Identifying sex differences in predictors of epicardial fat cell morphology

Figures & data

Table 1. Characteristics and fat cell sizes of total post-mortem cases and in female and male cases

Variable (median ± IQR)	Total (N = 118)	Male (N = 81)	Female (N = 37)	M vs. F P value
Case information
Age (years)	57.0 ± 23.5	56.5 ± 22.0	60.0 ± 32.5	0.38
BMI (kg/m^2)	27.2 ± 6.9	26.6 ± 6.4	27.9 ± 8.1	0.73
Fat cell sizes
Epicardial (x10^3 µm^2)	2.8 ± 1.9	2.8 ± 1.8	2.8 ± 2.1	0.75
Paracardial (x10^3 µm^2)	2.8 ± 2.4	3.1 ± 2.4	2.6 ± 2.5	0.47
Appendix (x10^3 µm^2)	5.2 ± 4.4^****####	5.4 ± 4.1^****####	4.3 ± 3.3^*##	0.08
Size ANOVA P value	<0.0001	<0.0001	<0.0001

Due to non-normal data distributions, variables are presented as median values ± interquartile ranges (IQR). Male (M) vs. female (F) differences (i.e. column differences) were assessed using Mann–Whitney test. Differences in median fat cell sizes within groups (i.e. row differences) were assessed using Kruskal-Wallis (ANOVA result as indicated) with Dunn’s multiple comparisons test. *P < 0.001, ****P < 0.0001 vs. group epicardial fat cell size, ^##P < 0.01, ^####P < 0.0001 vs. group paracardial fat cell size.

Figure 1. Correlation matrix of information available from all post-mortem cases and fat cell sizes (FCS).

Figure 2. Univariable correlations of fat cell sizes with body mass index (BMI) and age from all post-mortem cases after sex separation.

Female and male fat cell size correlations with BMI and age were performed using Pearson correlation. Regression line is presented for visual clarity. Relationships between BMI and epicardial, paracardial, and appendix fat cell sizes are shown from a, b and c, respectively. Sex-dependent associations of epicardial, paracardial, and appendix fat cell sizes with age are shown in d, e and f, respectively. Fat cell sizes for each adipose depot were natural log (ln) transformed to achieve a normal distribution. N = 37 females, N = 81 males.

Table 2. Stepwise linear regression for predictors of fat cell sizes in all cases and after sex separation

Adipose depot	Epicardial		Paracardial		Appendix
All cases
Predictor	β	P	β	P	β	P
BMI	0.015	0.047	0.029	0.001	0.053	<0.0001
Age	0.007	0.009	0.012	<0.0001	0.007	0.011
Female					−0.251	0.009
Adjusted R^2	0.070		0.203		0.290
Males
Predictor	β	P	β	P	β	P
BMI	0.019	0.047	0.038	0.0009	0.050	<0.0001
Age	0.008	0.027	0.012	0.003	0.009	0.013
Adjusted R^2	0.095		0.221		0.284
Females
Predictor	β	P	β	P	β	P
BMI	No predictors				0.052	0.001
Age			0.012	0.008
Adjusted R^2			0.163		0.242

Fat cell sizes were natural log transformed to achieve normal data distributions. Parameter estimates (β) are of transformed data. Model for All cases in included body mass index (BMI), age (years), female (yes). Model for Males and Females only included BMI and age. For variable to enter model α = 0.05, for variable to leave model α = 0.05. N = 118 post-mortem cases total, N = 81 males, N = 37 females.

Figure 3. Univariable correlations of fat cell sizes with the related fat volume from the subset of post-mortem cases.

Female and male fat cell size correlations with the anatomically related fat volume/area/thickness were performed using Pearson correlation. Regression line is presented for visual clarity. (a) Relationship between epicardial fat cell size with total epicardial fat volume. (b) Relationship between paracardial fat cell size with extra-pericardial fat volume. (c) Relationship between appendix fat cell size and visceral fat volume. Data were natural log transformed when raw data distribution was non-normal.N = 22 females, N = 48 males.

Table 3. Stepwise linear regression for predictors of fat cell sizes in subset of post-mortem cases with available fat volumes/area

Adipose depot	Epicardial		Paracardial		Appendix
All cases
Predictor	β	P	β	P	β	P
BMI			0.038	0.001	0.038	0.002
Age			0.016	0.0002
Female
Total epicardial fat	0.219	0.008
Extra-pericardial fat
Visceral fat					0.095	0.032
Adjusted R^2	0.088		0.236		0.288
Males
Predictor	β	P	β	P	β	P
BMI			0.041	0.006
Age			0.016	0.008
Total epicardial fat	0.273	0.003
Extra-pericardial fat
Visceral fat					0.205	<0.0001
Adjusted R^2	0.161		0.237		0.340
Females
Predictor	β	P	β	P	β	P
BMI	No predictors				0.046	0.021
Age
Total epicardial fat
Extra-pericardial fat			0.012	0.001
Visceral fat
Adjusted R^2			0.429		0.200

Fat cell sizes and volumes were natural log transformed according to achieve normal data distribution. Parameter estimates (β) are of transformed data. Model for All cases in included body mass index (BMI), age (years), female (yes), and the anatomically related fat deposition measure. Total epicardial fat volume was allocated the depot related to epicardial fat cell size. Extra-pericardial fat volume was allocated the depot related to paracardial fat cell size. Visceral fat volume was allocated the depot related to appendix fat cell size. Model for Males and Females included variables as above without female (yes). For variable to enter model α = 0.05, for variable to leave model α = 0.05. N = 70 post-mortem cases total, N = 48 males, N = 22 females.

Table 4. Stepwise linear regression analysis of fat cell sizes (FCS) as predictors of the size of fat cells in other depots in all cases

Adipose depot	Epicardial		Paracardial		Appendix
All cases
Predictor	β	P	β	P	β	P
BMI					0.034	<0.0001
Age			0.006	0.0007
Female					−0.187	0.005
Epicardial FCS			0.553	<0.0001	0.251	0.016
Paracardial FCS	0.62	<0.0001			0.508	<0.0001
Appendix FCS			0.425	<0.0001
Adjusted R^2	0.572		0.720		0.668
Males
Predictor	β	P	β	P	β	P
BMI					0.024	0.002
Age
Epicardial FCS			0.657	<0.0001
Paracardial FCS	0.661	<0.0001			0.680	<0.0001
Appendix FCS			0.448	<0.0001
Adjusted R^2	0.665		0.771		0.652
Females
Predictor	β	P	β	P	β	P
BMI					0.047	<0.0001
Age			0.010	0.002
Epicardial FCS			0.364	0.026	0.465	0.006
Paracardial FCS	0.600	<0.0001			0.415	0.004
Appendix FCS			0.420	0.001
Adjusted R^2	0.434		0.630		0.694

Fat cell sizes were natural log transformed according to achieve a normal distribution. Parameter estimates (β) are of transformed data. Model for All cases in included body mass index (BMI), age (years), female (yes), and all other fat cell sizes. Model for Males and Females included variables as above without female (yes). For variable to enter model α = 0.05, for variable to leave model α = 0.05. N = 70 post-mortem cases total, N = 48 males, N = 22 females.