st metatarsal inferior articular surface, ).

Table 8.

Age < 40 subgroup: distance features with |r| or |ρ| ≥ 0.25 and p < 0.05

Distance	Angle	Pearson r (p)	Spearman ρ (p)	Between-age p
1st sesamoid inferior → fat pad at 1st metatarsal head	IMA	0.35 (< 0.01)	0.29 (0.01)	< 0.01
Fat pad at 1st metatarsal head → proximal phalanx base (inferior)	DMAA	-0.34 (< 0.01)	-0.26 (0.03)	0.02
Calcaneus inferior facet (anterior) → calcaneal fat pad	HIA	-0.33 (< 0.01)	-0.29 (0.01)	0.01
Fat pad at 1st metatarsal head → proximal phalanx base (inferior)	IMA	-0.28 (0.02)	-0.30 (0.01)	0.34
Fat pad at 1st metatarsal head → 1st metatarsal inferior articular surface	HVA	-0.24 (0.05)	-0.29 (0.01)	0.80
2nd sesamoid centre → fat pad at 1st metatarsal head	HVA	-0.28 (0.02)	-0.28 (0.02)	0.54

Bolded coefficients Meet the threshold

Discussion

The role of the plantar fat pad in HV remains to be fully defined. Our findings suggest a relationship between fat pad–anchored morphometrics and deformity severity, but the precise mechanisms and clinical implications require further study.

The analysis of the cohort as a whole revealed no correlation between any of the plantar foot cushion distances and the standard HV severity angles. Despite this, subgroup analysis identified four sets of significant correlations, each specific to a subgroup. This may reflect the heterogeneity of the overall cohort and the relative homogeneity within each subgroup. These correlations were largely distinct and statistically different between subgroups.

For example, while the male subgroup demonstrated many (28) moderate negative correlations spanning the entire plantar cushion (Fig. 5), the female subgroup exhibited only two weak positive correlations, primarily involving the calcaneal fat pad, and one negative correlation involving the 1st metatarsal head. Similarly, the younger group showed both positive correlations (e.g., with the sesamoid bone) and negative correlations (e.g., with the calcaneal fat pad), both of which were statistically significant and unique to this group. In contrast, the older group exhibited only negative correlations between the 1st metatarsal fat pad and the distal 1st metatarsus, which were not significantly different from the younger group.

Each plantar fat pad was annotated using a single anatomical point corresponding to its maximal bony prominence the calcaneal tuberosity, base of the fifth metatarsal, and first metatarsal head. These regions represent the areas of highest physiological loading and are therefore most susceptible to compression and atrophy in hallux valgus. Selecting discrete points directly anchored to bony landmarks ensured consistent, reproducible measurements while minimizing annotation congestion and data redundancy that could arise from multi-point mapping. Moreover, this simplified approach supports our long-term goal of developing a practical, clinic-based assessment tool that can be easily adopted by clinicians for standardized radiographic evaluation.

Given that HV is a triplanar deformity that involves both soft tissue and bony alterations, and given that our distances were anchored on bony structures some of which change intrinsically with disease progression many of the observed signals likely reflect integrated alterations of the plantar fat cushion together with skeletal remodeling (e.g., decreased first-ray inclination and changes in the calcaneal axis) that make soft-tissue changes more radiographically apparent. This also suggests that these changes are triplanar and not exclusive to the sagittal plane. Recent investigations have applied multiplanar and triplanar quantitative analyses using both radiographs and weightbearing CT, emphasizing angular and positional relationships rather than area-based metrics [29, 30]. Although advances in imaging modalities and computer-aided design have enhanced the precision and reproducibility of these assessments, no prior studies have reported a method for quantifying defined distances in the sagittal plane for hallux valgus evaluation [1, 29–31].

As a related example, many significant correlations were identified between sesamoid-based fat-pad angles and hallux valgus parameters (HVA, DMAA, IMA, and HIA). These correlations may be exaggerated by the characteristic lateral shift of the sesamoids over the 1st metatarsal head rather than by true distance changes captured by fat-pad landmarks. In hallux valgus, the sesamoids migrate laterally relative to the metatarsal axis a displacement that is predominantly translational rather than angular.

Many correlations exhibited differing behaviors between Spearman and Pearson coefficients, indicating that some relationships were predominantly linear, whereas others showed stronger nonlinear patterns. As illustrated in Fig. 7a, the distance between the inferior calcaneal facet and the 5th metatarsal base fat pad displayed a relatively linear negative correlation with the severity of the hallux interphalangeal angle (HIA), a relationship more accurately captured by the Pearson coefficient. In contrast, other associations demonstrated nonlinear behavior (Fig. 7b and c) that was better represented by Spearman coefficients. For example, in Fig. 7b, the relationship between the fat pad at the 1st metatarsal head and the 1st metatarsal inferior articular surface versus the hallux valgus angle (HVA) showed Pearson r = − 0.35 (p < 0.01) and Spearman ρ = − 0.42 (p < 0.01), an absolute difference of 0.09. The curvature of this trend suggests an initial flattening of the plantar fat cushion early in the disease course, followed by an apparent rebound (increased measured angle or area) in more advanced deformity. Although this relationship may be influenced by a few extreme cases or potential outliers, the observed nonlinear pattern could reflect changes in severe hallux valgus related to first-metatarsal pronation and coronal-plane deformity, which alter the projected geometry on lateral radiographs.

Fig. 7.

Examples of linear and nonlinear correlations between plantar fat pad–related geometric features and hallux valgus severity measures. (a) The distance between the inferior calcaneal facet and the 5th metatarsal base fat pad shows a predominantly linear negative association with the hallux interphalangeal angle (HIA), exhibiting a stronger Pearson coefficient compared with Spearman. (b) The distance between the 1st metatarsal head fat pad and the 1st metatarsal inferior articular surface versus the hallux valgus angle (HVA) demonstrates a nonlinear pattern, where the relationship flattens at moderate deformity levels before steepening again in more severe cases captured more effectively by the Spearman coefficient (ρ = − 0.42) than by the Pearson correlation (r = − 0.35). (c) The distance between the 1st metatarsal head fat pad and the base of the proximal phalanx versus the intermetatarsal angle (IMA) similarly shows a nonlinear trend, in which the relationship flattens in more severe cases. In all panels, the blue line represents the nonlinear (quadratic) Spearman regression fit, the red dashed line indicates the linear Pearson regression fit, and the red shaded area shows the 95% confidence interval around the linear regression

The majority of associations were weak to moderate in magnitude. This likely reflects the chronic, slowly progressive nature of HV, inter-individual variability in soft-tissue composition, the influence of multiple coupled skeletal parameters, and measurement noise inherent to 2D projections of a 3D deformity. The most consistent changes localized to the forefoot fat cushion particularly the 1st metatarsal head fat pad with less prominent effects at the 5th metatarsal base and calcaneal heel fat pads. This regional specificity aligns with the clinical focus on first-ray mechanics in HV.

We evaluated all 57 normalized inter-landmark distances against each of the four standard radiographic angles (HVA, IMA, DMAA, HIA), computing both Pearson’s r and Spearman’s ρ for each distance–angle pair. We report only those associations that met the pre-specified thresholds ( |r| or |ρ|≥0.25 and two-sided p < 0.05 ); all other correlations did not meet these criteria and are not shown.

While our data indicate that hallux valgus severity may be associated with progressive change of the plantar fat cushion, we do not establish causality. These observations nonetheless raise the possibility that targeted management of fat-pad compromise (e.g., protective off-loading or soft-tissue–supportive interventions) could complement skeletal correction in selected patients an avenue that warrants prospective evaluation. The progressive deterioration of the plantar fat pad observed in certain HV subgroups may be critical to pain and function. Future therapeutic strategies, such as targeted soft-tissue support, must consider that while dynamic foot orthoses can help compensate for subtalar hyperpronation, care is needed, as some orthotic designs are noted to negatively affect dorsal or medial pressures [32].

Beyond the geometric considerations of the plantar fat pad, our findings are consistent with a broader body of evidence indicating that hallux valgus arises from complex, coupled interactions among osseous architecture, soft-tissue balance, and age-related structural degeneration. Foundational investigations have demonstrated progressive deformation of the first ray and metatarsophalangeal joint driven by abnormal loading patterns, ligamentous attenuation, sesamoid migration, and intrinsic muscle imbalance [33–39]. These studies consistently underscore that soft-tissue insufficiency particularly of the medial capsule, plantar structures, and sesamoid apparatus plays a decisive role in deformity progression and may precede overt skeletal malalignment. More recent work using advanced imaging modalities and biomechanical modelling has refined this perspective. CT-based three-dimensional reconstructions and weightbearing systems have shown substantial multiplanar rotation of the first metatarsal and sesamoids, along with alterations in plantar loading and soft-tissue tension [40–42]. Complementary ultrasound- and MRI-based analyses further demonstrate that plantar soft-tissue thinning, altered fat-pad elasticity, and degeneration of intrinsic musculature correlate with symptom severity and gait dysfunction [43]. Collectively, these data support the concept that plantar soft-tissue remodelling including fat-pad displacement, compression, and atrophy is not merely an epiphenomenon but contributes meaningfully to the biomechanical milieu that sustains or accelerates deformity. Our results, which identify subgroup-specific associations between fat-pad geometry and conventional hallux valgus radiographic parameters, extend this literature by providing sagittal-plane, distance-based quantitative evidence of plantar soft-tissue involvement that complements the previously described multiplanar skeletal and soft-tissue adaptations.

This study has several limitations that should be acknowledged. First, it was conducted at a single institution using a retrospective cohort design, which may limit the generalizability of the findings. As a cohort study, no longitudinal follow-up was available, precluding evaluation of deformity progression or the temporal relationship between radiographic and soft-tissue changes. Second, while three independent annotators performed landmark identification and consensus averaging was applied, those annotators were orthopedic residents, and manual annotation remains subject to inter- and intra-observer variability; automated or AI-assisted methods may improve reproducibility in future work. Third, as the analysis was based on two-dimensional radiographs, the results may not fully capture the complex three-dimensional geometry of the foot, and advanced imaging modalities such as CT, MRI, or ultrasound which provide more direct visualization of the plantar fat pad and intrinsic musculature were not included. Fourth, the cross-sectional design precludes causal inference. Finally, although radiographic parameters and geometric features were systematically analyzed, clinical correlations such as pain scores, plantar pressure mapping, or functional outcomes were not available, and the custom analytical framework (AGSV/PDM) has yet to undergo external validation, limiting direct translation to clinical practice.

Conclusions

Progressive Distance Mapping (PDM) may suggest subgroup-dependent associations between plantar fat–pad–anchored distance geometry and hallux valgus (HV) severity. While no universal cohort-level correlations were observed, PDM highlighted distinct morphologic patterns across age and sex strata, suggesting possible differences in the mechanical or soft-tissue adaptations underlying the deformity. These relationships were frequently nonlinear. Prospective, multicenter studies incorporating longitudinal and weight-bearing CT or plantar-pressure data are warranted to validate these findings and explore their potential clinical relevance.

Author contributions

H.M. contributed to **Methodology** , **conceptualization, formal analysis** and **writing – original draft** . M.K. participated in **conceptualization, data curation** and **writing – original draft** . A.M. contributed to **investigation** and **writing – review & editing** . F.Q. was involved in **methodology development** and **investigation** . D.R. contributed to **methodology** and **writing – review & editing** . L.A. was responsible for **conceptualization** and **writing – review & editing** . M.E. contributed to **conceptualization** and **data curation** . M.S. was involved in **data curation** and **writing – review & editing** . E.L. provided **resources** . M.Y. contributed to **resources** , **supervision** , and **project administration** . H.M. and M.K. contributed equally to this work. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work.

Funding

No specific funding was received for this study. The research was conducted as part of departmental academic activity without external financial support.

Data availability

The datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Institutional Review Board of Rabin Medical Center, Petach Tikva, Israel (IRB number: 0414-25-RMC, approved 18/08/2025). Patient consent was waived due to the retrospective nature of the study and the use of anonymized radiographic data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.