Baseline characteristics
Table 1 presents an analysis of a total of 2580 participants from the United States. Individuals exhibiting PhenoAgeAccel tended to be older, identify as Hispanic, and to have elevated levels of BMI, HbA1c, insulin, waist circumference, subcutaneous fat area, and visceral fat area. Furthermore, they demonstrated a higher prevalence of hypertension, cardiovascular diseases, and diabetes and were more likely to engage in smoking behavior. Additionally, these individuals had lower levels of HDL-C, education level, and income (P < 0.05).
Association between body fat area and PhenoAgeAccel
Table 2 illustrates the association between body fat area and PhenoAgeAccel. In this analysis, subcutaneous fat area (SFA) and visceral fat area (VFA) were categorized based on quartiles. VFA showed a marked positive association with PhenoAgeAccel after adjusting for variables such as age, gender, race,exercise, alcohol consumption, smoking status, marital status, education level, income, HbA1c, and histories of hypercholesterolemia, hypertension, diabetes, and cardiovascular disease. When compared to Q1 (the lowest quartile), the odds ratio for Q4 (the highest quartile) was 8.339 (95%CI: 5.559 to 12.510), with P for trend <0.001. For SFA, no statistically significant association was found at the Q2 level. However, significant associations were identified at both Q3 and Q4 levels. In Model 3, compared to Q1, the odds ratio for Q4 was 7.756 (95%CI: 5.250 to 11.458), with P for trend <0.001.
Table1 Baseline characteristics according to age and phenotypic age.
|
PhenoAge Deceleration/Stasis (n=1257)
|
PhenoAge Acceleration(n=1323)
|
P value
|
Age,years
|
37.94±12.34
|
39.37±12.32
|
0.044
|
Gender,%
|
|
|
0.331
|
Male
|
627(49.9)
|
693(52.4)
|
|
Female
|
630(50.1)
|
630(47.6)
|
|
Race,%
|
|
|
0.035
|
Mexican American
|
114(9.1)
|
148(11.2)
|
|
Other Hispanic
|
87(6.9)
|
120(9.1)
|
|
Non-Hispanic
|
1004(79.9)
|
974(73.6)
|
|
Other Race
|
52(4.1)
|
81(6.1)
|
|
Income,%
|
|
|
<0.001
|
Below poverty
|
132(10.5)
|
241(18.2)
|
|
At or above poverty
|
1032(82.1)
|
991(74.9)
|
|
Education level,%
|
|
|
<0.001
|
College graduate or above
|
490(39.0)
|
275(20.8)
|
|
High school grade or equivalent college
|
611(48.6)
|
860(65.0)
|
|
Less than 12th grade
|
156(12.4)
|
188(14.2)
|
|
Marital status,%
|
|
|
0.088
|
Married
|
640(50.9)
|
606(45.8)
|
|
others
|
557(44.3)
|
668(50.5)
|
|
Smoke ,%
|
460(36.6)
|
614(46.4)
|
0.002
|
Drink ,%
|
553(44.0)
|
527(39.8)
|
0.139
|
Regular exercise,%
|
1024(81.5)
|
1062(80.3)
|
0.646
|
TC, mmol/L
|
4.84±0.98
|
4.86±0.98
|
0.829
|
HDL-C, mmol/L
|
1.49±0.42
|
1.31±0.38
|
<0.001
|
LDL-C, mmol/L
|
2.86±0.87
|
2.96±0.87
|
0.084
|
Insulin, μU/mL
|
8.85±6.21
|
15.40±16.38
|
<0.001
|
HbA1c,%
|
5.28±0.38
|
5.77±1.17
|
<0.001
|
BMI, kg/m2
|
26.39±5.24
|
31.71±7.54
|
<0.001
|
WC, cm
|
91.47±13.78
|
104.68±17.82
|
<0.001
|
VFA, m2
|
0.84±0.49
|
1.25±0.65
|
<0.001
|
SFA, m2
|
2.82±1.42
|
4.00±1.87
|
<0.001
|
Self-report hypertension ,%
|
205(16.3)
|
382(28.9)
|
<0.001
|
Self-report diabetes,%
|
19(1.5)
|
139(10.5)
|
<0.001
|
Self-report hypercholesterolemia,%
|
280(22.3)
|
325(24.6)
|
0.408
|
Self-report cardiovascular diseases ,%
|
25(2.0)
|
73(5.5)
|
0.007
|
Continuous variables are presented as means ± standard deviations, and categorical variables are expressed as percentages. All analyses accounted for the complex survey design.
Abbreviations: TC, total cholesterol; LDL, low-density lipoprotein; HDL, high-density lipoprotein; HbA1c, Hemoglobin A1c; BMI, body mass index; WC, waist circumference; VFA, visceral fat area; SFA, subcutaneous fat area.
Table 2 Association of body fat area with phenoage acceleration.
|
|
Model1
OR(95%CI),P
|
Model2
OR(95%CI),P
|
Model3
OR(95%CI),P
|
VFA
|
Q1
|
Ref.
|
Ref.
|
Ref.
|
|
Q2
|
1.714(1.258, 2.334),
0.001
|
1.898(1.380,2.610),
<0.001
|
1.990(1.436,2.757),
<0.001
|
|
Q3
|
3.377(2.475,4.067),
<0.001
|
4.019(1.380,5.699), <0.001
|
4.053(2.864,5.736), <0.001
|
|
Q4
|
6.316(4.564,8.740), <0.001
|
8.713(5.956,12.748), <0.001
|
8.339(5.559,12.510), <0.001
|
|
P for trend
|
<0.001
|
<0.001
|
<0.001
|
SFA
|
Q1
|
Ref.
|
Ref.
|
Ref.
|
|
Q2
|
1.098(0.804,1.597), 0.557
|
1.149(0.827,1.597),
0.407
|
1.308(0.930,1.840), 0.123
|
|
Q3
|
2.338(1.717,3.182), <0.001
|
2.800(2.000,3.922), <0.001
|
3.020(2.133,4.276), <0.001
|
|
Q4
|
5.746(4.132,7.992), <0.001
|
7.796(5.356,11.346), <0.001
|
7.756(5.250,11.458), <0.001
|
|
P for trend
|
<0.001
|
<0.001
|
<0.001
|
Convert VFA and SFA into categorical data based on quartiles. Model1 did not include any covariable. Model2 was adjusted for age, race, gender. Model3 was based on Model 2 and adjusted for exercise, drink, smoke, marriage status, education level, income, HbA1c, hypercholesterolemia history, hypertension history, diabetes history, and cardiovascular diseases history.. All data accounted for complex survey designs.
Abbreviations: VFA, visceral fat area; SFA, subcutaneous fat area; OR, odds ratio; CI, confidence interval.
Subgroup analysis
Fig.3 displays that interaction tests on the data revealed a differential relationship between VFA and PhenoAgeAccel across subgroups defined by chronological age, gender, and diabetes status. In the chronological age subgroup (P-interaction = 0.002), this association was most pronounced in individuals aged 18-44 years. For the gender (P-interaction = 0.026) and self-report diabetes (P-interaction <0.001) subgroups, males and individuals with diabetes exhibited a stronger association between VFA and PhenoAgeAccel.
The relationship between SFA and PhenoAgeAccel varied across subgroups defined by chronological age, alcohol consumption, and diabetes status. In the chronological age subgroup (P-interaction = 0.035), this association was most pronounced in individuals aged 18-44 years. For the drink (P-interaction = 0.004) and self-report diabetes (P- interaction <0.001) subgroups, individuals who consume alcohol and those without diabetes exhibited a stronger association between SFA and PhenoAgeAccel.
RCS
Using chronological age as a subgroup, we conducted RCS tests to evaluate the dose-response relationship between body fat area and PhenoAgeAccel. Fig.4 demonstrates the results that in the population aged 18-59 years, both VFA and SFA exhibited non-linear associations with PhenoAgeAccel, with thresholds of 0.925m2 and 3.137 m2, respectively. When analyzed by age subgroup, the RCS revealed that for SFA and PhenoAgeAccel, non-linear relationships were observed in the 18-44 years and 45-59 years groups, with corresponding thresholds of 2.969 m2 and 3.394 m2. In contrast, for VFA and PhenoAgeAccel, a non-linear relationship was identified in individuals aged 18-44 years with threshold of 0.769 m2, while a linear relationship was noted in those aged 45-59 years with threshold of 1.220 m2.
Mediation effect analysis
As illustrated in Fig.5, HOMA-IR and HDL-C acted as partial mediators in the relationship between body fat area and PhenoAgeAccel. For HOMA-IR, as displayed in Table3, the mediating effect accounted for 10.2% of the relationship between VFA and PhenoAgeAccel in participants aged 18-59 years. When stratified by chronological age, the mediating effect of HOMA-IR was 13.4% in the 18-44 years and 6.2% in the 45-59 years. Additionally, HOMA-IR's mediating effect on the relationship between SFA and PhenoAgeAccel was measured at 4.7%. When stratified by chronological age, the mediating effect of HOMA-IR was 6.9% in the 18-44 years and 3.5% in the 45-59 years. Notably, the mediating effect of HOMA-IR is more pronounced in individuals aged 18-44 years compared to those aged 45-59 years.
Regarding HDL-C, as displayed in Table4, the mediating effect accounted for 15.3% of the relationship between VFA and PhenoAgeAccel in non-elderly adults (individuals aged 18-59 years). When stratified by chronological age, the mediating effect of HDL-C was found to be 8.4% in the 18-44 years old and 21.7% in the 45-59 years old. In addition, HDL-C's mediating effect on the relationship between SFA and PhenoAgeAccel was measured at 7.2%. Further analysis revealed that HDL-C did not exhibit a significant mediating effect in the 18-44 years old but demonstrated an effect of 11.6% in those aged 45-59 years old. Notably, the mediating effect of HDL-C was more significant in individuals aged 45-59 years than in those aged 18-44 years.
Table 3 Association of body fat area with PhenoAgeAccel mediated by HOMA-IR.
|
|
OR(95%CI)
|
Body Fat Area
|
Effect
|
Total
|
Age≥18, <45 years
|
Age≥45, <60 years
|
VFA
|
ACME
|
0.020(0.010,0.040)
|
0.037(0.023,0.060)
|
0.009(0.002,0.036)
|
|
ADE
|
0.177(0.133,0.208)
|
0.238(0.183,0.288)
|
0.142(0.086,0.191)
|
|
Prop. Mediated
|
10.2%
|
13.4%
|
6.2%
|
SFA
|
ACME
|
0.004(0.001,0.008)
|
0.007(0.003,0.012)
|
0.003(0.001,0.010)
|
|
ADE
|
0.086 (0.077,0.096)
|
0.091 (0.079,0.123)
|
0.074(0.054,0.092)
|
|
Prop. Mediated
|
4.7%
|
6.9%
|
3.5%
|
All models were adjusted for age, race, gender, exercise, drink, smoke, marriage status, education level, income, HbA1c, hypercholesterolemia history, hypertension history, diabetes history, and cardiovascular diseases history.
Abbreviations: ACME, stands for average causal mediation effects; ADE, stands for average direct effects; OR, odds ratio; CI, confidence interval.
Table 4 Association of body fat area with PhenoAgeAccel mediated by HDL-C.
|
|
OR(95%CI)
|
Body Fat Area
|
Effect
|
Total
|
Age≥18, <45 years
|
Age≥45, <60 years
|
VFA
|
ACME
|
0.030(0.019,0.041)
|
0.023(0.002,0.044)
|
0.033(0.016,0.049)
|
|
ADE
|
0.167(0.136,0.208)
|
0.251(0.199,0.307)
|
0.118(0.062,0.170)
|
|
Prop. Mediated
|
15.3%
|
8.4%
|
21.7%
|
SFA
|
ACME
|
0.007(0.003,0.010)
|
0.003(-0.002,0.008)
|
0.009(0.004,0.014)
|
|
ADE
|
0.084(0.073,0.095)
|
0.095(0.081,0.107)
|
0.067(0.049,0.084)
|
|
Prop. Mediated
|
7.2%
|
3.2%
|
11.6%
|
All models were adjusted for age, race, gender, exercise, drink, smoke, marriage status, education level, income, HbA1c, hypercholesterolemia history, hypertension history, diabetes history, and cardiovascular diseases history.
Abbreviations: ACME, stands for average causal mediation effects; ADE, stands for average direct effects; OR, odds ratio; CI, confidence interval.
Joint associations of body fat area and HOMA-IR, HDL-C with PhenoAgeAccel
We established the HOMA-IR threshold for diagnosing insulin resistance at 2.73[20]. As illustrated in Fig.6, we investigated the relationship between body fat area combined with HOMA-IR and PhenoAgeAccel across chronological age subgroups. The results indicated that in the interaction between VFA and HOMA-IR, individuals aged 18-44 years with HOMA-IR ≥2.73 and VFA >0.925 m² had a higher risk of PhenoAgeAccel compared to those with HOMA-IR <2.73 and VFA ≤0.925 m² (OR: 4.541, 95% CI: 3.273 to 6.301). In the participants aged 45-59 years, a significant interaction promoting PhenoAgeAccel occurred only when both HOMA-IR≥2.73 and VFA>0.925m2 (OR: 2.690, 95%CI: 1.739 to 4.161). Furthermore, in the analysis of SFA combined with HOMA-IR, the risk of PhenoAgeAccel was greatest among individuals aged 18-44 years exhibiting both HOMA-IR≥2.73 and SFA>3.137m2 relative to those with HOMA-IR<2.73 and SFA≤3.137m2 (OR: 5.705, 95%CI: 4.125 to 7.892). For participants aged 45-59 years, an interaction promoting PhenoAgeAccel was observed when both HOMA-IR>1.67 and SFA>0 .925m2 were present; notably, this group exhibited the highest risk for PhenoAgeAccel when combining conditions of HOMA-IR≥2 .73 and SFA>3 .137m2 compared to those with lower values (OR: 3.017 , 95%CI : 1.934 to 4.706 ).
We categorized HDL-C into four quartiles (Q1-Q4) based on its levels. As illustrated in Fig.7, we explored the association between body fat area and HDL-C in conjunction with PhenoAgeAccel across different chronological age subgroups. The findings indicated that in the interaction between VFA and HDL-C, for individuals aged 18-59 years, VFA≤0.925 m2 and HDL-C>1.60, ≤3.90 mmol/L were identified as protective factors for PhenoAgeAccel (OR: 0.576, 95%CI: 0.393 to 0.845), whereas VFA>0.925 m2 and HDL-C≥0.16, <1.09 mmol/L were recognized as risk factors. An elevate in HDL-C levels was associated with a lower risk of PhenoAgeAccel due to their interaction effects. For individuals aged 18-44 years, VFA≤0.925 m2 and HDL-C>1.60, ≤3.90 mmol/L also served as protective factors for PhenoAgeAccel (OR: 0.583, 95%CI: 0.377 to 0.903). For individuals aged 45-59 years, the interaction between VFA and HDL-C did not demonstrate statistical significance regarding the occurrence of PhenoAgeAccel. In contrast, for individuals aged 18-59 years, the combination of SFA≤3.137 m2 with HDL-C>1.60, ≤3.90 mmol/L was identified as a protective variable for PhenoAgeAccel (OR: 0.690, 95% CI: 0.482 to 0.988), whereas SFA>3.137 m2 combined with HDL-C≥0.16, <1.09 mmol/L constituted a risk factor for PhenoAgeAccel. Notably, a rise in HDL-C levels was linked to a lower risk of developing PhenoAgeAccel, attributed to their interaction effects. For those aged 18-44 years, SFA>3.137 m2 combined with HDL-C≥0.16, <1.09 mmol/L was also recognized as a risk factor for PhenoAgeAccel . For those aged 45-59 years , the combination of SFA≤3.137 m2 and HDL-C>1.60,≤3.90 mmol/L served as a protective factor against PhenoAgeAccel (OR: 0.517, 95%CI: 0.277 to 0.963), while SFA>3.137 m2 combined with HDL-C≥0.16,<1.09 mmol/L constituted a risk factor for PhenoAgeAccel.
Incremental predictive values of body fat area for PhenoAgeAccel
Table5 shows that in individuals aged 18-44 years, the basic model combined with VFA+HDL-C (AUC: 0.561), VFA+HOMA-IR (AUC: 0.671), SFA (AUC: 0.649), SFA+HDL-C (AUC: 0.636), and SFA+HOMA-IR (AUC: 0.698) demonstrated statistically significant differences compared to the basic model + VFA (AUC: 0.579) (P<0.001). The NRI and IDI tests further revealed that incorporating VFA+HOMA-IR, SFA, SFA+HDL-C, and SFA+HOMA-IR into the basic model enhanced its predictive capability for PhenoAgeAccel relative to the basic model + VFA (P<0.05). In individuals aged 45-59 years, the combination of the basic model with VFA+HDL-C (AUC: 0.578), VFA+HOMA-IR (AUC: 0.684), and SFA+HOMA-IR (AUC: 0.681) also exhibited statistically significant differences from the basic model + VFA (AUC: 0.607) (P<0.001). Additionally, the NRI and IDI tests indicated that introducing VFA+HOMA-IR and SFA+HOMA-IR into the basic model improved its ability to predict PhenoAgeAccel compared to the basic model + VFA (P<0.05).
Table 5 Incremental predictive values for PhenoAgeAccel.
Age≥18,<45years
|
AUC(95%CI)
|
P
|
IDI(95%CI)
|
P
|
NRI(95%CI)
|
P
|
Basic model
|
|
|
|
|
|
|
+VFA
|
0.579
(0.551,0.607)
|
Ref.
|
Ref.
|
Ref.
|
Ref.
|
Ref.
|
+VFA+HDL-C
|
0.561
(0.533,0.589)
|
<0.001
|
0.001
(-0.001,0.003)
|
0.288
|
-0.001
(-0.019,0.017)
|
0.876
|
+VFA+HOMA-IR
|
0.671
(0.644,0.697)
|
<0.001
|
0.021
(0.014,0.028)
|
<0.001
|
0.038
(0.001,0.070)
|
0.018
|
+SFA
|
0.649
(0.622,0.675)
|
<0.001
|
0.043
(0.033,0.053)
|
<0.001
|
0.132
(0.087,0.178)
|
<0.001
|
+SFA+HDL-C
|
0.636
(0.609,0.663)
|
<0.001
|
0.043
(0.033,0.053)
|
<0.001
|
0.125
(0.079,0.171)
|
<0.001
|
+SFA+HOMA-IR
|
0.698
(0.672,0.723)
|
<0.001
|
0.051
(0.040,0.062)
|
<0.001
|
0.130
(0.085,0.175)
|
<0.001
|
Age≥45,<60years
|
AUC(95%CI)
|
P
|
IDI(95%CI)
|
P
|
NRI(95%CI)
|
P
|
Basic model
|
|
|
|
|
|
|
+VFA
|
0.607
(0.572,0.643)
|
Ref.
|
Ref.
|
Ref.
|
Ref.
|
Ref.
|
+VFA+HDL-C
|
0.578
(0.542,0.614)
|
<0.001
|
0.011
(0.004,0.017)
|
0.001
|
0.019
(-0.019,0.058)
|
0.325
|
+VFA+HOMA-IR
|
0.684
(0.651,0.717)
|
<0.001
|
0.014
(0.007,0.021)
|
<0.001
|
0.038
(0.002,0.073)
|
0.039
|
+SFA
|
0.618
(0.583,0.653)
|
0.207
|
0.023
(0.012,0.033)
|
<0.001
|
0.088
(0.036,0.140)
|
0.001
|
+SFA+HDL-C
|
0.595
(0.559,0.630)
|
0.192
|
0.032
(0.021,0.044)
|
<0.001
|
0.097
(0.044,0.150)
|
<0.001
|
+SFA+ HOMA-IR
|
0.681
(0.648,0.714)
|
<0.001
|
0.031
(0.020,0.041)
|
<0.001
|
0.098
(0.046,0.150)
|
<0.001
|
The basic model involving age, gender, race, exercise, drink, smoke, marriage status, education level, income, HbA1c, hypercholesterolemia history, hypertension history, diabetes history, and cardiovascular diseases history.
Abbreviations: AUC, area under the curve; NRI, net reclassification improvement; IDI, integrated discrimination improvement; OR, odds ratio; CI, confidence interval.