Shape phenotype of thigh fat and muscle and risk of major adverse cardiovascular events after fragility hip fracture

Abstract

Background

Although sarcopenia has been recognized as a predictor of mortality in elderly patients with hip fracture, the association of thigh fat and muscle with cardiovascular (CV) outcome remains unclear. We examined the impact of computed tomography (CT)-derived shape features of thigh fat and muscle on major adverse CV events (MACE) in elderly patients with hip fracture.

Methods

We conducted a retrospective analysis of consecutive patients aged ≥65 years who presented with hip fracture confirmed on pelvic bone CT scan and underwent hip fracture surgery at our institution from April 2019 to December 2021. The cross-sectional area (CSA) and compactness (CM) of both the muscle and fat at the upper-thigh level were calculated from two-dimensional CT images using AVIEW Research (v1.1.38, Coreline Soft, Co. Ltd, Seoul, South Korea). The shape features of thigh fat and muscle were categorized into four groups based on the combination of CSA and CM: fat CSA (fat area [FA])/fat CM (FCM), muscle CSA (muscle area [MA])/muscle CM (MCM), FA/MCM and MA/FCM. In each of them, subjects were categorized into four subgroups: high CSA/high CM, high CSA/low CM, low CSA/high CM and low CSA/low CM. The primary outcome was MACE after 30 days of surgery, defined as a composite of all-cause death, acute myocardial infarction, stroke or hospitalization for heart failure.

Results

Of 356 patients enrolled (median age, 82 years; 76.7% females), 72 (20.2%) had MACE over a median follow-up of 13.1 months (ranges 5.9–21.0 months). Patients with MACE had a significantly lower median FA (193.7 vs. 226.2 cm², P < 0.0001) and FCM (0.443 vs. 0.513, P = 0.001) compared with those without MACE, but no significant differences were found in MA, MCM and FA–MA ratio between the two groups. In a multivariate Cox regression analysis, low FA (<240.1 cm²) (adjusted hazard ratio [HR] 2.99, 95% confidence interval [CI] 1.39–6.44, P = 0.005) and low FCM (<0.477) (adjusted HR 2.00, 95% CI 1.10–3.63, P = 0.023) were associated with an increased risk of MACE. Among the shape phenotypes of thigh fat and muscle, the thigh fat phenotype of low FA/low FCM (adjusted HR 3.13, 95% CI 1.81–5.42, P < 0.0001 [reference, high FA/high FCM]) was found to be an independent predictor of MACE.

Conclusions

In elderly patients with fragility hip fracture, thigh CT-derived measures of FA and FCM may provide useful prognostic information for predicting adverse CV outcomes.

Introduction

With the aging population, the incidence of hip fractures is increasing and its socioeconomic burden has been an important issue to both individuals and the community healthcare system.^1,2 The 1-year mortality rate of elderly patients with hip fracture ranges from 20% to 40% in the literature,^S1,3 and cardiovascular (CV) diseases (CVDs) have been the leading cause of death.⁴ Although earlier studies have reported that chronic comorbid conditions of elderly patients with hip fracture are associated with an increased risk of acute myocardial infarction (AMI), stroke and heart failure (HF) after surgery,^5–8 data on the predictors of adverse CV events in this population are limited.

Sarcopenia, the age-related loss of skeletal muscle mass and strength,⁹ increases the risk of subsequent fragility fracture occurring after low-energy trauma and it has been also associated with metabolic syndrome, which increases the risk of CVD morbidity and mortality.¹⁰ Particularly, the decline of muscle mass in lower limbs is twice that in the upper limbs¹¹ and the thigh is thought to be an area more susceptible to the effects of aging than lower leg or pelvic muscles.¹² In recent years, the significant association between thigh muscle mass and insulin resistance or incident type 2 diabetes mellitus (DM) has been also reported.¹³ Adiposity is associated with metabolic and CV risk profiles, but this relationship is complex and the susceptibility to cardiometabolic complications is not solely mediated by total body fat mass.¹⁴ Previous studies have demonstrated an increase in total body fat and redistribution of body fat (increased visceral fat, decreased subcutaneous fat and intermuscular fat infiltration) with aging,¹⁵ and the impact of total body fat mass has been largely dependent on the body fat distribution.¹⁶ In addition, the adequate expansion of subcutaneous fat to a positive energy balance has been considered as a beneficial physiological response that inhibits visceral or ectopic fat deposition. Lower limb fat is mainly stored as a subcutaneous adipose tissue in gluteofemoral (or thigh) region and its CV protective effects by acting as a metabolic buffer for the excess dietary lipids are well known.¹⁴ However, most studies among elderly patients with hip fracture used body mass index (BMI) as a proxy measure of adiposity, which does not reflect the amount and location of body fat.¹⁷

Although there have been some studies exploring the prognostic value of thigh composition in elderly patients with hip fracture, they have focused on predicting surgical complication, functional recovery or mortality rather than adverse CV events.^18–20 Few studies have examined the relationship between thigh fat and muscle measurements and CV outcomes following hip fracture. Given that pelvic bone computed tomography (CT) is routinely performed as an initial examination for suspected hip fracture, this opportunistic measurement of thigh composition without additional exposure to radiation is practical and reliable.

This study investigated the association of thigh composition with major adverse CV events (MACE) using shape features of thigh fat and muscle, cross-sectional area (CSA) and compactness (CM), derived from the routine preoperative pelvic bone CT images among elderly patients with fragility hip fracture.

Methods Study population

We conducted a retrospective cohort study of consecutive patients aged ≥65 years who presented to the emergency department of our institution with hip fracture confirmed on pelvic bone CT scan and who underwent hip fracture surgery from April 2019 through December 2021. This study was approved by the institutional review board (PC22RISI0041), and written informed consent from participants was waived due to the retrospective nature of the study.

Data collection and variables

All demographic and clinical data were extracted from the hospital electronic medical records and included age, sex, physical activity before trauma, pre-existing comorbidities (DM, hypertension, dyslipidaemia, coronary artery disease [CAD], atrial fibrillation [AF], HF, chronic kidney disease [CKD], stroke, dementia and chronic obstructive pulmonary disease), medications, laboratory findings at admission, echocardiographic parameters before surgery, fracture location and surgery type, and length of hospital stay. CAD was defined as the presence of at least one major epicardial vessel with ≥50% luminal stenosis on coronary CT angiography or invasive coronary angiography or a previous history of percutaneous coronary intervention (PCI) or surgical revascularization.^S2 AF was defined as an irregular heart rhythm with the absence of P waves on 12-lead electrocardiogram (ECG),^S3 and patients with newly detected AF on ECG during hospitalization, as well as those with previously known AF, were considered to have pre-existing AF.

Outcome measurements

The primary outcome was the first occurrence of MACE after 30 days of surgery, which was defined as a composite of all-cause death, AMI, stroke or hospitalization for HF. AMI was defined as symptoms of ischaemia with ST-segment elevation ≥ 1 mm in two or more contiguous leads, ST-segment depression ≥ 0.5 mm and/or T wave inversion ≥ 1 mm in two or more contiguous leads or new left bundle branch block and elevation of cardiac biomarkers (preferably troponin) with at least one value above the 99th percentile of the upper reference limit.^S4 Stroke was defined as a neurological deficit of presumed ischaemic vascular origin lasting ≥24 h confirmed on brain CT and/or magnetic resonance imaging.^S5 Hospitalization for HF was defined as the first readmission due to new-onset or worsening symptoms and signs of HF.

Computed tomography image acquisition

Pelvic bone CT was performed to diagnose hip fracture using 128-slice CT systems (Definition Edge and Somatom Force; Siemens Healthineers, Erlangen, Germany), with 120 kVp, automatic mAs modulation protocol, and field of view of 50 cm, matrix size of 512 × 512. Patients were scanned from the level of the third or fourth lumbar spine to the level of the midshaft of the femur in a supine position, and images were obtained craniocaudally without administration of intravenous contrast agents. The data were reconstructed with bone kernel (B69f) in 3 mm thickness without inter-slice gaps.

Body morphometric analysis of thigh fat and muscle

An experienced radiologist (L. S. W.) who was blinded to all the patient details selected the axial hip CT images at the level of the inferior tip of the ischaemia tuberosity and analysed the CT images for body composition measures using commercially available segmentation software of AVIEW Research (v1.1.38, Coreline Soft, Co. Ltd, Seoul, South Korea). The CSA (square centimetres) and CM of fat and skeletal muscle at the level of the upper thigh were assessed. Total fat area (FA) was measured using fat density thresholds from −190 to −30 Hounsfield units (HU). Skeletal muscle area (MA) was determined as the sum of lean soft tissues (non-bone and non-fat) within the deep fascia and was measured using predetermined thresholds from −29 to +150 HU (Figure 1). CM was defined as the ratio of the area of an object to the area of a circle (the most compact shape) with the same perimeter and was automatically calculated as below: [Image Omitted. See PDF]

The shape features of thigh fat and muscle were categorized into four groups based on the combination of CSA (FA or MA) and CM (fat CM [FCM] or muscle CM [MCM]), and in each of them, subjects were further categorized into four thigh composition phenotypes using the cut-off values of CSA and CM in thigh fat or muscle as follows.

• FA/FCM phenotypes: high FA/high FCM, high FA/low FCM, low FA/high FCM and low FA/low FCM;
• MA/MCM phenotypes: high MA/high MCM, high MA/low MCM, low MA/high MCM and low MA/low MCM;
• FA/MCM phenotypes: high FA/high MCM, high FA/low MCM, low FA/high MCM and low FA/low MCM; and
• MA/FCM phenotypes: high MA/high FCM, high MA/low FCM, low MA/high FCM and low MA/low FCM.

Statistical analysis

Continuous variables were expressed as mean ± standard deviation or median (interquartile range [IQR]) and compared using Student's t-test or the Mann–Whitney U test. Categorical variables were expressed as number (percentage) and compared by a χ² test or Fisher's exact test. Comparisons of variables across the body composition phenotype groups were performed by one-way ANOVA with Bonferroni's post hoc tests or the Kruskal–Wallis H test and the χ² test or Fisher's exact test, as appropriate. Receiver operating characteristic (ROC) curve analysis was performed to identify the optimal cut-off values of CSA and CM of thigh fat and muscle for predicting MACE. The cumulative incidence rates of MACE among the four composition phenotypes for thigh fat and muscle were revealed using Kaplan–Meier curves and compared statistically by log-rank tests. Hazard ratios (HRs) along with 95% confidence intervals (CIs) were calculated to evaluate the predictors of MACE after hip fracture using Cox proportional hazards models. Multivariate Cox regression analyses with the forward selection method were performed after adjusting for the covariates found to be significant in the univariate analysis or generally recognized variables known to affect the primary outcome. Statistical analyses were performed using SPSS Version 22.0 software (SPSS Inc., Chicago, IL, USA), and a two-sided P < 0.05 was considered statistically significant.

Results

Of the 384 elderly patients admitted to our hospital with hip fracture between April 2019 and December 2021, 28 subjects were excluded for the following reasons: death within 30 days after surgery (n = 6), malignancy at enrollment (n = 15), missing or inappropriate CT images for body composition analysis (n = 7) and insufficient medical records (n = 2). Finally, 356 patients (23.3% male; median age, 82.0 years [IQR, 76.0–86.0 years]) were eligible for analysis in this study. During a median follow-up of 13.1 months (IQR, 5.9–21.0 months), 72 patients (20.2%) experienced MACE after 30 days of hip fracture surgery. Of these, 67 (18.8%) patients died, 3 (0.8%) suffered from AMI and received PCI, 5 (1.3%) had stroke and 14 (4.0%) were hospitalized due to new-onset or worsening HF.

Patient characteristics

The baseline demographic and clinical characteristics of the overall population and comparisons between the MACE and non-MACE groups are summarized in Table 1. Compared with the non-MACE group, the MACE group was older (median age, 84.0 vs. 81.0 years, P = 0.001) and had a higher proportion of patients aged 80 years or older (73.6% vs. 53.9%, P = 0.002). Patients with MACE tended to be male (36.1% vs. 20.1%, P = 0.004); were more likely to have a history of CKD and dementia; and had a lower BMI, a higher Koval grade, lower haemoglobin and albumin levels and higher serum creatinine and C-reactive protein levels at admission compared with the non-MACE group. The median left ventricular ejection fraction (LVEF) in the total study population was 60.8% (IQR, 57.8–63.6%), and patients with MACE had lower LVEF with borderline statistical significance (P = 0.052) and greater E/e′ ratio than those without MACE. The intertrochanteric fracture was more frequently observed in the MACE group than in the non-MACE group, and patients with MACE had longer hospital stays. We found no statistically significant differences between the two groups in other comorbidity profiles, cardiac enzyme levels, surgery types and medications at discharge. Baseline characteristics according to the thigh composition phenotypes are also presented in Tables S2–S5.

Table 1 Baseline characteristics

Variables	Overall (n = 356)	MACE (n = 72)	Non-MACE (n = 284)	P valuea
Age (years)	82.0 (76.0–86.0)	84.0 (80.0–89.0)	81.0 (76.0–85.0)	0.001
>80 years, n (%)	206 (57.9)	53 (73.6)	153 (53.9)	0.002
Male, n (%)	83 (23.3)	26 (36.1)	57 (20.1)	0.004
Body mass index (kg/m2)	22.0 (19.9–24.4)	21.5 (18.9–23.4)	22.2 (20.1–24.9)	0.040
Koval grade	2.0 (1.0–4.0)	3.0 (1.0–4.0)	2.0 (1.0–4.0)	0.008
Comorbidity, n (%)
Diabetes mellitus	140 (39.3)	34 (47.2)	106 (37.3)	0.125
Hypertension	261 (73.3)	49 (68.1)	212 (74.6)	0.259
Dyslipidaemia	177 (49.7)	37 (51.4)	14 (49.3)	0.751
Coronary artery disease	28 (7.9)	8 (11.1)	20 (7.0)	0.252
Atrial fibrillation	30 (8.4)	9 (12.5)	21 (7.4)	0.164
Heart failure	20 (5.6)	7 (9.7)	13 (4.6)	0.146
Chronic kidney disease	36 (10.1)	14 (19.4)	22 (7.7)	0.003
Stroke	44 (12.4)	9 (12.5)	35 (12.3)	0.968
Dementia	64 (18.0)	21 (29.2)	43 (15.1)	0.006
COPD	8 (2.2)	4 (5.6)	4 (1.4)	0.056
Medications at discharge, n (%)
ACEi or ARB	164 (46.1)	28 (38.9)	136 (47.9)	0.171
Beta-blocker	26 (7.3)	5 (6.9)	21 (7.4)	0.896
Calcium channel blocker	145 (40.7)	23 (31.9)	122 (43.0)	0.089
Diuretics	72 (20.2)	15 (20.8)	57 (20.1)	0.886
Antiplatelet agent	122 (34.3)	30 (41.7)	92 (32.4)	0.139
Statin	168 (47.2)	34 (47.2)	134 (47.2)	0.995
Anticoagulant agent	22 (6.2)	4 (5.6)	18 (6.3)	1.000
Laboratory findings
Haemoglobin (g/dL)	11.3 ± 1.8	10.8 ± 1.9	11.4 ± 1.8	0.008
HbA1c (%)	6.5 (5.9–7.1)	6.6 (5.9–7.8)	6.3 (5.9–7.1)	0.072
Serum creatinine (mg/dL)	0.8 (0.6–1.3)	1.1 (0.7–1.5)	0.8 (0.6–1.1)	<0.0001
Albumin (mg/dL)	3.8 ± 0.4	3.7 ± 0.4	3.8 ± 0.4	0.001
C-reactive protein (mg/dL)	0.67 (0.14–3.31)	1.57 (0.17–4.68)	0.53 (0.14–3.06)	0.010
CK-MB (ng/mL)	2.2 (1.5–3.8)	2.7 (1.5–4.5)	2.2 (1.6–3.5)	0.877
Troponin I (ng/mL)	0.008 (0.005–0.015)	0.009 (0.005–0.018)	0.008 (0.005–0.013)	0.059
Echocardiographic parameters
LV ejection fraction (%)	60.8 (57.8–63.6)	59.5 (53.1–64.0)	61.0 (58.3–63.7)	0.052
E/e′avg	8.9 (7.0–10.9)	9.2 (7.8–11.9)	8.4 (6.7–10.7)	0.011
Fracture location, n (%)				0.017
Intertrochanteric region	157 (44.1)	42 (58.3)	115 (40.5)
Femur neck	145 (40.7)	24 (33.3)	121 (42.6)
Others	54 (15.2)	6 (8.3)	48 (16.9)
Surgery type, n (%)				0.751
Total hip arthroplasty	38 (10.7)	6 (8.3)	32 (11.3)
Hemiarthroplasty	105 (29.5)	21 (29.2)	84 (29.6)
Internal fixation	213 (59.8)	45 (62.5)	168 (59.2)
Length of hospital stay (days)	12.0 (9.0–16.0)	13.0 (10.0–17.3)	12.0 (9.2–15.0)	0.028

Abbreviations: ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; CK-MB, creatine kinase-myoglobin binding; COPD, chronic obstructive pulmonary disease; E/e′_avg, the ratio of early diastolic mitral inflow velocity to the averaged value of early diastolic mitral annulus velocities obtained from the septal and lateral sides of mitral annulus; HbA1c, haemoglobin A1c; LV, left ventricular; MACE, major adverse cardiovascular events.

^aComparisons between the MACE and non-MACE groups.

Table 2 demonstrates the CT-derived body composition measurements at the upper-thigh level of patients in the MACE and non-MACE groups. Patients with MACE had significantly lower FA (median, 193.7 vs. 226.2 cm², P < 0.0001) and lower FCM (median, 0.443 vs. 0.513, P = 0.001) than those in the non-MACE group. However, there were no significant differences in skeletal MA (median, 201.7 vs. 198.7 cm², P = 0.509), skeletal MCM (median, 0.660 vs. 0.650, P = 0.135) and FA–MA ratio (median, 1.1 vs. 1.0, P = 0.957) between the two groups.

Table 2 Computed tomography-derived upper-thigh body composition measurements

Variables	Overall (n = 356)	MACE (n = 72)	Non-MACE (n = 284)	P valuea
Fat area (cm2)	217.1 (175.0–259.0)	193.7 (152.6–232.4)	226.2 (179.6–266.0)	<0.0001
Fat compactness	0.494 (0.387–0.576)	0.443 (0.343–0.528)	0.513 (0.406–0.587)	0.001
Muscle area (cm2)	198.7 (175.4–227.2)	201.7 (176.9–232.0)	198.7 (174.9–226.1)	0.509
Muscle compactness	0.651 (0.584–0.717)	0.660 (0.608–0.734)	0.650 (0.577–0.711)	0.135
Fat–muscle area ratio	1.1 (0.8–1.3)	1.1 (0.8–1.4)	1.0 (0.8–1.3)	0.957

Abbreviation: MACE, major adverse cardiovascular events.

^aComparisons between the MACE and non-MACE groups.

Thigh composition measurements and major adverse cardiovascular events

Figure 2 illustrates the ROC curve analyses for the shape parameters of thigh fat and muscle to predict MACE in overall population. The optimal value of the thigh FA to predict MACE was <240.1 cm², with 86.1% sensitivity and 41.5% specificity (area under the curve [AUC] 0.639, 95% CI 0.569–0.709, P < 0.0001). Meanwhile, the optimal value of thigh FCM for predicting MACE was <0.477, with 65.3% sensitivity and 59.9% specificity (AUC 0.631, 95% CI 0.562–0.701, P = 0.001). However, the cut-off values of the thigh MA and MCM to predict MACE were not determined.

The univariate and multivariate Cox regression analyses for the association of thigh fat and muscle measurements with the risk of MACE are shown in Table 3. In the univariate analyses, low FA (<240.1 cm²) and low FCM (<0.477); old age; male gender; Koval grade; pre-existing AF, HF, CKD or dementia; serum levels of haemoglobin, haemoglobin A1c (HbA1c), creatinine, albumin, C-reactive protein and troponin I; LVEF; E/e′; and fracture location were related to the occurrence of MACE (all P < 0.05). After an adjustment for the confounding factors, both low FA (adjusted HR 2.988, 95% CI 1.388–6.436, P = 0.005) and low FCM (adjusted HR 2.000, 95% CI 1.100–3.634, P = 0.023), old age (adjusted HR 1.073, 95% CI 1.032–1.116, P < 0.0001), pre-existing AF (adjusted HR 3.673, 95% CI 1.654–8.157, P = 0.001) and CKD (adjusted HR 2.936, 95% CI 1.443–5.974, P = 0.003) were significantly associated with MACE after fragility hip fracture.

Table 3 Cox regression analyses for the risk of major adverse cardiovascular events after hip fracture surgery in elderly patients

Variables	Univariate analysis	Multivariate analysis
	Unadjusted HR (95% CI)	Adjusted HR	95% CI	P value
Low FA (<240.1 cm2)	4.021 (2.061–7.846)	2.988	1.388–6.436	0.005
Low FCM (<0.477)	2.533 (1.571–4.083)	2.000	1.100–3.634	0.023
MA (cm2)	1.001 (0.995–1.006)
Ln_MCM	2.306 (0.564–9.419)
Age (years)	1.060 (1.026–1.096)	1.073	1.032–1.116	<0.0001
Male	2.036 (1.258–3.296)
Body mass index (kg/m2)	0.938 (0.878–1.002)
Koval grade	1.241 (1.089–1.413)
Diabetes mellitus	1.446 (0.910–2.297)
Hypertension	0.725 (0.441–1.190)
Coronary artery disease	1.717 (0.822–3.583)
Atrial fibrillation	2.532 (1.255–5.110)	3.673	1.654–8.157	0.001
Heart failure	2.625 (1.200–5.742)
Stroke	1.041 (0.518–2.095)
Chronic kidney disease	3.063 (1.702–5.513)	2.936	1.443–5.974	0.003
Dementia	2.205 (1.322–3.678)
Medications
Antiplatelet agent	1.369 (0.857–2.187)
Statin	1.039 (0.654–1.651)
Anticoagulant agent	1.282 (0.466–3.525)
Laboratory findings
Haemoglobin (g/dL)	0.833 (0.743–0.933)
HbA1c (%)	1.354 (1.108–1.655)
Serum creatinine (mg/dL)	1.190 (1.082–1.309)
Albumin (mg/dL)	0.392 (0.232–0.663)
C-reactive protein (mg/dL)	1.069 (1.018–1.123)
Ln_troponin I	1.261 (1.009–1.577)
Echocardiographic parameters
LV ejection fraction (%)	0.973 (0.951–0.995)
E/e′avg	1.116 (1.040–1.197)
Fracture location
Intertrochanteric region	1.000 (reference)
Femur neck	0.566 (0.343–0.935)
Others	0.384 (0.163–0.903)	0.571	0.325–1.004	0.052

Abbreviations: CI, confidence interval; E/e′_avg, the ratio of early diastolic mitral inflow velocity to the averaged value of early diastolic mitral annulus velocities obtained from the septal and lateral sides of mitral annulus; FA, fat area; FCM, fat compactness; HbA1c, haemoglobin A1c; HR, hazard ratio; Ln, log-transformed; MA, muscle area; MCM, muscle compactness.

Shape phenotypes of thigh composition and major adverse cardiovascular events

Figure 3 shows the representative CT images of the four thigh composition phenotypes of FA/FCM. During the follow-up period, the MACE most commonly occurred in patients with low FA/low FCM, followed by those with low FA/high FCM, high FA/low FCM and high FA/high FCM in that order (33.6% vs. 19.8% vs. 13.8% vs. 6.1%, respectively, P < 0.0001) (Figure 4A). Moreover, patients with low FA/low FCM exhibited a significantly higher cumulative incidence rate for MACE than those in other fat phenotypes (log-rank P < 0.0001; Figure 4B). On the other hand, the thigh composition phenotypes of MA/MCM based on the median values (Figure S1) showed a similar incidence rate of MACE (P = 0.908; Figure S2A), and Kaplan–Meier curves for MACE did not significantly differ among them (log-rank P = 0.893; Figure S2B). When analysing by the thigh composition phenotypes of FA/MCM (Figure S3), patients with low FA were more likely to experience MACE than those with high FA regardless of MCM (P = 0.001; Figure S4A) and they had a higher cumulative incidence rate for MACE compared with those with high FA (log-rank P < 0.0001; Figure S4B). Of the thigh composition phenotypes of MA/FCM (Figure S5), the MACE more frequently occurred in patients with low FCM compared with those with high FCM irrespective of MA (P = 0.001; Figure S6A), and Kaplan–Meier curves revealed a higher cumulative incidence rate for MACE in the low FCM phenotypes than in the high FCM phenotypes (log-rank P < 0.0001; Figure S6B).

A multivariate Cox regression analysis for the association of shape phenotypes of thigh fat and muscle with the risk of MACE is presented in Table 4. The results of unadjusted and adjusted analyses are provided in Table S1. After adjustment for confounding factors, the thigh composition phenotype of low FA/low FCM (adjusted HR 3.131, 95% CI 1.808–5.422, P < 0.0001 [reference, high FA/high FCM]), along with old age (adjusted HR 1.066, 95% CI 1.026–1.108, P = 0.001), pre-existing AF (adjusted HR 3.467, 95% CI 1.576–7.625, P = 0.002) and CKD (adjusted HR 2.991, 95% CI 1.477–6.055, P = 0.002), and low serum albumin level (adjusted HR 0.471, 95% CI 0.239–0.928, P = 0.029) was associated with MACE.

Table 4 Association of shape phenotypes of thigh fat and muscle with major adverse cardiovascular events

Variables	Univariate analysis	Multivariate analysisa
	Unadjusted HR (95% CI)	Adjusted HR	95% CI	P value
FA/FCM phenotypes
High FA/high FCM	1.000 (reference)	1.000 (reference)
High FA/low FCM	2.688 (0.758–9.529)
Low FA/high FCM	3.830 (1.545–9.496)
Low FA/low FCM	6.762 (2.869–15.936)	3.131	1.808–5.422	<0.0001
MA/MCM phenotypes
High MA/high MCM	1.000 (reference)	1.000 (reference)
High MA/low MCM	1.029 (0.529–2.002)
Low MA/high MCM	1.206 (0.636–2.287)
Low MA/low MCM	0.930 (0.506–1.709)
FA/MCM phenotypes
High FA/high MCM	1.000 (reference)	1.000 (reference)
High FA/low MCM	2.012 (0.427–9.476)
Low FA/high MCM	6.375 (1.536–26.461)
Low FA/low MCM	7.499 (1.774–31.699)
MA/FCM phenotypes
High MA/high FCM	1.000 (reference)	1.000 (reference)
High MA/low FCM	2.283 (1.128–4.622)
Low MA/high FCM	1.066 (0.494–2.298)
Low MA/low FCM	3.322 (1.577–6.997)

Abbreviations: CI, confidence interval; FA, fat area; FCM, fat compactness; HR, hazard ratio; MA, muscle area; MCM, muscle compactness.

^aAdjusted for age, male gender, body mass index, Koval grade, diabetes mellitus, hypertension, coronary artery disease, atrial fibrillation, heart failure, stroke, chronic kidney disease, dementia, antiplatelet, statin, anticoagulant, haemoglobin, haemoglobin A1c, serum creatinine, albumin, C-reactive protein, log-transformed troponin I, left ventricular ejection fraction, E/e′_avg and surgery type.

Discussion

The major findings of our study were as follows. First, the occurrence of MACE after hip fracture in elderly patients was not uncommon and the shape features of thigh fat, FA and FCM, were related to MACE. Second, low FA and low FCM, along with old age, pre-existing AF and CKD, independently increased the risk of MACE. Third, among thigh composition phenotypes, the thigh fat phenotype of low FA/low FCM was a significant predictor of MACE following fragility hip fracture.

It is well known that elderly patients with hip fracture are more likely to experience adverse CV events compared with the general population. Chiang et al. found a higher risk of subsequent AMI in patients with hip fracture than in those without fracture.⁵ In a large, population-based cohort study, elderly patients were at increased risk of stroke for up to 10 years following hip fracture.⁷ However, most studies evaluating the predictors of adverse CV events in this population showed a lack of detailed data on fragile medical conditions^5–7 and focused on the early CV complications during hospitalization or the perioperative period.^21,22 Considering that thigh fat and muscle contribute to the cardiometabolic risk profile,²³ the assessment of thigh composition could provide useful clinical information on the future risk of CVD, as well as nutritional status. Significant relationships between age-related changes in thigh composition and mortality have been identified^24,25; however, its prognostic value as a predictor of MACE, especially in elderly patients after hip fracture, has not been elucidated. To the best of our knowledge, this is the first study investigating the associations of thigh fat and muscle with MACE among elderly patients with hip fracture.

Recently, Kim et al. conducted a retrospective cohort study of 876 elderly patients with proximal femur fracture and found that upper-thigh subcutaneous FA (SFA) measured on preoperative pelvic bone CT scan but not skeletal MA or muscle density was an independent predictor of 1-year mortality after surgery.²⁰ Although numerous research has revealed a significant relationship between sarcopenia and functional decline or mortality in older people,²⁶ several studies showed inconsistent results.^27,28 Furthermore, some prospective cohort studies of elderly individuals showed no significant associations between loss of lower extremity skeletal muscle mass and increased risk of long-term mortality.^29,30 Our study demonstrated a significantly lower upper-thigh FA in patients with MACE than in those without, but similar skeletal MA between the two groups, which is consistent with findings of the aforementioned studies. In this study, we additionally adopted CM, a computerized measure of two-dimensional shape feature for upper-thigh tissue composition, and found that, even with the same FA, patients with low FCM tended to have a higher incidence of MACE compared with those with high FCM (Figure 4). The CM is a quantitative morphological parameter to describe the compactness of an object relative to a sphere³¹ and has been used to characterize tumour phenotypes and to predict treatment response or overall survival in cancer patients.^32,33 In the present study, we classified the shape features of upper-thigh fat and muscle into four groups using CT-based CM in addition to CSA and identified the thigh fat phenotype of low FA/low FCM as an independent predictor of MACE in elderly patients with hip fracture.

A growing body of evidence has shown the role of adipose tissue as a complex endocrine organ secreting a variety of bioactive substances called adipokines that regulate energy homeostasis.³⁴ Previous studies have reported the beneficial effects of subcutaneous adipose tissue on the lipid and glucose metabolism, whereas ectopic fat depots outside of the subcutaneous tissue have been associated with pro-inflammatory cytokines linked to the risk of CVD.¹⁴ Intermuscular adipose tissue (IMAT) has been broadly defined as the visible lipid storage in adipocytes located within muscle fibres (also referred to as intramuscular fat) or between muscle groups (or perimuscular fat) underneath the deep fascia, which is generally considered to be an ectopic fat similar to abdominal visceral adipose tissue.²³ Thigh IMAT, along with thigh muscle loss, had a significant relationship with cardiometabolic risk profiles, including insulin resistance,^13,35 unfavourable lipid levels³⁶ and inflammation.^37,38 Low thigh subcutaneous fat was also independently associated with cardiometabolic risk.³⁹ Delmonico et al. reported the age-related increase in thigh IMAT with a progressive loss of thigh muscle and quality,²⁴ and these longitudinal changes in thigh composition showed a strong association with higher mortality.²⁵ However, the amount of IMAT is much smaller than that of subcutaneous fat, accounting for only 8% of the thigh adipose tissue.³⁵ In addition, particularly, intramuscular fat, under the limited resolution of CT, cannot be directly measured or accurately distinguished from perimuscular fat. Our study revealed different baseline characteristics according to the thigh composition phenotypes based on FA or MA and FCM (Tables S2 and S5) and found higher proportions of comorbidities in patients with low FCM compared with those with high FCM. Moreover, we measured thigh IMAT and SFA in a subset of the participants (n = 100) and evaluated the association of FCM or MCM with thigh IMAT and subcutaneous fat. As shown in Figure S7, thigh FCM was inversely correlated with thigh IMAT area (r = −0.446, P < 0.0001) and thigh IMAT/total FA ratio (r = −0.668, P < 0.0001) and positively correlated with thigh SFA (r = 0.694, P < 0.0001). Thigh MCM also showed significant negative correlations with thigh IMAT area (r = −0.532, P < 0.0001) and thigh IMAT/total FA ratio (r = −0.417, P < 0.0001) but had no relationship with thigh SFA (Figure S8). These findings suggest that thigh FCM may reflect thigh fat distribution and, accordingly, could be used as a surrogate marker of metabolic dysfunction related to adverse CV outcomes in elderly patients after hip fracture. Our study also confirmed independent associations of well-established CV risk factors including old age, AF, CKD and low serum albumin level with MACE in this population (Table S1).

The present study had several inherent limitations of a single-centre, retrospective study with potential biases that could affect clinical outcome. First, our data were derived from a population-based cohort of elderly patients with hip fracture, and detailed information on lipid profiles or insulin resistance-associated metabolic parameters at admission was not available. In addition, we did not assess sarcopenia-defining functional parameters such as handgrip strength or gait speed that are known to predict adverse CV events and mortality. However, muscle CSA has shown good correlations with indices of muscle function and also has been reported as an indicator of frailty.⁴⁰ Second, although we could not analyse muscle density in individuals, thigh muscle was measured based on predefined skeletal muscle HU thresholds that have been widely used to diagnose sarcopenia. Third, due to the limitation of manual region of interest (ROI) drawing of IMAT, we could not investigate the relationships of FCM with thigh IMAT and SFA in all subjects. The validation using IMAT measurement is needed in the future. Fourth, our study was conducted between 2019 and 2021 and overlapped with the COVID-19 pandemic; therefore, regular follow-up at outpatient clinics after hospital discharge was not available for some participants. Nevertheless, we obtained complete clinical outcome data of all patients via medical records review or telephone interview. Finally, changes in thigh tissue composition after hip surgery and the relationships with cardiometabolic risk factors were not assessed during follow-up.

Conclusions

In elderly patients with hip fracture, the shape features of thigh fat but not thigh muscle were associated with MACE occurrence after hip surgery, and the thigh composition phenotype of low FA/low FCM was an independent predictor of MACE. Our findings indicate that shape features of thigh fat combined with CSA and CM could be a useful prognostic marker for risk stratification of adverse CV outcomes and simultaneously reinforce the importance of close surveillance based on thigh fat phenotype after fragility hip fracture. However, the physiological implication of FCM in relation to thigh fat distribution, particularly thigh IMAT, should be validated in the future, and further large prospective studies with a long-term follow-up are needed to confirm the prognostic value of thigh FA and FCM for identifying elderly patients with hip fracture at a higher risk of MACE.

Acknowledgements

All authors of this manuscript comply with the guidelines of ethical authorship and publishing in the Journal of Cachexia, Sarcopenia and Muscle.⁴¹

Conflict of interest statement

The authors declare no conflicts of interest.