Association of serum albumin and obesity: Mediation effect of extracellular fluid, percent body fat and C-reactive protein

Abstract

This study aims to further validate the association of serum albumin and obesity in a larger population and explore potential underlying mediating factors. Our study utilized data from 5881 participants in the National Health and Nutrition Examination Survey (NHANES 1999–2004). Multivariable weighted logistic regression analyses were conducted to assess the association between serum albumin and obesity, while evaluating the potential mediating effects of extracellular fluid (ECF), percentage body fat (PFAT), and C-reactive protein (CRP). Secondary analyses encompassed restricted cubic splines (RCS) and subgroup analyses. Every 1 g/L increase in serum albumin concentration was associated with a 17% reduction in obesity prevalence (adjusted odds ratio = 0.83; 95% confidence interval (CI): 0.79–0.87; P < .001), after adjusting for age, sex, race/ethnicity, smoking status, drinking status, physical activity, hypertension, diabetes, calorie consumption, protein consumption, insulin, glycohemoglobin, TG, creatinine, ALT, and NHANES cycle. RCS analysis revealed no evidence of a nonlinear relationship (P for nonlinearity = .983). ECF mediated 66.3% (95% CI: 53.1–81.6), PFAT 31.7% (95% CI: 23.4–41.3), and CRP 11.1% (95% CI: 7.1–17.4) of the relationship (all P < .001). Sensitivity analyses revealed robust findings. Serum albumin levels demonstrated an inverse association with obesity odds. Mediation analyses confirmed mediating effects of ECF, PFAT and CRP, suggesting serum albumin may be associated with obesity through these mediators.

1. Introduction

Obesity is a chronic, progressive, Obesity has emerged as one of the most formidable global public health challenges of the 21st century,^[1] currently affecting over a billion individuals worldwide.^[2] Beyond being recognized as a significant risk factor for numerous chronic conditions, including cardiovascular disease,^[3] diabetes^[4] and so on, obesity itself has been declared a distinct disease entity that impairs quality of life, reduces life expectancy, and increases mortality risk.^[5]

A study found that lower albumin concentrations correlated with subsequent increases in body weight and fat mass.^[6] An investigation documented an inverse relationship between body mass index (BMI) and serum albumin.^[7] Additionally, another research identified an association between hypoalbuminemia and elevated BMI.^[8] These findings suggest an association between serum albumin and obesity, warranting further exploration of the potential factors contributing to this relationship to better understand this association.

Serum albumin, the most abundant plasma protein in the human body, is synthesized by hepatocytes.^[9] It fulfills multiple physiological functions, including maintaining colloid osmotic pressure (COP),^[10] transporting endogenous substances (such as fatty acids) and exogenous compounds,^[11] and exhibiting antioxidant and anti-inflammatory properties.^[12] Under conditions of low serum albumin, elevated extracellular fluid volume (ECF), resulting from reduced COP leading to water retention,^[13] contributes to increased body weight and BMI^[14]; Elevated free extracellular fat (PFAT), resulting from diminished fatty acid transport leading to excessive adipose accumulation, induces obesity ^[11]; and Adipose tissue deposition, particularly visceral fat, facilitates the secretion of proinflammatory cytokines such as interleukin-6.^[15] This subsequently stimulates hepatic production of acute-phase proteins including C-reactive protein (CRP), while concurrently suppressing albumin synthesis.^[16] Thus, we hypothesize that ECF volume expansion, increased PFAT, and chronic inflammation may constitute 3 key pathways linking low albumin to obesity.

This study aims to examine the association between serum albumin and obesity by analyzing data from the nationally representative, large-scale National Health and Nutrition Examination Survey (NHANES). Concurrently, it seeks to elucidate the potential mediating roles of ECF, PFAT, and CRP in this relationship, thereby providing a more comprehensive understanding of the underlying factors influencing this association.

2. Materials and methods

2.1. Study population and ethics

NHANES is an ongoing cross-sectional survey in the United States designed to assess the health and nutritional status of the civilian, non-institutionalized population at the national level, encompassing both adults and children.^[17] The survey employs a complex, stratified, multistage probability sampling design to ensure accurate representation of the target demographic characteristics.^[18] Survey data and detailed procedural documentation are publicly accessible at https://www.cdc.gov/nchs/nhanes/ (accessed March 6, 2025). For this study, data from 3 NHANES cycles (1999–2004) were utilized, focusing on adults aged ≥ 18 years with complete information on BMI, serum albumin, body composition (via bioelectrical impedance analysis, BIA), and dietary sample weights (Fig. 1).

Figure 1.

Flowchart of the study.

The NHANES protocol (#98-12, https://www.cdc.gov/nchs/nhanes/about/erb.html) was approved by the Ethics Review Board of the National Center for Health Statistics, Centers for Disease Control and Prevention (CDC). Written informed consent was obtained from all participants prior to their inclusion in the survey.

2.2. Body measurement and definition of obesity

At a mobile examination center, trained health technicians measured participants’ weight, height and waist circumference. BMI was calculated as weight in kilograms divided by the square of height in meters.^[19] Obesity is a BMI greater than or equal to 30.^[20]

2.3. Measurement of serum albumin

Albumin concentration was determined using the bromocresol purple (BCP) dye method. Processed serum samples were stored under appropriate refrigerated conditions (2–8°C) and transported to the Collaborative Laboratory Services Division for analysis. Measurements employed the DcX800 dual-color digital endpoint assay. During the reaction, albumin binds with the BCP reagent to form a chromogenic complex, with concentration quantified by measuring absorbance at 600 nm.^[21]

2.4. Covariates

Based on established clinical significance and existing literature, we comprehensively collected all known potential confounding factors affecting obesity.^[22] Demographic characteristics including age, sex, race/ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, others), cigarette smoking, alcohol drinking and physical activity were obtained through standardized questionnaires and in-person interviews. Dietary intake data encompassing calorie, protein, carbohydrate, and fat were collected via the 24-hour dietary interviews. At mobile examination center, standardized physical examinations and laboratory tests were conducted by clinical staff.

Smoking status was categorized as never (<100 lifetime cigarettes), former (≥100 cigarettes with current cessation), and current (≥100 cigarettes and currently smoking). Drinking status was defined as intake of ≥12 alcoholic beverages annually. Physical activity was classified into sedentary (no leisure-time activity), moderate (≥10 minutes of light-to-moderate exercise in the past 30 days), or vigorous (≥10 minutes of high-intensity exercise in the past 30 days). Chronic diseases were identified via structured self-report questionnaires, including hypertension, diabetes, and hyperlipidemia with verification by asking whether a physician had ever diagnosed the participant with these conditions. Systolic and diastolic blood pressure were measured using a mercury sphygmomanometer (W.A. Baum Co. Inc., Copiague). After the participant rested seated for 5 minutes, triplicate measurements were taken at 60-second intervals, typically on the right arm unless contraindicated. The final blood pressure value represented the mean of 3 readings. Body composition of participants, including ECF, intracellular fluid, total body water (TBW), extracellular fat (FAT), free extracellular fat (PFAT), and free muscle mass (fat-free mass), was assessed via BIA using the Xitron 4200 analyzer (Xitron Technologies, Inc., San Diego).

Serum glucose was measured using the hexokinase method using the Hitachi Model 917 multi-channel analyzer (Roche Diagnostics, Indianapolis). Insulin concentrations were measured by radioimmunoassay (RIA) using a double-antibody batch processing technique with multicrystal gamma counters (Models LB 2111 and LB 2104; Berthold Technologies, Bad Wildbad, Germany). Glycohemoglobin (HbA1c) was determined using boronate affinity high-performance liquid chromatography employing Primus CLC 330 and Primus CLC 385 instruments (Primus Corporation, Kansas City). Serum triglyceride (TG) were quantified using the glycerol phosphate oxidase method on the Hitachi Model 917 multi-channel analyzer. Using the Hitachi 704 analyzer (Roche Diagnostic, serum total cholesterol (TC) was determined enzymatically, Indianapolis), high-density lipoprotein cholesterol (HDL-c) was measured by both heparin-manganese precipitation and homogeneous direct methods. And low-density lipoprotein cholesterol was calculated using the Friedewald equation incorporating measured values of TC, TG, and HDL-c.

Using the Hitachi Model 917 multi-channel analyzer (Roche Diagnostics, Indianapolis), Creatinine was quantified using the compensated Jaffe reaction method, alanine aminotransferase (ALT) activity was quantified kinetically. CRP quantification was performed via particle-enhanced immunoturbidimetric assay on a BN-II automated nephelometry system (Model BNII; Siemens Healthcare Diagnostics, Marburg, Germany), strictly adhering to International Federation of Clinical Chemistry standards.

2.5. Statistical analysis

Accounting for the complex survey design of NHANES, all results were weighted to yield nationally representative estimates for the non-institutionalized civilian U.S. residents. The specific calculation method was as follows: sample weights for the 1999 to 2002 period were calculated as two-thirds of the four-year dietary weights, while those for the 2003 to 2004 cycles were derived as one-third of the biennial dietary weights.

Categorical variables are presented as unweighted counts (weighted percentages), while continuous variables are expressed as mean with standard deviation or median with interquartile range. Differences in baseline characteristics for categorical variables were compared using χ² tests, and for continuous variables, using t-tests or Wilcoxon rank-sum tests. The variance inflation factor was employed to assess potential multicollinearity among covariates, with a variance inflation factor value ≥ 5 indicating multicollinearity. Meanwhile, multiple imputation was performed for covariates with missing values.

Logistic regression models were employed to determine odds ratios (ORs) and corresponding 95% confidence intervals (CIs) for the association between serum albumin and obesity. Model 1 was unadjusted for covariates. Model 2 was adjusted for age, sex, race/ethnicity, smoking status, drinking status, physical activity, hypertension, diabetes, and NHANES cycle. Model 3 was additionally adjusted for calorie consumption, protein consumption, insulin, HbA1c, TG, creatinine, and ALT.

Restricted cubic splines (RCS) regression was implemented with knots positioned at the 10th, 50th, and 90th percentiles of serum albumin concentrations to evaluate nonlinear relationships between serum albumin and obesity. Furthermore, we investigated the potential mediating roles of ECF, PFAT, and CRP in the association. Mediation analyses were performed using Sobel tests, the bootstrap method, and the quasi-Bayesian Monte Carlo approach based on normal approximation.

Additionally, fully adjusted sampling-weighted subgroup analyses were conducted to examine associations using logistic regression models, with stratification by key variables, including age (18–35 vs ≥35 years), sex (male vs female), race/ethnicity (non-Hispanic white vs non-Hispanic black vs Mexican American vs others), smoking status (never vs former vs current), drinking status (no vs yes), physical activity (sedentary vs Moderate vs vigorous), hypertension (no vs yes), diabetes (no vs yes), calorie consumption (<2200 vs ≥2200 kcal/d), protein consumption (<75 vs ≥75 g/d), insulin (<10 vs ≥10 μU/mL), HbA1c (<5.2 vs ≥5.2%), TG (<1.06 vs ≥1.06 mmol/L), creatinine (<70 vs ≥70 μmol/L), and ALT (<20 vs ≥20 U/L). Likelihood ratio tests were performed to assess interaction effects across subgroups, with results graphically displayed using forest plots. To assess the robustness of our findings, sensitivity analyses were conducted by excluding extreme outliers with daily calorie consumption below 500 or exceeding 5000 kcal.

All statistical analyses were performed using R version 3.3.2 (http://www.R-project.org) and Free Statistics software version 2.3 (Beijing Free Clinical Medical Technology Co., Ltd., Beijing, China). Statistical significance was defined as a two-sided P-value < .05.

3. Results

3.1. Study population

Among the initial 31,126 participants in the 1999 to 2014 NHANES cycles, 17,061 were aged ≥ 18 years. However, 1730 participants lacked BMI data, 927 lacked serum albumin data, 8347 lacked BIA data, and 176 lacked dietary sample weights. After excluding participants missing any of the above data, a total of 5881 participants were included in the final analysis. Comprehensive inclusion and exclusion criteria are detailed in Figure 1.

3.2. Baseline characteristics

Table 1 presents the baseline characteristics of participants with or without obesity. Extrapolated to the national level, these findings represent approximately 94,774,506 U.S. adults. The mean (standard deviation) age of the participants was 31.5 (10.1) years, with 2851 (48.6%) being female. A total of 1608 participants (27.1%) were living with obesity, and this group exhibited significantly lower serum albumin levels compared to those who were living without obesity. The individuals with obesity tended to be older, female, higher anthropometric measurements such as BMI, weight, and waist circumference, elevated levels of body composition parameters including ECF, intracellular fluid, TBW, FAT, PFAT, and fat-free mass, higher systolic blood pressure, and diastolic blood pressure, non-Hispanic black, higher prevalence of hypertension, diabetes, and hyperlipidemia, elevated levels of key metabolic biomarkers comprising serum glucose, insulin, HbA1c, TG, TC, LDL-cholesterol, CRP, and ALT, lower mean height, less smoking, alcohol consumption, or vigorous physical activity, reduced nutritional intake including daily calorie consumption, protein consumption, and carbohydrate consumption, lower HDL-c, and creatinine. All variables demonstrated statistically significant differences between groups (P < .05), except for participants’ fat consumption.

Table 1.

Baseline characteristics of participants with or without obesity.

Variables	Total	Without obesity	With obesity	Statistics	P value
Unweighted sample size, n	5881	4273	1608	–	–
Weighted sample size, n (%)	94,774,506	69,106,621 (72.9)	25,667,885 (27.1)	–	–
Age (year), mean (SD)	31.5 (10.1)	30.6 (10.1)	33.9 (9.8)	125.642	<.001
Sex, n (%)
Male	3030 (51.4)	2352 (53.8)	678 (45.1)	77.589	<.001
Female	2851 (48.6)	1921 (46.2)	930 (54.9)
BMI (kg/m2), mean (SD)	27.2 (6.0)	24.3 (6.0)	35.0 (4.4)	10,469.535	<.001
Weight (kg), mean (SD)	77.7 (18.4)	70.0 (18.4)	98.4 (15.1)	5261.911	<.001
Height (cm), mean (SD)	168.9 (10.0)	169.4 (10.0)	167.5 (10.0)	39.916	<.001
Waist circumference (cm), mean (SD)	91.8 (14.6)	85.3 (14.6)	109.1 (10.6)	6408.948	<.001
FAT (kg), Mean (SD)	24.5 (11.7)	19.7 (11.7)	37.4 (11.3)	4886.863	<.001
PFAT (%), mean (SD)	31.1 (10.9)	28.5 (10.9)	38.1 (9.8)	1080.028	<.001
FFM (kg), mean (SD)	53.2 (13.7)	50.3 (13.7)	61.0 (13.9)	804.624	<.001
Glucose (mmol/L), mean (SD)	5.0 (1.4)	4.9 (1.4)	5.2 (1.4)	61.535	<.001
Insulin (μU/mL), median (IQR)	9.2 (6.3, 14.2)	7.9 (5.6, 11.3)	15.4 (10.5, 23.0)	645.204	<.001
HbA1c (%), mean (SD)	5.3 (0.8)	5.2 (0.8)	5.5 (0.9)	180.828	<.001
SBP (mm Hg), mean (SD)	111.0 (20.3)	109.3 (20.3)	115.3 (22.0)	100.97	<.001
DBP (mm Hg), mean (SD)	67.3 (15.2)	66.3 (15.2)	70.2 (16.3)	75.336	<.001
TG (mmol/L), median (IQR)	1.1 (0.7, 1.6)	0.9 (0.7, 1.5)	1.4 (0.9, 2.0)	362.692	<.001
TC (mmol/L), mean (SD)	4.9 (1.1)	4.8 (1.1)	5.1 (1.0)	83.255	<.001
HDL-c (mmol/L), mean (SD)	1.3 (0.4)	1.4 (0.4)	1.2 (0.3)	298.348	<.001
LDL-c (mmol/L), mean (SD)	2.9 (0.9)	2.9 (0.9)	3.1 (0.9)	41.949	<.001
CRP (mg/dL), median (IQR)	0.1 (0.0, 0.4)	0.1 (0.0, 0.2)	0.4 (0.2, 0.7)	1028.475	<.001
Creatinine (µmol/L), mean (SD)	71.3 (23.5)	71.9 (23.5)	69.8 (28.0)	9.983	.002
ALT (U/L), median (IQR)	21.0 (16.0, 30.0)	20.0 (15.0, 28.0)	25.0 (18.0, 35.0)	205.296	<.001
ECF (L), mean (SD)	16.6 (3.6)	15.7 (3.6)	18.9 (3.5)	1077.26	<.001
ICF (L), mean (SD)	22.9 (6.6)	21.6 (6.6)	26.3 (6.8)	664.944	<.001
TBW (L), mean (SD)	39.5 (10.0)	37.3 (10.0)	45.2 (10.0)	837.345	<.001
Race/ethnicity, n (%)
Non-Hispanic white	2409 (68.0)	1830 (69.5)	579 (64.0)	53.6	<.001
Non-Hispanic black	1301 (11.1)	855 (9.1)	446 (16.5)
Mexican American	1642 (9.7)	1175 (9.3)	467 (10.6)
Others	529 (11.2)	413 (12.1)	116 (8.9)
Smoking status, n (%)
Never	2633 (54.3)	1830 (53.8)	803 (55.7)	6.658	.036
Former	735 (16.1)	498 (15.7)	237 (17.4)
Current	1368 (29.5)	994 (30.5)	374 (26.9)
Drinking status, n (%)	3353 (75.1)	2441 (77.4)	912 (69.1)	44.003	<.001
Physical activity, n (%)
Sedentary	1936 (28.3)	1321 (26.0)	615 (34.4)	59.852	<.001
Moderate	1327 (25.6)	919 (24.3)	408 (29.1)
Vigorous	2615 (46.1)	2031 (49.7)	584 (36.6)
Hypertension, n (%)	509 (10.0)	240 (6.8)	269 (18.7)	180.955	<.001
Diabetes, n (%)	164 (2.4)	70 (1.4)	94 (5.2)	76.205	<.001
Hyperlipidemia, n (%)	688 (24.4)	404 (22.4)	284 (29.6)	23.593	<.001
Calorie consumption (kcal/d), mean (SD)	2402.9 (1131.8)	2442.1 (1131.8)	2298.6 (1077.8)	18.861	<.001
Protein consumption (g/d), mean (SD)	87.3 (47.7)	88.3 (47.7)	84.4 (44.8)	7.903	.005
Carbohydrate consumption (g/d), mean (SD)	302.3 (149.8)	308.6 (149.8)	285.4 (143.4)	28.249	<.001
Fat consumption (g/d), mean (SD)	87.5 (50.1)	88.1 (50.1)	86.1 (48.3)	1.797	.18
Serum albumin (g/L), mean (SD)	44.1 (3.4)	44.7 (3.4)	42.6 (3.3)	478.478	<.001
Serum albumin, n (%)
Q1 (<42)	1236 (18.0)	630 (12.6)	606 (32.6)	482.033	<.001
Q2 (42–44)	1200 (20.7)	822 (19.2)	378 (24.8)
Q3 (44–46)	1472 (25.4)	1133 (26.1)	339 (23.5)
Q4 (≥46)	1973 (35.9)	1688 (42.2)	285 (19.0)

Categorical variables are presented as unweighted counts (weighted percentages), while continuous variables are expressed as means (SD) or medians (IQR). Differences in baseline characteristics for categorical variables were compared using chi-square tests, whereas continuous variables were analyzed using t-tests or Wilcoxon rank-sum tests.

ALT = alanine aminotransferase, BMI = body mass index, CRP = C-reactive protein, DBP = diastolic blood pressure, ECF = extracellular fluid, FAT = fat mass, FFM = fat-free mass, HbA1c = glycohemoglobin, HDL-c = HDL-cholesterol, ICF = intracellular fluid, IQR = interquartile range, LDL-c = LDL-cholesterol, SBP = systolic blood pressure, SD = standard deviation, TBW = total body water, TC = total cholesterol, TG = triglyceride.

3.3. Association between serum albumin and obesity

After adjusting for age, sex, race and ethnicity, smoking status, drinking status, physical activity, hypertension, diabetes, calorie consumption, protein consumption, insulin, HbA1c, TG, creatinine, ALT, and NHANES cycle, elevated serum albumin demonstrated a significant inverse association with the odds of obesity (adjusted OR = 0.83; 95% CI: 0.79–0.87; P < .001). This association persisted when serum albumin was analyzed as a quartile-based categorical variable. Compared with individuals with the lowest quartile (Q1) of serum albumin (<42 g/L), the adjusted ORs for obesity in Q2 (42–44 g/L), Q3 (44–46 g/L) and Q4 (≥46 g/L) were 0.56 (95% CI: 0.39–0.71, P = .003), 0.38 (95% CI: 0.26–0.55, P < .001), and 0.2 (95% CI: 0.13–0.31, P < .001) (Table 2), respectively. Furthermore, RCS analysis revealed no evidence of nonlinearity (P for nonlinearity = .983) (Fig. 2).

Table 2.

Association between serum albumin and obesity.

			Model 1*		Model 2†		Model 3‡
Variable	Total	Event (%)	OR (95% CI)	P value	OR (95% CI)	P value	OR (95% CI)	P value
Albumin, (g/L)	5881	1608 (27.1)	0.81 (0.78–0.85)	<.001	0.82 (0.78–0.85)	<.001	0.83 (0.79–0.87)	<.001
Albumin (Quartiles)
Q1 (<42)	1236	606 (49.0)	1 (Ref)	–	1 (Ref)	–	1 (Ref)	–
Q2 (42–44)	1200	378 (32.5)	0.5 (0.37–0.68)	<.001	0.52 (0.37–0.73)	<.001	0.56 (0.39–0.71)	.003
Q3 (44–46)	1472	339 (25.1)	0.35 (0.26–0.47)	<.001	0.35 (0.25–0.49)	<.001	0.38 (0.26–0.55)	<.001
Q4 (≥46)	1973	285 (14.4)	0.17 (0.12–0.25)	<.001	0.18 (0.12–0.28)	<.001	0.2 (0.13–0.31)	<.001
P for trend	–	–	–	<.001	–	<.001	–	<.001

ALT = alanine aminotransferase, HbA1c = glycohemoglobin, TG = triglyceride.

^*Model 1 was unadjusted for covariates.

^†Model 2 was adjusted for age, sex, race/ethnicity, smoking status, drinking status, physical activity, hypertension, diabetes, and NHANES cycle.

^‡Model 3 was additionally adjusted for calorie consumption, protein consumption, insulin, HbA1c, TG, creatinine, and ALT. All VIF < 2.9; Tjur’s R² = 0.31; Power = 99%.

Figure 2.

Association between serum albumin and obesity odds ratio after full adjustment. The solid red line represents the smoothed OR for hypertension risk as albumin changes, and the shaded area indicates the 95% CI. Adjusted for age, sex, race/ethnicity, smoking status, drinking status, physical activity, hypertension, diabetes, calorie consumption, protein consumption, insulin, HbA1c, TG, creatinine, ALT, and NHANES cycle. ALT = alanine aminotransferase, CI = confidence interval, HbA1c = glycohemoglobin, OR = odds ratio, TG = triglyceride.

3.4. Stratified analyses based on variables

In multiple subgroups, stratified analysis was conducted to evaluate potential effect modifications on the association between serum albumin and obesity. After stratifying by age, sex, race/ethnicity, cigarette smoking, alcohol drinking, physical activity, hypertension, diabetes, calorie consumption, protein consumption, creatinine, ALT, no significant interactions were observed in any subgroups. While P < .05 for the interaction of age, and ALT, the findings may not reach clinical significance given considerations of the consistent directionality of associations and multiple testing (Fig. 3, Fig. S1, Supplemental Digital Content, https://links.lww.com/MD/R432).

Figure 3.

Association between albumin and obesity according to basic features. Except for the stratification component itself, each stratification factor was adjusted for all other variables, including age, sex, race/ethnicity, smoking status, drinking status, physical activity, hypertension, diabetes, calorie consumption, protein consumption, insulin, HbA1c, TG, creatinine, ALT, and NHANES cycle. ALT = alanine aminotransferase, HbA1c = glycohemoglobin, TG = triglyceride.

3.5. Sensitivity analysis

Following the exclusion of individuals with extreme calorie consumption (<500 or >5000 kcal/d), 5678 individuals were left, and the association between serum albumin and obesity remained stable. Serum albumin demonstrated an inverse association with obesity (adjusted OR = 0.81; 95% CI: 0.75–0.88; P < .001). Compared with individuals with the lowest serum albumin level (Q1), the adjusted ORs for obesity in Q2, Q3 and Q4 were 0.56 (95% CI: 0.38–0.83, P = .006), 0.38 (95% CI: 0.26–0.55, P < .001), and 0.19 (95% CI: 0.12–0.31, P < .001) (Table S1, Supplemental Digital Content, https://links.lww.com/MD/R432), respectively. In addition, the relationship between serum albumin and obesity remained stable in the unweighted data analysis (Table S2, Supplemental Digital Content, https://links.lww.com/MD/R432).

3.6. Mediation analysis

Figure 4 illustrates the underlying mediating roles of ECF, PFAT and CRP in the relationship between serum albumin and obesity. We observed that higher serum albumin was related to lower ECF (β = −0.17, 95% CI: −0.21 to −0.13, P < .001), PFAT (β = −0.40, 95% CI: −0.51 to −0.30, P < .001) and CRP (β = −0.04, 95% CI: −0.05 to −0.03, P < .001). Meanwhile, higher ECF (OR = 2.11, 95% CI: 1.99–2.25, P < .001), PFAT (OR = 1.17, 95% CI: 1.15–1.20, P < .001) and CRP (OR = 1.98, 95% CI: 1.56–2.51, P < .001) were significantly related to a higher odds of obesity.

Figure 4.

Mediation analysis of (A) Extracellular fluid, (B) Percent body fat, and (C) C-reactive protein in the association between serum albumin and obesity.

Furthermore, a significant mediating effect of ECF (proportion of mediation: 66.3%, 95% CI: 53.1–81.6, P < .001), PFAT (proportion of mediation: 31.7%, 95% CI: 23.4–41.3, P < .001) and CRP (proportion of mediation: 11.1%, 95% CI: 7.1–17.4, P < .001) in the relationship between serum albumin and obesity was observed.

4. Discussion

This study systematically examined the relationship between serum albumin and obesity in U.S. adults using a nationally representative NHANES dataset, while validating the mediating roles of ECF, PFAT, and CRP. Key findings include: Serum albumin demonstrated a significant inverse association with obesity indices; ECF, PFAT, and CRP all exerted statistically significant mediating effects on the albumin-obesity association; ECF emerged as the most influential mediator among the 3 variables; The inverse albumin-obesity relationship remained consistent across all subgroups, though subtle variations were observed among different demographic populations.

Previous studies have established a significant relationship between serum albumin and obesity. For instance, a study by Alessio Basolo et al^[6] involving 325 adult Native Americans revealed that lower serum albumin concentrations were associated with weight gain, with participants experiencing an average increase of 7.5 kg over a 6-year period. A retrospective analysis of clinical trial data by Jennifer Powers Carson and Jyoti Arora^[7] revealed the strongest inverse correlation between BMI and serum albumin among 334 American patients with comorbid overweight/obesity and diabetes. Stuart B. Prenner et al^[8] conducted a study involving 118 adults with heart failure with preserved ejection fraction (HFpEF), revealing that subjects with lower serum albumin exhibited higher BMI. These findings align with our observations of an inverse relationship between serum albumin concentrations and obesity prevalence in 5881 U.S. adults, suggesting that serum albumin may serve as a protective factor in the development and progression of obesity. Notably, our study further identified that ECF, PFAT, and CRP mediate the association between serum albumin and obesity, which provides a both comprehensive and novel research perspective.

A key finding of our study is the robust mediating effect of ECF in the serum albumin-obesity relationship, which aligns with the characteristic physiological phenomenon of ECF expansion frequently accompanying obesity. A study examining serum albumin in relation to fluid status among hemodialysis patients revealed that elevated albumin levels demonstrated strong inverse correlations with both the ECF/TBW ratio (r = −0.684) and actual body weight changes (r = −0.651).^[23] This finding and our mediation analysis results mutually reinforce each other. PFAT also demonstrated a significant mediating effect in the relationship between serum albumin and obesity. A study revealed a significant negative correlation between serum albumin levels and fat mass across both healthy individuals and patients with inflamed end-stage renal disease.^[24] The mediating role of CRP aligns with the physiological manifestation of chronic inflammation that frequently accompanies obesity. A study employing multivariable linear regression analysis demonstrated an inverse correlation between serum albumin levels and CRP concentrations^[25]; another investigation revealed a significant positive association between CRP levels and BMI.^[26]

Based on our findings and extant literature, we propose the following possible mechanisms to explain the inverse association between serum albumin and obesity: Physiologically, serum albumin serves as the primary contributor to COP, accounting for approximately 75 to 80% of its total effect.^[27] By drawing and retaining water molecules, it controls the distribution of ECF across the vascular and extravascular compartments.^[28] When albumin concentration decreases, COP declines proportionally, leading to reduced absorption of interstitial fluid. This imbalance consequently promotes interstitial fluid accumulation, manifesting as edema^[29] and weight gain; Serum albumin serves as the main carrier for fatty acids, binding and delivering them to tissues such as myocardial and skeletal muscle to provide cellular energy substrates.^[13] Under normal physiological conditions, each mole of serum albumin binds approximately 0.1 to 2.0 moles of fatty acids.^[30] Diminished albumin concentrations may disrupt fatty acid distribution and metabolism, thereby promoting ectopic lipid deposition^[31]; and Obesity can contribute to a state of chronic inflammation, characterized by the secretion of multiple pro-inflammatory cytokines from adipose tissue, including interleukin-6^[32] and tumor necrosis factor-alpha.^[33] Under inflammatory conditions, the liver prioritizes the synthesis of acute-phase proteins, such as CRP, while concurrently suppressing albumin production.^[34] This shift in hepatic protein synthesis consequently leads to a reduction in serum albumin levels.^[35]

Notably, subgroup analyses in this study revealed a stronger association between serum albumin and obesity in the female population. This may be related to the generally lower baseline serum albumin levels observed in women, as indicated in Table 1. Furthermore, the relationship between serum albumin and obesity was more pronounced among individuals with lower levels of ALT. Since serum albumin is primarily synthesized by hepatocytes,^[9] and elevated ALT levels serve as a sensitive indicator of hepatocellular injury,^[36] normal liver function is essential for maintaining adequate serum albumin concentrations.

Despite the study’s strengths, including a comprehensive analysis of a larger nationally representative sample and meticulous adjustment for multiple confounding factors, several limitations warrant Acknowledgments. First, although multivariate adjusted regression models and stratified sensitivity analyses were employed to comprehensively assess associations, residual confounding cannot be entirely excluded. While the NHANES dataset encompasses a broad spectrum of variables, not all potential confounders could be accounted for. Second, the cross-sectional design precludes definitive causal inference regarding the bidirectional relationship between serum albumin and obesity. Third, the study focused exclusively on the U.S. population; validation of the albumin-obesity association across other populations would significantly enhance the generalizability of the findings.

5. Conclusion

This study established a significant inverse association between serum albumin and obesity among U.S. adults while validating the mediating roles of ECF, PFAT, and CRP. These findings can be used to inform future studies aiming to better understand the association of serum albumin and obesity and to help inform guidelines for clinicians on how to correctly interpret and utilize serum albumin data for individuals with obesity. Future investigations should prioritize prospective designs and mechanistic studies to further elucidate the causal directionality of these associations.

Acknowledgments

The authors would like to thank Dr Jie Liu, Dr Haibo Li and the clinical scientist team for statistical support and study design consultations. Additionally, the authors would like to thank the National Center for Health Statistics (NCHS) and the Centers for Disease Control and Prevention (CDC) for making NHANES data publicly available. The authors also acknowledge the efforts of NHANES participants and data collection staff.

Author contributions

Conceptualization: Yan Wang.

Data curation: Yan Wang.

Formal analysis: Yan Wang, Qiang Lu, Lei Zhang, Guimei Wei, Xueyan Shen, Yan Peng.

Methodology: Yan Wang.

Validation: Yan Wang.

Visualization: Yan Wang.

Writing – original draft: Yan Wang.

Writing – review & editing: Yan Wang.

Supplementary Material

medi-105-e47802-s001.docx ^{(1MB, docx)}